Broadening the Frame by Matt Carlson

E. The Fundamental Attribution Error

One reason we might suspect that the just-world results reflect intuition rather than considered evaluation is that they cohere with decades of research on attribution. In particular, victim-blaming in the just-world experiments is an instance of a more general phenomenon, the “fundamental attribution error,” a term coined by Lee Ross to describe the findings of a 1967 study of Edward E. Jones and Victor Harris. In one of Jones and Harris’s experiments, subjects were given short essays on “Castro’s Cuba” that adopted either a pro-Castro or an anti-Castro stance. While some subjects were told that the pro- or anti-Castro position taken by the essay was chosen by the writer of the essay, others were told that the essay’s viewpoint was randomly assigned by an instructor. As one might expect, when a pro- or anti-Castro position appeared to be chosen by the writer, respondents overwhelmingly attributed the attitude expressed to the writer him- or herself. However, when respondents were told that an essay’s viewpoint was randomly assigned, subjects still tended to attribute the attitude expressed to the writer. Subjects thus failed to properly account for the situational constraints imposed on the writers. Even when fully informed about the background or situational factors that controlled behavior in this case, they tended to attribute behavior to the agents who engaged in the behavior directly.

The fundamental attribution error—a tendency of observers to attribute behavior to the dispositions of agents rather than situational factors, even when situational factors are an explicit part of a scenario presented—has been repeatedly confirmed over decades of research. Although various theories have been proposed to explain it, I suspect it arises largely from two human traits. First, humans are natural pattern recognizers, i.e., they naturally pick out “patterns” in their environment, often finding causal relations where there are, in fact, none. Such a tendency to identify causal patterns may inhibit attributions of behavior to background or situational factors because perceived patterns in people’s immediate sensory environments will likely short-circuit any process of attributing cause to unseen, situational forces. If one takes a pill and one’s headache disappears, an individual is likely to infer that the pill caused the outcome. Possible situational factors—e.g., that headaches go away on their own after a while—are unlikely to be accorded the same degree of plausibility because such inferences have been preempted by a pattern that more directly impacts one’s senses. Second, most “causal” relations perceived by humans are agentive in character, i.e., most events are seen as caused by human or human-like agents. Before the rise of science, for example, natural phenomena were almost universally attributed to human or human-like agency. As David Deutsch writes,

Most ancient accounts of the reality beyond our everyday experience were…anthropocentric. That is to say, they centred on human beings, and more broadly on people – entities with intentions and human-like thoughts – which included powerful, supernatural people such as spirits and gods. So, winter might be attributed to someone’s sadness, harvests to someone’s generosity, natural disasters to someone’s anger, and so on… In biology, it was once thought that living things must have been designed by a supernatural person, and that they must contain some special ingredient, a ‘vital principle’, to make them behave with apparent purposefulness (p. 42-3).

And as Jared Diamond observes about hunter-gatherer people, a decidedly non-market pricing context:

Hunter-gatherers over-generalize agency and extend it to other things that can move besides humans and animals, such as rivers and the sun and moon. Traditional peoples often believe those moving inanimate objects to be, or to be propelled by, living beings. They may also attribute agency to non-moving things, such as flowers, a mountain, or a rock. Today we label that as belief in the supernatural, distinct from the natural, but traditional peoples often don’t make that distinction. Instead, they come up with causal explanations whose predictive value they observe: their theory that the sun (or a god carrying the sun in his chariot) marches daily across the sky fits the observed facts. They don’t have independent knowledge of astronomy to convince them that belief in the sun as an animate agent is a supernatural error. That isn’t silly thinking on their part: it’s a logical extension of their thinking about undoubtedly natural things (p. 338).

The tendencies to (a) find causal patterns in one’s immediate sensory environment and (b) attribute causal relations to human agency suggest that humans may interpret social situations in terms of a kind of “grammar”—a blueprint for parsing scenarios into parts and relating the parts to each other. In particular, the tendency to see cause-effect patterns in the sensory environment immediately before one implies that people will neglect background information in their causal attributions. And the tendency to attribute cause to anthropocentric agency suggests that, if any agent is conceivably present in a scenario, people will attribute cause to that agent. Derogation of victims in just-world experiments is then explained as follows: when misfortune strikes, humans instinctively seek a causal agent; and if the only causal agent present is the victim him- or herself, blame, by default, goes to the victim.

More generally, people may seek anything in a scenario that can possibly be construed as agentive on the part of a victim. The fact that a person’s biography is often a black box from the perspective of an observer creates leeway to attribute a person’s misfortune to the individual him- or herself. If a person is dying of cancer, one can, if so motivated, believe that the individual’s lifestyle somehow contributed to it. If a person is economically destitute, one can, if so motivated, believe that the individual did not invest enough in education or work hard enough to avoid such a fate.

The just-world experiments appear to establish the boundary lines within which the fundamental attribution error occurs. At one extreme, when no causal agent apart from the victim of a misfortune is present, cause is assigned to the victim. This is an extreme case of the fundamental attribution error, since such an attribution not only neglects the situational constraints imposed on the individual but implicates agency where there is none. At the other extreme, when an agent who clearly causes a victim’s suffering is present, as in bullying scenarios, blame is directed to the appropriate causal agent, and the fundamental attribution error disappears. In both cases, cause is (a) found within the sensory foreground (the frame directly before the observer) and (b) attributed to human agency. Thus, victim-blaming per se is not associated with just-world beliefs (since just-world believers do not blame victims in bullying scenarios). Rather, victim-blaming is an extreme case of the fundamental attribution error.

Movies and television exploit this “grammar” of social situations. While absorbed in a dramatic scene, audiences ignore the background or situational elements they know to be present: that the characters and situations depicted aren’t real, that the people they see are actors, that the moments observed are likely interrupted frequently by directors yelling “cut,” that shots are carefully edited to create an illusion of drama, that plots of which scenes are part are typically constructed with no higher purpose than to hold the viewer’s attention, etc. Instead, all causes and effects are assumed by the viewer to be present within the drama itself. Moreover, fictional drama is all about the causal powers of human beings. Protagonists—an integral feature of fictional drama—are agents by definition, i.e., people who cause things to happen. It is true that things happen to protagonists, and indeed, misfortune that befalls a protagonist is typically the starting-point of a drama. But protagonists are seldom passive victims. And usually they act in ways that are at least partially, and often wholly, determinative of dramatic outcomes. Although we frequently meta-analyze plots, characters, actors, writers, and directors, after engaging in such analysis, we easily slip right back into the overtly fictional realm. The “grammar” of social situations snaps back into place.

This is not to say that metaphysical assumptions about the “world” are irrelevant to how we perceive social situations. Causation may run in both directions. That is, people prone to the fundamental attribution error may be particularly receptive to just-world ideologies, in particular, to religion. And once a just-world ideology is adopted, the ideology may itself influence how people perceive the world, rendering them more likely to attribute cause to anthropocentric than to situational factors. Religion and just-world ideologies may thus reinforce the tendencies that predispose people to adopt them in the first place. That is, they may be self-propagating memes.

Broadening one’s frame

Humans are open to the possibility that events may be caused by impersonal material forces. Hence, they have been able to utilize material forces to improve their well-being to a degree unprecedented in the animal kingdom—developing tools, pottery, cooking, hunting implements and techniques, agriculture, mechanical processes, and ultimately science. But the causal attributions that enabled these advances were hard-won, and the fruits of this type of reasoning took many millennia to realize. The overwhelming bias in the first instance is to attribute cause not to material but agentive forces. The fundamental attribution error (and by extension, victim-blaming in just-world experiments) thus seems unsurprising. Barring some learned or cultural recognition that immaterial non-agentive factors really are all there is to causation in many instances, humans’ reflexive attribution of cause to agentive forces is hard to shake.

But what about the subjects who do not blame victims in the just-world experiments? Clearly, people can broaden their frames and attribute cause to non-agentive factors outside their immediate sensory awareness. In the just-world experiments, if one broadens one’s frame—if a test subject reflects, for example, that she is involved in an experiment and that the scenario presented is an artificial concoction by researchers—then the perceived source of a victim’s sufferings may change, switching from visible agents to the researchers conducting the experiment or perhaps to the “world.” Although significant numbers of test subjects appear to take this step outside the frames presented to them and entertain such notions, this is not the most natural move for human beings. It is, after all, more cognitively taxing to attribute cause to uncertain, obscure background forces than to agents in the scene directly before one. However, pushing in the opposite direction is “modernity,” which may make broader situational factors increasingly difficult to ignore. Exposure to science, rational argument, commerce, multiple cultures, multiple religions, repudiations of traditional morality, improved knowledge of our place in the world and universe, etc., may condition people to question their intuitions and increasingly recognize potential causality in situational factors.

It is crucial to understand the context in which many of the just-world experiments have been conducted. Some researchers, Haidt included, have argued that the chronic use in psychology experiments of undergraduates who are western, educated and from industrialized rich democracies (W.E.I.R.D.) narrows the range of possible findings of such studies and thus the insights that can be gained from them. However, we should not lose sight of one respect in which experiments that use college students reflect real-world conditions. This is the underlying context: “market pricing,” in Fisk’s terms, or “the rational-legal mindset,” in Pinker’s. Participation in such studies is usually a market transaction, with students paid for their time in money or course credit. Also, student participants typically have no prior knowledge of fellow test subjects. Although college students may be of similar socioeconomic backgrounds and cultures, they generally are not of the same communities, i.e., not of the same tribes. Thus, they are for the most part strangers to one another, no less so than people generally are in advanced industrial societies. The market pricing context entails that the bases for sympathy associated with kin and community, and that are operative in communal sharing and equality matching contexts, are largely absent.

So, what does a market pricing context predict observers’ responses should be to the misfortunes of blameless victims? It cuts two ways. On the one hand, since the victims are, for the most part, neither kin nor fellow community members, subjects should not be disposed to feel especially sympathetic with them. On the other hand, since a market pricing context encourages people to broaden their frames, which should alert them to possible situational factors that may cause a victim’s misfortune, subjects may be more disposed to attribute a person’s suffering to outside, uncertain forces. Thus, the context predicts a bifurcated response, which is what we observe: about half of subjects blame the victim, and about half of subjects blame outside, situational factors.

As noted, there is no particular reason to assume that belief in a just world causes people to commit the fundamental attribution error in its most extreme form, i.e., victim-blaming in the just-world experiments. Subjects in these experiments are not asked to reflect on their theological or metaphysical views about the “world” and then decide whether a victim is blameworthy. They are simply asked to evaluate a victim of circumstance, and victim-blaming in such cases correlates with just-world beliefs. Thus, we may ask: why do people with a reflexive tendency to neglect situational factors in causal attributions tend to view the world as “just”?

One possibility is that a just-world view simply resonates with such individuals. If cause rests primarily with people in one’s immediate sensory awareness, why not assume the best about the “world”? Such an assumption likely has psychic utility in that it probably makes people happier.

It’s also possible that the fundamental attribution error is an evolved response, an adaptation. Humans are social creatures. Thus, a fair amount of what happens to human beings is caused by other humans, and so we might do well to be especially sensitive to this possibility. Additionally, possible background or situational factors are not immediately visible and are many and various. Picking out those background factors that may be operative in a given situation is cognitively more taxing than discerning causes that are personal and seemingly present in one’s immediate sensory foreground. In effect, attributing cause to personal rather than situational factors may yield more intellectual bang for the cognitive buck.

There is also the issue of variability. Throughout our evolutionary history, background or situational factors would have been relatively unchangeable. Bereft of science and technology, people were largely at the mercy of resource availability, climate, an absence of disease, the good will of neighboring tribes, etc. If something went awry in such circumstances, the one variable that people could affect was human behavior, either their own or others’. Thus, even if one’s misfortune is a product of circumstance, the rational response in such conditions is not to try to change the “world,” which is impossible in any case, but to alter one’s behavior, the only controllable variable in the situation. And to motivate a change in behavior, a focus on the agentive powers of human beings—including attributing blame to victims—could be helpful. It may, of course, be that regardless of what a victim does, the outcome will be the same. Even so, at the margin, there is at least some chance that an alteration of behavior can produce positive results. If a blight destroys a farmer’s crop, although the farmer couldn’t have prevented his misfortune, he can respond by planting a crop that isn’t susceptible to the blight. Such a response may be more likely if he receives opprobrium for his misfortune than if the community in effect insures him against such losses out of sympathy.

Additionally, in an environment in which humans are largely at the mercy of circumstance rather than active agents responsible for their fortunes, it may (somewhat ironically) have made sense to blame victims for their suffering. Why? Because victim-blaming enables one to focus one’s scarce attentional resources away from victims, who cannot be helped in any case, and toward more productive pursuits. Human attentional resources are limited and can in principle be allocated in any direction. To maximize one’s reproductive success, one should focus on just those activities likely to yield benefits. Blaming the victim for one’s misfortune frees one of the need to right situations that cannot be righted, of wasting resources trying to fix what cannot be fixed. When people are living on the edge of survival, and victims of circumstance cannot be helped, this may be adaptive.

In any case, although some of these stories may seem “just so,” it is clear that the fundamental attribution error enables people to efficiently perceive causes—those with a human source—that in most times and places would have been most relevant for humans to know about. These would be human causes. In the tightly bound social environments in which humans have predominantly lived, things that happen to people that people are most able to affect are caused by fellow humans. While the fundamental attribution error may lead to excessive false positives (attributions of cause to human agency where the latter is not involved), false positives are far less detrimental than false negatives (non-attributions of cause to human agency where such cause does exist).The fundamental attribution error is on the whole adaptive. It is beneficial to be guided in one’s attribution of cause and effect to a source that is likely to have been the cause and that one has some power to affect. Excessive attribution error only appears to become detrimental in relatively recent human history, when human society has evolved beyond band and tribal societies, so that peace and security depend upon relatively harmonious interactions between strangers and where humans have developed a capacity to affect the world through an understanding of the material and situational forces that are the true causes of many events that affect human beings.