Trap/Heuristic/No. 0846
Representativeness Heuristic
The representativeness heuristic is judging probability by resemblance to a familiar type rather than all relevant evidence. Named by Amos Tversky and Daniel Kahneman in 1972, this mental shortcut can lead people to overlook base rates, sample sizes, or category boundaries.
- Evidence
- Well established
- Read
- 6 min
- Links
- 10 connections
01You've seen this when…
- in life
An investment fund’s last six monthly returns rise steadily. The smooth chart looks like what you associate with skilled management, so you trust it more than six months of evidence warrants.
- at work
A founder has the hoodie, fast delivery, and ambitious pitch you associate with successful tech entrepreneurs. In the investment meeting, those familiar details stand in for evidence that this company will succeed.
- out in the world
A lottery draw produces 4, 5, 6, 7, 8, and 9. Viewers suspect something is wrong because winning numbers are supposed to look scattered.
02The idea
Something looks like a member of a category, so you judge that it probably belongs there. A person resembles your picture of an engineer. A company resembles a famous success. A sequence resembles what randomness should look like.
This shortcut is the representativeness heuristic: using resemblance to judge category membership or probability. Your mind answers the easier question of how well something fits a familiar type instead of weighing all the evidence about how likely it is.
Resemblance can be useful evidence. The mistake is letting it crowd out other information: how common the category is, how often outsiders share the same features, or how much data you have.
It differs from the availability heuristic. Availability asks what comes easily to mind. Representativeness asks what fits the pattern. A memorable plane crash can make flying feel dangerous through availability; a passenger who resembles a fictional villain can seem dangerous through representativeness.
03Why it happens
- A match arrives before a calculation. You can recognize a familiar type without deliberately working through probabilities. That quick impression feels like an answer, even when the question requires more information.
- A good description feels like strong evidence. Details make a person or event easy to picture. But a feature common among engineers may also occur among many non-engineers. Predictive value depends on the feature’s frequency among engineers and non-engineers.
- Small samples are expected to resemble the whole. If a coin is fair, a short sequence seems as though it should contain roughly equal numbers of heads and tails. Random samples can be lopsided. Expecting every handful to mirror the population contributes to the law of small numbers bias.
- Extra details can improve the fit while reducing the probability. A politically active person may seem more like a feminist bank teller than a bank teller. Yet the first category is contained within the second. Ranking the narrower description as more probable is the conjunction fallacy.
04A worked example
In a 1973 study, Daniel Kahneman and Amos Tversky gave participants short personality descriptions drawn from a pool of engineers and lawyers. One group was told the pool contained 70 engineers and 30 lawyers, while another was told it contained 30 engineers and 70 lawyers. Participants estimated the chance that each described person was an engineer.
What it looks like Careful individual assessment. Participants use details about each person to make individual judgments, giving those details priority over the group statistic.
What’s actually going on With the sketches in hand, judgments depended heavily on how much each person sounded like an engineer and much less on the composition of the pool. The same portrait could seem engineer-like whether engineers were the majority or the minority. Resemblance displaced the base rate: how common engineers were before considering the sketch.
What would have helped Starting with the pool’s proportions, then asking how strongly the description distinguishes engineers from lawyers. A sketch might contain useful clues; the starting odds remain relevant. Participants used the proportions much more appropriately when no personality description was supplied.
05How to spot it
06What to do instead
- Establish the starting rate first. Before reading the vivid description, find out how common the outcome is. When evaluating a project or an investment, or deciding whom to hire, use reference-class forecasting: begin with outcomes for a relevant group of comparable cases.
- Translate percentages into counts. In a hypothetical pool of 1,000 people, suppose 100 are engineers. A particular profile fits 80% of engineers and 10% of everyone else. That gives 80 matching engineers and 90 matching non-engineers. Even a strongly engineer-like profile leaves engineers slightly outnumbered among the matches.
- Test how distinctive the clue is. Ask how often the same feature appears when your guess is wrong. The useful quantity is the likelihood ratio, which compares the clue’s frequency under competing explanations. Bayes’ theorem combines that evidence with the starting odds.
- Check the size and boundaries of the evidence. Six good months provide less evidence than six good years. For a detailed story, check whether its events are all required to happen together. For a random sequence, distinguish one exact outcome from a broader pattern defined before seeing the result.
- Look for a distinguishing test. Replace another impression with evidence that separates the alternatives: a job-relevant work sample, a longer performance record, or a check of whether the proposed process actually occurred.
07When it isn’t a shortcut gone wrong
Similarity can guide accurate judgments. A mechanic recognizing a familiar failure pattern may be using diagnostic features learned through experience. The shortcut works better when the important features distinguish the alternatives in a stable setting with accurate feedback. That’s the concern of ecological rationality: whether a shortcut fits its environment.
Experimental wording matters too. In some conjunction tasks, people interpret a broad option as excluding the extra detail in the narrower option. Frequency formats can also improve probability judgments. So an incorrect answer is not always clean evidence that resemblance alone caused the error.
Check what recognizable patterns predict, using a relevant comparison group and accounting for how much evidence you have.
08Roots
In the early 1970s, Amos Tversky and Daniel Kahneman built probability problems out of ordinary materials, asking people to judge personality sketches and reason about coin tosses and small samples. The puzzles exposed a striking gap. People could know something about chance yet still expect its results to look like a tidy miniature of the process producing them.
Their 1972 paper, Subjective probability: A judgment of representativeness, named the shortcut. The later engineer-and-lawyer experiments made the problem especially clear: a few lines of biography could outweigh the stated composition of a group. The idea then appeared in accounts of how people predict outcomes and make investment decisions, and why they commit the conjunction fallacy.
An adaptive explanation is plausible but speculative. Quickly sorting unfamiliar things by resemblance could save effort and often produce useful guesses. This explanation leaves open why this particular shortcut evolved and whether every modern use is well adapted. The experimental account is firmer than the evolutionary story.
09How solid is this?
Classic experiments document judgments driven by resemblance despite relevant base rates, sample sizes, or category boundaries. Task wording, interpretation, and number format affect the errors, so resemblance is neither universally misleading nor the only explanation for every incorrect answer.
10Connections
- Often confused with Base Rate Fallacy, Availability Heuristic
- Countered by Bayes’ Theorem, Reference-Class Forecasting, Likelihood Ratio
- Can lead to Conjunction Fallacy, Gambler’s Fallacy, Law of Small Numbers Bias
- Part of Heuristic
- See also Ecological Rationality
11Origin and sources
Amos Tversky and Daniel Kahneman named and described the heuristic in their 1972 paper, Subjective probability: A judgment of representativeness.
- [1]Tversky, A., & Kahneman, D. (1972). Subjective probability: A judgment of representativeness. Cognitive Psychology, 3(3), 430–454.
- [2]Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80(4), 237–251.
- [3]Tversky, A., & Kahneman, D. (1983). Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Review, 90(4), 293–315.
- [4]Gigerenzer, G., & Hoffrage, U. (1995). How to improve Bayesian reasoning without instruction: Frequency formats. Psychological Review, 102(4), 684–704.
Suggest an edit· Updated 2026-10-02