Reasoning
Introduction to Cognitive Science, COMP 20090
LOGIC, n. The art of thinking and reasoning in strict accordance with
the limitations and incapacities of the human misunderstanding. The
basic [unit] of logic is the syllogism, consisting of a major and a minor
premise and a conclusion -- thus:
_Major Premise_: Sixty men can do a piece of work sixty times as
quickly as one man.
_Minor Premise_: One man can dig a posthole in sixty seconds;
therefore --
_Conclusion_: Sixty men can dig a posthole in one second.
This may be called the syllogism arithmetical, in which, by
combining logic and mathematics, we obtain a double certainty and are
twice blessed.
Ambrose Bierce, The Devil’s Dictionary
Introduction to Cognitive Science, COMP 20090
Are you a creature of reason?
Introduction to Cognitive Science, COMP 20090
Reason has traditionally been claimed as distinctly human, and
not to be found elsewhere in the animal world. However, recent
studies in this area show that, in lower levels, animals are
capable of some rational thinking.
Source: wikipedia
Introduction to Cognitive Science, COMP 20090
Reasoning does not help us in identifying goals or in
identifying problems.
Rather, it helps us derive new knowledge from that
which we already know
{
Premise: All men are mortal
Premise: Socrates is a man
Syllogism
Conclusion: Socrates is mortal
Introduction to Cognitive Science, COMP 20090
Inductive reasoning
The sun has risen in the east every morning
up until now
premise
Therefore the sun will rise in the east
tomorrow.
conclusion
Introduction to Cognitive Science, COMP 20090
Inductive reasoning
I have been fed every day up until
today (23 Dec)
premise
Therefore I will be fed tomorrow
(24 Dec)
conclusion
Introduction to Cognitive Science, COMP 20090
Inductive reasoning
That which is probably the case
Introduction to Cognitive Science, COMP 20090
Deductive Reasoning
‘Hard logic’
Conclusion follows from the premises
c.f. the Socrates syllogism...
Introduction to Cognitive Science, COMP 20090
Do the following enterprises use inductive or deductive
reasoning more?
1) Scientific Inquiry
2) Mathematics
Introduction to Cognitive Science, COMP 20090
Inductive reasoning
If the sky is getting very dark and
cloudy, it is probably going to rain.
Introduction to Cognitive Science, COMP 20090
Inductive reasoning
If I don’t eat something,
I will get hungry.
Introduction to Cognitive Science, COMP 20090
Inductive reasoning
If I turn on the light switch,
the light will come on.
Introduction to Cognitive Science, COMP 20090
If I turn on the light switch,
the light will come on.
Hypothesis
Confirmed?
Falsified?
Supported?
Evidence
Introduction to Cognitive Science, COMP 20090
Confirmation Bias
Many of us, a lot of the time, will tend to look for
evidence that supports our hypothesis (only).
Hypothesis
Confirmed? Falsified?
Supported?
Evidence
Introduction to Cognitive Science, COMP 20090
This is a widespread tendency and a problem
Introduction to Cognitive Science, COMP 20090
Confirmation Bias
Easily leads to stereotyping
Introduction to Cognitive Science, COMP 20090
Recall Skinner’s superstitious pigeons.
Could this be interpreted as a form of
‘Confirmation Bias’?
Introduction to Cognitive Science, COMP 20090
The Wason 2-4-6 task
2 4 6
These numbers conform to a general rule.
Figure out what it is. Generate other number
triples you think conform to the rule, and I will
confirm or disconfirm.
Introduction to Cognitive Science, COMP 20090
Did you look for confirmation of your
hypothesis, or did you look to falsify it?
Hypothesis: successive even numbers
Confirming evidence: 468
Falsifying evidence: 4 6 9
Introduction to Cognitive Science, COMP 20090
The 2-4-6 Task
• Wason (1960) devised a simple experimental paradigm with which to
test people’s strategies in hypothesis testing.
• In the task, participants were told that the three numbers “2-4-6”
confirm to a general rule that holds for a subset of all such number
triples.
• The participants’ task was to figure out what the general rule was.
• They did this by guessing other number triples (e.g. 4-6-8; 10-10-10,
6-4-2) and asking the experimenter whether or not that number triple
also confirmed to the rule.
• They also had to provide a reason for their guess (e.g. “I think the
rule might be numbers ascending in twos”)
• After each guess, the experimenter told the participant whether or not
the guess conformed to the rule.
• The rule was simply “three numbers in ascending order”
Introduction to Cognitive Science, COMP 20090
• Wason found that most participants took a long time to find the
correct rule. 28% of participants could not determine the rule at all.
• Participants tended to generate hypotheses that were more
restrictive than the actual rule (e.g. even numbers ascending in twos)
• Participants tended to put number triples to the experimenter that
conformed to their hypothesis (e.g. 4-6-8, 6-8-10,10-12-14). These
triples also conformed to the actual rule.
• The participants therefore gathered evidence that supported their
hypothesis about the rule (i.e. they showed confirmation bias).
• The only way for such participants to determine that the actual rule
was more general that their hypothesis was to generate number
triples that we false for their rule (e.g. 1-3-5, 1-2-3, etc).
• According to Wason, therefore, participants difficulty with the 2-4-6
task stemmed from their confirmation bias and their failure to falsify.
Note: it has been heavily debated whether this really stems from
‘confirmation bias’ or from peculiarities of this specific task
(maximally general rule)
Introduction to Cognitive Science, COMP 20090
Deductive reasoning
Introduction to Cognitive Science, COMP 20090
Deductive reasoning
• There is a much greater amount of research
conducted on deductive reasoning than on
inductive reasoning.
• This is primarily for reasons of practicality.
It takes many trials in order to make a
single induction. Also, inductive reasoning
is not precise or definite like deductive
reasoning is.
• (e.g. even in the 2-4-6 task you can never
be sure you know the rule precisely).
Introduction to Cognitive Science, COMP 20090
Deductive reasoning
• Conditional reasoning is
reasoning about propositions
using the logical relation known as
implication.
• Formally, it is based on
propositional logic
• ‘if….then’, ‘if…& only if’ (also
not, and, and or are used in
propositional logic)
Introduction to Cognitive Science, COMP 20090
A proposition P is either True or False
Given several propositions, we may be
able to validly infer other propositions.
Introduction to Cognitive Science, COMP 20090
Modus Ponens
Given these two propositions:
if p then q
p
We can conclude
q
if p then q if Socrates is a man then he is mortal
p Socrates is a man
therefore q therefore, Socrates is mortal
Introduction to Cognitive Science, COMP 20090
Modus Tollens
Given these two propositions:
if p then q
not q
We can conclude
not p
if p then q if there is fire then there is oxygen
not q there is not oxygen
therefore not p therefore, there is not fire
Introduction to Cognitive Science, COMP 20090
Modus Ponens Modus Tollens
Both are valid!
People tend to be better at using MP than MT
Introduction to Cognitive Science, COMP 20090
So are people good at reasoning?
Introduction to Cognitive Science, COMP 20090
Is this valid?
If Susan is angry, then I am upset
I am upset
Therefore, Susan is angry
Introduction to Cognitive Science, COMP 20090
Logical Fallacy
Invalid Inference
Introduction to Cognitive Science, COMP 20090
Fallacy 1: Affirmation of the Consequent
If p then q
q
therefore: p
Introduction to Cognitive Science, COMP 20090
Fallacy 1: Affirmation of the Consequent
If p then q
q
therefore: p
If I do well on the test, I will get drunk at the
weekend
I got drunk at the weekend
Therefore: I must have done well on the test (I can’t
remember)
Introduction to Cognitive Science, COMP 20090
Fallacy 1: Affirmation of the Consequent
If p then q
q
FALLACY!
therefore: p
If I do well on the test, I will get drunk at the
weekend
I got drunk at the weekend
Therefore: I must have done well on the test (I can’t
remember)
Introduction to Cognitive Science, COMP 20090
Fallacy 1: Affirmation of the Consequent
Note that I may or may not have done well on the test.
The argument is flawed. That’s all.
Introduction to Cognitive Science, COMP 20090
Fallacy 1: Affirmation of the Consequent
If I do well on the test, I will get drunk at the
weekend
I got drunk at the weekend
Therefore: I must have done well on the test (I can’t
remember)
Note that I may or may not have done well on the test.
The argument is flawed. That’s all.
Introduction to Cognitive Science, COMP 20090
Fallacy 2: Denying the Antecedent
If p then q
not p
Therefore not q
Introduction to Cognitive Science, COMP 20090
Fallacy 2: Denying the Antecedent
If p then q
not p
Therefore not q
If I do well in the test, I will get drunk
I failed the test
Therefore: I will not get drunk
Introduction to Cognitive Science, COMP 20090
Fallacy 2: Denying the Antecedent
If p then q
not p
FALLACY!
Therefore not q
If I do well in the test, I will get drunk
I failed the test
Therefore: I will not get drunk
Introduction to Cognitive Science, COMP 20090
antecedent consequent
if A then B
Affirming the antecedent (correct): A, therefore B
Denying the consequent (correct): not B, therefore not A
Affirming the consequent (wrong): B, therefore A
Denying the antecedent (wrong): not A, therefore not B
Even if A and B are both true, the first two ways of
reasoning are correct, and the second two are incorrect.
Introduction to Cognitive Science, COMP 20090
So are people good at reasoning?
• Marcus and Rips (1979): participants were presented
with examples of arguments involving modus ponens,
modus tollens, affirmation of the consequent and
denial of the antecedent.
• Participants were asked to judge each argument as
either valid or invalid.
• Participants often incorrectly judged modus tollens
invalid and incorrectly judged the two logical fallacies
valid.
Introduction to Cognitive Science, COMP 20090
The Wason selection task
A D 4 7
• Rule: If there is a vowel on one side of a card, then there is an odd
number on the other.
• Question: Which card(s) do you need to turn over to test that this
rule is true for these four cards?
• Approximately 5-10% of people (university students) get this right on
their first try.
Introduction to Cognitive Science, COMP 20090
A D 4 7
• Selecting the A resembles use of
the modus ponens argument: we
need to check that a vowel always
leads to an odd number to verify
the rule.
Introduction to Cognitive Science, COMP 20090
A D 4 7
The selection of the 4 resembles the use
of the modus tollens argument: we
need to check that not odd really does
imply not a vowel to verify the rule.
Introduction to Cognitive Science, COMP 20090
A D 4 7
Incorrectly selecting the 7 resembles the
affirmation of the consequent fallacy.
If p then q
q
therefore: p
Introduction to Cognitive Science, COMP 20090
People perform much
better if the propositions
and rules are meaningful
Introduction to Cognitive Science, COMP 20090
drinking drinking 16 years 20 years
a a old old
coke beer
• Each card has the age of a person and what they are drinking on
each side. Which card/s should you turn over to find out if someone
is drinking illegally?
• 60-70% of people make the correct selections on these kinds of
tasks (compared to 5-10% for the original task). However, the above
problem is logically identical to the original Wason selection task.
• Our ability to reason about conditional statements therefore shows
some interesting context effects. Meaning Matters!
Introduction to Cognitive Science, COMP 20090
Syllogistic reasoning
• A syllogism is a piece of deductive reasoning that consists of a
major premise, a minor premise and a conclusion. Syllogisms
have the following general from:
Major premise: All/Some/No M are [not] P.
Minor premise: All/Some/No S are [not] M.
Conclusion: Therefore, All/Some/No S are [not] P.
• Example:
All Frenchmen are cheese-eaters
All cheese-eaters are surrender-monkeys
Therefore, all Frenchmen are surrender-monkeys
• The above syllogism is logically valid (even if the conclusion is
untrue). The truth of a conclusion depends both on the validity of
the argument and the truth of the premises.
Introduction to Cognitive Science, COMP 20090
• There are 512 ways of combining the All, Some, No and not
quantifiers, producing 512 possible syllogisms. Most of these do not
produce logical arguments.
• As we saw for conditional reasoning, people often make errors when
making judgments about arguments involving syllogistic reasoning.
• For example, consider:
All French-men are cheese-eaters
Some cheese-eaters are farmers
Therefore, some Frenchmen are farmers
• The conclusion does not follow logically from the premises. However,
people often judge the argument to be a valid one. (Johnson-Laird,
1990)
• Such findings can be explained by belief biases – people make
judgements based on prior beliefs and knowledge rather than on the
rules of logic
Introduction to Cognitive Science, COMP 20090
Had we but world enough, and time,
Andrew Marvell: To This coyness, Lady, were no crime.....
His Coy Mistress .....
But at my back I always hear
Great poem.. Time's wingèd chariot hurrying near;
And yonder all before us lie
Deserts of vast eternity....
Which logical fallacy
does it commit? ....
Does that detract Now therefore, while the youthful hue
Sits on thy skin like morning dew,
from the argument? And while thy willing soul transpires
At every pore with instant fires,
Now let us sport us while we may,
Introduction to Cognitive Science, COMP 20090
