2. Chapter 12: Judgment and Reasoning
Lecture Outline
Judgment Heuristics
Anchoring
Detecting Covariation
Assessing the Damage
Confirmation and Disconfirmation
Logic
Decision Making
3. Judgment Heuristics
Induction (歸納) is a pattern of
reasoning in which one seeks to draw
general claims from specific bits of
evidence
Based on what you know about Allen, what is
likely to cheer him up today?
Based on what you know about cars, what is
the best kind to buy?
4. Judgment Heuristics
Heuristic (捷思)
Reasonably efficient and accurate
Gain efficiency lose accuracy
Judgment heuristics include
Attribute substitution
Availability heuristic
Representativeness heuristic
5. Judgment Heuristics
Attribute substitution is a strategy used
when we do not have easy access to a
desired piece of information
Instead, we base our decision on readily
available information (a proxy or index)
that we believe is correlated with the
desired information
7. Judgment Heuristics
Availability heuristic
Specific case of attribute substitution
Ease with which examples come to mind is
an index of frequency or likelihood
This class is easy—I know four people who got
A’s.
8. Judgment Heuristics
For instance, consider
In the English language, are there more
words that start with the letter “R” or with the
letter “R” in the third position?
Who washes the dishes more often, you or
your roommates?
Are more deaths caused by crimes or by
diseases?
9. Judgment Heuristics
The availability heuristic
More words that begin with “R”
Reality
More words with “R” in the third position,
But words that begin with “R” are easier to
bring to mind
10. Judgment Heuristics
The availability
heuristic may lead us
to believe that we
always do the
housework ourselves
11. Judgment Heuristics
The availability
heuristic may lead us
to think that more
deaths are caused by
crimes or accidents
than by disease
In reality, far more
deaths are caused by
disease
12. Judgment Heuristics
Availability heuristic
Group of students asked to recall past
episodes in which they had been assertive
(Schwarz et al., 1991)
One group gave 6 examples, and another, 12
examples
Which group then judged themselves to be
more assertive in general?
13. Judgment Heuristics
Those that gave 6 examples judged
themselves as more assertive
Easier to come up with 6 examples
14. Judgment Heuristics
The representativeness heuristic is
another example of attribute substitution
Member Prototype Category
resembles resembles
15. Judgment Heuristics
For instance, consider
Do you assume anything about someone if
you discover that he or she is a lawyer or an
engineer?
If a coin toss results in “heads” six times in a
row, what are the odds of getting “tails” the
seventh time?
If you hear an anecdote about a marathon
runner who has smoked for decades and is
perfectly healthy, does this mean that
smoking is safe?
16. Judgment Heuristics
The representativeness heuristic
All lawyers or all engineers are
homogeneous (a stereotype)
We assume that each individual member of a
category has the traits we associate with the
category overall.
17. Judgment Heuristics
The representativeness heuristic
The seventh coin toss is more likely to be tails
(the gambler’s fallacy), but the odds are still
50-50
18. Judgment Heuristics
The representativeness heuristic
Smoking must be okay for your health based
on one example (anecdotal evidence or “man
who” stories)
This is an example of reasoning from one
instance to the population.
19. Judgment Heuristics
Representativeness heuristic
Watched “prison guard” discussing his job (Hamill et
al., 1980)
One variable of the study was whether the
guard was compassionate or contemptuous.
Another variable was whether the participants
were told the guard was representative of all
guards.
What did participants later conclude about
prison guards in general?
20. Judgment Heuristics
Compassionate Contemptuous
Representative Compassionate Contemptuous
Not Representative Compassionate Contemptuous
Judgments were based on characteristics independent of whether
participants were told the guard was representative or not
21. Detecting Covariation
Covariation
Relationship between two variables
Negative or positive and can vary in strength.
Does college education lead to a higher paying
job?
Do you feel better when you have a good
breakfast?
22. Detecting Covariation
An illusory covariation
A perceived pattern such that
one variable predicts another
A study of Rorschach inkblots
found that even when fictitious
patients and fictitious
responses were randomly
paired, people believed they
had found patterns.
Other studies have found that
clinicians believed that
homosexuals and
heterosexuals interpreted these
images differently even though
the data did not show this.
23. Detecting Covariation
Confirmation bias
More responsive to evidence that confirms
one’s beliefs
Similar to overregularization by schemata
Essentially we ignore disconfirming data
24. Detecting Covariation
Big dogs are vicious
Notice examples that fit this pattern more
readily (biased attention)
Will recall examples that fit the pattern more
readily (biased memory)
25. Detecting Covariation
Another reason that estimates of
covariation can be inaccurate is a neglect
of base-rate information.
Base-rate information—information
about the likelihood of an event
Diagnostic information—does an
individual case belong to a category?
26. Detecting Covariation
Consider this example
Testing a new drug, in hopes that it will cure
hepatitis
Does taking the drug covary with a better
medical outcome?
27. Detecting Covariation
Results
70% of the patients taking the drug do recover
from the illness
Uninterpretable
If it turns out that the overall recovery rate is
70%, then our new drug is having no effect
whatsoever.
29. Detecting Covariation
Kahneman and Tversky (1973)
Base-rate information: 70 lawyers and 30
engineers
Diagnostic information (engineering):
“likes carpentry, sailing, math puzzles;
dislikes politics”
30. Detecting Covariation
The base-rate with no diagnostic
information = base rate
The base rate is not neglected!
Base-rate and diagnostic information =
diagnostic information.
The base rate is neglected!
31. Detecting Covariation
Is Tom an engineer? Does Tom resemble an engineer?
Representativeness heuristicWhat percentage are engineers?
Attend to base rate Base-rate neglect
32. Assessing the Damage
Imagine your friend has a system for
playing the lottery
What if she tells you it only worked the last
time she played or
She tells you it worked the last ten times she
played
Which do you believe?
Data set size and drawing conclusions
about a new category: One time might be
lucky; ten times is likely to be true.
33. Assessing the Damage
Dual-process models
System 1 refers to thinking that is fast,
automatic, and uses heuristics
System 2 refers to thinking that is slower,
effortful, and more likely to be correct
34. Assessing the Damage
Whether System 1 or System 2 is used
depends on the context of the decision
How much time is available for the decision?
How much attention and working memory are
available?
And how the problem is presented
What format are the data in?
Are statistical concepts primed?
35. Assessing the Damage
Emphasizing chance cues statistical
reasoning
A story about a restaurant assessment based
on a single meal chosen by one person
Informed assessment?
Now imagine that the person dropped his or
her pencil on the menu to pick that meal
Informed assessment?
36. Assessing the Damage
Background knowledge increases the
likelihood that participants will pay
attention to base rates
For instance, when predicting whether a
particular student will pass an exam,
participants do pay attention to the base-rate
information that only 30% of students pass
the exam
38. Assessing the Damage
Training can influence the likelihood of
reasoning with System 2
For instance, participants can be trained that
large samples of data are more reliable than
small samples
Taking a statistics class also improves
reasoning when sample size is important
39. Confirmation and Disconfirmation
Deduction start with general premises
and ask what follows
If you believe that red wine gives you
headaches, what follows from this?
If relationships based on physical attraction
never last, what follows from this?
40. Confirmation and Disconfirmation
Rooster wants to prove his crowing
causes the sun to rise
Confirming evidence: every day the
rooster crows, and the sun rises
Disconfirming evidence: one day he must
not crow and see what happens
41. Confirmation and Disconfirmation
Confirmation bias—more responsive to
evidence that confirms one’s beliefs and
less responsive to evidence that
challenges one’s beliefs
42. Confirmation and Disconfirmation
In a classic demonstration of confirmation bias,
Wason (1966) presented sequences like “2-4-6”
Several minutes to figure out rule (ascending
numbers two apart)
They only sought confirming evidence
Did not seek disconfirming evidence
A few did discover the rule
Sought disconfirming evidence
43. Confirmation and Disconfirmation
Selective Memory.
Gamblers betting on a football game
Wins are confirming evidence
Losses are remembered as near-wins
44. Confirmation and Disconfirmation
Belief perseverance is a tendency to
continue endorsing a belief even when
evidence has completely undermined it
45. Confirmation and Disconfirmation
Assigned to good or bad groups
Told about assignment
Bad judgers Good judgers
Ability to judge
whether authentic
Bad Good
Self-ratings of
social sensitivity
Low High
47. Logic
Is this syllogism
valid?
All P are M.
All S are M.
Therefore, all S are P.
In concrete terms:
All plumbers 水電工
are mortal 凡人 .
All sadists 虐待狂 are
mortal.
Therefore, all sadists
are plumbers.
48. Logic
The errors people make on syllogisms
tend to fall into predictable categories
One pattern is belief bias—if the
syllogism’s conclusion is something
people already believe to be true, they are
more likely to judge the conclusion as
following from the premises
49. Logic
Low-level matching strategy between the
words in the premises and those in the
conclusions (the atmosphere effect)
Some A are not X.
Some B are not X.
Therefore, some A are not B. (invalid)
51. Logic
modus ponens, affirming the antecedent
If P is true, then Q is true.
P is true.
Therefore, Q must be true.
Easiest form of logic
52. Logic
modus tollens, denying the consequent
If P is true, then Q is true.
Q is false.
Therefore, P must be false.
More difficult
53. Logic
Two common errors are affirming the
consequent
If P is true, then Q is true.
Q is true.
Therefore, P must be true. (invalid)
And denying the antecedent
If P is true, then Q is true.
P is false.
Therefore, Q must be false. (invalid)
54. Logic
If P is true, then Q is true.
P true P no true
Q true Modus ponens Illogical
Q not true Illogical Denial of the
consequent
55. Logic
If P is true, then Q is true.
Conditional statement Type of reasoning
P true, Q true Modus ponens
P not true, Q not true Denial of the antecedent
Q true, P true Affirmation of the consequent
Q not true, P not true Modus tollens
57. Logic
For both syllogisms
and conditional
statements, errors are
more likely when
Negatives are involved
The terms are abstract
(e.g., letters) and not
concrete
58. Logic
Wason’s four-card task
“If a card has a vowel on one side, then it must
have an even number on the other side”
Which cards must be turned over to test this
rule?
62. Logic
Why are some versions of the four-card problem
difficult and others easy?
Evolutionary psychologists suggest people can
“detect cheaters” who are not following rules of
social interaction
63. Logic
Alternatively, a pragmatic reasoning schema
may help explain the ease
These schemas involve “permission” or “cause
and effect” relations
64. Logic
Problem: “If a form says ‘entering’ on one side,
then the other side must include ‘cholera.’”
65. Logic
Permission schema
“If a passenger wishes to enter the country, he
or she must first receive a cholera inoculation.”
66. Logic
Necessary condition
“If Jacob passed his driver’s test, then it’s
legal for him to drive.”
Sufficient condition
“If Solomon is eligible for jury duty, then he is
over 21.”
67. Logic
Summary of logic
People commonly rely on reasoning
strategies that are different from the principles
of formal logic
Some of these principles are simple, such as
the “matching strategy”
Others are more sophisticated, such as a
“permission schema,” but may only be
triggered under the right circumstances
69. Decision Making
Many of our decisions follow the principle
of utility maximization, or choosing the
option with the greatest expected value
70. Decision Making
However, many decisions do not follow
this principle
For instance, consider the following
problem, as framed either in terms of lives
saved or lives lost
72. Decision Making
Framing changes the choices
Program A if the problem is “positively framed”
in terms of lives saved
Program B if the problem is “negatively
framed” in terms of lives lost
Identical utility
77. Decision Making
An alternative view is known as reason-
based choice, the idea that people make
a decision only when they detect what
they believe to be a persuasive reason for
making that choice
81. Decision Making
The orbitofrontal
cortex is essential for
evaluation of somatic
markers
Patients with damage
will make risky
decisions
82. Decision Making
Emotions play a role in decision making, through
what might be called affective heuristics
For instance, decisions that involve assessing
risk may depend on the feeling of dread of an
undesirable outcome, or anticipating the feeling
of regret for having made the wrong choice
Note that the latter involves predictions about
our future emotions, which are not necessarily
accurate
84. Decision Making
Decision making and happiness
Unable to forecast our future feelings
Would be better off having others make our choices
We end up “stumbling on happiness”
We end up stressed by the “paradox of choice”
86. 1. The fact that people report motor-vehicle
deaths as more common than diabetes
and homicides as more common than
stomach cancer reflects which of the
heuristics?
a) simulation heuristic
b) anchoring heuristic
c) availability heuristic
d) representativeness heuristic
87. 2. Which of the following is TRUE of
covariation?
a) A negative covariation indicates that there
is no relationship between two variables.
b) People tend to underestimate covariation
when they have theories about the
relationship between two variables.
c) Covariations are “all-or-none” and cannot
vary in strength.
d) Illusory covariations sometimes generate
prejudice toward groups of people.
88. 3. Which of the following is FALSE regarding
confirmation bias?
a) It works to bring our recollections into line
with our expectations.
b) It makes people more alert and
responsive to evidence that confirms their
beliefs than to challenging evidence.
c) Its effects are usually offset by our
general ability to think about covariation.
d) It makes us unlikely to seek
counterexamples.
89. 4. Poor diagnostic reasoning and illusory
correlations have been documented in all
of the following cases EXCEPT
a) individuals with considerable experience
in the domain being judged.
b) participants who have been offered cash
bonuses for accurate performance.
c) individuals for whom the stakes are very
high (e.g., doctors and financial advisors).
d) All of the above individuals demonstrate
these errors.
90. 5. According to the dual-process model of
reasoning, one mode of thought is ___,
while the other mode of thought is ___.
a) association driven; speedy
b) automatic; effortful
c) slower; effortful
d) automatic; effortless
91. 6. In the study in which people were asked to judge
their social sensitivity after being given false-
positive or negative feedback (but then debriefed),
participants were clearly influenced by
a) views they had of themselves before the
experiment.
b) the feedback they had been given, even though
they knew it was false.
c) the feedback they had been given, but only if they
had forgotten the debriefing that undermined this
feedback.
d) a memory search done after debriefing to help
them disconfirm the false feedback.
92. 7. In the context of a syllogism, what is a
matching strategy?
a) If the two premises match each other, the
conclusion is accepted.
b) If the conclusion matches the premises in
wording and structure, it is accepted.
c) Statements with the same structure are
all seen as identical.
d) People accept syllogisms when the
conclusions match their beliefs.
Notes de l'éditeur
Green is accurate reasoning. Red is inaccurate reasoning.
Correct answer: c
Feedback: We are more likely to read about motor-vehicle deaths or see them on the evening news and thus judge them as being more likely.
Correct answer: d
Feedback: Illusory covariations would have us detect a relationship that is not there and hence lead to prejudice.
Correct answer: c
Feedback: We make a lot of judgment errors when thinking about covariation. Hence, we are not good at it.
Correct answer: d
Feedback: Illusory correlations are pervasive even when people have vast experience in a domain.
Correct answer: b
Feedback: Heuristics are fast and efficient; reason-based choices are slower.
Correct answer: b
Feedback: People used heuristics to make judgments even when given information that clearly showed they were wrong.
Correct answer: b
Feedback: The matching strategy refers to wording.