Cognitive models of stereotype change (4): Motivational and cognitive influences

Three experiments investigating the effects of cognitive and motivational factors on stereotype change are reported. Trait ratings in all three experiments showed there to be greater stereotype-change when stereotype-inconsistent information was dispersed across many group members than when it was concentrated in only a few. A sorting task (Experiment 1) indicated that, in the concentrated conditions the stereotype disconfirmers were more strongly isolated from the rest of the group than in the dispersed conditions. Free recall protocols (Experiment 2) showed greater memory for the target exemplars when subjects anticipated interaction with a group member than when no interaction was anticipated. The questions subject chose to ask target group exemplars were also influenced by anticlpated future interaction. Subjects chose more stereotype-inconsistent questions when interactian was anticipated than when no interaction was anticipated. Experment 3 showed the impact of stereotype-inconsistent information to be greater when expectancies for the stereotyped group are weaker A cued-recall task yielded evidence of spontaneous subtyping. All these studies support the subtyping model, even in the presence of cognitive and motivational factors that might be expected to impede stereotype change.     

Cognitive load and person memory: the role of perceived group variability

Two studies investigated the effects of cognitive busyness and group variability on participants' memory for stereotype‐related information. In Study 1, participants formed an impression of an experimentally created group that was either homogeneous or heterogeneous in composition. While learning about the group, half of the participants were made cognitively busy, the others were not. The results supported our prediction that stereotypical efforts on memory are moderated by both the availability of processing resources and the variability of the target group under consideration. Under optimal processing circumstances, participants' recollections were dominated by the perceived variability of the group in question. That is, participants displayed preferential recall for stereotype‐consistent information when they believed the group to be homogeneous in composition, but a tendency to recall more stereotype‐inconsistent information when they considered the group to be heterogeneous in nature. Under sub‐optimal processing conditions, however, a different pattern emerged. Now, participants preferentially recalled stereotype‐consistent information regardless of the perceived variability of the group. These results were largely replicated in Study 2 when the perceived variability of a real social group was manipulated. We consider the implications of these findings for contemporary theories of stereotyping. Copyright © 1999 John Wiley & Sons, Ltd.     

The ISI Change phenomenon: When contradicting one stereotype changes another

When a population is perceived to consist of two mutually exclusive and exhaustive groups, the stereotypes that people form of these groups are likely to be interdependent. We predicted that stereotype-incongruent information about members of one group (target group) may then evoke change in the stereotype of another group (alternative group) even in the absence of any stereotype-incongruent information about members of the alternative group. We demonstrated this Indirect Stereotype-Incongruence Induced (ISI) Change phenomenon in two studies using novel groups (Study 1 and 2). The effect occurred when the incongruent information was about a competent target group (changing the stereotype about the less competent alternative group) but not when it was about an incompetent target group (Study 2). The ISI Change phenomenon implies that stereotypes may change even in the absence of directly stereotype-incongruent information.     

The impact of stereotype-incongruent information on perceived group variability and stereotype change

Three experiments showed increases in the perceived variability of social groups after perceivers received stereotype-incongruent information about group members. In Experiment 1, participants generated flatter distributions after exposure to incongruent information, compared with equally deviant congruent information, in the form of typical verbal materials. Experiment 2 indicated similar changes in dispersion after the presentation of numeric information about a single group member. In Experiment 3, the authors manipulated cognitive load at encoding or at the time group judgments were requested. Under conditions of cognitive constraint, stereotype-incongruent information ceased to promote more dispersed group representations. These results are consistent with the idea that incongruent information triggers more deliberative and comprehensive retrieval and generation of exemplars. The authors discuss the implications of these findings for stereotype change.     

Contrast effects in stereotype formation and change: the role of comparative context

Two experiments investigated the way in which the presence of a comparative or inter-group context during stereotype formation affects stereotype change, induced by subsequent disconfirming information. Participants learned about a focal group, after learning about one of the two context groups. After reporting their stereotypes about both groups, participants learned additional information about the focal group. This information described new group members who either confirmed or disconfirmed the group stereotype. Consistent with previous research, participants formed more extreme stereotypes about the focal group on dimensions that distinguished it from the context group (i.e., a contrast effect). In response to the subsequently presented disconfirming group members, a greater stereotype change was observed on dimensions that distinguished the focal group from the context group than on dimensions it did not. We argue that these effects are due to differences in perceived typicality of disconfirming group members.     

Cognitive models of stereotype change: (1). Generalization and subtyping in young people's views of the police

We report investigations of change in, and cognitive representation of young people's stereotypes of the police, in response to a police-schools liaison programme. This programme provides a real-life application of the ‘conversion’ model of stereotype change (in which stereotypes change radically in response to salient instances of disconfirming information). Study 1 revealed that school police officers were rated significantly more positively than the police in general, but that this view did not generalize to perceptions of the police in general. Stereotypes of the police became less positive over one year, although females were more positive than males, and school police officers were not judged typical of the category. Study 2 revealed that subjects categorized their school police officer separately from the police in general, and perceived him to share features with ‘caring and welfare’ professions, rather than other police officers and authority figures. Both studies converge on the limitations of the conversion model and tend to support the subtyping model (in which extremely disconfirming individuals are isolated from other group members).     

Atypicality and the two fundamental dimensions: Applying the negativity effect on warmth to group perception

The person perception literature has shown that negative information on warmth influences impression formation more than other kinds of information. In the present paper, we argue for the usefulness of using the knowledge accumulated on the negativity effect on warmth when studying how members of a group who are disconfirming the group stereotype are perceived. We show that negative divergent information on warmth is perceived as more surprising because it appears to be more discrepant than positive divergent information on warmth or than competence information. We also show how stereotype holders protect their stereotype by rating the surprising incoherent information as atypical. Results are discussed in terms of the necessity to apply our knowledge of the two fundamental dimensions of social perception and their properties to typicality ratings and, beyond them, to stereotype change research. Copyright © 2009 John Wiley & Sons, Ltd.     

Shift toward risk and heterogeneity of groups

Groups that were homogeneous and heterogeneous in terms of risk preferences were varied orthogonally across conditions of information exchange and discussion. The shift toward risk for heterogeneous group members was significantly larger than for homogeneous group members, who did not differ from a control condition. Further analyses showed that two possible alternative hypotheses, time differentials for discussion and regression toward the mean, were not tenable. The data support (a) the “at least as willing” interpretation of Brown's risk-as-value hypothesis, and (b) the necessity of constituting groups for heterogeneity and homogeneity with great care.     

Social categorization and the formation of group stereotypes: Further evidence for biased information processing in the perception of group-behavior correlations

An experiment tested the hypothesis that the mere categorization of people into social groups spontaneously instigates a mechanism through which group-relevant information is perceived and processed in a biased manner. This in turn may result in the erroneous perception of correlation between group and behavior. Subjects were initially assigned to be members of a minority group, a majority group, or were not assigned to a group. They were then presented with a series of statements that described members of the two groups performing either desirable or undesirable behaviors. Results showed that unaffiliated subjects perceived an illusory group-behavior correlation, indicating the operation of a cognitive bias to associate the minority group with distinctive behaviors. Subjects who were themselves members of the observed groups perceived illusory correlations that favored their own group, indicating a very different sort of bias. The results suggest that a categorization-based ingroup favoritism guided the manner in which group information was processed. These data lend support to the contention that social categorization spontaneously instigates specific cognitive mechanisms that contribute to group stereotype formation.     

Perceived entitativity,stereotype formation,and the interchangeability of group members

The authors investigated the effects of perceived entitativity of a group on the processing of behavioral information about individual group members and the extent to which such information was transferred to other group members. The results of 3 experiments using a savings-in-relearning paradigm showed that trait inferences about a group member, based on that member's behavior, were stronger for low entitative groups and for collections of individuals. However, the transference of traits from 1 group member to other members of the group was stronger for high entitative groups. These results provide strong evidence that the perception of high entitativity involves the abstraction of a stereotype of the group and the transfer of that stereotype across all group members. Implications for group impression formation and stereotyping are discussed.     

Group boundary permeability moderates the effect of a dependency meta‐stereotype on help‐seeking behaviour

Previous studies have found that when low‐status group members are aware that their in‐group is stereotyped as dependent by a specific out‐group (i.e. a dependency meta‐stereotype is salient), they are reluctant to seek help from the high‐status out‐group to avoid confirming the negative meta‐stereotype. However, it is unclear whether low‐status group members would seek more help in the context of a salient dependency meta‐stereotype when there is low (vs. high) group boundary permeability. Therefore, we conducted two experiments to examine the moderating effect of permeability on meta‐stereotype confirmation with a real group. In study 1, we manipulated the salience of the dependency meta‐stereotype, measured participants' perceived permeability and examined their help‐seeking behaviour in a real‐world task. Participants who perceived low permeability sought more help when the meta‐stereotype was salient (vs. not salient), whereas participants who perceived high permeability sought the same amount of help across conditions. In study 2, we manipulated the permeability levels and measured the dependency meta‐stereotype. Participants who endorsed a high‐dependency meta‐stereotype sought more help than participants who endorsed a low‐dependency meta‐stereotype; this effect was particularly strong in the low‐permeability condition. The implications of these results for social mobility and intergroup helping are discussed.     

Deviant but desirable: Group variability and evaluation of atypical group members

Two studies investigated how group variability affects reactions to atypical group members. In Study 1 (N = 65) we manipulated group variability and found that an atypical group member was evaluated more positively when the group was heterogeneous than when the group was homogeneous. In Study 2 (N = 276) we also manipulated group value and found a significant interaction whereby an atypical group member was evaluated more positively when the group was homogeneous and group members valued heterogeneity, but was evaluated more negatively when the group was heterogeneous and group members valued homogeneity. The results suggest that deviant or atypical members will not inevitably be rejected by the group, but rather that reactions to deviance are shaped and guided by the dynamic relationship between how the group is perceived by its members and their ideological beliefs about what is good for the group.     

A recurrent connectionist model of group biases

Major biases and stereotypes in group judgments are reviewed and modeled from a recurrent connectionist perspective. These biases are in the areas of group impression formation (illusory correlation), group differentiation (accentuation), stereotype change (dispersed vs. concentrated distribution of inconsistent information), and group homogeneity. All these phenomena are illustrated with well-known experiments, and simulated with an autoassociative network architecture with linear activation update and delta learning algorithm for adjusting the connection weights. All the biases were successfully reproduced in the simulations. The discussion centers on how the particular simulation specifications compare with other models of group biases and how they may be used to develop novel hypotheses for testing the connectionist modeling approach and, more generally, for improving theorizing in the field of social biases and stereotype change.     

Hidden profiles and the consensualization of social stereotypes: how information distribution affects stereotype content and sharedness

We examined the impact of the distribution of information regarding social groups on the formation of shared stereotypes within triads in two studies. Three‐person groups discussed which of three groups (A, B, and C) was the most able and the most sociable. In Study 1, some of the information about these three groups was available to all group members (shared) whereas the remainder was distributed among group members (unique). Based on the total profile, there was more evidence of group A being sociable and of group B being able than of A being able and B sociable. In Study 1 (n = 58), sampling was manipulated as ‘representative’ (information in line with the overall differences was shared) or ‘unrepresentative’ (only information contradicting these differences was shared). In a third condition, all items of information were shared. Emerging stereotypes were directly influenced by sampling of information independently of the discussion. As well as this, the discussion consensualized initial stereotypes. In Study 2 (n = 52), sampling was always unrepresentative and we manipulated the labels associated with the target groups in such a way that the stereotype associated with the label was either inconsistent or consistent with the overall differences between the target groups. In the inconsistent condition, participants were more likely to discuss information that violated stereotypical expectations, and to be less influenced by sampling as a result of discussion. Altogether, these findings suggest that information sampling directly affects the consensualization of social stereotypes. Copyright © 2003 John Wiley & Sons, Ltd.     

Protecting threatened identity: sticking with the group by emphasizing ingroup heterogeneity

In two studies (Ns=163, 164), the authors tested the prediction that perceptions of group variability can steer and guide the way that loyalty is expressed in times of identity threat. In both studies, participants were classified as lower or higher identifiers on the basis of their scores on a group identification measure, and manipulations involved group variability perceptions (homogeneous ingroup vs. heterogeneous ingroup) and threat to the ingroup. Higher identifiers presented with a homogeneous ingroup perceived more ingroup homogeneity under threat than when there was no threat. In contrast, higher identifiers who perceived the ingroup initially as heterogeneous perceived more ingroup heterogeneity under threat than in no threat conditions. Lower identifiers perceived more ingroup heterogeneity under threat (vs. no threat) irrespective of manipulated group variability perceptions. Discussion focuses on different ways that group loyalty can be expressed in times of identity threat.     

The Effects of Numerical and Logical Group Size on Computer-Mediated Idea Generation

A laboratory experiment investigated the effects of numerical group size and logical group size (i.e., group heterogeneity in regard to task demands) on the performance of groups addressing an idea generation task while using computer-mediated communication. Six different numerical group sizes were studied, which ranged from 5 to 10 members. Logical size was manipulated by varying the amount of task-relevant information given to a particular group member. Larger logical group sizes (heterogeneous groups) were induced by distributing unique task-relevant information among group members; smaller logical group sizes (homogeneous groups) were induced by providing all task information to all members. A significant main effect for numerical group size showed that larger groups outperformed smaller groups. Numerical group size interacted with logical group size resulting in greater performance gains for increased numerical group size within heterogeneous groups. Average contributions per group member diminished with increased numerical group size for homogeneous groups and increased for heterogeneous groups.     

Biased information search in group decision making

Research has shown that people prefer supporting to conflicting information when making decisions. Whether this biased information search also occurs in group decision making was examined in three experiments. Experiment 1 indicated that groups as well as individuals prefer supporting information and that the strength of this bias depends on the distribution of the group members' initial decision preferences. The more group members had chosen the same alternative prior to the group discussion (group homogeneity), the more strongly the group preferred information supporting that alternative. Experiment 2 replicated these results with managers. Experiment 3 showed that the differences between homogeneous and heterogeneous groups reflect group-level processes. Higher commitment and confidence in homogeneous groups mediated this effect. Functional and dysfunctional aspects of biased information seeking in group decision making are discussed.     

Accountability moderates member-to-group generalization: Testing a dual process model of stereotype change

According to contemporary models of accountability [Lerner, J.S., & Tetlock, P.E. (1999). Accounting for the effects of accountability. Psychological Bulletin, 125, 255-275], when individuals are warned that they will be held accountable for their decisions, both information processing and judgment vigilance increase. We used an established generalization paradigm [Garcia-Marques, L., & Mackie, D.M. (1999). The impact of stereotype incongruent information on perceived group variability and stereotype change. Journal of Personality and Social Psychology, 77, 979-990] to extend the application of these principles to the process of member-to-group generalization in stereotype change. As predicted, across the three studies (Ns = 60, 78, and 101), accountability was found to amplify generalization under control conditions, both when the member information was stereotypical (Experiment 1) and counterstereotypical (Experiments 2 and 3). Accountability was found to attenuate generalization (Experiments 2 and 3) when a meta-judgmental cue discredited the validity of the member information for the group judgment. Ancillary evidence from Experiments 2 and 3 suggests a mediational role for the cognitive fencing-off of the member information from the group schema. The implications of the observed interplay between stereotyping and meta-cognitions for theory and policy are discussed.