Vous êtes sur la page 1sur 17

Reasoning and Decision Making

Five general strategies


Reasoning and Logic
Two hypotheses
inherently logical
logic must be learned

Cognitive - reasoning.ppt 2001 Laura Snodgr

Reasoning and Decision Making


Evidence
categorical syllogisms
sources of error
conditional reasoning
sources of error

Bounded rationality

Cognitive - reasoning.ppt 2001 Laura Snodgr

Five General Strategies


Prototype matching
is an instance a member of a category

Representativeness heuristic
intuitive predictions
stereotypes
e.g. Is the man, with the glasses and tweed jacket who is reading a book at
the lunch table, a truck driver or a college professor?
base rate probabilities

Simulation heuristic
imagine a mental model

Cognitive - reasoning.ppt 2001 Laura Snodgr

Five General Strategies


Reasoning by analogy
similarity to past experiences or problems
problem of distinguishing valid from invalid analogies

Availability heuristic
frequency versus vividness of examples
examplar search
look for counter examples
how easily does an example come to mind

Cognitive - reasoning.ppt 2001 Laura Snodgr

Reasoning and Logic


Reasoning
the thinking involved in determining whether one proposition logically follows
from another
attempts to determine the validity of an argument or idea

Formal Logic
specify rules of inference that yield valid arguments
validity is independent of content (truth value)
how we ought to reason

How do we actually reason?


Do the rules of formal logic describe actual behavior or do they require
careful training?

Cognitive - reasoning.ppt 2001 Laura Snodgr

Two Hypotheses
Information processing stages of reasoning
Stage one - encode a representation in working memory
Stage two - retrieve set of rules to check logic

Humans are inherently logical


LTM contains a set of rules of inference equal to formal logic
when people make errors it is because:
encode problem incorrectly
fail to use inference rules and respond on some other basis

Humans are not inherently logical


LTM has no logic rules unless formally trained

Cognitive - reasoning.ppt 2001 Laura Snodgr

Two Hypotheses
Evaluate the hypotheses based on evidence from two
types of reasoning
categorical syllogisms
e.g. All M are P
All S are M
therefore All S are P
conditional reasoning
e.g. If P, then Q
P
therefore Q

Cognitive - reasoning.ppt 2001 Laura Snodgr

Categorical Syllogisms
Consist of a major and minor proposition (premise)
followed by a conclusion
A valid syllogism is one in which the conclusion
necessarily follows logically from the premises
There are four possible figures
M-P

P-M

M-P

P-M

S-M

S-M

M-S

M-S

S-P

S-P

S-P

S-P

Cognitive - reasoning.ppt 2001 Laura Snodgr

Categorical Syllogisms
There are four possible moods

affirmative-universal
negative-universal
affirmative-particular
negative-particular

e.g. All S are P


e.g. No S are P
e.g. Some S are P
e.g. Some S are not P

There are a total of 256 possible configurations


only 24 of these are valid !

Cognitive - reasoning.ppt 2001 Laura Snodgr

Sources of Error in Categorical Syllogisms


Atmosphere effect
respond to the mood of the premises
whenever at least one premise is negative (particular) then the conclusion
chosen tends to be negative (particular)
irrational - not use logic rules just maintain mood (heuristic)
generally true for the valid but can lead to invalid results

Content
especially with emotionally laden premises and conclusions
based on believability of conclusion
use biases and prior belief
All scientists are honest
All women are scientists
therefore: All women are honest

Cognitive - reasoning.ppt 2001 Laura Snodgr

10

Sources of Error in Categorical Syllogisms


Failure to accept the logical task
logic is theoretical NOT empirical
must ignore the truth value of each premise
error in encoding the nature of the problem
subjects evaluate the truth not the logic
e.g. Kpelle rice farmers
e.g. children
evidence that education may teach us when to reason

Failure to discriminate information given in the premise


from information retrieved from LTM
unintentionally supplement
e.g. when premise is Some men are honest; add but not all

Cognitive - reasoning.ppt 2001 Laura Snodgr

11

Sources of Error in Categorical Syllogisms


Incorrect conversion of premises

assumptions of symmetry
e.g. All A are B does not mean All B are A
real world examples help block incorrect conversions
e.g. All dogs are animals is not converted to All animals are dogs
real world examples may be biased by beliefs e.g. All republicans are rich
may convert to All rich people are republicans if you believe this

Forgetting premises
three types of common recall errors
omit a premise
displace terms from one premise to next
change quantifiers (some to all)

Cognitive - reasoning.ppt 2001 Laura Snodgr

12

Conditional Reasoning
Two parts
statement of the condition rule
if the first proposition is true then the second is true
statement of the truth value of one of the propositions
e.g.
If the world was flat, then you could fall off the edge
You fell of the edge
Therefore ?

Cognitive - reasoning.ppt 2001 Laura Snodgr

13

Conditional Reasoning
Form
If p, then q
p
therefore q

Name

Example

modus ponens
(valid)

If the objects is square, then it is blue


The object is square
The object is blue

If p, then q
modus tollendo
Not q
tollens (valid)
Therefore not p

If the objects is square, then it is blue


The object is not blue
The object is not square

If p, then q
denying the
Not p
antecedent
Therefore not q (invalid)

If the objects is square, then it is blue


The object is not blue
The object is not blue

If p, then q
Q
Therefore p

If the objects is square, then it is blue


The object is blue
The object is square

affirming the
consequent
(invalid)

Cognitive - reasoning.ppt 2001 Laura Snodgr

14

Sources of Errors in Conditional Reasoning


Negatives
people have trouble tracing the implications of negation
make more errors when a not is present
modus tollens and denying the antecedent

Confirmation bias

bias to seek answers (evidence ) that confirms supports rather than denies
problem in scientific hypothesis testing
stereotypes and prejudice
e.g. given 2, 4, 6, 8
generate sequences and guess the rule

Cognitive - reasoning.ppt 2001 Laura Snodgr

15

Sources of Errors in Conditional Reasoning


Confirmation bias continued
whenever we have an hypothesis (belief, prejudice) we seek confirmations
not disconfirmations
one piece of disconfirmatory evidence is much more informative than several
confirmatory pieces of evidence
problem: scientists have a confirmatory bias which interferes with progress

Cognitive - reasoning.ppt 2001 Laura Snodgr

16

Bounded Rationality
Human may be inherently capable of logic
However

use heuristics
not always accept logical task
processing and memory errors
problems with negatives
cruising or mindlessness not use our capacity

Cognitive - reasoning.ppt 2001 Laura Snodgr

17

Vous aimerez peut-être aussi