Farnam Street helps you make better decisions, innovate, and avoid stupidity.

With over 400,000 monthly readers and more than 93,000 subscribers to our popular weekly digest, we've become an online intellectual hub.

The Fundamental Attribution Error, or Why Predicting Behavior is So Hard

“Psychologists refer to the inappropriate use of dispositional explanation as
the fundamental attribution error, that is, explaining situation-induced behavior
as caused by enduring character traits of the agent.”
— Jon Elster


The problem with any concept of “character” driving behavior is that “character” is pretty hard to pin down. We call someone “moral” or “honest,” we call them “courageous” or “naive” or any other number of names. The implicit connotation is that someone “honest” in one area will be “honest” in most others, or someone “moral” in one situation is going to be “moral” elsewhere.

Old-time folk psychology supports the notion, of course. As Jon Elster points out in his wonderful book Explaining Social Behavior, folk wisdom would have us believe that much of this “predicting and understanding behavior” thing is pretty darn easy! Simply ascertain character, and use that as a basis to predict or explain action.

People are often assumed to have personality traits (introvert, timid, etc.) as well as virtues (honesty, courage, etc.) or vices (the seven deadly sins, etc.). In folk psychology, these features are assumed to be stable over time and across situations. Proverbs in all languages testify to this assumption. “Who tells one lie will tell a hundred.” “Who lies also steals.” “Who steals an egg will steal an ox.” “Who keeps faith in small matters, does so in large ones.” “Who is caught red-handed once will always be distrusted.” If folk psychology is right, predicting and explaining behavior should be easy.

A single action will reveal the underlying trait or disposition and allow us to predict behavior on an indefinite number of other occasions when the disposition could manifest itself. The procedure is not tautological, as it would be if we took cheating on an exam as evidence of dishonesty and then used the trait of dishonesty to explain the cheating. Instead, it amounts to using cheating on an exam as evidence for a trait (dishonesty) that will also cause the person to be unfaithful to a spouse. If one accepts the more extreme folk theory that all virtues go together, the cheating might also be used to predict cowardice in battle or excessive drinking. 

This is a very natural and tempting way to approach the understanding of people. We like to think of actions that “speak volumes” about others’ character, thus using that as a basis to predict or understand their behavior in other realms.

For example, let’s say you were interviewing a financial advisor. He shows up on time, in a nice suit, and buys lunch. He says all the right words. Will he handle your money correctly?

Almost all of us would be led to believe he would, reasoning that his sharp appearance, timeliness, and generosity point towards his “good character”.

But what the study of history shows us is that appearances are flawed, and behavior in one context often does not have correlation to behavior in other contexts. Judging character becomes complex when we appreciate the situational nature of our actions. The U.S. President Lyndon Johnson was an arrogant bully and a liar who stole an election when he was young. He also fought like hell to pass the Civil Rights Act, something almost no other politician could have done.

Henry Ford standardized and streamlined the modern automobile and made it affordable to the masses, while paying “better than fair” wages to his employees and generally treating them well and with respect, something many “Titans” of business had trouble with in his day. He was also a notorious anti-Semite! If it’s true that “He who is moral in one respect is also moral in all respects,” then what are we to make of this?

Jon Elster has some other wonderful examples coming from the world of music, regarding impulsivity versus discipline:

The jazz musician Charlie Parker was characterized by a doctor who knew him as “a man living from moment to moment. A man living for the pleasure principle, music, food, sex, drugs, kicks, his personality arrested at an infantile level.” Another great jazz musician, Django Reinhardt, had an even more extreme present-oriented attitude in his daily life, never saving any of his substantial earnings, but spending them on whims or on expensive cars, which he quickly proceeded to crash. In many ways he was the incarnation of the stereotype of “the Gypsy.” Yet you do not become a musician of the caliber of Parker and Reinhardt if you live in the moment in all respects. Proficiency takes years of utter dedication and concentration. In Reinhardt’s case, this was dramatically brought out when he damaged his left hand severely in a fire and retrained himself so that he could achieve more with two fingers than anyone else with four. If these two musicians had been impulsive and carefree across the board — if their “personality” had been consistently “infantile” — they could never have become such consummate artists.

Once we realize this truth, it seems obvious. We begin seeing it everywhere. Dan Ariely wrote a book about situational dishonesty and cheating which we have written about before. Judith Rich Harris based her theory of child development on the idea that children do not behave the same elsewhere as they do at home, misleading parents into thinking they were molding their children. Good interviewing and hiring is a notoriously difficult problem because we are consistently misled into thinking that what we learn in the interview process is representative of the interviewee’s general competence. Books have been written about the Halo Effect, a similar idea that good behavior in one area creates a “halo” around all behavior.

The reason we see this everywhere is because it’s how the world works!

This basic truth is called the Fundamental Attribution Error, the belief that behavior in one context carries over with any consistency into other areas.

Studying the error leads us to conclude that we have a natural tendency to:

A. Over-rate some general consideration of “character” and,
B. Under-rate the “power of the situation”, and its direct incentives, to compel a variety of behavior.

Elster describes a social psychology experiment that effectively demonstrates how quickly any thought of “morality” can be lost in the right situation:

In another experiment, theology students were told to prepare themselves to give a brief talk in a nearby building. One-half were told to build the talk around the Good Samaritan parable(!), whereas the others were given a more neutral topic. One group was told to hurry since the people in the other building were waiting for them, whereas another was told that they had plenty of time. On their way to the other building, subjects came upon a man slumping in the doorway, apparently in distress. Among the students who were told they were late, only 10 percent offered assistance; in the other group, 63 percent did so. The group that had been told to prepare a talk on the Good Samaritan was not more likely to behave as one. Nor was the behavior of the students correlated with answers to a questionnaire intended to measure whether their interest in religion was due to the desire for personal salvation or to a desire to help others. The situational factor — being hurried or not — had much greater explanatory power than any dispositional factor.

So with a direct incentive in front of them — not wanting to be late when people were waiting for them, which could cause shame — the idea of being a Good Samaritan was thrown right out the window! So much for good character.

What we need to appreciate is that, in the words of Elster, “Behavior is often no more stable than the situations that shape it.” A shy young boy on the playground might be the most outgoing and aggressive boy in his group of friends. A moral authority in the realm of a religious institution might well cheat on their taxes. A woman who treats her friends poorly might treat her family with reverence and care.

We can’t throw the baby out with the bathwater, of course. Elster refers to contingent response tendencies that would carry from situation to situation, but they tend to be specific rather than general. If we break down character into specific interactions between person and types of situations, we can understand things a little more accurately.

Instead of calling someone a “liar,” we might understand that they lie on their taxes but are honest with their spouse. Instead of calling someone a “hard worker,” we might come to understand that they drive hard in work situations, but simply cannot be bothered to work around the house. And so on. We should pay attention to the interplay between the situation, the incentives and the nature of the person, rather than just assuming that a broad  character trait applies in all situations.

This carries two corollaries:

A. As we learn to think more accurately, we get one step closer to understanding human nature as it really is. We can better understand the people with whom we coexist.

B. We might better understand ourselves! Imagine if you could be the rare individual whose positive traits truly did carry over into all, or at least all important, situations. You would be traveling an uncrowded road.


Want More? Check out our ever-growing database of mental models.