Now that we have a sense of the role of memory in how we understand people and events, we can move on to a second core process we use to gain understanding. This is the process by which we look for relationships of cause and effect. In our fender-bender example, the investigator involved is charged not only with obtaining a (let’s hope accurate) record of what happened but also with identifying what factors caused the outcome. In our own interpretation of events, we similarly seek to pair effects with their causes, so that we will know how to act and better predict outcomes in the future. When you see that your friend is really upset, you want to know what caused those feelings so that you can effectively console your friend, maybe help fix the problem, and know how to avoid the situation in the future. Fritz Heider pioneered this line of inquiry beginning back in the 1930s. At that time, the two dominant views of what causes humans to behave the way we do— psychoanalysis and behaviorism—placed little emphasis on our conscious thoughts. But Heider argued that to understand why people behave the way they do, we have to examine how they come to comprehend their social surroundings. To this end, Heider (1958) developed a common sense, or naive, psychology: an analysis of how ordinary people like you and me think about the people and events in our lives.

Working from a Gestalt perspective, Heider assumed that the same kinds of rules that influence the organization of visual sensations also guide most people’s impressions of other people and social situations. In one early study (Heider & Simmel, 1944), people watched a rather primitive animated film in which a disk, a small triangle, and a larger triangle moved in and out of a larger square with an opening. The participants were then asked to describe what they saw (FIGURE 4.3). People tended to depict the actions of the geometric objects in terms of causes, effects, and intentions, for example, “The larger triangle chased the smaller triangle out of the room [the larger square].”

This tendency led Heider to propose that people organize their perceptions of action in the social world in terms of causes and effects. He subsequently listened to how people talked about their social lives in ordinary conversation and obtained further support for this proposition. Specifically, people tend to explain events in terms of particular causes. Heider referred to such explanations as causal attributions.

126

Because causal attributions help people make sense of, and find meaning in, their social worlds, they are of great importance. When someone is bumped into in a bar, whether he attributes the bumping to malicious intent or mere clumsiness can be a matter of life or death. When an employee is late, whether the employer attributes that behavior to the person’s laziness or to her tough circumstances can determine whether she is fired or not. After a car accident, determining who was responsible can have profound financial consequences for the parties involved. If a woman shoots her abusive husband, a jury may have to decide if she did it because she feared for her life or because she wanted to collect on his life insurance. Whenever the economy is shaky, considerable debate arises over which political party’s policies are most responsible for the current economic gloom. In pretty much every domain of life, from bar jostling to jurors’ decision making to whom one votes for on election day, causal attributions play a significant role.

Basic Dimensions of Causal Attribution

Heider (1958) observed that causal attributions vary on two basic dimensions. The first dimension is locus of causality, which can be either internal to some aspect of the person engaging in the action (known as the actor), or external to some factor in the person’s environment (the situation). For example, if Justin failed his physics exam, you could attribute his poor performance to a lack of intelligence or effort, factors internal to Justin. Or you could attribute Justin’s failure to external factors, such as a lousy physics professor or an unfair exam.

The second basic dimension is stability: attributing behavior to either stable or unstable factors. If you attribute Justin’s failure to a lack of intelligence, that’s a stable internal attribution, because people generally view intelligence as relatively immutable. On the other hand, if you attribute Justin’s failure to a lack of effort, that is an unstable internal attribution. You would still perceive Justin as responsible for the failure, but you would recognize that his exertion of effort can vary from situation to situation. Stable attributions suggest that future outcomes in similar situations, such as the next physics test, are likely to be similar. In contrast, unstable attributions suggest that future outcomes could be quite different; if Justin failed because of a lack of effort, he might do a lot better on the next test if he exerted himself a bit more.

External attributions can also be stable or unstable. If you attribute Justin’s failure to a professor who always gives brutal tests or who is an incompetent teacher, you are likely to think that Justin will not do much better on the next test. But if you attribute Justin’s failure to external unstable factors such as bad luck or his girlfriend’s having broken up with him right before the test, then you’re more likely to think Justin may improve on the next exam.

As poor Justin’s example suggests, how we attribute a behavior—to internal or external, stable or unstable factors—affects both the impressions we form of the actor and the predictions we make about the actor’s future behavior. An internal attribution for a poor performance or a negative action reflects poorly on the actor, whereas an external attribution tends to let the actor off the hook. On the other hand, an internal attribution for a positive behavior generally leads to a positive impression of the actor, whereas an external attribution for a positive action undermines the benefit to the actor’s image. Attributions to stable factors lead to strong expectations of similar behavior in similar situations, whereas attributions to unstable factors do not.

127

Entity and Incremental Theorists

Inspired by her early work on achievement, Dweck and colleagues (Dweck, 2012; Hong et al., 1995) proposed that intelligence and other attributes need not be viewed as stable entities. Rather, they could be viewed as attributes that change incrementally over time. Dweck says that when we take the perspective of an entity theorist, we view an attribute as a fixed trait that a person can’t control or change. For example, we may see an attribute such as intelligence as a stable and enduring quality. But when we take the perspective of an incremental theorist, we believe an attribute is a malleable ability that can increase or decrease. For example, we may see an attribute such as shyness as a quality that, with the right motivation and effort, people can change (Beer, 2002). Take a moment to think about which attributes you think are fixed entities and beyond your control, and which attributes you think are changeable.

These theories have implications for how an individual interacts with, and responds to, the social world. Dweck and colleagues find that entity theorists, both children and adults, make more negative stable attributions about themselves in response to challenging tasks, and then tend to perform worse and experience more negative affect in response to such tasks. Moreover, they tend to eschew opportunities to change that ability even when the ability is crucial to their success. For instance, one study measured peoples’ theories about achievement and intelligence. Exchange students who had stronger entity theories expressed less interest in remedial English courses when their English was poor and improving would facilitate their academic goals (Hong et al., 1999). In contrast, incremental theorists view situations that implicate that ability more as opportunities to improve, to develop their skills and knowledge. This affects not just what we do for ourselves, but also how we treat others. When business managers had stronger incremental theories, they were more willing to provide mentorship to their employees and thus help their employees to improve (Heslin & Vanderwalle, 2008).

Although people generally have dispositional tendencies to hold either entity or incremental theories about human attributes, these views can be changed (Dweck, 2012; Heslin & Vanderwalle, 2008; Kray & Haselhuhn, 2007). For example, research has shown that convincing students that intelligence is an incremental attribute rather than a stable entity encourages them to be more persistent in response to failure, adopt more learning-oriented goals, and make fewer ability attributions for failure (Bergen, 1991; Burnette et al., 2012; Dweck, 2012).

128

How do people arrive at a particular attribution for a behavior they observe? Like most products of human cognition, casual attributions sometimes result from quick, intuitive, automatic processes, and sometimes from more rational, elaborate, thoughtful processes.

Whenever you ask people why some social event occurred, they can usually give an opinion. However, research has shown that people often don’t put much effort into thinking about causal attributions. People make a concentrated effort primarily when they encounter an event that is unexpected or important to them (Jaynes, 1976; Pyszczynski & Greenberg, 1981; Wong & Weiner, 1981). Such events are more likely to require some action on our part and are more likely to have a significant impact on our own lives, so it is more important to arrive at an accurate causal attribution.

Let’s imagine that when you were a kid, your mom always had coffee in the morning, and you came into the kitchen one morning and saw your mom put on a pot of coffee. If a friend dropped by and asked you why your mom was doing that, you would readily respond, “She always does that” or “She loves coffee in the morning”—the same knowledge that led you to expect her to do exactly what she did. However, if one morning she was brewing a pot of herbal tea, this would be unexpected, and you’d likely wonder why she was doing that instead of brewing her usual coffee. You would be even more likely to think hard about a causal attribution for her behavior if one morning she was making herself a martini.

129

Most events in our daily lives are expected; consequently, we don’t engage in an elaborate process to determine a causal attribution. According to Harold Kelley (1973), when events readily fit existing causal schemas, we rely on them rather than engage in much thought about why the events occurred. These causal schemas come from two primary sources. Some are based on our own personal experience, as in the mom-making-coffee example. Others are based on general cultural knowledge. If an American watches another person in a restroom passing a thin, waxed piece of thread between her teeth, little or no thought about why is generated, because it is a culturally normative accepted form of hygiene known as flossing. But consider our discussion in chapter 2 of cultural differences and think about how people garner prestige and self-worth from the cultural perspective. In the same way we might think little of someone flossing in a restroom, a Trobriand Islander wouldn’t blink an eye at a man building up a large pile of yams in front of his sister’s house and leaving them to rot. In that culture, this behavior is a way that people enhance their status and would not require any explanation. But if the Trobriander saw someone flossing, or if the American saw someone “yamming,” then a more elaborate process of determining an attribution would almost certainly ensue.

The Fundamental Attribution Error

When an event we observe isn’t particularly unexpected or important to us but doesn’t readily fit an obvious causal schema, we are likely to base our causal attribution on whatever plausible factor is either highly visually salient or highly accessible from memory. This “top of the head phenomenon” was illustrated in a set of studies by Shelley Taylor and Susan Fiske (Taylor & Fiske, 1975; Fiske & Taylor, 1991) in which participants heard a group discussion at a table. One particular member of each group was made visually salient. One way they accomplished this was by having only one member of the group be of a particular race or gender. Participants were asked how much each individual was responsible for the direction of the discussion. For instance, when the group included only one woman, she was viewed as most causally responsible for the discussion. Similar effects have been found if salience is achieved by having the person sitting at the apparent head of a table or if lighting is specifically focused on the person (McArthur & Post, 1977).

This reliance on visual salience was anticipated by Heider, who proposed that people are likely to attribute behavior to internal qualities of the person, because when a person engages in an action, that actor tends to be the observer’s salient focus of attention. Edward Jones and Keith Davis (1965) carried this notion further by proposing that when people observe an action, they have a strong tendency to make a correspondent inference, meaning that they attribute an attitude, desire, or trait that corresponds to the action to the person. For example, if you watch Ciara pick up books dropped by a fellow student leaving the library, you will automatically think of Ciara as helpful. These correspondent inferences are generally useful because they give us quick information about the person we are observing.

Correspondent inferences are most likely under three conditions (e.g., Jones, 1990). First, the individual seems to have a choice in taking an action. Second, a person has a choice between two courses of action and there is only one difference between one choice and the other. For example, if Sarah must choose between two colleges that are very similar except one is known to be more of a party school, and she chooses the party school, you may conclude Sarah is into partying. But you would be less likely to do so if the school she chose was more of a party school, but was also closer to her home, in a warmer climate, was known to be better in science, and was less expensive. Third, a correspondent inference is more likely when someone has a particular social role and acts inconsistent with that role. If a contestant in a game show such as The Price is Right wins a car but barely cracks a smile and simply says “Thank you,” you would be likely to infer she is not an emotionally expressive person. But if she jumps up and down excitedly after she wins the new car, you would not be as certain what she is like, because most people in that role would be similarly exuberant.

Although all three of these factors increase the likelihood of a correspondent inference, this tendency is so strong that we often jump to these correspondent inferences without sufficiently considering external, situational factors that may also have contributed to the behavior witnessed (e.g., Jones & Harris, 1967). Research has found that people’s tendency to attribute behavior to internal or dispositional qualities of the actor, and consequently underestimate the causal role of situational factors, is so pervasive that it is known as the fundamental attribution error, or FAE for short. The initial demonstration of the FAE was provided by Ned Jones and Victor Harris (1967) (FIGURE 4.4). Participants read an essay that was either strongly in favor of Fidel Castro (the longtime dictator of Cuba) or strongly against Castro. Half the participants were told that the essay writer chose his position on the essay. When asked what they thought the essay writer’s true attitude toward Castro was, participants, not surprisingly, judged the true attitude as pro-Castro when the essay was pro-Castro and anti-Castro when the essay was anti-Castro. However, the other half of the participants were told that the writer didn’t have a choice in whether to advocate for or against Castro; instead, the experimenter had assigned what side the writer should take. Logic would suggest that the lack of choice would make the position advocated by the essay a poor basis for guessing the author’s true attitude. However, these participants, despite knowing the essay writer had no choice, also rated his attitudes as corresponding to the position he took in the essay.

130

Many experiments have since similarly shown that despite reasons for attributing behavior largely, if not entirely, to situational factors, people tend to make dispositional attributions instead. One of the most compelling demonstrations was a study by Lee Ross and colleagues (1977), inspired by early precursors, which date back to the 1950s, to the present plethora of reality TV shows: quiz shows. Participants were randomly assigned the role of questioner, contestant, or observer, which ensured that the groups of students assigned to each role were equally knowledgeable. The questioners were asked to generate difficult questions that the contestants then had to answer. It is pretty obvious that coming up with tough questions from your own store of knowledge is a lot easier than answering difficult questions made up by someone else. Yet, both the observers and the contestants themselves concluded that the questioners were more knowledgeable than the contestants. Clearly these participants did not take sufficiently into account the influence of the situation—questioner and contestant roles—in making these judgments.

A compelling example of the FAE from everyday life is the tendency to think that actors are like their characters. Actors who play evil characters on soap operas have even been verbally abused in public! Leonard Nimoy, who played Mr. Spock in the original Star Trek television show and in subsequent movies, got so tired of being viewed as Spock when the series was at the height of its popularity that he wrote a book called I Am Not Spock. Similarly, slapstick comics such as Will Farrell have expressed frustration that people expect them to be wacky loons in real life. On the one hand, these errors make sense because we know these people only as their fictional characters. On the other hand, they are great examples of the FAE because we know that most of the time actors are saying lines written for them by someone else and are being directed with regard to their appearance, movements, and nonverbal behaviors.

The FAE has implications for how people judge others—for example, defendants in court—and for how they judge social issues pertinent to individuals and groups. If people are likely to make internal attributions to drug addicts, homeless people, and welfare recipients, they are probably more supportive of treating such people harshly and less likely to entertain ways to change environmental factors that contribute to these problems. Linda Skitka and colleagues (2002) found that, consistent with this reasoning, American political conservatives seem to be more susceptible to the FAE than liberals and are therefore less sympathetic to those low on the socioeconomic ladder. Likewise, conservatives judging wealthy people are led by the FAE toward attributions to those people’s abilities and initiative, rather than to their trust funds, connections, or lucky breaks.

How Fundamental Is the FAE?

Although the FAE is common when people make attributions for the behavior of others, this is not the case when making attributions for oneself. And if you think about visual salience, you should be able to (literally) see why. When we are observers, the other person is a salient part of our visual field, but when we ourselves are acting in the world, we are usually focused on our surroundings rather than on ourselves. This leads to what has been labeled the actor-observer effect: As observers we are likely to make internal attributions for the behavior of others, but as actors we are likely to make external attributions for our own behavior (Jones & Nisbett, 1971). When observing others, we attend primarily to them and not to their situation. In contrast, when acting ourselves, we are usually reacting to someone or something in our environment. When that is the case, such external factors are likely to be more salient.

Nisbett and colleagues (1973) asked people either why their roommates had chosen their majors or why they had chosen their own. Participants made internal attributions for the roommate’s choice: “Chris chose psychology because she loves analyzing people.” But when explaining their own choice, they emphasized attributes of the major: “Criminal justice is a fascinating field, and it allows me to consider a wide range of fields: police work, the FBI, law school, or teaching.” It is interesting that the actor-observer effect can be reversed by shifting the individual’s visual perspective. For instance, Storms (1973) replicated the actor-observer effect by showing that when pairs of participants sat across from one another and had a conversation, they generally thought that their partners were determining the things they talked about. They attributed the direction that the conversation took to the person they could see—their partner. Then he demonstrated that if shown a video playback of the conversation from the discussion partner’s perspective (now the actor was watching him- or herself talk!), participants were more likely to think that they were the ones steering the direction of the conversation.

131

The actor-observer effect has clear implications for interpersonal and intergroup relations. It suggests that when things are going badly in a relationship, whether it is a friendship, a marriage or an alliance between two countries, each actor is likely to view the external situation that is salient as responsible for the problems. This means that the friend, the marriage partner, or the other country is seen as the cause of the problem. And this, of course, can create and intensify finger-pointing and hostility between individuals and groups. Storms’s work suggests that one way to combat this cognitively based conflict intensifier is to make the other party’s perspective on the issues salient. This might defuse tensions by fostering greater consensus among different people regarding the causes of their problems.

There are, however, important qualifications to the actor-observer effect. For one, we are much more likely to make internal attributions for our successes (“I aced the test because I studied really hard”) but external attributions for our failures (“I bombed the test because the instructor is so awful he could not teach someone to tie their shoe”) (Campbell & Sedikides, 1999). We will discuss this tendency at greater length when we talk about self-esteem biases in chapter 6. Because of this tendency, the actor-observer difference is stronger for negative behaviors (Malle, 2006). In addition, research indicates that classic actor-observer asymmetries are even stronger in what we construe as unintentional behavior (Malle et al., 2007). When people have a strong intention for their behavior, they may be just as likely to make an internal attribution for their own actions as for someone else’s.

Does the FAE Occur Across Cultures?

Some social psychologists have proposed that the FAE is a product of individualistic cultural worldviews, which emphasize personality traits and view individuals as fully responsible for their own actions (Watson, 1982; Weary et al., 1980). In fact, the original evidence of the FAE was gathered in the United States and other relatively individualistic cultures. And it does seem clear, as we noted in chapter 2, that people in more collectivistic cultures are more attentive to the situational context in which behavior occurs. They also are more likely to view people as more influenced by, and indeed part of, the larger groups to which they belong. In fact, when directed to explain behavior, rather than focusing on an individual, people from more collectivistic cultures (e.g., China) generally give more relative weight to external, situational factors than do people from more individualistic cultures (Choi & Nisbett, 1998; Miller, 1984; Morris & Peng, 1994). Such research suggests that socialization in a collectivist culture might sensitize people to contextual explanations of behavior. More recent research also has suggested a weaker tendency to commit the fundamental attribution error among those of lower socioeconomic status (Varnum et al., 2011) and among Catholics compared with Protestants (Li et al., 2012).

Evidence of cross-cultural variation in attributional biases doesn’t necessarily mean that individuals raised in collectivist settings don’t form impressions of people’s traits and personalities by observing their behavior. Douglas Krull and colleagues (1999) found that when participants are asked to judge another person’s attitudes or traits on the basis of an observed behavior, people from collectivist cultures such as China are just as susceptible to the FAE as people from more individualistic cultures such as the United States. Around the globe, when the goal is to judge a person, we judge people by their behavior. But when the goal is to judge the cause of a behavior, there is cultural variation in how much the behavior is attributed to the person or the situation.

132

Although the FAE is very common and sometimes automatic, even people from individualistic cultures often take situational factors into account when explaining behavior. In the original Jones and Harris (1967) study, for example, participants who knew that the essay writer had no choice were less confident that the essay represented the writer’s true opinions. When might people consider these situational factors before jumping to an internal attribution?

To answer this question, Dan Gilbert and colleagues (1988) proposed a model in which the attribution process occurs in a temporal sequence of three stages.

This model predicts that people will be especially likely to ignore situational factors and to make the FAE when they are cognitively challenged and thus have limited attention and energy to devote to attributional processing.

Putting this model to the test, Gilbert and colleagues (1988) had participants watch a videotape of a very fidgety woman discussing various topics. The participants were asked to rate how anxious a person she generally was. The videotape was silent, ostensibly to protect the woman’s privacy, but participants were shown one- or two-word subtitles indicating the topic she was currently discussing. The videotape was always the same, but the subtitles indicated either very relaxing topics such as vacation, travel, and fashion, or very anxiety-provoking topics such as sexual fantasies, personal failures, and secrets.

If observers have sufficient resources, they should initially jump to an internal attribution for the fidgety behavior and view the woman as anxious, but in the condition in which the topics are anxiety provoking, they should make a correction and view the woman as less anxious. Indeed, as FIGURE 4.5 shows, this is what happened under normal conditions. But half the observers were given a second, cognitively taxing task to do while viewing the videotape. Specifically, they were asked to memorize the words in the subtitles displayed at the bottom of the screen. Under such a high cognitive load, these participants lacked the resources to correct for the situational factor (the embarrassing topics) and therefore judged the anxious-looking woman to be just as anxious if she was discussing sex and secrets as if she was discussing travel and fashion.

133

This three-stage model of attribution also helps us to understand how individual differences in the motivation to focus on possible situational causes can influence the kind of attributions people make. For example, Skitka and colleagues (2002) found that liberals generally are less likely than conservatives to view an AIDS victim as an irresponsible person. However, when cognitively busy, the liberal viewed the AIDS patient just as irresponsible as conservatives did. So when people have the motivation and resources, they often correct their initial leap to internal dispositional attributions, but when cognitively busy, they generally are not able make this correction.

So far, we have discussed primarily the relatively quick, automatic ways people arrive at causal attributions: relying on salience, jumping to a correspondent inference, and then perhaps making a correction. But what happens when we encounter very unexpected or important events? As we noted earlier, when people are sufficiently surprised or care enough, they put effort into gathering information and thinking carefully before making a causal attribution. Imagine a young woman eager about a first date with a young man she met in class and really liked; they were going to meet at a restaurant at 7:00 p.m. But he’s not there at 7:00, or 7:05, or 7:15. She would undoubtedly begin considering casual explanations for this unexpected, unwelcome event: Did he change his mind? Was he in a car accident? Is he always late?

Causal Hypothesis Testing

Putting conscious effort into making an attribution is like testing a hypothesis (Kruglanski, 1980; Pyszczynski & Greenberg, 1987; Trope & Liberman, 1996). First, we generate a possible causal hypothesis (a possible explanation for the cause of the event). This could be the interpretation we’d prefer to make, the one we fear the most, or one based on a causal schema (a theory we hold about the likely cause of that specific kind of event). A large paw print seems more likely caused by a bear than by a rabbit (Duncker, 1945). The causal hypothesis could also be based on a salient aspect of the event or a factor that is easily accessible from memory. Finally, causal accounts can also be based on close temporal and spatial proximity of a factor to the event, particularly if that co-occurrence happens repeatedly (Einhorn & Hogarth, 1986; Kelley, 1973; Michotte, 1963). If Frank arrives late to a party and soon after a fight breaks out, it is likely you’ll entertain the hypothesis that Frank caused the melee, especially if the co-occurrence of Frank’s arrival and fights happens repeatedly. The tendency of the co-occurrence of a potential causal factor (Frank’s arrival) and an outcome (a fight) to lead to a causal hypothesis (Frank causes fights) is called the covariation principle (Kelley, 1973).

Once we have an initial causal hypothesis, we gather information to assess its plausibility. How much effort we give to assessing the validity of that information depends on our need for closure versus our need for accuracy. If the need for closure is high and the need for accuracy is low, and if the initial information seems sound and fits the hypothesis, we may discount other possibilities. But if the need for accuracy is higher than the need for closure, we may carefully consider competing causal attributions, and then decide on the one that seems to best fit the information at our disposal. What kinds of information would we be most likely to use to inform our causal attributions?

134

Three Kinds of Information: Consistency, Distinctiveness, and Consensus

Harold Kelley (1967) described three sources of information for arriving at a causal attribution when accuracy is important: consistency (across time), distinctiveness (across situations), and consensus (across people) (FIGURE 4.6).

Imagine that a few years from now, Reese, an acquaintance of yours, goes to see Sylvester Stallone’s anxiously anticipated new film Rocky 9. (New regulations in boxing have made it legal to bring a walker into the ring.) Reese tells you, “You have to see this flick; it’s an unprecedented cinematic achievement.” By this time, it costs $30.00 to see a film, so you want to think hard before shelling out that kind of cash. Ultimately, you want to know if Rocky 9 is indeed a must-see. Or was it something about Reese, or the particular circumstances when she saw the movie? In other words, why did Reese love the movie?

Maybe there was a special circumstance when Reese saw the movie that led her to love it. So if she saw it a second time and didn’t like it much, Kelley would label the outcome low in consistency, and you would probably conclude there was something unique about the first time she saw it that caused her reaction. Maybe she is ultrasensitive to variations in popcorn flavoring, and there was a delectable extra dash of butter and salt on her popcorn. This would be an attribution to an unstable factor that might be very different each time Reese saw the movie. However, if Reese saw the movie three times and loved it each time, Kelley would label the outcome high in consistency across time, and you would then be likely to entertain either a stable internal attribution to something about Reese, or a stable external attribution to something about the movie. Two additional kinds of information would help you decide.

135

First, you could consider Reese’s typical reaction to other movies. If Reese always loves movies in general, or always loves boxing movies or Sly Stallone movies, then her reaction to Rocky 9 is low in distinctiveness. This information would lead you to an internal attribution to something about Reese’s taste in these kinds of movies rather than the quality of the movie. On the other hand, if Reese rarely likes movies, or rarely likes boxing or Stallone movies, then her reaction is high in distinctiveness, which would lead you more toward an external attribution to the movie.

Finally, how did other people react to the movie? If most others also loved the movie, there is high consensus, suggesting that something about the movie was responsible for Reese’s reaction. However, if most other people didn’t like the movie, you would be most likely to attribute Reese’s reaction to something about her. In short, when a behavior is high in consistency, distinctiveness, and consensus, the attribution tends to be external, whereas when a behavior is high in consistency but low in distinctiveness and consensus, an internal attribution is more likely. Research has generally supported Kelley’s model, although people often don’t seek all three types of information, and they more reliably use distinctiveness information than consensus information (e.g., McArthur, 1972).

The presence of other potential causes also can influence the weight that we assign to a particular causal factor. For example, not too long ago, most of the baseball-watching world was in awe of players such as Barry Bonds, Mark McGwire, and Alex Rodriguez, attributing their remarkable accomplishments to their incredible skills and effort. However, when allegations of steroid use surfaced, many fans’ discounted the players’ skills, attributing some of their accomplishments to an alternative causal factor: performance-enhancing drugs. The same process applied to Lance Armstrong, as we discussed in chapter 3. This tendency is called the discounting principle, whereby the importance of any potential cause of another’s behavior is reduced to the extent that other potential causes exist (Kelley, 1971; Kruglanski et al., 1978).

136

Motivational Bias in Attribution

What makes causal attributions so intriguing to social psychologists—and we hope to you as well—is that they are both important and ambiguous. They are important because they play such a large role in the judgments and decisions we make about other people and about ourselves. They are ambiguous because we can’t really see or measure causality; our causal attributions are based on guesswork because we rarely if ever have direct evidence that proves what caused a given behavior.

And if we did, it would probably tell us that the behavior was caused by a complex interaction of internal and external factors. For instance, any performance on a test is likely to be determined by a combination of the individual’s inherent aptitude, physical and mental health, amount of studying prior to the test, and other events in the person’s life commanding attention, as well as the particular questions on the test, the amount of time allotted for each question, and the quality of instruction in the course. And yet we typically either jump on the most salient attribution, or do a little thinking and information gathering, and then pick a single attribution, or at most two contributing factors, discounting other possibilities (Kelley, 1967).

137

As Heider (1958) observed, because causal attributions are often derived from complex and ambiguous circumstances, there is plenty of leeway for them to be influenced by motivations other than a desire for an accurate depiction of causality. Often the intention is to preserve positive images of those we like and negative views of those we dislike. We are likely to attribute a friend’s success to the friend’s good qualities; if an enemy does well, we’re likely to attribute it to dumb luck or cheating. Similarly, people generally believe that their country’s foreign policies are motivated by noble goals such as spreading freedom and prosperity rather than ignoble goals such as greed and revenge.

If we just want closure, we grab the simplest, most readily available attribution. One explanation of the FAE is that we want to attribute people’s behavior to stable traits so we can feel that we’ll be able to predict their behavior in the future (Gilbert & Malone, 1995). In fact, people attribute the behavior of another person more to that individual’s personality than to some other cause if they expect to have to interact with that person in the future (e.g., Miller et al., 1978). If we can believe we know what others are like, we will be more confident that we can get what we want from them.

Our attributions are also biased by our preferred views of the way the world works—our desire to maintain specific beliefs. Recall that people generally prefer to believe that the world is just, that good things happen to good people, and that bad things happen to bad people (Lerner, 1980). One way we preserve this belief is to view people as responsible for the outcomes they get. When people’s motivations are particularly high to believe the world is just, they are especially likely to blame people who have had bad things happen to them. This has been shown regarding people who have contracted STDs, rape victims, battered spouses, and poor people (e.g., Borgida & Brekke, 1985; Furnham & Gunter, 1984; Hafer & Begue, 2005; Summers & Feldman, 1984).

138