 Shaping

Shaping is the process of systematically and differentially reinforcing successive

approximations to a terminal behavior. Shaping is used in many everyday situations to
help learners acquire new behaviors.

For example, language therapists use shaping when they develop speech with a client
by first reinforcing lip movements, then sound production, and finally word and
sentence expression. Teachers working with students with severe disabilities shape
social interactions when they differentially reinforce eye contact, one-word greetings,
and conversational speech. A basketball coach shapes the foul-shooting behavior
of her players when she differentially reinforces accurate shooting from positions a
few feet in front of the basket to positions closer to the 15-foot regulation foul
line. Further, trainers use shaping to teach desirable behavior to animals for function
(e.g., loading horses into a trailer without injury to the horse or the attendant), or
for appeal or utility duty (e.g., teaching porpoises to execute show routines).

Depending on the complexity of a given behavior and the learner’s prerequisite

skills, shaping may require many successive approximations before the terminal
behavior is achieved. Achievement of the terminal behavior with respect to time,
trials, or direction is seldom predictable, immediate, or linear. If the learner emits
a closer approximation to the terminal behavior, and the practitioner fails to detect and
reinforce it, achievement of the terminal behavior will be delayed. However, if
a systematic approach is used—that is, if each instance of closer approximations to
the terminal behavior is detected and reinforced—progress can usually be attained
more quickly. Although shaping can be time-consuming, it represents an important
approach to teaching new behaviors, especially those behaviors that cannot easily
be learned by instructions, incidental experience or exposure, imitation, physical cues,
or verbal prompts.

practitioner using behavioral shaping differentially reinforces successive

approximations toward a terminal behavior. The end product of shaping—a terminal
behavior—can be claimed when the topography, frequency, latency, duration, or
amplitude/ magnitude of the target behavior reaches a predetermined criterion level.
The two key procedural components of shaping, differential
reinforcement and successive approximations, are described next.

 Differential reinforcement

When a heavy ball is thrown beyond a certain mark, when a horizontal bar is cleared
in vaulting or in jumping, when a ball is batted over the fence (and when, as a result, a
record is broken or a match or game won), differential reinforcement is at work.

– B. F. Skinner

Differential reinforcement is a procedure in which reinforcement is provided for

responses that share a predetermined dimension or quality, and in which
reinforcementis withheld for responses that do not demonstrate that quality. For
example, differential reinforcement is used by a parent who complies with his child’s
requests for an item at the dinner table when those requests include polite words such
 as “please” or “may I,” and who does not pass the requested item following requests
in which polite words are missing. Differential reinforcement has two effects:
responses similar to those that have been reinforced occur with greater frequency, and
responses resembling the unreinforced members are emitted less frequently.

When differential reinforcement is applied consistently within a response class, its

dual effects result in a new response class composed primarily of responses sharing
the characteristics of the previously reinforced subclass. This emergence of a new
response class is called response differentiation. Response differentiation as a result of
the parent’s differential reinforcement at the dinner table would be evident if all of the
child’s requests contained polite words.

The preferred strategy for decelerating an undesirable behavior is to reinforce an

acceleration target behavior that is an alternative to the deceleration target behavior, a
procedure called differential reinforcement. For example, to reduce a client’s criticism
of people, the client might be reinforced for complimenting others. As this example
shows, differential reinforcement changes the deceleration target behavior
(criticizing) indirectly by directly reinforcing the alternative acceleration target
behavior (complimenting).

Differential reinforcement works because the more the client engages in the
alternative behavior, the less opportunity the client has to engage in the deceleration
target behavior. Consider the case of a young girl with severe mental retardation who
frequently hit herself.1 To reduce her self-destructive behavior, she was reinforced for
using her hands to play with a puzzle. Differential reinforcement was effective
because while her hands were engaged in playing with the puzzle, she could not hit
herself.

The four major types of differential reinforcement, in order of most to least effective,
are differential reinforcement of

(1) incompatible behaviors,

(2) competing behaviors,

(3) any other behaviors, and

(4) a low frequency of the undesirable behavior.

 Successive Approximations

The practitioner using shaping differentially reinforces responses that resemble the
terminal behavior in some way. The shaping process begins with reinforcement
of responses in the learner’s current repertoire that share an important topographical
feature with the terminal behavior or are a prerequisite behavior for the final
terminal behavior. When the initially reinforced responses become more frequent, the
practitioner shifts the criterion for reinforcement to responses that are a closer
approximation of the terminal behavior. The gradually changing criterion for
reinforcement during shaping results in a succession of new response classes, or
successive approximations, each one closer in form to the terminal behavior than the
response class it replaces. Skinner (1953) discussed the critical nature of
successive approximations as follows:

The original probability of the response in its final form is very low; in some cases it
may even be zero. In this way we build complicated operants that would never
appear in the repertoire of the organism otherwise. By reinforcing a series of
successive approximations, we bring a rare response to a very high probability in a
short time. This is an effective procedure because it recognizes and utilizes the
continuous nature of a complex act.

Shaping Different Dimensions of Performance

Behavior can be shaped in terms of topography, frequency, latency, duration, and

amplitude/magnitude (see Table 1). Differential reinforcement could also be used to
teach a child to talk within a conversational decibel range. Let us suppose a behavior
analyst is working with a student who usually talks at such a low volume (e.g., below
45 decibels) that his teacher and peers have difficulty hearing him. Successive
approximations to 65 decibels (dB)—the amplitude of normal conversational speech
—might be 45, 55, and ultimately 65 dB. Differential reinforcement of speaking at a
minimum volume of 45 dB would place responses below that amplitude on extinction.
When the student is speaking consistently at or above 45 dB, the criterion would be
raised to 55 db. Likewise, when 55 dB and finally 65 dB are achieved, previous lower
amplitude levels are not reinforced (i.e., they are placed on extinction).

Table 1 Examples of Performance Improvements that Could Be Shaped in Various

Dimensions of Behavior

Topography (form of the behavior)

o Refining motor movements associated with a golf swing, throwing motion, or

vaulting behavior. Improving cursive or manuscript letter formation during
handwriting exercises.

Frequency (number of responses per unit of time)

o Increasing the number of problems completed during each minute of a math

seatwork assignment.
o Increasing the number of correctly spelled and appropriately used words
written per minute.

Latency (time between the onset of the antecedent stimulus and the occurrence of the

behavior)

o Decreasing compliance time between a parental directive to “clean your room”

and the onset of room-cleaning behavior.

• Increasing the delay between the onset of an aggressive remark and retaliation by a
student with

severe emotional disabilities.

 Duration (total elapsed time for occurrence of the behavior)

o Increasing the length of time that a student stays on task.

o Increasing the number of minutes of engaged study behavior.

Amplitude/Magnitude (response strength or force)

o Increasing the projected voice volume of a speaker from 45 dB to 65 dB.

• Increasing the height of the high jump bar for students enrolled in a physical
education class.

Prompting (Response Prompts, Stimulus Prompts)

 Stimulus Control

Reinforcement of an operant response increases the frequency of future responding

and influences the stimuli that immediately precede the response. The stimuli
preceding the response (i.e., antecedent stimuli) acquire an evocative effect on the
relevant behavior. In a typical laboratory demonstration of operant conditioning, a rat
is placed in an experimental chamber and given the opportunity to press a lever.
Contingent on a lever press the rat receives a food pellet. Reinforcement of the lever
press increases the frequency of lever pressing.

Researchers can make this simple demonstration more complex by manipulating other

variables. For example, occasionally a buzzer might sound, and the rat receives a
food pellet only when the lever is pressed in the presence of the buzzer. The buzzer
sound preceding the lever press is called a discriminative stimulus (SD, pronounced
“ess-dee”). With some experience, the rat will make more lever presses in the
presence of the buzzer sound (SD) than in its absence, a condition called stimulus
delta (SΔ, pronounced “essdelta”). Behavior that occurs more often in the presence
of an SD than in its absence is under stimulus control. Technically, stimulus control
occurs when the rate, latency, duration, or amplitude of a response is altered in the
presenceof an antecedent stimulus (Dinsmoor, 1995a, b). A stimulus acquires control
only when responses emitted in the presence of that stimulus produce reinforcement
more often than responses in the absence of stimulus.

Stimulus control should not be viewed as just an interesting procedure for laboratory

demonstrations. Stimulus control plays a fundamental role in everyday complex
behaviors (e.g., language systems, conceptual behavior, problem solving), education,
and treatment (Shahan & Chase, 2002; Stromer, 2000). People do not answer the
telephone in the absence of a ring. A person driving a car stops the car more often in
the presence of a red traffic light than in its absence. People who use both Spanish and
English languages will likely use Spanish, not English, to communicate with a
Spanish-speaking audience.

Using Prompts to Develop Stimulus Control

Prompts are supplementary antecedent stimuli used to occasion a correct response in

the presence of an SD that will eventually control the behavior. Applied
 behavior analysts give response and stimulus prompts before or during the
performance of a behavior. Response prompts operate directly on the response.
Stimulus prompts operate directly on the antecedent task stimuli to cue a
correct response in conjunction with the critical SD.

 Response Prompts

The three major forms of response prompts are verbal instructions, modeling, and

physical guidance.

Verbal Instructions

Applied behavior analysts use functionally appropriate verbal instructions as

supplementary response prompts. Verbal response prompts occur frequently in almost
all training contexts in the forms of vocal verbal instruction (e.g., oral, telling) and
nonvocal verbal instruction (e.g., written words, manual signs, pictures).

Teachers often use vocal verbal instructional prompts. Suppose a teacher asked a

student to read the sentence, “Plants need soil, air, and water to grow.” The student
reads, “Plants need . . . Plants need . . . Plants need . . .” The teacher could use any
number of verbal prompts to occasion the next word. She might say, “The next word
is soil. Point to soil and say soil.” Or she might use a rhyming word for soil. For
another example, Adkins and Mathews (1997) taught in-home caregivers to use vocal
verbal response prompts to improve voiding procedures for two adults with urinary
incontinence and cognitive impairments. The in-home caregiver checked for dryness
each hour or every 2 hours between 6:00 A.M. and 9:00 P.M. The caregiver praised
dryness, asked the adult to use the toilet, and provided assistance as needed when the
adult was dry at the regularly scheduled check. This simple response prompt
procedure, which was introduced following a baseline condition, produced in oneof
the adults a mean 22% reduction of grams of urine collected per day in wet diapers
during the 2-hour prompted voiding condition, and a mean 69% reduction during the
1-hour prompt condition. The second adult received only the 1-hour prompted voiding
condition, which resulted in a mean 55% reduction of urine collected per day.

Modeling

Applied behavior analysts can demonstrate or model the desired behavior as a

response prompt. Modeling can effectively prompt behaviors especially for learners
who have already learned some of the component behaviors required for the imitation.
Modeling is an easy, practical, and successful way for a coach to show a player an
appropriate form for shooting a basketball through a hoop when the player already
can hold the ball, raise it over his head, and push the ball away from his body.
Few teachers would use modeling to teach a child with severe disabilities to tie her
shoes if she could not hold the laces in her hands. In addition, attending skills are
important. The learner must observe the model to enable the imitation of the
performance. Finally, modeling as a responseprompt should be used only with
students who have already developed imitative skills. The use of models to assist in
the development of appropriate academic and social behavior has been demonstrated
repeatedly.
 Physical Guidance

Physical guidance is a response prompt applied mostoften with young children,

learners with severe disabilities, and older adults experiencing physical
limitations. Using physical guidance, the teacher partially physically guides the
student’s movements, or physically guides the student throughout the entire
movement of the response.

 Stimulus Prompts

Applied behavior analysts have frequently used movement, position, and redundancy

of antecedent stimuli as stimulus prompts. For example, movement cues can help
a learner discriminate between a penny and a dime by pointing to, tapping, touching,
or looking at the coin to be identified. In the coin discrimination task, the teacher
could use a position cue and place the correct coin closer to the student. Redundancy
cues occur when one or more stimulus or response dimensions (e.g., color, size,
shape) are paired with the correct choice. For instance, a teacher might use a color
mediation procedure of associating a numeral with a color, then link the name of a
color to an answer for an arithmetic fact (Van Houten & Rolider, 1990).

 Transfer of Stimulus Control

Applied behavior analysts should provide response andstimulus prompts as

supplementary antecedent stimuli only during the acquisition phase of instruction.
With the reliable occurrence of behavior, applied behavior analysts need to transfer
stimulus control from the response and stimulus prompts to the naturally existing
stimulus. Applied behavior analysts transfer stimulus control by gradually fading
stimuli in or out, gradually presenting or removing antecedent stimuli. Eventually, the
natural stimulus, a partially changed stimulus, or a new stimulus will evoke the
response. Fading response prompts and stimulus prompts is the procedure used to
transfer stimulus control from the prompts to the natural stimulus, and also to
minimize the number of error responses occurring in the presence of the natural
stimulus.

Transferring Stimulus Control from Response Prompts to Naturally Existing Stimuli

Wolery and Gast (1984) described four procedures for transferring stimulus control from
response prompts to natural stimuli. They describe these procedures as mostto- least prompts,
graduated guidance, least-to-most prompts, and time delay.

Most-to-Least Prompts

The applied behavior analyst can use most-to-least response prompts to transfer stimulus
control from response prompts to the natural stimulus whenever the participant does not
respond to the natural stimulus or makes an incorrect response. To apply most-to-least
response prompts, the analyst physically guides the participant through the entire
performance sequence, then gradually reduces the amount of physical assistance provided
as training progresses from trial to trial and session to session. Customarily, most-to-least
prompting moves from physical guidance to visual prompts to verbal instructions, and finally
to the natural stimulus without prompts.
Graduated Guidance

The applied behavior analyst provides physical guidance as needed, but using graduated
guidance she will immediately start to fade out the physical prompts to transfer stimulus
control. Graduated guidance begins with the applied behavior analyst following the
participant’s movements closely with her hands, but not touching the participant. The analyst
then increases the distance of her hands from the participant by gradually changing the
location of the physical prompt. For example, if the applied behavior analyst used physical
guidance for a participant’s hand movement in zippering a coat, she might move the prompt
from the hand to the wrist, to the elbow, to the shoulder, and then to no physical contact.
Graduated guidance provides the opportunity for an immediate physical prompt as needed.

Least-to-Most Prompts

When transferring stimulus control from response prompts using least-to-most prompts, the
applied behavior analyst gives the participant an opportunity to performthe response with the
least amount of assistance on each trial. The participant receives greater degrees of
assistance with each successive trial without a correct response. The procedure for least-to-
most prompting requires the participant to make a correct response within a set time limit
(e.g., 3 seconds) from the presentation of the natural SD. If the response does not occur
within the specified time, the applied behavior analyst will again present the natural SD and a
response prompt of least assistance, such as a verbal response prompt. If after the same
specified time limit (e.g., another 3 seconds), the participant does not make a correct
response, the analyst gives the natural SD and another response prompt, such as a gesture.
The participant receives partial or full physical guidance if the lesser prompting does not
evoke a correct response. Applied behavior analysts using the least-to-most response prompt
procedure present the natural SD and the same time limit during each training trial. For
example, Heckaman, Alber, Hooper, and Heward (1998) used instructions, nonspecific verbal
prompts, modeling, and physical prompting in a least-to-most 5-second response prompt
hierarchy to improve the disruptive behavior of four students with autism.

Time Delay

To produce a transfer of stimulus control, most-to-leastprompts, graduated guidance, and

least-to-most promptsoccur as consequences to gradual changes in the form, position, or
intensity of a response evoked by the natural stimulus. Conversely, as an antecedent response
prompt, the time delay procedures use only variations in the time intervals between
presentation of the natural stimulus and the presentation of the response prompt. Constant
time delay and progressive time delay transfer stimulus control from a prompt to the natural
stimulus by delaying the presentation of the prompt following the presentation of the natural
stimulus.

The constant time delay procedure first presents several trials using a 0-second delay—that

is, the simultaneous presentation of the natural stimulus and the response prompt. Usually,
but not always (Schuster, Griffen, & Wolery, 1992), the trials that follow the
simultaneous prompt condition apply a fixed time delay (e.g., 3 seconds) between the
presentation of the natural stimulus and the presentation of the response prompt
(Caldwell, Wolery, Werts, & Caldwell, 1996).
The progressive time delay procedure, like constant time delay , starts with a 0-second
delay between the presentation of the natural stimulus and the response prompt. Usually, a
teacher will use several 0-second trials before extending the time delay. The number of
0- second trials will depend on the task difficulty and the functioning level of the participant.
Following the simultaneous presentations, the teacher will gradually and systematically
extend the time delay, often in 1-second intervals. The time delay can be extended after a
specific number of presentations, after each session, after a specific number of sessions, or
after meeting a performance criterion.

 Transfer of Stimulus Control Using Stimulus Control Shaping

The preceding sections focused on response prompts that do not change the task
stimuli or materials. The stimulus control shaping procedures presented here modify
the task stimuli or materials gradually and systematically to prompt a response.
Supplementary stimulus conditions are faded in or faded out to transfer stimulus
control from the stimulus prompt to the natural stimulus. Stimulus control shaping can
be accomplished with stimulus fading and stimulus shape transformations.

Stimulus Fading

Stimulus fading involves highlighting a physical dimension (e.g., color, size, position)

of a stimulus to increase the likelihood of a correct response. The highlighted
or exaggerated dimension is faded gradually in or out.

Krantz and McClannahan (1998) faded out scripts (i.e., the words Look and Watch
me) embedded in photographic activity schedules. The embedded scripts prompted
the social exchanges of children with autism. The words Look and Watch me were
printed with 72- point font and bold letters on white 9-cm note cards. Krantz and
McClannahan began fading out the words by removing one third of the script card,
then another third. Sometimes, during the script fading, portions of letters were still
shown on the card such as a part of an o in Look. Finally, the scripts and cards were
removed.

Stimulus Shape Transformations

The procedure for stimulus shape transformations uses an initial stimulus shape that
will prompt a correct response. That initial shape is then gradually changed to form
the natural stimulus, while maintaining correct responding. For example, a program to
transform the shape of stimuli to teach number recognition could include
the following steps:

The shape of the stimulus prompt must change gradually so that the student continues
to respond correctly. In teaching word identification, using stimulus
shape transformations could include the following steps: Figure 9 (page number 422)