Automatic and Controlled Components of Stereotyping, Prejudice, and Self-Regulation
Central questions about stereotyping and prejudice are the extent to which they occur without a person’s awareness or intent, and the extent to which they can or cannot be controlled. Do we exhibit subtle prejudicial behavior of which we are not aware? Where do we get these biases? How and when are we able to overcome them? In our research, we examine the ways in which automatic and controlled processes interact with and constrain one another to produce stereotyping and prejudice. Current work is focusing on how these processes contribute to stereotype/prejudice development and change, the inhibition of stereotyping and prejudice, and stereotypic biases in memory, particularly as they relate to failures in eyewitness testimony. Other research is examining the interaction of automatic and controlled processes in a broader context of self-regulation that includes addictive behavior, phobias, aggression, emotion and judgment, etc. One tool we use in this research is the Quad model of automatic and controlled processing. The model is a multinomial model that separates the automatic and controlled components of behavior from performance on a single task (see below). In other research, we are using tools of cognitive neuroscience to understand the automatic and controlled components of social behavior. We are using both event-related potentials (ERPs) and functional magnetic resonance imaging (fMRI) to examine these questions.
For a Quad model tutorial, click here.
For Excel model templates, click here.
***This link contains 2 zipped templates. In the first template, categories in the compatible blocks that share the same equations (White+Good; Black+Bad) are collapsed. In the second template, categories in the compatible blocks that share the same equations remain separated. The categories may be collapsed if model fit is not significantly reduced by uncollapsing the categories or if the numbers of errors in the categories with the same equations do not differ from one another (as assessed by chi-square). In both templates, OB is assessed only from Black & White items (in incompatible blocks) but not from good and bad items. The models assume that AC drives responses to Good & Bad items. These templates also estimate the w statistic, a measure of effect size for model fit. The templates demonstrate use of the model with 3 separate conditions. Conditions may be deleted or added, as needed.
***I highly recommend the MultiTree software program for designing and modeling data. It can be downloaded for free, along with supporting documents here.
***In applying multinomial models, it is sometimes necessary to take into account inter-subject variability. The best way to do that at present is to either use individual-level parameter estimates or to conduct latent class modeling within the HMMTree program developed by Christoph Klauer and Christoph Stahl. HMMTree may be downloaded for free, along with an article describing the use of latent class modeling here.
Here is a .eqn equations file, an Excel file that shows the order of the response categories, and a .mdt data file for using the Quad model with either MultiTree or HMMTree. The .eqn and .mdt files are saved as .txt files here for downloading ease, but must be renamed with the .eqn and .mdt extensions to use in Multitree and HMMTree. Compatible response categories are separated in these files.
In another line of work, we are studying how stereotypes form—that is, how particular attributes come to be associated with different groups. We have proposed that, when people learn about a new group, attention is directed toward those features of the group that most clearly differentiate it from groups that are already known. Other information about the new group (including information about attributes that may be highly descriptive of the group) is not attended to. Thus, what people actually learn about a group is not necessarily the features that best describe the group; rather, what is learned are the features that best distinguish the group from other groups.
One consequence of this process is that the stereotypes that are formed about novel groups are particularly strong (because so much attention is devoted to only a few key differentiating attributes). Importantly, any feature that causes one group to be learned later than another group will lead to the formation of stronger stereotypes about the later group. For example, this may be one explanation for why minority group stereotypes are generally more prevalent and strongly held than majority group stereotypes. Because the features of majority groups are typically learned first, minority groups must be differentiated from majority groups. In doing so, attention is directed only at those attributes that permit differentiation, and the stereotypes that form around these attributes are quite strong.
Face Perception and Memory
In still other research, we are investigating many aspects of face perception and memory. One project aims to understand the phenomenon of hypo-descent—the tendency for people of mixed-race origin to be categorized as members of minority rather than majority groups (Barack Obama is seen as black not white). Though there are many factors that contribute to this tendency, we are particularly interested in the cognitive processes that contribute to it. Our model of stereotype formation (described above) also provides a means to explain hypo-descent. Among the features that are learned first about majority (vs. minority) group members are physical facial features. One result is that the physical features of minority groups may be associated especially strongly with those groups because they are attended to in order to differentiate minority group appearance from majority group appearance. This appears to be one reason why mixed-race people are perceived as belonging to minority groups.
Another aspect of face perception we are studying is the Own-Race Effect. Hundreds of studies have demonstrated that people of all races tend to more accurately recognize members of their own group than members of other groups. The implications for eye-witness testimony are significant. One explanation for this phenomenon is that we have greater expertise in differentiating among members of our own group. An alternative explanation that has received increasing support is that we are simply more motivated to attend carefully to in-group members and, therefore, subsequently recognize them more successfully. Our research has examined how variations in social context, group identification, and perceived group power moderate the extent of the Own-Race Effect.
Stereotypic Biases in Person Perception
We also examine the ways in which stereotypes and prejudices influence the impressions we form of other people. We are interested in how these beliefs affect the things we notice about another person, how we interpret their behavior, how we make judgments about them, and what we remember about them.
Much of this research also has been concerned with the efficient nature of stereotyping. One thing we know about stereotypes is that people tend to rely on them to a greater extent when their processing resources are depleted. Whether due to tiredness, task difficulty, or anxiety, situations that decrease the availability of processing capacity increase the use of stereotypes. We are interested in how this functional aspect of stereotyping influences the way we attend to, understand, judge, and remember other people.
The Mental Representation of Stereotypes and other Social Knowledge
Another area of research is concerned with understanding the mental representation of stereotypes and other social knowledge. Exactly what types of knowledge are activated in memory when we use a stereotype (or make judgments about another person or ourselves)? We are particularly interested in the extent to which stereotypes about a group are based on knowledge of particular group members' behavior or are based on abstract knowledge about what the group is like as a whole. This research has important connections to our work on stereotype efficiency. The factors that make stereotype use efficient also influence the manner in which other people’s behavior is represented in memory. In turn, these representational differences have important implications for how stereotypes may be changed.
Stereotype Inhibition and Individuation
Despite their efficient use, there are many situations in which we would rather not be influenced by our stereotypes. For both personal and social reasons, we often feel the need to avoid stereotypic thought. But how successful are these attempts? Recent research suggests that it is not so easy to suppress unwanted thoughts. Ironically, by focusing on these unwanted thoughts, we actually increase their mental accessibility. As a result, unwanted thoughts often "rebound," having greater influence than if we had never tried to suppress them.
Much of our research in this area has attempted to identify the conditions under which people will spontaneously engage in stereotype suppression and the conditions under which that suppression will and will not have these unwanted consequences. More recent endeavors have focused more on individuation as a means to avoid stereotypic responding. When we individuate, our impressions are based on the whole array of information that we have about another person, rather than simple group membership. We have been examining how intrinsic motivations to avoid prejudice encourage a variety of different kinds of individuating behaviors relating to attention (which attributes/behaviors of another person capture our attention), attributional processes (how do we explain others' behavior), and memory processes.