Actor–observer asymmetry (also actor–observer bias or actor–observer difference) is a bias one exhibits when forming attributions about the behavior of others or themselves. When explaining their own behavior, people are more likely to attribute their actions to the particular situation rather than their personality, also known as a situational attribution. However, when an observer is explaining the behavior of another person, they are more likely to attribute this behavior to the actors' personality rather than situational factors, also known as dispositional attribution. For example, a politician explaining why they voted against war may say it is because war is not needed, a situational factor. On the other hand, a person judging why the politician voted in this way may say it is because the politician is too liberal, a personality trait.
Sometimes, the actor–observer asymmetry is defined as the fundamental attribution error, defined as when people tend to explain behavior using internal, personal characteristics rather than the external factors or situational influences.However, Malle (2006) highlights that these two phenomena should be distinguished because the fundamental attribution error refers to inferring stable internal traits from behaviour, whereas actor-observer asymmetry specifically refers to explanations of behaviour.
Actor-observer asymmetry is often explained using perspectives and salience. When forming attributions, perspective highlights the situation, and what is occurring around the perceiver is most salient. As a result, perceivers may be more likely to make attributions based on these salient situational factors. However, when judging someone else, their behaviour is more salient than the situation. This may explain the greater chance of making dispositional attributions. Furthermore, when making judgements on one's own behaviour, much more information regarding the self is available, including knowledge of past behaviour. On the other hand, when judging others' behaviour, much less information is available. This lack of quality information likely also contributes to differences in attributions made.
The specific hypothesis of an actor–observer asymmetry in attribution was originally proposed by Edward Jones and Richard Nisbett, who stated that "actors tend to attribute the causes of their behavior to stimuli inherent in the situation, while observers tend to attribute behavior to stable dispositions of the actor". Supported by initial evidence, the hypothesis was long held as firmly established. However, a meta-analysis of all the published tests of the hypothesis between 1971 and 2004 found that there was no actor–observer asymmetry of the sort that had been previously proposed. The author of the study interpreted this result not so much as proof that actors and observers explained behavior exactly the same way but as evidence that the original hypothesis was fundamentally flawed in the way it framed people's explanations of behavior as either stable dispositional attributions or situational attributions.
Considerations of actor–observer differences can be found in other disciplines as well, such as philosophy (e.g. privileged access, incorrigibility), management studies, artificial intelligence, semiotics, anthropology, and political science.