Questions: Cognitive Biases in Judgment Under Uncertainty

Question 1

Linda is 31, single, outspoken, and very concerned with social justice. Which is more probable? (A) Linda is a bank teller. (B) Linda is a bank teller and a feminist activist.

• A. Option B — the description fits a feminist teller so much better that the probability is higher.
• B. Option A — P(A) ≥ P(A and B) always holds; every feminist teller is also a teller, so teller alone is at least as probable.
• C. They are equally probable because feminist teller is simply a more specific version of teller.
• D. Option B — adding true, fitting information to a description always increases its probability.

Answer: (B)

This is the conjunction fallacy, driven by the representativeness heuristic. The description makes 'feminist teller' feel more probable because it fits the narrative better — but the basic probability rule P(A) ≥ P(A and B) is unconditional: the set of all bank tellers must include all feminist tellers plus any non-feminist tellers. Adding a condition can only maintain or reduce probability, never increase it. Option D states the bias explicitly and is wrong for the same reason. The Linda problem shows how narrative coherence overrides logic.

Question 2

A disease affects 1 person in 1,000. A diagnostic test is 99% accurate (1% false positive rate). A patient tests positive. A physician estimates there is roughly a 99% chance the patient has the disease. Which cognitive error is the physician making?

• A. Availability bias — the vividness of a positive test result inflates its perceived reliability.
• B. Base rate neglect driven by the representativeness heuristic — the test's accuracy is salient while the disease's rarity is ignored.
• C. Anchoring bias — the 99% figure in the test's accuracy anchors the probability estimate upward.
• D. No error — 99% test accuracy means a positive result is 99% reliable.

Answer: (B)

Applying Bayes' theorem: out of 1,000 patients tested, about 1 is a true positive and about 10 are false positives (1% × 999). So only roughly 1 in 11 positive tests (~9%) reflects actual disease. The physician focuses on the accuracy figure — which feels representative of a reliable result — and ignores the base rate (1 in 1,000). This is base rate neglect: the representativeness of 'accurate test → positive result' crowds out the prior probability. Option D is the misconception being tested.

Question 3

Anchoring effects on judgment persist even when subjects are explicitly told the initial anchor value is random and irrelevant to the question.

• T. True
• F. False

Answer: True

In Tversky and Kahneman's classic studies, subjects who spun a wheel rigged to land on 10 or 65 gave systematically different estimates of the percentage of African nations in the UN — despite knowing the wheel was random. Awareness does not eliminate anchoring because the bias operates in the automatic, fast-processing system that generates the initial estimate before deliberate reasoning can intervene. This is what makes anchoring practically important: debiasing requires structural interventions (removing the anchor, requiring explicit base rate information) rather than mere awareness.

Question 4

The availability heuristic produces accurate frequency estimates when events are salient and easy to recall, because salient events are usually frequent.

• T. True
• F. False

Answer: False

The heuristic is a useful rule of thumb precisely because frequency and memorability are often correlated — but the correlation breaks down when events are memorable for reasons other than frequency: emotional salience, novelty, media coverage, and personal relevance all inflate availability without reflecting actual rates. Shark attacks are far rarer than deaths by vending machine but vastly more memorable. The heuristic systematically misestimates exactly those cases where society's attention and resources are most misallocated — dramatic, visible risks are overweighted relative to chronic, statistical ones.

Question 5

Why do cognitive biases in probability judgment persist even when people are aware of them, and what does this imply about effective debiasing?

Model answer: Cognitive biases arise from fast, automatic heuristic processing that generates judgments before deliberate reasoning can intervene. Awareness engages slower, deliberative reasoning, but typically too late to override the initial biased estimate — you can recognize the Linda problem as a conjunction fallacy after the fact while still having 'felt' option B was more probable. Research shows awareness reduces bias only marginally. What works better are structural interventions: presenting base rates explicitly and prominently, requiring decision-makers to consider the opposite hypothesis, using checklists that force consideration of alternatives, and redesigning choice architectures to remove anchors.

The practical implication is that individual education is a weak debiasing tool. The biases are features of the cognitive architecture, not merely failures of knowledge. Effective debiasing changes the environment of the decision — the structure of how information is presented — rather than relying on individuals to mentally correct for biases they know they have.