LEARNING TO BE AN “ECOLOGICAL” COACH

DEVELOPING ATTUNED & ADAPTABLE COACHING SKILLS

ROB GRAY, PH.D.

Copyright © 2024 by Rob Gray, Ph.D./Perception Action

Consulting & Education
LLC
All rights reserved.
No part of this book may be reproduced in any form or by
any electronic or mechanical means, including information
storage and retrieval systems, without written permission from
the author, except for the use of brief quotations in a book
review.
All figures are reproduced with permission from rights holders.
To Sara, Angus, Molly, and Jonah. Thank you for all the
support, love, laughs, and fun!
CONTENTS

Preface: From A Theory Of Skill To Guiding Others In

Becoming Skillful
Knowledge “Of” & “About” & The Problem Of Path
Dependency In Coaching
The Value (And Limitations) Of Using Skill Acquisition Theory &
Research In Coaching
An Ecological Approach To Coaching – The Coach As A
Learner & Explorer
“Having Your Cake And Eating It Too”: Applying Representative
Learning Design
Designing Effective Constraints & Using The Cla
Making A Skill Easier For A New Learner
Coaching To The Individual Athlete & Working With Groups
Educating A Coach’s Attention & Learning To Observe
Movement
Perceiving The Affordances Of Others
Using Imagery, Demonstration & Verbal Instruction
What To Do Next: Adapting Practice
Creating A Culture And A Form Of Life
Inspiring Motivation, Autonomy & Confidence
Embracing The Chaos And Adapting To The Constraints Of
Coaching
Rethinking How We Track & Assess Progress
Tools For Coach Education: Workshops, Mentorships, &
Constraining The Coach
Notes
About The Author

PREFACE: FROM A THEORY OF

SKILL TO GUIDING OTHERS IN
BECOMING SKILLFUL
Who was the best coach you ever had? Mine was a hockey
coach named Glenn Philips whom I worked with when I was
9-10 years old. Growing up I was an ice hockey goaltender. I
played until my second year of college when I realized I was
much more likely to make a career by focusing on a classroom
than an ice rink! Like most Canadian boys my age, I started
skating and playing hockey almost as soon as I could walk. I
spent many hours on an ice rink my dad made in our
backyard by standing out in the cold with a garden hose at
night. Thanks, old man! I was originally a defenseman when I
started playing organized hockey because the team already had
a goalie. But a few days into the season, his dad pulled him
off the team because he was getting so stressed out before
games, that he was getting physically ill, developing ulcer-like
symptoms. Our coach asked: “Does anyone want to take his
spot?” -I immediately raised my hand! I like to say that this
early exposure to hard pieces of rubber flying at my head led
to my eventual fascination with tau and perceiving time to
collision – it was necessary for my survival!
But back to Coach Phillips. What made him an effective coach?
I still remember one of the first practice activities he had me
do. To be an effective hockey goaltender you can’t just stand
back in the net and wait to stop shots. You need to “close
down the angle” with the shooter – move away from the
comfort of your net and out towards them. Place your hand
on the page you are reading right now. Notice that you can
still see a lot of the page. Now move your hand towards your
face. Less page visible – less open net to shoot at!
The way that I had been taught to “close down the angle” up
to that point was through a lot of instruction telling me where
to be and when. It was honestly something I wasn’t motivated
to focus a lot of attention on at that point – it was much
more fun to make spectacular saves with your glove or stick!
But then came Coach Phillips.
In the activity, he positioned a group of shooters in a
semi-circle around me. His instructions were: “When I pass the
puck to a shooter all you are allowed to do is skate out
towards them. You can’t move your glove, body, or stick to
stop their shot in any way”. Well, this sounds dumb, I thought.
But quickly something amazing came to light. Hardly any of the
shooters scored goals. Most missed the net completely while
others shot the puck right at me.
Coach Phillips was a master at creating this type of formative
learning environment. He helped me to experience things for
myself instead of just telling me things. I can still picture the
wry smile on his face when I turned to look at him at the end
of the “close the angle” activity. While other coaches constantly
shouted things, he let his unique practice activities do most of
the talking for him. He had a very different view of what it
means to educate someone. Rather than trying to transmit
knowledge to me, he took the approach Carl Woods and I
discussed in a recent paper
1
: educere, “to lead out or guide”.
And, although he would never use these words, of course, he
was a master at “constraining to afford”. Instead of telling me I
had to cut down the angle on the shooter, he took away some
of my movement solutions so that I would explore different
ones. If I started to wander too far from my net, he would
pass the puck to the shooter a little more quickly. He was
what I would call an “ecological coach”.
The Ecological Athlete & Coach
In my first two books, “How We Learn to Move”
2 and
“Learning Optimal Movement”3
, the focus was on how athletes
become skillful – promoting the Ecological Approach to Skill
and Ecological Dynamics Theory4
. In presentations, I typically
boil this down to six key principles:
1)
Athlete-Environment Symmetry. Skill is all about the
relationship the athlete creates and maintains with their
environment. It is not all, asymmetrically, about what’s in their
head. This is where the term “ecological” comes from – we are
always including the performer’s ecology (their physical
surroundings, the context) in looking at skill.

2)
Direct Perception of Affordances. We pick up opportunities
for action (affordances) directly from our environment without
the need to process, interpret, or add to the information (e.g.,
from previous experiences). As JJ Gibson5 so wonderfully put
it: our environment structures the sensory information we
receive in meaningful ways.

3)
Information-Movement Coupling. We control our actions
prospectively by establishing and maintaining a relationship
between the information we pick up from the environment and
some aspect(s) of our movement. We don’t predict, plan, or
program.

4)
Self-Organization with respect to Constraints. The
movements and positions of our body that we use to execute
an action occur through a process of self-organization (the
body finds its own solution if you will) NOT via a central
executive in our head that tells our body what to do. Critically,
this self-organization is shaped by the task (e.g., size of playing
area), environmental (e.g., presence of wind), and individual
(e.g., level of fatigue) constraints.

5)
“Repetition without Repetition”. Successfully repeating a
performance outcome (e.g., always getting a basketball in the
hoop) is NOT achieved by repeating the same movement over
and over. We need variability in our movement from execution
to execution to allow us to adapt to the variability in the
internal and external environment.

6)
Direct Learning Through Experience. With practice and
experience, we become more skillful by changing our
relationship with our environment and adapting our
information-movement control laws. Skill does NOT come from
an accumulation of knowledge in the form of mental models
and motor programs.
In this book, I want to switch from focusing on skill
development in the athlete to the skill development of coaches.
How can one become an “ecological coach”? That is, coaching
in a manner that aligns with these key principles. Not
surprisingly, given that it is based on a conceptualization of skill
that is very different from what most of us were taught (so
much so, that I called it a “revolution” in my first book), the
methods of coaching we need are going to be very different
too! How can coaches develop skills in manipulating constraints,
picking up the affordances of the athletes we are working with,
observing movement, stepping in with instruction to guide
exploration, co-adapting practice, using demonstration, teaching
in group settings, use data to inform their practice design?
Notice I am calling these things “skills”. We are not going to
just acquire knowledge about these things. We are going to
focus on how coaches can develop their ability to do them, in
a specific context. This has been identified as a key limitation in
most coach development and education programs – as we will
discuss more in the next chapter, we tend to focus on what to
coach rather than how to coach it
6
.
So, not only are we going to learn to coach in a manner
consistent with the ecological approach we are going to get all
meta and take an ecological approach to develop our coaching
skills. We are going to add constraints on the ways we can
coach, and purposely add variability to the coaching
environment – to challenge the coach not just the athlete. We
will educate our attention to different sources of information
from our athletes’ interaction with the practice environment,
etc…but we will get to that!
As the subtitle suggests, in this book I want to focus on two
aspects of coaching in particular – attunement and adaptability.
As I will discuss in more detail in Chapter 1, we spend a lot of
time in coaching looking at our athletes through a filter. The
filter created by our knowledge about what things “should look
like”. We look for “errors” to correct based on some ideal
technical model we have been taught. We view our athletes to
find solutions for them – after all, we are the coaches – we
are the ones who are supposed to know what to do!
In this book, I want to remove this filter and help coaches
perceive (become attuned to the information in) the practice
environment directly. There is information in the environment
available to a coach that can be used to achieve their goal of
making their athlete more skillful. Information about an athlete’s
affordances and action capabilities. Information about an
athlete’s exploration of different movement solutions and how
we might help guide them in this search. Information about
how they are adapting (or not) to different constraints they
face. Information about whether they are using specifying
information to control their movement. Not only does this not
require interpretation and supplementation via our “knowledge
about what it should look like” filter – this filter can blind us
from picking this information up. We need to remove it and
become better attuned to what is right there in front of us – a
rich environment full of information.
Over the past few years, I have had the opportunity to observe
many very high-level coaches (MLB, NFL, NBA, NHL, Premier
League, Olympics, etc.) and I have come to a strong
conclusion: the secret sauce of coaching is what you do next.
Anybody can set up a practice drill they saw on YouTube or
read in a book. The real art of coaching comes in what you
do next. What do you say (or not say!) based on how the
athletes are performing in your drill? What do you change
about the drill? How do you use the feedback and reflections
from the athletes? After this drill is done do you just stick with
what you have pre-planned for that day, or might you do
something different based on what you saw? Highly skilled
coaches are highly adaptable. They are, as Olympic rowing
coach Craig Morris so wonderfully puts it, prepared, but not
planned7
:
It is a journey of learning to pay attention; of being prepared,
skillfully stretching toward things that are there, but not
over-planned, remaining open and responsive to emergent
possibilities along the way.

But Why Would I Want to Be an “Ecological Coach”?

When I talk about applying theories of skill acquisition to
coaching, I often get the reaction: “I don’t need a theory, I
just use the methods and tools that work for me”. But in
every one of these cases, after talking to the person for a bit,
it becomes clear that they are guided by a theory. They just
aren’t aware of it. The key principles that have dominated
coaching for years and years (focus on “the fundamentals”,
correcting technique, improving through strict repetition, the
coach has the answers, etc.) are all ASSUMPTIONS about how
we become skillful just like those in the ecological approach.
They have just been used for so long that many people now
treat them as facts. As I will discuss in more detail in Chapter
2, all coaching is guided by an underlying theory of how we
think people learn.
The next typical response I get is: “Well, there is a reason we
have stopped questioning the underlying assumptions we use in
coaching - the methods work!”. Yes, indeed they do. No one is
denying that using prescriptive coaching, repetition, etc. will lead
to improvements in performance. Human beings are learning
machines. Almost anything we do, including nothing (AKA the
famous placebo effect), will lead to performance improvements.
We are not talking about whether something does or doesn’t
work. We are talking about relative effectiveness. More and
more there are restrictions on practice time – whether it's due
to availability of facilities or concerns with player workload. Is
there a way that we can coach that not only leads to
performance improvements but also makes athletes more
adaptable to changes in constraints, better decision makers, less
injury-prone, and more creative? A way that is more fun,
engaging, and motivating for the young athlete? As I continue
to catalog on this webpage8
, there is an ever-growing body of
evidence to suggest that “ecological” coaching has many benefits
(along with superior performance) above and beyond
prescriptive coaching (which, by the way, does lead to improved
performance in most of these studies just not as much). To
put it frankly, all coaching is guided by a theory of how we
learn so why not pick the better one?
But if you are not convinced, I hope you will still read on. We
are going to explore lots of different novel and innovative tools
for coach development that I think can be highly beneficial for
anyone.
Dissatisfaction with the Dominant Approach to Coach Education
Research examining coach education has revealed a consistent
theme: the dominant model of formal coach education is
broken. In a recent article published by Selimi et al. (2023)7,
comments about development courses made by coaches were
striking: coaching courses were seen as ‘rudimentary’,
‘outdated’, ‘repetitive’, and needing ‘an absolute revamp’.
Furthermore, they did not effectively address the complexities of
modern coaching. Another criticism is that coaches spend a
significant amount of time learning in settings that do not
represent their sport. To reiterate a frustration expressed in my
first book: there must be a better way!
“The Year of the Coach” - My Hidden Exploration Agenda
Writing this book has been a bit different from the first two.
For those, I reached a spot on my exploration path and
suddenly picked up an affordance (e.g., “I could provide a
resource for people to start learning about the ecological
approach to skill”). Here, I picked up the affordance (providing
a resource for ecological coach development) very early in the
process - a few days after I finished my last book! This is the
hidden agenda behind my declaring 2023 to be the “year of
the coach” on the Perception & Action Podcast. Having
identified the affordance I wanted to realize, I created multiple
learning opportunities for myself by inviting many great coaches
and instructors for podcast interviews. I was interested in
gathering information from their experiences in using an
ecological approach. What worked well right away? What is still
a struggle? As you will see there are some quotes from these
interviews in different places in the book.
This book has also been largely informed by the amazing
opportunity I have had over the past two years – serving as a
skill acquisition specialist for the Boston Red Sox. Working
firsthand with so many amazing coaches has taught me that
becoming an ecological coach is as much an art as it is a
science. But, as I hope to convince you, I think it's time we
dropped the strict dichotomy between those two things,
especially when it comes to coaching.
So, thanks again for sharing my journey! Cheers for now and
keep em’ coupled – for the coach not just for the athlete this
time!

1
KNOWLEDGE “OF” & “ABOUT” &
THE PROBLEM OF PATH
DEPENDENCY IN COACHING
Q uick, tell me everything you know about bicycle riding. You
could probably go on and on for several minutes. Most bicycles
have two wheels and brakes, some have gears, you can ride
on a road or use a mountain bike on a trail, etc. Now tell me
what you know about how to ride a bike. No, cancel that. Just
answer one question for me: what direction do you move to
make a bike go to the left when riding at full speed? Well,
that’s obvious, I move to the left – turning the handlebars in
the direction I want to go. If that was your answer, I am here
to tell you that you are wrong. As is explained in detail here
1
,
your initial movement to go left is to the right!
I use this example to illustrate James Gibson’s critical distinction
between “knowledge of” and “knowledge about” our
environment2
. Knowledge of facilitates knowing HOW to
execute an action because it involves the pickup of information
and affordances used to control action directly. It can only be
demonstrated through doing because it involves a relationship
between information and movement. It is scaled relative to our
individual constraints and action capabilities. It involves knowing
whether an object is hittable, reachable, catchable, etc. for us,
not that is 10 feet away, moving at 50 mph etc.
Knowledge about facilitates our conceptual, indirect
“understanding” of our environment. It is demonstrated either
verbally or by using symbols or pictures (e.g., drawing a
football play out on a whiteboard). It is scaled in abstract units
of physics (meters, feet, miles per hour, etc) rather than being
body-scaled because it refers to the typical, the ideal. In terms
of action, “knowledge about” can only be used to describe what
to perceive and what to do (e.g., turn right when you get to
the 50-yard line).
While writing this book I have been asking bike riders at our
local park the question I posed to you at the start of the
chapter – while they are resting not riding! To date, over 90%
of them have got it wrong. Yet, every single one of them
successfully navigated the bend leading into the park. Their
perceptual-motor system knew the correct answer even though
“they” didn’t! They have knowledge of steering a bike that is
completely at odds with their knowledge about it. Good thing
we have self-organization because the boss is out to lunch on
this one!
This effect works in the other direction too. I have a lot of
knowledge about baseball batting. To be immodest, I
understand the constraints, information, attractors etc. as well as
probably anyone in the world. But I have a very strict rule I
use when doing on-field demonstrations – get someone else to
swing the bat! If I did my own demos, I would immediately
lose all credibility because my knowledge of hitting is very low.
I do not have nearly enough experience performing the skill to
use the information to effectively control my movements at an
elite level. Nor have I learned enough about my own action
capacities and individual constraints to effectively pick up
affordances and calibrate my movements.
This distinction is so significant that it is frequently argued that
these two types of knowledge involve completely different areas
of the brain! As first proposed by Milner and Goodale3 in
their two streams hypothesis, the visual information going from
our eyes to our brain seems to split into two distinct streams
or pathways after it leaves our primary visual cortex (illustrated
in Figure 1.1). The ventral stream, which goes from the occipital
lobe down to the temporal lobe on the side of your head,
seems to handle knowledge about. It is involved in the
identification and recognition of objects. The dorsal stream,
which goes from the occipital lobe up to the parietal lobe near
the top of your head, does knowledge of. It handles the control
of action.
A diagram of a brain Description automatically generated
Figure 1.1 – The Two Streams Hypothesis
Evidence consistent with this can be seen from research on
brain-damaged patients. For example, take the case of H.M.
who had damage to his ventral stream resulting from
experimental surgery attempting to control his epileptic seizures
4
. One of the tasks he was asked to perform after the
surgery was mirror drawing, which is actually a great task to
demonstrate the different types of knowledge and processes like
calibration. Start by making a little course on a piece of paper
like the star-shaped one shown in Figure 1.2. Now take it and
place it beside a mirror. Your task is to draw a line with a
pencil, staying within your course boundaries. You lose a point
every time you hit a boundary. But you must do it while
looking at your hand in the mirror instead of directly at it.
Think about the two types of knowledge here. Most people
immediately have some knowledge about this task. They can
intuit that the mirror is going to cause the image of their hand
to be reversed so they can describe what to do: “when I am
drawing the line on a path that goes to the right, I am going
to need to move my hand to the left”. But if you try this for
yourself you will immediately see that this does little to help
you! Most people hit the boundaries a lot and are very slow at
completing the course. At first, anyways! You don’t have
knowledge of how to do this task effectively. That comes
through actual experience doing it. Specifically, you need to
re-calibrate the information-movement control law for drawing
(change the relationship between what you see and how you
move). Over time, with practice and the development of
knowledge of, we get faster and more accurate at mirror
drawing.
A graph of a graph of a graph Description automatically
generated with medium confidence
Figure 1.2 – The Mirror Drawing Test
This was also the case for H.M. Over the course of several
sessions, he got faster and more accurate too (as illustrated on
the right side of Figure 1.2). His intact dorsal stream
accumulated knowledge of
. But what about his knowledge
about? Could he better describe what he was doing while
performing the mirror drawing task? No. He could not even
remember that he had done the task ever before! Every day
the experimenter showed up, H.M. was completely convinced he
was performing the mirror drawing task for the very first time.
The dissociation between knowledge of and about can also be
seen with patient DF, who developed brain damage from
carbon monoxide poisoning5
. When asked to make a verbal
judgment about whether a line was vertical or tilted 20 deg to
the right, DF could not reliably tell the difference. But when
asked to post a letter into a mail slot that was positioned at
different angles, she perfectly adjusted her hand orientation to
the orientation of the mail slot. Her knowledge about was
disrupted while her knowledge of was completely intact.
Why is this relevant to coach development? Because the vast
majority of coach development programs involve accumulating
knowledge about coaching. That is, we learn to be able to
describe and conceptualize what we would do in working with
an athlete. Coaches acquire a lot of knowledge about what to
coach. Don’t use blocked practice. Make sure to use external
focus of attention instructions. They get very few opportunities
to develop knowledge of coaching. And, as we have just seen,
knowledge about is not sufficient for actually being able to
perform an action – doing it, in a specific context, with a
specific set of constraints. This problem can be seen in a
meta-analysis of 285 coach professional development programs
conducted by Lefebvre and colleagues6. The vast majority
(>200) were focused on developing coaches’ sport-specific
knowledge about (
e.g., movement fundamentals, technical and
tactical skills), while very few (<25) provided training on how to
apply this knowledge (i.e., develop knowledge of
).
I think this effect can also be seen in data from research
examining a coach’s preferred methods for acquiring knowledge
7
. By far, the most popular method identified (55% of
participants) was interactions with other coaches (watching them
run practice and having peer discussions). The more structured
and formal ways we present knowledge to coaches, including
formal education (1.5%), professional development activities
(5.8%), and reading books and articles (13.6%), were much less
popular. One of the reasons why I think this preference exists
is because these less popular methods do very little to teach
coaches knowledge of – how to actually do it, in context. We
know from a large body of research and anecdotal evidence,
that one of the best ways for an athlete to become more
skillful is by observing others performing the action8
. Coaches
have figured this out too!
I think it’s also notable that coaches ranked running their own
practices quite low (4.9%) as a preferred method for acquiring
coaching knowledge. This mirrors something I had said before
when talking about athletes – “the game is typically NOT the
best teacher”. It involves performing not learning. For coaches,
running practices are typically all about making athletes more
skillful not making yourself more skillful. For the latter, we are
going to have to move away from the typical practice and add
some things (constraints!) designed specifically for coach
development. This is exactly where we are headed in this book!
Adapted vs. Adaptive and Path-Dependency in Coaching
Why? One of my favorite words to use in my role as a skill
acquisition consultant. A couple of years ago, I was watching
my stepson engage in pickleball practice. The coach, who I later
learned was in his first year of coaching, was running a drill
designed to improve serving. For it, players waited in lines to
make a serve toward hula hoops placed on the ground on the
other side of the court. Every time they hit the ball into a
hoop their team got a point. After practice, the coach
approached me because he had heard about what I do for a
living. And I was happy to talk with an eager young coach!
I began with my favorite question: why? Why do you run the
serving drill that way? His response was: “Players need to be
able to move the ball around the service box so, for example,
they could make their opponent play a backhand or short hop
the ball with a deep serve. The rings help them learn to do
that”. I asked him whether he had ever thought about using
other players instead of rings. For example, you could have a
player stand in a particular location and then tell the server
“You get one point if you make them play it on their
backhand” or “you get a point if they have to move backward
to play the ball”. For me, this is a much-preferred activity
because it teaches the server to pick-up the affordance (based
on information from their environment) for “forcing a backhand
return” and allows another player to practice service returns
instead of spending that time waiting in line. His response to
this was one that I hear very often that makes me cringe:
“Well, that’s the way we have always done it. That’s the way I
was taught”.
My issue here is not so much that he was using the “wrong”
activity. My problem is not so much with any one drill – it is
the resistance to move off that drill and explore. Considering a
variant of the activity and trying something different seemed so
foreign to him. He was adapted not adaptive. This is what I do
when I want to teach serving - stuck in a place.
In an excellent paper by Wood and colleagues9
, which I will
be referring to a lot in this book, they discuss the difference
between these two concepts. Coach education often results in
coaches becoming adapted (“to adjust or modify”) to a specific
environment. We learn what to do when coaching through a
process of developing knowledge about. That is, the
tried-and-true coaching methods. The problem with being
adapted is that is a static state. We adjust and then stay at
this new place. There is little room for innovation, creativity,
and sensitivity to individual differences or changing constraints
(e.g., technology, equipment, or rules).
It also leads to another big problem that exists in a lot of
coaching: path dependency. We do what we do because that’s
the way we were taught and that’s the way our coach was
taught, and so on and so on. Coaching methodology continues
on an unchallenged path. Think about all the major changes
we have seen in the practice environment in recent years. The
proliferation of monitoring and feedback technologies. The
growth of remote coaching. Dealing with fewer in-person
sessions, larger groups, wider ranges in ability, etc. We can’t
just always use the same methods we were taught. We need
to be adaptive.
Being adaptive requires continuous engagement with the practice
environment – picking up information about the affordances of
others and about how our athletes are exploring. Being open to
there being multiple possible solutions to the movement
problems you create– rather than just the one “correct” one.
Thinking beyond what is going on in the current practice
session to what one might do in the future. It is about letting
things emerge and reveal the path ahead dynamically rather
than having a pre-set plan that you must stick to. For those
paying attention, the distinction between being adapted vs
adaptive in coaching exactly parallels the distinction between
adjustability and adaptability in skilled performance I discussed
in Chapter 8 of my first book.
This problem of path dependency was summed up nicely in a
recent article by Smith et al10: “This may lead some coaches
to implement methods of practice without necessarily
understanding the why behind the chosen activity and result in
a form of content regurgitation from a position of authority.”
We have lost the “why” in coaching.
The Dangers of the “Knowledge About” Lens
Along with the problem of limiting adaptability and creativity in
coaching, the overemphasis on knowledge about creates an
even more serious problem – it creates a lens that can cloud
our observations. This is articulated wonderfully by Olympic
slalom canoe coach Craig Morris in his article “On the wisdom
of not-knowing” 11:
I was often found verbally-constraining functional movement
solutions by continually attending to an athlete’s performance
through the lens of error correction: comparing what I was
looking at against a preconceived ‘model of excellence’. For me,
the seismic shift in what it meant to pay attention as a coach
came when athletes started reporting feelings of roboticisation;
lacking presence during competition by focusing too intently on
trying to enact my pre-race plan. Perhaps I was confusing the
map with the territory? With great discomfort, I indeed
discovered that it was my knowledge about how I thought
things should be done in Canoe Slalom that was constraining
the growth of the athlete’s knowledge of it
.

Knowledge about what performance should look like can

dramatically shape what we see and consequently our
interactions with our athletes. As Craig Morris puts it
11: “there
appears to be a wisdom in not-knowing, a wisdom that keeps
one open – responsive – to what the world has to share, such
that they can get to know it a little better than before”.
The idea that we need to acquire a lot of knowledge about
something before we can do it goes part and parcel with the
dominant information processing view of skill – I can’t perceive
the ambiguous world around me without prior knowledge and
a mental representation with which to interpret it. It is trying to
impose some certainty over a complex, uncertain, and highly
unpredictable world!
We will spend much more time considering how a coach can
become adaptable and better connect with the practice
environment in future chapters, but I next want to consider
what role skill acquisition research and theory might play in this
journey.

2
THE VALUE (AND LIMITATIONS) OF
USING SKILL ACQUISITION THEORY
& RESEARCH IN COACHING
T hat works in theory but not in practice. Most sports science
research has little actual relevance to coaching. These are two
sentiments that I very commonly encounter in talking with
coaches. And, you may be surprised, but I am not here to
completely convince you otherwise! Many of the issues coaches
have with skill acquisition research and theories are completely
valid. But that doesn’t mean these things don’t have any value.
Before we get into the nitty-gritty of being an “ecological coach”
I think we need to make peace with the sports scientists (like
me). Part of coach development involves learning how to use
sports science research effectively. When and how is it useful?
When should we be cautious about over-generalizing from the
lab to the field? How are we going to use data to inform our
practice?
The Link Between the Lab & the Field: How Skill Acquisition
Science is NOT being used in most Practices
As a person who has published his fair share, I can tell you
that one of the most depressing things you come to realize is
how little coaches read journal articles. In the study of
preferred ways of acquiring coaching knowledge mentioned in
the last chapter, less than 2% of coaches identified journal
articles as their preferred method. But is that surprising? We
will look at the multitude of reasons why this is the case in the
next section but first I want to pick at this wound a bit more.
While it might be true that coaches don’t access the information
directly by reading journal articles, they could still get the
information other ways and adopt practice interventions that are
consistent with the main conclusions of skill acquisition research.
Well, sadly, this does not seem to be the case either! Not only
do coaches not seem to be incorporating the key principles of
instruction and practice design from research, but they often go
completely against them. This can be seen in several (you
guessed it) published research studies that have examined and
analyzed what coaches do during practice. Yes, I do see the
irony here.
In a paper published in 20101
, Paul Ford and colleagues
examined the practice activities and instructional behaviors of 25
youth soccer coaches over 70 different practice sessions. They
found several ways in which practice was highly inconsistent
with contemporary skill acquisition research. First, it was found
that on average, 65% of practice time was spent in “training
form” activities versus 35% in “playing form” activities. Training
form activities were defined as those focused on improving
physical fitness and technical skills through unopposed and
decomposed drills (e.g., dribbling around cones, running lines)
while playing form activities were things like small-sided games
and practicing different phases of play that emphasize
decision-making and tactical parts of the game. Two-thirds of
practice on training form! This is particularly lamentable given
that, if a coach really wants to emphasize it, many of the
things that fall within this category can be practiced at home
and do not require a full team to complete. As Ford et al
conclude: “A conservative view would suggest that more practice
time should be spent in playing form than training form
activities, whereas a more radical view would be that only
playing form activities should be employed.”
The second main finding was that most coaches sure do like to
talk a lot! In the study there were high levels of instruction,
feedback, and management, irrespective of the activity in which
players engaged. In other words, there were limited
opportunities for players to self-organize their movement and
develop autonomy (a key component in motivation that will be
discussed more in Chapter 13).
Finally, there were very few observed differences in the
structure of practice across age levels and groups. In other
words, there was little evidence that coaches utilized any
progressions in training such as the evidence-supported
practices of scaling the equipment and/or playing area2 for
younger athletes, adjusting the level of variability in practice to
align with the inherent variability of the performer3
, and
periodizing practice based on the stage of coordination and
control4
. To sum up, the authors conclude: “A significant lag
has been identified between the generation of cutting-edge
research evidence and its application in coaching and coach
education”.
At a more fine-grained level, research examining the content of
verbal instruction used in coaching has found that it is highly
inconsistent with the large body of research on the focus of
attention. As I proclaimed in my last book: “There is a relative
tidal wave of research showing that the use of internal cues is
not the most effective way to coach”5
. Despite this large body
of evidence6
, research examining the use of cues and
instructions in coaching has consistently found that the vast
majority are internally focused: 85% in track and field7
, 71%
in boxing8 and 70% in baseball9
. Not only not heeding the
research evidence – doing exactly the opposite of what it says!
So, let’s consider a few reasons why this might be the case.
Problems with Access
I think we have all experienced a situation where we find an
article online that sounds interesting, we click on the link to
find out we can read it by paying the low, low price of only
$50. Heck, this sometimes happens to me when I am trying to
find a copy of one of the articles I wrote! Frankly put, the old
model of scientific publishing is a scam. Before you pay for
such a ridiculous price try contacting the author directly – they
are allowed to send you a completely free preprint copy and
most are more than happy to do it.
But while I agree articles being hidden behind paywalls have
been a problem in the past it is becoming less and less of one.
With the proliferation of open access papers, the alternative
ways to access the material (e.g., Twitter/X threads, YouTube
videos, Podcasts!), I think there is a lot of great information out
there for free now. I now have over 500 podcast episodes
waiting there for you. So, I don’t think this is the primary
problem anymore.
Lack of Generalizability from the Lab to the Field
A much more serious issue in applying published research in
coaching is that a lot of tasks used in papers are completely
unlike real sporting tasks. For example, many motor learning
studies involve participants learning a sequence of keystrokes on
a keyboard, moving a robot arm through a force field or
reaching to pick up a stationary object off a table. The problem
here is that we are trying to generalize results from “simple”
tasks to “complex” ones.
In their excellent 2002 review paper, Wulf and Shea10 define
simple tasks as ones that are discrete (that is, not affected by
the environment), are not heavily reliant on feedback, have low
degrees of freedom (so involving fewer body parts and/or
fewer movement components), and have lower time pressure
demands. Conversely, complex skills tend to be more
continuous, have more degrees of freedom that require
coordination, are heavily affected by an ever-changing
environment, and have high temporal demands. Let’s compare
key pressing and baseball batting. The former involves one
body part (the hand), a constant external environment (the
keys are always in the same place, or the force is constant on
any given trial), and not very strict time demands (typically in
these studies any response with 1.5 seconds is accepted).
Baseball batting, on the other hand, involves coordinating the
movement of the entire body, reacting to a highly variable
environment (different pitch speeds, types. and locations) and
has severe temporal demands (the performer has less than 0.5
sec to act and the required margin of error for success is
±10ms11). These are two very different animals!
This difference can also be seen in the results. For example,
let’s consider the classic blocked vs random training effect
found in motor learning research. Early research on this topic,
largely based on the use of simple tasks, has produced the
now classic pattern illustrated in Figure 2.1
Chart, line chart Description automatically generatedFigure 2.1 –
Typical blocked vs random practice results
As shown in this figure, random practice typically leads to a
slower rate of acquisition of the skill but better learning in the
long term, in retention and transfer tests. This has been
explained by the Contextual Interference hypothesis – the
cognitive interference created by trying to develop motor
programs for multiple skills at the same time results in more
elaborate and effective motor programs in the long run.
However, a recent meta-analysis of studies that have examined
complex sports skills12, has shown that the pattern illustrated
in Figure 2.1 does not occur reliably – there are no significant
differences between blocked vs random practice. Furthermore,
the effect is not moderated by other variables like age and skill
level as has been proposed in other studies13. For example,
there was no evidence to support the commonly held belief
that younger or new learners need to progress through a stage
of blocked, isolated practice (i.e., focused on the “fundamentals”)
before variability can be introduced. As we will look at in more
detail in Chapters 6 & 7, for complex skills the simple blocked
vs random dichotomy does not typically apply. Instead, this
research suggests that more subtlety is required in finding the
optimal level of practice variability for an individual athlete.
Another “classic” motor learning finding that does not seem to
hold up when we move from simple lab tasks to more complex
sport-like tasks is the frequency of feedback effect. When we
perform a task like hitting a baseball there are two basic types
of feedback we can receive concerning the outcome of our
action. The first, called intrinsic feedback, is information
obtained from our five senses. When we make solid contact
with a baseball, we see it go flying out onto the field, we hear
a loud “crack of the bat” sound, and we feel a slight vibration
in our hands. Conversely, when we don’t hit it solidly, we see
the ball land weakly on the ground, we hear a dull “thud” type
sound and often feel a very strong (sometimes painful)
vibration in our hands. The other type of feedback we can
receive about the performance outcome is called extrinsic
feedback. This includes any feedback that we cannot get
ourselves, from our senses, including verbal feedback from a
coach (“great hit” or “you got under that one”) or the output
of the ever-increasing array of technologies like launch monitors,
radar guns, markerless motion capture systems, etc.
Early research on performance feedback revealed a somewhat
surprising and counterintuitive finding: too much extrinsic
feedback is a bad thing! In a typical experiment14, participants
are given the simple task of producing a target level of force
(say, 20 N) by pushing down on a sensor with their hands.
They are given extrinsic feedback about their performance via a
digital display of the force output. In such experiments, it is
common to find that those who receive this feedback after
every trial show poorer acquisition (in retention tests) as
compared to those who receive it in only 50% of the trials. As
first proposed by Salmoni et al. (1984)15 in what is now called
the “guidance hypothesis”, this effect is thought to occur
because too much extrinsic feedback impairs our ability to
become attuned to intrinsic feedback (e.g., the pressure we feel
in our fingertips) and to learn to do our own adjustments, an
essential part of motor learning. We essentially become
overdependent on a coach or some technology for adjusting
our movements.
But again, this simple story does not seem to hold up when
we look at more complex, sports-like tasks. For example, Wulf
and colleagues16 examined the effect of feedback frequency for
the task of learning slalom skiing in a ski simulator. Specifically,
participants were given extrinsic feedback about the time of
force onset (a key performance indicator for slalom skiing) via
a digital display either on every trial (100%) or on 50% of the
trails. There was also a control group that received no
feedback. It was found that the group that had received the
most feedback (100%) during practice showed superior learning
of the skill on a retention test (where the feedback was not
present), whereas the 50% feedback group showed no such
performance gains, and the control group even demonstrated a
performance decrement. Thus, unlike the results for simple
tasks, there was no indication that the 100% extrinsic feedback
participants developed a dependency on the extrinsic feedback
that reduced the learning effectiveness of this condition. Rather,
the opposite seemed to be true. Since this study, there have
been several others that have also found this “reversal of the
guidance hypothesis” for more complex tasks (reviewed by
McKay et al., 2022
17
).
To understand why this occurs, let me tell you a little story
about baseball batting cages. When I started working with MLB
teams, I noticed that many were using a technology called
HitTraxTM in their batting cages. This system is comprised of
a sensor that picks up the flight of the ball leaving the player’s
bat and a video-game-like display that shows a virtual ball flying
out into a simulated field. So, it essentially shows where the ball
would have gone (how far, what direction, how fast) if the
batter were on a real field and not in the batting cage. This
system clearly provides extrinsic performance feedback, thus, my
initial reaction as a “skill acquisition expert” (based on what I
knew about the guidance hypothesis) was that we didn’t want
to use this on every swing. My thought was that it would
impair the batter’s ability to tune into intrinsic feedback and
hurt their ability to make their own adjustments at the plate
(e.g., change how they swing on the next pitch after producing
a weak grounder on the last pitch). But after looking at the
data for a while, the results were similar to those of Wulf and
colleagues – batters showed greater improvements when
HitTraxTM was turned on for every swing.
Looking more closely, I came up with a simple hypothesis for
this effect: the intrinsic visual feedback batters receive about the
outcome of their swing in a batting cage is very poor and
often misleading because it is not representative of what occurs
in the game. This is particularly the case when we consider the
vertical launch angle of the ball coming off the bat, illustrated in
Figure 2.2. Because of the way a batting cage is constructed,
balls with a relatively low launch angle (say 5-10 degrees) will
travel the furthest into the cage and thus seem to be,
intrinsically, the best hits. Indeed, a common thing I have
observed over the years is that without any extrinsic feedback
like HitTrax, many batters will start deliberately hitting a lot of
balls at low launch angle that go all the way to the back of the
cage because they look better. Conversely, a ball hit at the
same speed but with a steeper launch angle (say 25 degrees)
will travel a few feet, hit the top of the batting cage, and then
fall to the ground 20 feet in front of the batter. Not very
impressive!
But what happens in the game is exactly the opposite. Figure
2.2 (bottom panel) shows the relationship between launch angle
and a batting statistic called wOBA (which estimates the
projected number of runs that would be scored by the hit)
using MLB game data. For a ball hit at 95 mph, the wOBA
for a launch angle of 25 degrees is about 250 points higher
than for a launch angle of 10 degrees. This is because a ball
that is hit hard at a launch angle of 25-30 degrees is likely to
be a homerun while the same hard hit at 5-10 degrees has a
good chance of being caught by a fielder. Thus, in this case,
giving a high frequency of extrinsic feedback seems to be a
good thing.
My hypothesis is somewhat like one of the explanations
proposed by Wulf and colleagues in their ski simulator study.
Namely, they argued that the reversal of the feedback
frequency effect may have occurred for their task because the
intrinsic feedback received for a complex motor task like
learning to slalom ski is too complex (there is too much
information to pick up) for it to be effective for new learners
who have yet to educate their attention to this information.
When we ski, intrinsic feedback is not simply the amount of
pressure on our fingertips (like in a simple force production
task) but instead, it involves picking up the precise timing of
multiple forces throughout the entire body. In previous studies,
my colleagues and I have shown that because novice learners
are often more internally focused (that is, their attention is
directed to the movement of their body, rather than the effect
their movements are having on their environment), they are
often relatively insensitive to the intrinsic feedback about
performance outcome received for complex tasks like hitting a
ball19. Thus, I think we need to consider the complexity and
representativeness of the intrinsic feedback in the training
environment and the stage of the learning of the individual
athlete before deciding whether it should be removed from
some trials.
A baseball field with a graph Description automatically generated
A baseball field with a graph Description automatically generated
Figure 2.2 – Unrepresentative intrinsic visual feedback from
batting cages. The bottom panel reproduced using data Clemens
(2020)18.
For me, these effects illustrate a key point that we will return
to throughout this book. Rather than there being some simple,
universal laws about skill acquisition methodology, coaches need
to consider the entire system and context. Because we are
working with athletes that are complex systems, it is rarely the
case that statements like “we should never use internal focus of
attention instructions” or “too much extrinsic feedback is bad”
will always universally apply. We will consider what this means
for applying theory to coaching in a little bit but let’s first
address a few more issues with applying research.
Research is about Groups, I am Coaching Individuals
Another related problem with a lot of sports science research,
which I lamented in my first book in a section I called “What
Gets Lost in the Average”, is that most published research is
group-based while we coach individuals. This seems to be an
example of the conflict between the “art” and “science” of
coaching.
In graduate school, my supervisors drilled into me the principles
of the scientific method – what is required to make sound
conclusions from a research study. For skill acquisition research,
this included things like (i) giving the same instructions to all
the participants within the same treatment group in your study,
(ii) making sure all participants and groups had the same
amount of practice time, (iii) trying to make sure participants in
the different groups were matched in terms of their training
history, (iv) using statistics based on the group means with a
large enough sample size so that the data is not skewed by
one individual performer, and (v) remaining objective as
researcher - think of the typical scientist in a white lab coat.
Now, think about coaching an athlete. As most will
acknowledge, there seems to be an art to doing this well. We
do NOT want to use the same methods, instructions, feedback,
etc. for every athlete. We want to be adaptable and change
these things based on what we are seeing in front of us. We
also need to consider where an athlete comes from – their
previous experience and intrinsic dynamics. We don’t want to
use methods that will result in the best results for the group,
on average. Instead, we want to get the best out of every
athlete or, if anything, get the most out of the outliers in the
group to produce a small number of elite athletes. Finally, as
we will discuss in more detail in the next chapter, we need to
recognize that, as a coach, we are part of the system never
completely objective and impartial.
This has led to, what I believe, is a beneficial change in our
skill acquisition research methods from traditional group-based
experiments to what has been termed “contextualized skill
acquisition research”20-22. This initiative, led by one of my
co-authors Luis Uehara involves direct and active involvement
by the researcher rather than some effort to maintain
objectivity. Furthermore, analyses are focused on changes within
individuals over many executions rather than on averages from
groups of athletes that participate in a very small amount of
training. A multi-method approach is used which attempts to
capture the entire system for an athlete (e.g., including the
social constraints imposed by their family) rather than just one
part of it. I will discuss some of this type of research in more
detail in Chapter 12.
Let’s next consider the value of a good theory.
The Role of Skill Acquisition Theory in Coaching

While throughout most of this chapter, I have been giving

examples of why we can’t coach all our athletes in the same
way and why there are no simple, universal laws for the
methods that we use as coaches, for me, this does NOT mean
that are no underlying skill acquisition principles that guide our
coaching. One of the things I was asked to do in my role with
the Boston Red Sox was to create a WWH poster (that now
hangs in our player development facility) where these letters
stand for; “what we believe”, “why we believe it”, and “how we
coach it”.

What I came up with is essentially the same as the six key

principles of the ecological approach I described in the preface.
For example, I believe that allowing an athlete to self-organize is
a more effective way to learn and develop a skill than
prescribing a movement solution for them. I believe that for an
athlete to perform effectively under ever-changing constraints
(which are present in all sports to some degree) we need to
promote good movement variability in practice, not suppress
variability through lots of strict repetition. I believe that any
“fundamentals” of a sports skill will emerge if we use
appropriate constraints and do not need to be taught first in
an isolated, decomposed manner. And so on. These principles
are my north star. Every time I am asked to evaluate a new
training technology or a new practice drill, I ask myself does it
align with the ecological principles of skill acquisition (AKA what
I believe). Every time I am asked about adjustments we can
make to training for a specific athlete in a specific context, I
ask myself: does this align with what I believe about skill? How
I use different coaching methods (from instruction and imagery
to constraints to video feedback) – same question! My theory
(ecological dynamics) is the framework on which all my skill
acquisition recommendations hang. That is the true value of
having a skill acquisition theory as a coach – it gives you a
principled, and evidence-based reason for the decisions you
make. When faced with that critical coaching question “What do
I next?” you have a guide to help you come up with an
answer amongst the essentially infinite number of possible
answers. This was expressed nicely by Ji Yi Chow and
colleagues23
:
To make sense of practical activities, implement effective
organization and make efficient use of time, coaches, teachers
and sport practitioners need to rely on a model of the learner
and the learning process.
And, in an article I will discuss in detail in Chapter 13 by
Stone et al.:
Theory has played a pivotal role in shaping “why I coach the
way I coach” whilst also enhancing confidence in my approach.
Furthermore, it has invited me to co-create this “why” with
athletes, sharing our perceptions of what the sport is and
reflecting upon how congruent our practice methodologies are
in preparing for skilled performance in competition. In essence
it is an anchor of reference for practitioners in the loop of
“observe to design, design to observe”.
But isn’t this just “It Depends” coaching?
At this point, I want to address a common misconception: the
fact that there are no universal laws of skill acquisition (e.g.,
external cues, a low frequency of extrinsic feedback and
random practice are always better) DOES NOT mean that I
am arguing for “it depends” coaching or at least one popular
conception of that phrase24.
If by “it depends” we mean that we need to adapt our
methods (instructions, feedback, practice sequencing, number of
repetitions, use of demonstration, difficulty level, etc.) based on
the individual athlete, the particular sport we are coaching, and
the specific constraints of the practice environment (e.g., how
many athletes we are coaching at once) then I would argue
that ecological dynamics and associated coaching methods like
the CLA and nonlinear pedagogy are the very definition of this
phrase. And I agree wholeheartedly with this meaning. Yes, we
certainly do need to be adaptable like this and adjust which
coaching methods we use and how we use them based on the
specific context. This is the knowledge “of” coaching we
explored in the last chapter.
In my batting cage example, accepting the fact that we needed
to provide extrinsic feedback at a high frequency did not
require me to change my principles of motor learning (e.g., that
athletes learn best through self-organization). Similarly, if I
occasionally need to provide internally focused instructions (e.g.,
“Why don’t you try bending your back leg a bit more”) as a
guide I am not changing my fundamental beliefs about the best
way to learn a skill. In both cases, I am changing the
constraints based on the specific context and situation and
allowing the athlete to self-organize.
But if by “it depends’ we mean that we need to change our
theory of skill acquisition (e.g., whether we try to promote
self-organization vs prescribe an ideal solution, whether we
encourage variable movement solutions vs. promote low
variability through strict repetition, whether we need to learn
the motor program for the “fundamentals” first so we can build
on it, etc.) depending on the situation24
, I do NOT agree and
that is not what I am arguing for here at all. When you allow
yourself to shift back and forth between incompatible motor
learning theories25, you lose your north star, and your
guiding framework crumbles.
For example, research looking at coaches who attempt to make
a shift to being an “ecological coach” (discussed in detail in
Chapter 14) has shown that one of the biggest challenges is
sticking to using the approach despite lots of pressure to revert
to the traditional coaching methods that were passed down to
them
26
. As I discussed in my first book, we are going to be
honest with ourselves about what learning really looks like – it
takes time, it does not follow a simple straight-line path with
outcomes that are easily predictable from the inputs we add as
coaches. They are not simple metrics we can use to show
improvement. We need to give self-organization the time and
space it needs, guiding it when we can. If we allow ourselves
the possibility that this individual athlete would learn better if we
just told them what to do (i.e., we switch theories) then we will
not allow this to happen. For me, allowing yourself to move the
goalposts by switching between different theories leads to weak
and often counterproductive practice interventions.
As an example, let’s look at some research that has studied
how giving a performer an explicit solution affects their implicit
learning. Up front, I will acknowledge this work again involves a
simple reaching task so its implications for complex sports tasks
are limited, but I think there is still a useful message here. In
these studies, participants are asked to move a cursor on a
screen toward a target. The cursor is controlled by the
movement of the person’s hand via a joystick or motion
tracker. To investigate adaptation and learning, a rotation is
added to the mapping between the movement of the joystick
and the movement of the cursor. So, for example, if the
person moved the joystick straight up, the cursor might go 45
deg to the right. Previous research has shown that participants
can learn to compensate for this rotation and hit the target in
two different ways: explicit and implicit learning. The explicit
component of learning can be assessed by asking participants
to indicate their aim point before they start moving. So, for
example, if you asked me to move the cursor to the top of
the screen and I told you I planned to aim 45 deg to the left
it would indicate that I had figured out the manipulation and
adopted an explicit strategy to counteract it.
However, this research has also shown that participants also
learn to adapt implicitly (that is, without awareness or any
explicit strategy they can verbalize). This can be seen by the
fact that even though participants indicate an aiming point they
often do not actually move there. The difference between the
indicated aiming point and where they move has been used as
a measure of implicit learning. The type of learning can also be
assessed by looking at aftereffects. If participants are using only
explicit learning for this task, if you take the rotation away and
tell them that you have, there should be no aftereffect – they
will hit the target because they can just consciously stop
changing their aiming strategy. For implicit learning, however,
there should be a large aftereffect as an error in the opposite
direction to the original rotation. To alter implicit learning, the
participant is going to have to recalibrate their
information-movement control law by doing some practice trials
without the rotation present.
Another way to think of this is to consider playing golf on a
windy day. If there was a strong crosswind from left to right,
there would likely be two processes going on to allow you to
keep hitting the ball straight. An explicit strategy of “I am going
to start the ball out a bit to the left so the wind carries it back
to the middle” (possibly given to you by your coach) and an
implicit learning process whereby the calibration between
information and movement is changed to alter the path of your
swing and the flight path of the ball – for example making the
ball work from right to left into the wind. Do these two
processes complement each other? How do they interact?
This question was addressed in a 2011 study by Benson and
colleagues27. Fifty-four participants were randomly assigned to
one of two groups: an implicit learning and an explicitly
instructed group. The task again involved moving a joystick to
move a cursor toward a target on a screen. Participants in the
Explicit group were shown a diagram of a clock face and told
that their movement would be rotated by 30 deg, were then
shown that if they moved towards 12 o’clock the cursor would
go to 1 o’clock and were told they could compensate for this
by aiming for the 11 on the clock. So, in other words, their
“coach” gave them an explicit movement solution. Within the
experiment, there were also randomly placed catch trials in
which the rotation was turned off. Participants in the explicit
group were told to not use the aiming correction strategy for
these trials. Participants in the Implicit group were given no
instructions at all before the rotation trials. They were told
there would be catch trials and for these, they should attempt
to hit the target in the same way as on the other trials. There
was a total of 300 practice trials with rotation for both groups
after which there were a set of trials with no rotation to
measure aftereffects.
What was found? Initially, the explicit strategy worked effectively
at reducing errors to almost zero. However, it came at a cost
– participants in the explicit group were slower to react and
had lower precision as compared to those in the implicit group.
The authors propose that the longer reaction times reflect the
increased cognitive load required to use a conscious, explicit
strategy. Participants in the explicit group also showed
significantly smaller aftereffects in the catch and post-training
trials suggesting that less recalibration and implicit learning was
occurring. There was also evidence that many participants in
the explicit group eventually abandoned using the explicit
strategy in favor of using the more effective and less
demanding implicit recalibration. So, in sum, this study shows
that while adopting an explicit strategy for correcting movement
errors can be advantageous in getting rid of the errors more
quickly, it comes at the cost of being slower, more cognitively
demanding, and less precise. More importantly, it can, in some
cases, attenuate recalibration and implicit learning – a more
effective strategy in the long run. Evidence that giving a
performer an explicit, prescriptive movement solution impairs
implicit learning, and recalibration has also been shown in other
studies28,29. This line of research is nicely summed up by
one of the titles of these papers: “Implicit adaptation
compensates for erratic explicit strategy in human motor
learning”29.
For me, these studies exemplify the problem with “it depends”
(on a theoretical level) coaching. If you fundamentally believe
that we control our movements using a top-down explicit
movement solution that is passed down from a coach (i.e., we
“boss” our body around as I described in How We Learn to
Move) then that is not compatible at all with the belief that
learning occurs through a process of implicit, self-organization
and recalibration. And switching between these two approaches
is going to be highly counterproductive. It is valuable for a
coach to have one clear theoretical approach and decide which
of these approaches to skill acquisition fits better with your
personal view and experiential knowledge. This will help guide
you more effectively in your practice design.
A similar conflict arises when we consider how the information
processing, traditional approach, and ecological approach to skill
acquisition address the so-called “fundamentals” of a sports skill
30. In the traditional approach, it has long been believed that
these need to be coached first before playing the game. This is
thought to involve the development of a motor program for
performing a decontextualized, low-variability/low-complexity
version of the skill (through lots of strict repetition) before the
athlete is faced with the higher complexity of the game. For
example, we learn how to “dribble” by going through a course
of cones before we play soccer. We learn to hit off a tee
before we play baseball. In golf, we first to learn “swing” by
practicing on flat ground before we play on a course with
different lies and surfaces. The assumption is that we need the
motor program for these “fundamentals” first then we will learn
when to use it and how to adjust it later by training in more
realistic game-like situations. This is what I called training for
“adjustability” in my first book.

In the ecological approach, it is proposed that if there are any

“fundamentals” of a skill they will emerge while the athlete is
engaged in representative practice scaled appropriately for their
age and skill. The focus here is on adaptability – the ability to
find a movement solution for different combinations of
constraints, right from the beginning, rather than learning to
adjust a motor program for an unrepresentative version of the
skill learned in the early part of training.
This difference can be seen in a recent study by Deuker and
colleagues31
. In this study, young soccer players were trained
following a traditional, decontextualized approach (e.g. learning to
dribble around cones, learn to pass through cones, etc.).
Players were also given prescriptive instruction about the correct
technique and were given corrective feedback when they
deviated from this. A second group followed an ecological
approach practicing in context-rich small-sided games. Task
constraint manipulations included variations in instruction (e.g.,
“maintain possession” or “play over the defense”), rules (points
for scoring or keeping the ball), the number of players, and
the size of the playing area.
Pre- and post-training tests were deliberately slanted toward the
traditional approach in that they involved simple, out-of-context
tests of dribbling, passing, and shooting ability on a simple
course. What was found? Both groups showed similar (and not
statistically different) improvements from pre-post training on all
the tests. So, in other words, the ecological group learned these
basic skills without having them pulled out of the game and
trained separately. Furthermore, when the authors looked at the
practice data in more detail, they found that the traditional
group had a total of 777 touches of the ball while generating
these performance gains while the ecological group had only
354 touches. So, playing in the small-sided game was a more
efficient (and likely more enjoyable!) way to learn the
“fundamentals” as compared to decontextualized, isolated
practice. It took fewer repetitions (touches) to produce the
same overall results. As one final example, I refer readers to
my blog post on the two different ways the connection ball
constraint can be used in coaching baseball32
.
Again, I cannot see how you can blend or switch between the
two theories effectively33
. They are at complete odds as to the
most effective way to develop skill. To quote an excellent
commentary paper by Kearny et al34.
Abstracting parts of different pedagogies can result in
fragmented and mutated ‘folk pedagogies’: incoherent
assumptions about learning and coaching which can be difficult
for coach development to overcome.
And from a blog post by Hydes et al at the Constraints
Collective35: “Using a ‘pick & mix’ approach to coaching
methodology is not coherent and logical”.
Contemporary Internal, Representation-Based Theories of Skill
Acquisition
A criticism of ecological dynamics that has been raised since I
wrote my first book is that many of the problems with the
traditional approach to skill acquisition are only issues with older
theories in this area37
. In particular, it is Schmidt’s schema
theory in which movements are controlled via internal
representations (i.e., motor programs/schema) in the brain.
More contemporary information-processing theories of skilled
motor action address issues of one ideal technique, the
nonlinearity of learning, and embodied perfection (all issues that
I raised in my first book). However, although these theories are
major improvements, they seem to lack specificity in terms of
how they should be applied to sport and contain unnecessary
components when we can get the job done with direct
perception! But let’s have a look at these newer internal
model-based accounts of skill. Note, in reviewing these theories,
I am going to focus on specific attempts to apply the theories
to sports coaching rather than give a comprehensive discussion
of the pros and cons of the theory.
Predictive Processing
Predictive processing is predictive, internal-model based theory
of perceiving and acting, first proposed by Andy Clark37. The
key difference with this theory is that instead of trying to
predict the events in the outside environment (e.g., where a ball
with land and a when), it is proposed that our brain is trying
to predict the sensory information or feedback that will be
received as the result of our motor actions. The predictions are
generated by internal models that use our past experiences.
These models get updated when there is some discrepancy
between the predicted feedback and the actual feedback we
receive. This removes many of the problems associated with
trying to predict a future event I discussed in Chapter 5 of my
second book. To quote from Lohse & Hodges38:
When we act there is thought to be a corollary of this action
plan (termed efference copy), which enables “forward model”
predictions about the action’s expected sensory consequences
operating largely outside of awareness. Expectations (and
forward models) improve with practice, and we become better
at generating accurate predictions about how a movement will
feel, look or sound.
This idea of predicting the sensory feedback one should receive
was actually first proposed in Schmidt’s schema theory, but
predictive processing argues that it is the main signal we use
for the control of action. O’Brien et al. (2023)39 have
proposed that predictive processing can be applied to basketball
coaching. The main idea put forth in their paper is that a
coach can support skill acquisition by purposely creating
situations where there will be a discrepancy between the
predicted and actual sensory feedback. In other words, we are
trying to create situations that surprise the athlete because there
are unexpected variations from the already-established internal
patterns that athletes have learned over time. To quote the
authors:
Rather than being told where to run and when and how to
pass, a coach can add the element of “surprise” or novelty to
the session such as orchestrating opposition players to randomly
lose possession of the ball in mid-play, suddenly state that there
is only 2 min left on the clock to score, or, without warning,
arbitrarily switch from a block of 4v4 small-sided play to a
defensive overload scenario of 4v5. Each of these situations
provides players with something novel, allowing them the
opportunity to update and stabilize their movement models.
On the surface, this sounds exactly like the type of
manipulations we used in the CLA. However, the authors argue
it goes beyond that by helping the coach decide which
constraints need to be manipulated in practice and when.
Specifically, they propose that to effectively manipulate
constraints a coach needs to “look through the eyes” of their
athletes. That is they need to: consider how their athlete's
brains generate objects of perception in a top-down manner,
not by accumulating and combining input signals, but rather, by
issuing predictions “based on hierarchical generative models that
rely on prior probabilities and likelihood estimates”.
Sounds complicated! I don’t agree with this view at all. In the
ecological approach, we do already look at things from our
athlete's perspective. From their behavior, and their movements
made under different constraints, we try to understand the
information they are using to control their behavior, their
control laws (how they use this information), and their
calibration (how well their actions take into account their
individual action capabilities). From this, we try to design
practices to Educate Attention, Educate Intention, and influence
perceptual-motor Calibration – the three components of Direct
Learning40. The strength of this approach is that it is
grounded in observable things – the information from the
environment and our athlete’s actions. I am not required to
make guesses about unknowable internal mental states such as
trying to know whether the athlete was surprised or not or
what they predicted was what happened. As will be discussed
in detail in Chapter 11, the CLA gives a much clearer and
simpler account of how to design and adapt constraints.

Active Inference
An extension of predictive processing that has more to say
about the control of action is called Active Inference41. In this
theory, the prediction of the sensory feedback is modeled as a
Bayesian process in which we start with a prior prediction or
estimate. For example, using baseball bating as an example, the
prior prediction about the upcoming pitch could be based on
information from the count (# balls and strikes), the previous
history of pitches the batter has faced, and the pitcher’s
tendencies (e.g. how often they throw a fastball). This prior is
then updated using information that becomes available as the
event unfolds. So, first, using the kinematics of the pitcher’s
windup (commonly called “advance cues”) and then using visual
information from the actual ball flight (e.g. tau). So, it is a
model that combines offline and online information. But critically
NOT direct and indirect perception42
. Finally, the prior for the
next event is updated based on the sensory feedback that was
actually received, which in Bayes theory is called the posterior.
In active inference, our internal Bayes models can be used to
predict new sensory data (e.g. the feedback expected when
executing an action), or can work backward to infer the hidden
states that likely caused some observation (known as model
inversion). Any prediction error in the sensory feedback serves
as a signal (called “free energy” from the work of Karl Friston
41) that we use to control our movements.
So, imagine, for example, we are reaching to pick up an object
and we predict the visual feedback that will be received is that
our hand is aligned with the object's position. If the visual
feedback we actually receive is that our hand is lower than the
object – that is a prediction error (or free energy) that
compels us to act to correct it. Thus, active inference allows for
continuous online control which can deal with things like
changes in an object’s trajectory (e.g. due to a sudden gust of
wind). This is a major upgrade to traditional ideas of predictive
control where a motor program runs off ballistically and cannot
be altered after it starts. Free energy is a signal that is
continuously available for the system, so it allows for continuous
online adjustments to minimize the prediction error. Stated
another way - active inference, is the extension of predictive
processing to the use of actions to minimize future surprises
(i.e., prediction errors).
To my knowledge, there has been only one published attempt
to apply this to sports: Harris et al.’s 43 active inference model
of anticipation. The main goal of this effort was to see if an
active inference model can explain the key findings from the
anticipation literature – for example, that when anticipating,
athletes combine multiple sources of information in a manner
based on their reliability. So, for example, when anticipating the
direction of a tennis serve, a player will use advanced kinematic
cues more heavily early in the event and weight information
from the ball flight more heavily later in the event.
Harris and colleagues model anticipation of pitch type in
baseball, specifically whether the pitcher will throw a fastball or
a changeup (i.e. a slow ball). To quote the authors:
Within the model, the batter has prior beliefs about hidden
states such as contextual information (which pitch is more
likely?). These then take the form of categorical probability
distributions that are updated through integrating beliefs about
past states (which pitch have I just faced?), current sensory
information (what pitch do kinematic cues indicate?) and beliefs
about future states (which pitch is likely to come next?). The
selection of actions by the observers in the model are driven
by the need to minimize future or expected free energy with
their action choices.
Here is where my problem with this approach. Like most
internal models it does not have a lot specific to say about the
control of action. This is a predictive model that can be used
to explain anticipation experiments involving people making
passive predictions e.g., Saying the word “fastball” or
“changeup”. Not actually swinging a bat. So, its effectiveness in
explaining this body of research is not all that surprising. But it
has little to say about how you swing the bat to hit a
changeup of a fastball. The output of the model is a binary,
categorical prediction about pitch type. What do we do with
that prediction? Do we have pre-stored and fastball and
changeup swings that we just initiate? What about hitting a low
fastball vs a high fastball?
The big advantage of the ecological approach is that it solves
the problem of perception and action at the same time! For
example, in my prospective control model of baseball batting44
,
the perceptual information is used to control the action by
establishing a coupling relationship. For example, see the tau
gap model of hitting I outline in my second book. The batter is
continuously coupling their bat movement to the information
from the environment (with this relationship also being
influenced by offline information via the specifics of the control
law). I don’t see how this active inference model gives specific
information about how fast to move the bat, when to start the
swing, what direction/angle to move it, etc. All these problems
are addressed in prospective control.
This becomes even more of an issue when we try to apply
these ideas in practice. Let me try to use a specific baseball
batting example to illustrate. One problem I have worked on
with a lot of batters is trying to improve thier ability to hit
4-seam fastballs with “good ride”. What the heck does that
mean?! Well, some pitchers in baseball (Gerrit Cole from the
NY Yankees is a good example) can put an abnormally high
amount of backspin on the ball when they throw a fastball.
While the MLB average for the spin on that pitch is about
2200 rpm, Cole throws his 4-seam fastball with nearly 2500
rpm of spin. Why does that matter? A pitch with more back
spin is less affected by gravity – spin makes a projectile resist
the force of gravity and travel straighter – this is why we try
to throw an American football in a spiral (that is, put spin on
it). Back to baseball – because Gerrit Cole’s 4-seam fastball has
more spin than average, it will fall less due to gravity and will
cross the plate higher (by a distance equal to roughly 2-3
baseballs) than a pitch at an equivalent speed thrown by a
pitcher that can only put an average spin on the ball. This
causes batters to swing under the pitch and either pop it up in
the air or miss it completely.
But, aha, I am sure proponents of active inference are yelling
right now – yes, this can be perfectly explained by the model.
The batter’s prior (which is presumably based on the ball flight
for an average pitcher) is wrong and needs to be adjusted.
Yes, that does seem like a good basic description but what, as
a coach, do I do with that with the player in front of me?
Just tell them: update your prior? Wouldn’t simply giving the
batter knowledge about this difference in ball flight (e.g. by
watching a video of Cole pitch or telling them it won’t drop as
much on the way to the plate) be expected to change the
prior and solve the issue – I can tell you from personal
experience that it does not!
Approaching this problem instead from an ecological dynamics
perspective, I first try to identify the information-control law the
batter is using. That is, look at the relationship between the
aspect of their movement they are controlling (in this case,
what is called the vertical attack angle of the bat) and some
information from the flight of the ball (for example, its initial
launch angle at release, its rate of change of vertical direction,
its bearing angle, etc). From this, I can understand, what
specifically needs to be changed about the batter’s swing to hit
this pitch successfully. For example, they might be controlling
their swing using information that is non-specifying about the
height the pitch will cross the plate. As I discussed when
talking about predictive processing, I am focusing on the
observable (information, movement) not trying to infer internal
mental states.
From there I use a CLA approach to guide the batter to
search for different movement solutions (i.e. different
information-movement control laws). For example, I might add
the constraint of early occlusion of the ball flight so that the
batter must rely on the rate of change of direction or bearing
angle to control their swing instead of the ball’s launch angle at
release. Or I could alter the constraints by changing their task
goal – getting them to practice hitting the ball directly into the
ground to change the calibration between the information and
their movement. Finally, I might add a visual reference mounted
beside the plate with baseballs cut in half and numbers drawn
on them nailed to it, as illustrated in Figure 2.3. I would give
the batter the task of saying which ball number (i.e. height) the
pitch crossed the plate. As will be discussed in Chapter 5, this
is an example of using a constraint to augment the information
or feedback a performer receives. Finally, I might give them the
task of only swinging at pitches within a certain range of
heights (i.e., ball number 2-4) to augment the affordance of
making the pitcher get the ball down in the zone before they
swing.
A baseball field with a number of numbers Description
automatically generated with medium confidence
Figure 2.3 – Constraint to Provide Augmented Information
about Pitch Height
So, from an ecological dynamics perspective, I can come up
with lots of ideas – many of which I have tried and seen
results with. I can’t see how the active inference model of
batting can be used to generate similar ideas for practice
design, at least not yet. I think active inference (moving to
reduce prediction error) could give a reasonably plausible
explanation for the control of action. But ecological dynamics
can do the same, in a much simpler way using only observable
variables with much more applied value to coaches.
Recommendations for Using Sport Science in Coaching
Ok, given all these limitations and caveats what is a poor coach
supposed to do?! Here are some general tips I would give for
making the most effective use of sports science research in
coaching.
1.Pick a Side
I know it’s not a popular viewpoint, but I really do believe that
to get the most out of your athletes you need to pick a side in
the skill acquisition debate. I think sitting on the fence and
playing both sides is ineffective. For example, can we use
prescriptive instruction, task decomposition, and strict repetition
to promote an ideal technique in some practice sessions and
allow your athlete to self-organize in others? I think this is an
ineffective (and is really an impossible) stance to take. You
either believe that skill emerges from the process of
self-organization, or you don’t. Either you believe that internal
models or representations are involved in the control of actions
in sports, or you don’t. Either you believe that the key to
sporting success is developing a repeatable, ideal technique or
you believe we need to constantly adapt our movement to
ever-changing environmental conditions.
That is not to say you must coach in the same way all the
time. Not at all. For me, instead of playing both sides of the
prescription vs self-organization theoretical fence, it is much
more appropriate to think of the same types of manipulations
in different ways that align with one of the other views. For
example, manipulations like changing field and ball size or
adding different rules to practice can either be thought of as
changing the constraints to promote self-organization or
changing the conditions to promote a desired technique. Not
both! Similarly, making practice more game-like is either for the
purpose of supporting an athlete’s understanding through
changing, constructing or enriching knowledge structures and
internal models in a process of guided discovery. Or you are
doing it to support the process of attunement and the
development of coupling between information and movement.
Not both! As I hope to illustrate in the coming chapters, there
are Ecological Dynamics methods to address all of the problems
a coach faces.
I think it is valuable for a coach to have one clear theoretical
approach and decide which of these approaches to skill
acquisition fits better with your personal view. This will help
guide you in more effectively in designing practice. From Stone
and colleagues26:
Theory is the foundation upon which good practice is built. It
provides a solid and reliable base upon which practitioners can
design learning experiences. When it is most effective, theory
sits in the background “unnoticed”, but practitioners can
continually refer to it for guidance and reassurance. Powerful
theoretical ideas such as self-organization and affordances help
practitioners to continually monitor and question their activities
(and assumptions about learning).

2.Make sure research findings align with your experiential

knowledge – Believe in your approach.
Over the years I have got a lot of flak for using the word
“believe” (like I just did above) when I talk about skill
acquisition research. A common retort I will get is: “We are
taking objective scientific evidence NOT belief or faith, which are
subjective”. I respectfully disagree. I recall an interaction I had
early in my career with one of the true GOATs of sports
science - Lew Hardy from Bangor University. I was presenting
some of my work and afterward, I did what I thought every
good scientist is supposed to do: I talked at length about the
limitations and qualifications and did a lot of waffling – “well
this might work if” kind of thing. Lew looked me straight in
the eye and said to me: “While I respect you qualifying your
findings, do you believe in them? If you were working with a
team and had to bet a mortgage payment on whether your
manipulation would work or not, would you?” I will never
forget that. For any coaching method to be effective, you need
to truly believe in it!
Yes, my opinions about the best way to acquire a skill I have
written about in these three books now are heavily based on
scientific evidence, but still, they are also beliefs. I honestly
believe (and am now paying part of my mortgage) based on
these beliefs. I think part of the reason I am such an advocate
for this approach is that it aligns very well with my personal
experiences (e.g., working with Coach Phillips). Coaching
following an ecological approach is more effective, allows for
more individuality and creativity, and is more fun for the
athletes. I am sure some say that is just confirmation bias, but
as I documented in my first book and this podcast episode33
,
I started as a firm believer in the information processing
approach and spent most of my time looking for evidence to
support it. It was when I started working as a consultant with
professional teams that I began to see things that changed my
beliefs.
So, as I coach, I think it’s important what the scientific
evidence is saying resonates with you. That is, does it fit with
what you have experienced as a coach? Is it something you
truly believe will work? Along these same lines, in the immortal
words of Bruce Lee:
3.“Absorb what is useful, discard what is useless, and add what
is specifically yours”.
Coaching is a journey. It is an iterative process where we are,
as Craig Morris says “prepared not planned”. We should
continually adapt our methods to the ever-changing constraints
we face. For that reason, I think you must take out of
research what you think will work given where you are on
your journey and your constraints.
For example, I was recently speaking to a bunch of coaches
about practicing hitting (spiking) in volleyball. They were hesitant
to change the drills they had been doing (having players wait
in line so the coach could toss the ball for them to hit) to a
more ecological approach (e.g., small-sided games) because they
were worried about the number of reps the players would get.
Instead of getting into a big debate, I made two suggestions for
changes instead. First, can we add some more variability to the
drills by having three different locations where the players line
up, so they are required to approach the net from different
angles? Second, can we think about the representativeness of
the information for a second? Instead of just throwing the ball
in the air, could you have another coach throw it to you, and
then you set it for the player? And can we add a couple of
blockers on the other side of the net? I discuss this example in
more detail in Chapter 4.
As a consultant/skill acquisition specialist, I always try to meet
coaches where they are. If they use a lot of isolated,
decontextualized drills, can we add a little variability and make
the drill more representative and game-like in terms of the
information and action fidelity involved (discussed more in
Chapter 4), like in my volleyball, example? If instead, a coach is
already using a lot of small-sided games we might get into the
nitty gritty of designing some new ones. The same is true in
some of the coaching workshops I do (which I will discuss
more in Chapter 16). When I work with coaches who use a lot
of internally focused technical coaching cues, I don’t tell them
they have to stop cold turkey. Instead, we work together to try
and think about changing their favorite ones into external cues
or analogies that can be used instead. In that way, we are at
least giving them some alternatives and adding t to their
coaching toolbox.
I think my batting cage story is a good example of this. I
knew what the research findings said but when there was a
disagreement with the current practice I didn’t just say “You
have to do it this way because that’s what the research says”.
I tried to look at the whole context and concluded that it does
not apply in this case.
I think you need to do the same as a coach. Take from the
research findings what is useful to you now (at the current
point in your journey) and you can add to your current
practice.

4.“You can’t learn how to do brain surgery in a weekend

seminar”

One of the biggest complaints I hear about using an ecological

dynamics approach in coaching is that there are too many big
words involved – too much terminology that needs to be
learned. Attractors, Metastability, and Degeneracy, oh my! While
I am a little sympathetic to this, I will point out that skill
acquisition is a science that requires some effort to learn. The
title of this section is a quote that Frans Bosch likes to use
frequently. Dynamical systems theory is complex – you aren’t
just going to grasp it right away after reading a couple of
books or papers.
But, more seriously, in my work as a skill acquisition specialist,
I have really seen the value that comes from using the
terminology consistently and appropriately. For example,
referring to practice interventions as “constraints” keeps
everyone on the same page with what we think the
mechanisms are involved (i.e. self-organization) and their
purpose. The same goes for affordances and attractors. How
you use language in coaching is part of something we will dive
into in much more detail in Chapter 12 – the form of life you
create around your athletes. The person who first introduced
the concept, Wittgenstein45
, discussed this concept in terms of
language:
To grasp the meaning of a word in any given context it is
necessary to pay attention to the various non-linguistic activities
and practices engaged in by that group; since it is within this
context that any given language is used and any given
language will be interwoven with such activities and practices. It
is the use of words together with these non-linguistic activities
that make up ‘language games’. Speaking a language is part of
an activity and so of a form of life.
In other words, the meaning of language is inextricably bound
with how we act and vice versa. A form of life consists of the
behaviors, attitudes, values, beliefs, practices, and customs that
shape the communities we live in. In sports, every practice
activity and drill we do is inseparable from the culture and
customs in which they are embedded.
Ok, with all these preliminaries out of the way, let’s (finally!) get
more into the details of understanding coaching from an
ecological perspective…
3
AN ECOLOGICAL APPROACH TO
COACHING – THE COACH AS A
LEARNER & EXPLORER

I s there a better way to conceptualize the process of coaching

and learning to coach? One that considers the entire system in
which a coach works and provides a model for coach
development beyond simply sitting in a classroom and acquiring
knowledge about your sport? In this chapter, let’s consider how
we can understand coaching behavior and development from
an ecological dynamics perspective in a way that mirrors the
skill acquisition model for an individual athlete. This model is a
modified version of the one described in the paper by Wood
and colleagues1
.
Figure 3.1 illustrates an alternative, ecological dynamics-based,
account of coaching behavior. First, just as was the case for
the athlete, coaching behavior (which coaching methods are
used, how, and when) emerges from a set of interacting
constraints and a process of self-organization. Based on Newell’s
constraints model2 I discussed in my first book; these include
task constraints that are highly specific to the particular sport
and individual athlete(s) being coached. Things like the rules
and goals of the sport, the individual constraints of the athlete
(e.g., their physical dimensions, intentions/motivations, and action
capabilities), and the sport-specific technology and equipment
available to the coach. Another way to think of these is that
they represent “what the coach has to work with” for the
specific coaching task.
A diagram of a diagram Description automatically generated
Figure 3.1 – An Ecological Model of Coaching. Adapted from
Wood et al (2022).
We also have the individual constraints the coach brings to the
practice field. This includes any previous “knowledge of”
experience from coaching and/or playing the sport, “knowledge
about” their sport and skill (e.g., biomechanics, model of the
skill, strategy), and their general understanding and knowledge
about skill acquisition. Even though I have picked on it a lot so
far in the book, a coach’s “knowledge about” their sport does
play an important role in coaching, of course! This model
serves to frame their intentions for practice (e.g., what
constraints do we want to manipulate and when, what they are
going to look for in their athlete’s movement patterns). From
Stone et al.:
Theory acts as a constraint upon coaching practice, guiding
attention from the infinite possibilities for practice design and
interventions that can create a random experience that confuses
athletes. With a shared theoretical approach practitioners and
performers have a coordinated approach to why they practice
the way they practice enabling co-adaption with an anchored
reference point amidst what is inherently a dynamic journey.
There are also environmental constraints - general properties
that would apply no matter what sport or athlete is being
coached. For example, practical constraints of the practice
environment like the available space and time, and the number
of athletes being coached. They also include any available
opportunities for expanding the coach’s knowledge of and about
their sport through coach education. That is, both formal,
structured coach education (e.g. workshops targeted at
developing “knowledge about”) and more informal opportunities
to develop “knowledge of” (e.g., running a practice activity with
a mentor coach’s team).
Finally, in this model, we are going to acknowledge and explore
the socio-cultural constraints present in the coaching
environment. These include things like pressure from parents,
governing bodies, athletes, and other organizations (e.g.,
requirements for quantifying improvements and winning). An
example of this I have experienced many times when coaching
youth sports is the social pressure from parents to do what
they think coaching is supposed to “look like”. For many
parents, stopping the practice a lot, giving tons of technical
instructions and corrections to athletes, and having athletes
organized in orderly lines is what they associate with good
coaching (likely because that is the way they were taught
themselves). In my experience, letting athletes explore movement
solutions, play, and pick up affordances freely in something like
a small-sided game has often been met with “when are you
going to start coaching” comments. This, of course, contributes
to the strong path dependency in coaching we discussed in the
last chapter.
Within this model, coaches pick up information from the
coaching environment and use this to act (e.g., choose the next
instruction, change the practice activity, or give feedback). But
just as was the case for the athletes, this is a reciprocal,
coupled relationship where their actions serve to alter the
environment and thus the information they pick up. Effective
coaching involves establishing a relationship with the coaching
environment such that the emerging coaching solution results in
effective skill development of the athlete. As we will explore in
future chapters, there are several places where we can develop
and improve our coaching skills. These include:
(i)
Improving the ability to design practice activities through
appropriate task analysis, employing representative learning
design, and effective constraints manipulation (Chapters 4-7)

(ii) Improving coaches' ability to pick up information from the

practice environment (i.e., their attunement to things like
information about affordances available to their athletes) and
use this to effectively adapt their coaching behavior (Chapters
8-11).
In other words, mastering the art of “what to do next”.

(iii) Influencing the social-cultural constraints of the coaching

environment. This will involve creating a “form of life” around
one’s practice that emphasizes autonomy, self-regulation, and
psychological safety (Chapters 13-14)

(iv) Developing effective and actionable “knowledge about”

coaching. This includes learning to more effectively use
compatible theories (e.g. self-determination theory, deliberate
practice, the challenge point hypothesis), biomechanical
knowledge/models of one’s skill, performance data, and training
technology (Chapters 15). In this section, we will also consider
novel ways to track and measure skill development that
consider its nonlinear nature.

(v) Creating contextualized and situated opportunities to

improve “knowledge of” coaching and “clean” one’s lens
(Chapter 16)
Within this process of learning to coach, it will also be
important for us to consider what stage of development the
coach is in.

A 3-Stage Model of the Development of Coaching Skill

The stages of coaching skill development can be broadly
conceptualized using Newell’s three-stage model of skill
development for a performer2
. As illustrated in Figure 3.2,
The first stage involves establishing a basic coaching solution.
Although, in theory, we could coach a particular sports skill in
an infinite number of possible ways the solution will be shaped
by the particular constraints placed on the coach (illustrated in
Figure 3.1). Because there are many similarities between the
constraints faced by different coaches working in the same
sport, effective solutions will have a similar “topology” or basic
structure that connects the parts involved – much, in the same
way, effective movement solutions used by athletes for a
particular sporting skill (e.g. swinging a baseball bat) will share
a similar general structure or topology. Within traditional coach
education, this basic coaching solution is not found through a
process of exploration – it is handed down from another coach
or coach developer.
The second stage involves replicating the success of this
coaching solution by exploiting it by establishing the basic
parameters. Here we are talking about adjusting things like the
number of repetitions (or hopefully “repetitions without
repetition”), duration of a practice, the type and order of
constraints manipulated, and the amount of variability added to
practice.
In the final stage, adaptation and optimization, the coach learns
to tailor methods to get the most out of the individual athlete
they are working with. In this stage, coaches become more
attuned to the information from individual athletes and improve
their ability to individualize practice. This optimization stage is
also of particular importance when working with elite athletes
for which we are well beyond coaching the basic pattern of
coordination.
A diagram of a process Description automatically generated
Figure 3.2 – A 3-Stage Model of Coach Development
In this view, just as was the case for an athlete, a high level
of coaching skill comes from having a good functional
relationship with the coaching environment. A coach’s success
will depend on the goodness of fit of this relationship.
Throughout the book, we will look at how different coaching
methods (e.g., instruction, demonstration, constraints
manipulation, etc.) can be adapted for these different stages of
coach development

A Changing View of Knowledge Transmission - From

Indoctrination to Education.
A fundamental change in coaching that comes from this model
is the nature of knowledge transmission from the coach
developer to the coach and from the coach to the athlete. The
dominant model of coach education has long been based on
what Lave has called the “culture of acquisition”3
. This is the
idea that becoming an effective coach is a cognitive process that
involves first acquiring a body of second-hand, decontextualized,
and objective knowledge. This “knowledge about” coaching is
stored in the coach’s brain in the form of rules, concepts, and
representations that can be later applied in practice in the
“proper” context when the time is ‘right’. In other words, the
role of a coach developer is to transmit the knowledge that is
in their head into yours so that you can use this to interpret
and process the ambiguous information from the coaching
environment. We are back to the view of perception as indirect
(i.e., something that must be enriched with previous knowledge
and memory) that James Gibson so strongly fought against.
We are back to the traditional view of education as the passive,
asymmetric process of knowledge transmission from teacher to
student. Instructing not guiding.
It sounds harsh, but if we follow this approach coach education
becomes largely a process of indoctrination with the outcome
being a high degree of conformity and path dependency. To
quote a recent article by Selmi and Woods4
:
Its secondary nature constrains one’s search, narrowing their
focus towards the application and enactment of what has been
prescribed for them by another. In other words, focusing too
much on knowing about coaching may limit a coach’s
opportunity to grow their knowledge of it. When one is
examining the world for oneself there is no limit to the scrutiny
– one can look as carefully as one wishes, and one can always
discover new information. But this is emphatically not the case
with [transmitted] second-hand information.
In this book, I want to take what Selmi and Woods recently
called a situated approach to coach development. One that is
based on Ingold’s concepts of educare (educating as leading
out) and enskillment (“knowledge of” that comes from exposure
to and experience with one’s environment)5
. Instead of just
regurgitating and reciting secondary information, we will focus
on continuing to develop a coach’s relationship with their
environment – their attunement and adaptability.
Let’s next dive into the different components of this model –
beginning with practice design.

4
“HAVING YOUR CAKE AND EATING
IT TOO”: APPLYING
REPRESENTATIVE LEARNING
DESIGN
I n this chapter, I want to examine how a coach can learn to
apply the basic principles of representative learning design to
their own practice. This will be key to achieving what we are
all looking for – a positive transfer of training. That is, ensuring
that we are doing in practice makes our athletes better in the
game. To understand this, we first need to revisit what it
means for an athlete to be skillful in the ecological approach.

Remember that in our new view of skill acquisition, the

purpose of practice is NOT to build up some internal
representation or motor program for a particular movement
that we can pull out later and use automatically. In other
words, the purpose of practice is NOT for the athlete to learn
a specific motor solution or set of solutions. The purpose of
practice is for the athlete to learn to solve movement problems
– that is to self-organize into a movement solution that will
achieve their goal under different sets of constraints. We are
looking for dexterity (adaptability), not automaticity. We are
trying to teach the athlete to softly-assemble the movement
degrees of freedom into a solution NOT hard wire them into a
reflexive routine. To understand this, I first want to consider
the difference between the concepts of specificity and
representativeness.
Breaking Loose from the Shackles of Specificity of Practice
The term “specificity” as it is applied to sports practice has a
somewhat convoluted history. In the context of skill acquisition
research, one of its most prominent uses was in the pioneering
work by Luc Proteau and colleagues in the late 1980’s/ early
1990’s. In a series of very clever studies, these researchers
investigated the learning of simple manual pointing tasks under
different viewing conditions1, 2
. One group of participants
learned the task under normal, full-vision conditions. That is,
they could see both the target to be hit and their hand for the
entire duration of the movement. Another group learned the
task under restricted vision conditions in which they could only
see the target they were aiming at. Their hand was not visible
during the movement. Following 2000 (!) trials of practice, both
groups were then tested under two conditions: full vision and
target-only visible.
The typical pattern of results found in these studies is illustrated
in Figure 4.1. The group that learned under full vision (FV)
conditions struggled when the vision of their hand was taken
away in the post-training test. This is not surprising.
Information that was available during practice (e.g., the tau of
the visual gap between their hand and the target – See
Chapter 2 of my second book) and likely was used in their
information-movement control law was now gone. The more
surprising finding is what happened to the group that trained
under restricted, target-only visible conditions. They showed
poor performance under full vision conditions. Since there was
no visual information about their hand position available during
training, this group likely learned to use proprioceptive
information about the position of their hand. So, why then did
they perform so poorly in the full vision test? The
proprioceptive information about hand position is still available
plus there is now additional useful visual information. Adding
more useful, task-relevant information hurt performance for this
group! What is going on here?!

A graph with black dots and white text Description

automatically generated
Figure 4.1 – Specificity of Practice in Learning.. Based on data
from Proteau and colleagues2

Building on earlier ideas emphasizing the importance of the

similarity between the practice and test environments for
transfer4,5, Proteau and colleagues proposed the “specificity of
practice hypothesis” to explain this effect. This hypothesis
proposes that learning is highly specific to the information used
to ensure optimal performance during practice. When we add
more information after the skill has already been learned (or
make the conditions easier in some way), not only does the
information not seem to be used but it can hurt performance!
This basic finding has been shown with other manipulations too.
For example, it has been shown that training under conditions
of high anxiety improves subsequent performance but only
when tested under conditions of high anxiety6
. Although the
specificity of practice hypothesis has been challenged in
subsequent research over the years, the basic hook was set -
learning was highly specific to the conditions of practice,
therefore, if we want transfer of training to competition we
need to practice under conditions that are specific to the
competition environment.

So, in popular vernacular, the term “specificity” has come to

mean the extent to which practice resembles competition with
the assumption being that we should try to make practice have
as high a degree of specificity as possible. As a few examples:

“Simply put, the principle of specificity states that how you train
should mimic the skills, movements, and action required to
perform and excel in the game, activity or event you are
participating in” 7

and

“The general hypothesis, that we attempt to make those

conditions in acquisition practice like those expected in the
criterion “test” performance, is an old one based on common
sense” 8

and

“The principle of specificity implies that to become better at a

particular exercise or skill, one should perform that exercise or
skill. For example, a runner should run to improve running
performance”. 9

Practice like you play. Coaching 101.

But the concept of “specificity of practice’ has a few very
important limitations. First, it does not provide much guidance
about what exactly needs to be the same between practice and
competition for transfer to occur. Is it everything? The use of
the word “specific” implies the answer is every single detail. So,
does that mean the only way you can get better is through
actually playing the real game?

The second issue, if we stick to the rule of “specificity of

practice”, is that it does not allow a coach the flexibility to
implement some training interventions that have proven to be
effective. This includes things like the principles of overload and
underload10. Research has consistently shown that training
with equipment that is heavier or lighter than what will be used
in competition can have benefits11. But this violates the
principle of specificity – why in the world would I practice
pitching a 6 oz baseball if I and never going to be asked to
do it in a game played with a regulation 5 oz ball? Another
example is the use of visual occlusion in training (which I
discussed in Chapter 13 of my first book). Inconsistent with the
basic idea of Proteau’s “specificity of practice hypothesis”, even
though an athlete will perform under full vision conditions in
competition, selectively removing visual information at certain
times12 and of certain locations13 in practice can benefit skill
acquisition. Finally, the concept of exaggerating information and
amplifying affordances – the idea of “sending invitations in pink
envelopes” I discussed in Chapter 6 of my second book (e.g.
the proven benefits of using small-sided and conditioned games
for team sports) would again seem to go against the idea of
specificity.

Related to this point, is the final problem with the concept of

specificity: the game is not always the best teacher. That is,
practicing in a highly specific, game-like manner is not the best
way to learn a new skill. Playing a full game will typically not
give an athlete enough repetitions (repetitions without repetition)
of a skill to get better at it. Think about a soccer match. How
many times does a striker get to shoot in a game or how
does a goaltender get to make a save? Just playing a game
also greatly limits our ability to implement deliberate practice,
where we focus on improving an athlete’s weaknesses.

I think the reason we have latched so strongly on to the idea

of specificity is because it was a good fit with the dominant
theory of skill acquisition. That is, that skill lives in the
performer’s head. The goal of practice is to build up automatic
motor programs, mental models, memory structures, etc. of the
skill that they can be pulled out in the game. The goal of
practice is to perfect the ideal solution. What benefit would it be
to develop a solution for a set of conditions that will not occur
in the game? The answer can be seen if we let go of this
not-very-useful concept of “specificity” and turn to the far
superior “representativeness”.

Having Your Cake and Eating It Too: The Key Ingredients of

Representative Learning Design.

When speaking about the concept of representativeness, I like

to use the analogy of eating a birthday cake. To enjoy the
cake do we need to eat the whole thing (i.e. the specificity
answer)? No! We can just take a slice. But we need to make
sure the slice has all the key ingredients, so it faithfully
represents the whole cake. We don’t want to just scrape icing
off the top! Representative learning design (RLD) is taking a
slice out of the game so that a particular part of the game can
be focused on for a while and developed. We are deliberately
moving away from the game for a while for a purpose. But
we are going to make sure the slice still has all the key
ingredients that the full game (the cake) has. This will ensure
we achieve the main goal of RLD: to maximize the likelihood
that the skills being practiced will transfer to competition. In
RLD, our cake has four key ingredients that relate directly to
the key principles of Ecological Dynamics I discussed in the
preface:
1) Information & Information-Movement Coupling

So often, in traditional practice, we focus on developing

technique NOT on becoming skillful. That is, we spend way
more time on the action while mostly ignoring perception. The
distinction between technique and skill was recently discussed in
a paper by Bennet and Fransen14.:
Simply put, we can consider a technique to be but one of the
possible movement solutions to a movement problem faced by
an athlete. Skill, on the other hand, represents the extent to
which an athlete can adapt that technique to produce behavior
that suits the task’s demands (i.e., is functional) and results in
performance optimization (i.e., is beneficial).
In the ecological approach, being skillful is not solely about the
ability to generate a particular action (aka a technique such as
a baseball or golf swing). Equally important is the ability to pick
up information from one’s environment which specifies what
action will be effective and when to use it. Skill is a relationship
between information and movement. We can’t train action in
the absence of the information that specifies how and when it
can be used. We need to “keep ‘em coupled”!
How, as a coach, do you know whether the information
available to an athlete in practice is representative? There are a
few important qualities I look for. First, representative
information is dynamic: the opportunities for action it specifies
appear and disappear over time. Second, it specifies events as
they will occur in the game, not just in practice. Third, it is
variable – it doesn’t specify the same affordance (opportunity
for action) on each repetition. I will use examples to illustrate
these shortly, but let’s first look at the other ingredients.
2) Action fidelity
Along with being coupled to specifying information, there are
other important criteria for how the action is performed. Does
the action involve coordinating similar degrees of freedom? Does
it have similar temporal and spatial constraints? Are there
similar requirements for transferring force? Is it nested within
the other actions that the athlete must perform?
As an example, I have seen it recommended that golfers can
improve their game by practicing catching a tennis ball
bouncing off a wall with one hand. While this might be fun, the
task dynamics (intercepting a moving object with a simple hand
movement) are completely different than those involved in
hitting a golf ball (contacting a stationary object with a tool
using force generated by the movement of the whole body). As
a simple rule, if I can’t tell what sport you are practicing for
while watching you train, it's likely not representative!
3) Affordances and decisions (plural!)
For a practice task to be representative one of the most
important ingredients is presenting the athlete with a landscape
of affordances such that they must decide in the moment
which movement solution to use and when to use it. Critically,
we don't need to present all possible action opportunities that
could occur in the game – again we can just take a small
slice. What we want to ensure is that there is some degree of
unpredictability (i.e., the athlete doesn’t know the exact action
they will perform before the drill begins) even if there are only
a few possible options.
A good example of this can be seen in a recent study by
Caldeira and colleagues15. In this study, experienced volleyball
players were trained in the task of approaching the net to
spike the ball. One training group practiced in a traditional,
repetitive/blocked practice manner. Before each trial, players
knew where the ball would be set and where the blockers
would be positioned. After practicing this play several times in a
row they switched to a different position at the net and to
different blocker locations. There was a total of three different
locations/blocker positions. For the unpredictable/representative
training group, one of these three plays was chosen randomly
on every trial. The main dependent variable was the number of
successful hits, measured before and after training.
What was found? In a post-training test, the unpredictable
training group had significantly more successful attacks as
compared to the predictable group (62 vs 45%). Digging deeper
it was also found that the unpredictable training group had
more spatiotemporal variability in their attacking movements
suggesting they were exploring more possible movement
solutions. As the authors of this study conclude:
The task variability provided by the unknown, but yet limited,
actions of the block, guided the athletes to become perceptually
attuned to the blockers’ actions and channeled them to find
functional movement solutions under rapidly changing
performance constraints.
On the surface, this finding is also consistent with the results of
a study by Mercado-Palimino et al
16. which compared
planned/predictable vs unplanned/unpredictable movements in
volleyball blocking. This study revealed more functional
movement variability (leading to less ground reaction forces, and
smaller knee flexion angles – both markers for ACL injury) in
the unplanned condition. As I discuss in more detail in Chapter
6, reducing the number of affordances/options to a small
number like this is a great example of using task simplification
(as opposed to decomposition) to make practice easier for a
lesser skilled performer.
This study also illustrates the two other qualities of the
decisions we are looking for in representative learning design.
First, decisions should be emergent – that is they should arise
through the performer's coupled interactions with the practice
environment, rather than be pre-determined before the drill
starts. Second, decisions should be embodied – they should
depend on the performer's individual constraints (body
dimensions and action capacities). A player that can jump high
has more options for blocking compared to one that can’t.
4) Emotional context
The final ingredient for RLD is ensuring that practice generates
some of the same emotions that are experienced in competition.
In particular, those related to feelings of anxiety and
performance pressure. Are there consequences for whether the
athlete succeeds or fails in achieving their goals? Is there similar
time pressure to act? Is there social pressure and competition?
Examples of Employing Representative Learning Design
In this section, I want to look at a few examples of applying
RLD that I have completed in my work as a consultant. The
first step in this process is always to perform a task analysis
(of both the game and practice task) and to assess the degree
of representativeness. After that, I meet with the coaches to
discuss ways we can change things.
As discussed by Buszard and colleagues17
, there are three
different ways we can evaluate the representativeness of
practice. The first method involves conducting a notational
analysis of practice (i.e., documenting what is being practiced)
so that the frequency at which different events/conditions are
practiced and the context in which they occur can be related
to competition. The second method involves using the
Representative Practice Assessment Tool (RPAT) developed by
Krause and colleagues18. An example of this tool, which I
discussed in detail in Chapter 16 of my second book, is
provided below. An RPAT assessment checks for the key
ingredients of RLD I described above. Another variant of this
(the Alive Movement Problem Checklist) was provided in a
recent article by Shawn Myszka and colleagues19 and is shown
in Figure 4.2. Finally, a more detailed analysis can be
conducted using tracking technology. For example, we could
compare GPS tracking data from units worn during practice
with data from Hawkeye camera technology (e.g. as is currently
used in tennis and baseball) to explore how the number and
timing of accelerations and decelerations compare between
practice and competition.
Keep the problem-solution relationship intact
Present a task disposition (emotional context) representative
of the competitive practice environment
Contains relevant sources of information for the athlete to
regulate movements
Keep perception, cognitions, and actions deeply intertwined
Maintain a practical and relevant goal as an intention to
channel the movement solution
Allow for the continuous (re) organization of system
degrees of freedom
Require the athlete to authentically connect to the problem
in their own unique way
Maintain a certain level of unpredictability, requiring the
athlete to actively make decisions as needed
Present emerging and decaying affordances
Change the problem in some meaningful way each time it
is faced (repetition without repetition)
Figure 4.2 – Alive Movement Problem Checklist. From Myszka
et al. 19

Let’s dive into the examples…

1)Baseball infield practice.

Starting with the task analysis, the characteristics of traditional

baseball infield practice I observed were as follows:

-Self-toss fungos: a coach throws a ball up into the air (to

themselves) and then hits it towards the fielders (shown in
Figure 4.3, top panels)
-The player that will field the ball is told before the ball is hit
that they will be the one to field it (either verbally or
through a hand gesture)
-If a throw is made after the player fields the ball (on about
60% of observed drills), the player that fields the ball knows
who they are throwing the ball to before the ball is hit (e.g.,
1
st base). This is varied in a predictable/blocked fashion.
-Typically, there are no baserunners on any other time
constraints on the fielding action.
-There is some variability in how the ball is hit (e.g., # of
hops before it reaches the fielder, whether it is hit to their
left or right, etc.).

A collage of a person swinging a bat Description automatically

generated
Figure 4.3 – Self-toss (top) vs front toss (bottom) fungos.

To assess the representativeness of this type of practice, I first

looked at game data to determine when baseball infielders
typically make errors20
. Errors are 60% more likely to occur
when there is a runner on base. They also become less likely
with 2 outs. Why would the presence of a runner or the
number of outs matter? Because, with a runner on base
and/or less than 2 outs, the fielder is required to pick up
affordances (e.g., can I turn a double play or should I just get
the batter out) and make a decision during the play (i.e. which
bag to throw to). The next important effect I found was that
plays on which errors occur have an average leverage index of
around 1.2. The leverage index is a measure of the importance
of the situation to the overall outcome of the game with 1.0
being average. Therefore, errors tend to occur more often
in more important game situations suggesting that emotional
context (pressure) plays a role. Finally, looking at the running
speed of the player that hit the ball, the rate of errors for a
fast runner (30 feet/sec, 98th percentile speed) is significantly
higher than for a slower runner (24 feet/sec, 5th percentile
speed)21. So, in other words, the time constraints matter:
fielders are less successful in completing the play when they
have less time (i.e. the runner is faster). Clearly, there was a
disconnect between what we were practicing and the situations
that were the most challenging in a game.
Finally, I completed an RPAT assessment as shown in Figure
4.4

A screenshot of a grid with red circles Description automatically

generated
Figure 4.4 - Representative Practice Assessment Tool (RPAT)
for Baseball Fielding

Details about my scores were as follows:

While in some cases the plays practiced were targeted at a
specific player's weakness (e.g., the coach would hit more ball
to the left if that is a play that they struggled with) this was
typically not the case and the overall goal of the session was
just “fielding practice”. Too vague.
The use of self-toss fungoes is problematic for two reasons.
First, the information that specifies the time to contact
between the bat and ball (the ball’s vertical drop rate) is
non-specifying in the game (in which the ball approaches the
bat longitudinally rather than from above). Second, when the
hitter tosses the ball to themselves it results in the ball being
hit squarer so that it has less spin and travels slower and
straighter than a ball hit in the game. Thus, the constraints
on the ball flight are different.
Typically, what the player is required to do is completely
predictable before the ball is hit so no emergent decisions are
required.
Players do not need to adapt their intentions as much as we
would like. For example, very few balls are hit between
players where there is uncertainty about who will field it.
Yes, the player is required to adapt their movement solution.
For example, moving forwards vs backward, play on the
backhand, playing on the forehand, playing on a short hop,
etc. However, this is usually done in a very blocked,
predictable manner so could be improved.
The lack of time constraints makes the challenge level
relatively low.
The task dynamics of fielding the ball are fairly representative,
however, because the player is frequently not required to
throw the ball, it is likely to result in differences in the
movement solution since the act of catching is not nested
within the act of throwing.
There are no real consequences to making or not making
the play and the lack of time pressure makes the task very
different from the game. In some cases, this type of fielding
practice is done in front of fans so there is some
performance pressure.

Recommended Changes
Switch to using front-toss fungos. That is, have another
coach toss the ball to the coach that it is hitting, as shown
in the bottom panel of Figure 4.3. It doesn’t really matter
how far away it is thrown from or whether it is
underhand/overhand – we are trying to make practice
representative for the fielder NOT the hitter. This makes the
information and ball flight characteristics more game-like and
also adds some variability to where the ball is hit.
Hit the ball to a random location so the play is not
predictable. As in the Caldeira et al. study, this does not
require all possible options (e.g. the coach could hit the ball
just to the right side of the field).
Add some sort of time constraint on the play. After
discussion with the coaches, we decided that having
baserunners regularly be a part of fielding practice was
impractical due to workload/injury concerns. So, we put a
time clock on the field with a loud buzzer when time
expired.
Keep a score for plays to create some pressure/competition.
Was the throw accurate? Did it beat the timer?

2) Volleyball spiking/hitting practice

As I discussed briefly in Chapter 2, when I was at a

conference a couple of months ago I had a couple of coaches
ask how they could make the typical spiking/hitting practice
more representative. The main features of this type of practice
identified were:

Players stand in a line waiting to hit the ball

The ball is thrown in the air by a coach rather than being
set by another player
There are typically no blockers and the location (at the
net) at which the spike occurs is varied in a blocked
fashion.
There is no game context surrounding the play (e.g. no
score, history of the previous shots made in that point,
etc).

As the result from the Caldeira et al. study shows, this is a

less effective way to learn this skill. My RPAT assessment is
shown in Figure 4.5.

Details about my scores were as follows:

No specific, individualized practice goals for players.
The information is very different. The player cannot track the
ball from pass to set to spike. The information they use to
time their spiking action (the arm swing of the person
throwing the ball) cannot be used in the game.
There are no decisions required.
The intention is the same on every execution and does not
need to be adapted.
Very little adaptation of the movement solution is required
because the set is always in an ideal location, the player is
always approaching the next from the same angle and there
are no blockers.
The challenge level is too low. The action is successful on
almost every attempt.
The task of hitting the ball is the same but no adjustments
are required.
There is no real pressure or competition.

A grid with red circles and numbers Description automatically

generated
Figure 4.5 - Representative Practice Assessment Tool (RPAT)
for Volleyball Spiking/Hitting

Based on this analysis, I recommended using a small-sided

games (SSG) approach, however, the coaches expressed two
concerns about completely abandoning their spiking drill. First,
they worried that in a SSG, players would not get enough
opportunities to spike/hit to show improvement in this skill.
Second, they felt that the variability present in an SSG would
be too much for a new learner to acquire “the fundamentals”
of how to spike the ball. They needed to make the task easier
for them. Well, as will be discussed shortly, I think there is
evidence that contradicts both of these beliefs, I was
sympathetic to their views and tried to “meet them where they
are” in their coaching journey. Here are my recommendations.
1.
Have another coach throw the ball to a coach who will set
the ball to the hitter. This will make the specifying information
for timing the action more like it is in the game. It will also
add a small amount of (beneficial) variability to the locations of
the set.
2.
Instead of one line of players waiting for their turn to spike,
have three lines. I gave a couple of different options here. First,
players would be told which line was going to hit next. While
predictable, this at least gives some variation in the approach
angle into the hitting action. The other option was to have
players from all 3 lines approach the net and the one that hits
the ball be an emergent decision based on where the ball was
set.
3.
Have a few blockers on the other side of the net that are
varying their position
4.
Individualize practice by giving players more reps with plays
they have trouble with in games (e.g., approaching the ball
from the left).
5. Create a competition based on successful hits
3) Training in Martial Arts and Boxing
In recent years I have been very excited to see an ecological
approach being adopted in martial arts and MMA instruction.
This has been challenging because the traditional theory of skill
acquisition has seemingly been the undisputed champion in this
sport for a very long time. When most people hear the words
“marital arts training” they picture Bruce Lee performing the
same striking action over and over against a wood block.
Although, as discussed in this great article22
, Bruce Lee was
actually a closet ecological approach guy.
One of the problems I get asked about a lot in this area is
how to teach a beginner the proper stance for boxing or
striking without just prescribing the solution to them. In looking
at the way this is traditionally taught, here are the key features
we can identify
The athlete is given detailed technical internally-focused
instructions about where to place their feet and hands
(e.g., “keep both feet pointing in the same direction,
parallel to each other at a 45-degree angle” “point
shoulder towards bag”, “keep elbows in”) as shown in
Figure 4.6.
Typically, the stance is taught completely out of context
with no opponent present. If the athlete is required to
strike it is against a heavy bag.
There is no goal or intention to the action other than
performing the action itself.
There are few opportunities to practice nested affordances
(e.g. defending after striking).

A person wearing boxing gloves Description automatically

generated
Figure 4.6 – The correct boxing/striking stance

The RPAT assessment for this type of drill would be something

like:

A screenshot of a survey Description automatically generated

Figure 4.7 - Representative Practice Assessment Tool (RPAT)
for Practicing the Boxing Stance

Recommendations: Boxing Tag

What is the alternative to hitting a heavy bag and having a

coach prescribe and correct the ideal stance? Playing tag! This
activity involves having opponents play a game together where
each is required to try to tag their opponent on their shoulder
or belly, well at the same time, avoiding being tagged
themselves. Based on how the stance is emerging (more on
that next chapter) the coach can manipulate the task
constraints including which body part should be tagged, the size
of the ring, the staring positions, etc. When I interviewed him
for my podcast, MMA coach Scott Sievewright explained this
activity:
We play a little shoulder tap or belly button tag game. When
someone initially comes in, I really don't need to tell them how
to stand at all. That tends to emerge pretty organically. They
know not to have their feet stuck together. They know not to
stand with their back to their opponent. So relatively quickly, an
effective box stance appears, and it's really quite nice to watch
how. Even in a space of a short few weeks or sessions this
real organic footwork starts to emerge.
In boxing tag, movements are purposeful and functional. They
are driven by the information a boxer will use in competition
(the position and distance of their opponent). They must adapt
their intentions and actions based on the constraints and
specifying information from their opponent. And, just like the
game we play in the schoolyard, boxing tag is challenging and
fun!
But, most of all, this activity includes one of my favorite
features: the performer gets to learn the “why” themselves.
Why do we keep our chin tucked and our elbows close to the
body? We could have a coach prescribe these positions to us
and (maybe) tell us why we are doing it. Or, in activities like
the boxing tag, we can learn the “why” ourselves. We can
learn the key elements of a boxing stance that tend to produce
more successful performance outcomes and ones that tend to
produce less successful ones. In other words, we will gain
through experience (through our own successes and failures)
knowledge of some of the invariants involved in boxing! In my
experience, learning to use a particular movement pattern
because you have experienced consequences is much more
powerful than doing it because a coach told you so. It also
allows for much more room for adaptation to individual
constraints (e.g. standing with a more open or closed stance).
Research on RLD
Is there any evidence to support the idea that more
representative practice leads to better transfer of training to
competition? Or any evidence that employing RLD leads to
more game (competition)-like movement solutions? Along with
the volleyball study described above, there have been a small
number of studies that have addressed this question.
In 2019, Krause and colleagues23 compared three different
levels of representativeness (low, medium, and high) for
practicing the tennis serve. The low representativeness group
(RPAT score = 22) practiced the serve with no opponent on
the other side of the court. The medium group (RPAT score =
33) practiced serving to an opponent who attempted to return
the ball, at which point the rally immediately stopped. Finally,
the high group (RPAT score = 40) practiced the serve against
an opponent and then played out the point. Before and after 6
weeks of training, participants completed a simulated match play
test. It was predicted that the higher representative practice
would result in more variability in the serve placement,
direction, and speed to create a positional advantage against the
opponent (i.e., more evidence of the adaptability of movement
patterns to changing constraints). It was also predicted that the
low-representative practice group would prioritize just speed
over these other tactics.
What was found? The relationship between increasing
representativeness and skill acquisition was not linear, rather
different behaviors emerged. That is, there was no overall trend
that the highly representative group always focused on ball
placement rather than speed. Instead, when hitting 2nd serves
in match play, the low and moderate representative groups
prioritized speed over placement while the high representative
group prioritized placement over speed. I think this study
highlights an important point we will touch on again in Chapter
11: we are dealing with a complex system here so the
behaviors that emerge and the adaptations athletes make are
not going to be perfectly predictable from the changes made to
practice. Constraints do not cause behaviors. They do not
produce perfectly predictable and deterministic changes to a
performer’s movement pattern. They lead to complex emergent
behaviors.
Barris et al. investigated the representativeness of diving practice
21. Six elite springboard divers each participated in two testing
sessions: in a dry-land training facility and in the pool. Divers
performed the same springboard dive phases for a reverse dive
take-off. However, in the dry-land condition divers only
completed a partial dive (one somersault) and landed feet first
on a foam mat. In the pool-based training, divers completed
traditional wrist first entries from a 3 m springboard. The
preparation phase of five randomly selected reverse take-offs
was captured for each participant. There were some major
differences in the take-offs made in the two training
environments. In the aquatic environment, divers exhibited
significantly greater step lengths and jump heights before
takeoff. This again highlights the importance of maintaining the
nesting of actions in practice.
We will look at more research on RLD in future chapters.
Conclusions
The real power of RLD is that gives the coach a principled
and deliberate way to allow their practice to deviate from the
game. This is what we want. We don’t just want to play the
full game all the time – we want to take slices out of the
game to focus on them and improve them. We want to find
new, representative problems for our athletes to solve to
increase their dexterity and adaptability. Next, we will explore
one of the best methods for achieving this: the constraints-led
approach (CLA).

5
DESIGNING EFFECTIVE
CONSTRAINTS & USING THE CLA
I n this chapter, I want to focus on one of the key
approaches we can utilize to make our practice more
representative - the constraints-led approach (CLA). This is the
coaching methodology I find myself using the most and find the
easiest to teach to other coaches. I want to start by reviewing
the logic and theoretical rationale of the CLA. Next, we will
consider the basic principles for designing effective constraints.
Guiding Self-Organization & Clearing Up Misconceptions
It is exciting to see the CLA grow in popularity as a coaching
method and being adopted by so many coaches, however,
there are a few misconceptions about it that remain that I
would like to clarify. First, I often see traditional coaching
contrasted with ‘constraints-based coaching”
1,2 with one of the
main conclusions of this type of comparison being that using
constraints manipulations is nothing new – it is something
coaches have employed for a long, long time. Well, I am here
to tell you that is completely true. As illustrated in Figure 5.1,
ALL coaching methods involve constraint manipulations! Every
single one. They always have and always will. There is literally
nothing else you can do as a coach!
Diagram of a diagram of a model of a theory Description
automatically generated with medium confidence
Figure 5.1 – Relationship Between Newell’s Constraints Theory
and the CLA
Think about it. As Karl Newell defined them
3
, constraints
include the equipment, the playing area, the number of players,
the rules, instructions, demonstrations, and feedback. What else
can a coach manipulate in practice? Nothing! What differs
between traditional “constraint manipulations” and the CLA is
the purpose of the manipulations. The “why”. Traditionally,
constraints have been manipulated in practice to get an athlete
(or members of a team) to move in a certain way. Execute the
ideal technique. Execute the play exactly as it is drawn up on
the whiteboard. In other words, they have been used in a
prescriptive fashion – giving the athlete(s) the solution. In the
CLA, they are, of course, used for a very different purpose: to
encourage the athlete to self-organize and find their own,
authentic movement solution by giving the athlete a problem to
solve.
The other issue I see discussed a lot is whether there is
anything “correct” or “incorrect” about a movement solution
and to what extent the ecological approach allows the coach to
intervene in this self-organization process. In my view, even
though there is no one ideal solution, effective movement
solutions do have certain key properties. Whether we call these
invariants – key properties of the movement solution that are
non-negotiable. Or we use Kelso’s term: movement topology4
referring to the basic spatio-temporal structure of the solution.
Adopting an ecological approach does not mean that a coach
must throw away all their knowledge about what an effective
movement solution for their skill contains (or looks like) and/or
ignore insights that can be gained from things like
biomechanical analyses. Doing this in the name of “promoting
self-organization” would be like throwing the baby out with the
bath water. This is why my preferred metaphor for the CLA is
“guiding self-organization”. In the CLA, we don’t just want to
let the athletes do whatever they feel like and struggle to find
success or potentially injure themselves – we want to use our
knowledge about (understanding of the skill supplemented with
analytical information) combined with our ever-growing
knowledge of (knowledge about how to coach it) to help them
find an effective solution.

The Logical Flow of the CLA

Start with a Problem. To most effectively apply the CLA, we
need to start with a problem: something we don’t like about
the athlete’s current movement solution. The two main reasons
we might want to change an athlete’s movement solution are:
(i) it is resulting in sub-optimal performance outcomes and/or
(ii) it has the potential to cause injury. If neither of these is an
issue, then we should leave it alone! As my father used to say
“There is more than one way to skin a cat” (aka embrace the
bliss of motor abundance). We don’t want to suppress
individuality and authenticity in the name of orthodoxy.
Starting with a problem like this is critical because it gives us a
principled way to choose the type of constraint manipulation(s)
we are going to use in our practice design amongst the almost
infinite number of possibilities. It also emphasizes a critical
point: effective use of the CLA is deliberate and individualized -
targeted at the weaknesses of a specific athlete. In most cases
using the same constraint for every athlete you are working
with will not be an effective use of practice time (more on this
in Chapter 7). Unless, of course, we are working on a
team-level problem like not enough ball movement in basketball.
But we will get to that.
So, Step #1 in the CLA is to identify an aspect of the
movement solution you would like to change for an individual
athlete. When we work with elite athletes, the aspect we want
to change is likely to be much smaller and specific to a smaller
subset of game conditions. Let’s now look at an example from,
you guessed it, baseball (I will have some examples from other
sports too, I promise!).
Example #1 – Increasing Bat Speed in a Baseball Swing
John Soteropolis is a hitting coach with the Boston Red Sox
that I have been lucky enough to work with for the past
couple of years. One of the movement solution problems he
has focused on a lot in the past few years is slow bat speed.
Why is this a problem? It is a simple fact that, in baseball, the
faster you swing the bat, the harder you hit the ball. As
discussed in my second book, it also gives you longer to pick
up visual information from the ball in flight. Adding 5mph to
your bat speed increases the exit velocity of the ball leaving the
bat by 6 mph. The harder you hit the ball, the more value
you add to your team and the more you increase their chance
of winning. For example, increasing exit velocity by 6 mph from
95 to 101 mph results in an increase in wOBA by nearly 200
points. Before joining the Red Sox, John documented a specific
example of using the CLA to address this slow bat speed
problem with Lars Nootbar from St Louis Cardinals5.
Nootbar’s swing had an average bat speed of 67 mph, about 5
mph below the MLB average – clearly a suboptimal movement
solution for a major league hitter. He had identified a problem!
Supplement Observation with Analytics. The second step in
applying the CLA is to try and understand the underlying
cause(s) of the problem in more detail. This can be done
through a combination of the coach’s knowledge/observation
and technology. This is exactly what John did. From
observation, he identified the key issue as being the load in
Nootbar’s swing (i.e., how he shifts weight onto his back foot
before moving forward toward the ball), shown in Figure 5.2.
He next used a motion tracking technology called the K-Vest™
to get more detailed information as shown in Figure 5.3. From
this output, we can see, that Nootbar’s pelvis was rotating (65
deg) more than his torso (12deg). The caused the timing of the
coiling of his body to be out of sync. The net result was a
down-the-line effect (i.e., later in the movement solution) that
there was very little separation between his hips and shoulders
when he started the forward movement toward the ball.
John explains why this is problematic:
Think of the core muscles as a rubber band. When the pelvis
rotates in front of the torso, hip-to-shoulder separation is
created and stretches the rubber band (internal and external
obliques stretch). Next, as the pelvis decelerates, the torso
accelerates, which would be the rubber band “snapping” into
contact (internal and external obliques contract). With 8° of
hip-to-shoulder separation, Lars was barely stretching his rubber
band, and losing out on potential elastic energy.
A collage of a person swinging a bat Description automatically
generatedFigure 5.2 – Lars Nootbar’s Load Phase of the Swing
before (left) and after (right) CLA training. Reproduced with
permission from Driveline Baseball.
This description also exemplifies my preferred way of observing
an athlete’s movement (which will be discussed more in
Chapter 8): viewing movement solutions through the lens of
attractors6
. Specifically, his movement solution does not include
one of the key attractors for effective movement (discussed in
detail in Chapter 15) – proximal to distal action. When we
move, we want body parts closer to the center of the body
(e.g., the pelvis) to complete their rotation prior to the rotation
of parts further away from the center of the body (e.g., the
torso). Again, note the difference between the problem John
identified and an ideal movement solution: He is not saying
that Lars should rotate this much, and at this time – he has
identified a multi-joint coordination property (separation) that is
not emerging.
A screen shot of a baseball game Description automatically
generated
Figure 5.3 – Biomechanical Data Used to Support Swing
Analysis. Reproduced with permission from Driveline Baseball.
Design the Constraint. Step #3 is for the coach to identify a
constraint that can be added to practice that will destabilize the
athlete’s movement solution. The term “destabilize” again comes
from an attractor-based analysis. An attractor is a very stable
state in a movement pattern that a performer has settled into.
To promote a change, we need to destabilize this attractor.
Another way to think about it is this: we are trying to identify
something we can add to (or change about) the task of hitting
such that the player will be unsuccessful using their current
moment solution. This lack of performance success in hitting
provides a stimulus to explore and try something different.
In this case, John used one of my favorite approaches to the
CLA: converging constraints. That is, identifying more than one
type of practice (constraint manipulation) that will produce the
desired outcome. The two main task constraints he identified
were using a longer bat and utilizing the starting position of
having the legs crossed (called “hook ‘ems” because the legs
are hooked over each other). Again, in John’s words:
Hook ‘Em and Long Bat were staples of our training with Lars
Nootbaar. The Hook ‘Em drill cleans up his load/stride, putting
him in a position where his pelvis and torso are coiled evenly.
The long bat forces him to rotate efficiently and hit the ball
flush at a contact point farther in front of home plate.
The reason why I like to use converging constraints like this is
because in my experience it increases the chance the changes
to the movement solution will be sticky (i.e., will still be there
under pressure of the game instead of the performer reverting
back to their old solution).
Encourage (but still guide!) self-organization
. Once we add the
new constraints to practice (i.e. we created a new movement
problem for the performer to solve) we want to let them find
their own unique movement solution to produce the goal
performance outcome. That is, we want to encourage them to
explore. There are a few different ways this can be done. The
first is by choosing constraints that amplify the affordance we
want the athlete to take. Remember, as a coach, we are not
going to deny the fact that we know something about the new
movement solution we would like our athletes to have. We
want to pick constraints that are going to lead the athlete in a
certain direction in their search for a new solution. This is
precisely why John choose the “hook em” constraint. It
amplifies the affordance (aka encourages the athlete to take the
opportunity) to swing from a load with a better coil and larger
hip-shoulder-separation. Notice here, John didn’t tell Nootbar to
swing that way and, of course, the Hook-em position was not
how he actually swung in the end (see Figure 5.2). John let
the constraint do the talking for him. This is the second way
we want to encourage self-organization: use very minimal (if
any) technical instructions or cueing.
Finally, a very effective way to guide/encourage exploration is
by purposely adding some variability to the constraints in the
new movement task. For John, this involved using some
sessions with weighted Axe™ bats. These include an underload
bat that is 20% lighter than an average MLB bat, a barrel
overload bat which is 20% heavier (with the extra weight being
in the barrel) and a handle overload bat which is 20% heavier
(with the extra weight being in the handle). This variability
encourages the performer to explore a larger area in the
movement space in order to adapt to these new constraints.
Provide Feedback. The final step in using the CLA is to provide
clear feedback to the athlete. Remember we are trying to get
the athlete to self-organize into a new movement solution that
will satisfy some performance goal. Therefore, the athlete needs
to know whether or not they are achieving that goal. Not
achieving this goal is one of the things that is going to
incentivize them to try something different! Feedback can come
in three different types:
Knowledge of Process (KP). Feedback about some aspect of the
movement solution that is related to the final performance
outcome but is not about the outcome itself. In our baseball
batting example, process feedback involved using a Blast Motion
Sensor™ to display the bat speed to the hitter after every
swing. John and I talked about the possibility of reducing the
frequency at which we provide his feedback (based on the
“guidance hypothesis” and associated research discussed in
Chapter 3). But after a bit of trial and error, we realized it was
more effective to provide it on every trial. Again, I think this is
the case where the intrinsic feedback about bat speed is not
reliable enough.
Knowledge of Results (KR). Feedback about the actual goal the
performer is trying to achieve (e.g. did the ball go in the hoop
in basketball, did they score in soccer, did they get a hit in
baseball). For the same reasons discussed in Chapter 3 in my
analysis of batting cages, John used a HitTrax™ display to
show ball flight and exit velocity. Again, this is important for the
player to be able to evaluate the effectiveness of any new
movement solution they are trying.
Transition Feedback. The last type of feedback refers to the
change in the coordination pattern that will need to occur in
the future in order for the task goal to be achieved successfully
7
. In other words, instead of giving feedback about current or
past movements, it provides information about the transition
from the current ineffective movement solution to one that will
be effective. Ideally, we want this to be something that the
performer experiences immediately – for example, see my
discussion of using the connection ball in pitching in my first
book. But, I have found that is not always possible. For
training bat speed, John provided transition feedback through
the biomechanical analyses from K-Vest. That is, he
intermittently provided Nootbar with plots like the one shown in
Figure 5.3 so he could get feedback about how his movement
solution was changing. Of course, we have to be very careful
about how and when we present this type of biomechanical
feedback to athletes or we could induce an internal, technical
focus of attention in our athlete.
The Difference Between Self-Regulation & Self Organization &
How They Can Work Together
As a coach, we want to use these different types of feedback
to promote self-regulation in our athletes. Self-regulation refers
to the ability of an athlete to plan, monitor, and evaluate their
practice activities when attempting to achieve a performance
goal8
. These things are clearly not the same as the processes
we referring to when we talk about self-organization of
movement. They refer to setting goals for practice, making sure
the practice plan is stuck to, and evaluation of the practice
outcomes. Not the control of movement. Another way to think
about it is self-regulation is a process that serves to educate
and maintain intention which, of course, acts as a individual
constraint for self-organization.
Summary. The net results of John’s coaching can be seen in
Figure 5.3 (bottom panel). Nootbar increased his hip-to-shoulder
separation from 8 to 24 deg and his average bat speed from
67 to 73 mph. I think this is an excellent example of using the
CLA and illustrates a key point: we are encouraging the
performer to develop an integrated movement solution which
includes intertwined processes of perception, cognition, and
action. Again in John’s words:
Mechanics are complicated, and saying “Lars added +5mph of
bat speed because he gained 10° hip-to-shoulder separation”
wouldn’t be completely accurate. While more hip-to-shoulder
separation was definitely a contributing factor, encouraging him
to swing fast during training and monitoring his bat speed with
a Blast sensor played a huge role in boosting his intent.

Example #2 – “Curing” a Slice in a Golf Swing

The second example I want to look at is a movement solution
problem I get asked about a lot: a slice in a golf swing. A slice
(defined as a shot that curves away from a player’s dominant
hand) hurts performance for a couple of reasons. First, a sliced
shot is hard to keep in the fairway. Second, it does not carry
as far. So, finding a movement solution that makes the ball fly
straighter is likely to improve a player’s score in a round of
golf. We have identified why we want to change the movement
solution (i.e., the problem).
To get more information about the cause of this issue we can
have the golfer hit in front of a Trackman launch monitor.
In most cases, the output will be like what is shown in Figure
5.4. A swing with a slice typically has a relatively large, negative
club path (-11.9) meaning that the golfer has an out-to-in swing
path (right to left for a right-handed golfer). The second issue
is the position of the club face at the point of contact – the
large positive value of “face to path” (8.0) indicates that the
club face is open (facing to the right) at contact. So how can
we apply a CLA to fix these issues with the golfer’s movement
solution (i.e., their swing)?
A screenshot of a phone Description automatically generated
Figure 5.4 – Trackman Output for a Sliced Shot
The first type of constraint that I commonly use is one that
augments the sensory feedback the golfer gets. For example, I
take foot powder spray and apply it to the club face. This
allows the golfer to better pick up information about the
location of ball-club at contact. Many golfers who slice tend to
make contact on the heel of the club instead of in the middle
of the club face. Next, I give the golfer a ball with a pattern
(e.g., a checkboard) on it. This allows them to pick up more
information about the ball’s spin direction as it leaves the club.
With these two constraints in place, I give the golfer a new
task goal: “Swing so the point of contact is in the center of
your club and so that the ball spins back towards you, not to
the right or left”.
The next constraint I like to use is a swing path barrier to
change the club path. As shown in the Trackman data, a
common cause of a slice is an out-to-in swing path which
results in sidespin on the ball as the club cuts across it. To
encourage the golfer to explore a new swing path, I place an
object (e.g. a club head cover) on the outside of the ball
(furthest from the golfer). I then give the instruction: “Swing so
that you don’t hit the object”. In my research, I have shown
that this type of barrier constraint can also be highly effective
for changing the path of a baseball swing9
.
The final manipulation I use is to constrain body posture at the
beginning of the movement, similar to John’s hook-ems for a
baseball swing. Specifically, I give the golfer instructions about
the grip on the club I would like them to adopt. I will instruct
(and demo) that I would like them to alternate between having
a strong, neutral, or weak grip on the club – making 5 swings
and then switching to a different grip. A strong grip is when
the “V” formed by their thumb and index finger is turned
towards their back shoulder. A weak grip is when the “V” is
turned towards their front shoulder. Notice how I am not just
giving them the solution I think will work (e.g., a strong grip)
and having them repeat that over and over. The goal of this
constraint is not to “correct a mechanical flaw” but rather to
allow the athlete to learn more about the relationship between
the action and the performance outcome. That is why I have
them also practice the weak and neutral grip and allow for
some variability. For a newer golfer, I might also do this by
changing the ball flight goal. That is, have them deliberately try
to hit the ball to the right and to the left. There are also some
important differences in my language: instead of “you must” or
“you should” I say “Why don’t you try”. I am using my
knowledge to guide them in a direction as they search through
the solution space, but I want them to experience it for
themselves.
Notice here that I am not saying anything about the mechanics
of the swing (e.g., “keep your lead arm straight” or “rotate
around your back hip more”). These bodily-focused, technical
instructions should be avoided as much as possible for several
reasons: (i) they induce the inferior internal focus of attention
10, (ii) athletes are not very good at making small changes to
their techniques based on this type of instruction10
, and (iii) it
can interfere with their ability to find a movement solution that
works for their own, individual constraints. If a coach does
have a good reason to believe that a technical change like this
might help (e.g., they are not achieving one of the invariants
for a good golf swing), I think there are much better ways to
give this information to the athlete – discussed in Chapter 15.
Example #3- Team Level Constraints, Small-Sided and
Conditioned Games (SSCG) in Soccer
There are several different ways we can use constraints to
address team-level issues. First, let’s look at using a CLA for a
team that is having difficulty in advancing the ball up the field
and is instead just maintaining possession. The SSCG game I
might use is shown in Figure 5.5.
A football field with red and white circles and a ball Description
automatically generated
Figure 5.5 – Small-Sided Game to Improve Advancing the Ball
from The Defensive End
The key constraints manipulations used in this game are:
-A reduction in the number of players. Six defensive, including
the goalkeeper), five offensive, plus 1 neutral player (aka a
floater) that joins the team in possession of the ball. A
reduction in the size of the playing area (using only half the
field). The purpose of these manipulations is to reduce the
overall stability of the system and increase the ability of one (or
a small group of players) to disrupt it by moving into space,
thus amplifying the affordance of advancing the ball.
-I am a big fan of including neutral players in SSCG games
because it amplifies the often dynamic and transient nature of
affordances. For example, an opportunity to score may
suddenly appear when the neutral player joins one of the
teams. It also gives the person playing this position the unique
opportunity to practice transitioning between the offensive and
defensive phases of the game.
-Small goals are placed at center field to give the defending
team something to shoot for. This gives a directionality to their
ball movement (i.e., the affordance to score) so that they don’t
just pass to keep possession. It also encourages players to
provide cover for each other as they attack the ball carrier.
- The final constraints I add are specific rules to amplify the
affordances. For example, giving the offensive team more points
for goals on an intercepted pass to encourage pressing as a
team. Sometimes I will also add rules like a red team cannot
pass to the goalkeeper, effectively splitting the field in half,
leading to further instability.
The second example is an SSCG (shown in Figure 5.6)
designed to improve team defense, especially in outnumbered
situations. The key constraints manipulations are as follows:
A screenshot of a video game Description automatically
generated
Figure 5.6 – SSCG to Improve Team Defense
-Reduction of the number of players (4 on 4, with 2 neutral
players) and the size of the playing area (half the width of a
regulation field with two channels on the side) to reduce system
stability.
-The two neutral players join the team in possession of the ball
but are restricted to only move within their channel at the
edges of the field. This gives the defense more information
about how their positioning influences the outcome of the play.
For example, if a player is attacking from the channel, should I
go into the channel or stay in the central area?
-The presence of the neutral players creates more opportunities
to experience unnumbered attacks that can occur suddenly
from different locations on the field.
-Using rule constraints, we can also amplify specific affordances.
For example, adding a rule that the defense gets 1 point every
time they force the attack into one of the channels or the
offense gets one point every time, they can cause a defensive
player to move into a different channel
The use of the channels here is another good example of the
difference between representativeness and specificity. Of course,
there are going to be no physical lanes you must run in a real
game (i.e., the activity is low on specificity). We are deliberately
moving away from the game to work on specific things, while
still ensuring that we have the key ingredients.

Summary and Other Examples of Constraints Manipulations

Table 5.1 lists and describes all the different principles for
manipulating constraints discussed in this chapter. As I
mentioned, one of the key goals in using the CLA should also
be to identify converging constraints wherever possible so I will
often try to combine more than one of these. In the coming
chapters, we will explore several more examples of the CLA.
We will also see how we can adapt constraints (e.g. change the
rules of a SSCG) based on how the athletes are responding to
our manipulations and will consider ways we can get our
athletes involved in co-designing and co-adapting constraints
along with the coach.
Before looking at some research on the CLA I want to make a
critical point we will dive into detail in Chapter 11. The CLA is
deliberately a fuzzy method of coaching, in that is not meant to
be used in a highly prescriptive (for the coach) and formulated
way where we have a lookup table of which constraint to use
and when. We can never fully predict how our athletes will
respond to any constraint we add ahead of time. The behavior
that emerges will be the result of an interaction between the
other constraints present in the system. For the CLA to be
effective it needs to be an iterative process where the coach
doesn’t just “set it and forget it” but instead is willing to step
in and make adjustments to the constraint manipulations.
Principle
Description
Converging Constraints
Using different constraint manipulations to address (i.e., that
converge on) the same issue
Constraining to Amplify Affordances
Using constraints manipulations to make some affordances
(passing vs holding the ball) more inviting to the performer
than others
Constraining to Augment Sensory Feedback
Adding a constraint that provides the performer with more
process and/or performance feedback (e.g. markers on a golf
ball or baseball that make it easier to perceive spin direction)
Constraining the Task Outcome Goal
Changing the task goal (e.g., the direction the ball should go)
so the performer is required to produce a specific outcome
Constraining to Increase Knowledge of the Solution Space
Add constraints that encourage a wide range of movement
solutions (e.g., performing a golf swing with a range of
different grips) to give an athlete the opportunity to learn the
relationship between their movement pattern and a particular
performance outcome (e.g. a weaker grip tends to make a
golf ball slice)
Constraining Body Starting Posture/Positions
Add a constraint that the athlete must begin a movement
from a particular position (e.g. crossed legs to discourage
over-rotation during the back swing)
Constraining with Equipment
Adding a constraint like a smaller, more bouncy ball or
heavier stick or bat
Scaling equipment
A specific type of equipment constraint manipulation that is
designed to simplify the task for a newer and/or younger
learner (discussed more in Chapter 6).
Allowing the Athlete to Experience the Why
Add a constraint to help the athlete experience the key
invariants of a skill (e.g. why it is more effective to jump off
of one leg instead of the other when doing a layup in
basketball)
Co-Designing, Co-Adapting Constraints
Allowing the athletes to be involved in the process of
designing and choosing constraints (discussed more in
Chapters 13 & 14)
Table 5.1: Example Principles for Designing Constraints

Some Research Demonstrating the Benefits of CLA

Manipulations
In this section, I want to review some of the published studies
that have been done on these different types of constraint
manipulations. This is not meant to be a comprehensive review
but rather to highlight some supporting research. More
examples of CLA research will be discussed in other chapters
where there is a specific fit.
Constraining to Increase Knowledge of the Solution Space
I recently conducted a study that investigated the benefits of a
wide exploration of the solution space (Gray, in press11). In
this study, I used my baseball batting virtual environment (VE)
to compare two different types of training: practicing the “right”
way and practicing the “wrong” way. Traditionally, in sports, we
spend most of the practice time trying to do the “right” things.
That is, producing the desired performance outcomes (e.g.,
hitting a golf shot down the middle of the fairway, shooting a
basketball into the net, hitting a baseball into fair play). An
alternative is to practice doing the “wrong” thing (i.e., an
undesired outcome) sometimes. This could give the performer
more information about the solution space and available
affordances.
In Experiment 1, the “Right” way group always attempted to
“hit the ball hard into fair play” (i.e. the desired outcome) and
were given technical instructions about the swing and corrective
feedback from an experienced coach. This condition was
designed to be highly like traditional baseball batting practice.
The “Wrong” way group received no technical instruction and
had different outcome goals for each session including some
undesired ones: hit the ball as far to the right (or left) as you
can, pop the ball up into the air, and drive the ball into the
ground. In Experiment 2, the “Right” way group did not
receive explicit instructions or corrective feedback. Furthermore,
in this second experiment, the “Wrong” way group only
practiced the undesired goals. All groups trained for 6 weeks.
In pre and post-tests, batters completed simulated games in
which the pitch speed, type, and location were varied, an
umpire called balls and strikes, and the fielders shifted locations
(e.g., they all moved to the left or right side of the field – see
Figure 5.7).

Diagram Description automatically generatedFigure 5.7 –Shift

and away swings11. A: The left-handed batter (i.e., standing
on the right side of the plate) swings at a pitch on the inner
half of the plate (i.e., “inside”), an action that has a higher
probability of hitting the ball in the direction of more fielders.
B: The left-handed batter swings at a pitch on the outer half
of the plate (i.e., “outside”), an action that has a higher
probability of hitting the ball in a direction away from most of
the defenders.

What was found? First, in terms of batting performance, batters

in the “Wrong” way group had a significantly larger increase in
batting average and homerun rate, and a significantly larger
decrease in strikeout % after training, as compared to both the
“Right” way group and a control group (which just did their
regular baseball practice). This was the case in both
experiments. Think about that. Batters in the “Wrong” way
group were better at achieving the desired outcome of baseball
batting (hitting the ball hard into fair play) even though they
practiced producing these outcomes much less or not at all
(Experiment 2).
These effects found here challenge the traditional
conceptualization of skill acquisition in sports where it is
proposed that the goal of practice is to develop memory
structures (e.g., motor programs, motor memory, predictive
Bayesian models) for producing the desired outcome. From this
perspective, the “Wrong” way training used in the present study
would be expected to impair skill acquisition because it should
lead to the development of internal models for actions that
produce undesired outcomes in baseball batting like foul balls,
pop flies, and ground outs. At the very least, it is taking
valuable practice time away from developing models for actions
that produce the desired outcome. Instead, the present findings
are more consistent with the conceptualization of motor skill
acquisition as an adaptive, problem-solving process where the
performer is learning to find different movement solutions (i.e.
move to different areas of the solution space) to achieve their
goal under different constraints12, 13.
Why did these effects occur? To address this, I next looked at
some of the underlying process variables. A couple of main
effects could be seen. First, the “Wrong” Way” group explored
a wider range of movement solutions during training (the range
of horizontal and vertical attack angles of their swing were
significantly larger) and had more functional variability in their
swing in the post-test. An Uncontrolled Manifold Analysis
(explained in Chapter 9 of my first book) revealed they had a
larger increase in “good” variability. Presumably, this helped
them to find a solution in the face of the varying constraints in
the test phase (i.e., achieve “repetition without repetition”).
Finally, the “Wrong” way group showed evidence that they
have learned to better perceive the actions afforded by different
pitch locations. After training (referring to the definition given in
Figure 5.7), batters in the “Wrong” way group made fewer
swings that would likely result in the ball being hit into the
defense (a “Shift” swing) and more swings that would lead to
the ball being hit away from it (an “Away” swing).
Constraining to Augment Sensory Feedback
A good example of this type of constraint can be seen in the
study by Verhoeff and colleagues, which looked at using the
CLA to coach the power clean in weightlifting14. The study
involved case studies of individual athletes with different
technical issues. Each participant was filmed, and a coach
identified undesired movement patterns (i.e. ones that violate
the key invariants of a power clean). For example, some
participants used movements that involved excessive looping of
the bar away from the body during the lift. This is considered
to be maladaptive because it is breaking the kinetic chain
leading to less force production.
Two different constraints were used to address this issue; thus
this study also demonstrates the value of using converging
constraints The first involved placing agility poles in front of the
bar (that the performer would hit if they moved the bar too
far forward). The second constraint manipulation involved
placing chalk on the bar and giving participants the task of
making sure they left chalk on their thighs during the lift. Both
of these constraints provide the lifter with augmented feedback
about the position of the bar during the lift. The training
involved 2 sessions per week for 6 weeks. Using the CLA
approach was found to increase the athlete’s single rep
maximum for the power clean by roughly 15%.
Constraining by Changing Equipment
In a study published by Brocken and colleagues15
, the effect
of equipment modification for training field hockey was
examined. A sample of 129 female, youth field hockey players
from 15 different teams participated. For this study, a crossover
design was used in which athletes participated in both the
control and treatment conditions, with half of the participants
doing the control first then the experimental, and the other half
doing the reverse. The training lasted seven weeks with skills
tests at weeks 1, 4, and 7. Teams in group A completed four
60-minute training sessions using a modified field hockey ball
between weeks 1 and 4 and then switched to training with a
regulation ball between weeks 4 and 7. A regulation field
hockey ball was modified so that it had an uneven mass
distribution. This produced more variable and unpredictable
motion of the ball. Teams in group B did the opposite. The
skills test involved stickhandling the ball through a course of
cones and then taking a shot on the goal with the total time
used as the main dependent variable.
The training involved typical ball control, passing, shooting drills,
and some small-sided gameplay. As the authors note, the main
purpose of the equipment modification here is to increase
movement execution redundancy. That is, to force the athlete to
solve the same movement problems (dribbling through the
same course, passing the same distances, shooting on the same
size goal) by using different movement solutions. Due to the
variability in the ball’s movement, the same pattern of
coordination could not be used on each attempt.
What was found? The results showed clear benefits of using
the modified ball in training. Group A, which used it first,
showed a significant decrease in time to complete the course
between testing periods at week 1 and 4, but not between
weeks 4 and 7 when they used the regulation ball. The
changes in time were 5.5 sec reduction between weeks 1 and 4
and less than a 1 sec reduction between weeks 4 and 7.
Group B, which did the opposite training order, showed the
exact opposite pattern of results. Between weeks 1 and 4
(when they trained with the regulation ball), there was no
significant change in time, while between weeks 4 and 7 their
time decreased by nearly 6 seconds on average.
The authors sum up the results of this study nicely:
The inherently less predictable rolling of the modified hockey
ball in this study presumably creates a constantly changing
challenge, forcing the player to control the action by coupling to
the momentary unfolding information. In addition, the more
erratic rolling of the ball is thought to induce a more active
exploration, creating a larger movement repertoire (i.e.
degeneracy), which allows a learner to achieve the same task
goal in different ways and to better adapt to different situations.
Another example of manipulating equipment as a constraint is
one by Sara Santos and colleagues looking at the effects of
manipulating the type of ball used during small-sided soccer
games16
. Specifically, 12 youth soccer players were asked to
play in 4v4 and 6v6 games under four different ball conditions:
a regulation soccer ball, a smaller handball, a non-spherical
rugby ball, and a condition in which the ball type was changed
every 2 min. The main goal of this study was not to assess
the training benefits of these manipulations but rather to
analyze how they change coordination and movement during
the actual training. Variables analyzed were divided into physical
ones (like total distance covered) and tactical ones (spacing
between players, number of successful passes, etc.).
What was found? Overall, the effects of using different ball
types were larger when there were more players on the field in
the 6v6 side games – presumably because there were more
opportunities for action to be affected. Looking first at creative
and tactical behaviors, as compared to using a regulation ball,
when using the smaller handball there was less exploration of
the space. The authors argue that this occurred because a
smaller ball affords easier passing and shooting opportunities.
Interestingly, the rugby ball had a similar effect for different
reasons. There was again less evidence of exploration of space
and the distance between players was smaller. But, in this case,
it was due to the increase in difficulty of controlling and
passing the ball. Due to its odd shape, players had to focus
more on individual ball control and less on the positions of
teammates and opponents. Results for the mixed condition were
similar to the rugby ball only – as compared to the regulation
ball, there were a lower number of successful actions like
passing and dribbling, lower fluency, lower distance, and higher
irregularity in the distance to the players.
The results of this study are a good example of something I
mentioned earlier: it is difficult to exactly predict what the effect
of particular constraint manipulation will be because constraints
are interactive. It’s fine and good to say “I am going to
challenge the players by throwing in a weird ball”, but as we
can see from this study this could be detrimental depending on
what you are trying to achieve as a coach. If your goal is to
improve team play, then mixing in different balls might be
counterproductive because it makes the athlete focus more on
their own individual dribbling technique rather than picking up
opportunities for action from the environment. Instead, if your
goal is to improve your players’ ball control skills, this type of
ball modification is likely to be more effective. It will also
depend on the skill level of your athletes (e.g. their individual
constraints). Using different balls in team play could be more
effective for more highly skilled players that are better able to
adapt to the less predictable movement. My best advice is:
don’t spend too much try trying to design the perfect CLA
practice on a white board. Try it and see what happens!

Co-designing and Co-adapting Constraints

In a recent study, Domenico et al17 investigated the effects of
CLA on training the countermovement jump (CMJ), which is a
vertical jump performed after loading force in a squat position.
There are two basic variants of the CMJ: bound-arm (where
the arms must remain stationary) and free-arm (where arm
movement is unrestricted). Thirty-six participants were recruited
from a local gym and were split into two training groups.
Participants were asked to perform both variants of the CMJ
before and after 12 weeks of training. One group (Cognitive
training group) focused on the jump technique and leg strength
exercises following a standard weight training progression. The
second group (the CLA) met as a group to discuss different
ways of performing the movement and the exercises to be
used, and their progressions were decided by the coaches and
athletes together.
What was found? For the bound-arm version of the CMJ, both
groups improved their jump height and there were no
significant group differences. For the free-arm version of the
CMJ, the CLA group showed a significantly greater (over 2x)
increase in CMJ after training, as compared to the Cognitive
Training group. The authors proposed that the results for the
two different variants occurred because there are more possible
movement solutions that can be used in the free-arm CMJ.
Next, let’s consider how we can use the CLA to reduce the
complexity and difficulty of a task for a new learner.

MAKING A SKILL EASIER FOR A

NEW LEARNER
D rinking from a fire hose. Jumping in the deep end. There
are lots of different analogies we could use to describe the
challenge of playing a sport for the first time. There seems to
be so much going on and everything is happening so fast! It’s
just too much. To learn the skills necessary to succeed in
sports we need to make things easier and less complex for a
new learner. No one is denying that!
Traditionally, this problem has been conceptualized using
cognitive load theory1
. This theory is based on the classic
information processing model2 of how we take in and use
information, shown in Figure 6.1. Information from our external
environment (which is thought to be impoverished and
insufficient to act) passes through three different types of
memory to enhance it: sensory memory, working memory, and
long-term memory. Sensory (aka iconic) memory filters out
information from our environment (e.g., sights and sounds not
related to the task we are performing) and stores a trace of
task-relevant information just long enough so that it can be
passed into working memory.
Information in working memory is either processed (stored in
long-term memory) or discarded (forgotten). Working memory
is thought to be capacity limited: specifically, it can only hold
7±2 items3
. People with a good memory can hold nine things
in working memory while those with poor memory can only
hold five. That is one of the reasons phone numbers are seven
digits. Processing in working memory involves moving the
information into long-term memory (encoding) using
schema-based memory structures. A schema is a way to
organize information into related classes (e.g. we have a schema
for a “dog” that includes all its relevant characteristics). Once
they are developed, schemas help to reduce the load on the
system by allowing us to store information in chunks. For
example, whereas a novice basketball player must encode the
position of all nine other players on the court, the more
experienced player only needs to hold two things in working
memory: the schema for the offense (“we are in a triangle
offense) and the schema for the defense (“they are playing
man-to-man”)4
. Stated another way, taking in new information
that draws upon your existing knowledge (your schema) serves
to expand the capacity of your working memory. As we will
see in Chapter 8, schemas are thought to exist both for
concepts (e.g., “dog” and animal”) and for actions (e.g., “hitting”
and “throwing”).
A diagram of a working memory Description automatically
generatedFigure 6.1 – Information processing model on which
Cognitive Load Theory is Based.
The other main capacity limitation in this model is our attention.
To move information from working memory into long-term
memory we need to focus our attention on it and rehearse it.
Think, repeating a phone number in your head over and over
so you don’t forget it. With enough practice (and the
development of a more detailed schema) encoding and
retrieving information is thought to become automatic and no
longer requires attention. So, to sum up, based on cognitive
load theory, the problems a performer faces when first learning
a new skill are due to attentional and memory limitations.
Based on this theory, three primary ways can be used to
make a task easier for a new learner. All of these involve some
sort of task decomposition – that is, breaking apart a complex
skill in some way to make it a simpler task with less cognitive
load. The first is task decoupling. The main goal here is to
reduce the attentional and memory demands on the system by
reducing the amount of information. One of the most common
ways this is achieved in sports training is through
perception-action decoupling. We attempt to develop the schema
for processing perceptual information separately from developing
the schema for the motor action. For example, we can take a
skill like dribbling a basketball or soccer ball out of the game
context to make it easier. That way there is less information
coming into sensory memory and fewer things to hold in
working memory. Instead of having to process information
about teammates and opponents moving around a field, we
only need to process information about a handful of stationary
cones. That way, the athlete can focus their limited attentional
resources on developing the schema (aka motor program) for
the action. This decomposition also reduces the cognitive
workload on the performer by removing decision-making (i.e.,
which action schema should I retrieve from memory). Finally,
this makes it easier for the performer to repeat the motor
action (thought to be critical for developing action schema in
the traditional, information processing approach).
We can also use the complimentary approach of attempting to
develop the perceptual schema separately from the action
schema by using one of the many decontextualized
“sports vision” training tools that are on the market now. For
example, computer-based training tools like Neurotracker and
S2 Cognition purport to develop perceptual abilities like
“selective attention” and “perception speed”. Here the cognitive
load is reduced by using simple, out-of-context stimuli and
reducing the required action to a button press (instead of a
more complex movement like a kick or a swing).
The second method we can use to reduce cognitive load for
an athlete is part-task training (aka part practice). In this
approach, we divide an action into distinct phases and focus on
training just one of these phases. Examples include training just
the ball toss in a volleyball or tennis serve, practicing a golf
backswing by pausing at the top of the swing, or training just
the loading phase of a baseball pitching delivery. Another
common example of this is anticipation and decision-making
training using temporal occlusion where the performer only
executes the first part of the action. For example, having a
tennis player watch a video of an opponent serving and then
step in the direction of the serve5
. Here we are attempting to
reduce cognitive load by taking a complex perceptual-motor skill
and reducing it in smaller (discrete) movements.
There are two different ways the skill can be broken into parts,
as illustrated in Figure 6.2. Segmentation separates a skill into
phases that occur in sequence (i.e., are separable in time). For
example, training the backswing phases of a golf swing
separately from the downswing or practicing the ball toss in
tennis serve without hitting the ball. Fractionation involves
separating and training phases of a skill that normally occur
simultaneously when the full action is performed. For example,
training the knee bend in a basketball jump shot or tennis
serve separately from the movements of the upper body.
A drawing of a person playing tennis Description automatically
generated
Figure 6.2 – Task Decoupling and Part Task Training
(Segmentation and Fractionation) in Tennis Training
The final method for making a skill easier based on cognitive
load theory is Scaffolding. Much like scaffolding on a building,
this technique involves building a complex skill by breaking it
into sub-skills. So, instead of breaking a skill apart into
movement phases like in part-task training, we are breaking it
apart at the task level. These sub-skills are then trained
separately which allows the learner to progressively build new
sub-skills on the foundation of one’s already learned. Another
way of thinking of this is reducing the problem space, that is
the gap between where the learner is currently at and the
desired goal, by creating a simpler version of the skill. A
sporting example of this is illustrated in Figure 6.3. Instead of
training the full, complex skill of baseball batting we first train
the sub-skill of swinging the bat fast (aka bat speed). This
might involve a drill where we have a player hit a ball off a
tee or launched by a pitching machine set at a constant speed
and angle. Thus, we simplify the cognitive load required to use
visual information to adjust the trajectory of the bat. Once the
player has established some proficiency in this sub-skill, we can
add the bat-ball sub-skill on top of this. For example, the
player now must swing hard at pitched balls varying in speed
and trajectory (but are still all “strikes”) which requires using
perceptual information to get the bat to the moving ball. We
can then add to this the sub-skill of making swing decisions
(i.e., swinging at pitches that will be in the strike zone vs
inhibiting the swing for ones that are not) and finally ball flight
(e.g., hitting the ball in the air or to a particular side of the
field.).
A person standing on scaffolding Description automatically
generated
Figure 6.3 – Scaffolding Sub-Skills in Baseball Batting
As another example consider using scaffolding to teach someone
how to tackle a ball carrier in rugby
6 or American football.
For this, we could first reduce the task to executing the
sub-skill of the player getting their feet in front of the ball
carrier. Once they can do this, we could next work on getting
their head in the correct (and safe) position to make a tackle.
Finally, we could work on having a strong wrap on the
opponent and leg drive through the tackle. A critical aspect of
this is that we need enough practice (strict repetition) for the
subskill to become automatic before we move on to the next
sub-skill. Scaffolding in this manner is meant to reduce the
cognitive load by allowing a player to focus on only one or two
movements (in each sub-skill) rather than the whole thing at
once.
Problems with the Cognitive Load Theory Approach

The problem with all these methods of making a skill easier is

that they rely on the assumption that the perceptual, cognitive,
and motor processes during the execution of a sports skill
interact in a manner consistent with a linear system. Such
systems have the key feature of superimposition. Simply put,
this means that we don’t expect there to be any significant
interactions between the inputs to the system. Thus, we can
just break the skill into parts and then add the parts back
together. The act of hitting a baseball is equal to sum of the
separable components of perceiving the ball flight and swinging
the bat (decoupling). The movements involved in a full tennis
serve are equal to the movements in the phase of the ball
tossing plus the phase of the forward racquet swing (part-task
training). Tackling an opponent in rugby is equal to the
sub-skill of getting your body in front of the ball carrier plus
the sub-skill of wrapping them in a tackle (scaffolding). The
whole is equal to the sum of the parts. Therefore, we can
reduce the cognitive load of learning a new skill by breaking it
apart in some way and then just piecing it back together after
each part is learned.
This, of course, goes against one of the fundamental
assumptions of the ecological approach to skill – that human
beings are complex, non-deterministic systems in which the
outputs (plural!) are not predictable from the inputs.
Fundamentally, this is the acceptance that there are interactions
between the inputs to the system. Perception and action are
reciprocally coupled. As I move (e.g., drive my body forward or
turn my head) during a baseball swing the information available
from the ball flight changes. The information that I pick up
from the ball flight is continuously used to alter the swing (i.e.
prospective control). NOT perceive then ACT (i.e., separable
processes represented by boxes is a flow diagram),
Perception-Action, inseparable, coupled, interacting processes.
Similarly, how I toss a ball during a tennis serve depends on
the movement solutions available for the racquet swing. My
movements during the forward swing of the racquet depend on
the trajectory of the ball during the toss (which varies
significantly from serve to serve7
). The phases (and sub-skills)
all interact with each other in a non-linear manner. The
presence of such interactions also shoots down the principle of
superimposition – the whole is not equal to the sum of the
parts. We can’t decompose skills into parts and just put them
back together because the parts interact with each other. The
output is not just the linear sum of the inputs.
These limitations have been largely born out in the findings
from research examining these different approaches to reducing
cognitive load. As predicted by ecological dynamics, this work
has shown that the characteristics of the part are different
when it is performed separately as compared to when the part
is embedded in the full skill. For example, as discussed in
Chapter 11, of my first book, when we decouple
perception-action in soccer goaltenders we get completely
different gaze behaviors
8
. Similar effects have been found in
baseball batting when decoupling perception-action by having the
batter hit off a tee9
. Finally, decoupling perception -action has
been shown to produce different forces and muscle activation
during sidestep and crosscut maneuvers in Australian Rules
Football10. Further evidence in support of this can be seen by
the fact that there are larger differences between elite and
lesser-skilled performers for a coupled task as compared to a
decoupled one
11,12.
Research comparing part-task training has found similar effects.
Wickens and colleagues13 conducted a meta-analysis of the use
or part-task training. In this analysis, 37 studies in which
transfer of training was measured were included. Overall, there
was a negative cost of using part-task training with on average
the part-task groups doing about 20% worse after training, as
compared to whole task control groups. Using segmentation to
create the part task resulted in significantly less cost as
compared to using fractionation. However, and returning to the
issues discussed in Chapter 3, the tasks included in this review
were almost all simple motor tasks. For complex tasks
(requiring the coordination of multiple degrees of freedom), all
the ways we can split or subdivide a skill will involve some sort
of fractionation. Due to the non-linear nature of skill,
segmentation is not possible.
This can be seen when we look at research on part task
training for sports skills. Figure 6.4 shows the ball toss in a
volleyball serve when performed in isolation (left panel) and
when it is performed as part of the full action of serving the
ball (right panel) 14. The toss is significantly lower and less
variable in the part task, suggesting that the player will have to
re-learn this phase of the skill in whole-task training after it has
been developed (and automatized) through part-task training. A
similar result was found by Reid and colleagues when
examining part task training of the tennis serve 15
. Vertical
displacement of the ball at its zenith increased significantly
during part-task training as compared with the whole task
performance. Furthermore, the temporal associations between
racket and ball motion seen during the execution of the full
skill were weakened during part-task training.
A graph of training and ball height Description automatically
generated
Figure 6.4- Ball trajectory during the toss phase of a volleyball
serve when performed under part-task (PT) and whole-task
(WT) conditions. Based on results from Davids et al.
14
Finally, research examining the use of scaffolding to learn the
different sub-skills in hitting a baseball or cricket ball (by using
a pitching machine instead of a real pitcher/bowler) has shown
similar effects. Kuo and colleagues16 found that, in comparison
to hitting off a real pitcher, batters lift their lead foot off the
ground earlier and shift their body weight more slowly when
hitting a ball projected by a machine as compared to hitting of
a pitcher (see also
17
). Similarly, in cricket batting, Pinder and
colleagues18 found an earlier initiation of the backswing, front
foot movement, downswing, and front foot placement when
facing a real bowler compared to the bowling machine. The
peak height of the backswing was higher when facing the
bowler, along with a significantly larger step length. See also
19.
An interesting recent study examined the effects of scaffolding
on learning to ride a bicycle20
. Traditionally, bike riding is
taught in a scaffolded manner using training wheels. The
sub-skills of moving forward and steering are separated from
the sub-skill of balancing the bike. An alternative is to use a
balance bike. As shown in Figure 6.5, a balance bike does not
have pedals, or training wheels, and is scaled so that the
children's feet can touch the ground from the sitting position.
This allows them to use their feet to propel themselves forward
and maneuver the bike. In contrast to the bike with training
wheels, the balance bike allows children to place their feet on
the ground whenever they feel instability (a constraint designed
to remove the rate limiter associated with the fear of falling).
Another advantage is having no pedals to get in the way when
learners use their feet to facilitate motion. Simultaneously, the
balance-bike design allows the child to couple their postural
regulation actions with the bicycle movements: the child learns
to maintain balance on the bike from the first moment. We do
not break the skill into separate parts of balancing and
propelling the bike forward.
The study by Merce et al, assessed the transfer of training to
riding a regular bicycle. Training on a bicycle with training
wheels led to a success rate of 75% for the achievement of
independent cycling on a conventional bicycle, as compared to a
100% success in the balance bike group. The balance bike
participants were also significantly faster in learning to
self-launch, ride, brake, and cycle independently, as compared
to participants in the training wheels group.
A child riding a bike on a road Description automatically
generated
A child riding a bike on a path Description automatically
generated
Figure 6.5– A bicycle with training wheels vs a balance bike
The issue here is that task decomposition (i.e., breaking a skill
into parts in some way) not only simplifies it and reduces the
cognitive load, but it also fundamentally makes it a different skill
. This is likely to greatly reduce positive (or even produce
negative) transfer of training effects.
Understanding the Challenges of Learning a New Skill from an
Ecological Perspective
From an ecological approach, there are three main challenges
that a new learner faces. (Note, I am giving these an order
here but they all must be learned at the same time).
1)Educating Attention to Specifying Information.
The first issue involves learning to use specifying information
sources that will allow a task to be performed successfully
across a wide range of conditions. This requires becoming
attuned to higher-order information sources in the perceptual
array. This is a process called education of attention in Direct
Learning21 (discussed in Chapter 12 of my first book). For
example, in hitting an approaching object we must learn to
move away from swinging when the ball is a certain distance
from us (a lower order variable) and switch to controlling our
movements using one that specifies the time to contact (i.e.
tau).
By higher order, we are not referring to variables that are
constructed in our brain by combining two or more lower-order
variables. This construction only occurs at the level of us
describing the information sources. As proposed by Gibson, we
pick up these information sources directly. We don’t compute
them from lower-order sources. I tested this idea in one of the
first experiments I ever ran as a graduate student
22. In this
study, we asked participants to judge the time to contact for
objects approaching from two different distances (1m and 5m).
Two objects were “thrown” at their head, and they had to say
which would have arrived sooner. There are two different ways
we could make this judgment. First, we could construct the
higher-order variable (time to contact, TTC) by first estimating
the lower-order variables of speed and distance (like you were
taught in math class). Second, we could perceive TTC directly
by picking up tau. In the study, we separated these two
options by using lenses to magnify the object at the far
distance by a factor of 5. Speed was held constant. Thus, the
lower order variable of the perceived distance of the two
objects was the same, while the higher order variable (tau) was
different. Consistent with the idea that we don’t construct tau.,
time contact judgments were different for the object projected
from 5m and the one projected from 1m, even though all
participants reported that they appeared to be at the same
distance.
If we use the task decomposition methods discussed earlier in
this chapter, we can exacerbate this problem of education of
attention by creating visual motor mappings. These are
relationships between information and movement that will allow
the skill to be performed successfully in the over-simplified
practice conditions but not under a different set of task
constraints. An example of this can be seen in the typical
approach we use when we teach someone to catch or hit a
ball. We often start by doing soft underhand tosses with little
variation in ball speed. This creates a visual-motor mapping
between the ball’s angular size (or distance) and its time to
contact. Thus, the new learner can use the strategy of
controlling the closure of their hand or glove based on the
ball’s size (distance). But as soon as we move to more realistic
conditions, with varying ball speeds, we break this visual-motor
mapping and this control strategy no longer works.
Instead, we want to educate the learner’s attention to
information that is specific to the aspect of the event they are
trying to control. For catching, the specifying variable is tau.
Tau species when an object will hit the glove (and it needs to
be closed) no matter what the object’s distance, speed, size,
shape, etc. Stated another way, a tau value of 2.5 is specific to
the event that an object will arrive in 2.5 seconds. If we
encourage a learner to use a non-specifying information source
(e.g., by creating a visual-motor mapping through task
decomposition) they will again have to unlearn this before
moving to use the specifying variable.

2) Solving the degrees of freedom problem.

As first identified in the seminal work of Bernstein23, another
challenge for a new learner is solving the degrees of freedom
problem. This can be stated as: Among the almost infinite
possible ways (e.g., combinations of joint angles, forces, speeds,
etc.) a task can be performed, how does the performer choose
one? To illustrate this, let’s consider the example of learning to
do a power-clean movement in weightlifting, illustrated in Figure
6.6. How much should we bend our knees to start the lift?
When should we start to move the weight up to our
shoulders? Should we throw the weight out from our body to
generate momentum? These are just a few of the many control
problems a new learner must solve. As a coach, we need to
somehow reduce the size of the possible movement solution
space.
A person lifting weights in different poses Description
automatically generatedFigure 6.6 – Degrees of Freedom (DF)
Problem in performing a power clean
Again, we could do this through task decomposition –
separating the skill into distinct sub-phases or sub-skills and
training those separately. Or by giving the learner the solution
by prescribing a particular technique through verbal instruction.
But, another way of helping the learner to solve the degrees of
freedom problem is through the use of the CLA. Remember we
call them constraints because they take away movement
solutions – thus making the degrees of freedom problem easier
to solve. An example of this can be seen in the weightlifting
study by Verhoeff et al. discussed last chapter. There are only
certain movement solutions in the power clean that will make is
so that the bar touches my thighs (and leaves chalk on them).
Quoting a study by Moy et al.24 I will discuss shortly,
Constraints channel an individual’s exploration within a narrower
area (limited number of movement solutions) of the practice
environment toward desired functional movement solutions
3)Calibrating & Adapting to Individual Constraints
Finally, a new learner needs to adapt their movement solution
to be effective given their individual constraints. This involves
developing their affordance perception – what actions are
possible for me? Where are my action boundaries?
Task Simplification- The Ecological Way to Make the Task
Easier
In the Ecological Approach, we are still faced with the same
problem introduced at the start of this chapter – most sports
skills are far too complex for a learner to start performing the
full task right away. We still need some way to make it easier
for them. But, in the ecological approach, we are not going to
achieve this by breaking things apart. Instead of using task
decomposition, we are going to use task simplification. To
illustrate the difference between the two, consider the difference
between scaling vs cropping an image (illustrated Figure 6.7).
Imagine that you insert an image into a document, and the
image is too large. You could make it smaller by cropping
(pulling on only one side) or scaling (pulling on all sides at the
same time). When we crop, we distort the image. We break it.
We lose part of it. The same thing happens when we
decompose a skill. And continuing with the analogy, once we
crop something we can’t get the original, full image back (there
is no un-crop function – go ahead, try it on Figure 6.7!).
In the ecological approach, we want to make the image smaller
(the skill is easier and more manageable for the learner) by
scaling. We want to make it easier while keeping it whole. Then
we can just scale it back to the full thing. Stated another way:
we should make the task easier while still meeting the key
criteria of representativeness we discussed in Chapter 4. Let’s
look at some examples of how this can be achieved using the
CLA.
A diagram of a complex Description automatically generated
with medium confidence
Figure 6.6 -Cropping (left) vs Scaling (right) an Image
Examples of Task Simplification
1)
Shifting Action Boundaries by Equipment Scaling
One of the main challenges that can occur for a new learner
(especially one who is younger and less physically developed) is
that the movement solution needed to produce a basic level of
proficiency is beyond their action boundary. Two common
examples are a child trying to shoot a regulation 22oz
basketball at a regulation 10-foot hoop and a child trying to hit
a regulation 10 oz volleyball over at regulation 7’ 11 inch volley
ball net. In both cases, the desired movement solution we are
trying to teach the player (i.e., a jump shot, spiking the ball) is
not possible to execute given the individual constraints of the
child.
As I discussed in detail in my first book one effective way to
combat this is through equipment scaling. E.g., reducing the
height of a basketball rim or volleyball net, using lighter or
lower compression balls, etc. Research examining equipment
scaling has shown that this results in faster and more effective
skill development as compared to comparable task
decomposition methods (e.g., hitting a volleyball against a wall
instead of over the net)25
.
Equipment scaling should not only be thought of to address
limitations related to physical dimensions (e.g., height and arm
length) but also those related to action capacity in older
athletes.
2)
Removing Rate Limiters by Changing Rules
A rate limiter is an individual constraint that limits the rate at
which a performer acquires and develops a movement skill
through exploration of the solutions space. For example, in
learning to ride a bicycle, fear of falling is something that limits
the ability of a new learner to explore different movement
solutions. Using a balance bike is, of course, an example of
changing the equipment constraints to reduce this.
In a recent study by Moy et al.
24, the authors proposed that
the rule which allows only one serve in volleyball (in contrast,
to tennis, where two serves are allowed) may be such a rate
limiter. The rules that permit only a single serve may be
actively discouraging the performer from exploring their action
capabilities and taking risks to search for alternative movement
solutions due to a fear of making a mistake and forfeiting a
point. In the study, 40 male high students competed in two
different training games: a regulation (single serve) game and a
modified game that allowed for two serves. The order of
gameplay in this within-subject design was counterbalanced. The
main dependent variables were: serve speed, accuracy, and type
(overhand vs underhand).
What was found? In the two-serve game, participants had
significantly higher serve speed and less accuracy on the first
serve. As compared to the one-serve game, they also attempted
significantly more jump serves. Thus, the task constraint of
adding the two-serve rule seemed to successfully encourage
more exploration of the solution space. It will be important for
future research to investigate transfer from practicing under the
two-serve rule to playing the regulation one-serve game and
run a longer study to see if participants could eventually find a
movement solution with higher accuracy under the two-serve
constraints.
This is an excellent example of how we can use the CLA to
make a skill easier for a new learner without decomposing it
into isolated drills like serving a ball against a wall or practicing
only the ball toss.
3)
Changing the Required Task Precision
Another way that we can simplify a skill without breaking apart
is by changing the movement precision required for a
successful performance outcome. A famous example of this
approach is “Tiger Par” – the system Tiger Wood’s father Earl
Woods devised (in conjunction with Tiger’s first coach Rudy
Duran)26. Instead of doing lots of isolated drills focusing on
swing mechanics before letting him play golf (i.e. teaching “the
fundamentals”), Earl made golf easier without breaking it apart.
Specifically, he adjusted the number of shots required to make
par on the course based on Tiger’s individual constraints. For
example, for a course that was par 54 (designed for an
average adult amateur who can hit the ball 225 yards), the par
was increased to 67 for Tiger, who at the time could only hit
the ball 180 yards. As Tiger grew and began to hit the ball
further, “Tiger par “was adjusted downwards.
4)
Constraining the Opponent
As we saw in Chapter 4, it is critical in MMA training that we
learn to nest affordances. For example, we need to be able to
throw a punch or strike our opponent in a way that does not
leave us completely open to a counterattack. Typical
decomposed drills where we practice “on air” or against a
heavy bag do not allow for this nesting to occur. Nor does it
allow coupling our movements (e.g. a particular striking or
kicking movement) to the information we will use to control
these actions in competition (e.g. an opponent holding their
hands low or standing a certain distance away). How then can
we make MMA fighting easier for a new learner, especially
when the person they are matched with is of a much higher
skill level?
One effective way that we can simply (not decompose) the task
is by constraining their opponent in how they can defend or
attack. For example, we could add a rule that the opponent
can only defend with their arms and not use their legs. In
practice, I have found that when doing this it's better to
constrain to a small number of options, rather than
constraining the opponent to do only one solution. This
prevents the interaction from becoming too predictable for the
new learner – remember some degree of unpredictability (and
thus, emergent decision-making) is an essential feature in
representative learning design.
A similar approach can be seen in the volleyball study by
Caldeira et al., discussed above, in which three different options
for attacking in volleyball were interleaved.
5)
Scaling Variability Appropriately
If you have reached this point, hopefully, I have convinced you
of the value of adding variability to practice! I like to say this is
the “lowest hanging fruit” in improving practice design. But
critically, the amount of variability we add needs to be
appropriate for the individual athlete we are working with. This
can be conceptualized using the Challenge Point Hypothesis if
we consider practice variability as a source of task difficulty –
adding variability increases difficulty. We expect the U-shaped
relationships shown in Figure 6.7. The optimal challenge point
(CP) will require a higher level of variability for an expert as
compared to a novice
A diagram of a performance diagram Description automatically
generated with medium confidenceFigure 6.7: Optimal Challenge
Point (CP) as a Function the Amount of Variability for Different
Skill Levels.
The relationship between variability and learning was
investigated in a recent study by Caballero et al.
27 Groups of
22 novice participants were trained in a ball-throwing task
under three different levels of task variability. The low variability
group was trained to throw to targets that were dispersed in a
radius (from trial to trial) of 0-20 cm. The medium variability
group practiced with a dispersion from 0- 40 cm and the high
variability group practiced under a dispersion of 0-60cm. These
three groups were compared to a control group that always
threw the ball to the same target location. The different groups
are represented visually in Figure 6.8.
A group of numbers on a white background Description
automatically generated
Figure 6.8 – Dispersion of throwing targets for the control (A)
and three different variability groups.
What was found? All three variability groups improved more
after training than the control group. It’s the lowest-hanging
fruit for improving most practice! There were no significant
differences between the three levels of variability. This is
perhaps not surprising because the pre and post-test involved
throwing a ball toward a target in a constant location (e.g., A
in Figure 6.7). We would expect high variability in practice to
better promote dexterity (i.e., adaptability of the movement
solution for changing constraints) which was not assessed by
these tests.
But what I want to focus on is an analysis looking at the
relationship between the performer's inherent variability and the
training benefits. Inherent variability was quantified by measuring
something called a Lag-1 autocorrelation. This is a measure of
the similarity between consecutive throws that was assessed in
the pre-test. Another way to think of it is the amount of
variability the performer brings to training. A low Lag-1
correlation means consecutive throws go to different locations –
what we might expect from a new learner (i,e., inconsistent
performance outcomes). This analysis revealed that, for
participants in the low variability training group, there was a
significant correlation between the amount of intrinsic variability
and the amount of pre-post training improvement. This provides
some evidence for the idea that performers with more inherent
variability need less variability added through practice
manipulations.
A method for finding the optimal level of variability for different
performers will be discussed next chapter.
6)
Starting With a Variant of the Skill That Has Less Degrees
of Freedom
Another way we can make a task easier is by reducing the
degrees of freedom involved but we have to be careful not to
reduce them so much that it creates a new task. An example
of this can again be seen in the way Earl Woods instructed
Tiger. He used the approach of starting him on the green and
then progressively working back to the tee (i.e., the opposite of
what we normally done in golf). That is, he started with putting
(a task that has a lower number of degrees of freedom) and
finished with tee shots (more degrees of freedom). The driver
was the last club that Tiger learned to hit.
7) Changing Individual Constraints – Increasing Action Capacity
A final way we can make a task easier for a new learner is
by increasing their action capacities (i.e., moving their action
boundaries). Giving an athlete more action capacity gives them
the potential to be more skillful by increasing the number of
available movement solutions. Examples include increasing the
maximum jump height for a basketball player and increasing
the maximum bat speed for a baseball player. See Chapter 6 of
my second book for a discussion of how to best design action
capacity training.
Oh Fundamentals, Where Hast Thou Ought Gone?
One sticking point I frequently have with coaches about
adopting an ecological approach is: “What about the
fundamentals?”. The idea that we can let kids start playing
basketball before we teach them the “proper” way to shoot or
dribble a ball is a hard sell! I don’t think we need to
completely abandon the idea of “the fundamentals” we just
need to change what we mean when we use that term.
Instead of focusing on technique (i.e., the specifics of the
movement solution), we should be focusing on what is required
in the movement outcome. Specifically, ask what are the key
features of successful movement outcomes for my skill? In
other words, what are the key invariants? That is, what must
be there (i.e., cannot vary) for an execution to be successful? If
we think this way, we see that “the fundamentals” are not
defined at the level of the technique (e.g. your elbow must be
90 deg) but rather at the level of movement outcome (e.g. the
ball must align with the target).
Berstein’s way of conceptualizing this was to break movement
variability into two parts: elemental variables and essential
variables. An example, for baseball pitching, is shown in Figure
6.9. This figure shows the relationship between different types
of variability. At the far right, we have variability in the
performance outcome. In almost all cases, we want this to be
low i.e., we want to consistently hit the target we are aiming at
whether it’s the catcher’s mitt in baseball, the bullseye in darts,
or the center of the hoop in basketball. But how is this
achieved?
A diagram of a person's hand and a person's hand Description
automatically generated
Figure 6.9 – Elemental and Essential Variables in Movement.
An elemental variable is an aspect of the movement that is free
to vary from execution to execution to allow for functional
variability. In baseball pitching this includes things like the
maximum internal shoulder rotation, elbow flexion, and lead
knee angle. Essential variables are aspects of movement that
must be invariant to achieve low variability in the performance
outcome. For pitching, this is the final position of the hand
when the ball is released. So, in other words, we can use lots
of different combinations of elemental variables as long as they
keep the variability of the essential variable (release point) low.
Evidence for this can be seen in a couple of studies of baseball
pitching that used the permutation method. In this method,
multiple joint angles from pitching delivery are fed into a
forward biomechanical model that predicts the resultant position
of the hand at ball release and/or final ball position. The critical
manipulation involves comparing the variability in position
predicted by the model under the two different conditions
illustrated in Figure 6.10: (1) when the joint angles used in the
model come from the same delivery (SAME) and (2) when the
joint angles come from different deliveries (DIFFERENT). The
first of these studies, discussed in detail in my second book, is
one by Matsuo and colleagues28. To briefly review, in this
study, joint angles (including the elbow, shoulder, pelvis, knee,
and ankle) were measured for 12 semi-professional pitchers
asked to throw 10 maximum velocity fastballs. They found that
the variability in the angle of the hand at the point of release
was nine times higher for the DIFFERENT conditions, as
compared to the SAME conditions. This suggests that there is
considerable variability in these joint angles (i.e., the elemental
variables) from pitch to pitch (or else the results would have
been the same for the SAME and DIFFERENT conditions) and
the joint angles are forming a motor synergy (discussed more
next chapter) in order to stabilize the essential variable of final
hand position.
A picture containing timeline Description automatically generated
Figure 6.10 – The Permutation Method. Comparing variability
produced when we take joint angles from the SAME vs
DIFFERENT deliveries

The second study, by Kusafuka et al.

29, again used a
permutation method except the variables entered into the model
were the release parameters (e.g. the horizontal and vertical
positions of the hand, initial ball speed, elevation angle and spin
axes of the ball at release, etc.). These were then used to
predict the final position of the ball when it crossed the plate
(i.e. the performance outcome in Figure 6.9). Data was taken
from 12 former professional pitchers throwing fastballs. The
main comparison again was between SAME (all release
parameters taken from same delivery) and DIFFERENT (release
parameters taken from different deliveries) conditions. What was
found? There was no significant difference in outcome variability
between the SAME and DIFFERENT conditions. This occurred
because these Essential variables (i.e. the ball release) did not
change significantly from pitch to pitch (i.e. they are invariant).
So, to sum up, if we want to teach “the fundamentals” of a
skill we need to make sure we are focusing on the correct
thing. We want to make sure that our athletes develop
Bernstein’s essential variables (i.e., the invariants) and not get
lost in the weeds of the specifics of the technique, the elemental
variables (i.e., the variants, which can and need to vary from
execution to execution). I will look at the concept of invariants
and how a coach can identify them in more detail in Chapter
8.
Do We Have to Prescribe “the Fundamentals” to an Athlete?
Another sticking point for a lot of coaches is moving past the
idea that we need to prescribe the fundamentals to an athlete.
Even if we change our view from focusing on Elemental
variables to focusing on Essential Variables, many coaches feel
like they still need to give a new athlete one of possible
movement solution (e.g. one set of Elemental Variables that they
know works) or they will never find it on their own. A growing
body of evidence suggests that is simply not the case.
First off, we have the study discussed in Chapter 3 by Deuker
et al. 30 which compared prescriptive instruction of passing,
shooting and dribbling skills in soccer to a CLA approach that
focused on playing small-sided games. The key question
addressed was not: would the CLA lead to more dexterous,
adaptable players who are better decision-makers in the game
(the main proposed benefit of using the CLA)? Rather, it was
the “fundamentals” question: would both groups perform equally
well on out-of-context tests of technical skill (e.g. dribbling
around cones). In other words, tests were biased towards the
prescriptive training. Both groups showed similar (and not
different statistically) improvements after training. Stated another
way, the CLA group learned the “fundamentals” of soccer (i.e.,
found a functional set of essential variables) equally well as a
group that had the fundamentals (i.e. one specific solution)
prescribed to them.
A similar result was found in another soccer study be Esposito
and colleagues31. In this study, 30 players were randomly
assigned to either a CLA group or a prescriptive training
group. Before and after training, participants completed the
Loughborough Soccer Passing Test (LSPT)
32 . A video of this
test can be seen here
33
. The test involves players responding
to verbal commands to pass the ball in a circuit of cones
within a grid. So, again, the study evaluated how well an
ecological approach prepares athletes for decomposed,
out-of-context assessment of technical skill. Again, there were no
significant group differences in pre-post changes in performance
on the LSPT.
Finally, Lindsay et al34 compared prescriptive, linear training
with a nonlinear (CLA) approach to coaching the power clean
in weightlifting. Sixteen novice weightlifters were randomly
assigned to either a linear or nonlinear (CLA) training condition,
both of which lasted for four weeks. Both types of training
were designed to address a common flaw seen with novices –
looping or throwing the bar away from the body to get it up
in the air.
For the linear group, the training involved explicit instructions,
repetitious practice, and providing feedback to correct errors to
direct learners toward an ‘ideal’ technique. This was
decomposed into different phases for the action. For example,
instructions for the Transition phase were: “When the bar gets
just above the knees, forcefully push the hips forward with a
slight bend in the knees. Shift your body weight onto the
middle of the foot, keep your heels on the floor”.
For the nonlinear group, a CLA approach like what was used
in the Verhoff study (described in Chapter 5 above) was
employed. This included putting chalk on the barbell and
placing poles in front of the participants. The lifters were then
given the instructions to either “lift so thar bar leaves a chalk
mark on your thighs” or “lift to not hit the poles”.
This is a perfect example of changing from technical
fundamentals toward a focus on movement outcome invariants.
In the nonlinear group, instruction is not directed at the specific
technique (what the ‘hips’, “knees’ and “feet’ are doing) but
rather the essential variables in the movement outcome (what
the bar is doing).
The main dependent variable was the forward displacement of
the bar during the lifts. A cluster analysis was also performed
on the kinematics data to classify how many different
movement solutions were used during training. What was
found? Both types of training were equally successful at
reducing the forward displacement of the bar as there were no
significant group differences. The cluster analysis revealed that
the number of distinct movement patterns (roughly 7 on
average) explored in training was not significantly different for
the two groups either. Thus, it seems as though participants in
the linear group did not actually stick with the one solution
they were prescribed. But, the key point here is again, that in
the absence of prescriptive instructions, well-designed constraints
can serve to guide a learner to find “the fundamentals” and
away from flawed movement solutions on their own without
getting lost in a sea of self-organization!

7
COACHING TO THE INDIVIDUAL
ATHLETE & WORKING WITH
GROUPS
I n this chapter, I want to explore ways in which a coach can
balance two competing environmental constraints: the need to
individualize practice and the constraint of having to run
practices with groups of players of different skill levels.
Individualizing coaching is one of the central tenets of the
ecological approach to skill. We need to consider an individual
athlete’s organismic constraints and intrinsic dynamics. We need
to create affordances in the practice environment that depend
on individual effectivities (body dimensions and action capacities).
A gap between two opponents or a ball moving at a certain
speed does not offer the same opportunity for action to every
player. And, finally, we need to challenge the athlete at the
right level so that they get information for learning. In the first
part of this chapter, I want to explore ways in which we can
most effectively tailor our practice design to individual athletes.
In the second part of the chapter, we will explore ways to
manage practice with groups of athletes.

Individual Differences – The Inconvenient Truth of Skill

Acquisition
As we already saw in Chapter 3, the predominant research
methods used in the field of skill acquisition have essentially
been designed to minimize (ignore) the individual differences in
how people acquire a new motor skill. This flies in the face of
the common observation from coaches that there are “fast
learners” and “slow learners”, athletes are more and less
“coachable”, etc. In their recent review of how skill acquisition
research is being applied, Williams and Hodges1 identified:
“Consider individual differences in how learners respond to
different interventions “ as one of their key action points. To
quote the authors:
Paradoxically, while researchers have become proficient at
controlling everything to examine how generally a single factor
impacts on performance and learning, we have largely turned a
blind eye to individual differences that exist between learners.
The importance of individualizing practicing is one of the key
messages that I take from Anders Ericsson’s seminal work on
deliberate practice2
. Figure 7.1 represents some of the findings
Ericsson observed across a wide range of sports including
basketball, soccer, ice hockey, and baseball. As we continue to
practice a sport as a young athlete there seems to be a
crossroads that occurs depending on the extent to which we
engage in maintenance practice (i.e., practicing those things we
are already good at, which is generally more fun) and
deliberate practice (i.e., practice designed to specifically focus on
our weaknesses, which is typically less enjoyable). This is also
the point at which we see a separation between athletes who
will go on to be elite and those who will be near elite. Since
each athlete is going to have their own individual strengths and
weaknesses, deliberate practice is synonymous with individualized
practice.
A graph of a line Description automatically generated with
medium confidenceA diagram of a course of practice
Description automatically generated
Figure 7.1 – Types of Practice and Their Relationship to Skill
Development. Adapted from Ericsson & Pool (2017)2
.
So how can we best put individualized, deliberate conditions
into our practice?
Methods for Individualizing Practice
There are three methods that I like to use to manipulate
constraints to achieve deliberate practice for different athletes. Of
course, to work on an athlete’s weakness we need to know
what these are first! This is best done as a two-stage process.
The first stage involves evaluating an athlete’s relevant action
capacities and individual constraints to identify their key rate
limiters. As discussed in Chapter 7, a rate limiter is an
individual constraint that limits the rate at which a performer
acquires and develops a movement skill – it’s a like a
handbrake on skill development.
A good example of this can be seen in the case of baseball
player Lars Nootbar discussed in Chapter 4. After an initial
evaluation, John identified bat speed as the biggest rate limiter
for his hitting performance, with the ability to load effectively in
the backswing being the specific cause of the lack of bat speed.
Another way to think of these rate limiters is that they are the
“big rocks” in skill development that we need to move first!
Working on Lars’ ability to hit the specific types of pitches (e.g.
a curveball vs slider), pitches in certain locations (inside vs
outside) or to hit balls to different locations on the field are all
going to have a relatively minor effect on performance if he
doesn’t hit the ball very hard. Figure 7.2 shows a pattern of
results that I have observed when studying the gaze behavior
of baseball batters. For the three players shown, improving the
final fixation error (e.g. the separation between your eyes and
the ball at the point of contact) resulted in them hitting the ball
harder (what we call “game power”). But their bat speed (what
we call “raw power)” set the ceiling on these gains.
A diagram of a game error Description automatically generated
Figure 7.2 – Rate Limiters and Baseball Batting
Rate limiters like this can be addressed through a combination
of training designed to improve action capacity (e.g. improve
rotational velocity in the weight room) and by varying the task
constraints (e.g. the service rule in the volleyball study discussed
above) that help remove some of the psychological rate limiters
(e.g., fear of failure).
After identifying and addressing any rate limiters, I next look at
how an athlete’s performance varies as a function of the key
task constraints for their sport (e.g. distance, speed, time of
game, etc.). This can be derived either from game/competition
statistics or by using something I like to call a diagnostic
constraint. A diagnostic constraint is something you manipulate
as a coach to improve your understanding of your athlete, as
opposed to something you manipulate to try and make them
more skillful. Figure 7.3 illustrates two hypothetical examples of
data that we might get from this process. The left panel shows
a basketball player’s shooting percentage from different locations
on the court. The right panel shows a baseball player’s hitting
performance (wOBA) for different pitch locations. In both cases,
we can see that the players have clear weaknesses we want to
target in practice.
A diagram of a basketball court Description automatically
generated
A screenshot of a computer Description automatically
generated
Figure 7.2 – Left: A Hypothetical Basketball Percentage Shot
chart. Right: Hypothetical Batting Zone Chart.
Armed with this knowledge, let’s look at the ways we can be
deliberate and individualize practice.

Practical Ways to Individualize Challenge

1)
Constrained Choice
As we will look at more in Chapter 13, getting an athlete
involved in practice by allowing them to choose what they are
going to work on can be a powerful tool for enhancing
motivation and autonomy. However, we don’t want to give
them complete control over what they practice. Research
examining the practice diaries3 of athletes has shown that, left
to their own devices, most will choose maintenance over
deliberate practice because it’s more enjoyable and rewarding. I
have observed this personally in how I train as a distance
runner. Instead, of doing workouts quicker than my race pace
(to improve speed) or much slower than my race pace (to
recover), I tend to do most of my runs at a pace slightly
slower than I do in a race. This allows me to cover a larger
distance in a reasonable amount of time which is very
satisfying.
A good compromise is using constrained choice. This is where
the coach selects a small number of practice activities and lets
the athlete choose which one they want to do in a particular
session or the order in which they are completed. Let’s look at
a research study that examined the latter. In a paper published
in 20204
, An et al compared two conditions for training golf
putting. Both groups were given the same three practice
activities. The first activity was a visual cue condition in which
small colored markers were placed on the putting green. These
were designed to help performers focus on developing equal
length swing (i.e., a 1:1 ratio between the length of the back
swing and follow-through). The second activity involved the use
of a metronome playing an auditory tone (at 60 bpm)
designed to facilitate a focus on developing a constant swing
tempo. The third activity was the constraint of a chest bar
placed between the arms in front of the chest to provide tactile
cues and constrain the performer to use a pendulum-like
motion. The only difference between the two training groups
was that in the Choice group participants selected the order in
which they completed these three practice activities, while
participants in the control group completed them in a preset
order. Following training, the Choice group had a roughly 20%
higher putting accuracy score. They also gave higher ratings of
perceived confidence.
Using a constrained choice paradigm seems to achieve the best
of both worlds. We get the powerful motivational effects of
choice while still ensuring that our practice is deliberate. That is,
we can use constraints targeted at a specific weakness or rate
limiters. For example, building on the study just described one
could imagine using different postural constraints for individual
athletes (instead of everyone using the chest bar). This is
essentially what I do in my baseball training…. One athlete
might be asked to cross their legs while another is asked to
spread their feet as wide as possible (depending on my analysis
of their swing).
Using the examples illustrated in Figure 7.2, we could give a
basketball player the choice between taking shots from position
2, 5, or 7 as opposed to just letting them shoot from
anywhere on the floor they choose. Or for the baseball player,
we could have two pitching machines set up to throw either
inside or low pitches (both areas of weakness for them) and
give them the choice of which they would rather practice first.
2)
Win, Advance, Lose, Stay
Another effective way to individualize practice is to set up
different stations with activities and have the athlete move
between them based on their performance. In this approach, if
the performance criterion for a given practice activity is met (a
“win”) then a switch to a different activity occurs on the next
trial, while if it is not (a “loss”) then the same task is practiced
again on the next trial. For example, we could have a
basketball player shoot from a particular location on the floor
and not move to another location until they make 2/3 shots.
The basic rationale here is that repeatedly practicing the same
task when the performer is already doing it successfully (as in
a lot of blocked practice) is a waste of time. Conversely, moving
away from one task to another (and adding more variability)
when success has not yet been achieved is potentially poor for
motivation and learning.
The effectiveness of this win-move, lose-stay for training
basketball shots was investigated by Porter et al in 20195.
Practice was comprised of 360 trials in total, with 60 trials
completed during each of the six practice sessions. As illustrated
in Figure 7.3, shots were practiced from four different locations
on the floor, varying in both distance and angle to the hoop.
There were three different training groups. During each session,
the blocked group undertook 15 trials at one of the four
locations before switching to the next. The random group
undertook all 60 trials following a predetermined quasi-random
order, with no consecutive trials occurring at the same location.
The win-move, lose-stay (WMLS) group undertook all 60 trials
following the same order as the random group, however, they
only switched shooting locations following a successful shot –
otherwise they repeated the same shot from the same location.
Pre and post-training tests involved 20 trials (5 at each shot
location). Post-tests included both an immediate test (right after
the final block of training) and a delayed test (7 days after the
final practice). There was also a transfer test that involved 10
shots from the free throw line (a location not practiced during
training).
What was found? While there were no group differences for
the immediate post-test, the WMLS group had a significantly
higher shooting percentage in the delayed post-test (51) as
compared to either the blocked (48) or random (46) groups.
Similar results were found for the transfer tests: WMLS (58),
blocked (48), and random (47).

A diagram of a basketball court Description automatically

generatedFigure 7.3 – Shot locations in the Porter et al study
3)
Adjusting Variability Through Constraint Switching & Taking
Advantage of Learning at Different Time Scales
The third method I find effective in individualizing training is
adjusting the amount of variability for each performer,
particularly the variability created by switching between different
task constraints. As discussed in Chapter 6 (and illustrated in
Figure 6.7), it is proposed that there is a level of variability that
is optical for learning for each individual athlete. I try to ensure
that we are spending most of our time training in this optimal
region. For example, in bat speed training, a common approach
that is being used by many teams now is to have the batter
hit with three different bats8
: an underload bat (20% lighter
than a normal bat), handle-overload bat (20% heavier than a
normal bat with the weight added to the handle), and
end-overload bat (20% heavier than a normal bat with weight
added to the end of the bat). Going back to Chapter 4, we
are giving the batter new representative problems to solve.
Figure 7.4 shows the bat speed data I collected using this type
of training over a period of several weeks. In this figure, we
can see that learning seems to be happening at two different
time scales7
. There is a slow learning process – across several
weeks bat speed is increasing overall (which is what we want)!
This occurs because the batter is finding new movement
solutions for swinging the bat through a process of
self-organization. But there are quicker improvements. In each
session, every time we hand the player a new bat, their swing
starts a bit slower and then increases. This is an example of a
well-known effect in motor learning called the warm-up
decrement.
A graph showing the growth of a stock market Description
automatically generated with medium confidenceFigure 7.4 –
Slow and Fast Learning in Bat Speed Training
The warm-up decrement was first discovered by Jack Adams
in 19528
. In his studies, he used a rotary pursuit task which
involves a small target mounted on a wheel spinning around
and around and the participant moving a stylus to stay in
contact with the target for as long as possible. So, it is a test
of eye-hand coordination. Participants completed five days of
practice on this task with 36 trials per session. The dependent
measure of performance was the time on target, that is, how
long they could keep the stylus in contact with the target.
In training, there was mostly a continuous improvement in
performance with diminishing returns that could be fit by your
traditional learning curve (i.e. slow learning). For example, on
day 1 of the testing, participants showed a large improvement
over the 36 practice trials from 3 sec time on target to about
7 sec. While on day 5, the last day of training, they still
improved but with a much more modest gain from 7 to 8 sec.
However, every time participants started a new session of
practice it took them a while to “get back up to speed”. For
example, at the end of Day 1, time on target was over 7
seconds but at the start of Day 2, it was only 5.5. It took
about 6-7 trials to get back to the performance level they had
achieved on the previous day of practice. It’s as if they needed
to warm up a little – thus the name given by Adams. But
interestingly the size of this warm-up decrement got smaller
and smaller with each day of training. So, while it took 6-7
trials to warm up at the start of Day 2, it only took
participants 2-3 trials to warm-up at the start of Day 5. So
here we have a classic example of changes occurring at
different time scales…improvements in time on task were
occurring over the course of days (i.e. slow learning) but also
within the first few trials of each practice session (i.e., fast
learning)
But, of course, you are probably wondering: why does this
warm-up effect occur? Experiments that followed the ones done
by Adams ruled out a couple of possibilities. First, it does not
seem to be a physical effect. If you allow participants to move
around, stretch, etc. before starting a new day’s session you
still get a warm-up decrement. It also does not seem to be an
example of forgetting. The most commonly accepted explanation
for it now is Nacson and Schmidt’s Activity Set Hypothesis
which was proposed in 1971
9 They proposed that rather than
the warm-up decrement being due to some loss of coordination
or skill for the specific task, it is due to a loss in the
participant’s preparation to respond (what we now classify as
an individual constraint). This preparation includes three things:
having the appropriate level of arousal/motivation, having
attention focused on the right things, and having appropriate
expectancies about the task. According to Nacson & Schmidt,
these three things, which form the activity set, get adjusted to
optimal levels during practice but will be lost during a rest
period. A common example of this would be a performer
showing up at practice over-hyped with too high a level of
arousal. So, it will take a few trials on each new session to
properly adjust this activity set leading to the warm-up
decrement. With enough practice, the activity set gets
incorporated into our control laws for movement. Therefore, it
has less and less of an effect over time.
The key point for our purposes in this Chapter is that the rate
at which we switch between task constraints is a source of
variability that can be manipulated to get the level of challenge
appropriate for an individual athlete. So, for example, if we
have a baseball batter who is swinging fast and hitting the ball
hard on for most of the trials of a practice session, I will have
them switch the bat they are using more often (say, every 5
swings) as compared to an athlete that is not performing as
well. Doing this is going to make the warm-up decrement play
a larger role. I also find this to be an easy technique to
implement because all players are doing the same thing in a
practice session (they are not doing something completely
different) you are just changing the rate at which they are
asked to switch constraints. Other examples of how I have
used this are varying the rate at which basketball, handball, or
soccer players switch between shot locations (e.g., the one’s
shown in Figure 7.3) in a practice. So, for example, one might
switch every 5 shots while another switches every 10. Or, in
martial arts training, I will vary how often the fighter switches
between different starting positions relative to their opponent.
For this, I use the 70% rule I discussed in my second book –
we want all performers to be about 70% successful. If their
success rate is higher than this level, switch more often. If it’s
much lower than 70%, switch less.
The Challenge of Individualizing in Groups
How can we achieve all this when we have 10 athletes
standing in front of us with large individual differences in
ability?! While there is no simple fix, I think there are some
best practices we can use. And like with most things in the
ecological approach, we can view this as an opportunity rather
than a problem! So, here is what I try to focus on when I
run a practice.
Constraining to Afford in Both Directions.

When working in a group that includes one or more athletes

who are much more experienced than others, I like to think
back to being a kid playing street hockey in Canada and ask:
how can we change the game to even the playing field? How
can we change the constraints for different players so that we
can give the more experienced athlete a new movement
problem to solve while at the same time opening up new
affordances for the lesser-skilled players when they play against
each other? Here I am talking about things like using different
equipment, creating different rules for scoring (e.g., the more
experienced athlete has to put it in a smaller area of the net
or court), having two players always guard one in a small-sided
games scenario, adding a rule that the more-experienced athlete
can only use certain movement solutions (e.g., they have to
defend their arms behind their back in an MMA training
activity), etc.

2. Taking Advantage of Variable Observational Learning Models

.

Although it can seem like having athletes of very different skill

levels is problematic it actually creates the ideal situation for
observational learning. As will discussed more in Chapter 10,
research has shown that we learn best when exposed to a
mixture of peer (watching other athletes that are close in skill
level to ourselves) and expert models10
. The former allows
learners to gain information from ineffective movement solutions
and gaining better information about what they themselves are
doing when executing the skill. Viewing an expert gives learners
the opportunity to pick up information about new affordances
and new possibilities for movement solutions. So, I try to have
the more experienced athletes do demonstrations for the
lesser-skilled as part of..

Create Learning Opportunities by Allowing Athletes To Coach.

Allowing athletes to coach each other in practice activities can

be a very powerful tool for learning. Especially, when they
know that they will be asked to do it beforehand. In a really
interesting set of studies, Keith Lohse and colleagues11 have
shown that when athletes expect that they will have to teach a
skill later, they learn the skill more quickly and effectively
themselves. Results suggest that when expecting to teach we
attend to different things and pick up different information
when learning the skill ourselves. One of the things I love to
do when this is happening is walk around a listen to the things
the kids are telling each other. This can be a great way to
“steal” cues and analogies that are more relatable to younger
athletes! Sometimes my examples of programming VCRs and
playing Pac-Man fall on deaf ears.

4. Use Co-Adaptive Practice Design.

Instead of having a practice activity that is too easy for the

more-experienced athletes (e.g., just put the ball in play in a
baseball batting activity) or having one that is too hard for
less-experienced ones (e.g., try to hit the ball to the opposite
field), I like to let the athlete choose their intention for each
execution. For example, I might ask a tennis player: “What are
you going to try to do with this serve?” Typically, you get
answers that reflect the skill level (e.g., “make sure it lands in”
vs “make my opponent return it on their backhand”) but you
can also guide this as a coach by making suggestions. Giving
the athlete a choice has both the benefits of promoting
autonomy while also creating a more effective level of challenge
for each individual.

Taking Advantage of the Opportunity to Do Some Pressure

Training.

There is a lot of research showing that adding pressure to

training can lead to better motor learning12. So, I like to take
advantage of groups to create one of the most effective types:
social pressure. While creating competitions (the first person to
hit 10 balls in fair play wins) can be effective, creating social
consequences for not performing well (e.g., my favorite –
having the person that gets the fewest number of hits make a
speech in front of the team) seems to better simulate the type
of pressure that makes some athletes fail in competition. To
take into account different skill levels I either have different
constraints (i.e., scoring systems) for the different athletes or
put them in small teams.

6. Using Different Methods for Matching Opponents.

Finally, in the common situation where athletes are training in
pairs, I like to explore the methods with which the pairs are
selected. For example, letting the athletes choose, letting the
coach choose, and using some random method like picking a
number out of a hat.

Ok, let’s take a break from designing practice. What should a

coach do once it starts? What information can we pick up
from observing practice, and how we can use that to adapt
practice?

8
EDUCATING A COACH’S ATTENTION
& LEARNING TO OBSERVE
MOVEMENT
" We need to coach towards something”
- Dan Pfaff, Altis

I f you think about it, picking up the right visual information

and using it optimally is as critical for a coach as it is for an
athlete. And when coaches do it “live” in practice there are
some similar constraints as those faced by the athlete. The
unfolding movement being viewed can be very fast and have a
very short duration. Also, like the problem faced by the athlete,
the differences in movement and posture they are looking for
can be incredibly small. In this chapter, I want to examine
different ways that coaches can pick up and utilize information
from observing the movement of their athletes.
The Traditional View: Mental Models, Schemas & Motor Skill
Diagnosis
What are good coaches picking up when they observe an
athlete and how are they using this information? Traditionally
this process of perceptual expertise in coaching has been
conceptualized as the coach looking for flaws or errors in the
movement based on their own internal, mental model of the
correct technique. This is captured in the model of Motor Skill
Diagnosis developed by Pinheiro and Simon1
. The main idea
proposed is that for every skill there is a correct technique with
some small acceptable bandwidth around it. To capture this
idea, the authors again appeal to the concept of a memory
schema. As we saw in Chapter 6, a schema is a structure in
long-term memory that we use to organize, classify, and chunk
new information.
Used in this context, a memory schema (that is, an internal
model for a general class rather than one specific action) allows
some variation in the skill. As Richard Schmidt once said a
schema provides a “reference correctness” rather than a specific
recipe for the movement2
. Using a schema to process the
incoming information also reduces the load on a coach’s
working memory and attention, which otherwise will be
overloaded.
Within their motor skill diagnosis model, Pinheiro and Simon
include three distinct phases: cue acquisition, cue interpretation
and diagnostic decision. Note, importantly, the use of the word
“cue” here. It is consistent with the indirect perception nature
of this internal model approach. That is, the coach is not
picking up information that directly specifies what needs to be
changed about the athletes’ technique, but rather they are
getting cues or clues that need to be interpreted and
processed.

The first stage of the model involves acquiring cues that can
trigger the recognition process in the schema. The cues are
comprised of spatial and temporal patterns from the movement.
The choice of cues is not arbitrary or random but is based on
systematic observation of the motor skill, looking specifically for
discrepancies between the actual performance and the ideal
model.
The second stage of the model, cue interpretation, involves
finding possible meanings for the cues through processes of
extrapolation and interpolation. That is, the coach is using their
memory store of previously experienced movement executions
to interpret the new one they are observing. The coach does
not necessarily have to have seen the specific technical flaw
before to be able to diagnose it. They can do this by
extrapolating from previous observations.
The third stage involves making inferences about the specific
technical error observed and generating feedback or a new
practice activity to correct it. For those paying attention, this
schema is essentially the “corrective lens” that Craig Morris was
referring to in the article I discussed in Chapter 2. Finally, this
stage involves making inferences about the specific technical
error observed and generating verbal feedback or a new
practice activity to correct it.
In 2021, Fetisova and colleagues3 compared the internal
models/schema used by tennis coaches when evaluating a
tennis serve. Four novice coaches (with 2-4 years of
experience) and four expert coaches (all with >20 years of
coaching experience) were interviewed and asked to participate
in a Think Aloud protocol (i.e., narrating while they were
observing their athletes) for the study. As illustrated in Figure
8.1, it was found that the internal models used by the coaches
had six meaning units.
A diagram of a tennis player Description automatically generated
A diagram of a diagram Description automatically generated
with medium confidence

Figure 8.1 – The Internal Models of a Tennis Serve Used by

Novice (top) and Expert (bottom) coaches. Based on data from
Festova et al3.

Although there are some key differences, that I will discuss

shortly, it is clear there are also a lot of similarities between the
internal models for the novice and expert coaches. Within the
technical elements, body elements, and phases categories,
coaches were primarily looking at key body positions within the
action (stance, grip, knee flexion, hip, forearm, back) that occur
in distinct phases (load, contact, ball toss, finish). This, of
course, fits well with the part-task training and perception-action
decoupling approaches to make a sports skill easier, discussed
in Chapter 6. Not only does breaking the skill into parts make
the skill easier for the athlete, it also reduces the cognitive load
during observation by the coach.
The main differences between the internal models used by
novice and expert coaches were in the reasoning chains,
concepts, and key flexion points components. Expert coaches
brought in holistic concepts such as rhythm, elastic reaction,
“good pace” and extension. They also included key transition
(or flexion) points between the phases with acceptable
bandwidths: “
90 degrees between the forearm and shoulder in
the trophy position”, and “knee flexion is 110 degrees plus or
minus 10 degrees”. The findings led the authors to conclude
that the main attribute of tennis coaching expertise is not pure
technical knowledge but rather the perception of technique as a
whole movement and the ability to understand the connection
between technical elements.
There are some important limitations to observing movement
using this type of internal model. First, for the most part, the
coach breaks the skill into a distinct set of positions and
transitions between these positions. Thus, they are breaking a
complex motor skill that requires coordinating multiple degrees
of freedom into key positions of single (or a small number of)
degree(s) of freedom (e.g., “knee flexion of 110 degrees”).
Stated another way, they are reducing the skill to a set of
lower order variables (angles and position) instead of a higher
order variable that captures the pattern of coordination (e.g.
relative phase). This is problematic because it again assumes
linear interactions and linearity of scaling within the system.
That is, the overall skill can be predicted by understanding the
component parts (technical elements). As first emphasized by
Bernstein, with his discussion of “context-conditioned variability”
(see Chapter 2 in my first book for more details), the effect a
single degree of freedom (e.g., one joint angle or the amount
contraction in one muscle) will have on the overall movement
pattern is highly dependent on the context.
Another problem can be seen in the concept of movement
bandwidth. While this does allow for some individual variation in
the skill, it is expressed in terms of a single degree of freedom
with some acceptable range (variance) around some ideal value:
“knee flexion is 110 degrees plus or minus 10 degrees”. This
does not allow for any variation in the overall pattern of
coordination. For example, it is possible for a movement to
produce a successful outcome with a degree of freedom that is
out of its idea bandwidth if there is some compensation
(co-variation) from other degrees of freedom in the overall
coordination pattern. The concept of bandwidth assumes that
variation in technique comes around a single joint rather than
variation involving multiple joints (i.e. the coordination pattern).
As we will see below, having a bandwidth when observing a
movement is very different than looking for invariants in the
movement.
Finally, using this type of internal model does not do a good
job of capturing functional variability or motor synergies within
the system. A motor synergy occurs when the different degrees
of movement work to compensate for each other during a
single execution. For example, whether or not a particular
shoulder angle is part of an effective tennis serve depends on
what the associated elbow angle is for the same serve. If I
rotate my shoulder slightly less (e.g., because of fatigue), I can
compensate for this with more elbow extension. Looking at
movement as a series of positions (with some bandwidth) does
not capture how they might work together. Critically, a motor
synergy is expressed in terms of a manifold (a 3D volume)
rather than a bandwidth (a one-dimensional variable).
In his review of the concept of motor schema, Karl Newell4
identified a few other general issues that also apply to using a
schema for observing and interpreting movement. The first is:
where do these schemas come from in the first place? This is
one of the problems of indirect perception James Gibson first
identified: “Knowledge of the world cannot be explained by
supposing that knowledge of the world already exists”5
. If we
need this memory structure to observe an athlete, then how
could we ever build this structure in the first place? This is
circular logic. There must be information in the movement that
directly specifies whether it will be effective without the need for
interpretation using past knowledge. Second, although schemas
eliminate memory storage problems (the coach does not need
to have a store in memory for every possible variation of a
sports skill), they are replaced by a computational problem.
There are too many interacting factors (degrees of freedom) to
be able to predict what the ideal technique will be for an
individual athlete.
Furthermore, it is also critical to consider the corresponding
sensitivity of performers to implement kinematic changes in their
movement patterns. For instance, if the visual sensitivity of a
coach is superior to the athlete’s kinesthetic sensitivity, any
feedback provided at the most precise level of visual sensitivity
will be beyond the level that performers can use effectively.
Stated another way, even though a coach may be able to
detect small deviations from some ideal (visual sensitivity)
athletes may not have the kinesthetic sensitivity (i.e. “feel”) to
correct such deviations. A good example of this can be seen in
the Giblin et al study6 of the ability (or rather, inability) of
tennis players to make small changes to their movements
during a serve (discussed in Chapter 7 of my first book).
Another issue associated with this internal model approach is
that it does not seem to be consistent with research on the
gaze behavior of coaches7
. In a study published in 2020,
Waters and colleagues explored the gaze behavior of coaches of
different experience levels. Twenty-two sprint coaches (10
experts with more than 10 years of experience and high-level
accreditation and 12 novice coaches) were compared to 12
sport biomechanists with no coaching experience. Participants
were asked to make verbal commentaries about running
technique while watching videos. While doing this, their gaze
behavior was measured with an eye tracker. The videos created
were of athletes from a range of different skill levels and two
different sports (rugby and track & field).
What was found? There were no significant group differences
in gaze behavior. There was also some evidence that all
participants were using a visual pivot strategy when viewing the
videos. That is, they were keeping their eyes fixed on one point
so that they could view the body motions in peripheral vision.
The authors conclude:
These findings don’t seem to support the idea that coaching
observation is driven by internal mental models. If this were the
case we would have expected to see clear gaze behavior
differences (for coaches at different levels).
Finally, is there any evidence that expert coaches can identify
the objectively better technique (as assessed by the athlete’s
actual performance) by observing the technical elements using
an internal model? The results of this type of analysis have
been mixed. For example, in a study published by Sparrow in
19928
, it was found that there was no difference in the ability
of expert vs novice coaches to estimate the handicap of a
golfer from observing their technique. However, in a different
study by Leas and Chi
9
, experienced coaches' ratings of the
overall movement of swimmers were significantly correlated with
their race times.

The Ecological Approach to Observing Movement as a Coach

In the ecological account, coach observation involves a process
of directly picking up information from the athlete’s movement
pattern which specifies whether it will be effective (in achieving
the athlete’s goal/intention) or not. This involves detecting
invariants in the movement solutions of our athletes. Or
equivalently, as discussed in Chapter 6, the coach should focus
on the Essential movement variables instead of the Elemental
ones. So, when I watch a baseball pitcher, I focus on detecting
whether the key invariants (e.g., not breaking the kinetic chain,
consistent position of the hand at ball release) are present or
not, rather than the details of how those are produced (i.e., the
pitcher’s arm slot or the length of their stride). Fundamentally,
this is recognizing that there is no one ideal technique but
rather there are multiple different movement solutions (that
must have some common features) that will be effective. Again,
this is very different than looking for a solution bandwidth.
Invariants in Movement
Let’s clarify the concepts of invariants in movement a little
more. The term comes from Gibson’s analysis of perceptual
information (discussed in Chapter 2 of my second book). It has
long been argued that we need the help of internal models to
interpret sensory information because it is ambiguous. For
example, when an object’s image size increases how do we
know whether this occurred because the object increased in
physical size (like a balloon being inflated), moved closer to us,
or a combination of both? The change in image size cannot be
used to distinguish between these different events in the outside
world.
Gibson noted that we can easily solve this problem by picking
up different, higher-order sources of information from the
environment. Specifically, if we use the ratio of an object’s
image size (S) and the angle it forms with the horizon (β), we
can directly distinguish between these different events (see
Figure 2.2 in my second book). How do we know what this
information source is? Easy, we look for perceptual invariants –
information that remains constant for every event of a
particular type. For example, every time an object approaches
us, the S/β ratio will remain constant. This is the invariant –
across all events of this type (approaching object) it does not
vary.
The same logic can be applied to understanding the control of
action. For every jump shot that goes in the basket, golf drive
that goes down the center of the fairway, and baseball pitch
over 95 mph there will be some invariants in the movement
pattern. Things that successful movement solutions have in
common. But just as was the case with Gibson’s analysis of
perceptual information, these will be higher-order aspects of the
movement (for example, things like relative phase and motor
synergies) not lower-order variables (e.g., the angle or range of
motion of the knee joint).
On a much simpler level, we can see this every time we view
a sports skill. All classes of movement (e.g. hitting vs throwing)
have invariant characteristics of the relative motions of the body
and limbs. Otherwise, we would not be able to know what we
are looking at! Kelso refers to this as a movement topology. A
topology is a set of geometric properties and spatial
relationships that remain unaffected by overall size and shape.
These topologies are present when we plot movements in terms
of variables of the solution space. For example, Figure 8.2
shows the solution space (created by looking at the shoulder
and elbow joints) for overhand volleyball serves made by 2
experts (members of the French National team) and 2 novices
(undergraduate students with no previous volleyball experience)
10.
There are a few different things we can see in these solution
space plots. First, note that they all have roughly the same
shape – they all look like fishhooks rather than circles or
squares, for example. This similarity is what allows us to watch
a player and say “they are serving a volleyball” instead of “they
are throwing a football”. Why are these similar? Because the
people performing the action all share some common
constraints. They are all serving using the human body, with
the same size ball, over the same net, etc. So, we should
expect there to be some similarities between their solutions.
Second, although they have common overall shapes, there are
some clear differences between the novice and expert
movement solutions. This is again to be expected. An
experienced volleyball player will have different action capacities,
be attuned to different sources of information, and have
different information-movement control laws. Finally, the two
experts' movement solutions don’t have the same shape.
Different specific combinations of elbow and shoulder angles
(the elemental variables) can be used to produce the same
essential variable (e.g. the hand arriving at the same instant the
ball does).
Figure 8,2 – Movement topologies in solution space for a
volleyball serve. Based on data from Temprado et al
10
.

Another important feature of a movement topology is that is

scale invariant. In other words, the same overall relationships in
solution space will still be present when we serve the ball 70
mph instead of 50 mph. Or if we serve the ball short or deep.
Obviously, we should be expected to see some difference in the
movement solution for these different task goals, but the overall
topology (the invariants) should still be there.
Another way to think about invariants is in terms of attractors
– they represent points of stability in the solution space that
we are drawn into. This idea will be discussed in more detail in
Chapter 15.
Identifying the Invariants for Your Skill
How, as a coach, do we go about separating the wheat from
the chaff in movement, that is separating the elemental variables
(that for the most part, we want to ignore when observing our
athletes) and the essential variables (the key invariants that we
want to guide self-organization toward)? I am aware that most
coaches are not going to have access to motion-tracking
systems, but if you do I have some ideas for you in Chapter
15! Luckily, identifying invariants does not have to be done at
the level of movement topologies in solution space.
Fundamentally, it just requires a task analysis – looking at a
sports skill in terms of specifically what is trying to be achieved
and how. In my interview with Greg Souders11, he gave a
great description of how he does this in Brazilian Ju-jitsu:
So an example is if we're in a standing situation, immobilization
has a huge effect once I get the hips on the ground. The hips
are the seat of power and the seat of balance. We move our
hips through connection with the floor. So if I put your hips
on the floor, I've done most of the destabilization that's going
to occur, as it relates to starting to immobilize you.
And so. That is what we call an invariant. It's an invariant
feature of the human body as it relates to staying mobile and
producing power. And so once I learned, I started looking for
those. Okay, well, that's the first one. And then degeneracy was
important because the question then becomes how many
different ways can I line up my structure to achieve this end?
How many different ways can I solve the problem? How can
put those hips down on the floor in as many varied ways as
possible?
So, Greg identified getting the hips down on the floor as a key
invariant to immobilizing an opponent. He then guides his
athletes in finding their own (degenerate) movement solutions
that produce this movement outcome. Another example he
gives is the invariant for getting submission by attempting to
break an opponent’s arm:
So the human arm breaks in two ways, we either twist it
when it's bent to break it, or we hyperextend it when it's
straight to break it. There's no other way the arm breaks,
that's it. It's an invariant property of an arm. So rather than
show them the optimized technique to straighten an arm, we
give them the opportunity to exist in that position where they
have to attempt to straighten an arm. And we play games
where they get to learn to adapt to the variation present in
that system.
Notice how Greg’s descriptions are focused on outcomes
(breaking and immobilizing), not the specifics of the techniques
used to produce those outcomes. He recognizes what his
athletes are trying to achieve: movement solutions with key
invariants in the face of incredible variance (e.g. changes in the
size, position, etc. of your opponent). And he designs his
practice to help his athletes find these.
I think this is the best way to identify the invariants for your
sports skill as a coach. Think about your skill and ask yourself:
what must be present in the movement outcome for success?
Like Greg Souders’ laws of arm breaking, can you identify how
the laws of physics influence movement solutions in your sport?
Below I give a few examples I use. Note, that I am deliberately
trying to keep the discussion simple so that we do not need
an advanced physics degree to understand the ideas.
The Laws of Force & Collision
Newton’s Second Law
As we all learned in high school physics class, Newton’s Second
Law states:
F(force)=m(mass)*a(acceleration)
Whether we are talking about a club, racquet, bat, ball, or your
hand - if you want it to move quickly (i.e., accelerate) you
need to put force into it! Where does the force come from in
most sports? The ground! So, we can state a simple invariant
for most sports skills: if we want to make something move
quickly we need to transfer the force effectively from the
ground into it. Because of our anatomy, this is best achieved
by creating a kinetic chain. As first proposed in 1875 by Franz
Reuleaux, our body contains rigid, overlapping segments that
are connected via joints which creates a system whereby
movement at one joint produces or affects movement at
another joint in the kinetic link. Our body parts are like links in
a chain: as one part moves it causes the next link in the chain
to move and so on. The lower extremity and trunk generate
and transfer energy to the upper extremity. Coordinated lower
extremity muscles (quadriceps, hamstrings, hip internal and
external rotators) provide a stable base for the trunk (core
musculature) to rotate and flex.
Maintaining this kinetic chain is a key invariant for most sports
skills. When we break it, we don’t generate as much force, so
we don’t generate as much speed. Furthermore, in many cases,
this also increases the chance of injury. For a specific example,
see my discussion of forearm flyout in baseball pitching from
my first book. There are a few different things a coach can
look for:
1)Is the athlete effectively loading (i.e., taking force from the
ground)? Specific things to look for:
-Are they stretching the elastic band and creating separation in
their fully loaded position, as discussed in Chapter 5?
-Are they maintaining contact between their foot/feet and the
ground for as long as possible or are they coming out of their
load (e.g. by spinning on their back foot)?
2)Is the athlete transferring force effectively?
-Is the force transfer roughly proximal to distal? That is, does
movement of the large body parts (proximal to the center of
our body) occur before movements of the smaller body parts
(distal from our center)? For example, rotation of the pelvis
occurs before the rotation of the torso which occurs before the
rotation of the arms. This can be done via either coarse
viewing or by measurement of the kinematic sequence12
(discussed in Chapter 15).
3)Are they decelerating effectively?
-Are they spinning off at the end of the movement? Is there
force leakage? That is, are they continuing to move after the
action is complete in a different direction than the object they
hit or stuck is going? This is wasted force that wasn’t put into
the ball and has to be dissipated. An example of a baseball
pitcher exhibiting this behavior is shown in Figure 8.2.
A person in a red shirt and cap Description automatically
generated
Figure 8.2 – Force Leakage in Baseball Pitching. Instead of
moving in the direction the ball is pitched, the pitcher spins so
this ends up with his back to plate.
Newton’s Third Law
Newton's third law of motion states that in a collision between
two objects, both objects experience forces that are equal in
magnitude and opposite in direction. Stated another way, if we
want an object we hit to go straight we need to hit it straight!
This is why I consider the variables described in Chapter 5 for
“curing a slice” in golf to be invariants. They are breaking this
fundamental rule of physics. This boils down to three essential
variables:
1)
The club must hit the ground behind the ball
2)
The club path must be straight
3)
The ball must be struck in the center of the club
What type of swing do you use to create these outcomes I am
not too concerned with, unless, of course, it breaks the kinetic
chain in some way. But, the golfers out there are probably
saying: what about playing a draw or a fade? For these shots,
an experienced golfer deliberately does not use a straight club
path to put side spin on the ball and make the ball move in a
certain direction. This is an example of the final stage of
coaching shown in Figure 3.2, the adaptation and optimization
phase. At this stage, the coach might work with the athlete to
violate (break) some invariants to produce some different (more
advanced) performance outcomes.
The Laws of Optics
Here we are talking about pretty obvious things. Light travels in
straight lines which means (in the absence of a reflecting
surface) we can’t look in a completely different direction from a
visual information source (e.g., down at the ground) and still
pick it up. A related invariant is one that I discussed in detail
in Chapter 3 of my second book: gaze stabilization. The basic
invariant is: to pick up visual information effectively we can’t let
our body movements cause our gaze direction to shift. Two
common examples I see in baseball often are: when a batter's
gaze shifts upwards when they land on their lead foot and
when the same happens for an outfielder as they land hard on
the ground before the ball is hit.
The Laws of Projective Motion
As an example of an invariant that can be derived from these,
consider the law:
or as stated in words, the final height of a projectile ( depends
on three things: gravity (g, which we can’t change, although
see my discussion of spin on 4-seam fastballs in Chapter 3 –
again another example of how optimizing performance can
involve violating the invariants), how long it takes to get to its
destination (t, which will be roughly the same for pitches of a
given type), and the height at which it is released (
). This
gives the invariant illustrated in Figure 6.9: if a pitcher wants to
consistently hit their target, their release point (
) must not
vary.
Learning to Pick up Invariants
Once we have identified them, how do coaches learn to
perceive them? Well, similar to the athlete, we need to educate
their attention to these information sources through practice.
And just like with the athlete, this will be facilitated by adding
variability to the observation conditions. Viewing from different
angles, different body types, different ages, and skill levels. As
Sparrow & Sherman discussed, the goal is to learn to pick
topologies and scaling, rather than specific cues.
Another effective thing I like to use is something I have already
mentioned a couple of times: diagnostic constraint manipulations.
So, the idea here is to add a specific constraint like a heavier
bat or wider stance. Observe what happens. If they are still
successful in the performance outcome (i.e., hitting the ball
hard) what about the swing remains invariant? If they were
unsuccessful what was missing once you added the constraint?
Next chapter we turn to another key skill for a coach to
develop when observing the movements of their athletes –
detecting their affordances.

9
PERCEIVING THE AFFORDANCES OF
OTHERS
C an the athlete you are working with make it through that
gap in the defenders? Can they get to that flyball? Can they
break free of that hold? An important part of effective coaching
is detecting the action opportunities available to the athlete you
are working with. We are talking about perceiving affordances,
but in this case not for ourselves but for others. Recall, that an
affordance is an opportunity for action conveyed by information
from the environment. Research has shown that we are very
good at picking up affordances for ourselves. We can detect
whether objects are reachable, throwable, climbable, etc. f
or us
quite accurately. But what about perceiving the affordances of
others? Can we pick up information from the environment to
determine whether an object is reachable, throwable or
climbable for another person? How might we do this? This has
important implications for coaching – being able to detect the
action capacities of and affordances available to an athlete you
are working is invaluable in designing practice. So, let’s have a
look at a few studies that explored this topic.

As per usual, there are two very different views of how we

might perceive the affordances of others. The first, aligned with
information processing approach, proposes that we do this by
process of mental simulation. That is, when we observe another
person performing an action it activates the neural mechanism
(mental models, motor programs, etc.) that we use to produce
the same action ourselves. This fits with the very popular
concept of mirror neurons1 which respond both when
producing an action and viewing someone else producing that
same action. The alternative ecological view is, of course, that
perception of the affordances of other’s is information driven.
That is, we pick up information directly from our environment
that specifies the action capacities of others.
There are a couple of simple differences we should see
between these two theories. First, if we are understanding other
people’s actions through a process of mental simulation, then
we might predict that we would be poor at judging the
affordances of others with very different action capacities from
ourselves. For example, a much taller or shorter person.
Second, if are using mental simulation we should need to see
the action being performed (so that it triggers the appropriate
mirror neurons) to perceive the affordances of others. Neither
of these would be predicted in the ecological approach. Because
we are picking up information from the environment, we are
perceiving another’s affordances in terms of what they can do,
and we shouldn’t need to see the action performed. For
example, if a person is standing still there is information that
can be used to detect whether they can reach an object of a
particular height.
These two different explanations were compared in 2008 study
by Ramenzoni et al2
. In this study, participants were asked to
make judgments about the affordance of reachability for
another person. That is, they viewed another person just
standing still with their arms at their side and were required to
adjust the position of an object that was hung from the ceiling
until it was at the maximum height that would reachable for
that the person. After the judgement was made, participants
were asked to try to reach objects at different heights so that
their actual maximum reach height could be calculated.
There were two manipulations added to tease apart the
different theories. First, participants came to the study in pairs
and recruited such that one was shorter (150– 160 cm) and
one was taller (180–190 cm). If we buy the simulation
approach this should make it difficult for the participants to
make accurate affordance judgments. The other manipulation
was to have the participant making the affordance judgment
stand on platforms of different heights – so, unbeknownst to
them, while viewing the other person they stood at the same
level, 7.5 cm or 15 cm higher. Why would this matter? Because
in the ecological approach, we don’t perceive the world in
terms of units from physics – for example, we don’t perceive
height in terms of feet or meters. We perceive it relative to our
own individual constraints (i.e., embodied perception) -specifically,
we perceive height in terms eye heights. That is, we perceive
another object or person relative to the height our eyes are
above the ground. So, for example, if an object is 12 feet
above the ground, it would be perceived as being 2 eye heights
tall for a person that is 6 foot and roughly 2.5 eye heights for
a person that is 5 feet tall. The same physical height is
perceived differently because we don’t see the world in terms
of physical dimensions – we see it in terms of what it affords
us (e.g., can I touch the other person on the top of their
head?). For the present purposes, the change in the height of
the platform the observer was standing on should change their
perception of another person’s affordances because it alters
their eyeheight. Finally, participants were also asked to judge
the affordance of reachability, for themselves. For this, they
moved the object on the rope to a point where they could just
touch it with no other person there.
What was found? First, participants were very accurate in
judging the affordances for both themselves and the other
person. In these types of studies, we typically calculate the ratio
of the perceived to actual value (called a  or Pi number) –
so, in this case, the judged maximum reach height to the
actual measured maximum. In all conditions, Pi numbers were
close to 1.0 indicating accurate affordance perception. So, we
can judge what is reachable for another person, even if they
are much taller or shorter than were are. Note, of course, that
when making the affordance judgment for the other person,
the judger never saw the action being performed. The person
just stood there. They never even lifted their arm, never mind
reaching for something above. And the judgments were still
accurate! Finally, as predicted, there was a systematic effect of
eyeheight on the affordances judgments both for judging one’s
own and the other persons’ maximum reach height.
The next study I want to look at dives deeper into
understanding the information we use to judge the affordances
of others and brings us into the realm of sports. In a study
published by Weast and colleagues in 20143
, the authors were
interested in what role the kinematic information (i.e. body
movements) we pick up from others might influence affordance
perception. For this, they looked at maximum reaching height
while jumping. So what was the highest point a person can
reach while jumping which if you think about it is highly
relevant in sports – for example, a basketball player detecting
whether an opponent will be able to block their shot or not.
The second affordance they looked at was horizontal long
jumping distance. If we can’t observe a person doing these
actions, how might we perceive what they are capable of?
To solve this problem, Runeson proposed the specification of
dynamics principle4
. He proposed that because our movement
is lawfully related to the forces that generated it, then it follows
that information about those forces should be available for a
perceiver through the movement kinematics (i.e., defined as the
features or properties of a moving object). So, for example, we
should be able to pick up information about a performer’s
ability to generate vertical propulsive force by viewing them
performing actions like walking or doing squats. Because their
movement kinematics for these actions will be directly related to
the forces generated, there is information about force in the
kinematics. We should also expect there to be different levels of
attunement to such kinematic information depending on
experience. For example, in athletes vs non-athletes.
To test these ideas, the authors created point-light videos of
actors performing four different actions: squatting, walking,
twisting, and standing on one leg. For those unfamiliar, a point
light display is created by putting little lights on the different
joints of a person and filming them moving in the dark. Check
it out here
5
. Critically, the only information in a point light
display is the movement kinematics (the relative motion between
the different body parts). The first prediction of the study was
that reports for both maximum reaching while jumping (RWJ)
and horizontal long-jumping distance (LJ) would be more
accurate after viewing point-light videos of actors performing
movements related to these affordances (squatting and walking),
as compared with movements unrelated to these affordances
(twisting and balancing on one leg).
The other manipulation in this study was sporting experience.
For the study, the authors recruited both athletes (from
basketball and soccer) and non-athletes. They predicted that the
athletes would be superior to non-athletes in perceiving the
affordance because long-term experience in a sport leads to
attunement to kinematic information (i.e. education of attention).
What was found? The key results are shown in Figure 8.1.
First, as predicted,  numbers (i.e., ratios between the
perceived and actual value) for squatting and walking were
closer to 1 than ratios for twisting and balancing on one leg,
confirming the author’s prediction that perceptual reports would
be more accurate after movements related to performing a
vertical and a horizontal jump were viewed. In terms of
sporting experience, the main prediction was also supported by
the results. For both the jump-to-reach and long jump
judgments athletes were significantly more accurate in perceiving
the affordances of others than non-athletes.

A comparison of a bar graph Description automatically

generated|
Figure 8.1 – Perception of Affordance of Reaching
While Jumping (RWJ) in Others using Kinematic Information.
Based on data from Weast and colleagues3
In a 2018 study, Thomas and colleagues6 examined the jump
to reach affordance in more detail. One way that we could
perceive another person’s ability to do this would be to just
pick up the lower order affordances, that is their maximum
reach height while not jumping and their maximum jump
height while not reaching, and then just add these two together
in some cognitive process that combines them. But, if you think
about it, that is not likely to be very accurate because the act
of reaching is nested within the act of jumping. So, for
example, if I swung my arms really hard to get higher in the
air on my jump that might impair the timing and accuracy of
my reach to the object while in the air. As we saw in Chapter
6 in my critique of the common practice of task decomposition
in training the two things need to work together.
The alternative in the ecological approach is that we just pick
up the higher-order affordance (jumping to reach) directly. We
don’t compute it from its lower order parts, much in the same
way as we don’t compute tau by getting an object’s distance
and then dividing it by it’s speed. We just pick up the tau
ratio directly.
To test this idea, Thomas and colleagues asked participants to
make separate estimates of three different things: (i) the
maximum height a person standing in front of them could
reach without jumping, (ii) the maximum height they could
jump without reaching, and (iii) the maximum height they could
reach while jumping. They also made these judgments for
themselves. In all cases, they moved a target hanging from the
ceiling to the judged maximum height.
What was found? The main comparison of interest in the study
was how participants judge the affordance of jumping to reach
compared to the additive model estimate based on combining
the lower-order reaching and jumping affordances. For both
judgments of one’s own ability and the ability of another
person, there were significant differences between the two. The
authors concluded that:
The results also support Gibson’s notion of information. The
affordance of jump-reaching height is specified by ambient,
optical information independent of related affordances.
Regardless of whether jump-reaching affordances for the self or
for another person are perceived, lower-order affordances were
not cognitively combined with a linear function to produce
them. Furthermore, even when the lower-order properties are
themselves affordances, they did not additively combine to
produce the affordance investigated in this task
A similar effect can be seen in the study of the perception of
maximum throwing distance for others published by Ji and Pan
in 20197
. Here participants were asked to pick a ball for
another person that would result in their furthest throw. Again,
they were very good at this and the affordance judgment was
again based on higher-order information about the ball’s mass
and size.
Finally, this concept was further investigated in a study by
Wagman and colleagues (2018)8 . Performing a given action
(e.g., changing a light bulb) often requires performing a
subordinate action (e.g., climbing a ladder), which itself requires
performing additional subordinate actions (e.g., stepping on an
individual rung of the ladder), and so on. In terms of
perceiving the affordances of others, if we are sensitive to
nested affordances, this means that not only can we use direct
information to detect the action capacities of another person
(which we have already seen many examples of so far in this
chapter) but we can also pick up impending changes in action
capabilities.
To investigate this, the authors looked at perceiving the
affordances of reachability of another person under different
nested task constraints to test whether the perceiver would
exhibit prospective sensitivity not only to the actor’s current
action capabilities but also to impending changes in the actor’s
action capabilities. The conditions were to reach; (i) with the
fingertips of the outstretched hand while standing on the floor,
(ii) with the fingertips of the outstretched hand while standing
on a (visible) stepstool, or (iii) with the distal end of a (visible)
stick to be held in the actor’s outstretched arm. Actors did not
actually perform any of these behaviors or any of the required
subordinate behaviors (e.g., standing on the step stool, picking
up the stick, raising his or her arm). Rather, the actor merely
stood next to the step stool or the stick (depending on the
condition) with his or her arms at his or her sides while the
perceiver reported the actor’s maximum reaching height in that
condition.
What was found? Pi numbers were almost exactly 1 for all
three conditions suggesting that we can perceive these types of
nested affordances directly again from information in the
environment.
For me, this brings into question the validity of using things like
functional movement screens to assess athletes. For most of
these, we are measuring simple, lower-order affordances rather
than the higher-order nested ones we see in most sports.
As an example of how this can be applied to sports, consider
how Greg Souders coached an MMA athlete with short arms,
D:
I don't know. We haven't measured his arms, but they're a
little bit short. He does not strangle with his arms from the
back very well. His legs are a little bit longer than his torso.
So, let's get our legs involved and have the legs produce more
pressure than the arms. So, we just organized around that. He
was really struggling and how this appeared when we play
submission-only events. Submission only is there's no score.
And then if the time limit goes off and no players submit each
other, they go into an overtime situation where they start on
each other's backs. D was not winning. He could not strangle
these people.
So, we're like: let's get your legs involved. They're better.
They're longer and they're better for strangulation. So, what we
did was he taught him how to create what we call the “arm in
arm out” scenario. The idea is that in order to strangle with
the legs, we want a head and arm included in the system. So,
if I'm able to take one of the arms and connect it to the head
and close something around this space, I can cause
compression at the shoulder on one side and the neck on the
other. If we compress these two. areas together this
compresses the carotid arteries, stopping blood flow to the
brain, and they go unconscious.
So as you can see, the arms are not long enough to create
meaningful compression. Now some people can, but this can be
difficult for a guy with short arms. But if we look at my legs,
I'm able to get my legs around this area, very easily. I can
close my legs around that structure and squeeze to a particular
effect.

As a coach, you can and should be looking for the action

opportunities that are available to your athletes. Based on this
we can do two things – work on increasing these through
appropriate action capacity training and or guide the athlete to
movement solutions that will be possible within their action
boundaries (e.g. Greg guiding D to a movement solution
involving the legs).
Next, let’s take a different view of some tried and true methods
for coaching an athlete to achieve an affordance.
10
USING IMAGERY, DEMONSTRATION
& VERBAL INSTRUCTION
A common misconception about being an ecological coach is
that it reduces the size of the coach’s toolbox taking away
some of the coaching methods that, on the surface, do not
seem to fit with this approach. For example, using mental
imagery (aka visualization), explicit verbal instruction and
demonstration. Traditionally, all these methods have been
conceptualized from an information-processing point of view.
Imagery and demonstrations are used to build up an athlete’s
internal representations (or mental models) of a skill. While
explicit instruction is for prescribing an ideal movement solution.
But, as I hope to convince you in this chapter, we don’t need
to throw the baby out with the bath water! All of these
evidenced-based methods can be equally well understood (and
implemented) from an ecological approach. Let’s start with
mental imagery.
An Ecological, Enactive Account of the Use of Mental Imagery
(MI) in Sports
As most people know, there is a lot of research demonstrating
the benefits of mental imagery (i.e., trying to imagine or
visualize a successful performance outcome)1 and several
famous athletes have commented that this was critical to their
success. The classic example comes from golf great Jack
Nicklaus2
:
I never hit a shot, not even in practice, without having a very
sharp, in-focus picture of it in my head. First, I see the ball
where I want it to finish, nice and white and sitting up high
on the bright green grass. Then, the scene quickly changes,
and I see the ball going there: its path, trajectory, and shape,
even its behavior on landing. Then there is a sort of fade-out,
and the next scene shows me making the kind of swing that
will turn the previous images into reality.
In a meta-analysis including data from 36 sports studies,
Lindsay et al
3 found that MI had a significant effect on
performance, with a medium effect size (g=0.476). MI combined
with physical practice (PP) had a significantly larger effect than
using just PP or MI alone. As a specific example of one of
these studies, Smith et al4 compared the different types of
training for improving bunker shot performance in golf. There
were three training groups: MI (visualized 15 bunker shots per
week for 6 weeks), PP (physically performed the same number
of shots), and MI+PP (did a combination of both). MI+PP
resulted in 22% increase in performance, PP alone resulted in a
13% increase, and MI alone resulted in a 7% increase.
MI in sports is traditionally thought of as an act of mentally
simulating an action by manipulating an internal representation
in memory. The proposed mechanism for why it has a benefit
on performance is functional equivalence - when we
appropriately imagine performing a skill, the same areas in are
brain become active that are active when we perform the same
skill. This activity serves to strengthen the connections in the
brain for performing the action. On the surface, this is a
challenge for the ecological approach because there is no
currently available information from the environment to support
direct perception while an athlete is using MI. At least it seems
like there isn’t!
How might we understand this important ability using concepts
like affordances and skilled intentionality instead of mental
representations? What seems to be problematic for an ecological
account is that direct perception is restricted to the present,
whereas imagery is seemingly accessing things from the past
(using memory representations) so we can act in the future.
This idea that direct perception is locked in the present is
incorrect and is something that Gibson5 wrote about:
A perceptual system that has become sensitized to certain
invariants (information) and can extract them from the stimulus
flux can also operate without the constraint of the stimulus flux.
What he is saying here is direct perception is not just about
picking up currently available sensory information and acting on
it in the present – it is about establishing a relationship with
one’s environment. And this relationship, this skillful coordination
with the landscape of affordances in the environment, has both
a past and a future. For example, through extensive practice,
Jack Nicklaus established a relationship with the environment of
the golf course. He educated his attention to specifying
information and established effective control laws for making golf
shots. He developed skilled intentionality to maintain an optimal
grip on multiple affordances (e.g., hitting the ball close to the
hole, laying up, hitting to the middle of the green - see Chapter
9 of my second book for more detail on this concept).

In the skilled intentionality framework, mental imagery is simply

an extension of this relationship. It is maintaining a state of
action readiness to enact affordances, a coordination with one’s
environment even when the information used to establish that
coordination is not present. As Gibson first proposed, once a
perceptual system becomes attuned to the information from the
environment, this skillful attunement can be used in the
absence of the information. What we imagine when engaged in
mental imagery is dependent both on our abilities and the
affordances offered by the environment and therefore they are
still coupled. In this sense, MI does not require representations
of the external environment.
The ecological account proposes that MI is not anything special
– it is part and parcel with just perceiving and acting. Having
skilled intentionality involves maintaining a grip on a set of
affordances (which shape how you will act in the future)
without always having current information present. It does not
have to be some complex mental process we engage in
involving the use of internal representations. but rather just
from a skillful coordination we have developed with our
environment.

An Ecological Approach to Using Mental Imagery (MI) in

Practice
As discussed in detail in a paper by Lindsay and colleagues3,
the design principles of nonlinear pedagogy (NLP) provide
practical guidelines on how practitioners could help learners
develop attributes such as adaptability using mental imagery.
For example, rather than using an MI script that depicts a
specific movement form, it could incorporate details of nets that
differ in height across tennis practice (e.g., manipulated task
constraints) and tell learners to focus on the trajectory of the
ball as it goes over the net (i.e. movement outcomes instead of
specifics of the technique). This would require learners to adapt
movements continuously to the different net heights to achieve
performance outcomes. Further, instead of having learners use
MI to hit to a target, details could describe an actual opponent
e.g., specifying information related to their body position. Such
details would capture the design principle of representativeness
discussed in Chapter 4.
Given that MI training occurs within a simulated environment,
practitioners and learners can ‘manipulate’ task constraints
through the scripts delivered during practice. For example,
when looking to develop tennis return shots, scripts could be
designed to ‘manipulate’ the height of the net or the position of
an opponent that changes as practice progresses. For example,
the MI script could be: “As you set up to receive serve, take
note of the height of the net, it is higher than usual. As the
serve approaches you and you set up for your return shot,
take note of where your opponent is standing, try and hit the
ball up and over the net to the space outside your opponent’s
reach”. In this way, the script aims to include information to
provide a performance context for the learner to explore, rather
than describing a specific movement pattern to reproduce.
The authors suggest that MI could also utilize analogy-based
instructions (which as discussed in Chapter 7 of my first book,
can be more effective in supporting self-organization). For
example, imagery scripts may include statements like the
following: “Imagine striking the ball in the shape of a rainbow”
or “Shoot the ball like you are reaching into a cookie jar on
the shelf”. Again, coaches should design MI scripts that detail
movement outcomes rather than a prescription of a specific
movement form.
In this paper, Lindsay and colleagues also include a study
comparing the effectiveness of linear/prescriptive and nonlinear
mental imagery training for learning the power clean (PC) in
weightlifting. This design of this study mirrors the study by the
same authors (discussed in Chapter 6) which compared
linear/prescriptive and nonlinear physical practice for the same
skill6
. Fourteen participants were split evenly into two groups.
Both groups participated in eight MI practice sessions
(approximately 30 minutes in duration) over a period of four
weeks. When in the start position of the movement, participants
would listen to either a linear or nonlinear recorded script that
guided them through the five MI trials.
The linear MI training was designed based on the idea that
during skill acquisition learners should be directed toward an
‘optimal’ technique and this is best achieved through repetitive
practice. Therefore, linear MI scripts involved details of what is
considered an ‘optimal’ PC technique and were movement form
(rather than movement outcome) focused. Participants in the
linear condition received prescriptive MI scripts according to
different phases of the lift. For example, the second pull phase:
“As your lower body extends imagine being right up on your
toes” and the turnover of the barbell onto the shoulders:
“Imagine bending your elbows and pulling your body under the
bar”.
For the nonlinear condition, scripts were focused on the
movement outcomes, manipulated constraints, and utilized
analogy-based instructions to guide self-organization. This
included MI instructions such as “imagine trying to flick the
bottom of your shirt as you pull upwards” and “imagine
exploding upwards like you are jumping straight up”. The
manipulation of task constraints in the nonlinear MI scripts
included chalk on the barbell and poles in front of the barbell
(i.e. the same ones used in the CLA condition in the physical
practice study described in Chapter 6). Participants were told to
either imagine moving the bar without hitting the poles in front
of them or imagine leaving a chalk mark on their thighs with
the barbell. These constraints were also physically present
during MI practice.
What was found? Results were highly similar to the physical
practice study by the same authors (discussed in Chapter 6).
Equivalent improvements in performance (assessed in terms of
forward bar displacement) were found for both MI groups.
Regarding the quantity of exploration, no significant differences
were observed between linear and nonlinear conditions – the
cluster analysis again revealed that both groups used a similar
number of different techniques during training. The authors
conclude:
The present study highlights the potential benefits of utilizing a
NLP approach to MI to encourage learners to explore
movement solutions that align with a learner's capabilities
without negatively impacting performance. It may be beneficial
for MI practitioners to consider designing practice that allows
deviations from prescribed technical models to facilitate learners'
inherent exploration of individual task solutions.
Guidelines for using MI in Practice
To sum up this section, here are the basic guidelines I would
give for using an ecological approach to MI:
Use co-design in creating the MI script. That is, have the
athlete write down everything they experience when executing
the skill.
Keep it short and simple – the script should detail an action
of about 15-20 seconds, maximum
Combine MI with physical training of the same skill, where
possible. However, it should be noted that one area in which
MI is particularly effective is during injury rehabilitation7
when the ability to do PP is limited.
Image from the 1
st person not the 3
rd person perspective.
The former ensures the information used will be more
representative of what will be used in the game/competition.
Focus on movement outcomes, not movement technique
Manipulate constraints in MI scripts
Adapt the script as the athlete improves, in a similar manner
to how we adapt constraints, variability, etc during PP
Try to make the environment where the MI will be
performed as similar to what will be imaged as possible (e.g.,
physically include any constraints that will be present in MI
script).

Demonstration as Rate Enhancer

Demonstrating a skill to the athlete (and the observational
learning it evokes) has long been a popular and effective
training method used by coaches8
. Again, demonstration has
traditionally been understood in terms of an internal model
approach to skill. Similar to what was discussed above for
imagery, observing someone else performing an action is
thought to activate (and thereby, strengthen) the same areas of
the brain (i.e. internal representations) we use when we execute
the skill ourselves. But again, demonstration and observational
learning can also be understood using an ecological approach.
One way of conceptualizing the role of demonstration in skill
development in an ecological framework is to think of it as a
rate enhancer. As the name implies, a rate enhancer is the
opposite of a rate limiter. It is a constraint that serves to speed
the rate of skill acquisition by helping the learner find an
effective movement solution more quickly by guiding them to a
particular area of the solution space. For example, Scully and
Newell (1985)9 have proposed that observing a model would
have its greatest effects early on because it would speed the
rate at which the learner assembles a new movement topology
(i.e., during the coordination phase of learning) and allow them
to more quickly to advance to the next stage of skill
development where the main focus is the parameterization of
the movement pattern, (what Newell calls the control stage).
Another way of thinking about this is that the demonstration
makes it such that the learner has to spend less time broadly
searching for a movement solution and can spend more time
working on refining the solution. In Newell’s terminology, it
accelerates the shift from coordination to control. Demonstration
can also help a new learner find a temporary movement
solution that, although it may be very imprecise, slow, and
clunky, allows them to achieve some initial progress. This
solution is eventually discarded when the learner becomes more
experienced. A good example of such rate enhancement is
Bernstein’s concept of freezing degrees of freedom. As described
in Chapter 4 of my first book, not using some degrees of
freedom early in learning (e.g. not bending your elbow on a
volleyball serve) allows for some temporary success, with the
likelihood that these degrees of freedom will be “freed” later on.
At this point, you might be wondering: if demonstration can act
as a rate enhancer in this way, wouldn’t it be even better to
prescribe an initial movement solution they could use via verbal
instruction? Good question. The difference between the two is
that (when used in the right way) demonstration can support
self-organization while prescription typically does not. As will be
detailed in the guidelines section below, two of the keys to
using “ecological demonstration” as a coach are: (1) not giving
a lot of narration (particularly, about the details of the
technique, body movements, and positions) while you are doing
it, and (2) demonstrating the skill while varying the task
constraints. This creates two important features that support
self-organization. First, it supports the development of invariants
in movement. In viewing the same skill demonstrated under
differing constraints, the athlete can pick what remains the
same in the movement pattern. Second, it supports adaptation
to individual constraints. The athlete can take what is relevant
to them out of the demonstration rather than being told “watch
this” or “do this”.
The effect of demonstration on the early stages of motor
learning was investigated in a study by Horn and colleagues in
2007
10
. The main question of interest was: does
demonstration serve to constrain the movement solution
adopted by the learner and accelerate the coordination phase?
Sixteen participants were asked to learn a novel throwing
movement. Imagine executing a baseball pitch except that you
must release the ball backhanded (like you are slapping
someone in the face with the back of your hand).
Prior to training, all participants were given two instructions
focused on the required outcome of the movement. First, they
were told that the back of the hand should face the target at
ball release. Second, to prevent underarm throwing, the
participants were told that the ball should be released from a
position in which the wrist was located above the height of the
elbow. After three pre-test trials, participants were split evenly
into two groups: the control group and model group. The
control kept practicing the throwing tasks based on the verbal
instructions. The model group watched a life-size video of five
demonstrations (with no verbal commentary) of a model
executing the throw.
What was found? In terms of performance, the model group
showed an immediate, significant increase in ball speed over the
first three trails after observing the model. There were no
changes in speed after this immediate boost. The control group
showed no significant change in ball speed throughout the
entire acquisition phase.
In terms of the movement solution, Figure 10.1 shows the
results for one participant in the demonstration group (left
panels) and one participant in the control group (right panels).
This figure shows the solution space created by looking at the
elbow and shoulder angles. Immediately, in the first 3 trials
after viewing the video, the participant in the demonstration
group produced a very different movement topology than was
seen in the pre-tests (for both groups) and in the early trials
of training for the control group (bottom right panel). This
movement topology, which was maintained in the rest of the
training, was also similar to that of the person doing the
demonstration. So, in other words, the demonstration served to
quickly guide the performer to an effective area of the solution
space. As the authors state:
It seems like the main effect is that observing the
demonstration seemed to immediately constrain the model
group’s coordination, most likely shifting it to a different (and
based on the ball speed results) more optimal location in the
perceptual-motor workspace. So, demonstration does appear to
be a rate enhancer for skill acquisition, in particular in terms of
the coordination phase.
Figure 10.1: Movement solutions (elbow and shoulder angles)
used in a reverse throwing task. Left panels: Before and after
watching a demonstration. Right panels: The same two time
points with no demonstration. Based on data from Horn et al.
10

Related to the issue of what a coach says during a

demonstration, Al-Abood et al. (2002)
11 investigated the
relationship between instruction type, demonstration, and visual
search (i.e. gaze behavior) for learning a free throw in
basketball. Sixteen novice basketball players were split into two
equal groups. All participants viewed video demonstrations of an
expert model making free throws. For the movement form
group, this was accompanied by instructions that directed
attention to the model’s movement patterns. For the movement
outcome group, the video demonstration was accompanied by
instruction that directed attention to the ball trajectory relative
to the basket. Tests of free throw performance were given
immediately before and after the demonstrations.
What was found? In terms of performance, the movement
outcome group exhibited a significant improvement in score
between pre-test and post-test. The movement form group
showed no such improvement. In terms of gaze behavior, while
watching the video, the movement effects group made more
fixations (15) of longer duration (446 msec
) than the
movement form group (11, 329 msec). Finally, the movement
form group spent proportionally less time viewing areas outside
the model’s body than the movement effects group. This
supports the idea that narrating a demonstration with a lot of
“technical instructions” serves to over-constrain the exploration
of the solution space – in this case, the search of the
information space with gaze.

Guidelines for Using Demonstration

Demonstrate a skill under different constraints to allow the
athlete to pick up the invariants.
Be quiet! Do not give lots of technically focused commentary
while doing the demonstration.
Before doing the demonstration use language like “you could
try” instead of “you should (or must)..”. We are trying to
guide self-organization, not prescribe a solution.
Using the task simplification principles discussed in Chapter 6
to make sure the demonstration is appropriate for the skill
level/age of the athlete.
Athletes benefit more from video demonstrations that are
presented at real-time as compared to slow-motion
presentations12.
Take advantage of the benefits of self-modeling13 (i.e.
watching yourself perform an action). For example, take video
examples of effective executions from a previous game or
competition and show the athlete. Again, be sure not to
draw their attention to specifics of the technique (through
your commentary) but focus on the movement outcome
instead.

Guiding search and self-organization with verbal instructions

While we tend to want to treat the use of verbal instructions
as something qualitatively different than other coaching methods
14, as first proposed by Newell and Ranganathan (2010) 15
,I
think it makes much more sense to treat them as just another
type of task constraint. Like all constraints, the purpose of a
verbal instruction given by a coach should be to take away (or
de-stabilize) a particular movement solution and encourage the
athlete to search for a different one. So, instead of giving or
prescribing a solution to the athlete, we are guiding them in
their exploration. The specifics of the use of language are very
important here. The instruction “Why don’t you try bending
your knees” is an attempt to guide a performer away from a
solution that involves keeping their legs straight (that the coach
believes will be ineffective) and encouraging them to explore a
different area of the solution space, that they may or may not
stay in. To quote Newell and Ranganathan :
Instructions may help to make the search process more
efficient by eliminating non-solutions to the task - especially
when tasks become more complex. Instructions, therefore, have
the potential to guide and constrain the search process in the
learning task.
Or as Otte and colleagues16 state:
Verbal information should instead be mainly used as an
augmented informational constraint to guide an athlete’s search
activities. When learning to learn to move, it is the athlete who
needs to use information to solve a performance problem and
not the coach providing verbal information to solve the problem
for an athlete
If you look at research on different types of verbal instruction
and which ones lead to more effective motor learning, the
results are consistent with this conceptualization. Using external
focus of attention instructions (e.g., “focus on the ball leaving
the bat” or “pushing off from the ground’) encourages an
athlete to focus on the outcomes of their movements (i.e. the
essential variables discussed in Chapter 6) while using, less
effective, internal focus of attention instructions (e.g., “focus on
the snap of your wrist” or “pushing off with your lead foot”)
encourage a focus on technique (i.e. the elemental variables).
Finally, analogy instructions get an athlete to focus on the
invariant properties (the topology of the movement) rather than
the details of how that movement was produced. A few studies
provide support for this ecological view of the role of coach
instructions.
For example, using constraints to guide the process of
self-organization was investigated in a study by Lafe and Newell
(2022)17. Thirty-two participants (split equally into two groups)
were asked to learn a two-handed force production task. The
goal force on each trial (displayed as a red line on a computer
screen) was given by a function that depended on the ratio of
force in the two hands, not the total force. The force being
generated by the participant was displayed as a yellow line.
Both groups were told that their goal was to adjust the force
in their hands so that the two lines on the display fell on top
of each other. Before the learning phase began, the instruction
group was given the additional instruction: “The error between
the red and yellow lines will be reduced by maintaining equal
pressure between each hand.”
What was found? For the instruction group, there was a
significant reduction in performance error (the difference
between the actual and target force output) across the three
blocks of training. There was no significant change in error for
the no-instruction group. How did the instruction influence the
search for a solution? Figure 10.2 (top panel) shows the
solution space (created by plotting the force in the left hand vs
the force in the right hand). As can be seen in this figure, the
instruction group explored a much smaller area of the solution
space, but more consistently produced effective solutions with
equal force in the two hands (which is, of course, why they
had a lower performance error). The no-instruction group
explored a much larger area of the space but also produced
more ineffective solutions.
What we can see here is the essential tradeoff that a coach
must face when they decide to give a cue or instruction. When
we give a very specific instruction about the solution we are
looking for (as was the case in this study) they become
proficient at the task more quickly but at the cost of not
exploring some parts of the solution space that may work
better for them in the long run. For example, referring to
Figure 10.2, solutions with 6-12N in each hand.
A diagram of a graph Description automatically generated
A graph with black dots Description automatically generated
Figure 10.2. Effect of instructions on performance and search in
a force production task (top). Relationship between total
performance error and size of the search area for the
Instructions Group (bottom). Based on data from Lafe &
Newell17
This point is further emphasized when we look at the
relationship between the size of the search area and the
performance error found in this study, shown in Figure 10.2
(bottom panel). Even though participants in the instruction
group searched less of the solution space overall, it was still the
case that a greater search (i.e., more movement variability)
resulted in a lower total performance error.
Another example of how instructions can constrain the search
for a movement solution can be seen in the volleyball study by
Singh et al. (2022)18. Twenty experienced volleyball players
(who competed at the collegiate and/or professional level) all
completed a serving task under three different instruction
conditions, in a within-subjects design. Players performed
overhand serves towards a 120 cm diameter target on the
other side of the court. The target was subdivided into smaller
regions with more points awarded for serves closer to a central
bullseye which was 30cm in diameter. The three instructions
were: internal (“focus on your hand while contacting the ball”),
proximal external (“focus on contacting the middle of the ball”),
and distal external (
“focus on hitting the bullseye”). Previous
research has shown that, for experienced athletes, instructions
that direct attention externally to a location further from the
body (distal) are more effective than instructions that direct
attention to external locations close to the body (proximal) 19.
From an ecological perspective, the further our attention is from
our body, the less likely we are to try to do top-down control
of our movements (as opposed to allowing self-organization to
occur).
Along with comparing the performance score, for the three
groups, the authors performed an uncontrolled manifold (UCM)
analysis, which separated the total movement variability (from
trial to trial) into “good variability” which are variations that still
meet the performance goal of hitting the target and “bad
variability” which take the performer away from the
performance goal (e.g., because the ball is hit too short).
What was found? First, there was a significant effect of
instruction on serving accuracy, with the external distal
instructions resulting in the best performance. Mean accuracy
points per serve were as follows: proximal-distal (1.82), proximal
external (1.42), internal focus (1.27). Figure 10.3 shows the
results of the UCM analysis. As can be seen in this figure, the
type of instruction did not affect the total movement variability.
Instead, it improved the efficiency of search. That is participants
in the external-distal conditions spent more time searching the
space of effective solutions that would achieve the task goal (i.e.,
good variability) and less time searching through the space of
solutions that would not achieve the task goal (i.e., bad
variability).

Guidelines for Using Verbal Instructions

Where possible direct an athlete’s attention to the effect of
their movements (i.e., externally) as opposed to movement
technique or form itself.
Use instructions to send your athletes toward a particular
area of solution space (so that they can explore) as opposed
to sending them to a particular solution within the solution
space (i.e., prescription)
Use instructions sparingly in the early stages of learning a
skill. As proposed by Otte et al
16 instructions should be
limited to a minimum in the Coordination Training and Skill
Adaptability Training stages, because in these early stages of
learning, we want broad exploration of the solution space.

A graph of different sizes and colors Description automatically

generated with medium confidence
Figure 10.3 – Effect of instruction type on “good” and “bad”
variability in volleyball serving. Based on data from Singh et al.
(2022) 18
.

11
WHAT TO DO NEXT: ADAPTING
PRACTICE
T he taste isn’t quite right. Add a pinch more of this or a
teaspoon more of that. Feels a bit spongy to the touch –
better cook 5 minutes longer. Something is off - time to pour
the whole thing down the drain and start over! A key to being
a master chef is going beyond what was written in the recipe
on paper (i.e., the precise amount of each ingredient and
cooking time), trusting your senses, and being adaptable. That
is, being willing to deviate from the original formulation of the
plan and go in a slightly (or very!) different direction. The
same applies to being an ecological coach. Designing effective
practices is only half the battle. Effectively using coaching
methods like the CLA requires learning the art of “what to do
next”.
Why is this so important? Because we are dealing with a
complex system! Accepting the ecological dynamics account of
skill development means that we need to have a different view
of causality. We need to abandon the idea that we can
perfectly predict what effect a constraint will have on our
athlete’s performance. Constraints do not cause behaviors.
Remember, a complex system is non-deterministic – that is to
say the outputs are not perfectly predictable from the inputs.
When we add a constraint to practice (e.g., an extra player or
a smaller net), the new task constraint is going to interact with
all the other constraints present in the system (individual,
environmental, and the other task constraints), making the
behavior that emerges less predictable.
If you adopt the CLA as a coach, I can almost guarantee you
will experience this for yourself! You will design an incredibly
clever constraint manipulation on a piece of paper or a
whiteboard and what emerges on the field will be completely
different than you expected. One of the most common
examples of this occurs when we try to add a simple rule as a
constraint. For example, a common affordance that I see
coaches of team sports wanting to encourage is more
movement or passing of the ball on offense. Like many
coaches, when first faced with this issue I thought the solution
was easy: add a constraint that the team must “pass the ball 3
times before they shoot”.
If you have tried this, you have probably seen all the different
and creative ways players come up with to shortcut this rule.
Players make three 1-foot passes to each other at the top of
the court of field and then start playing. Players pass the ball
off the backboard to themselves. Players pass the ball off an
opponent’s back to themselves. And so on. “You never said
what kind of pass it had to be coach!” This is a case where
the opportunity for action (the affordance) I had hoped they
would take (passing the ball to teammates on the other side of
the court or close to the basket to create scoring opportunities)
did NOT emerge from my clever constraint.
What should a coach do next when faced with a similar
situation? Are there any guidelines we can give to a coach in
this situation? Was there a better way to design the constraint
in the first place? I will detail what I did with my basketball
activity later in this chapter but first let’s look more at some of
the questions that coaches can face when using an ecological
approach.
“Unanswered” Questions about using CLA
The “fuzziness” surrounding the details of using an ecological
approach to coaching was recently discussed in a paper by
Kennedy & O’Brien (2024)1
. The authors express their main
frustration as follows:
Despite the reassurances of leading academics that coaches
would develop expertise through an exploratory trial-and-error
learning process in which they become better at “choosing
which constraints to use, how much to manipulate them by,
and how long to leave them in place”, further guidance may be
beneficial to help coaches avoid a process of throwing
constraints against the wall to see what sticks.
The authors identify three key questions, in particular, that they
feel require further guidance.
Q1: How should ecological designers understand and address
athlete errors?
In the traditional, prescriptive approach to coaching, there is
heavy emphasis placed on identifying and correcting “errors” in
our athlete’s performance. Kennedy and O’Brien define an error
as: “a momentary mismatch (or difference) between an
intended movement pattern and a produced movement
pattern”. Implicit in this definition is the assumption that the
athlete intends to produce the ideal movement pattern (or
some variant of it adjusted for the specific conditions) given to
them by their coach. The specific example they give is: what
should a field hockey coach do if one of their players is
looking down at the ball while dribbling? In the traditional
approach, this is a clear error that a coach should immediately
step in and correct with some verbal instruction (e.g., “keep
your head up”).
How do we think about “errors” in the ecological approach?
First off, an athlete does NOT intend to produce a specific
movement pattern. Instead, they intend to produce a specific
performance outcome that aligns with our goals. We intend to
make a 3-point shot in basketball because our team is way
behind in the game. We intend to stickhandle through
defenders in field hockey to get an open shot on the net. The
specific movement pattern that we use to achieve these goals
emerges through a process of self-organization with respect to
the constraints.
My favorite current example of the difference between the two
comes from the last Olympic Games in Paris. In the gold medal
final, Team USA basketball was in a very close game with
France. In the final quarter, Steph Curry made four 3-point
shots to seal the victory. For those that have not seen it his
final shot, it can be viewed here
2
. Do we really think that
Curry intended to produce that movement pattern? A fadeaway,
one-footed, off-balance shot over two defenders! Even if that
was his intention (which we could never know), no basketball
coach in the world would teach a player to shoot that way! So,
was that an “error”? No, it wasn’t!
One of the reasons that Steph Curry is one of the greatest
3-point shooters of all time is his incredible dexterity. His ability
to effectively adapt his movement solution to constraints is
unmatched. For example, on the shot shown in the video, he
needed to adapt his jump shot solution to the following
constraints: (1) fatigue (he had played nearly a full game), (2)
pressure (this was a chance for him to win his first gold
medal), (3) two tall defenders guarding him, (4) movement to
his right after making the behind the back dribble. I find it
very hard to believe that that was the movement pattern he
intended to produce when he received the ball.
To be successful, an athlete is often going to have to a
produce a movement pattern that is very different from the
ideal (intended) one they were taught by their coach. If you
need another example, take a look at Patrick Mahomes’
behind-the-back pass from this year’s NFL preseason3 Again,
was this really how he intended to pass a football or was it an
emergent adaptive solution? Is this an error?
Let’s return to the example the authors give (a field hockey
player looking down while stickhandling). First off, I can’t let
this pass without commenting on the irony here! The main
method we use to teach the fundamental technique of
stickhandling in the traditional decomposition approach (i.e.,
dribbling through cones) is likely to be the main culprit in
producing the error they describe. But either way, from an
ecological perspective, what should we do if we saw this?
What we have here is a movement solution that is missing one
of the key invariants (or essential variables) for stickhandling:
the player is in a position to pick up information from the
other players on the field and the spaces between them. As
with all invariants, there are different specific movement
techniques or forms (i.e. elemental variables) that could achieve
this (e.g. looking down at the ground only intermittently,
keeping one’s eyes up the entire time, moving the ball further
in front of you, etc.). As a coach, what should we do if the
player’s movement solution is missing this invariant?
First off, we need to consider what stage of learning the player
is in. If they are novices, it is possible that looking down is a
necessary rate enhancer for them to achieve some initial
proficiency. But at some point, we would want to add a
constraint that encourages them to move away from this
solution. But when and how do we do this? Here is one of
the first tips I will give in this chapter…

Using the 70% Rule as a Guide

Imagine that we set up a practice activity where novice players
have to dribble through three opponents who have their sticks
out, but the defenders are not allowed to move their sticks to
try and steal the ball (i.e., task simplification by constraining the
opponent). Our hypothetical novice player can do this
successfully on 4/10 (40%) attempts. In other words, they are
not achieving the success rate we are looking for to challenge
them optimally4
. The task is too difficult. As a coach, we
could give them another set of reps to see if they can achieve
a higher success rate or we could make the task easier (e.g.
by taking away one of the defenders, increasing the spacing
between them, or by taking away their sticks). Now, let’s say,
with some more practice, they can successfully get through the
defenders without losing the ball on 9/10 (90%) of attempts.
Now they are performing above the optimal challenge level.
That is a signal to increase the difficulty (by changing the
constraints). Let’s consider two different cases:
Case #1 – The player is achieving a 90% success rate and is
looking down at the ground most of the time. Here we want
to add a constraint that takes away this solution and
encourages the player to get their eyes up (i.e., use the basic
logic of the CLA). For example, we could add a teammate who
moves into the play to receive a pass or allow the defenders to
move to steal the ball. We could also use the ChinUp goggles5
(i.e., visual occlusion) I discussed in my first book. We would
not want to add constraints like using a smaller, more bouncy
ball (e.g. a golf ball) or allow the defenders to swipe with their
sticks to steal the ball because both of these are likely to
encourage looking down. In other words, they are not
well-designed constraints for this specific situation.
Case #2 – The player is achieving a 90% success rate, and is
looking up most of the time, but they are letting the ball get
too far ahead of them (i.e. they aren’t protecting it very well).
Here we might want to add the constraint of using the golf
ball. The bouncier, more difficult to control the ball, might
discourage them from letting it get too far away from their
body.
So, to sum up, what do we do when we see the athlete using
a movement solution that is missing one of the key invariants
we have identified for our skill? Follow the 70% rule! If they
aren’t yet achieving that level of success, we shouldn’t do
anything about it as it may be serving as a rate enhancer. If
anything we might want to change the constraints to lower the
task difficulty. Once the athlete gets above the 70% success
rate, we want to add a constraint to make it more difficult. But
not just any constraints. Following the logic of the CLA, we
want to add a constraint that de-stabilizes (aka makes it very
difficult to achieve a high level of success in the task using) the
current movement solution and encourages the athlete to
explore.
Q2: What should ecological designers do when undesired skills
emerge?
The authors define an undesired skill as: “unwanted (or
unhelpful) movement patterns (or solutions) formed over time
(e.g. bad habits)”. So, in their terminology, an error is a
momentary, one-time thing whereas an undesired skill is more
permanent and stable. So, using the example from Chapter 6
in my first book, as I understand it, if Tim Tebow through a
pass that wasn’t a nice tight spiral that would be an error,
while if he held his hands too low when he dropped back to
pass that is an undesired skill. In the language of ecological
dynamics, an undesired skill is associated with a deeper, more
stable attractor.
The example the authors give is a basketball player who is
jumping in the air and then looking to make a pass. This is
an “undesired solution” because, although sometimes it can lead
to a successful pass, it increases the chance the player will
make a turnover. Once in the air, they have very little time to
find an open teammate and deliver the pass. If they come
back down to the ground without getting rid of the ball, they
will be called for traveling and their team will lose the ball.
As noted by the authors, in traditional coaching, this undesired
movement solution would be corrected via “the delivery of
direct instruction of ideal movement patterns and repetitive
trial-and-error practice” early and often. The fact that the player
successfully made a pass using this technique on some attempts
would be ignored and they would be told “Don’t jump in the
air before you pass”, which they would then be asked to
repeat multiple times.
The problem with this approach, from an ecological perspective,
is we are ignoring the possibility that “jumping in the air to
pass” may be a functionally adaptive movement solution for this
individual player. It is possible that their individual constraints
allow them to use this movement solution successfully under
some constraints, whereas other players cannot. Even if we
truly believed that no player could master this particular
movement solution, where does it end? What else is an
“undesired skill”? Poor footwork? Being off balance when you
shoot? Shooting over multiple defenders? This is a very slippery
slope that quickly leads us back to prescribing the one, ideal
technique and coaching the dexterity out of players like Steph
Curry and Patrick Mahomes.
As a coach, we should consider any movement solution that
contains the invariant features we have identified as being a
potentially adaptive solution for the particular athlete we are
working with. Instead of trying to get rid of it, we should be
using the CLA to challenge it. That is, we should keep
changing the constraints to make it more and more difficult to
use this solution, without eliminating it as a possibility.
Imagine we start with a basic 3 v 3 small-sided game and our
hypothetical player is completing roughly 90% of his “jump in
the air” passes. We can add more players on the floor, allow
different defenses (e.g. double teams, zone), add a shot clock,
play an unbalanced game with more defenders, etc. If the
player is still achieving a high success rate with this solution,
then we should let it be! If their success rate goes down
during this process (e.g. it drops to 50% when we go to
playing 5v5), this may alone encourage them to find a new
solution. Or, at this point, we could use the CLA to move them
away from it e.g., by adding a rule that the defensive team
gets 2 points for every turnover. Let me address a few
questions I commonly get at this point.
1)Isn’t there a danger in creating Tim Tebows?
That is, isn’t it likely that we will create strong attractors for
movement solutions that work early in the learning process but
will not be successful later on (and will be more difficult to
change at that point)? I am not worried about this happening
if we continue to move practice up the representativeness
continuum and use the 70% rule as a guide. Attractors for
ineffective movement solutions do not arise because we fail to
step in to prescriptively correct them. Instead, they arise
because we spend too much time practicing in conditions of
low representativeness with a very high-performance success
rate. Without having been there to see it, I can almost
guarantee you that Tim Tebow did a lot of practice with no
defensive pressure in which he made nearly 100% of his
throws. In other words, his movement solution of holding the
ball too low was not sufficiently challenged.
2)But isn’t an “undesired skill” the same thing as an invariant?
Why do I call keeping your eyes up while you stickhandle an
invariant but not passing a basketball with both your feet on
the ground? This is an important point. We need to be very
stringent when we label something an invariant! By definition: if
there is one athlete that can achieve success using that
movement solution then it is NOT invariant (i.e., something that
must be present). The key to identifying real invariants is
grounding them in the laws of physics, as discussed in Chapter
8. F=ma. To create acceleration, we must effectively transfer
force. Light travels in straight lines. In the absence of a
reflecting surface, we can’t see something if our eyes are
pointing in a completely different direction. The human arm
only breaks in certain ways, etc., and so on! There is no
physical law that I know of that says I can’t pass the ball to
my teammate when I am in the air.
3)But what about the potential for injury?
Isn’t it possible that a movement solution may be functionally
adaptive (i.e. allow for a high degree of performance success)
but is likely to cause an injury in the future? Shouldn’t we try
to use the CLA to move an athlete away from that solution?
Yes, indeed. Aspects of a movement solution that are protective
against injury (e.g. distributing force when you land back on
the ground) should be classified as invariants. But, in my
experience, this is not as big of a problem as most would think
because many of the movement solutions that have the
potential to cause injury also hurt performance in some way.
They often represent local minima in the solution space. That is,
the performer is successful under this specific set of constraints
but there is another solution waiting out there to be found that
would be effective under an even broader range of constraints.
Let’s look at a couple of examples.
As I discussed in Chapter 7 of my first book, one of the
causes of elbow injuries in baseball pitchers is forearm flyout.
This occurs when they break the kinetic chain during the
delivery and put all the force in their poor elbow ligament. But
this would likely also come at the performance cost of slower
pitch speed. Similarly, if a basketball player lands on one foot
after attempting to block a shot, this increases the potential for
knee or ankle injuries. But it would also make them slower to
go for a rebound as compared to landing with a good,
well-distributed base of force on two feet. But, either way, if as
a coach we identify a movement solution that we feel could
lead to injury we should be using the CLA to encourage the
performer to find another one, even if it does lead to initial
performance success.
4)But aren’t you saying that one of my athletes should be able
to keep doing a particular movement solution while I try to
guide another athlete away from it?
Yes! We are back to the inconvenient truth of having to
individualize our coaching. What might be a functionally adaptive
movement solution for one player (that we want to keep
allowing them to do but challenging them with constraints) is
likely to NOT be for other players (we want to encourage them
to use other solutions using the CLA). It all depends on
whether they are successful using it and whether that success
level is maintained when we change the constraints. This is, of
course, something kids do on the playground all the time:
“Stop shooting 3-pointers Rob, you haven’t made one all
game!”
Q3: How should ecological designers design in variability using
constraints?
The final question asked in this paper concerns the amount of
variability we add to practice. The authors express their
frustration as follows:
We suggest that instead of ecological designers having to make
their “best guess” as to how much variability to design into the
performance environments, more guidance be given about how
to correctly identify the critical information to implement the
constraints-led approach, particularly for those less familiar with
the underlying theoretical aspects of the approach.
For example, if you are a parent coaching your kid’s basketball
team for the first time and want to set a practice activity for
learning jump shots how do you know: how many different
locations on the court should I start with, how often should I
switch between these locations, when should you add new
constraints (e.g. a defender in their face), how should combine
shooting drills with other skills like passing? What should I do
if some kids are great shooters and others can barely get the
ball to the net? Etc.
Before attempting to answer these questions, I want to address
an issue that really bugs me! In many of the critiques like this,
the authors attempt to hold the ecological approach to an
unfair standard that they don’t hold their own traditional
methods to! This year, I have been faced with some of the
challenges just described: I was coaching my stepson’s flag
football team for the first time. In the app that shows our
game schedule (called Mojo if you want to look for yourself)
there is, as Kennedy and O’Brien suggest, “a detailed list of
specific skills and drills to help guide the process of designing
and delivering training environments that support
age-appropriate skill development” For example, there are
categories for throwing, taking handoffs, and flag pulling with
videos of drills you can use for each. But nowhere in this app
is there any information that answers the questions raised at
the start of this section. How many reps? When should I move
to a higher-level drill? How often should I switch? Unless I am
missing something, the precise answers to these questions the
authors are asking the ecological approach to provide are also
not addressed in the traditional approach!
Focusing just on the question of how much variability in
constraints to design into practice, this issue was discussed in
Chapter 6 (e.g. see Figure 6.7) in the section looking at scaling
variability appropriately. Again, the 70% rule is your friend here.
A novice that brings a lot of their own inherent variability will
most likely not achieve a high success rate if we, for example,
change the location on the floor they shoot from often.
Variability is a source of task difficulty. So, we need to dial it
back until they are performing at the optimal challenge level.
Then we can dial it back up as they become more skillful.
So, the answer to this question is: we should always be
designing some variability into practice (“lowest hanging fruit”
remember) but at a level that results in the optimal level of
challenge.
Pulling this all together, let’s “add some texture” to the CLA
and summarize the different guidelines discussed in this chapter.

Some (More) Basic Guidelines on Using the CLA

Designing the Initial Constraint
1)Be as Specific as Possible in Defining the Problem
Let’s return to my failed “you have to pass 3 times before you
shoot” constraint. The issue here was that my definition of the
problematic movement solution that I wanted to try and move
the team away from was not specific enough. I wanted them
to stop holding the ball and instead pass it around to their
teammates. But not just any pass! More specifically, my goal
was to get them to pass the ball so that the defensive players
would have to move around, thus, opening lanes for us to
drive to the basket and score. So, I was looking for passes
that moved the ball to the other side of the court, changed the
direction of flow, occurred once we were on the offensive end,
etc.
To achieve this, I added the constraint illustrated for the
small-sided soccer game in Figure 5.6. I drew out vertical lanes
on the floor and constrained certain players to move within
them. This combined with the “3 pass” rule served to get the
type of ball movement I was looking for!

2)Wherever Possible use a Physical Constraint Over a

Rule-Based One.
I have found that, in general, physical constraints (e.g. barriers,
different size balls, changing boundaries, more players) are
more effective than rule-based ones. For example, imagine I
made a rule that said “For a basket to count you have to
shoot it with a high arc on it”. When players went out on the
court, what motivation would they have to use a movement
solution to satisfy this constraint? To do what the coach told
them! Just like my “pass 3 times before shoot” rule, there is
no inherent reason to follow it.
If instead, I had given the defenders long sticks they could use
to block shots, the players would be motivated to use the high
arc solution because that would be an effective way to get the
ball to the basket. While rule-based constraints can be effective
and sometimes are the only way to de-stabilize a particular
movement solution, I think physical constraints let new
movement solutions emerge more naturally, in a way that gives
players more autonomy instead of being told what they can
and cannot do.
3) Use Constraints that are Relevant to your Athlete’s Current
Performance
As will discussed in more detail in Chapter 13, one thing that
can make a practice activity more motivating for an athlete is
to make it timely. That is to make it pertain, to a performance
issue they are currently experiencing. To achieve this end, I like
to use an approach I call “Play Afford Play”. Here is an
example I used when running a kids’ pickleball practice one
day.
I come to practice and the first thing I say is: we're going to
start today by playing some games right away. No drills. Right
to playing. Instead of waiting until the end of practice to
reward kids with playing games – let’s do it right away.
Immediately this puts the kids in a good mood! While these
games are going on, I walk around, and look for things players
are struggling with. So, for example, the first time I did this I
noticed that a lot of the kids were missing serves. Specifically,
they were losing points because they were hitting the serves
too long.
When this initial round of games ended, instead of following
some pre-determined plan, we worked on serves. In other
words, I adapted practice to what was happening in front of
me. Specifically, I used a CLA to guide players to a movement
solution that keeps the ball in. For example, I might add an
instructional constraint in the form of the analogy: “swing your
racquet like it’s following a rainbow” (to add topspin to the
ball). Or I might use the constraint of a piece of rope across
the top of the net that the players must hit the ball under (to
give them augmented information). After doing these practice
activities, we immediately went to playing games so the kids
could put these new movement solutions into action.
What I am trying to achieve here is to motivate
self-organization. Give the athlete a reason for wanting to
change, find a new movement solution, and adapt. Instead of
just showing up at practice and saying we need to work on
forehands or backhands, I let the design of practice emerge
from what they are struggling with. In Craig Morris’ words I
was attempting to: “skillfully stretch toward things that are
there..”
4)
 Think in terms of designing constraints that encourage
interactions between the athlete and their environment not just
techniques.
The other thing I try to do is, when I'm thinking about
planning practices I ask myself: can I describe what we're
going to work on today in terms of an interaction or a
relationship (between the athlete and their environment) rather
than just the skill or action itself? So instead of saying, we're
going to work on backhands, we're going to work on serves, I
say we're going to work on keeping your opponent deep in
the court, we're going to work on moving your opponent
around, we're going to work on preventing your opponent from
smashing the ball at you. So, interactions, not just techniques.
As a coach, this gets you thinking about designing practice in
terms of affordances and how this can be amplified using
constraints. I think this is particularly useful for a new coach
when they are faced with the challenge expressed by Kennedy
and O’Brien:
To figure out what affordances are available to their athletes in
the performance environment and how to dynamically
manipulate the environment, task, and personal constraints in
order to shape skill development.
When and How Should I Change a Constraint?
1)Maintaining an optimal level of challenge
With regards to the question of “when”, as we have discussed,
the main tool at your disposal is the 70% rule. When we
deviate from this ideal level of challenge in either direction we
should consider changing a constraint. But, there is also more
to it than that. As a coach, we need to observe our athletes to
see whether or not they are exploring the movement space
and whether or not they are taking the affordances we want
them to. Are your athletes trying something new? Are they
trying to make longer passes? Pass the ball through that
opening? Swing hard? Etc. If this is not happening, we may
want to consider..
2)
When should I completely abandon a constraint?
Let’s go back to my example with the soccer team. As a
coach, you create a practice activity with a smaller playing area
and add a rule that the team must make at least 3 passes
before taking a shot on goal. After several minutes you are not
seeing any changes in passing behavior. Your activity is not
working. What should you do?
The first thing I think needs to be done is a careful
examination of why you think the activity is not working. What
specifically are your athletes doing or not doing that is making
you feel this way? Do you know specifically what you are
looking for as a coach? In my soccer example, what does it
mean to improve at passing? What is involved in being a good
passer in soccer? Well, here are a few of the things it involves
if we only consider the passer: the ability to effectively scan the
environment while handling the ball and receiving a pass
(instead of looking at your feet), attunement to the action
capabilities of your teammates (if I make a quick through ball
with player A be able to get to it?), sensitivity to the perceptual
information needed to making passing decisions and guide the
act of passing (i.e. direction and speed of the pass) –
specifically the size of gaps and rate of change of gap size and
interpersonal distance, moving the ball quickly etc. and so on.
Again, one of the main ways that I think the problem of
ineffective practice activities can be avoided is by going through
this kind of thought process before you implement a constraint.
Next, we need to address the second problem I see with a lot
of practice activities: clearly defining and communicating what it
means to be successful in the activity to the athletes involved.
When you introduce an activity, it can sometimes be easy to
create a conflict with the athlete’s natural motivations for playing
the game. For example, in soccer, an attacking player's
motivation is to score. If you introduce a constraint that they
have to make at least three passes before shooting without
clearly explaining why, what is their motivation for doing this?
All it is doing is artificially delaying what they really want to do.
This can lead to a very common human behavior: satisficing.
Satisficing occurs when a performer chooses the first option,
they find that is just good enough (in other words, it is
satisfactory) instead of continuing to search for an optimal
solution. If there is no real motivation to follow the instructions
you give, people are extremely good at finding easy loopholes
that will be just good enough.
In the lab, I used to see this all the time in my perception
experiments. For example, if I am doing a time-to-contact
experiment and ask a participant which of two objects would
hit you sooner – what is their motivation? – in most cases, it
is just to be finished with study so they can leave and not
have to repeat trials. So, if I leave them the option of doing
something else that’s easier (for example, basing their judgment
on distance) they will. Obviously, this type of satisficing behavior
can greatly compromise the effectiveness of a practice activity.
In my example, the goal of adding the 3 -pass constraint is to
give players more experience with the perceptual information
and task dynamics involved. If they are satisfying this constraint
by making a bunch of quick back-and-forth passes this will not
be achieved.
As a coach, there are two ways you can avoid this type of
dysfunctional satisficing. The first is to clearly explain the “why”
of each activity to the players involved. Or better yet, design
the constraints so they experience it themselves as discussed in
Chapter 5! This will hopefully change their motivation and lead
to them judging the simple shortcuts to not be satisfactory
solutions.
So as a coach, we have designed what we think is an effective
practice activity (both in terms of knowing what we are looking
for and getting the players on board), what happens when the
players still don’t seem to be making the adaptations and
coming up with the movement solutions we are hoping for?
Well, first let’s talk about the reasons why this poses a
challenge. To start with we have the infamous
performance-learning paradox. That is, players looking like they
are struggling and not “getting it” might actually be a better
indicator your activity is leading to improvement than if
everything was running smoothly – learning is messy,
remember. Then we have, of course, nonlinearities and plateaus
in learning such that it may look like there is no progress at
all and then suddenly you see an improvement. This can be
driven by a bifurcation in dynamical systems terminology where
a performer becomes attuned to a higher-order perceptual
information source.
Several types of improvement or learning can occur without
any observable change in performance. A good example of this
is the phenomenon of overlearning where performance hits a
plateau, but the efficiency of performance is still improving. That
is, they can produce the same level of performance with a
reduced amount of effort. For example, even though
performance outcomes are staying the same, it is possible that
the performer could be developing more efficient attentional
control and gaze behavior strategies such that if you switched
to testing them in different conditions (say, a dual task or
under high pressure) you will see that they are making gains
by being able to handle such conditions. All this is to say that
effective motor learning is not as easy to spot as most people
think! Give it time!
Next, a coach should pay attention to the stability of behavior
during practice activities. Think of stability as a continuum. At
one end we have completely stable performance where there is
not much variability and athletes are executing the same actions
over and over. This situation is often what inspires a coach to
come up with a new practice activity in the first place – that
is, athletes are stuck in a certain pattern of execution and not
continuing to improve or adapt. This can also happen when a
new practice activity does not motivate an athlete to change
their existing perceptual-motor solution. Another and very
different reason why this pattern of overly stable behavior can
occur with a new activity is if you have not established a
culture of psychological safety in your practices (discussed more
in Chapter 12). In other words, you have not conveyed to
athletes that it’s Ok to look bad and make mistakes.
At the other end of the continuum, we have complete instability
where behavior is highly variable and almost random. This can
occur when the demands of the new practice activity are too
high for a performer’s ability – in other words, you are
pushing them too much. We need to dial back difficulty (and
variability) using the methods described in Chapter 6.
The final thing I think is useful to consider to help determine
whether or not a new practice activity is working or on not is
to introduce perturbations. That is a new constraint that
suddenly and unexpectedly alters the state of the
athlete-task-environment. For example, for one of the reps add
a time constraint or add a dual task that requires players to
shout something out, or add an extra player to one of the
sides. A sudden alteration to a system like this can accomplish
two things. First, as I mentioned, perturbations like these can
reveal learning that has occurred that was not visible in
unperturbed conditions. Second, it can also help to sometimes
nudge a performer out into a different area of solution space.
What if we have done all of these things – carefully examined
the process, looked at the stability of performance, added
perturbations, etc. – and we still do not see the type of
adaptations in behavior the practice activity was designed to
produce? Should we stick with the activity and just hope
self-organization occurs? For how long? Should we just
abandon it and move on to something else?
Well, I would make a couple of points here. First, we know
from motor learning research, that if there is sufficient feedback
and information for learning available to a performer there
should be some evidence of adaptation (or at least an increase
in the variability of behavior) very quickly. For example, in
experiments where I have asked batters to adjust to a new bat
weight6
, I see changes within a couple of swings – as long as
there is clear feedback. When I take feedback about the swing
away you either see no adjustments or high variability – so
missing the sweet spot of stability. Another example, I will give
is from my study looking at using a CLA in batting7
. Even
though I saw no evidence of performance outcome changes,
there were significant differences in movement variability by the
second block of training. So, I think it is reasonable to expect
to see something fairly quickly!
But if you are not, I would recommend doing one last thing
before you give up: change the initial conditions of the activity.
Changing where a performer starts in the perceptual-motor
landscape can have a large effect on the solution they come up
with. So, if your activity is not working, try moving it to a
different part of the practice session, try changing up who
starts with the ball, give different body starting postures, etc.
Doing this changes both the task constraints, and most likely
the individual constraints of the performer (for example, how
fatigued they are, their level of motivation), etc. The key point
is that when we start with a different set of initial conditions it
encourages a performer to take a different route through
perceptual-motor space to find the appropriate movement
solution.
But if none of this works, don’t be afraid to abandon a
practice activity. Good coaching involves experimentation. Even
failed experiments can be valuable if you are willing to learn
from them.
Before I end this chapter, there are a couple more points I
would like to make.
On the Strictness of the 70% Rule
Throughout this chapter, I emphasized the importance of aiming
for a success rate (the optimal level of challenge) and called it
a “rule”. Must it be that precise? Do we need to give an
athlete 10, 50, or even 100 reps so that we can accurately
calculate their success rate? No, think of it as an optimal range
rather than a single point. 70 plus or minus 5 or so. If you
are running a practice, and your athletes are successful on 2
out of 3 attempts (i.e., 66%) or three out of four attempts (i.e.,
75%) you are right where you want to be!
Relatedly, I often get asked about how this optimal level of
challenge applies to sports with a much lower success rate. For
example, hitting a baseball ball or making a 3-pointer in
basketball (where 30-40% is considered elite). Remember we
are talking about practice here, not the game. Even though the
game might involve a much lower success rate, I think the
70% rule is something we still want to aim for early in practice.
Once we reach the optimization stage of coaching an athlete
(and the constraints become much closer to competition-like)
then we can again “break the rules” and deviate from this ideal
level of challenge.
Stop Looking for the CLA Recipe
As I have tried to emphasize several times in this chapter,
there will never be a prescriptive, do this first and then that,
manual for coaching the CLA. Or at least, there will never be
one that doesn’t go against the theory on which the CLA is
based. Be a master chef, adapt and iterate, and don’t just
follow recipes! Thus, I will end this chapter the same way
Stone et al.
8 end their paper on advice for applying an
ecological approach in coaching (discussed in detail in Chapter
14):
We realise that many of you might also find yourself
disappointed if you expected a detailed recipe of how CLA or
NLP can be enacted from this paper (chapter). However, as
we have explained, such pre-determined and mechanistic
solutions are counter to the theoretical assumptions that
underpin the ED (ecological dynamics) approach

12
CREATING A CULTURE AND A
FORM OF LIFE

A re we going to be a “run and shoot” offensive team? Or

one that wins with defense? Are we going to emphasize hitting
a lot of home runs or focus on “getting them on, getting them
over and getting them in”. In our gym, are we going to focus
on being disciplined, respectful and doing what the instructor
says OR are we going emphasize autonomy, playing and having
fun? These are all examples of the “culture” that gets formed
around a team or group of athletes that train together. As a
coach, how can we best build a culture that meets with our
goals and values? How can we best understand this idea and
where does it fit in with our constraints-based model of
coaching discussed in Chapter 3?
In ecological dynamics terms, what we are talking about here is
a “form of life”. This concept was first introduced by the
philosopher Wittgenstein1
. A form of life consists of the
behaviors, attitudes, values, beliefs, practices, and customs that
shape the communities we live in. These sociocultural practices
serve to constrain the emergence of specific behavioral patterns.
So, one way we can think of them is as a type of
environmental constraint in our coaching model.
Wittgenstein first discussed the concept of “form of life” to
explain the use of language. He emphasized that we can’t
understand the meaning of words separately from the context
in which they are used. Thus, it is an idea, that fits perfectly
with the ecological approach. It also emphasizes the importance
of not just what we say in creating a culture but in what we
do!
How does this apply to sports? Well critically, just like language,
every practice activity or drill we do is inseparable from the
culture and customs in which they are embedded. A detailed
example of this can be seen in Brazilian soccer as examined in
a study by Uehara et al
2
. Brazilian players are known for
playing Joga Bonito (which translates to “the beautiful game”).
A style of play that emphasizes flair, deception, creativity, and
flaunting rules (e.g. diving to draw a yellow card, and scoring
goals in unique ways). How might this style of play have arisen
from their “form of life”?
A revered figure in Brazilian culture is the Malandro (or
Malandra for a female) which translates to “trickster”. These
anti-heroes, celebrated in music and movies, used their
malandragem (or “cunning”) to fight government corruption and
oppression. They went around the law by often being deceptive
and manipulative to get things done. The patterns of movement
associated with the Malandra are called Ginga, which translates
to “body sway”. These are atypical, fluid body movements used
to deceive an opponent. So, you can see clear parallels between
the traits revered in the Brazilian culture and the play on the
field. Other examples of this types of socio-cultural influence in
sports can be seen in baseball in Asia, and rugby in New
Zealand, for example.
The influence of this “form of line” on the emergence of skillful
behavior can be best understood in terms of the concept of a
field of affordances, introduced by van Dijk and Reitveld3.
Recall, this is the idea that at any given instant there are
multiple invitations for actions calling out to us. Critically, for
what we looking at in this chapter, these affordances are
materially entangled. That is, they are embedded in a
socio-cultural context which constrains the likelihood that the
performer will or will not accept the invitation and realize the
affordance. Let’s dig into this idea of how the strength of the
invitation an affordance can change in a bit more detail.

Why are Some Affordances More Inviting than Others?

This discussion of “form of life” raises some key points for
coaches to understand within the ecological approach: why are
some affordances more inviting to an athlete than others? Why
do our athletes choose to accept the invitation to realize some
affordances and ignore others? And how might we influence all
of this as a coach?
As discussed in an article by Rob Withagen and colleagues4,
affordances invite us to act but we also clearly have agency
over whether we choose to do so. Throughout the history of
ecological psychology, many have equated the concepts of
affordances and perception-action coupling as suggestions that
those of us who follow an ecological approach are mechanist
behaviorists, in the Skinnerian sense. That is, just as Skinners’
dogs initiated an automatic response of salivating to the
stimulus of the bell, we are arguing that the behavior of
athletes (their movements) is automatically driven by information
from the environment. That is information about affordance
causes the behavior to realize the affordance without the
human being having much say or choice in the matter!
Stimulus (information) causes response (movement pattern).
Gibson himself was, of course, very opposed to this view. That
is why we always refer to affordances as opportunities or
possibilities for action and not causes of them. In Gibson’s
framework, the environment is conceived not as a collection of
causes, but as a manifold of action possibilities.
There is an important distinction between this concept of
agency and the idea of embodied perception (discussed in
Chapter 5 of my first book), as it relates to affordances. When
there is a mismatch between information and action capabilities
– for example, I look at set of stairs with risers that are much
higher than my leg length – I don’t perceive the affordance of
climbability, at all. It’s not that I am not accepting the
affordance or that it is less inviting – it does not exist for me.
It is important to make another, related point here. In Gibson’s
view, affordances are permanent – as long as they are within
our action boundaries they will also be there. For example, a
chair of a certain height always affords sitting to you. What
changes is how these affordances align with your intentions or
goals which are temporary and ever-changing. As human
beings, we have some agency over which affordances we
choose to accept and try to realize. For example, standing on a
chair to reach something off a high shelf is an affordance (an
opportunity to reach higher) that we might accept while cooking
a meal at home but that we will likely not accept while viewing
a priceless Louis 14th chair at a museum. Even though the
affordance is present in both cases.
As Withagen and colleagues put it:
The environment is not a neutral manifold of action possibilities
the agent simply chooses from; rather, the environment can
invite a certain action or even urge a person to do something.
That is certain affordances are more inviting than others to us.

There are several factors that can influence the degree to

which certain affordances invite behavior. The first is the
amount of effort required to achieve the affordance. This was
nicely demonstrated in a study I highlighted in my first book
when looking at the embodied perception research done by
Frank Eves5
. In one of his early studies, Frank and his
colleagues investigated what factors influence people’s choice to
take the stairs or the escalator in a shopping mall, when both
options are available. After they made the choice, he asked
them to judge the steepness of each and found that people
who chose the escalator perceived the stairs to be steeper than
those who chose the stairs. In follow-up studies, he also found
that people who chose the escalator tended to be heavier,
older, carrying heavier objects like a backpack, or were
energy-depleted.
In looking at these results I don’t think we would want to
claim that the people who choose to use the escalator did not
perceive the affordance of “climbability” of the stairs.
Presumably, in most cases, if there were no escalators they
would have climbed them and not just turned around and
went home. Both the escalator and the stairs offered
opportunities for action – the affordance of getting to the next
level of the mall. It was just the case that the affordance from
the escalator was more inviting to them because presumably it
was perceived that it would require significantly less effort to
realize. Objects like stairs that are too near our action
boundary (of climbing vs not being able to climb) do not invite
as much as ones that are well within our action boundary.
Another important point here – affordance can only invite
behavior if they are perceived in the first place. If, for example,
a particular affordance is outside our action boundary due to
our action capacity it will not be perceived and thus cannot be
invited. As a sporting example, this is essentially what we are
trying to achieve when we scale equipment for young athletes.
When we use a lower hoop in basketball, we make the
affordance of making a one-handed layup perceivable and thus
inviting (as opposed to shooting with two hands and putting
their whole body into it) by pushing it within an action
boundary when it was not before.
The second factor that influences how inviting an affordance is
that some affordances are more important for our survival
from an evolutionary perspective. These include affordances for
nutrition, shelter, and avoiding danger. In the context of sports,
this can create a difficult challenge in injury recovery: the
affordances of acting to avoid reinjury or pain can be much
more inviting than the affordances a player needs to realize
(e.g., accelerating through a gap in defenders) if they want to
return to their pre-injury level of performance.
The third factor is where “form of life” or culture comes in. As
a culture, we have agreed on the appropriateness of different
affordances. For example, the standing on a chair example I
discussed earlier is not an acceptable action to execute in a
museum. This again is something that a coach can influence
through the culture or form of life they create around a team.
For example, is it acceptable for an offensive player to join an
offensive breakout? Is it acceptable for an offensive player to
walk back on defense? Even though these affordances might
not be explicitly encouraged or discouraged by a coach, their
degree of “invitingness” is shaped by things like how they are
praised/criticized, how often they are practiced, or whether they
are a focus of a constraints drill designed to influence them.
The fourth factor is personal history. Just in the way our
information-movement control laws change with experience so
does how inviting a particular affordance is to us. For example,
if we accept the affordance to make a cross-court pass and it
gets intercepted, that affordance will be less inviting to us next
time.
How Can a Coach Create a “Form of Life” Around Their
Team
Through our behaviors (both what we do and what we say)
on the practice field, we as coaches can change how inviting
particular affordances are or are not. Before getting to some
specific ideas for achieving this end, let’s look at what not to
do first!
You Can’t Copy & Paste a “Form of Life”
As discussed in detail in the article by Rothwell and colleagues
6
, a common mistake that we make is to try and copy and
paste the culture from a team in another country (or region)
onto our system. This is something that seems to happen after
every Olympics or international competition like the World Cup.
An example of this, described in detail by O’Sullivan et al7.,
occurred when after qualifying for the World Cup in 2006,
Football Federation Australia decided to import a “Dutch
model”. Australian soccer chief Steve O’Connor’s tells you
everything you need to know about how successful this was:
I don’t think you can import systems wholesale from one
country to another and expect it to work. There are the ways
the Dutch do things, the Germans, the French and they have
been very successful. But there are all different reasons for
that.
The problem here is one that I have talked about throughout
this book: linear thinking. When we try and copy and paste
another country’s style like this, we are again ignoring the
interactions that occur within a complex system. Specifically, we
are ignoring the principle of socio-material entanglement. A
country is not just playing a particular style. They are playing a
style that is embedded within and influenced by the
socio-cultural constraints of their society. As Rothwell and
colleagues state:
Simply imitating the traditional practices of another nation may
present performance challenges without first exploring,
understanding, and embracing the form of life that influences
the factors that lead to another nation's success in competitive
sport.
For example, imagine trying to just copy and paste the
Brazilian Joga Bonito onto a more conservative country like
Iran. Most people would agree that this would not work! But I
think they would make that conclusion for the wrong reasons.
They would assume that the Iranian players are not as skillful
enough e.g. they have not spent as many hours on the ball by
themselves working on their dribbling skills. As you might
guess, I would propose a different explanation: this would not
be a failure because of the lack of decomposed ballhandling
practice, but rather because the “form of life”, and thus which
affordances are most strongly invited, are completely different
between the two countries.
So, the first thing we need to do when thinking about the
culture we want to create on our team or at our gym is to
consider how the form of life we want to create fits within the
greater socio-cultural constraints of the society we live in. I
think we also need to put a bit more thought into what exactly
we want to emulate from another country. Rather than just
trying to copy and paste the superficial elements from another
system I think it is much better to focus on the key principles
of play.
Principles of Play
Principles of play can be thought of as the desired qualities that
a coach has for the movement solutions their team uses.
Examples of attacking and defensive principles of play in soccer
are shown in Table 12.1.
Note, similar to the concepts of invariants, principles of play do
not prescribe a particular technique that has to be used (e.g.,
always dribbling on the outside of the foot, one player moving
here, another moving there), but instead the key properties of
the movement outcome the coach is looking for. It is important
to note, however, that principles of play are not the same as
invariants – rather than being things that must be there
(grounded in the laws of physics) they are more desired
properties based on the coach’s philosophy. But I think they
can be used in the same way in coaching.
Attacking Principles
Defending Principles
Width
Stretch the opponents laterally across the field.
Delay
Slow down the attack to let the defensive reorganize when
outnumbered
Mobility
Interchange positions and move to create space
Cover
To provide close support for the first pressuring defender
Penetration
Get inside and behind the defense’s shape
Balance
Covering the space away from the ball to limit attacking
options
Support
To keep the position with support in front of, on the side,
and behind the ball
Compactness
To assemble as quickly as possible to protect areas of
vulnerability to attack
Table 12.1: Example Principles of Play in Soccer
Principles of play are often conceptualized and taught from an
information-processing perspective. For example, it is assumed
that they require the player to develop a mental model and
conceptual understanding of each of the principles. But there is
an alternative way to develop and use them. As Marcus
DiBernando explains in his great blog post8 on the topic:
The principles of play should be very simple ideas that the
coach can guide the player’s attention naturally to in a training
session. The principles can also be carried out using constraints
in training sessions.

Promoting Principles of Play Through the CLA

For example, let’s look at using the CLA to guide athletes to a
movement solution that is consistent with the principle of
“width”. This can be supported using the instructional and
questioning task constraints. For example, the coach can stop
the game for a moment, asking where is it less crowded on
the field, where are there more advantages for us, and where
might there be a better chance to score. By questioning the
players, the coach can educate the attention of the players to
another area of the field, making them more attuned to
switching the point of attack. Questioning is something that I
haven’t discussed much in this book. I think it can be a very
effective tool if it has one key feature: the player(s) answer by
doing, not just by saying. Remember we are trying to promote
knowledge of not knowledge about.
The coach could also use other task constraints. For example,
creating two wide channels on the field that the outside backs
must occupy in possession (see Figure 5.6), guarantees the
attacking team’s shape has width. Using the SSCG shown in
Figure 5.6 guides a team to solutions consistent with the
principles of “penetration” and “support”.
Breaking the Rules
An important part of the form of life we create around or
team or group of athletes, along with the key principles of play
or invariants we want them to follow, are acceptable ways they
are allowed to break these rules. From Vaughn et al9
.
Therefore, the development of expertise requires that a
participant is not only well aware of underpinning sociocultural
constraints or functionality of practice (i.e. usefulness,
effectiveness, appropriateness or adequacy) but also knows how
to diverge from them with novelty (i.e. originality).
Building a Culture of Failure, Er Psychological Safety
Across all three of my books, I have tried to emphasize that
failure is an essential part of learning! If your athletes are
doing everything perfectly in practice, achieving their
performance goals on every execution, they are performing
NOT learning. There is a place and time for this type of
practice (discussed more below) but if we want to focus on
skill acquisition (i.e., we want our athletes to explore new
movement solutions or find a new scaling of an existing
solutions) then we need to have some failure. NOT achieving
our performance goal (and the associated information for
learning that comes with it) on some attempts is the essential
motivation that we need to drive us to explore and change!
To support this, a coach must create a culture around their
athletes in which it is OK to fail. The technical term for this
psychological safety, which can be defined as a shared belief
held by members of a team that it’s OK to take risks, to
express their ideas and concerns, to speak up with questions,
and to make mistakes — all without fear of negative
consequences. Notice within this definition we are also creating
an environment that encourages our athletes to ask “why”. This
again fits perfectly with the model of coaching discussed in
Chapter 3. We want our athletes to be active, self-regulating
participants in their skill development NOT passive receivers of
information who just do things because the coach tells them to!
There are a few different ways a coach can promote
psychological safety:
1)Reduce fear of failure in practice. Performers typically play it
safe when they fear negative reactions and avoid risk-taking
that could lead to making mistakes. This does not mean there
should be no consequences for actions in practice. As I
mentioned, in my second book having consequences is
important for doing pressure training. But if possible make it
more of a fun game than something your athletes dread (e.g. if
you are last you have to make a funny speech in front of the
team instead of running laps).
2)
Reduce the practical consequences of failing. One of the
things I have noticed that can limit exploration is athletes’
worry that they will slow practice down when they make a
mistake (e.g. while someone runs to pick up the ball after an
errant pass). To remove this, I like to have balls ready to be
immediately put back into play.
3)
Encourage athletes to play an active role through questioning.
For example, ask them their opinion about how a particular
practice activity went and if there are ways, they think it could
be improved.
Blending Skill Development & Game Preparation
While we are on the topic of failure, one of the real challenges
I have found in working with high-level athletes is how to best
support the often-competing goals of skill development and
preparing to perform. As we have been discussing, skill
development requires a significant amount of failure which can
conflict with a player’s need to feel confident about their ability
to perform when they go into a game or competition. To
address this issue, I have tried to create a culture around the
teams I work with in which we have three distinct goals for
practice. These goals, which are adapted from Lohse and
Hodges's Extended Challenge Point Hypothesis10, are
illustrated in Figure 12.1 and can be described as follows:
A diagram of a performance analysis Description automatically
generated with medium confidence
Figure 12.1 – The three different goals of practice. Adapted and
modified from Lohse and Hodges10

1)Practice to Learn (L). This is the main goal we have

discussed throughout the book. This is a practice designed to
improve an athlete's performance by encouraging them to
explore other movement solutions or working to more effectively
use the one they already have (i.e., scaling or optimization).
Here we want to follow the basic principles of representative
design, the CLA, and the Challenge Point Hypothesis. That is,
we want to add constraints to create representative problems
for our athletes to solve and have them operating at around a
70% success rate. So we have a high level of task difficulty
and a relatively high failure rate. As discussed in Chapter 4, in
many cases we are also likely to have a lower level of
specificity. For example, we might want to use bats, balls, clubs,
field sizes, numbers of players, and other constraints that are
totally unlike what the athlete will face in a game or
competition. We are taking a slice of the game to work on it
and improve it.
2)
Practice to Perform(P). The goal of this type of practice is to
prepare an athlete to achieve the best possible performance
they can in the upcoming game or match, against a specific
opponent. Here we do want a higher level of specificity. For
example, we might have a baseball batter hit against a pitching
machine set to throw the types of pitches the starting pitcher
in that night’s game throws. Or, we might have a soccer or
football team practice defending against the specific type of
offense our upcoming opponent plays. The level of task
difficulty is still relatively high because we want the main task
constraints (e.g. speeds, time available, space, etc.) to be
competition-like. But the difficulty won’t typically be as high as
when we practice to learn, because we are not using principles
like overload.
3)
Practice to Build Confidence (C).
Here the goal of practice is
to make the athlete feel good! We want them performing at a
very high level, with a relatively low failure rate so that they
can feel confident for the upcoming match. The level of
specificity will be somewhat moderate – we want the basic
conditions to be similar to the game but the speeds, amount of
variability, etc. to be at lower levels to allow for more success.
In the environment I try to create, we follow two basic
principles: (i) clear intent and (ii) progression/blending of goals.
First off, we need to recognize which one of these goals we
are trying to achieve. I have no problem at all with practice
activities designed to build confidence but if that is all we do,
then that is a problem! Second, as we approach the game or
competition (for example, on game day itself) I try to
encourage a progression and blending of these goals. This
concept is illustrated in Figure 12.2.
A diagram of a performance Description automatically generated
Figure 10.2 – Blending The Goals of Practice In Game Day
When a player first shows up on the game day (or the day
before if you compete in the morning) I want them to focus
on Training to Learn. My sales pitch I use to sell the player is
that we are going to "Break you down, then build you back
up”. When we have a some time before the game or
competition, that is the time we are going to challenge you and
you are going to have some failure, because there will be time
for you to have a lot of success and build your confidence
back up later. Here we are going to use individualized CLA
activities focused on improving your weaknesses. We are going
to let the athlete have some say in how they practice (i.e.,
constrained choice of which constraint they do first). In sports
where we play a lot, like baseball or basketball, if we don’t
make time for this on game days, when are we going to do
it?
After completing the Train to Learn phase of the day, the next
thing I like to do is hold Advance meetings where we give
players the scouting information we have about our opponent.
We watch videos and go over some tendencies. In my
experience, these meetings are typically held much closer to
game time but I like to change that. The purpose of advance
meetings should be to give players specific intentions (i.e., make
certain affordances more inviting) based on the opponent.
Examples include: “close out on their 3-point shooters’, “make
the pitcher get the ball down in the strike zone”, “make that
player use his left hand”, etc. Then, critically, we want to allow
our athletes to practice implementing these intentions.
This is where the blending into the Train to Perform phase
comes in. We want to create practice activities that include
some of the specific characteristics of our opponent. Initially, we
can have the athlete work on the same things they did in the
Train to Learn phase then have them practice the specific goals
we laid out for them in the Advance Meeting.
Finally, as we get closer to the game and there are fans at the
stadium watching what we do, we can shift into the Train to
Build Confidence phase. Go ahead and hit those 70 mph
pitches right down the middle of the plate for home runs or
make ten -3-pointers in a row shooting from the same spot
with no defender. Our goal here is to build your confidence.
For this to be effective we need to change an athlete’s view of
what a good pre-game routine involves from “I repeat these
exact same things every time” to a “break you down, build you
back up”, repetition without repetition perspective. We need to
create a culture of challenge, or “practicing hard to make the
game easy” and deliberate practice.

Once More About Practice Co-Design

Throughout this book, I also keep referring to the principle of
co-design – the basic idea that the athlete and coach create
and adapt the conditions of practice together. There are so
many potential benefits to making this part of your culture In a
great article on the topic, Woods et al. (2020)11 explain:

Co-designed activities ensure that self-organization and the

search for functional movement solutions remain at the
forefront of each activity. There is no prescription of optimal or
correct solutions given to the athlete by a coach/teacher/parent.
Involving athletes in this design process has: unlocked
experiential knowledge, increased the ownership of their learning
environment, and deepened their ‘knowledge of’ the competitive
environment.

13
INSPIRING MOTIVATION,
AUTONOMY & CONFIDENCE

"
The most powerful performance-enhancing drug ever invented
is belief”
- Me

T he above quote is one that I have used in talks many times

over the years. Although I spend a lot of my time getting lost
in the weeds of visual information and movement dynamics,
there is no denying the incredible power that factors like
motivation and confidence have on performance.
One of the most incredible, yet grotesque, examples of this is
the “drowning rat experiments” conducted by Curt Richter in
the 1950’s1
. Here is what he found, in a nutshell. Richter’s
experiments investigated how long it takes a rat to drown when
you place it in a bucket of water. He compared two basic
conditions. When just placed in the water, wild rats (who are
very good swimmers) would swim around the surface for a
short time, then go to the bottom and drown within a matter
of a few minutes. Next, he tweaked the conditions slightly. He
again started by placing the rats in the water in the same way
but once they started to head toward the bottom, he pulled
them out, dried them off, and let them rest and recover for a
bit. When he put them back in the water, how long did it take
them to drown? Instead of only lasting a few minutes, these
rats would swim for days! What changed? The second set of
rats had a belief (or hope) that they would be saved. This
changed their intention, which changed their action.
I often hear people say that ecological dynamics does a great
job of explaining the perceptual control of action but is not
equipped to handle effects like this. To explain the role of
things like confidence, belief, and motivation in sports many
believe that we need to bring in other (typically indirect
perception-based, “mental”) concepts. To that, I say: nonsense!
These factors are individual constraints that shape the
emergence of coordination solutions just like the size of the ball,
the number of players on the field, and the weather do.
The ecological dynamics conceptualization of these factors is
discussed nicely by Ian Renshaw and colleagues2
. In this
paper, the focus is on intrinsic motivation – actions that are
driven by internal instead of external rewards. The authors
begin by noting how ineffective prescriptive, repetition-based
coaching is in supporting intrinsic motivation. Think about the
typical drill-based activity, like dribbling around cones, punching
air, or hitting a ball of a tee. They are boring and monotonous.
To get athletes to complete such activities, coaches must
provide several sources of external motivation. We might give
our athletes a pre-practice speech emphasizing the importance
of “sacrifice and struggle” in becoming skillful. In baseball, I
have heard coaches tell players that going through monotony
(e.g. by doing the same drill over and over) is a rite of
passage! During the activity, many coaches resort to yelling
things like: “Come on get going” or “Keep it up”. Finally, we
use the promise of something more rewarding (“After we finish
these drills, we can scrimmage for a few minutes at the end of
practice”). Such coaching methods tend to make athletes
externally motivated: they complete drills to not get yelled at or
punished (e.g., running laps), and receive praise (a “good job”
or a high five) or rewards. There is a lot of research showing
the inferiority of external motivation in comparison to intrinsic
motivation. Luckily, using an ecological approach to running
practice (i.e. nonlinear pedagogy/the CLA) addresses several of
these issues.
One of the most popular well-supported theories in sports
science is related to motivation. Specifically, Self Determination
Theory (SDT), is a meta-theory combining a few different ideas,
first put together in a package by Deci & Ryan
2 in the
mid-80’s. It proposes that our level of intrinsic motivation is
driven by three basic human needs: Autonomy, Competence,
and Relatedness. Relatedness concerns the need to feel
connected, involved, supported, and consequently experience
satisfying interpersonal relationships. Competence reflects a belief
in one’s abilities and capacity to control outcomes. Finally,
Autonomy represents the desire to express choice and not to
feel controlled or compelled to do something. When these needs
are being met, we act with intrinsic motivation. Practicing is
self-satisfying.
Renshaw and colleagues propose that there is a natural fit
between the ecological approach and SDT. Specifically, these
psychological needs emerge through a process of
self-organization in the face of constraints, just like movement
solutions do. Psychological goals need not be made explicit to
learners. Instead, they can be built into the design of practice
sessions. To quote the authors:
In learning environments, where these psychological needs are
met, adaptive self-determined motives emerge; where they are
not met individuals become frustrated, and maladaptive motives,
behaviour and movement patterns emerge. Thus, the way in
which a tennis serve is coached will affect its expression as a
movement skill along with any resulting emotions and motives.
Using an ecological approach supports the emergence of the
basic needs in several ways. Encouraging athletes to explore in
practice to find their optimal solutions to performance problems
serves to build confidence and autonomy. Autonomy is
enhanced by allowing individuals to be in control of the learning
process. Competence is developed by allowing the athlete to find
movement solutions that are better matched with their own
action capabilities and individual constraints. This is in stark
contrast to the traditional approach of telling an athlete “the
right way” to do it. Finally, relatedness is enhanced through the
interaction between the coach and athletes that arises during
the process of practice co-design.
A good example is the use of task simplification, discussed in
Chapter 6. Appropriately scaling constraints to allow the optimal
level of challenge, helps athletes develop feelings of competence
while still competing in realistic, representative environments.
Keeping an athlete connected with their practice environment
and their teammates (instead of using isolated, decomposed
drills) can help enhance feelings of relatedness.
Next consider how an ecological approach can be used to
develop relatedness within teams. Team spirit or “team
chemistry” are typically thought to be add-ons that are created
through social outings and forced team-building, activities rather
than something that can be promoted via the design of learning
tasks within practice sessions. One way to address this is by
using small-sided games. Along with amplifying information and
affordances, reducing the number of players on the field
increases the importance of the contributions of a single player.
Players also get more information about how their actions do
or do not support their teammates. This all serves to support
the development of relatedness within a team.
Supporting the development of these basic needs and intrinsic
motivation in our athletes is naturally supported by several of
the concepts we have already looked at in the book. For
example, when we take into account the individual constraints
and capacities of our athletes and incorporate individualization
into our practice design (as discussed in Chapter 7), we can
avoid undermining feelings of competence. This is something
kids do very naturally when playing backyard games with
players of different ages and sizes.
To summarize, motivation, confidence, and belief in oneself are
information-driven percepts. They emerge from the practice and
training environments we create.
Coaching Methods that Align with SDT
The Power of Choice
In Chapter 7, we saw that allowing athletes to make a choice
in how they practice (e.g., the order of the constraints used in
learning to putt in golf) resulted in improved motor learning.
Allowing an athlete to make a choice (an essential part of
co-design), helps to support the need for autonomy. Let’s look
at some more research that has addressed some specific
questions related to this topic:
1)Does the Choice Need to be Task Relevant?
When we get an athlete involved in practice design by allowing
them to make a choice about it, does it need to be something
relevant to their performance (e.g., what size bat they use?) or
can we use something that is likely to be task-irrelevant (e.g.,
the music playing in the training room). There are a few
different studies that tackled this issue:
Lashanlou et al.
4
, conducted a golf putting study to address
this question. Thirty-six participants were evenly split into two
groups that both practiced for 60 trials. The relevant-choice
group chose the color of the ball that they putt with. The
task-irrelevant group chose which of two paintings was hung on
the wall in the practice room. What was found? From pre-post
training, the task-relevant group had a significantly greater
improvement in putting performance (mean score of 3), as
compared to the task-irrelevant group (2.5).
Wulf et al
5
. examined the type of choice in learning to throw
a lasso. In Experiment 2 of the study, 42 participants were
equally split into three groups. At the start of training, all
participants were shown a 60-second video demonstration of
the skill (with no narration!). The irrelevant choice group was
able to choose the color of the mat (blue, green, or pink)
placed under the target cone they were throwing at. Note the
change in terminology here – this isn’t a typo! The choice of
color (labeled as irrelevant in Wulf et al study) was a “relevant”
choice in the golf putting study just described. More on this
shortly. The relevant choice group had the choice of whether
they viewed the 60-second video demonstration before each
block of trials in the learning phase. Importantly, each
participant in the irrelevant choice group was yoked to a
participant in the relevant group. Participants in the irrelevant
group viewed the demo videos on the same schedule as
chosen by their yoked participant. Participants in the relevant
group used the mat color chosen by the irrelevant group. A
control group went through the same training but did not
choose either the mat color or when the demo videos were
presented. What was found? The task-relevant and irrelevant
choices had similar benefits on learning. In the retention phase,
there were no significant differences in performance between
these two groups, and both performed significantly better than
the control group, which had no choice in how they practiced.
Arbinga and colleagues6 addressed the question of choice
relevance in a recent dart-throwing study. 90 novices were split
into four training groups. Prior to being assigned to a group, all
participants were shown seven darts (differing in color and
feather design) and were asked to rank them from 0 (least
liked) to 6 (most liked). The training groups were as follows:
Choice-Like (choose between their two highest-rated darts),
Choice-Dislike (choose between their two lowest-rated darts),
Assigned-Liked (assigned the dart that had the same ranking,
for them, as a yoked participant in the Choice Like group), and
Assigned-Dislike (assigned the dart that had the same ranking,
for them, as a yoked participant in the Choice Dislike group).
The training involved throwing 90 darts, split into blocks of 15,
practiced on the same day. What was found? The two choice
groups (both Like and Dislike) significantly outperformed the
two, yoked groups that did not have a choice. The two groups
that performed with their preferred/liked dart (whether they
had a choice or not) outperformed the two groups that
performed with a dart they disliked.
So, to sum up, I think we can take away some key messages
here. First, allowing athletes to make a completely irrelevant
choice (i.e., that does not pertain to any of the objects they
are going to interact with, like the paintings on the wall) does
NOT seem to create a learning benefit. Second, allowing an
athlete to choose between the objects they will interact with
(e.g. the color of darts, balls, or matts) does seem to result in
better skill learning. This is the case even when we do not
expect the choice to be relevant in terms of changing the
information or task dynamics, like occurs with other constraints
we might manipulate. For example, we do not expect the
information-movement control law or calibration to change when
we switch a dart from green to red like it would need to if we
switch the dart from light to heavy. What seems to matter
more is whether the choice matters to the individual athlete
(e.g., whether they get to choose something they like or care
about). When a person perceives a choice, they are given to
be unimportant, it is been argued that this can increase feelings
of insecurity and decrease motivation7
. Finally, these choices
seem to have an even larger effect when the athlete is tested
under high-pressure conditions8
.
2)Does this Work for Experienced/Expert Athletes?
Do the types of choices matter when we are training experts?
Does making a seemingly insignificant choice like the color of
an object have an effect on more experienced performers who
are likely to have a better understanding of their skill and why
things are being manipulated in practice?
This question was addressed in a recent study by Shooli et al9
. Fifty-six basketball players (who had at least 10 years of
playing experience and currently played in the first division of
an adult league) were split into one of two groups who
completed free throw training. The choice group was allowed to
choose between practicing with a ball of one of three different
colors (green, red, and blue) while the no-choice group trained
using a ball color yoked to one of the participants in the choice
group. The training (80 shots) used different types of
attentional instructions. What was found? Surprisingly, the choice
group significantly outperformed the outperformed the no-choice
group.
So, the bottom line is that choice does seem to matter, even if
it seems irrelevant and insignificant to you as a coach. Just
make sure it matters to your athletes!

14
EMBRACING THE CHAOS AND
ADAPTING TO THE CONSTRAINTS
OF COACHING
A s someone who just went back to youth sports coaching,
after being away from it for several years (and focusing on
coach education), I have experienced many challenges with
adopting an ecological approach firsthand. As coaches, we have
a lot of constraints on us. The need to balance short-term
success with long-term skill development. The feedback and
comments from parents, managers, etc. Individualizing practice
within groups, etc. Let’s go through some of the issues that
were identified in the great articles by Stone and colleagues1
and Chow et al.
2
.
What are the Main Struggles Coaches Face in Adopting an
Ecological Approach?
The Time Required for Effective Practice Design and Iteration
Despite being viewed as a “set and forget it” type of coaching,
using the ecological approach effectively takes quite a bit of
time. As I discussed in Chapter 11, we need to put in some
time specifically defining the goals of a CLA practice activity (i.e.,
what specific aspects of the performer’s current movement
solution do we want to destabilize by adding a constraint).
Should the constraint be a physical one, an instruction, a rule,
etc? How much variability in constraints should we start with?
Even though the end product is likely to be different from our
plan, we still need to take time to do the initial CLA planning.
This is something that I have observed a lot in working as a
coach educator within a sports organization. Understanding the
basics of ecological dynamics theory such that you can
effectively “copy and paste” a CLA activity you read about or
saw online can happen pretty quickly. But moving to the next
level where you can make new CLA activities that are
individualized for the athletes you are working with - that takes
a bit more theoretical knowledge so that you can understand
the logic of the CLA.
Similarly, once practice is over, we need to take time to
systematically reflect on how well things went, keeping track of
any adaptations we made, and incorporating them into future
practice. What worked? What failed? Relying purely on your
memory of what happened in practice, without tracking things
at the moment, is not the most effective way to do this in my
experience. Instead, I use my phone as a notebook or voice
recorder to make quick observations I can compile and reflect
more on later.
As a way to facilitate these processes, I would highly
recommend you get involved in a “mastermind” type group
with other coaches. This keeps you accountable with regard to
the reflection on and continued iteration of your practice, with
support and advice from people dealing with similar issues. I
have seen advantages to having both sport-specific coaching
groups and ones that include coaches from multiple different
sports. The former allows you to get into the nitty gritty of
sport-specific movement problems and constraints. The latter
gives you more novel ideas from people working on similar
issues in a very different context. If you are interested in
joining a multisport version of a coaching group, I host one
twice a month for supporters of my podcast3
.
Dealing with the Loss of Control & Pressures to Retake It
I think we have all heard them. The comments from the
sidelines when we adopt an ecological approach in practice:
“When are you going to start coaching” “Aren’t you going to
teach them any basic skills like how to shoot or how to
dribble”. This is one of the main findings by Chow, Stone et al.
Another reason why coaches may choose not to adopt ED
approaches could be associated with the expectations from
parents and others around, especially for paid programs. For
example, a coach once shared how he agreed with this
approach, but he was hesitant to implement as parents may
complain if it looked like he was not teaching and if the class
looked messy.
I experienced this myself when I ran practices as an assistant
coach for my 11-year-old stepson's basketball team earlier this
year. There were two particular issues that I wanted to focus
on in one of the practices. The first was team communication
on defense. After timeouts, we were frequently caught in “I
thought you had him” type situations. To address this, I came
up with (what I thought) was the clever idea of mixing up the
teams every 5 minutes without using any way to indicate who
was on each team visually (i.e., no shirt vs skins). This
exaggerated the need to communicate verbally and by pointing
and gesturing on defense (which is what we wanted!). Initially,
there was a lot of confusion and chaos but the behaviors I
was looking for eventually emerged, despite the “you know I
have pinnies (i.e. colored shirts) in my car” comments from the
sideline. The second issue was that the play in our league was
quite physical, and the refs didn’t call many fouls. So, when we
scrimmaged or played small-sided games, I again exaggerated
this by deliberately keeping my whistle in my pocket and letting
them play. This was met with “Is your whistle broken?” “Do
you know how it works” comments.
Excepting the ecological to coaching necessitates giving up some
control as a coach. After all, we call it “self-organization” not
“coach organization”. Treating an athlete as a complex system
means that we cannot demonstrate simple cause and effect in
our coaching methods. Things get a bit chaotic. Your players
are not always taking the affordances you had hoped they
would. All the while you are getting comments from the
sidelines from people indoctrinated in the traditional view of skill
acquisition. When this happens, it is so tempting (and easy) to
revert to layup lines, dribbling around cones, using agility
ladders, and giving a solution to your athletes via prescriptive
instruction. It will make everyone a lot happier! It will restore
your image as the “boss” of the team. There is evidence it will
even make the players themselves more satisfied.
A recent application of the CLA I have been involved with is
its use in police training. If you think about, there is not a
better example of the need for being adaptable in the face of
completely unpredictable constraints, than policing. Initial
evidence seems to support this view. For example, Koerner et
al. (2021)4 compared traditional prescriptive training with a
CLA approach for teaching police recruits to defend against
knife attacks. The CLA group always practiced against an
opponent with constraints manipulations including the size of
space, angle of attack, weapon, and number of attacks. The
prescriptive group were taught the techniques for defending and
practiced these “on air” (i.e. with no opponent) with lots of
correction from the coach. Paralleling the findings in sports, the
results showed a clear benefit for CLA training in terms of the
number of attacks successfully defended, time to disarm the
attacker, and performance in a transfer condition that involved
defending against a surprise attack. What I want to focus on
here, though, are the reflections of the learners. Despite these
clear learning benefits, comments included:
What I would have wished for was that we had learned a
technique. You didn’t quite know whether it’s right or wring or
what you should do, you just did something and it actually
works! But a technique would not have been bad.
And..
That one (a participant in the prescriptive group) gets feedback
about whether the defense was done right. I think such
feedback is very important, because you do not look at yourself
from the outside. You are in the situation, you do know exactly
what you have done now in retrospect.
So, even the learners themselves, who have experienced all the
amazing benefits of the ecological approach that have I been
touting throughout this book, sometimes want the coach to take
control and be in charge.
Within this domain of police training, even researchers seem to
want to stubbornly hold on to their traditional beliefs about
learning in the face of completely contradictory evidence. For
example, a recent study showed that police officers trained and
taught in the same academy (for over 200 hours) had a high
degree of variability in the specifics of how they responded to
the same event5 (e.g., where they parked their car, their
approach angle, when they drew their gun, etc.). This is, of
course, what we expect in the ecological approach – instead of
executing a set procedure, the officers are demonstrating
“repetition without repetition” and adapting to the constraints.
But, despite this result, one of the authors' main conclusions
was that the officers had “insufficient training and experience to
demonstrate shared knowledge (mental models)”. The
dominant, traditional approach to skill is truly firmly rooted!
This is why I am so strongly and stubbornly opposed to the
multi-theoretical, “pick n mix”, “it depends” (theoretically)
approach to coaching discussed in Chapter 2. If I allow myself
the option of switching to being prescriptive, decomposing, and
using strict repetition, why wouldn’t I in the face of all this
incredible pressure to switch? As Stone et al. acknowledge:
The biggest challenge is the fear of “letting go” on the part of
the practitioner. There is typically a strong desire to prescribe
how learning should exactly occur for the learner which leads
to a teacher-centred approach. The practitioner prescribes
instructions that dictates the expected movement form expected
from learners without accounting for individual differences in the
learning context. There is a need to be the holder of
knowledge and experience on how learning should be
undertaken. Thus, it would need a mindset change on the part
of the practitioner to be “comfortable being uncomfortable” to
let learners explore their own movement solutions to accomplish
task goals set out in the session.
I know there are some out there saying “We should give them
what they want”. That is, if trainees want to be prescribed a
solution (as seemed to be the case in the knife defense study)
shouldn’t we give it to them? No, I don’t believe so. While we
want the athlete to be involved in their skill development,
sometimes we know better. If our knowledge of and about as
a coach (i.e., our model or theory of learning) goes against
what the athlete thinks they want then I think we need to do
some re-education, which leads to my main recommendation for
dealing with this loss of control.
Educate the Athletes and the Parents About your Approach
One of the most rewarding things I have done in my role as a
skill acquisition specialist is to develop and deliver a presentation
to the athletes I work with called “What Learning Really Looks
Like”. The purpose of this presentation is to set expectations
for what the practice environment is going to look like and to
change some of the entrenched, out-of-date views of skill
acquisition. In this presentation I try to hit on these key points:
1)
To Get Through Easy You Have to Go Through Hard. To
become skillful and be able to do things effortlessly in
competition requires being challenged and failing in practice. We
don’t get better when we do everything perfectly every time –
aka “Perfect practice DOES NOT make perfect”. For this, I love
to use the image developed for a Mountain Dew advertisement
by Richard Wagner (shown here
6
). This wonderful image
shows a skateboarder landing a difficult trick perfectly, mixed in
with images of all the different ways he fell/failed while learning
to do it.
2) We Need to Practice with Intent to Work on Things That
We Aren’t Already Good At
! Here I am emphasizing ideas
from deliberate practice (individualized practice focused on your
weaknesses) and the different goals of practice discussed in
Chapter 12. I emphasize that our intentions serve to organize
our actions and the importance of having SMART goals
(specific, measurable, attainable, relevant, and time-bound).
3)Stop Thinking About Your Mechanics/Technique – Let the
Coach Worry about it Instead. Here we are talking about the
detrimental effects of an internal focus of attention and
introduce how your coach is going to guide you in finding new
techniques using the CLA.
4)
Practice Hard to Make the Game Easy. We are deliberately
going to put you into conditions that are more difficult than
you will face in the game, and you are going to fail sometimes.
This will make performing in competition easier. To quote Steph
Curry: “We overload in our workouts, so the game slows down
in real life”.
5)
Play the Long Game: You Won’t Likely See Gains Every
Day. Here I emphasize the nonlinearity of learning. Sometimes
we will have to go through periods of suboptimal movement
solutions and high instability to get to a better place in the long
run. I emphasize that great athletes are adaptable.
6)
I Don’t Know. Let’s Find Out Together Despite these
attempts to re-educate I still get players saying things like: “Just
tell me how I need to swing to get the ball in the air”. So, I
like to try to nip that in the bud by introducing ideas of
self-organization.
I have received a lot of positive feedback about these
presentations. And it was particularly rewarding for me when I
saw one the Boston Red Sox players, Trison Casas, say the
following in an article7
:
Every swing is different...Even in batting practice, every single
pitch is different. Having a variety of swings to match those
planes is what makes great hitters great, and leaves good
hitters to just be average.
Adaptability and dexterity! :)
In terms of educating parents and other interested parties that
aren’t on the field, I have decided that before my next season
as a flag football coach, I am going to give a comparable “This
is How I Coach and Why?” type of presentation. Keep an eye
out on my YouTube channel (@RobGrayASU) – once I have
designed it, I will record it. and make it freely available.
The Challenge of Assessment & Measurement
This last big challenge faced by ecological coaches is one of the
most difficult ones to handle: how do I know (and show!) that
my athletes are improving when coached via an ecological
approach? This is something I am going to dive into more next
chapter.

15
RETHINKING HOW WE TRACK &
ASSESS PROGRESS

I f we are not coaching an athlete to hit specific milestones of

fundamental movement goals on a checklist or looking for linear
increases in some performance metric then how do we know
whether or not they are progressing? How do we know
whether motor learning is occurring? Traditional assessment
tools in coaching – our skills tests – usually involve
measurement in impoverished, out-of-context performance
environments. We try to forecast future in-game success based
on the ability to perform isolated skills. We are measuring the
repeatability of one movement solution (technique) rather than
the ability to adapt our movement solutions (dexterity, skill)2.
These tests place a very major constraint on coaches’ ability to
align with the key principles of the ecological approach. In
many ways, we need to be as creative in our assessments of
athletes as we are in designing practice.
What we need is a range of assessment tools that assess
things like exploration of the solution space, sensitivity to
affordances, and adaptation to constraints. While, as a field,
ecological dynamics is not quite to the point where such metrics
are readily available there is encouraging early work in this area
which includes suggestions for how we might design our own.
So, I want to start this chapter by looking at these novel ideas.
I will then turn to ways that we can use existing metrics (e.g.,
biomechanics data) in novel ways that align with the ecological
approach.
Assessing Exploration
Notational Analysis of Affordances
An interesting attempt to move assessment in this direction is
Lopez-Felip and Turvey’s functional semantics for sport1
. This
is essentially a new approach to the notational analysis2 of a
game using video/GPS data that assesses the affordances taken
by a player instead of just their number of shots, passes, and
drives (taken without considering the context). The basic idea is
illustrated in Figure 15.1. The top panel of this figure illustrates
the situation where forward B moves into an open space and
midfielder A passes him the ball. Instead of asking just whether
this pass was successful or not (which depends on a multitude
of factors) we can ask whether it was afforded or not (i.e., we
can separate the process, the decision, from the performance
outcome). To do this, we can look at three things: (1) the
information (e.g. what is the size of the gap between the
defenders and how fast is it closing?), 2) the relevant
effectivities (e.g., what is the maximum speed and precision with
this player A can pass the ball), 3) other contextual
factors/information (e.g., the location on the field where this
event is occurring).
This latter factor is illustrated in the bottom panel of Figure
15.1. Here the field is divided into quadrants and three zones (
intervention, help, and cooperation). Each of these quadrants
(I-IV) and three zones define the affordances available to the
defender based on their distance from the ball carrier. In the
Intervention Zone, the defender’s proximity to the ball primarily
affords the overarching activity of intervening (e.g., stealing the
ball). Players in the mutual help zone can support the nearest
defender's ability to intervene (e.g., by following the principles of
play of “cover” and “compactness”, discussed in Chapter 12 but
cannot currently directly intervene themselves). Finally, players in
the cooperation zone cannot directly support intervention but
instead, cooperate by focusing on principles of play like
“support” and “balance”.
Lopez-Felip and Turvey break down the affordances for the
situation illustrated in Figure 15.1 Referring to the top panel,
before player B makes their run, A has the ball in Region III.
B is in Region IV surrounded by opposing players and at
some distance from the ball. The dynamics of the current
situation would not place A in a situation X that affords A the
act of passing to B. Next, B perceives that his or her current
zone in Region IV does not afford to receive a pass from A. B
moves to change the situation. Specifically, B moves to become
momentarily unmarked from his or her defenders and, in so
doing, creates a momentary open space. The new situation
affords A the opportunity to pass to B on this new occasion.

A diagram of a football field Description automatically generated

Figure 15.1 – A Functional Semantics for Notional Analysis of
Sports Performance. Based on analysis from Lopez-Felip and
Turvey
1
Assessing Visual Exploratory Behavior
Along with having our athletes explore the solution space for
different movement solutions, we also want them to explore the
information space to educate their attention to higher order,
specifying information for controlling actions. One way this can
be assessed is by looking at the extent to which an athlete is
visually exploring the playing environment. I looked at some of
the research on this topic (done by Geir Jordet and colleagues)
in Chapter 3 of my second book.
Since this original work, there have been several studies that
provide support for the idea that head-turning/scanning
behavior can be used as an exploration metric, but with some
important limitations. For example, Jordet and colleagues3
investigated the scanning behavior of 27 elite professional
football players in an English Premier League club. Players were
filmed across 21 matches, producing a total number of 9,574
individual ball possessions for analysis. Using a model that
accounts for individual player differences and pass difficulty, it
was found that the more a player scans, the higher the
probability of completing a pass. Aksum et al. 4
, analyzed the
scanning behavior used by players in 2018 UEFA European
U17 and U19 Championship semi-finals and finals. The older
players (U19) players performed more scans than younger, U17
players. There was also again a positive relationship between
scan frequency and pass success.
Aksum and colleagues5 performed a more detailed analysis of
scanning behavior in 11 vs. 11 real-game environments using
mobile eye-tracking. More specifically, they looked at the
duration and content of the scans (e.g. whether they looked at
opponents or teammates). What was found? The players’
scanning duration was influenced by the ball context and the
action undertaken with the ball at the time of scan initiation.
Furthermore, fixations were found in only 2.3% of the scans.
Additionally, the results revealed that the stop point is the most
information-rich part of a scan and that the players had more
opponents than teammates inside their video frame during
scans.
But there were also some caveats. First, there is evidence that
this metric obeys Goodhart’s Law6
, which states that when a
measure becomes a target, it ceases to be a good
measure. Specifically, it has been reported that, since it is being
assessed, youth players have now started using rapid head
turns in which they don’t actually look at anything on the
playing field so that they can get a good score in the “head
turn count” measure7
. Second, research suggests while it is
important, visual scanning behavior might not have as big of an
effect on passing performance as the original research suggests.
For example, in a study looking at the scanning behavior of 35
midfielders participating in the Euro 2016 championships, Phatak
and Gruber8 found a significant positive relationship between
scan rate and passing success rate and a significant negative
relationship between scan rate and turnover rate. However, the
scanning variable only accounted for 4% of the variance in the
in-game pass completion and turnover rate. Finally, it has been
shown that visual scanning behavior can not be used to
distinguish super elite (players from the UEFA Champions
League Squad of the Season in the 2018–2019 season) and
elite players (teammates of the super elites) 9
.
Despite these limitations, I think visual scanning behavior
(assessed via head turning or using mobile eye tracking which
is becoming more and more affordable and easy to use) could
be an effective way to assess to what extent the constraints we
use in practice encourage exploration. For example, I have
found a relationship between pre-pitch exploratory behavior and
infielder performance (e.g., range, conversion rate) in baseball.
Measuring Adaptability to Constraints
A concept I have mentioned a few times already is the
Diagnostic Constraint. This follows a general philosophy I have
when evaluating athletes: I am less concerned with inter-athlete
comparisons (i.e., how your movement solution compares to
player X) than I am about intra-athlete comparisons (i.e., how
your movement solution for constraint A compares to your
movement solution for constraint B, how do you adapt?). For
example, whether a baseball player’s swing looks like Shohei
Ohtani’s is less relevant to me than how your swing for a
fastball looks compared to your swing for a curveball. There
are lots of reasons why the former might differ that I can’t do
very much about (e.g., you are a lot shorter than Ohtani) while
the latter difference reflects your dexterity and adaptability,
which I can influence!
As an example, I recently started analyzing the in-game bat
tracking data that is now available for MLB players10, to see
how batters adapt their swing speed and swing length as a
function of the count (i.e., # of balls and strikes). An example
of this can be seen in Figure 15.2. This figure shows the % of
time this batter (one of the better hitters I worked with) gets
his “A swing” off relative to the MLB average. An “A Swing” is
defined as one that is in the 95% percentile for both bat speed
and swing length. Notice how when this hitter gets ahead in
the count (more balls than strikes, so counts of 2-0, 2-1, 3-0,
3-1), they have more “A” swings. This is the kind of adaptation
of movement solution to changing constraints we want! When
the situation affords hitting the ball hard (because the batter
can’t strike out if they miss) they swing harder. For more of
this type of analysis, check out the great work of Coach Josh
Rodrigues11. I think this type of diagnostic constraint analysis
can be applied to most sports – take a simple aspect of the
movement solution you can easily measure (speed, time,
amplitude, etc.) and see how it (hopefully) systematically varies
as a function of changes in the constraints.
A diagram of a number of individuals Description automatically
generated with medium confidence
Figure 15.2 – Percentage of “A Swings” (95 percentile for bat
speed and length) relative to the MLB average in different
counts.
Dynamical Systems Measures of Individual & Team Performance
An interesting recent attempt to model and assess sports
performance is the dynamical systems model of
attacker-defender in soccer by Brink et al. (2024)12
. In this
model, illustrated in Figure 15.3, the attacker with the ball and
the closest defender are treated as a coupled dynamical system.
Overall movements of the attacker and defender (i.e.,
accelerations) are based on the relative strength of the different
attractors and repellors in the system. For example, the attacker
is attracted to move toward the goal while also trying to get
away from (i.e., repelled by) the defender. The defender on the
other hand is attracted to both the attacker with the ball and
to their goal (to protect it). A key part of this model is that it
incorporates the individual constraint of “aggressiveness” –
which is determined by the relative weighting of the attraction
to the goal and repulsion by the defender (e.g., a very
aggressive player would weigh the attractor to the goal more).
In a game situation, the strength of these attractors and
repellers would also vary based on the relative distances
between the attacker, defender, and goal.
A diagram of a attack Description automatically generatedFigure
15.3 – Dynamical Model of Attacker-Defender Interaction in
Soccer. Based on Brink et al. (2024).
The authors fit this model to data collected from The San Jose
Earthquakes Major League Soccer (MLS) team during 17 home
games from the 2019 season. Data were analyzed for four
different players. The model provided a good fit to the data,
but the most interesting aspect was the other analyses of how
varying the model parameters changed the outcomes for each
player. For this, the authors created a dribbling score which
depending on how much each dribbling maneuver: (i)
decreased the distance to the goal and/or (ii) increased the
angle to the defender. As would be expected increasing the top
running speed (a variable in the model) by 25% was predicted
to increase the dribbling score for all players, but by different
amounts (ranging from 2% for player 4 to 10% for player 1).
Conversely, changing the “aggressiveness” parameter had very
different predicted effects. A 50% increase in this parameter led
to an increase in dribbling scores for two of the players and a
decrease for the other two.
Overall, I think this model presents some very intriguing ways
of assessing and tracking performance that align well with the
ecological approach to skill development. For example, the
dribbling score quantifies the ability of a player to take
affordances (getting closer to the goal and protecting the ball
from the defender) that align well with some of the principles
of play discussed in Chapter 12. For example, the overall
dribbling score essentially is a measure of “penetration”. The
“aggressiveness” parameter provides a way to assess calibration,
that is how well a player's movements are adjusted for their
individual action capacity (e.g., top running speed). The authors
also propose this could be used to develop game plans:
The framework presented here can be used to construct game
plans such as which defender to target, which attacker should
be the dribbling focal point, where on the pitch should the
dribble start, etc. and provide a quantitative metric to assist
scouting by assessing the dribbling skills of any player in terms
of measurable behavioral and physical parameters.
Duarte Araujo and colleagues13-15 have also effectively applied
dynamical systems models to team play in sports. For example,
these authors have shown how different concepts from
ecological dynamics can be used to explain the emergence of
synergies in team play. Recall a synergy occurs when the
different degrees of freedom in a system (in this case, the
player’s on a team) work together and have compensatory
variability to maintain the performance goal. They demonstrate
how the following system properties of a synergy can be used
to understand team play.
1)
Dimensional Compression, a process resulting in independent
degrees of freedom being coupled (i.e., moving together) so that
the synergy has fewer degrees of freedom than the set of
components from which it arises (so essentially a type of
freezing);
2)
Reciprocal Compensation, if one element does not produce
its function, other elements should display changes in their
contributions so that task goals are still attained (e.g. a
defender can switch to playing offense, the “mobility” principle
of play);
3)
Degeneracy, structurally different components performing a
similar, but not necessarily identical, function with respect to
context.
As a more specific examples, consider the study by Duarte et
al16 which examined team coordination tendencies in
association football near the scoring zone. The dynamical
systems measures of interest in this study were the location of
the team centroid (i.e., team center, illustrated in Figure 15.4,
top panel) and surface area (i.e., occupied space), illustrated in
Figure 15.5, top panel). As can be seen in Figure 15.4 (bottom
panel), the centroids for the offense and defense demonstrated
a strong symmetric relation (i.e., their distances to the goal tend
to follow each other) but there was a crossover of the centroid
distance (created by a pass) moving the offense’s centroid
closer to the goal just before a goal was scored. Analysis of the
surface area of each team did not reveal a clear coordination
pattern between sub-groups. However, as can be seen in
Figure 15.5, a large reduction in SA for the defensive team
(which violates the “width” principle of play) often immediately
precedes a goal.
A football field with a diagram Description automatically
generated
A graph of a ball control Description automatically generated
with medium confidence
Figure 15.4 – Top: Distance of Team Centroid (dc
) from the
Goal. Bottom: Change in centroid distances of the offensive and
defense leading to a goal. Based on Duarte et al16.
A green field with red balls Description automatically generated
A graph of a ball control Description automatically generated
Figure 15.5 – Top: Team Surface Area (SA). Bottom: Change
in surface areas of the offense and defense leading to a goal.
Based on Duarte et al
16.
Using Biomechanical Data to Assess Invariants and Attractors
One of the major technological innovations coming to a lot of
sports now is the ability to do biomechanical analysis from
in-game data. Markerless motion tracking systems like Kinatrax
TM and HawkeyeTM are making this easier and easier. But
what should we do with this data? The initial impetus was
around identifying the details of the optimal technique (e.g. what
stride length produces the best pitch velocity in baseball, what
amount of knee bend results in the most accurate 3-point
shooter in basketball). Hopefully, to no one who is reading this
book’s surprise, this has been unsuccessful. So, what should we
do instead? The answer is that we need to put the bio back in
biomechanics. Specifically, we can use biomechanics data to look
for attractors in the system
Recall, that an attractor is an area of movement solution space
to which we get pulled into because it represents a point of
stability in the system. This creates a topological structure with
the solution space called an invariant set. Every attractor is an
invariant set. The two are directly linked. So, when we look for
invariants in a movement solution (as discussed in Chapter 8)
often we are looking for attractors. Attractors can occur for
several reasons but one of the most prominent is our individual
constraints, namely our anatomy. An athlete will get “attracted"
to certain movement patterns under certain constraints imposed
by their body, the environment, and the task. For example, if I
increase my speed of locomotion toward a target, I will first
walk, and then walk a bit faster, and then still faster, and then
I will suddenly shift into running, a totally different pattern of
movement. This transition between attractors is called a phase
shift.
Frans Bosch has identified a few universal (because they arise
from the constraints imposed by our anatomy) attractors
associated with human movement. One of the most effective
uses of the sports biomechanics data is to look for evidence of
these attractors – to complement a coach’s observation of the
invariants described in Chapter 8. Let’s look at an example.
Recall, that proximal to distal action is a movement solution for
which body parts near our center (the pelvis) reach their
maximum velocity before parts further from our center. This
attractor can be evaluated by looking at kinematic sequencing in
our movements. For example, Donna Scarborough and
colleagues have studied this in baseball pitching, as illustrated in
Figure 15.6. This figure shows the angular velocity for five
different body parts: the pelvis, trunk, upper arm, hand, and
forearm. The first thing to take note of is the scale on the
y-axis (up to 5000 degrees/second!). The rotations involved in
pitching are some of the fastest human movements ever
recorded. Moving from left to right on the x-axis, we can see
when the peak velocities for the different body parts occur. For
the most part, the movement from this pitcher shows the
proximal to the distal attractor. The pelvis (1) hits its maximal
velocity first, followed by the trunk (2), followed by the upper
arm (3). Then we see a slight deviation in the order as the
maximal rotation of the hand (5) occurs before the maximal
rotation of the forearm (4).

A diagram of a variety of skeletons Description automatically

generated
Figure 15.6 – Kinematic Sequencing in Baseball Pitching. Based
on data from Scarborough et al. 16.

In a couple of studies, Scarborough et al.

17,18 have looked at
the relationship between this kinematic sequencing and variables
such as pitch velocity, and injury susceptibility. Although there is
a fair degree of variability in the specific sequence used (e.g.,
12345 vs 12354 as shown in Figure 15.6) there were some
general patterns consistent with the importance of this attractor.
For example, when the maximal pelvis rotation occurs after
maximal trunk rotation (i.e., 2 occurred before 1 in Figure 15.6)
this was associated with increased external shoulder and elbow
torques, which can increase the risk of anterior shoulder and
ulnar collateral ligament injuries, respectively. Furthermore, the
torque force in the shoulder was lowest for the “almost”
proximal to distal sequence (12354). Finally, for each pitcher,
pitch velocity was highest for their most proximal-to-distal
sequence. I think using motion tracking data to assess attractors
and invariants like this is an effective way to use this
information.

16
TOOLS FOR COACH EDUCATION:
WORKSHOPS, MENTORSHIPS, &
CONSTRAINING THE COACH

P ulling this all together, what are some of the tools we can
use to support the development of the ecological coach? In this
chapter, I want to focus on three things that I have found to
be particularly useful in supporting this journey: interactive
coaching workshops, mentorships/apprenticeships, and using the
CLA for coach development.
Interactive Co-Education Practice Design Workshops
One of my favorite things to do as a skill acquisition specialist
is to facilitate practice design workshops. So, instead of just
listening to me give a lecture about the CLA and ecological
dynamics theory, we get down to the nitty-gritty of designing
specific constraint manipulations (including instructional ones) for
specific issues coaches face. My goal with these is to address
the gap in coach education between teaching “what to coach”
and “how to coach it”, discussed in Chapter 1. The focus of
the workshops is on the latter, where the process of thinking
through the CLA design (and understanding its logic) is more
important than the specific practice activities we come up with
in the end.
The effectiveness of these types of workshops has been
investigated in two studies by Andrew and colleagues. The first
1 looked at workshops designed to change how coaches use
cues and instructions. This was focused on two things: 1)
shifting coaches from the predominant use of internally-focused
cues to using more externally-focused ones (to be consistent
with the “tidal wave” of research evidence), and 2) reducing
the number of instructions coaches give while the athlete is
performing the action (called concurrent instructions) which
have shown to be less effective in supporting self-organization
as compared to instructions given before and after the action2
.
Another important element of these workshops is that they
were co-created. That is the researchers worked with the key
stakeholders in the team (e.g., the Head of Sports Science, the
Academy Director, and/or the General Manager) to decide
which information to include and how to best connect it with
the specific context and form of life of the team they were
working with.
In their study, Andrew et al created and evaluated this type of
workshop for soccer coaches. In the first phase of the
workshop, the researchers systematically observed and
videotaped five coaches running their practices to evaluate how
instructions were used and to provide feedback to the coaches
(e.g., the proportion of different types of instructions illustrated
as pie charts).
In the second phase, skill acquisition specialists lead 60-minute
educational workshops. For these, external vs. internal focus of
attention were defined, explained, and key research findings
were discussed. To relate to the environment, soccer-specific
examples of focus of attention instructions were provided to
coaches (e.g. for passing: ‘
Use the inside and outside of your
foot
’ vs “‘
Move the ball left and right’). Coaches were
encouraged to be active in this process by discussing the
principles of instructional behaviors (peer-to-peer learning and
coach-to-researcher learning) to establish how, why, and when
they could change their previous instructional behaviors.
What was found? Figure 16.1 (top panel) shows the timing of
instructions given by the coaches. As predicted, most
instructions (64%) were concurrent. There were no significant
differences in the timing of instructions pre- and post-workshop.
Figure 16.1 (bottom panel) shows the content of the instructions
before and after the workshop. There was a significant change
in cue content. Following the workshop, coaches used
significantly more (
26% increase) external focus of attention
cues and significantly fewer internal focus cues (25%
decrease).
I think the discrepancy in these findings for cue timing and cue
content can be understood by appealing to the distinction
between knowledge about and knowledge of, discussed in
Chapter 1. Changing from internal to external focus of attention
instructions would seem to be something that can be supported
by traditional “knowledge about”classroom instruction (as was
used in this study) – you are changing what they are
coaching. Timing likely needs to be taught by focusing on
“knowledge of”. That is, you need to practice it in situ, doing it
in context because we are trying to change more about “how”
to coach.
The second study3 looked at workshops designed to increase
the use of games-based, tactical activities (e.g. small-sided
games) and reduce the use of drill-based, technical activities
(e.g. dribbling around cones, ball handing “on air”) in soccer
coaches. The same basic workshop format was used. Practice
sessions were first observed and analyzed. During the
workshops coaches were: (i) shown the research comparing
these two types of activity, (ii) given specific feedback about
their usage, and (iii) involved in a discussion of ways they
could change their practice design. Data on the types of
practice activities used were collected post-workshop and 3
months later, to assess retention. What was found? The results
are shown in Figure 16.2 (top panel). For drill-based activities,
there was no significant change in their usage from pre- to
post-workshop. For games-based activities, there was a
significant increase in their usage both immediately after the
workshop (by about 17%) and 3 months later.

A graph showing the time of a performance Description

automatically generated with medium confidence
A diagram of a cross between two different types of focus
Description automatically generated
Figure 16.1 – Cue timing (top) and content (bottom) before
and after a coach education workshop. Based on data from
Andrew et al1
.

In a second experiment, the authors evaluated the effect of a

comparable workshop designed to increase the use of practice
activities that have an active decision-making (ADM) component.
As shown in Figure 16.2 (bottom panel) this was also
successful as the use of ADM activities increased significantly
(by about 13%) from pre to post workshop. However, note
that the time for these activities was taken out of warmup,
fitness, and transition activity time – the time devoted to
NADM did not change significantly. The same was true from
the drill-based activities shown in the top panel. So perhaps
more work is needed to get coaches to reduce the use of
these technical and NADM drills as recommended by Ford et
al..4 and discussed in Chapter 2
So, in sum, research on the effectiveness of this type of coach
education workshop is very promising. They do seem to
promote behavioral change in coaches! This agrees with my
own experience using them. Critically, however, I think it’s
important to think about when they will and will not be
effective. I think they are a good fit when we want to change
“offline” coach behaviors (e.g. practice design, choosing an
appropriate constraint, analyzing practice data) that are more
focused on what to coach. If we want to change “online”
behaviors (i.e., things that occur in the practice like adapting
constraints and the timing of feedback) these are going to have
to be taught more in situ, to support the development of
“knowledge of”.
Either way, I would highly recommend you try one of these
coach education workshops (either with the coaches in your
organization or if you are a solo practitioner, coaches from
other organizations). If you are looking for a skill acquisition
specialist to facilitate a workshop, please don’t hesitate to reach
out!

A graph of activity and activity Description automatically

generated
A graph of a bar graph Description automatically generated
with medium confidence
Figure 16.2 – Using of different types of practice activities
before and after coach education workshops. Based on data
from Andrew et al
3
.

Mentorships, Apprenticeships & The Situated Coach

It may have taken us 16, long-winded chapters to get here but
I think we can come to a simple (and maybe obvious)
conclusion: coaches are more likely to effectively learn their
craft by practicing it in context - By doing it. Meaningful
learning is situated – it occurs within a specific context. It
shapes and is shaped by the “form of life” and the community
of practice with which is entangled. As proposed in a recent
article by Selmi and Woods5
, situated coach learning involves
two interwoven dimensions: (1) exposure to real-world contexts
via an apprenticeship/mentorship relationship and (2) legitimate
participation within communities of practice.
For exposure to real-world context, we need to fundamentally
shift coach education from educative experiences in
decontextualized settings (i.e., classrooms) toward short-term
apprenticeships with a coach mentor. That is, we need to do
more actual coaching when learning to coach – developing
knowledge of. The mentor (or coach developer) can serve a
few important roles including purposely manipulating the
constraints on the coach (more on that at the end of this
chapter), providing systematic and objective observation of what
worked and what didn’t, and helping coaches to self-regulate
(plan, reflect and revise their practices).
Consistent with the principles of ecological dynamics, the role of
a mentor in a more situated approach to coach education
would be one of collaboration and guidance without specification
or prescription (i.e., coach it like this)6
. This could be achieved
using what Rudd et al.
7 refer to as ‘soft’ pedagogical
approaches, like nudging or questioning, which encourage less
experienced coaches to attend to things directly.
A community of practice (CoP)
4 can be defined as a group of
people ‘who share a concern, a set of problems, or a passion
about a topic and who deepen their knowledge and expertise
in this area by interaction on an ongoing basis. Instead of
complaining under your breath about parent’s comments and
the demands placed on you by a manager, take the
opportunity to engage. Give a “This is How I Coach and
Why?” presentation, like was discussed in Chapter 14. Educate
and provide them with opportunities for more useful feedback.
Constraining the Coach
In the coaching model laid out in Chapter 3, the methods used
by a coach are emergent. That is, they arise from a process of
self-organization in the face of a particular set of constraints.
Therefore, it is possible to use the same CLA logic we use for
skill development in our athletes for skill development in
coaches. That is by changing constraints we can encourage
coaches to explore the coaching solution space and find new
and more effective solutions much in the same way we do with
our athletes.
Some of these constraint manipulations will happen naturally.
For example, in the football practice I ran the other night I
had only six kids show up (it was Fall break). I wanted to
work on shared affordances, communication, and coordination
when playing defense. But I quickly realized that practicing
coordination between our five starting defensive players wasn’t
going to work too well when there was only 1 offensive player
for them to cover. I had to adapt. I split the field into smaller
chunks and worked on coordination between smaller groups of
players instead.
We (or our mentor/coach developer) can also do this more
purposely. We can follow the same CLA logic I laid out in
Chapter 5. We can purposely add a constraint to the practice
environment that takes away a coaching solution and
encourages coaches to explore. For example, one coaching
solution that I find myself wanting to destabilize often is the
frequent use of overly technical, internally focused feedback
given to players. The constraint I like to add is music, played
loudly. The athletes love it and it makes it so comments like
“keep your elbow straight” and “your knees bent” can’t be
heard very easily. When the coach must go right up to the
athlete to give them an instruction, they say less. Or even
better, they start trying to make their point by changing the
constraints.
Another fun one I like to use is the “replace cones with
players” coaching constraint. So, every time a coach wants to
use a cone they must put a player there instead. For example,
if they want to use a cone to make a basketball player take a
certain angle to the hoop for a layup, put another player there
instead. They are allowed to tell the “cone players” what they
can and can’t do. I find this becomes a natural way to
encourage the coach to use the “constrain the opponent”
method, discussed in Chapter 6. Even though they could say
“Just stand there like you are a cone”, they quickly see the
other possibilities (e.g. Just stick your arms out, or you can
only move to your left).
Another one of my favorites is the “Take away a rule” coach
constraint. Here the coach must choose which of the regulation
rules to remove from a scrimmage in practice. For example, in
basketball, they could remove the traveling or double dribbling
rule. They could make it so goaltending is allowed or remove
the shot clock. This helps to build a “constrain to afford”
mindset. If I change this it's likely to encourage players to do
that.
Finally, I use the “anything but the regulation ball” coach
constraint. For this, I show up to practice with as many
different balls as I can find: soccer balls, rugby balls, tennis
balls, beach balls, volleyballs, etc. I tell the coach they can use
any of these except the one that was meant for their sport!
Hopefully, you can see there a lots of different fun ways you
can go with this. Your athletes will probably have a lot of ideas
too! But ideally, you want to individualize – design a constraint
based on the individual coach you are working with. For
example, for coaches that are in love with technical,
skill-and-drill type activities, constrain them to design a SSCG
that could teach the same skill.
______________________________________________
_____
Ok, we have come to the end, coaches. Thus, completes my
“learn” trilogy. Athletes learning basic patterns of coordination to
athletes optimizing their movement solutions to (now) coaches
helping to facilitate these processes. I hope my unique
perspective (being a researcher studying skill acquisition in a lab
setting, a consultant who must find effective ways to apply this
knowledge on the field, and now a coach again, myself) has
provided you with some ideas to run with. Thanks for reading,
coach! Cheers for now and keep em’ coupled.
NOTES
Preface
Woods, C. T., Rudd, J., Gray, R., & Davids, K. (2021).
Enskilment: An ecological-anthropological worldview of skill,
learning and education in sport. Sports Medicine-Open, 7(1),
1-9.
Gray, R. (2021). How We Learn to Move: A Revolution in
the Way We Coach & Practice Sports Skills. PACE
Publishing.
Gray, R. (2022). Learning to Optimize Movement: Harnessing
the Power of the Athlete-Environment Relationship. PACE
Publishing.
Button, C., Seifert, L., Chow, J. Y., Davids, K., & Araujo, D.
(2020). Dynamics of skill acquisition: An ecological dynamics
approach. Human Kinetics Publishers.
Gibson, J. J. (1979). The Ecological Approach to Visual
Perception. Hillsdale, NJ: Lawrence Erlbaum & Associates.
Selimi, E., & Woods, C. T. (2024). Enskilment into the
coaching landscape: towards a situated approach to coach
education in Australian football. Sport, Education and Society,
1-14.
Morris, C. E., Davids, K., & Woods, C. T. (2022). On the
wisdom of not-knowing: reflections of an Olympic Canoe
Slalom coach. Sport, Education and Society, 1-12.
https://perceptionaction.com/comparative/
Selimi, E., Lascu, A., Serpiello, F., & Woods, C. T. (2023).
Exploring football coaches’ views on coach education, role,
and practice design: An Australian perspective. Plos one,
18
(5), e0285871.

Knowledge “Of” & “About” and the Problem of

Path-Dependency in Coaching

Turning Point Activity

https://www.exploratorium.edu/cycling/brakes3.html
Gibson, J. J. (1966). The Senses Considered as Perceptual
Systems. Boston, MA: Houghton Mifflin.
Milner, D., & Goodale, M. (2006). The visual brain in action
(Vol. 27). OUP Oxford.
Squire, L. R. (2009). The legacy of patient HM for
neuroscience. Neuron, 61(1), 6-9.
Whitwell, R. L., Milner, A. D., & Goodale, M. A. (2014). The
two visual systems hypothesis: new challenges and insights
from visual form agnosic patient DF. Frontiers in neurology, 5
, 255.
Lefebvre, J. S., Evans, M. B., Turnnidge, J., Gainforth, H. L.,
& Côté, J. (2016). Describing and classifying coach
development programmes: A synthesis of empirical research
and applied practice. International Journal of Sports Science &
Coaching,
11(6), 887-899.
Stoszkowski, J., & Collins, D. (2016). Sources, topics and use
of knowledge by coaches. Journal of sports sciences, 34(9),
794-802.
Horn, R. R., & Williams, A. M. (2004). Observational
learning. Skills acquisition in sport: Research, theory, and
practice, 175-203.
Wood, M. A., Mellalieu, S. D., Araújo, D., Woods, C. T., &
Davids, K. (2022). Learning to coach: An ecological dynamics
perspective. International Journal of Sports Science &
Coaching, 17479541221138680.
Smith, K., Burns, C., O’Neill, C., Duggan, J. D., Winkelman,
N., Wilkie, M., & Coughlan, E. K. (2023). How to coach: A
review of theoretical approaches for the development of a
novel coach education framework. International Journal of
Sports Science & Coaching, 18(2), 594-608.
Morris, C. E., Davids, K., & Woods, C. T. (2022). On the
wisdom of not-knowing: reflections of an Olympic Canoe
Slalom coach. Sport, Education and Society, 1-12.

2. The Value (and Limitations) of Using Skill Acquisition Theory

and Research in Coaching
Ford, P. R., Yates, I., & Williams, A. M. (2010). An analysis of
practice activities and instructional behaviours used by youth
soccer coaches during practice: Exploring the link between
science and application. Journal of sports sciences, 28(5),
483-495.
Buszard, T., Reid, M., Masters, R., & Farrow, D. (2016). Scaling
the equipment and play area in children’s sport to improve
motor skill acquisition: A systematic review. Sports medicine, 46
,
829-843.
Farrow, D., & Buszard, T. (2017). Exploring the applicability of
the contextual interference effect in sports practice. Progress in
brain research,
234, 69-83.
Otte, F. W., Millar, S. K., & Klatt, S. (2019). Skill training
periodization in “specialist” sports coaching—an introduction of
the “PoST” framework for skill development. Frontiers in Sports
and Active Living, 61.
Gray, R. (2022). Learning to Optimize Movement: Harnessing
the Power of the Athlete-Environment Relationship. PACE
Publishing.
Chua, L. K., Jimenez-Diaz, J., Lewthwaite, R., Kim, T., & Wulf,
G. (2021). Superiority of external attentional focus for motor
performance and learning: Systematic reviews and
meta-analyses. Psychological Bulletin, 147(6), 618.
Porter, J., Wu, W., & Partridge, J. (2010). Focus of attention
and verbal instructions: Strategies of elite track and field
coaches and athletes. Sport Science Review, 19(3-4), 77.
Halperin, I., Chapman, D. W., Martin, D. T., Abbiss, C., & Wulf,
G. (2016). Coaching cues in amateur boxing: an analysis of
ringside feedback provided between rounds of competition.
Psychology of Sport and Exercise, 25
, 44-50
van der Graaff, E., Hoozemans, M., Pasteuning, M., Veeger, D.,
& Beek, P. J. (2018
). Focus of attention instructions during
baseball pitching training. International Journal of Sports Science
& Coaching, 13
(3), 391-397.
Wulf, G., & Shea, C. H. (2002). Principles derived from the
study of simple skills do not generalize to complex skill learning.
Psychonomic bulletin & review, 9
(2), 185-211.
Watts, R. G., & Bahill, T. (2000). Keep your eye on the ball:
curve balls, kunckleballs, and fallacies of baseball. W.H Freeman.
Ammar, A., Trabelsi, K., Boujelbane, M. A., Boukhri, O., Glenn,
J. M., Chtourou, H., & Schöllhorn, W. I. (2023). The myth of
contextual interference learning benefit in sports practice: A
systematic review and meta-analysis. Educational Research
Review, 100537.
Brady, F. (2008). The contextual interference effect and sport
skills. Perceptual and motor skills,
106(2), 461-472.
Winstein, C. J., & Schmidt, R. A. (1990). Reduced frequency
of knowledge of results enhances motor skill learning. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 16
(4), 677.
Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984).
Knowledge of results and motor learning: a review and critical
reappraisal. Psychological bulletin, 95
(3), 355.
Wulf, G., Shea, C. H., & Matschiner, S. (1998). Frequent
feedback enhances complex motor skill learning. Journal of
motor behavior
, 30(2), 180.
McKay, B., Hussien, J., Vinh, M. A., Mir-Orefice, A., Brooks,
H., & Ste-Marie, D. M. (2022). Meta-analysis of the reduced
relative feedback frequency effect on motor learning and
performance. Psychology of Sport and Exercise, 61, 102165.
https://blogs.fangraphs.com/a-sweet-spot-by-any-other-definition/
Castaneda, B., & Gray, R. (2007). Effects of focus of attention
on baseball batting performance in players of differing skill
levels.
Journal of Sport and Exercise Psychology, 29
(1), 60-77.
Uehara, L., Button, C., Falcous, M., & Davids, K. (2016).
Contextualised skill acquisition research: A new framework to
study the development of sport expertise. Physical Education and
Sport Pedagogy,
21(2), 153-168.
Uehara, L., Button, C., Araújo, D., Renshaw, I., & Davids, K.
(2018). The role of informal, unstructured practice in developing
football expertise: the case of Brazilian Pelada.
Journal of
Expertise, 1
(3), 162-180.
Uehara, L., Davids, K., Pepping, G. J., Gray, R., & Button, C.
(2022). The role of family and football academy in developing.
International Sports Studies, 44(2), 6-21.
Chow, J. Y., Davids, K., Button, C., & Renshaw, I. (2021).
Nonlinear pedagogy in skill acquisition: An introduction.
Routledge.
Collins, D., Taylor, J., Ashford, M., & Collins, L. (2022). It
depends coaching–The most fundamental, simple and complex
principle or a mere copout?. Sports Coaching Review, 1-21.
378 – Achieving Balance (and Integration?) in Skill Acquisition
Research & Theory https://perceptionaction.com/378/
Stone, J. A., Rothwell, M., Shuttleworth, R., & Davids, K.
(2021). Exploring sports coaches’ experiences of using a
contemporary pedagogical approach to coaching: An
international perspective. Qualitative Research in Sport, Exercise
and Health, 13(4), 639-657.
Benson, B. L., Anguera, J. A., & Seidler, R. D. (2011). A
spatial explicit strategy reduces error but interferes with
sensorimotor adaptation. Journal of neurophysiology,
105(6),
2843-2851.
Mazzoni, P., & Krakauer, J. W. (2006). An implicit plan
overrides an explicit strategy during visuomotor adaptation.
Journal of neuroscience, 26
(14), 3642-3645.
Miyamoto, Y. R., Wang, S., & Smith, M. A. (2020). Implicit
adaptation compensates for erratic explicit strategy in human
motor learning. Nature neuroscience, 23
(3), 443-455.
Newell, K. M. (2020). What are fundamental motor skills and
what is fundamental about them?. Journal of Motor Learning
and Development,
8
(2), 280-314.
Deuker, A., Braunstein, B., Chow, J. Y., Fichtl, M., Kim, H.,
Körner, S., & Rein, R. (2023). “Train as you play”: Improving
effectiveness of training in youth soccer players. International
Journal of Sports Science & Coaching, 17479541231172702.
https://perceptionaction.com/connectionball/
Why I Switched Sides to the Ecological Approach
https://perceptionaction.com/383-2/
Kearney, P. E., Logan, O., Dunton, A., & Coughlan, E. K.
(2024). The adaptable coach–A critical review of the practical
implications for traditional and constraints–led approaches in
sports coaching: A commentary.
International Journal of Sports
Science & Coaching, 19(3), 1255-1261.

https://www.theconstraintscollective.com/bloghome/can-coaches-pic
k-amp-mix-ideas-from-different-scientific-theories-to-inform-their-pr
actical-applications
Lindsay, R., & Spittle, M. (2024). The adaptable coach–a
critical review of the practical implications for traditional and
constraints-led approaches in sport coaching. International
Journal of Sports Science & Coaching, 17479541241240853.
Clark, A. (2013). Whatever next? Predictive brains, situated
agents, and the future of cognitive science. Behavioral and brain
sciences, 36
(3), 181-204.
Hodges, N. J., & Lohse, K. R. (2022). An extended
challenge-based framework for practice design in sports
coaching. Journal of Sports Sciences, 40
(7), 754-768.
O’Brien, K. A., Kennedy, A., & O’Keeffe, M. J. (2023). The
Element of Surprise: How Predictive Processing Can Help
Coach Practitioners Understand and Develop Skilled Movement
in Sport Settings. International Sport Coaching Journal, 1(aop),
1-10.
Jacobs, D. M., & Michaels, C. F. (2007). Direct learning.
Ecological psychology,
19(4), 321-349.
Parr, T., & Friston, K. J. (2019). Generalised free energy and
active inference. Biological cybernetics, 113(5), 495-513.
Online vs Offline Information VS Direct vs Indirect Perception.
https://perceptionaction.com/382-2/
Harris, D. J., Arthur, T., Broadbent, D. P., Wilson, M. R.,
Vine, S. J., & Runswick, O. R. (2022). An active inference
account of skilled anticipation in sport: Using computational
models to formalise theory and generate new hypotheses. Sports
Medicine, 52
(9), 2023-2038.
Gray, R. (2021). Approaches to visual-motor control in baseball
batting. Optometry and Vision Science
,
98
(7), 738-749.
Tonner, P. (2017). Wittgenstein on forms of life: a short
introduction. E-Logos Electronic Journal for Philosophy.

3. An Ecological Approach to Coaching – The Coach as a

Learner & Explorer

Wood, M. A., Mellalieu, S. D., Araújo, D., Woods, C. T., &

Davids, K. (2022). Learning to coach: An ecological dynamics
perspective. International Journal of Sports Science &
Coaching, 17479541221138680.
Newell, K. M. (1985). Coordination, control and skill. In
Advances in psychology
(Vol. 27, pp. 295-317). North-Holland.
Lave, J., & Wenger, E. (1991). Situated learning: Legitimate
peripheral participation
. Cambridge University Press.
Selimi, E., & Woods, C. T. (2024). Enskilment into the
coaching landscape: towards a situated approach to coach
education in Australian football. Sport, Education and Society,
1-14.
Woods, C. T., Rudd, J., Gray, R., & Davids, K. (2021).
Enskilment: An ecological-anthropological worldview of skill,
learning and education in sport. Sports Medicine-Open, 7(1),
1-9.

Having Your Cake and Eating It Too: Using Representative

Learning Design

Proteau, L., Marteniuk, R. G., & Lévesque, L. (1992). A

sensorimotor basis for motor learning: Evidence indicating
specificity of practice. The Quarterly Journal of Experimental
Psychology, 44
(3), 557-575.
Proteau, L. (1992). On the specificity of learning and the role
of visual information for movement control. Advances in
psychology
, 85, 67-103.
Gray, R. (2022). Learning to Optimize Movement: Harnessing
the Power of the Athlete-Environment Relationship. PACE
Publishing.
Thorndike, E. L. (1931). The evolution of learning in recent
times: Future possibilities.
Logan, G. D. (1988). Toward an instance theory of
automatization. Psychological review, 95
(4), 492.
Lawrence, G. P., Cassell, V. E., Beattie, S., Woodman, T., Khan,
M. A., Hardy, L., & Gottwald, V. M. (2014). Practice with
anxiety improves performance, but only when anxious: evidence
for the specificity of practice hypothesis.
Psychological research,
78
, 634-650.
https://www.healthline.com/health/principle-of-specificity
Schmidt, R. A., Lee, T. D., Winstein, C., Wulf, G., & Zelaznik, H.
N. (2018). Motor control and learning: A behavioral emphasis.
Human kinetics.
https://www.verywellfit.com/principle-of-specificity-definition-312037
5
Hellebrandt, F. A. (1958). Application Of The Overload
Principle To Muscle Training In Man1.
American Journal of
Physical Medicine & Rehabilitation, 37(5), 278-283.
Caldwell, J. M. E., Alexander, F. J., & Ahmad, C. S. (2019).
Weighted-ball velocity enhancement programs for baseball
pitchers: a systematic review. Orthopaedic Journal of Sports
Medicine, 7
(2), 2325967118825469.
Oudejans, R. R. (2012). Effects of visual control training on
the shooting performance of elite female basketball players.
International Journal of Sports Science & Coaching, 7(3),
469-480.
Dunton, A., O’Neill, C., & Coughlan, E. K. (2019). The impact
of a training intervention with spatial occlusion goggles on
controlling and passing a football. Science and Medicine in
Football, 3
(4), 281-286.
Bennett, K. J., & Fransen, J. (2023). Distinguishing skill from
technique in football. Science and Medicine in Football, 1-4.
Caldeira, P., Paulo, A., Veloso, A., Infante, J., Davids, K., &
Araújo, D. (2024). How functional movement variability
facilitates successful skill adaptation during the volleyball attack.
International Journal of Sports Science & Coaching,
19(2),
668-676.
Mercado-Palomino, E., Richards, J., Molina-Molina, A., Benítez,
J. M., & Espa, A. U. (2020). Can kinematic and kinetic
differences between planned and unplanned volleyball block
jump-landings be associated with injury risk factors?. Gait &
posture, 79
, 71-79.
Buszard, T., Maloney, M., Krause, L., & Oppici, L. (2020).
One day at a time: Steps to optimising the daily training
environment for talent development. In
Talent Identification and
Development in Sport (pp. 66-80). Routledge.
Krause, L., Farrow, D., Reid, M., Buszard, T., & Pinder, R.
(2018). Helping coaches apply the principles of representative
learning design: validation of a tennis specific practice
assessment tool.
Journal of sports sciences,
36(11), 1277-1286.
Myszka, S., Yearby, T., & Davids, K. (2023). (Re)
conceptualizing movement behavior in sport as a
problem-solving activity. Frontiers in sports and active living, 5,
1130131.

https://www.beyondtheboxscore.com/2013/6/24/4456142/when-is
-an-error-more-likely-to-occur-in-a-game

https://www.thedynastydugout.com/p/sprint-speed-and-the-impact-
on-hitting
Myszka, S., Yearby, T., & Davids, K. (2023). Being Water:
how key ideas from the practice of Bruce Lee align with
contemporary theorizing in movement skill acquisition. Sport,
Education and Society, 1-17.
Krause, L., Farrow, D., Pinder, R., Buszard, T., Kovalchik, S.,
& Reid, M. (2019). Enhancing skill transfer in tennis using
representative learning design. Journal of Sports Sciences, 37
(22), 2560-2568.
Barris, S., Davids, K., & Farrow, D. (2013). Representative
learning design in springboard diving: Is dry-land training
representative of a pool dive?. European Journal of Sport
Science
, 13
(6), 638-645.

5. Designing Constraints & Using the CLA

1.
Smith, K., Burns, C., O’Neill, C., Duggan, J. D., Winkelman,
N., Wilkie, M., & Coughlan, E. K. (2023). How to coach: A
review of theoretical approaches for the development of a novel
coach education framework. International Journal of Sports
Science & Coaching, 18(2), 594-608.
2.
Lindsay, R., & Spittle, M. (2024). The adaptable coach–a
critical review of the practical implications for traditional and
constraints-led approaches in sport coaching. International
Journal of Sports Science & Coaching, 17479541241240853.
3.
Newell KM. Constraints on the development of coordination.
In: Wade MG, Whiting HTA, editors. Motor development in
children: aspects of coordination and control. Amsterdam:
Martin Nijhoff; 1986. p. 341–61.
4.
Kelso, J. S. (2007). Understanding complex systems.
Springer Complexity.
5.
https://www.drivelinebaseball.com/2022/04/st-louis-cardinals-how-l
ars-nootbaar-added-8-mph-of-bat-speed/
6.
Bosch, F. (2020). Anatomy of Agility: Movement Analysis in
Sport.
7. Newell, K. M. (2003). Change in motor learning: a
coordination and control perspective. Motriz, 9
(1), 1-6.
8.
Duivenvoorden, J., Van Der Kamp, J., Van Hilvoorde, I., &
Savelsbergh, G. (2021). Self-regulated learning strategies during
self-controlled practice in physical education.
Journal of Physical
Education and Sport,
21(2), 894-903.
9.
Gray, R. (2020). Comparing the constraints-led approach,
differential learning and prescriptive instruction for training
opposite-field hitting in baseball. Psychology of Sport & Exercise,
51, I01797.
10. Giblin, G., Farrow, D., Reid, M., Ball, K., & Abernethy, B.
(2015). Exploring the kinaesthetic sensitivity of skilled performers
for implementing movement instructions. Human Movement
Science
, 41, 76-91.
11. Gray, R. (In press). The Value of Practicing the “Wrong”
Way: Skill Development & Affordance Perception Through
Broad Exploration of the Solution Space.
12. Araújo, D., & Davids, K. (2011). What exactly is acquired
during skill acquisition?. Journal of Consciousness Studies,
18
(3-4), 7-23.
13. Myszka, S., Yearby, T., & Davids, K. (2023). (Re)
conceptualizing movement behavior in sport as a
problem-solving activity. Frontiers in sports and active living, 5,
1130131.
14. Verhoeff, W. J., Millar, S. K., Oldham, A. R., & Cronin, J.
(2020). Coaching the power clean: A constraints-led approach.
Strength & Conditioning Journal, 42(2), 16-25.
15. Brocken, J. E. A., Van Der Kamp, J., Lenoir, M., &
Savelsbergh, G. J. P. (2020). Equipment modification can
enhance skill learning in young field hockey players. International
Journal of Sports Science & Coaching, 15
(3), 382-389.
16. Santos, S., Coutinho, D., Gonçalves, B., Abade, E.,
Pasquarelli, B., & Sampaio, J. (2020). Effects of manipulating
ball type on youth footballers’ performance during small-sided
games. International Journal of Sports Science & Coaching,
15
(2), 170-183.
17. Di Domenico, F., D’Isanto, T., Esposito, G., Aliberti, S., &
Raiola, G. (2023). Exploring the Influence of Cognitive and
Ecological Dynamics Approaches on Countermovement Jumping
Enhancement: A Comparative Training Study.
Journal of
Functional Morphology and Kinesiology, 8
(3), 133.

Making a Skill Easier for a New Learner

Sweller, J. (1988). Cognitive load during problem solving: Effects

on learning. Cognitive science, 12(2), 257-285.
Atkinson, R.C. and Shiffrin, R.M. (1968). 'Human memory: A
Proposed System and its Control Processes'. In Spence, K.W.
and Spence, J.T. The psychology of learning and motivation,
(Volume 2). New York: Academic Press. pp. 89–195.
Miller, G. A. (1994). The magical number seven, plus or minus
two: Some limits on our capacity for processing information.
Psychological Review, 101(2), 343–352.
Allard, F., Graham, S., & Paarsalu, M. E. (1980). Perception in
sport: Basketball. Journal of sport and exercise psychology, 2
(1), 14-21.
Singer, R. N., Cauraugh, J. H., Chen, D., Steinberg, G. M., &
Frehlich, S. G. (1996). Visual search, anticipation, and reactive
comparisons between highly-skilled and beginning tennis players.
Journal of Applied Sport Psychology,
8
(1), 9-26.
https://www.rugbycoachweekly.net/rugby-drills-and-skills/tackling/s
caffolding-approach-to-tackling#
Giblin, G., Farrow, D., Reid, M., Ball, K., & Abernethy, B.
(2015). Exploring the kinaesthetic sensitivity of skilled performers
for implementing movement instructions. Human Movement
Science
, 41, 76-91.
Dicks, M., Button, C., & Davids, K. (2010). Examination of gaze
behaviors under in situ and video simulation task constraints
reveals differences in information pickup for perception and
action. Attention, Perception, & Psychophysics
,
72
, 706-720.
McCue, M., & Wattie, N. (2021). To Tee, Or Not to Tee:
Exploring the Relationship between Representative Practice and
Performance in Highly Skilled Baseball Hitters. Journal of
Expertise/March
,4
(1).
Tidman, S. J., Lay, B., Byrne, S., Bourke, P., & Alderson, J.
(2020). Reducing anterior cruciate ligament injury risk factors
by training perception: How vital is maintaining the
perception-action coupling?. ISBS Proceedings Archive, 38(1),
440.
Huesmann, K., Loffing, F., Büsch, D., Schorer, J., &
Hagemann, N. (2022). Varying degrees of perception-action
coupling and anticipation in handball goalkeeping. Journal of
motor behavior
, 54
(4), 391-400.
Travassos, B., Araujo, D., Davids, K., O'hara, K., Leitão, J., &
Cortinhas, A. (2013). Expertise effects on decision-making in
sport are constrained by requisite response behaviours–A
meta-analysis. Psychology of sport and exercise, 14(2), 211-219.
Wickens, C. D., Hutchins, S., Carolan, T., & Cumming, J.
(2013). Effectiveness of part-task training and increasing-difficulty
training strategies: a meta-analysis approach. Human Factors
, 55
(2), 461-470.
Davids, K., Kingsbury, D., Bennett, S., & Handford, C. (2001).
Information-movement coupling: Implications for the organization
of research and practice during acquisition of self-paced
extrinsic timing skills. Journal of sports sciences,
19(2), 117-127.
Reid, M., Whiteside, D., & Elliott, B. (2010). Effect of skill
decomposition on racket and ball kinematics of the elite junior
tennis serve. Sports Biomechanics, 9
(4), 296-303.
Kuo SY, Cheng KB, Chiu HT, Huang YC, et al. Differences in
Baseball Batting Movement Strategies Between Facing a Pitcher
and Pitching Machine. XXIV CongressInter Soc of Biomech
2013;15-16
Cork, A., Justham, L., & West, A. A. (2010). Batter's
behaviour during training when facing a bowling machine and
when facing a bowler.
Proceedings of the Institution of
Mechanical Engineers, Part P: Journal of Sports Engineering
and Technology, 224(3), 201-208.
Pinder, R. A., Renshaw, I., & Davids, K. (2009).
Information–movement coupling in developing cricketers under
changing ecological practice constraints. Human movement
science, 28(4), 468-479.
Gibson, A. P., & Adams, R. D. (1989). Batting stroke timing
with a bowler and a bowling machine: A case study. Australian
Journal of Science and Medicine in Sport,
21(2), 3-6.
Mercê, C., Davids, K., Catela, D., Branco, M., Correia, V., &
Cordovil, R. (2023). Learning to cycle: a constraint-led
intervention programme using different cycling task constraints.
Physical Education and Sport Pedagogy, 1-14.
Jacobs, D. M., & Michaels, C. F. (2007). Direct learning.
Ecological psychology,
19(4), 321-349.
Gray, R., & Regan, D. (1999). Do monocular time-to-collision
estimates necessarily involve perceived distance?. Perception, 28
(10), 1257-1264.
Latash, M. L., Bernstein, N. A., & Turvey, M. T. (2014).
Dexterity and its development. Psychology Press.
Moy, B., Renshaw, I., & Gorman, A. D. (2024). Applying the
constraints-led approach to facilitate exploratory learning of the
volleyball serve. Journal of Sports Sciences,
42(6), 511-518.
Buszard, T., Reid, M., Masters, R., & Farrow, D. (2016).
Scaling the equipment and play area in children’s sport to
improve motor skill acquisition: A systematic review. Sports
medicine, 46, 829-843.
Renshaw, I., Arnott, P., & McDowall, G. (2020). A
constraints-led approach to golf coaching . Routledge.
Caballero, C., Barbado, D., Peláez, M., & Moreno, F. J. (2024).
Applying different levels of practice variability for motor learning:
More is not better. PeerJ, 12, e17575.
Matsuo, T., Jinji, T., Hirayama, D., Nasu, D., Katsumata, Y., &
Morishita, Y. (2020). Consistent hand dynamics are achieved by
controlling variabilities among joint movements during fastball
pitching. Frontiers in Sports and Active Living,
2
, 579377.
Kusafuka, A., Kudo, K., & Nakazawa, K. (2022). Control of
accuracy during movements of high speed: implications from
baseball pitching. Journal of Motor Behavior, 54
(3), 304-315.
Deuker, A., Braunstein, B., Chow, J. Y., Fichtl, M., Kim, H.,
Körner, S., & Rein, R. (2024). “Train as you play”: Improving
effectiveness of training in youth soccer players. International
Journal of Sports Science & Coaching, 19(2), 677-686.
Esposito, G., Ceruso, R., Aliberti, S., & Raiola, G. (2024).
Ecological-Dynamic Approach vs. Traditional Prescriptive
Approach in Improving Technical Skills of Young Soccer Players.
Journal of Functional Morphology and Kinesiology, 9
(3), 162.
Le Moal, E., Rue, O., Ajmol, A., Abderrahman, A. B.,
Hammami, M. A., Ounis, O. B., ... & Zouhal, H. (2014).
Validation of the Loughborough Soccer Passing Test in young
soccer players. The Journal of Strength & Conditioning Research
, 28(5), 1418-1426.
LSPT https://youtu.be/yEkGskRD-50?si=UoDWIuLUYK7YmmBE
Lindsay, R. S., Komar, J., Chow, J. Y., Larkin, P., & Spittle, M.
(2023). Is prescription of specific movement form necessary for
optimal skill development? A nonlinear pedagogy approach.
Research Quarterly for Exercise and Sport,
94(3), 793-801.

Coaching to the Individual Athlete & Working with Groups

Williams, A. M., & Hodges, N. J. (2023). Effective practice

and instruction: A skill acquisition framework for excellence.
Journal of Sports Sciences, 41
(9), 833-849.
Ericsson, A. & Pool. R (2017). Peak: Secrets from the New
Science of Expertise. HarperOne
Starkes, J. L., Deakin, J. M., Allard, F., Hodges, N. J., &
Hayes, A. (2014). Deliberate practice in sports: What is it
anyway?. In The road to excellence (pp. 81-106). Psychology
Press.
An, J., Lewthwaite, R., Lee, S., & Wulf, G. (2020). Choice of
practice-task order enhances golf skill learning. Psychology of
Sport and Exercise,
50, 101737.
Porter, C., Greenwood, D., Panchuk, D., & Pepping, G. J.
(2020). Learner-adapted practice promotes skill transfer in
unskilled adults learning the basketball set shot. European
Journal of Sport Science
, 20(1), 61-71.
https://www.drivelinebaseball.com/axe-bat/
Newell, K. M., Liu, Y. T., & Mayer-Kress, G. (2001). Time
scales in motor learning and development. Psychological review
, 108
(1), 57.
Adams, J. A. (1952). Warm-up decrement in performance on
the pursuit-rotor. The American journal of psychology
, 65(3),
404-414.
Nacson, J., & Schmidt, R. A. (1971). The activity-set
hypothesis for warm-up decrement. Journal of Motor
Behavior,
3
(1), 1-15.
Ste-Marie, D. M., Law, B., Rymal, A. M., Jenny, O., Hall, C.,
& McCullagh, P. (2012). Observation interventions for motor
skill learning and performance: an applied model for the use
of observation. International Review of Sport and Exercise
Psychology, 5
(2), 145-176.
Daou, M., Lohse, K. R., & Miller, M. W. (2016). Expecting
to teach enhances motor learning and information processing
during practice. Human Movement Science,
49, 336-345.
https://perceptionaction.com/79-2/

Educating a Coach’s Attention & Learning to Observe

Movement

Pinheiro, V. E., & Simon, H. A. (1992). An operational model

of motor skill diagnosis. Journal of Teaching in Physical
Education,
11(3), 288-302.
Schmidt, R. A. (1975). A schema theory of discrete motor
skill learning. Psychological review, 82(4), 225.
Fetisova, Y., Zois, J., Spittle, M., & Dawson, A. (2021).
Coaches’ internal model of the tennis serve technique:
Knowing or understanding?. International Journal of Sports
Science & Coaching, 16(3), 568-584.
Newell, K. M. (2003). Schema theory (1975): Retrospectives
and prospectives. Research quarterly for exercise and sport,
74
(4), 383-388.
Gibson, J. J. (1979). The Ecological Approach to Visual
Perception. Hillsdale, NJ: Lawrence Erlbaum Associates.
Giblin, G., Farrow, D., Reid, M., Ball, K., & Abernethy, B.
(2015). Exploring the kinaesthetic sensitivity of skilled
performers for implementing movement instructions. Human
Movement Science
, 41, 76-91.
Waters, A., Panchuk, D., Phillips, E., & Dawson, A. (2020).
Experiential knowledge affects the visual search behaviors of
sprint coaches and sport biomechanists. Frontiers in Sports
and Active Living, 2
, 95.
Sparrow, W. A., & Sherman, C. (2001). Visual expertise in
the perception of action. Exercise and sport sciences reviews
,
29(3), 124-128.
Leas, R. R., & Chi, M. T. (1993). Analyzing diagnostic
expertise of competitive swimming coaches. In
Advances in
psychology
(Vol. 102, pp. 75-94). North-Holland.
Temprado, J., Della-Grasta, M., Farrell, M., & Laurent, M.
(1997). A novice-expert comparison of (intra-limb)
coordination subserving the volleyball serve. Human movement
science, 16(5), 653-676.
Greg Souders Interview: https://perceptionaction.com/495-2/
Scarborough, D. M., Bassett, A. J., Mayer, L. W., &
Berkson, E. M. (2020). Kinematic sequence patterns in the
overhead baseball pitch. Sports biomechanics, 19(5), 569-586.

9. Perceiving the Affordances of Others

Rizzolatti, G. (2005). The mirror neuron system and its

function in humans. Anatomy and embryology,
210(5),
419-421.
Ramenzoni, V. C., Riley, M. A., Shockley, K., & Davis, T.
(2008). An information-based approach to action
understanding. Cognition, 106
(2), 1059-1070.
Weast, J. A., Walton, A., Chandler, B. C., Shockley, K., &
Riley, M. A. (2014). Essential kinematic information, athletic
experience, and affordance perception for others. Psychonomic
bulletin & review, 21, 823-829.
Runeson, S., & Frykholm, G. (1983). Kinematic specification of
dynamics as an informational basis for person-and-action
perception: expectation, gender recognition, and deceptive
intention. Journal of experimental psychology: general
,
112(4),
585.
Point light walker:
https://www.biomotionlab.ca/html5-bml-walker/
Thomas, B. J., Hawkins, M. M., & Nalepka, P. (2018).
Perceiver as polar planimeter: Direct perception of jumping,
reaching, and jump-reaching affordances for the self and
others. Psychological Research, 82, 665-674.
Ji, H., & Pan, J. S. (2019). Can I choose a throwable object
for you? Perceiving affordances for other individuals. Frontiers
in Psychology, 10, 2205.
Wagman, J. B., Stoffregen, T. A., Bai, J., & Schloesser, D. S.
(2018). Perceiving nested affordances for another person’s
actions. Quarterly Journal of Experimental Psychology,
71(3),
790-799.

Using Imagery, Demonstration & Instruction

Ladda, A. M., Lebon, F., & Lotze, M. (2021). Using motor

imagery practice for improving motor performance–a review.
Brain and cognition, 150
, 105705.
Nicklaus, J. (2007). Jack Nicklaus: My Story. Simon and
Schuster.
Lindsay, R. S., Chow, J. Y., Larkin, P., & Spittle, M. (2023).
Creating adaptable skills: a nonlinear pedagogy approach to
mental imagery. Journal of Sport Psychology in Action,
14(3),
170-180.
Smith, D., Wright, C. J., & Cantwell, C. (2008). Beating the
bunker: The effect of PETTLEP imagery on golf bunker shot
performance. Research quarterly for exercise and sport,
79(3),
385-391.
Gibson, J. J. (1979). The Ecological Approach to Visual
Perception. Hillsdale, NJ: Lawrence Erlbaum Associates.
Lindsay, R. S., Komar, J., Chow, J. Y., Larkin, P., & Spittle, M.
(2023). Is prescription of specific movement form necessary for
optimal skill development? A nonlinear pedagogy approach.
Research Quarterly for Exercise and Sport,
94(3), 793-801.
Multhaupt, G., & Beuth, J. (2018). The Use of Imagery in
Athletic Injury Rehabilitation. A Systematic Review. German
Journal of Sports Medicine/Deutsche Zeitschrift fur Sportmedizin
, 69
(3).
Holmes, P., & Calmels, C. (2008). A neuroscientific review of
imagery and observation use in sport. Journal of motor
behavior
, 40
(5), 433-445.
Scully, D. M., & Newell, K. M. (1985). Observational-learning
and the acquisition of motor-skills-toward a visual-perception
perspective. Journal of human movement studies, 11(4),
169-186.
Horn, R. R., Williams, A. M., Hayes, S. J., Hodges, N. J., &
Scott, M. A. (2007). Demonstration as a rate enhancer to
changes in coordination during early skill acquisition. Journal of
Sports Sciences,
25(5), 599-614.
Al-Abood, S. A., Bennett, S. J., Hernandez, F. M., Ashford, D.,
& Davids, K. (2002). Effect of verbal instructions and image
size on visual search strategies in basketball free throw shooting.
Journal of sports sciences, 20(3), 271-278.
Al-Abood, S. A., Davids, K., Bennett, S. J., Ashford, D., &
Marin, M. M. (2001). Effects of manipulating relative and
absolute motion information during observational learning of an
aiming task. Journal of Sports Sciences,
19(7), 507-520.
Jennings, C. T., Reaburn, P., & Rynne, S. B. (2013). The
effect of a self-modelling video intervention on motor skill
acquisition and retention of a novice track cyclist's standing
start performance. International Journal of Sports Science &
Coaching,
8
(3), 467-480.
Smith, K., Burns, C., O’Neill, C., Duggan, J. D., Winkelman, N.,
Wilkie, M., & Coughlan, E. K. (2023). How to coach: A review
of theoretical approaches for the development of a novel coach
education framework. International Journal of Sports Science &
Coaching,
18
(2), 594-608.
Newell, K. M., & Ranganathan, R. (2010). Instructions as
constraints in motor skill acquisition. In
Motor learning in
practice (pp. 37-52). Routledge.
Otte, F. W., Davids, K., Millar, S. K., & Klatt, S. (2020). When
and how to provide feedback and instructions to athletes?—How
sport psychology and pedagogy insights can improve coaching
interventions to enhance self-regulation in training. Frontiers in
psychology
, 11, 549431.
Lafe, C. W., & Newell, K. M. (2022). Instructions on task
constraints mediate perceptual-motor search and how movement
variability relates to performance outcome. Journal of Motor
Behavior,
54(6), 669-685.
Singh, H., Shih, H. T., Kal, E., Bennett, T., & Wulf, G. (2022).
A distal external focus of attention facilitates compensatory
coordination of body parts. Journal of Sports Sciences,
40(20),
2282-2291.
McNevin, N. H., Shea, C. H., & Wulf, G. (2003). Increasing
the distance of an external focus of attention enhances learning.
Psychological research
, 67
, 22-29.

What To Do Next - Adapting Practice

Kennedy, A., & O’Brien, K. A. (2024). Adding texture to the

Art of constraints-led coaching: a request for more
research-informed guidelines. Sports Coaching Review, 1-20.
Steph Curry, 2024 Olympics:
https://youtu.be/xf-zV1uTkW0?si=nL_rcvVOp3nFRgvy&t=217
Patrick Mahomes, behind the back pass:
https://www.youtube.com/watch?v=IHkvak_oToc
Guadagnoli, M. A., & Lee, T. D. (2004). Challenge point: a
framework for conceptualizing the effects of various practice
conditions in motor learning. Journal of motor behavior
, 36
(2), 212-224.
https://chinupgoggles.com/
Scott, S., & Gray, R. (2010). Switching tools: Perceptual-motor
recalibration to weight changes. Experimental brain research,
201
, 177-189.
Gray, R. (2018). Comparing cueing and constraints
interventions for increasing launch angle in baseball batting.
Sport, Exercise, and Performance Psychology, 7
(3), 318.
Stone, J. A., Rothwell, M., Shuttleworth, R., & Davids, K.
(2021). Exploring sports coaches’ experiences of using a
contemporary pedagogical approach to coaching: An
international perspective. Qualitative Research in Sport,
Exercise and Health, 13
(4), 639-657.
 Creating a Culture and Form of Life

Tonner, P. (2017). Wittgenstein on forms of life: a short

introduction. E-Logos Electronic Journal for Philosophy.
Uehara, L., Button, C., Saunders, J., Araújo, D., Falcous, M.,
& Davids, K. (2021). Malandragem and Ginga: Socio-cultural
constraints on the development of expertise and skills in
Brazilian football. International Journal of Sports Science &
Coaching,
16(3), 622-635.
Van Dijk, L., & Rietveld, E. (2017). Foregrounding
sociomaterial practice in our understanding of affordances:
The skilled intentionality framework. Frontiers in psychology,
7
, 1969.
Withagen, R., De Poel, H. J., Araújo, D., & Pepping, G. J.
(2012). Affordances can invite behavior: Reconsidering the
relationship between affordances and agency. New ideas in
psychology
, 30(2), 250-258.
Eves, F. F., Thorpe, S. K., Lewis, A., & Taylor-Covill, G. A.
(2014). Does perceived steepness deter stair climbing when
an alternative is available?. Psychonomic bulletin & review,
21(3), 637-644.
Rothwell, M., Davids, K., & Stone, J. (2018). Harnessing
socio-cultural constraints on athlete development to create a
form of life. Journal of Expertise, 1
(1).
O’Sullivan, M., Vaughan, J., Woods, C. T., & Davids, K.
(2024). There is no copy and paste, but there is resonation
and inhabitation: Integrating a contemporary player
development framework in football from a complexity sciences
perspective. Journal of sports sciences, 1-10.
https://coachdibernardo.com/2023/09/30/constraints-variability-
academic-theory-and-player-development-in-soccer/
Vaughan, J., Mallett, C. J., Potrac, P., Woods, C., O'Sullivan,
M., & Davids, K. (2022). Social and cultural constraints on
football player development in Stockholm: influencing skill,
learning, and wellbeing. Frontiers in Sports and Active Living,
4
, 832111.
Hodges, N. J., & Lohse, K. R. (2022). An extended
challenge-based framework for practice design in sports
coaching. Journal of Sports Sciences, 40
(7), 754-768.
Woods, C. T., Rothwell, M., Rudd, J., Robertson, S., &
Davids, K. (2021). Representative co-design: Utilising a source
of experiential knowledge for athlete development and
performance preparation. Psychology of Sport and Exercise,
52
, 101804.
 Inspiring Autonomy, Motivation & Confidence

https://www.psychologytoday.com/gb/blog/kidding-ourselves/20
1405/the-remarkable-power-of-hope
Deci, E. L., & Ryan, R. M. (2012). Self-determination theory.
Handbook of theories of social psychology,
1
(20), 416-436.
Renshaw, I., R Oldham, A., & Bawden, M. (2012). Nonlinear
pedagogy underpins intrinsic motivation in sports coaching.
Lashanlou, M. B., & Dehghanizade, J. (2024). Comparison of
Learning Golf Putting Skill in Different Choices: A Test of the
OPTIMAL Theory. Journal of Motor Control and Learning, 6
(1).
Wulf, G., Iwatsuki, T., Machin, B., Kellogg, J., Copeland, C., &
Lewthwaite, R. (2018). Lassoing skill through learner choice.
Journal of motor behavior
, 50(3), 285-292.
Arbinaga, F., Fernández-Ozcorta, E. J., Ch