If there is a gang of hoodlums hanging out at the ATM, I'm likely going to wait to take out money. If I invite a girl from the hippy dorm out to dinner, I'd probably check to see if she is vegetarian. In Jussim's (Reference Jussim2012) book, his Stereotype Rationality Hypothesis implies that it is rational and reasonable for me to make these decisions – to be accurate we should use stereotypes when there is no relevant individuating information. But when there is individuating information that conflicts with the stereotype, relying on the stereotype would be unreasonable. Maybe I overhear the gang of hoodlums discussing the fate of the Baudelaire siblings. Maybe the girl from the hippy dorm is wearing a leather jacket. If this is useful but not definitive information, I should still rely on the stereotypes. But if I have clear, credible, reliable, and abundant evidence that the hippy dorm girl is a carnivore –maybe she told me – then I can jettison the stereotype.
I suggest that the Stereotype Rationality Hypothesis is only, well, partially right. (That makes it partially wrong.) I agree it is rational to rely on stereotypes, but in the complexity of real world social interactions, most of our individuating information is going to be non-definitive, and even when it seems to be definitive, it seems that way because it invokes additional stereotypes. It is hard to get away from stereotype use, which is why it is essential to investigate the relative accuracy of the different kinds of stereotypes we create and rely on. And I agree that this project has been neglected. In its place arose a strange but wildly popular story about how we come to understand others.
In developmental and comparative psychology, and in philosophy of mind, there has been a tendency to emphasize human accuracy in interpretation of people and prediction of behavior. The notion driving almost 40 years of research in theory of mind, mindreading, and folk psychology is that for accuracy, we need to understand the beliefs and desires of others; it is thought we need to be mindreaders.
Jussim's characterization of social psychology as emphasizing inaccuracy in social perception, and bias in stereotype, helps us understand why developmental and comparative psychologists have focused so much attention on theory of mind/mindreading – namely, if humans are really good at interpreting people and predicting their behavior, and stereotypes are inaccurate when it comes to person perception, they can't play a role in our typical, accurate, folk psychological reasoning. The motivated denial of stereotype accuracy not only led us away from investigating how stereotypes work in human cognition, but because we are largely successful when coordinating behavior with members of our community, the denial of stereotype accuracy led us toward a worse theory of how we understand other people – theory of mind.
Stereotypes can be seen as a kind of inductive reasoning, at least for those stereotypes that arise from one's experience with reality. Theory of mind is a kind of theoretical reasoning, in which we construct theories about the invisible mental causes of human behavior based on limited behavioral evidence. I have argued that there is reason to think inductive reasoning is more accurate than theoretical reasoning in this case (Andrews Reference Andrews2012). Why worry about accuracy in theory of mind? First off, any one piece of behavior can be caused by a number of different mental states. The interviewer might offer the candidate the job because she thinks he is the smartest, the most sociable, or the cutest. She might not even know why she hired him! Thinking about someone's reasons for action – their beliefs – can trigger confirmation biases. Since the holism of the propositional attitudes causes an intractability problem, the relationship between observable behavior and the propositional attitudes that presumably cause behavior would be too complex to allow for timely, much less accurate, prediction of behavior (Zawidzki Reference Zawidzki2013). The unmitigated search space would be too great. We need to limit search space in order to practically use our theory of mind ability. Apperly (Reference Apperly2010) thinks we can limit the search space by appealing to scripts of typical behavior, and Zawidzki thinks that our ancestors’ practices of mindshaping, which led to cohesion in our community, and differences between different communities, limits the search space for each community. I endorse another option – we use stereotypes to mark the probabilities of particular actions, beliefs, goals, and so forth. Since I think that the role of theory of mind in understanding other people, and predicting behavior, has been wildly over-exaggerated, I'm sympathetic to Jussim's project of showing how stereotype use can be accurate.
But I don't think he goes far enough.
This is because Jussim thinks “people rely on stereotypes only hesitantly and reluctantly. Only when they have no individuating information or when the individuating information they do have is irrelevant or ambiguous do they use stereotypes to any substantial extent. Stereotypes, apparently, generally function not as a first option but, instead, as a best guess of last resort when there is little else to go on” (p. 381). This claim defends the Stereotype Rationality Hypothesis, and in turn Jussim thinks it is defended by studies that show that with more information, people do not default to a stereotype. He refers us to Locksley et al. (Reference Locksley, Borgida, Brekke and Hepburn1980), which found that when people lack relevant individuating information they judge a man as more assertive than a woman, but after observing a woman interrupting a dominating student, she is judged as assertive as a man. Findings like this do not challenge the view that the stereotypes operate as a baseline, however. We may find that for women, interrupting a dominating student gets her labeled as assertive, but that for a man, the same behavior gets him labeled as a good teacher. Background knowledge matters, and it can shape how we interpret movements.
The fact that stereotypic judgments don't simply assert themselves as the full and final story doesn't mean that they are not still in play. Indeed, I think that stereotypes are an important part of the practice of folk psychology, which involves constructing and manipulating models of individuals and groups. Our individual models consist of a variety of information, including personality traits, stereotypes, personal history, social role, relationships, situational factors, goals, emotions, and so forth (Andrews Reference Andrews2012). When we are engaged in person perception, we manipulate the model we are building of that individual. Knowing someone's gender identity is part of that model, and knowing someone's cultural background is another. So is knowing what they've done in the past, how messy their office is, and so forth.
Not only does the ability to deviate from a single stereotype in the face of additional information not undermine the importance of stereotypes, but, I'd like to suggest, the additional relevant information will often be based in other stereotypes. Much of what is referred to as individuating information is nothing more than more stereotypes. Let's go back to the definitions. Jussim endorses the definition of stereotype as “a set of beliefs about the personal attributes of a social group” (p. 302), and by “social groups” he means living human groups. While racial groups and gender groups are the ones that usually come to mind when talking about stereotypes, this definition reflects a much wider approach to stereotypes, and reflects the diversity of groups that one can form stereotypes about. Indeed, while discussing studies demonstrating evidence of (and against) stereotype accuracy Jussim includes the following social groups: business majors, ballet dancers, sorority members, dorm residents, day/night people (and members of political parties).
We have a definition of stereotypes, and a number of examples of social groups that are stereotyped. We also have a definition of individuating information as “information particular to a target person, rather than his or her group membership … it includes features such as a person's personality, preferences, tastes, attitudes, accomplishments, experiences, competencies, and behaviors” (p. 362). Examples of individuating information of different sorts include: physical appearance, dress, height, facial expressions, test results, student performance, assertive outburst, and tidiness of room or office. I want to challenge the distinction between information particular to a target person and group membership, since we can make social groups out of anything; we can also – and almost certainly do –form stereotypes about the kinds of people who share these “individuating” properties. We have stereotypes about good-looking people, hipsters, short people, smiley people, good students, and so forth. Stereotypes can also enjoy Venn diagram relationships (or fail to in interesting ways); we may have a consensual stereotype for male and female, and a different unique stereotype for Asian male that does not merely overlap the male and Asian stereotypes. In other cases, we might lack a stereotype for the various salient group memberships, so an Asian male hipster might be seen as more Asian male, or more hipster, depending on the context. Or he might be seen as some overlapping combination of the two stereotypes. What this suggests is that Jussim's Stereotype Rationality Hypothesis underestimates our need for stereotypes in social cognition. It is rational and reasonable for me to use reliable stereotypes in person perception, but I don't need to jettison the stereotype when I gain additional relevant or even “definitive” information about the person, especially since that may just invoke another stereotype. Instead, our deliciously complex cognitive capacities use all the relevant information, allowing us to construct rich models of other people. The hippy girl might love steak, but only when the cow was grass fed.
That brings me to my final issue. Social cognition is a triangle between two individuals and a particular context. The context, which includes the goals of the perceiver, is an important variable when it comes to person perception. The same black youth may appear to be a good student sitting in a college classroom, and a worrying threat on the street – even to the same perceiver. The studies of racial stereotypes that ask teachers to make predictions of their students are limited to the triangle of teacher–student–classroom. The teacher's stereotype of black youth may be accurate when it comes to her students, not only because she is familiar with them, but because she is likely motivated to see them in a particular way (part of the context). So it would be an overgeneralization to say that the teachers have accurate stereotypes of black youth. Rather, teachers may have accurate stereotypes of black youth in the classroom. When Republicans and Democrats use wildly inaccurate stereotypes about one another, their motivation to see one another in a certain way is also part of the context. If we care to examine the range of accuracy and inaccuracy in stereotypes (and I hope we do), we need to create a taxonomy of the varieties of stereotype types, one that reflects the order and breadth of the stereotypes themselves and the relationships between them, as well as the contexts in which they are created and used. We need a discipline of stereotype studies.
If there is a gang of hoodlums hanging out at the ATM, I'm likely going to wait to take out money. If I invite a girl from the hippy dorm out to dinner, I'd probably check to see if she is vegetarian. In Jussim's (Reference Jussim2012) book, his Stereotype Rationality Hypothesis implies that it is rational and reasonable for me to make these decisions – to be accurate we should use stereotypes when there is no relevant individuating information. But when there is individuating information that conflicts with the stereotype, relying on the stereotype would be unreasonable. Maybe I overhear the gang of hoodlums discussing the fate of the Baudelaire siblings. Maybe the girl from the hippy dorm is wearing a leather jacket. If this is useful but not definitive information, I should still rely on the stereotypes. But if I have clear, credible, reliable, and abundant evidence that the hippy dorm girl is a carnivore –maybe she told me – then I can jettison the stereotype.
I suggest that the Stereotype Rationality Hypothesis is only, well, partially right. (That makes it partially wrong.) I agree it is rational to rely on stereotypes, but in the complexity of real world social interactions, most of our individuating information is going to be non-definitive, and even when it seems to be definitive, it seems that way because it invokes additional stereotypes. It is hard to get away from stereotype use, which is why it is essential to investigate the relative accuracy of the different kinds of stereotypes we create and rely on. And I agree that this project has been neglected. In its place arose a strange but wildly popular story about how we come to understand others.
In developmental and comparative psychology, and in philosophy of mind, there has been a tendency to emphasize human accuracy in interpretation of people and prediction of behavior. The notion driving almost 40 years of research in theory of mind, mindreading, and folk psychology is that for accuracy, we need to understand the beliefs and desires of others; it is thought we need to be mindreaders.
Jussim's characterization of social psychology as emphasizing inaccuracy in social perception, and bias in stereotype, helps us understand why developmental and comparative psychologists have focused so much attention on theory of mind/mindreading – namely, if humans are really good at interpreting people and predicting their behavior, and stereotypes are inaccurate when it comes to person perception, they can't play a role in our typical, accurate, folk psychological reasoning. The motivated denial of stereotype accuracy not only led us away from investigating how stereotypes work in human cognition, but because we are largely successful when coordinating behavior with members of our community, the denial of stereotype accuracy led us toward a worse theory of how we understand other people – theory of mind.
Stereotypes can be seen as a kind of inductive reasoning, at least for those stereotypes that arise from one's experience with reality. Theory of mind is a kind of theoretical reasoning, in which we construct theories about the invisible mental causes of human behavior based on limited behavioral evidence. I have argued that there is reason to think inductive reasoning is more accurate than theoretical reasoning in this case (Andrews Reference Andrews2012). Why worry about accuracy in theory of mind? First off, any one piece of behavior can be caused by a number of different mental states. The interviewer might offer the candidate the job because she thinks he is the smartest, the most sociable, or the cutest. She might not even know why she hired him! Thinking about someone's reasons for action – their beliefs – can trigger confirmation biases. Since the holism of the propositional attitudes causes an intractability problem, the relationship between observable behavior and the propositional attitudes that presumably cause behavior would be too complex to allow for timely, much less accurate, prediction of behavior (Zawidzki Reference Zawidzki2013). The unmitigated search space would be too great. We need to limit search space in order to practically use our theory of mind ability. Apperly (Reference Apperly2010) thinks we can limit the search space by appealing to scripts of typical behavior, and Zawidzki thinks that our ancestors’ practices of mindshaping, which led to cohesion in our community, and differences between different communities, limits the search space for each community. I endorse another option – we use stereotypes to mark the probabilities of particular actions, beliefs, goals, and so forth. Since I think that the role of theory of mind in understanding other people, and predicting behavior, has been wildly over-exaggerated, I'm sympathetic to Jussim's project of showing how stereotype use can be accurate.
But I don't think he goes far enough.
This is because Jussim thinks “people rely on stereotypes only hesitantly and reluctantly. Only when they have no individuating information or when the individuating information they do have is irrelevant or ambiguous do they use stereotypes to any substantial extent. Stereotypes, apparently, generally function not as a first option but, instead, as a best guess of last resort when there is little else to go on” (p. 381). This claim defends the Stereotype Rationality Hypothesis, and in turn Jussim thinks it is defended by studies that show that with more information, people do not default to a stereotype. He refers us to Locksley et al. (Reference Locksley, Borgida, Brekke and Hepburn1980), which found that when people lack relevant individuating information they judge a man as more assertive than a woman, but after observing a woman interrupting a dominating student, she is judged as assertive as a man. Findings like this do not challenge the view that the stereotypes operate as a baseline, however. We may find that for women, interrupting a dominating student gets her labeled as assertive, but that for a man, the same behavior gets him labeled as a good teacher. Background knowledge matters, and it can shape how we interpret movements.
The fact that stereotypic judgments don't simply assert themselves as the full and final story doesn't mean that they are not still in play. Indeed, I think that stereotypes are an important part of the practice of folk psychology, which involves constructing and manipulating models of individuals and groups. Our individual models consist of a variety of information, including personality traits, stereotypes, personal history, social role, relationships, situational factors, goals, emotions, and so forth (Andrews Reference Andrews2012). When we are engaged in person perception, we manipulate the model we are building of that individual. Knowing someone's gender identity is part of that model, and knowing someone's cultural background is another. So is knowing what they've done in the past, how messy their office is, and so forth.
Not only does the ability to deviate from a single stereotype in the face of additional information not undermine the importance of stereotypes, but, I'd like to suggest, the additional relevant information will often be based in other stereotypes. Much of what is referred to as individuating information is nothing more than more stereotypes. Let's go back to the definitions. Jussim endorses the definition of stereotype as “a set of beliefs about the personal attributes of a social group” (p. 302), and by “social groups” he means living human groups. While racial groups and gender groups are the ones that usually come to mind when talking about stereotypes, this definition reflects a much wider approach to stereotypes, and reflects the diversity of groups that one can form stereotypes about. Indeed, while discussing studies demonstrating evidence of (and against) stereotype accuracy Jussim includes the following social groups: business majors, ballet dancers, sorority members, dorm residents, day/night people (and members of political parties).
We have a definition of stereotypes, and a number of examples of social groups that are stereotyped. We also have a definition of individuating information as “information particular to a target person, rather than his or her group membership … it includes features such as a person's personality, preferences, tastes, attitudes, accomplishments, experiences, competencies, and behaviors” (p. 362). Examples of individuating information of different sorts include: physical appearance, dress, height, facial expressions, test results, student performance, assertive outburst, and tidiness of room or office. I want to challenge the distinction between information particular to a target person and group membership, since we can make social groups out of anything; we can also – and almost certainly do –form stereotypes about the kinds of people who share these “individuating” properties. We have stereotypes about good-looking people, hipsters, short people, smiley people, good students, and so forth. Stereotypes can also enjoy Venn diagram relationships (or fail to in interesting ways); we may have a consensual stereotype for male and female, and a different unique stereotype for Asian male that does not merely overlap the male and Asian stereotypes. In other cases, we might lack a stereotype for the various salient group memberships, so an Asian male hipster might be seen as more Asian male, or more hipster, depending on the context. Or he might be seen as some overlapping combination of the two stereotypes. What this suggests is that Jussim's Stereotype Rationality Hypothesis underestimates our need for stereotypes in social cognition. It is rational and reasonable for me to use reliable stereotypes in person perception, but I don't need to jettison the stereotype when I gain additional relevant or even “definitive” information about the person, especially since that may just invoke another stereotype. Instead, our deliciously complex cognitive capacities use all the relevant information, allowing us to construct rich models of other people. The hippy girl might love steak, but only when the cow was grass fed.
That brings me to my final issue. Social cognition is a triangle between two individuals and a particular context. The context, which includes the goals of the perceiver, is an important variable when it comes to person perception. The same black youth may appear to be a good student sitting in a college classroom, and a worrying threat on the street – even to the same perceiver. The studies of racial stereotypes that ask teachers to make predictions of their students are limited to the triangle of teacher–student–classroom. The teacher's stereotype of black youth may be accurate when it comes to her students, not only because she is familiar with them, but because she is likely motivated to see them in a particular way (part of the context). So it would be an overgeneralization to say that the teachers have accurate stereotypes of black youth. Rather, teachers may have accurate stereotypes of black youth in the classroom. When Republicans and Democrats use wildly inaccurate stereotypes about one another, their motivation to see one another in a certain way is also part of the context. If we care to examine the range of accuracy and inaccuracy in stereotypes (and I hope we do), we need to create a taxonomy of the varieties of stereotype types, one that reflects the order and breadth of the stereotypes themselves and the relationships between them, as well as the contexts in which they are created and used. We need a discipline of stereotype studies.