The fundamental attribution error, and the Castro essay that showed it

The fundamental attribution error is the tendency to explain another person's behaviour by their personality while giving too little weight to the situation they were in. A colleague walks in late and the first thought is "disorganised", not "cancelled train". Edward Jones and Victor Harris showed the pattern in 1967 with essays about Fidel Castro, and Lee Ross gave it its name in 1977.

Below are the definition in plain words, the Castro experiment, how the error differs from correspondence bias and the actor-observer asymmetry, why it happens, and what the critics say. The article also flags where popular retellings of the experiment disagree with the researchers' own accounts.

• The fundamental attribution error is the habit of reading another person's behaviour as a sign of their traits or attitudes while underweighting the situation.
• In Jones and Harris's 1967 study, readers judged an essay writer to hold the view the essay argued even when told a debate coach had ordered the writer to argue it. The inference was weaker than for free-choice writers, but it was there.
• Ross named the pattern in 1977. Jones thought the name "overly provocative and somewhat misleading".
• Correspondence bias is often used as another name. Some researchers separate the two: correspondence bias is the observed pattern, and the fundamental attribution error is a proposed explanation of it.
• The actor-observer asymmetry compares how people explain their own and other people's behaviour. It is a separate claim, and a 2006 meta-analysis found no overall asymmetry.

What the fundamental attribution error means in simple terms

The APA Dictionary of Psychology defines the fundamental attribution error as the tendency to overestimate how far a person's behaviour is determined by their lasting personal characteristics, attitudes or beliefs and, correspondingly, to minimise the influence of the surrounding situation, such as financial or social pressure. The same entry gives correspondence bias as another name for it, and adds that the tendency seems to be more common in some societies than in others.

Attribution theory studies how people explain behaviour, and it starts from a simple split. A dispositional explanation points to the person: traits, motives, attitudes. A situational explanation points to the circumstances: pressure, roles, rules, bad luck.

The fundamental attribution error is leaning too far toward the first kind of explanation. Part of the reason is visibility. Gilbert and Malone call it the invisibility problem: you can film what a person does, but the pressure of an audience shows up nowhere in the footage.

Take the late colleague. The quick explanation is about the person: careless, disorganised, uninterested. The slower one is about the situation: a cancelled train, or a school run that overran. The error is not that personality never matters. It does. The error is taking the trait explanation without checking the situation, even when the situation fully accounts for what happened.

The Castro essay: how Jones and Harris showed it in 1967

The classic demonstration is Edward Jones and Victor Harris's paper The attribution of attitudes, published in 1967. As Daniel Gilbert describes it in a chapter on Jones's work, participants read an essay that supported or opposed Fidel Castro's regime. Some were told that the writer had freely chosen to write it. Others were told that a debate coach had ordered the writer to write it. Everyone then estimated how much the writer personally supported or opposed Castro.

Jones and Harris were testing a theory that predicted a clear difference: readers should infer the writer's real attitude when the writer had a free choice, and should not when the position had been dictated. Gilbert calls the prediction timid. The first half held: readers inferred strong pro-Castro and anti-Castro attitudes from the free-choice writers. The second half failed. Readers also drew inferences about the writers who had been assigned their side. The inferences were much weaker, but they were there. Gilbert writes:

Jones and Harris called this the observer bias, as Gilbert and Malone's 1995 review records. The phrase "fundamental attribution error" came later.

Retellings disagree on how the writer's side was assigned. Some popular accounts say a coin toss decided it, and a 2015 paper by Walker, Smith and Vul says a course instructor did. Gilbert and Malone's review and Gilbert's chapter both describe a debate coach, and this article follows them. Every version keeps the point: readers knew the position was assigned and still took the essay as evidence of the writer's own view.

The finding has been replicated many times. Gilbert and Malone record that the paradigm produced dozens of careful replications and extensions in the following decade. A 2017 replication with 145 college students in the Philippines used an essay about a reproductive health law. In the no-choice condition, readers still judged the writer to favour the side the essay took.

Where the name comes from: Ross in 1977

Lee Ross introduced the phrase in a 1977 chapter, The intuitive psychologist and his shortcomings. In Gilbert and Malone's account, Ross made two points. First, without the bias many classic experiments would be dull. Studies such as Milgram's drew attention because people underestimate how strongly a social situation can push, and so are surprised by the results. Our guide to Milgram's obedience experiment shows how far the setup moved the obedience rate. Second, the same underestimation was shared by the participants in those experiments, which made it a key to behaviour across many seemingly unrelated studies.

Ross also co-authored a vivid demonstration that year. In two experiments, one participant wrote difficult general-knowledge questions and another tried to answer them. Questioners were consistently rated as more knowledgeable than their partners, although the role let them pick questions on subjects they knew well. The bias was stronger among answerers and uninvolved observers than among questioners.

Jones himself preferred another name. In a 1988 passage quoted by Gilbert, he wrote:

Gilbert adds that Jones renamed his observer bias the correspondence bias.

Fundamental attribution error vs correspondence bias vs actor-observer bias

These terms are easy to mix up. The table separates what each one claims.

Fundamental attribution error and correspondence bias are the pair most often used as synonyms, and the APA Dictionary treats them that way. Gilbert and Malone define the correspondence bias as the tendency to draw inferences about a person's unique and enduring dispositions from behaviours that can be entirely explained by the situations in which they occur.

Bertram Gawronski (2004) argues for keeping them apart. In his analysis the correspondence bias is the pattern: perceivers draw dispositional inferences from constrained behaviour. The fundamental attribution error is the most common explanation of the pattern: perceivers assume that situations have little influence on behaviour. His review finds no evidence for that explanation and concludes that several different processes produce the pattern. If a course asks you to distinguish them, use correspondence bias for what is observed and the fundamental attribution error for the proposed cause, and note that many authors use them interchangeably.

The actor-observer asymmetry is a different claim. Introduced by Jones and Nisbett in 1971, it says that people explain their own behaviour with situational causes and other people's behaviour with personal causes, as Malle's 2006 meta-analysis puts it. The fundamental attribution error, as usually defined, is about how observers judge other people. A definition that adds "while explaining their own behaviour by the situation" has folded the actor-observer idea into it.

The ultimate attribution error carries the idea to groups. Thomas Pettigrew proposed it in 1979 as part of an analysis of prejudice. Our guide to social identity theory covers the ingroup and outgroup vocabulary.

Two neighbouring ideas concern the self rather than observers. Attributional style is a habit of explaining one's own setbacks, and locus of control is an expectation about whether outcomes depend on your own actions. Our locus of control guide compares them.

Why the fundamental attribution error happens

Research does not point to a single cause. Gilbert and Malone's review describes four mechanisms that produce different forms of the bias.

Four sources of the bias

• Lack of awareness. Observers cannot weigh a situation they do not know about, and many situational forces leave no visible trace.
• Unrealistic expectations. An observer who knows the situation may still misjudge how strongly such a situation usually changes what people do.
• Inflated categorisations. Knowing the situation can change how the behaviour is perceived, so that it seems more extreme than it was and the excess is credited to the person.
• Incomplete corrections. Observers seem to draw the dispositional inference first and then adjust for the situation, and the adjustment is the fragile step.

Why mental effort matters

Gilbert, Pelham and Krull (1988) tested the idea that the correction takes effort. In Gilbert and Malone's description, observers watched a video of a woman who seemed nervous while talking to a stranger, with subtitles saying what topic she was discussing. Observers who were not busy rated her as less anxious by nature when the topic was anxiety-provoking. Observers who were busy rehearsing word strings rated her as anxious by nature whatever the topic.

One design does not settle the matter, but it fits the claim that the first inference is cheap and the correction suffers when attention is taken.

Culture shifts the size of the effect

Joan Miller's 1984 study compared Indian and American adults and children. At older ages, the Americans referred more to general dispositions and less to contextual factors than the Hindu participants did, and references to dispositions grew much more with age among the Americans. Miller traced the differences to contrasting cultural conceptions of the person.

The effect does not vanish elsewhere. The Philippine replication above, in a mainly collectivist society, still found readers inferring the writer's attitude from the essay. Treat culture as a difference of degree, not a clean divide.

Examples of the fundamental attribution error in everyday life

These are illustrations, not studies. The trait explanation may even be right. The error is accepting it without checking the situation.

• Lateness. A colleague arrives late and is filed as careless. The delayed train is invisible to everyone in the room.
• Driving. A driver cuts across your lane and becomes "reckless". Only the car is in view, not whatever caused the swerve, such as a missed exit.
• Exams. A classmate fails a test and the verdict is "didn't try". The two night shifts that week never reach the conversation.

Is the fundamental attribution error really an error?

The observation behind the term has held up. In his 1998 chapter, Gilbert judged that most researchers probably agreed the correspondence bias is a fairly robust and general phenomenon. Three points of criticism concern what the observation means.

The actor-observer half does not hold on average. Malle's meta-analysis of 173 published studies found average effect sizes from d = −0.016 to d = 0.095, close to zero. The asymmetry held only under some conditions, such as hypothetical events or actor and observer being intimates. It held for negative events, with a reverse asymmetry for positive events, so Malle concluded that across valence no actor-observer asymmetry exists. That is a finding about comparing self and other, not about the observer-side pattern in the Castro studies.

The label "error" depends on the benchmark. Walker, Smith and Vul argue that classic studies were judged against a deterministic standard: if a situation fully dictates behaviour, the behaviour tells you nothing about the person. Most situations do not fully dictate behaviour. They show that the results of Jones and Harris, and of a 1982 study by Quattrone, are consistent with an ideal Bayesian observer who treats situations as probabilistic influences. This is one modelling argument, not a consensus.

The usual explanation is contested. As covered above, Gawronski found no evidence that the pattern is caused by a belief that situations have little impact.

So the pattern is well supported, while its explanation, its label and the self-versus-other half of popular definitions are debated.

Learning the terms for an exam

Fundamental attribution error, correspondence bias and actor-observer bias are easy to swap when you learn them one at a time. They stick better when you learn them against each other, as our guide to memorising psychology terms explains. In Psychology World you can save "fundamental attribution error", "correspondence bias" and "actor-observer bias" to a Social psychology list and drill them against each other.