16-10-2025
Heuristics and Biases
ED 323: Behavioural Economics
Instructor: Prof. Sunny Bhushan
Heuristic - a mental shortcut
• Heuristics are mental shortcuts or rules of thumb that people use to
make decisions. They differ from optimization in that they typically
involve a limited information set and a more computationally
tractable decision method.
• The recognition heuristic - if one of two objects is recognized, infer
that the recognized object is more likely to be the target object.
• Eg: predict which of two players will win tennis match.
1
16-10-2025
Heuristic - a mental shortcut
• There is substantial evidence that people use heuristics.
• People don’t normally calculate conditional probabilities using Bayes’ rule.
Instead, the heuristics might approximate Bayes rule under certain conditions.
• Heuristics are often accurate and tractable, but in some
environments can lead to error.
• Tversky and Kahneman (1974) defined three classic
heuristics:
• representativeness, availability and anchoring.
Representativeness heuristic
• Suppose you wish to estimate the probability that an event or
person belongs to a certain class.
– “What is the probability that event A belongs to class B?”
– “What is the probability that process B will generate event A?”
• Under the representativeness heuristic, people evaluate
probabilities by the degree to which A is representative of
(similar to) B.
• People judge probability by similarity or stereotype matching.
2
16-10-2025
Representativeness heuristic
• Tversky and Kahneman (1974) provide the following
example:
– An individual who has been described by neighbor as follows:
• “Steve is very shy and withdrawn, invariably helpful, but with little interest in
people, or in the world of reality. A meek and tidy soul, he has a need for order and
structure, and a passion for detail.”
– What is the probability that Steve is a librarian?
• assessed by the degree to which he is representative of, or similar to, the stereotype of a
librarian.
Representativeness heuristic
• This approach to the judgement of probability leads to neglect
of statistical principles causing serious errors.
– Base-rate neglect (insensitivity to prior probabilities of outcomes)
• Personality description of group of 100 professionals – engineers and lawyers.
• Group A : 70 engineers and 30 lawyers
• Group B : 30 engineers and 70 lawyers
3
16-10-2025
Representativeness heuristic
• This approach to the judgement of probability leads to neglect
of statistical principles causing serious errors.
– Sample size neglect (Insensitivity to sample size)
• A town is served by two hospitals. Large hospital – 45 babies born /day. Small
hospital – 15 babies born/day
• Now, if 50% of all babies are boys. But exact percentage varies from day to day. In
one year, each hospital recorded the days on which more than 60% of the babies
born were boys. Which hospital you think recorded more such days?
• Response:
– The large hospital : 21
– The small hospital : 21
– About the same : 53
Representativeness heuristic
• This approach to the judgement of probability leads to neglect
of statistical principles causing serious errors.
– Conjunction fallacy (thinking specific conditions are more probable)
• In considering tosses of a coin, people regard the sequence
– H-T-H-T-T-H more likely than the sequence
– H-H-H-T-T-T which does not appear random, and
– H-H-H-H-T-H which does not represent the fairness of the coin.
4
16-10-2025
Availability heuristic
• Under the availability heuristic, people assess the frequency of a
class or the probability of an event by the ease with which they can
recall instances or occurrences.
– Assessing risk of heart-attack among middle-aged people by recalling
such occurrences among people you know.
– Probability of success of a business venture by imaging various
difficulties it could encounter.
• Availability is a useful clue for assessing frequency or probability
because instances of large classes are usually recalled better and
faster than instances of less frequent classes.
Availability heuristic
• Availability is also affected by factors other than
frequency and probability. So, complete reliance
availability may lead to predictable biases.
– Biases due to the retrievability of instances.
– Biases due to the effectiveness of a search set.
– Biases of imaginability.
– Illusory correlation.
10
5
16-10-2025
Availability heuristic
• In one experiment, Tversky and Kahneman (1974) gave
experimental subjects lists of names.
– In some lists, the men were more famous than the women, and in other
lists, vice versa.
– After viewing the list, they were asked whether the list had more men
or women.
• For each list, the subjects judged that the sex with more famous
names was more common. Those names were more available in
their minds.
11
Adjustment and anchoring heuristic
• When using anchoring and adjustment, people estimate by
starting from an initial value and adjust from that value to
obtain the final estimate.
– Suppose you know the odds of outcome A, and want to estimate the
odds of outcome B. Anchoring and adjustment implies that you will
start with the odds of outcome A and adjust to obtain the odds of
outcome B.
12
6
16-10-2025
Adjustment and anchoring heuristic
• The accuracy of anchoring and adjustment depends on the anchor’s quality
and the size of the adjustment from the anchor.
– The quality of the anchor relates to the strength of the correlation between the
anchor and the quantity being estimated.
• Empirically, people tend to use weak or irrelevant anchors.
– The size of the adjustment then needs to account for the relationship between
the anchor and the quantity being estimated.
• Empirically, observed adjustments from the anchor are too small.
• Example:- Tversky and Kahneman (1974): Percentage of African countries
in the United Nation.
13
The Cab Problem
(Tversky & Kahneman, 1982)
• The Scenario:
– A hit-and-run accident occurred at night involving a cab.
– In the city:
• 85% of the cabs are Green,
• 15% are Blue.
– A witness identifies the cab as Blue.
– The court tests the witness’s accuracy under similar conditions and finds
they are 80% reliable (i.e., correctly identify color 80% of the time).
• The Question:
– What is the probability that the cab involved in the accident was actually Blue?
14
7
16-10-2025
The Cab Problem
(Tversky & Kahneman, 1982)
What People Usually Answer Normative (Bayesian) Reasoning
• Most people say “80%”, • According to Bayes’ Theorem:
matching the witness’s
reliability.
• They ignore the base-rate
information (85% Green,
15% Blue).
15
The bias-variance trade-off
• Heuristics, while often associated with errors, can be powerful and
effective decision-making tools due to the bias-variance trade-off.
• The bias-variance trade-off suggests that as model complexity
increases, bias tends to decrease, but variance (sensitivity to data
fluctuations) increases. There’s an optimal level of complexity that
balances these two factors.
• Simple heuristics can sometimes outperform more complex
decision-making strategies by finding a better bias-variance trade-
off, leading to lower overall error despite using less information.
– The “Take the Best” heuristic
– The gaze heuristic
16
8
16-10-2025
Understanding bias and variance
• Bias: -
– Bias is the degree to which there are erroneous assumptions in your model.
– The classic case of bias is when you have failed to include a relevant predictor.
• Variance: -
– The inclusion of too many predictors can lead to what is called variance, which is
an error that arises because of the sensitivity of the model to fluctuations in the data
you use to develop the model.
– It ultimately involves giving too much weight to irrelevant or marginally relevant
information.
• Predict the level of dropout in a school
17
• An unbiased predictor will tend to center on the
target. A low variance predictor will tend to cluster. A
low variance, low bias estimate is the best outcome.
18
9
16-10-2025
19
Overconfidence
• “Perhaps the most robust finding in the psychology of
judgment and choice is that people are overconfident.”
De Bondt and Thaler (1995)
• Overconfidence is a robust finding in judgment and decision-
making research, but it actually encompasses three distinct
phenomena:
– Overprecision,
– Overestimation, and
– Overplacement.
20
10
16-10-2025
Overconfidence
• Overprecision is the tendency to believe our predictions
or estimates are more accurate than they are, often
demonstrated in confidence interval tasks.
• Overestimation is the belief that we can perform better
than we realistically can, typically occurring for difficult
tasks but not for easy ones.
• Overplacement is the erroneous judgment that we are
better than others, which varies with task difficulty.
People tend to over-place on easy tasks and under-place
on difficult ones.
21
Thank you!
22
11