between nodes can be positive or negative; positive activations facilitate the connection between nodes, and negative connections inhibit connection. Given repeated stimulus exposure and experience, the activation of a few nodes within a pattern can lead to the completion of prior learned patterns; triggering nodes a andb alone can complete the pattern by stimulating candd. Pattern activation is constrained by the initial informa- tion input, and by prior weights (positive or negative) between nodes. Multiple simulta- neous patterns of activation occur at the same time, reflecting people’s capacity for concurrent on-line cognitions.
Applications of PDP models to social psychology have been few thus far, largely because theoretical models are still developing. However, a few programs of research apply PDP models to stereotyping phenomena. One particular new trend in the stereotyping literature is the focus on both facilitatory and inhibitory mechanisms that underlie stereo- typic thought (see Bodenhausen & Macrae, 1998, for review). Extending the emphasis on both positive and negative nodal connections, recent research has examined the vari- ables that undermine or dampen stereotypes (e.g., Macrae, Bodenhausen, & Milne, 1998;
Macrae, Bodenhausen, Milne, & Ford, 1997; Monteith, 1993), rather than focusing solely on variables that set them in motion. Another new trend in stereotyping research, derived from PDP models, is the explicit focus on constraint satisfaction processes (Kunda
& Thagard, 1996; Miller & Read, 1991), in which activated nodes (e.g., individuating traits or personal motives) can counter the overwhelming effects of other nodes (e.g., cat- egorical beliefs) when forming impressions and making judgments.
Social and cognitive psychology offer several models of mental representation, each model overlapping significantly with others, yet providing unique explanations and pre- dictions. Prototype, exemplar, associative network, and PDP models have contributed valuable insight into the cognitive structures that sustain stereotypes. They indicate that stereotypes emerge from basic cognitive units that store prior beliefs and expectations, and are thus embedded within our mental architecture.
The utility of social categories
As Lippmann (1922) remarked in the chapter’s opening quotation, humans cannot handle the complexity of their environment, and therefore “reconstruct it on a simpler model”
(p. 16) to function within such complexity. Gordon Allport’s (1954) The Nature of Prej- udiceelaborated on Lippmann’s theme of simplifying the world, introducing a cognitive perspective to the stereotyping literature. So remarkable were his insights on the mental processes guiding stereotypes that even the most recent empirical findings can trace their theoretical origins to his text, and his insights are taken for granted.
Allport (1954) argued, in a now familiar analysis, that categorization of objects is adap- tive and necessary to function effectively: “A new experience must be redacted into old categories. We cannot handle each event freshly in its own right” (p. 19). People catego- rize furniture such as tables and chairs; kitchen devices such as cookware and eating uten- sils; places such as restaurants and office buildings; and people such as professors, lawyers, and housewives. Categorization allows people to extract meaning from environmental objects by attending to a few diagnostic cues, rather than perceiving every attribute of every object – a time-consuming and labor-intensive task. In Allport’s words, “All cate- gories engender meaning upon the world. Like paths in a forest, they give order to our life-space” (1954, p. 171).
According to Allport’s analysis, the processes guiding categorization are the same for all objects. Upon encountering objects in the environment – whether furniture, utensils, locations, or people – individuals first selectcertain characteristics that define the object, thenaccentuatethose characteristics in their formed impressions (overlooking other char- acteristics), and finally interpretthe object by generalizing from those particular charac- teristics (see Allport, 1954, chapter 10 for more discussion). Although subsequent research refined this outline of stages (e.g., Brewer, 1988; Fiske & Neuberg, 1990, reviewed shortly), Allport’s initial work suggested that stereotyping follows basic rules associated with mere categorization, from which no perceivers or objects are exempt.
Although by now most current stereotyping research hinges on the role of categoriza- tion, Allport noted a fundamental distinction between categories and stereotypes. Cate- gories refer to associated concepts, properties, or objects that overlap in meaning or purpose. Stereotypes refer to exaggerated beliefs associated with a category of people that function to rationalize behavior toward that category (see Allport, 1954, chapter 12).
Stereotypes are specific consequences of the more general categorization process. The importance of this distinction lies in the difference between basic mental processes ensuing from categorization, versus interpersonal and social processes ensuing from gen- eralized beliefs about group members.
Models of category-based stereotyping
Social cognition researchers developed models to track the progression of people’s per- ceptions about others, from initial categories that promote stereotypes to piecemeal impressions that incorporate detailed information. The two most commonly cited models
are Brewer’s (1988) dual process model of impression formation and Fiske and Neuberg’s (1990) continuum model of impression formation. Superficially, both share similar fea- tures, particularly their emphasis on categorization and its effects on subsequent infor- mation processing, but vary in specific theoretical postulates (Fiske, 1988). This subsection extracts the core themes from the vast empirical research supporting one or both of these models: (a) Perceivers automatically categorize other people; and (b) whenever possible, perceivers interpret information about others according to their initial categorization, which can result in stereotypes. However, when motivated, perceivers can (c) make use of category-inconsistent information to revise their categorical beliefs; as well as (d) view others as individuated beings, rather than as stereotyped category members. Note the temporal nature of these themes – as proposed in the original theoretical models, each theme represents a discrete stage in the impression formation process.
Automatic categorization. Perceivers categorize other people immediately upon meeting them. Repeated findings, using computer-aided timing techniques, reveal that initial cat- egorization can occur within milliseconds after first encounter (Banaji & Hardin, 1996;
Devine, 1989; Dovidio, Evans, & Tyler, 1986; Zàrate & Smith, 1990). Perceivers typi- cally categorize others using obvious, visually salient cues, usually based on race, gender, and age (McCann, Ostrom, Tyner, & Mitchell, 1985; Perdue & Gurtman, 1990; Stangor, Lynch, Duan, & Glass, 1992); but other initial categories are also common, such as body size (Crandall, 1994; Ryckman, Robbins, Kaczor, & Gold, 1989), physiognomy (Kleck
& Strenta, 1980; Zebrowitz, 1997), and social roles (Macrae, Bodenhausen, & Milne, 1998). Initial categorization often occurs outside of perceivers’ awareness, and the effects of categories on perception can go unnoticed (Macrae, Milne, & Bodenhausen, 1994).
Once a target is placed within a category, numerous cognitive effects that facilitate stereotyping can take immediate hold (see Fiske & Taylor, 1991). For example, perceivers minimize differences between the target and other category members (Taylor, 1981), and ascribe stereotypic attributes to the person (Devine, 1989; Dovidio et al., 1986). Hence, upon categorizing a new acquaintance as a lawyer, perceivers might infer that he or she possesses stereotypical characteristics, such as ambition, intelligence, dishonesty, and greed.
The effects of initial categorization can have undeniably negative consequences for targets, who are perceived as interchangeable category members rather than as unique individuals, but this process confers advantages to the perceivers. By using automatic cat- egories to assist their perceptual processes, people can make quick judgments and con- serve cognitive energy for other tasks (Macrae et al., 1994). Thus, initial categorization organizes perceivers’ complex environments, directing attentional flow along an orderly trajectory.
Information interpretation. After automatic initial categorization, perceivers sometimes engage in more thoughtful processing. This depends on motivation to exert cognitive effort beyond the categorization stage, as well as available information for forming impres- sions (Fiske & Neuberg, 1990). With modest amounts of motivation, perceivers make
use of additional information (Erber & Fiske, 1984; Neuberg & Fiske, 1987; Ruscher &
Fiske, 1990), but are biased by their initial categories.
Following categorization, perceivers’ attention flows primarily to category-consistent information (Hamilton, Sherman, & Ruvolo, 1990). So after categorizing someone as
“Black” or “female,” perceivers tend to pay close attention to attributes consistent with those categories. Category-based information processing increases when perceivers have limited attentional resources – that is, when they have little time or energy to think accu- rately (Macrae, Milne, & Bodenhausen, 1994).
Although perceivers bias information processing toward category-consistent informa- tion, other types of information exert influence on impression formation. Perceivers can encounter three types of information when forming impressions of others: information that matches the category, information irrelevant to the category, and information that disconfirms the category. When information matches the category, expectations are con- firmed and categories strengthened (Hamilton et al., 1990; Oakes, Turner, & Haslam, 1991); stereotypes proceed from the alleged fit between target information and prior beliefs (cf. Snyder, 1984). When information is irrelevant to the category, perceivers either overlook it (Belmore, 1987; Fiske, Neuberg, Beattie, & Milberg, 1987) or interpret it according to category-based expectations (Hilton & Von Hippel, 1990; Nelson, Biernat,
& Manis, 1990); stereotypes thus capitalize on informational ambiguity. When infor- mation disconfirms the category, perceivers tend to perceive that information as non- representative of the general category (Krueger & Rothbart, 1990; Kunda & Oleson, 1995); stereotypes can thus “explain away” deviant information (e.g., Weber & Crocker, 1983). According to meta-analyses, perceivers tend to recall expectancy-inconsistent information only under very specific conditions (Stangor & McMillan, 1992), for example, when expectancies are weak, when incongruencies are strong, and when perceivers have explicit impression formation goals.
People assimilate information into their pre-existing beliefs about the category. This work resonates with Fiske and Neuberg’s (1990) idea of confirmatory categorization, whereby perceivers preserve their categories and prior beliefs via selected information searches. Stereotypes appear resilient to irrelevant and incongruent information, as per- ceivers find creative ways of reinterpreting discrepant information. But as we will see, stereotyping is not inevitable. Provided sufficiently disconfirming information and suffi- cient motivation to pay attention, perceivers can alter their categorical beliefs and view others as individuals, not only as category members.
Revising categorical beliefs. When highly motivated to attend and when provided adequate information, perceivers can modify both their a priori expectations and the very nature of their categories. Thus, stereotype change can occur by revising categorical beliefs through motivated attention to information (Hilton & Von Hippel, 1996).
Several models postulate the processes by which categories can change. An early model of change contrasted two processes (Rothbart, 1981). According to the bookkeeping process, people change their stereotypes gradually over time, by attending to category- discrepant targets and incorporating new information into that category. Categorical expectations change incrementally by averaging new information with prior beliefs.
According to the conversion process, people change their stereotypes more rapidly, as a
result of encountering highly discrepant category members. Accordingly, one prominent category member can individually alter the nature of the category (see Weber & Crocker, 1983, for empirical tests of this distinction).
Therefore, stereotypes can change when people attend to and incorporate new infor- mation into prior categories. Perceivers can form more specific categories subsumed within the broader stereotype, known as subtyping. Conversely, they can develop elabo- rated beliefs about the category, referred to as subgrouping (see Rothbart, this volume, chapter 3), which can ultimately dilute the stereotype. The latter process, wherein cate- gory variability diminishes stereotypic beliefs, presents the more promising avenue for stereotype change.
Forming individuated impressions. Models of impression formation posit that perceiving individuals according to their unique personal traits, rather than category-consistent attributes, represents the most individuated form of person perception (Brewer, 1988;
Fiske & Neuberg, 1990). However, these models suggest that individuation occurs under rare instances; even subtyping and subgrouping processes prioritize categories over indi- vidual traits. Category-based perception is more common because full individuation takes enormous effort from the perceiver.
Under rare conditions, perceivers are motivated to form piecemeal impressions that integrate all the target’s attributes. This motivation can come from outcome dependency (Erber & Fiske, 1984; Neuberg & Fiske, 1987), accountability (Tetlock, 1992), accuracy goals (Chen, Schechter, & Chaiken, 1996; Neuberg, 1989), and personal values of fairness (Monteith, 1993), in addition to other social motives (see Fiske, 1998, for review).
To arrive at an individuated impression, perceivers must proceed through all stages of perception just described – from categorization to information search and interpretation to recategorization. When no single category can explain the target person, perceivers treat the category as just another attribute. Perceivers appraise the individual’s unique attributes, incorporating them into a piecemeal impression (cf. Anderson, 1981). Cate- gories, thus, reduce to just another piece of information, and carry equal weight with other idiosyncratic information in the impression formation process (Fiske & Neuberg, 1990; see also Kunda & Thagard, 1996).