Dixon et al. provide an interesting overview of how relations between social groups may be influenced by beliefs and attitudes held by members of one group towards the members of another group. They refer to negative attitudes of this kind as “prejudices.” A major argument is that certain positive intergroup attitudes should also be considered as prejudices, because they may impede political progress toward what the authors call “social justice in the fuller sense” (Abstract, para. 2).
From our perspective of the neuropsychology of skill learning and performance, we found this argument incomplete, because no distinction was made between attitudes based on accurate versus inaccurate statistical models. In fact, only three of the eight definitions of prejudice in Table 1 (see Dixon et al.) consider whether an attitude is “faulty” or “unjustified.” Distinguishing between more and less accurate statistical models appears essential for understanding prejudices in this broader sense, that is, as predictive models about how other people and the world in general tend to behave, because they are likely to differ in their underlying mechanisms and behavioral consequences.
Evolution has endowed the human brain with many systems to acquire and update models of the external world. Some of this learning is explicit and cognitive, relying on personal experiences, as well as on second- and third-hand information acquired from various channels. Much of it also occurs implicitly from mere exposure, without intention to learn and with limited or no conscious awareness of what is being learned. Implicit learning has been widely studied for a range of phenomena, from simple perceptual and motor sequences to language and social cognition (Frith & Frith Reference Frith and Frith2012; Perruchet & Pacton Reference Perruchet and Pacton2006; Shanks Reference Shanks, Lamberts and Goldstone2005). Finally, learning itself is influenced by our evolutionary heritage. Numerous studies have shown that we are biologically prepared–prejudiced by evolution, as it were–to learn certain associations and that we are contraprepared to learn other associations. These biases are functional from the perspective of natural selection (Seligman Reference Seligman1970). As a simplistic example, humans are predisposed to learn a fear of snakes and spiders, but not of flowers (Öhman Reference Öhman2009).
It appears evident that these mechanisms are also sources of prejudices in the narrower sense discussed in the target article and, therefore, that both the cognitive and emotional aspects of intergroup attitudes in many instances reflect knowledge about the actual behavioral tendencies of groups. Furthermore, such knowledge is likely to be highly useful. For example, there are established statistical differences between men and women in a number of variables; how they express emotions (e.g., Kring & Gordon Reference Kring and Gordon1998), spend their time (e.g., Shanahan & Flaherty Reference Shanahan and Flaherty2001), how often they cry (e.g., Lombardo et al. Reference Lombardo, Crester and Roesch2001), and which types of work they prefer (e.g., Lippa Reference Lippa2010). The genetic, cultural, and environmental sources of these differences need not concern us here. The point is that the differences as such are highly informative about what people feel and do in various situations and can be used to modify one's actions to avoid conflict and social tension (cf. Gigerenzer Reference Gigerenzer2010).
A common objection to this line of reasoning is that individuals deviate considerably from the typical behavior of the many and multifarious groups to which they belong. This may make prediction inaccurate in each particular case and can hence lead to an unfair and inappropriate treatment of that person. Three comments in relation to this point are in order. The first point is essentially a reality check. In many social situations, both information about the other individual and the time to make a decision are severely limited. The brain is thus forced to make decisions based on necessarily imperfect predictions of the likely outcomes of the available behavioral alternatives. Secondly, the dangers of faulty overgeneralizations may not be as large as is sometimes suggested. Empirical studies on stereotypes suggest that people do not believe that all members of a group, but rather a larger proportion of that group compared to other groups, share the stereotypical properties (McCauley & Stitt Reference McCauley and Stitt1978). We can easily embrace the statistical statement that men are taller than women without expecting every man to be taller than any woman. Thirdly, although actions based on statistical predictions may result in unfair treatment of individuals, the effects of this can go both ways. Individuals that deviate in one direction from the group mean on a socially relevant variable may get more negatively treated than they would if they had been evaluated solely from their merits as individuals. However, by the same token, individuals that deviate in the opposite direction will receive a more positive treatment than they deserve.
Needless to say, many prejudices are based on blatantly wrong beliefs about other groups. We wholeheartedly support the notion that it would be beneficial for humankind to reduce intergroup prejudices that are negative, incorrect, and nonadaptive. Nevertheless, we believe that prejudices involving positive biases towards in-group members are of special interest both theoretically and practically. A recent comprehensive argument for that ethnocentric bias may increase inclusive fitness through kin selection has, for example, been provided by Salter (Reference Salter2007). Group cohesion without kinship gradients may also be adaptive under many circumstances (Hamilton Reference Hamilton1971). Although there is scientific controversy around these ideas, they appear highly relevant to any theory of prejudice.
In conclusion, we think that the study of prejudice may benefit from considering statistical learning and evolutionary perspectives. To use the terminology of Zawadzki (Reference Zawadzki1948): a “monistic” treatment of prejudices that solely regards their contents as “emotional hallucinations,” uncorrelated with the realities of group behavior, appears unlikely to be fruitful. Accurate models of the social, neuropsychological, and evolutionary mechanisms underlying prejudice are necessary for evaluating the feasibility and likely outcomes of various interventions designed to reduce prejudice. All of this will be essential information when considering political issues, but science and politics are better kept separate.
You have
Access
Dixon et al. provide an interesting overview of how relations between social groups may be influenced by beliefs and attitudes held by members of one group towards the members of another group. They refer to negative attitudes of this kind as “prejudices.” A major argument is that certain positive intergroup attitudes should also be considered as prejudices, because they may impede political progress toward what the authors call “social justice in the fuller sense” (Abstract, para. 2).
From our perspective of the neuropsychology of skill learning and performance, we found this argument incomplete, because no distinction was made between attitudes based on accurate versus inaccurate statistical models. In fact, only three of the eight definitions of prejudice in Table 1 (see Dixon et al.) consider whether an attitude is “faulty” or “unjustified.” Distinguishing between more and less accurate statistical models appears essential for understanding prejudices in this broader sense, that is, as predictive models about how other people and the world in general tend to behave, because they are likely to differ in their underlying mechanisms and behavioral consequences.
Evolution has endowed the human brain with many systems to acquire and update models of the external world. Some of this learning is explicit and cognitive, relying on personal experiences, as well as on second- and third-hand information acquired from various channels. Much of it also occurs implicitly from mere exposure, without intention to learn and with limited or no conscious awareness of what is being learned. Implicit learning has been widely studied for a range of phenomena, from simple perceptual and motor sequences to language and social cognition (Frith & Frith Reference Frith and Frith2012; Perruchet & Pacton Reference Perruchet and Pacton2006; Shanks Reference Shanks, Lamberts and Goldstone2005). Finally, learning itself is influenced by our evolutionary heritage. Numerous studies have shown that we are biologically prepared–prejudiced by evolution, as it were–to learn certain associations and that we are contraprepared to learn other associations. These biases are functional from the perspective of natural selection (Seligman Reference Seligman1970). As a simplistic example, humans are predisposed to learn a fear of snakes and spiders, but not of flowers (Öhman Reference Öhman2009).
It appears evident that these mechanisms are also sources of prejudices in the narrower sense discussed in the target article and, therefore, that both the cognitive and emotional aspects of intergroup attitudes in many instances reflect knowledge about the actual behavioral tendencies of groups. Furthermore, such knowledge is likely to be highly useful. For example, there are established statistical differences between men and women in a number of variables; how they express emotions (e.g., Kring & Gordon Reference Kring and Gordon1998), spend their time (e.g., Shanahan & Flaherty Reference Shanahan and Flaherty2001), how often they cry (e.g., Lombardo et al. Reference Lombardo, Crester and Roesch2001), and which types of work they prefer (e.g., Lippa Reference Lippa2010). The genetic, cultural, and environmental sources of these differences need not concern us here. The point is that the differences as such are highly informative about what people feel and do in various situations and can be used to modify one's actions to avoid conflict and social tension (cf. Gigerenzer Reference Gigerenzer2010).
A common objection to this line of reasoning is that individuals deviate considerably from the typical behavior of the many and multifarious groups to which they belong. This may make prediction inaccurate in each particular case and can hence lead to an unfair and inappropriate treatment of that person. Three comments in relation to this point are in order. The first point is essentially a reality check. In many social situations, both information about the other individual and the time to make a decision are severely limited. The brain is thus forced to make decisions based on necessarily imperfect predictions of the likely outcomes of the available behavioral alternatives. Secondly, the dangers of faulty overgeneralizations may not be as large as is sometimes suggested. Empirical studies on stereotypes suggest that people do not believe that all members of a group, but rather a larger proportion of that group compared to other groups, share the stereotypical properties (McCauley & Stitt Reference McCauley and Stitt1978). We can easily embrace the statistical statement that men are taller than women without expecting every man to be taller than any woman. Thirdly, although actions based on statistical predictions may result in unfair treatment of individuals, the effects of this can go both ways. Individuals that deviate in one direction from the group mean on a socially relevant variable may get more negatively treated than they would if they had been evaluated solely from their merits as individuals. However, by the same token, individuals that deviate in the opposite direction will receive a more positive treatment than they deserve.
Needless to say, many prejudices are based on blatantly wrong beliefs about other groups. We wholeheartedly support the notion that it would be beneficial for humankind to reduce intergroup prejudices that are negative, incorrect, and nonadaptive. Nevertheless, we believe that prejudices involving positive biases towards in-group members are of special interest both theoretically and practically. A recent comprehensive argument for that ethnocentric bias may increase inclusive fitness through kin selection has, for example, been provided by Salter (Reference Salter2007). Group cohesion without kinship gradients may also be adaptive under many circumstances (Hamilton Reference Hamilton1971). Although there is scientific controversy around these ideas, they appear highly relevant to any theory of prejudice.
In conclusion, we think that the study of prejudice may benefit from considering statistical learning and evolutionary perspectives. To use the terminology of Zawadzki (Reference Zawadzki1948): a “monistic” treatment of prejudices that solely regards their contents as “emotional hallucinations,” uncorrelated with the realities of group behavior, appears unlikely to be fruitful. Accurate models of the social, neuropsychological, and evolutionary mechanisms underlying prejudice are necessary for evaluating the feasibility and likely outcomes of various interventions designed to reduce prejudice. All of this will be essential information when considering political issues, but science and politics are better kept separate.