25.6: Cognitive Mechanisms

Different Types of Processing

There is increasing evidence that the human mind consists of several subsystems that work relatively independently of one another. For example, there seems to be a special system for recognizing faces that is largely independent of the perceptual system for recognizing other sorts of objects (one piece of evidence for this is that people with certain kinds of brain damage lose one ability while the other remains intact). Subsystems are not just somewhat independent; they can sometimes even arrive at rather different conclusions about the same thing.

Although the general topic of cognitive (mental) subsystems is fascinating, only a couple of views about alternative types of cognitive processes are relevant here. Patricia Devine has argued that our thinking about other people goes on in two different ways.

First, there is a relatively automatic, habitual mode of thinking that reaches conclusions quickly, involves little or no conscious thought, and over which we have little control. This system is useful, because we often need to draw conclusions fast; we don’t have the time or attentional resources to reason in painful detail about everything we encounter, and this system lets us do a lot of social navigation on automatic pilot. But since it lies beyond our control, it can arrive at conclusions we would not endorse if we thought about things carefully; and since it lies outside our conscious thought, it can influence us without our awareness.

Second, there is a more self-conscious mode of thought that is more reflective, and that we can control; it’s what we normally think of as careful, conscious thought. It can lead to more well-reasoned conclusions than the automatic system provides.

A Two-Stage Model of Information Processing

Devine puts these ideas together in a two-stage model of our processing of information about people. In the first stage, automatic processing draws quick conclusions. Here, stereotypes can exert a strong influence. In the second stage, conscious processing can tell us that a conclusion we came to automatically is unsupported or even wrong, and we can then modify or ignore it.

Suppose, for example, that Wilma isn’t prejudiced against African Americans. Still, she has grown up in a society where the stereotype of black people as hostile is very common, so she has been exposed to it over and over. Under various conditions (e.g., when she sees a black person, or hears about one), this stereotype may be triggered automatically; it influences her thought without her intention. It’s automatic. But since Wilma is not prejudiced, her controlled, conscious processing can step in and remind her that the stereotype is unfair and inaccurate, and so she can avoid acting on it. By contrast, if Wilbur is prejudiced, this automatic, stereotypical reaction may be the end of the matter; he won’t go on to think about it more carefully. In cases where we don’t have much time or motivation to think about something consciously, e.g., because we are busy thinking about something else, we may only go through the first stage of processing. Devine’s research suggests that under these conditions, even people who are relatively unprejudiced may still be influenced by common stereotypes. In such situations, a stereotype is activated automatically, and people don’t go on to make the corrections they normally would have if they had given the matter conscious thought. Since the automatic processing occurs outside of conscious control, however, they aren’t aware of this. This gives us a theoretical framework to understand implicit bias.

Evidence for the Two-Stage Model

In one study, Devine used a test to divide white subjects into a high-prejudiced group and a low-prejudiced group. She then displayed words on a computer screen very quickly so that subjects were not consciously aware of them, but the words were up long enough that they did register unconsciously. One half of each group received words that are commonly associated with stereotypes of black people (‘jazz’, ‘hostile’, ‘Harlem’); the other half got completely neutral words. She then asked the subjects to read a story about a fictional character named ‘Donald’.

The story is designed to be ambiguous enough that it is possible for readers to interpret Donald’s actions in various ways. Race is not mentioned in the story. But even low-prejudiced subjects who had seen the words involved in the black stereotype interpreted Donald in a much more negative way than those who saw the neutral words. The words associated with the stereotype somehow activated negative feelings. This automatic processing, completely outside the conscious awareness and control, influenced how they interpreted Donald and his actions. Although the details of Devine’s model aren’t fully accepted by everyone working in her field, several subsequent studies confirm the basic idea that cultural stereotypes can operate automatically to influence how we think, without our even being aware of it. One moral of this research is the importance of careful, conscious thought, since without it we may arrive at conclusions that are quite different from the ones we would want to draw.

Levels of Generality

There also seem to be differences in processing that are based on the levels of generality of the things that we think about. The social psychologists Susan T. Fiske and Steven Neuberg argue that the attitudes we have concerning general categories or groups may differ a good deal from the attitudes we have toward their individual members. Many people claim to dislike lawyers in general, but they like their own (individual) lawyer just fine. Lots of people express contempt for politicians, but they like those that they know about; for example, most incumbents are popular in their home districts or states and win when they run for reelection. This is relevant here, because prejudices are often general, directed against “them”—that’s just what they are like.

The psychologist Robert Abelson suggests that there are two different subsystems here, and that they communicate poorly enough that they often lead us to have inconsistent attitudes (e.g., disliking a group while liking its members). A classic study by Richard LaPiere in 1934 illustrates the idea. LaPiere traveled around the United States with a young Chinese couple. They were accommodated in all but one of the two hundred restaurants and hotels where they stopped.

Later Lapiere wrote these places and asked if they would be willing to accept Chinese guests. Of those who answered, 90% said “No.” There are probably many reasons for this, but one may be that the respondents had negative views about the general category of Chinese, but they didn’t have any difficulty with these two Chinese individuals.

If something like this is right, it helps explain why a positive interaction with individual group members may do little to change a person’s general attitude toward the group (we will see additional explanations below). The reason is that the cognitive subsystem that deals with generalities and groups is to some degree unaffected by the subsystem that deals with specific individuals. If so, then it may be more effective to change prejudices against groups by changing how people think about the group’s history or achievements, rather than only by getting people to interact with group members; the latter is important, but it may not be enough because it may not affect the cognitive subsystem that is responsible for how we think about groups.