You are on page 1of 42

This article was downloaded by: [Tufts University]

On: 08 October 2014, At: 13:18


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

European Review of Social


Psychology
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/pers20

Social Categorization and


Stereotyping: A Functional
Perspective
a a
Ad van Knippenberg & Ap Dijksterhuis
a
University of Nijmegen
Published online: 15 Apr 2011.

To cite this article: Ad van Knippenberg & Ap Dijksterhuis (2000) Social


Categorization and Stereotyping: A Functional Perspective, European Review of
Social Psychology, 11:1, 105-144, DOI: 10.1080/14792772043000013

To link to this article: http://dx.doi.org/10.1080/14792772043000013

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all
the information (the “Content”) contained in the publications on our
platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy,
completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of
the authors, and are not the views of or endorsed by Taylor & Francis.
The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and
Francis shall not be liable for any losses, actions, claims, proceedings,
demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in
relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/terms-and-
conditions
Downloaded by [Tufts University] at 13:18 08 October 2014
Chapter 4

Social Categorization and


Downloaded by [Tufts University] at 13:18 08 October 2014

Stereotyping: A Functional
Perspective
Ad van Knippenberg and Ap Dijksterhuis
University of Nijmegen

ABSTRACT
Social categorization and stereotyping are discussed as normal perceptual
processes upon which perceivers rely for an adequate understanding of every-
day social events. Social categorization and stereotyping are treated as strongly
linked. It will be argued that in many situations social categorizations are used
because perceivers expect the categories involved to differ in important respects.
Thus, spontaneous social category selection is guided by the content of the per-
ceiver's stereotypes. It is further argued that stereotype activation and applica-
tion are largely governed by functional principles based on, for instance, the
perceiver's task or goals, the usefulness of the stereotype in a situation, and indi-
vidual differences in stereotypic beliefs. Subsequently, the functional determi-
nants of stereotype content are discussed. In the final section of this paper, we
pay attention to some of the mechanisms that operate in information selection,
perception, and memory in the service of the preservation of stereotypes.

What is it then, I pressed her further, that goes on between normal people, from
which she feels herself excluded? It has to do, she has inferred, with an implicit

Correspondence should be addressed to: Ad van Knippenberg, Department of Social Psychology,


PO Box 9104,6500 HE Nijmegen. The Netherlands.
E-mail: vanknippenberg@psych.kun.nl.
European Review of Social Psvchology. Volume 11. Edited by Wolfgang Stroebe and Miles Hewstone.
0 Zoo0 John Wiley & Sons Ltd.
106 AD VAN KNIPPENBERG A N D AP DIJKSTERHUIS

knowledge of social conventions and codes, of cultural presuppositions of every


sort. This implicit knowledge, which every normal person accumulates and gen-
erates throughout life on the basis of experience and encounters with others,
Temple seems to be largely devoid of (Sacks, 1995, p. 270).

In a way, social categorization and stereotyping are inseparable. If you use one
concept, the other is implied. Perceivers categorize other people, e.g., as
women, or as social psychologists, or as soccer hooligans, because they believe
that it makes a difference.That is, perceivers use social categories because they
have stereotypes about them. Or else, why would they? Conversely, when per-
ceivers stereotype other people, they must have categorized them first. As
Downloaded by [Tufts University] at 13:18 08 October 2014

stereotypes pertain to social categories, e.g., hooligans are aggressive,the social


category (hooligan) must be identified before one can apply the stereotype
(aggressive).
In our study of social categorization and stereotyping we adopt a functional
perspective (cE Allport, 1954; Tajfel, 1969). We look at social categories and
stereotypes as adaptive cognitive structures that we can hardly do without in
everyday social life because they enable us to understand quickly and quite
adequately what happens around us. Social categorization and stereotyping
are profoundly involved in comprehending ordinary social situations. That
is precisely what the opening statement we borrowed from Oliver Sacks is
meant to express. Although the realm of implicit knowledge, of which Temple,
the autistic person Sacks described, is devoid, encompasses more than stereo-
types, the description elegantly captures the nature of stereotypes: i.e., the sort
of things normal people have come to “know” about social groups. It would
be a tremendous handicap not to have stereotypes (or, more generally “cul-
tural suppositions of every sort”) because, as Sacks goes on to say, “Lacking
it, she has instead to ‘compute’ others’ intentions and states of minds, to try to
make algorithmic, explicit, what for the rest of us is second nature” (1995, p.
270).
In the present chapter, we elaborate on the knowledge function of social
categories and stereotypes. That is, we discuss these cognitive structures pri-
marily as useful tools for understanding social events. The emphasis will, there-
fore, be on the way in which social categorizations and stereotypes describe
and explain social reality. In analysing their epistemic function, two major
components may be discerned. On the one hand, social categorization and
stereotyping serve cognitive purposes in that they help the perceiver to con-
struct a meaningful and coherent interpretation of social events. This may be
labelled the “cognitive” function of a social category system. On the other
hand, social categorizations are used because they “accurately” reflect impor-
tant interstimulus differences and similarities.They provide social agents with
relevant distinctions between social stimuli in relation to their goals, motives,
and tasks. Social categorizations and stereotypes are useful in the sense that
CATEGORIZATION AND STEREOTYPING 107

(and to the extent that) they enable perceivers to act adequately in their social
environment. One might call this the pragmatic function of a social category
system (see, e.g., Tajfel, 1972). It will be evident, however, that although we
may analytically distinguish between these two components of the knowledge
function of social categories and stereotypes, they are part and parcel of the
same process. From the perspective of the perceiver, these functions merge
into an integrated, meaningful and psychologically valid perception of social
events.

SOME CONCEPTUAL AND THEORETICAL ISSUES


Downloaded by [Tufts University] at 13:18 08 October 2014

When we discuss social categorization and stereotyping, we need to have at


least some working definitions of these basic concepts. Social categorization
may be understood as the subjective classification of people into social
groups in a manner that is meaningful to the perceiver (Tajfel, 1972; van
Knippenberg, 1984). A social category, then, is a meaningful social group. In
our view, and consistent with Tajfel’s ideas on these issues (e.g., Tajfel, 1969,
1972,1981; Tajfel & Turner, 1979;Tajfel & Wilkes, 1963), a social categoriza-
tion may be considered to be meaningful to the extent that it co-varies with,
or “explains”, characteristics of people or events associated with them (van
Knippenberg & Ellemers, 1990). In brief, social categories are meaningful in
one or other context to the extent that one expects or perceives intercategory
differences and intracategory similarities. In fact, the use of the term “mean-
ingful” in this context is just another way to express the knowledge function
of social categories.
What is a stereotype? In very general, terms, stereotypes are mental repre-
sentations of social groups (Stangor & Lange, 1994;Stangor & Schaller, 1996),
i.e., images of groups in our minds.This definition is, however, still rather non-
descript. Exactly what information about the social group in question is being
mentally represented in the stereotype? Although in our own research we
have mainly focused on stereotypical traits, we prefer a rather loose concep-
tion of what stereotypes are made of. For all we know, mental representations
of social groups may contain exemplars, prototypical group members, typical
traits, and all sorts of other attributes, including attitudes, preferences, and
probably also typical behaviours, of the group or of the group members (see
e.g., Mackie, Hamilton, Susskind, & Rosselli, 1996). In summary, a stereotype
contains knowledge of what the group or members of the group in question
are like.
In line with our present concern with the cognitive and pragmatic func-
tions of social categories and stereotypes, we will focus on the perceived co-
variation between category membership and relevant stimulus features. We
share this emphasis on co-variation with self-categorization theory (Turner,
108 AD VAN KNIPPENBERG A N D AP DIJKSTERHUIS

1987) as well as with several social cognitive approaches to stereotyping (e.g.,


Ford & Stangor, 1992; Ryan, Park, & Judd, 1996). However, due to our focus
on epistemic aspects, we tend to neglect other functions and aspects of social
categories.These other functions and needs include the need for positive social
identity, the need to justify our value system, to preserve the status quo, and
to validate our cultural world view (see, e.g., Hoffman & Hurst, 1990; Jost &
Banaji, 1994;see also Stangor & Schaller, 1996). Following from our narrowed
focus, we do not incorporate larger conceptual systems or belief systems in our
analysis of social categorization and stereotyping. Also, motivational compo-
nents, such as striving for positive social identity, fall outside the scope of the
present chapter. The advantage is, of course, that this selectivity enables us to
Downloaded by [Tufts University] at 13:18 08 October 2014

analyse the knowledge function of social categories and stereotypes in greater


depth.
We believe that, in a way, the knowledge funcion of stereotypes is more basic
than some of the other functions, in the sense that these other functions may
be posited only because social categories and stereotypes serve as vehicles for
social knowledge. For instance, it is assumed that people strive for positive
social identity (Tajfel, 1978;Tajfel & Turner, 1979,1986).Therefore, perceivers
want ingroups to be better than outgroups, which may render ingroup stereo-
types more positive than outgroup stereotypes (Ellemers & van Knippenberg,
1997; van Knippenberg & Ellemers, 1993). As a result, the activation of
ingroup-related terms (we, us) enhances the accessibility of positive words,
while outgroup-related primes (they, them) enhance the accessibility of nega-
tive words (Perdue, Dovidio, Gurtman, & Tyler, 1990). Consistent with Tajfel’s
(1978) definition of social identity as “that part of an individual’s self-concept
derived from his membership in a social group” (p. 63), these motivational
effects on ingroup and outgroup stereotypes ultimately derive from the fact
that stereotypes tell us something about the characteristics of groups and
group members. In a nutshell, stereotypes can play an important role in the
perceiver’s positive social identity precisely because they provide knowledge
about what groups and group members are like (cf. Berger, 1966; see also
Smith & Henry, 1996).
In line with our conceptualization of stereotyping as a normal functional
process, social categories and stereotypes include the whole gamut of social
groups one can possibly come across, i.e., men and women, old and young,
Black and White, Canadians, Italians and Japanese, doctors and nurses, police-
men and policewomen, bookkeepers, bank clerks, cardrivers and pedestrians,
social psychologists and soccer hooligans, literally any social category you can
think of. Each of these social categories is associated with stereotypical
expectancies. For instance, we know that old people may walk slowly, nurses
take care of the sick, social psychologists are intelligent and disciplined, and
soccer hooligans are loud, aggressive and not intelligent. Social categories and
stereotypes have to do with everyday knowledge. This includes the “knowl-
CATEGORIZATION A N D STEREOTYPING 109

edge” that it may take you longer to buy stamps if there are some elderly
people in the queue in front of you at the post office. Generally, stereotypes
are not problematic. But sometimes they are. Probably that is to some extent
the case in all situations in which perceivers differ as to the content of a
stereotype. And stereotypes are most likely particularly problematic if the
difference in view is between ingroup and outgroup, a topic we are not going
to discuss in the present chapter (but see van Knippenberg, 1984; Ellemers &
van Knippenberg, 1997).
Starting from this general background, in the remainder of this chapter we
aim to discuss some issues concerning social categorization and stereotyping.
In part I we explore the factors that affect social categorization processes. We
Downloaded by [Tufts University] at 13:18 08 October 2014

will discuss in detail the top-down and bottom-up processes in the activation
and use of social categories in social perception. In part I1 we will first, in
line with a functional view on stereotypes, discuss some of the principles that
govern the activation and application of stereotypes. Then we look at the
determinants of stereotype content, again from a functional perspective.
Finally, in Part 111, we will pay attention to some of the mechanisms that
contribute to stereotype persistence.There is much in the way in which stereo-
types operate that sustains their own preservation. In various stages of infor-
mation processing stereotypes seem to have self-confirmatory properties.

PART I: SOCIAL CATEGORIZATION


Spontaneous Social Categorization:An Experimental Paradigm and
a Theoretical Framework
Phenomenologically, social categories are something “out there”. Imagine an
observer describing a few quite mundane events: “An elderly couple crosses
the street. In the distance, a man runs to catch the bus, and the postman rides
past on his bicycle”. In all these ordinary everyday events, quick effortless cat-
egorizations have been made without the observer realizing that these per-
ceptions were not just data-driven. Given the same data, it is conceivable that
someone else could have seen different things: “There were two well-dressed
persons crossing the street. A jogger ran on the pavement, and there was a
man on a bike with a curly moustache”. Actually, one could think of several
more ways to describe these same events.
In most social situations, there are usually many different ways in which the
people that are present can be categorized. Social psychological research into
spontaneous categorization has mostly made use of the so-called “who said
what” or name confusion paradigm. In this paradigm, devised by Taylor, Fiske,
Etcoff, and Ruderman (1978), participants hear or read the statements made
by several stimulus persons who allegedly take part in a group discussion.
110 AD VAN KNIPPENBERG A N D AP DIJKSTERHUIS

These stimulus persons belong to different social categories (as conveyed by


photographs, or first names, or in other ways). Afterwards, participants have
to indicate for every statement who made it (hence, the “who said what” para-
digm). The errors participants make in allocating statements to sources reveal
the spontaneous occurrence of social categorizations. In brief, a social catego-
rization is likely to have occurred to the extent that within-category confu-
sions exceed between-category confusions. Taylor et al. (1978) showed that
within-gender confusions occurred more often than between-gender confu-
sions and that within-race confusions occurred more often than between-race
confusions, suggesting that gender and race categorizations were used to orga-
nize social information in memory.
Downloaded by [Tufts University] at 13:18 08 October 2014

Thus, as has been corroborated in later studies (e.g., Hewstone, Hantzi, &
Johnston, 1991;Stangor, Lynch, Duan, & Glass, 1992), it seems that gender and
race categorizations (and maybe also age) tend to be salient in many social
situations, with the gender categorization probably being the most prominent
among them (Stangor et al., 1992; see also Fiske, 1998;Fiske, Haslam, & Fiske,
1991; Mackie et al.. 1996; van Knippenberg, van Twuyver, & Pepels, 1994).
Why are these categories so salient? What does, in general terms, determine
category salience?
Category salience is theoretically explained as being a function of accessi-
bility and fit. More specifically, categories become salient when they are acces-
sible and at the same time they fit the data (cf. Bruner, 1957; Oakes, 1987).
Accessibility refers to the ease with which a category or construct comes to
mind or, in more specific terms, accessibility refers to the probability that
a category or a construct will be used given appropriate external stimuli (cf.
Bargh, 1994; Bruner, 1957; Higgins, 1989;Wyer & Srull, 1986). Category acces-
sibility may originate from a history of frequent activation, due either to socio-
cultural conditions rendering a social categorization culturally prominent, or
to stable personal constructs or persistent motivations that colour a perceiver’s
view of the world (Kelly, 1955). Such recurrent patterns of activation may
result in the “chronic accessibility” of a category or construct (Higgins, King,
& Mavin, 1982). In contrast, there may be “situationally” or “momentarily”
accessible social categorizations (Higgins, 1989; Higgins et al., 1982) due to
recent activation, or to specific external or internal cues, or to temporary goals
or motives, that make the perceiver particularly susceptible to a specific social
categorization in the situation at hand. Both the chronic and the situational
accessibility of a social category may foster its use in a given situation (van
Knippenberg et al., 1994).
The actual application of a category to a given stimulus configuration
depends not only on accessibility, but also on fit. Fit may be defined as the
match between stimulus features and category specifications (Bruner, 1957;
Oakes, 1987). Oakes distinguished two forms of fit. Structural fit refers to the
co-variation between the categorization and characteristics of the stimuli. For
CATEGORIZATION A N D STEREOTYPING 111

example, if the people in category A are tall, and the people in category B are
small, then there is structural fit. Apart from these structural differences, par-
ticipants may perceive “normative fit”: that is, they see social categories as
meaningful social entities that differ in terms of “specific, normatively appro-
priate dimensions” (Oakes, 1987, p. 131). Thus, normative fit encompasses the
perceiver’s expectancy that the social categories involved are different with
regard to stimulus features that are relevant in that situation. As will become
evident from the empirical overview below, it makes sense to distinguish
between fit as a feature of the actual perceptual data (i.e., structural fit) on the
one hand, and the stereotypical expectancies of the perceiver with regard to
what he or she is about to see (i.e., normative fit) on the other hand. One might
Downloaded by [Tufts University] at 13:18 08 October 2014

say that, functionally speaking, structural fit represents the pragmatic function
of social categorization and stereotyping. That is, the co-variance of our social
categories with relevant stimulus features enables us to behave in a socially
adequate manner. Conversely, normative fit serves a cognitive function in that
it enables us to understand our social environment in terms of knowledge
accumulated in past experiences. Both components may, independently, foster
category salience.

The Role of Chronic and Situational Accessibility


The effect of category accessibility on the salience and actual use of social cat-
egories in person perception and memory has received only scant explicit
attention in empirical research. Chronic accessibility has been put forward to
explain the prominence of sex and race categorizations observed in several
studies (e.g., Stangor et al., 1992). In one of our own studies (van Knippenberg
et al., 1994), we made an explicit comparison between gender categorization
(as an instance of a chronically accessible categorization) and studentlteacher
categorization (as an example of a generally less accessible categorization). In
this experiment, the participants were shown a simulated group discussion to
which each of the 12 discussion participants (stimulus persons) contributed
one statement pertinent to the discussion topic. The stimulus persons in this
simulated group discussion were male and female teachers and male and
female students, so that they could be categorized by their gender as well as
in terms of their academic status (as student vs. teacher). After the participants
had been presented with the statements the stimulus persons had made in the
group discussion, they were given a surprise memory task in which they had
to indicate the source of each statement. The spontaneous use of the gender
and studentkeacher categorizations in processing the discussion information
was derived from the errors participants made in their statement allocations.
The results of the name confusions across various discussion topics are sum-
marized in Table 4.1. O n average, participants made 5.18 within-gender errors
(e.g., mistakenly allocating a statement made by a female stimulus person to
112 A D VAN KNIPPENBERC A N D AP DIJKSTERHUIS

Table 4.1 The effect of discussion topic on gender and academic status categorization
(adapted from van Knippenberg et al.. 1994)
~~ ~

Gender categorization Studentkeacher categorization


Within errors Between errors Within errors Between errors
Issue
Gendcr 5.81 1.68 3.88 3.33
University 4.65 2.93 5.19 2.4R
Irrelevant 5.06 2.04 3.75 3.13
Mean 5.18 2.24 4.32 3.00
Downloaded by [Tufts University] at 13:18 08 October 2014

another female) and only 2.24 between-gender errors (e.g.. confusing a female
source with a male source). The same allocation data revealed 4.32 within-
academic-group confusions and 3.00 between-academic-group confusions.
Using the difference (diff.) within- minus between-category errors as an index
of categorization, it thus turned out that the presumably chronically more
accessible gender categorization had been used to a greater extent (diff. = 2.94)
than the studentheacher categorization (diff. = 1.32).Thus, this study provided
some suggestive evidence that chronic category accessibility enhanced the use
of the category involved.
In two of their experiments, Stangor ef af. (1992) tried to enhance category
salience by priming the categories in question (e.g., by giving explicit instruc-
tions to pay attention to either the gender differences or the race differences),
thus attempting to enhance their situational accessibility. These priming
manipulations failed, however. to enhance the actual use of these categoriza-
tions in a subsequent who-said-what study, presumably because the catego-
rizations studied (i.e., gender and race) were already chronically highly
accessible. which made it difficult to enhance their accessibility further by tem-
porary measures. In a later study, van Twuyver and van Knippenberg (1995)
selected categorizations that were chronically less accessible to begin with (as
established in a pilot study). In this experiment, the stimulus persons in the
simulated group discussion could be categorized according to university major
(psychology vs. law) and university town (Nijmegen vs. Amsterdam). Prior to
the presentation of the group discussion, the priming manipulation was
administered by having participants in an ostensibly unrelated study fill out
an extensive questionnaire about attitudes and preferences of either psychol-
ogy and law students or Nijmegen and Amsterdam students. It was found that.
on average, participants made relatively more within-group errors (5.53) and
fewer between-group errors (3.54) for the primed categorization than for
the non-primed alternative categorization (4.88 within category errors and
4.05 between category errors). Thus, priming seems to foster category use
CATEGORIZATION AND STEREOTYPING 113

(diff. = 1.99), presumably by enhancing the situational accessibility of the


primed categorization compared to the non-primed alternative categoriza-
tion (diff. = 0.83).

Stereotypical Expectancies Enhance Category Use


When we introduced the term “fit” above, we suggested that it makes sense to
distinguish between structural fit, as “something in the data” (i.e., actual co-
variance between stimulus features and the categorization), and normative fit,
as something the perceiver expects (i.e., “stereotypically” expected co-variance
between stimulus features and the categorization). We want to emphasize here
Downloaded by [Tufts University] at 13:18 08 October 2014

that stereotypical expectancies are important determinants of spontaneous


categorizations, at least in the who-said-what paradigm, but probably in any
categorization paradigm in which expectancies shape perception and memory
(see also Part I11 of this chapter).
Observing a simulated group discussion, as experimental participants in the
who-said-what paradigm do, is a sequential “on-line” process in which a social
categorization-selected at early stages of the process-may be used to
organize perception in order to make sense of what is going on. Thus, the
dominant social categorization will tend to be imposed prior to an apprecia-
tion of what stimulus persons are actually contributing to the discussion (that
is, prior to the establishment of actual co-variance between stimulus features
and the categorization in the perceived event). Thus, the perceiver’s stereo-
typical expectancies may play a decisive role in the selection of the social cat-
egorization used to process the group discussion, largely irrespective of
whether or not there is actual co-variance between relevant stimulus features
and the categorization.
In the van Knippenberg et al. (1994) experiment briefly described earlier,
we manipulated expectancies by varying the discussion topic. By means of a
pilot study among students from the same population as our later participants,
we selected three topics: one that we called the gender issue, because men and
women were expected to have different views about it (i.e., “Positive dis-
crimination in favour of women”), one university issue, because student and
teachers were expected to differ (i.e., “More severe consequences of course
evaluations for teachers”) and one neutral issue, on which neither men and
women nor students and teachers were expected to differ (i.e., “More rigor-
ous measures against social welfare fraud”). Participants were randomly allo-
cated to a simulated group discussion in which 12 stimulus persons (male and
female teachers and male and female students) discussed one of these topics.
In Table 4.1 the results of the source confusion data are shown. As hypothe-
sized, a strong gender categorization (5.81 within-gender errors, 1.68 between-
gender errors; diff. = 4.13) was observed when the participants watched a
discussion about the gender issue while there was virtually no studentlteacher
114 A D VAN KNIPPENBERG A N D AP DIJKSTERHUIS

categorization in this condition (diff. within- minus between-academic-


status errors = 0.55). Conversely, in the university issue condition, the aca-
demic status categorization (diff. = 2.71) was significantly stronger than the
gender categorization (diff. = 1.72). It is interesting, by the way, that partici-
pants watching the discussion on the neutral topic (on which neither catego-
rization was expected to yield a difference in attitude) still showed a stronger
gender categorization (diff. = 3.02) than studentlteacher categorization (diff. =
0.62). We come back to the implications of this latter finding in the next
section.
On the basis of these results of van Knippenberg et al. (1994) we may con-
clude that participants spontaneously use a social categorization which they
Downloaded by [Tufts University] at 13:18 08 October 2014

expect to co-vary with relevant features of the stimulus persons (in this case,
their attitude towards the issue discussed). It is worth emphasizing that these
stereotypical categorizations occurred irrespective of any real co-variance
between these categorizations and the data (i.e.. the positions advocated in
the stimulus persons‘ discussion statements). In other words. the existence of
a stereotypical expectation that the people to be observed differ in terms of
relevant features in the situation suffices to make that categorization salient
and to make perceivers actually use that categorization in perceptually
structuring the perceived events. Actual co-variance seems unnecessary for
this effect to occur, that is, the described effect of discussion topic on sponta-
neous social categorization did not interact with the manipulated distribution
of standpoints of the stimulus persons (the effects of which we will discuss
later).

Some Social Categorizations Have Normative Fit in Many Situations


In the neutral issue condition described above, neither men and women nor
students and teachers were expected to have different attitudes (as indicated
in the pilot study). Yet, the source confusions of the participants in the
experiment revealed that they nevertheless used the gender categorization
in processing the simulated discussion. Similarly, spontaneous gender and
race categorizations have been observed in several other studies in the absence
of explicit differential (stereotypical) expectations concerning category
members’ attitudes (Taylor et uf.. 1978; Stangor et af.. 1992). So perceivers
appear to organize their perception and memory spontaneously in terms of
gender and race categories, despite the fact that these categories seem irrel-
evant in that specific perceptual context. Earlier (cf., van Knippenberg rt af..
1994), we attributed these findings to the high chronic accessibility of the
categories involved, which made perceivers use them as a default option.
However, a recent study by Blanz (1999) suggests that there may be sorne-
thing like a “general expectancy of intercategory differences” (or “general”
normative fit) that may tempt people to use that categorization even though
there is no specific apparent reason to use it.
CATEGORIZATION A N D STEREOTYPING 115

The three topics that had been studied in the van Knippenberg el al. (1994)
study, were preselected on the basis of a pilot study. The gender issue (“posi-
tive action in favour of women”) was selected because participants indicated,
on average, that they expected men and women to have different opinions
about this issue. This led us to argue that the gender categorization was used
in processing the “positive discrimination” discussion because our partici-
pants-maybe virtually every one of them-expected to see intercategory dif-
ferences. On the neutral issue (“social welfare fraud”) the pilot study showed,
on average, no expected gender difference. Yet, it is conceivable that there
were nevertheless many participants who expected men and women to differ,
one way or the other, in their views on this topic (without yielding a clear
Downloaded by [Tufts University] at 13:18 08 October 2014

average difference in one specific direction), and it may be the case that these
participants were responsible for the significant gender categorization in the
neutral issue condition.
Our present argument is based on the recent work of Blanz (1999), who
introduced a sort of “generalized expectancy of intercategory differences”
(which he called “stored stereotype” or “stored meta-contrast”, Blanz, 1999,
p. 44) to explain the prevalence of chronically accessible categories (such as
gender) across many situations. To summarize his view, perceivers need not
expect specific differences between the categories (e.g., concerning the atti-
tude towards the issue discussed), but it may suffice that they see the cat-
egories as different in many respects to make this categorization salient in a
great variety of situations.
The predictive value of this “generalized expectancy” was empirically sup-
ported. Blanz (1999, expt. 1) found a significant gender categorization in a
name confusion experiment in which the topic of the simulated discussion
was ?he application of Berlin for the Olympic Games in 2000”. In a post-
questionnaire he measured the participants’ generalized stereotypical expec-
tations (“stored meta-contrast ratio”) by means of a seven-point scale item:
“In my opinion there are generally strong differences between males [. . .] and
females [. . .]” (Blanz, 1999, p. 56). This generalized expectancy correlated
highly ( r = 0.58) with the observed gender categorization (i.e., with the dif-
ference within minus between gender errors). It is worth noting that this
general expectancy of gender differences was a better predictor of the gender
categorization than the specific expectancy that men and women would have
different attitudes towards the issue discussed, i.e., the application of Berlin
for the Olympics 2000 ( r = 0.38). Also, after applying a median split on the
general expectancy measure, Blanz found that the observed significant gender
categorization was completely due to participants having a high score on this
generalized expectancy measure, while it was absent for participants with a
low score. So, in other words, gender-as a chronically salient social catego-
rization-was used by those participants who believed that men and women
differed “in general”. In addition, it appeared that participants who did not
have these expectations showed no significant gender categorization.
116 AD VAN KNIPPENBERG AND A P DIJKSTERHUIS

It seems plausible that the spontaneous categorizations in terms of gender


observed in other research (e.g., Stangor et al., 1992; van Knippenberg et al.,
1994) were also due to those participants who expected men and women to
differ “in general”. More generally speaking, it is plausible that empirical
demonstrations of the spontaneous use of chronically accessible social cat-
egories, even in the perception of events in which there is no specific reason
to expect intercategory differences, may be ascribed to those participants who
(implicitly or explicitly) believe that the categories are different in many
important respects.
There are a few suggestive indications that individual differences in beliefs
or expectancies concerning the categories involved play a role in the sponta-
Downloaded by [Tufts University] at 13:18 08 October 2014

neous use of some chronic social categorizations. For instance, it was found
that gender categorizations were used less by androgynous people than by sex-
typed people (Frable & Bem, 1985;Taylor & Falcone, 1982). It was also shown
that individual differences in racial prejudice were positively related to social
categorizations in terms of race, that is, high-prejudice participants used race
categorizations, while low-prejudice participants did not (Stangor et al., 1992,
expt. 3; see also Secord, Bevan, & Katz, 1956). It is quite probable that these
individual difference measures correlate with generalized beliefs about inter-
category differences and intracategory similarities on relevant dimensions.
Thus, in line with a functional view on social categorization, in these cases too
perceivers probably spontaneously use those social categorizations they
expect to co-vary with important features of the stimuli, i.e., they use the social
categories that are meaningful to them in that social situation.

The Relationship between Accessibility and Nonnative Fit


As outlined above, it seems that chronically accessible categorizations like
gender and race tend to be used by perceivers who stereotypically believe that
there are many important intercategory differences. A theoretically interest-
ing implication of this observation may be that chronic accessibility, or “cat-
egory chronicity”, is particularly characteristic of perceivers who have these
stereotypical beliefs (cf. Blanz, 1999). That is, although chronic category
accessibility and stereotypical expectancies are conceptually distinct, they may
theoretically and empirically be strongly related.The belief that two categories
are generally different becomes stronger to the extent that there are
more dimensions of intercategory difference in the perceiver’s mental reper-
toire. If there are many such dimensions of difference, and these pertain to
important social domains, the likelihood that these categories will be used will
increase. As a result, the categorization in question may become chronically
accessible.
Thus, chronic accessibility may develop over time in perceivers who believe
CATEGORIZATION AND STEREOTYPING 117

the categories involved to differ in many important respects. Just like trait
chronicity (Higgins et al., 1982), the chronic accessibility of social categories
is probably an individual difference variable that affects the perception and
interpretation of social situations by the individual involved in important ways.
Also like trait chronicity, chronic category accessibility is presumably not a
dichotomous variable, dividing the world of perceivers into, for example, sexist
and gender-blind people, but a continuous variable (Higgins & Brendl, 1995).
People most likely vary as to the degree to which (or the number of dimen-
sions on which) they believe, e.g., men and women to be different and they
will tend to apply this categorization to social events accordingly.
It may thus be argued that social categorizations like gender and race are
Downloaded by [Tufts University] at 13:18 08 October 2014

used by perceivers for whom these categories have normative fit for a great
variety of social situations. Such chronically accessible categories have a wide
range of application, that is they are chronically accessible to those perceivers
that believe these categories to be pertinent to many different aspects of social
life. It may be useful t o speak of “general normative fit” to indicate the stereo-
typical expectancies associated with such chronically accessible social catego-
ries, in order to distinguish them from generally non-salient social categories
that only become salient in specific social situations in which they have nor-
mative fit. In the latter case, we may speak of situationally accessible social
categories that tend to be evoked by situational cues in specific situations, and
that are usually appropriate to be used in these specific situations, that is. they
have “specific normative fit”.

Structural Fit Sometimes Generates “Stereotypes” Where They


Did Not Exist
So far we have elaborated on the importance of accessibility and normative
fit in eliciting spontaneous social categorizations. It appears that what we see
is very much determined by the concepts that come to mind easily (accessi-
bility) and their expected appropriateness (normative fit). These top-down
processes strongly determine the social categorizations we use in structuring
our perceptions of social events. What, then, is the contribution of bottom-up
factors? More specifically, what is the contribution of “structural fit”, that is,
of real co-variance between the categorization and the data?
It makes sense to use a social categorization that has a high level of struc-
tural fit. For instance, if an observer of a simulated group discussion notices
that all members of group A are in favour of some specified attitudinal
position and all members of group B are against (that is, there is perfect struc-
tural fit), the A/B categorization seems highly useful in order to understand
properly what is going on. The effects of structural fit on spontaneous social
categorizations have, however, received only scant empirical attention. Van
118 A D VAN KNIPPENBERG A N D AP DIJKSTERHUIS

Knippenberg et al. (1994) included “structural fit” as a factor in their experi-


ment, by varying the standpoints advocated by the stimulus persons. Three
levels of fit were induced. In one condition there was malelfemale fit, that is
male and female stimulus persons expressed opposing viewpoints (e.g., all
males were in favour and all females against, or the other way around); in a
second condition there was studentheacher fit, and finally there was a condi-
tion with no fit at all (i.e., zero correlations between the categorizations and
the stimulus persons’ standpoints). This fit manipulation had a strong effect
on the observed social categorizations (see Table 4.2). Male/female fit elicited
a strong gender categorization (diff. within- minus between-gender errors =
5.13), while in this condition there was no studentkeacher categorization
Downloaded by [Tufts University] at 13:18 08 October 2014

(diff. = -0.03). Studentkeacher fit facilitated the use of the academic status cat-
egorization (diff. = 2.84) while it reduced the gender categorization (diff. =
1.39). In the no-fit condition, there was a moderate gender categorization
(diff. = 2.12) and a clearly weaker studentheacher categorization (diff. = 1.25).
It should be noted that the fit manipulation in this study was rather extreme,
that is, in the conditions in which there was structural fit the categorization
fully “explained” (r = 1.0) the positions the stimulus persons advocated in the
discussion (i.e., there was perfect fit). In a recent study, Blanz (1999, expt. 2)
used a weaker fit manipulation (contrasting r = 0.5 with r = 0), and found no
effect of this fit manipulation on the categorizations he used (school pupils vs.
college students; home town), unless the categorization in question was also
primed (in a way similar to ours, cf., van Twuyver & van Knippenberg, 1995).
Blanz thus showed that weak structural fit alone may not be sufficient to make
perceivers use a social categorization; it helps if this categorization is also made
salient by priming.
Thus, it appears that either spontaneously, or helped by priming with the
category, perceivers seem quite apt to notice structural fit and to use the cat-
egorization involved in processing the observed event. Interestingly, both our
structural fit effect and the primed structural fit effect obtained by Blanz may

Table 4.2 The effect of structural fit on gender and academic status categorization
(adapted from van Knippenberg et al., 1994)

Gender Studentlteacher
categorization categorization
Within Between Within Between
errors errors errors errors
Fit
Male/female fit 6.40 1.27 3.58 3.61
Studentlteacher fit 4.17 2.78 4.96 2.12
No fit 4.92 2.80 4.44 3.19
CATEGORIZATION AND STEREOTYPING 119

tell us something about the genesis of a stereotype. Inherent in the name con-
fusion paradigm is that perceivers have this categorically organized informa-
tion stored in their memories, at least for a little while. It is worth noting that,
if you take our definition of stereotypes seriously, this means that in the short
period in which the participants processed the discussion information, a
stereotype has been formed (because a stereotype is defined as the mental
representation of a social group, and because the memory of the observed
group discussion satisfies this description; it is a stereotype). Such stereotypes
formed on the spot may be short-lived on all those occasions in which the
co-variation of category and stimulus features turns out to be merely acci-
dental, i.e., when this experience of co-occurrence is not repeated on subse-
Downloaded by [Tufts University] at 13:18 08 October 2014

quent occasions. But when the observed regularity is encountered more often,
the stereotype may further develop and strengthen, and evolve into a stable
cognitive structure over time (see part I1 on stereotype content; Dijksterhuis
& van Knippenberg, 1999;see also Sherman, 1996).Although this form of asso-
ciative learning may not be the only way in which stereotypes are acquired-
it may not even be the most important way-the studies cited show that
perceivers may under the right circumstances be quite capable of forming new
stereotypes.

Social Categorizations Tend to Be Competitive


As we explained at the beginning of part I of this chapter, in most social
situations the people present can be categorized in different ways. In most of
the name confusion studies discussed above, that feature of real-life situations
was mimicked by using crossed categorizations. That not only makes the situ-
ation more realistic, but it also enables us to study the question of which of
the available categories will actually be activated. An interesting feature of
categorization that we have so far not discussed is that categorizations tend to
be mutually competitive. That is, if one of them gets the upper hand, the
other(s) tend to be pushed aside or inhibited.
Theoretically, one could apply several social categorizations at the same
time. But from a functional viewpoint, that might be counter-productive. The
combined knowledge of many categories activated simultaneously might
seriously undermine our conceptual understanding of an observed event. An
example may illustrate this point. Take, for instance, a Dutch social psycholo-
gist who is also a dedicated fan of the Dutch national soccer team. If you
happen to come across this guy dressed in an eyecatching orange outfit, yelling
and shouting while kicking empty beer cans around, it suffices to activate the
knowledge that he is a Dutch soccer fan (note that orange is the colour of the
Dutch soccer team) in order to understand what’s going on. Now, things might
become quite confusing if you simultaneously activated the knowledge that
he is a social psychologist, since these are mostly intelligent and disciplined
120 A D VAN KNIPPENBERG A N D A P DIJKSTERHUIS

people. So, on such an occasion, while it is highly functional to activate the


Dutch soccer fan category-and the attendant stereotype-it is very confus-
ing to also activate the social psychologist category. In fact, for a proper under-
standing of the event, we had better not simply ignore the social psychologist
categorization, but actively suppress it.
Macrae, Bodenhausen, and Milne (1995; see also Bodenhausen and Macrae,
1998) investigated this multiple-category problem. They argued that, upon
meeting a fat Irish priest, we use one category (e.g., Irish) and actively inhibit
the others (e.g., man, priest, fat). That is, if for some reason one category (e.g.,
Irish in the context of an England-Ireland rugby match) is more appropriate
than the others, this particular category will be activated and used at the
Downloaded by [Tufts University] at 13:18 08 October 2014

expense of the others. These other categorizations are then actively prevented
from getting access to the perceptual interpretation process. In one of their
studies, Macrae et 01. (1995) either primed their participants with the female
stereotype, or with the Asian stereotype, or they were not primed (i.e.. the
control group). Later, the primed participants watched a video of a Chinese
woman, while the control group saw a video clip from a wildlife show. In
a subsequent lexical decision task, it was shown that only one social category-
the one that was primed-was activated. The other social category was inhib-
ited.That is. compared to the participants in the control group, the participants
primed with the Asian stereotype showed enhanced activation of this stereo-
type (faster responses) and reduced activation of the female stereotype
(slower responses) and, conversely, participants primed with the female stereo-
type showed enhanced activation of the female stereotype and reduced
activation of the Asian stereotype. In sum, people (generally) do not
apply multiple social categories at the same time. Instead, several factors
(e.g., context, individual differences) will render one of the available social
categories more salient at the expense of the activation level of its
competitors.
There are some indications of similar competitive categorization patterns
within the name confusion paradigm. It seems that, in studies in which there
are two social categorizations available, the use of one of the two implies that
the other one is used less. Van Twuyver and van Knippenberg (1999) found a
moderate, but significant, negative within-cell correlation (on average r =
-0.25) between the use of the gender categorization (as indicated by within-
minus between-category confusions) and the use of the studentheacher cate-
gorization. Such a negative interdependence (or functional antagonism, cf.,
Turner, 1987) between the use of different social categorizations may be
a surface manifestation of competitive cognitive facilitation and inhibition
processes. Note that one could use two categorizations simultaneously, that is,
perceivers may on occasion use subtypes, or subcategories, cf. van Twuyver and
van Knippenberg (1998). Within the confines of our present functional argu-
ment, this would be predicted to occur to the extent that perceivers believe
CATEGORIZATION AND STEREOTYPING 121

that this particular conjunction of categories is informative, that is, has nor-
mative fit.
Elsewhere (Dijksterhuis & van Knippenberg. 1998), we have argued that
the inhibition of competing social categories (i.e., competing knowledge) is
highly functional. Inhibition plays an important role in language comprehen-
sion. When we read the phrase “he thrusts the spade into the soil”, we easily
understand that the protagonist is working with a digging implement and not
with a playing card.This latter, alternative meaning of “spade” is actively inhib-
ited on reading this sentence (cf. Gernsbacher & Faust, 1991). Similarly, social
categorizations are elicited in the service of a specific interpretation of stimu-
lus input. We categorize in order to “understand” what happens around us.
Downloaded by [Tufts University] at 13:18 08 October 2014

Alternative categorizations may create ambiguity; they tend to blur percep-


tion. In order to get a clear picture, alternative ways of interpreting events
must be pushed down, they must be inhibited. Thus, both in language com-
prehension and in social perception, inhibition is an important process because
it helps to disambiguate perception.

Social Categorizations:Some Conclusions


In part I of this chapter, we have elaborated on the antecedents of sponta-
neous social categorizations. In line with Bruner’s (1957) and Oakes’s (1987)
theorizing, the accessibility of a social categorization in the perceiver’s mind
appears to be an important determinant of social category selection. Some
social categorizations, such as gender and race, seem to be accessible all the
time. that is, they are chronically accessible, at least to some perceivers. Other
categorizations that are generally less accessible can be made temporarily
accessible by priming, which enhances their salience and use in the perception
of social events (van Twuyver & van Knippenberg, 1995).
We have further argued that in many situations categories are used that
have normative fit, i.e., people tend to use social categories which they expect
to co-vary with relevant interstimulus differences in the situation at hand.
Gender and race are examples of social categorizations that have normative
fit for many people in many situations. Other categorizations, e.g., students and
teachers, have specific normative fit, that is, one expects them to be useful in
specific social situations, e.g., when students and teachers discuss issues that
pertain to the university. It seems that in all these cases spontaneous category
elicitation is guided by stereotype content. The subjective stereotypical expec-
tation that pertinent intercategory differences and intracategory similarities
exist may well be the most important determinant of everyday social category
use.
We suggested that the extent to which a social categorization, such as
gender, has “general” normative fit, i.e., is subjectively believed to co-vary with
important stimulus features (e.g., appearance, traits, attitudes, or whatever a
122 AD VAN KNIPPENBERG A N D A P DIJKSTERHUIS

perceiver may be interested in), may be regarded as an individual difference


factor. It seems plausible that these individual differences in normative fit go
hand in hand with “category chronicity”. That is, it seems plausible that per-
ceivers who believe, for example, that men and women differ in many respects,
tend to use this categorization across a great variety of social situations and,
due to its frequent use, the gender categorization becomes chronically acces-
sible. It may thus be argued that, eventually, perceivers will preferentially acti-
vate those social categories they have strong stereotypes about. Thus, the
existence of strong stereotypes in a perceiver’s mental repertoire may be a
causal factor in the use of the social category in question in the perception of
social events.
Downloaded by [Tufts University] at 13:18 08 October 2014

Finally, there is some evidence in the literature that the use of one social
categorization tends to inhibit the use of other categorizations. In a sense, the
meaningful interpretation of social reality (that is, a particular emphasis on
specific intercategory differences and intracategory similarities) tends to push
aside other potentially meaningful ways of interpreting the same stimulus
situation.
So far, we have studied categorical perception processes mainly from the
perspective of spontaneous categorization. Below, we will look into some com-
plementary mechanisms that operate in stereotyping.

PART 11: STEREOTYPE ACTIVATION


AND APPLICATION
When one meets a person belonging to a social category one has stereotypes
about, does that inevitably lead to the activation of the associated stereotype,
and to the application of this knowledge in perception and judgement? We
will discuss these issues below. First, we look into some of the basic determi-
nants of stereotype activation.

Is Stereotype Activation Inevitable?


There has been a debate in the literature as to whether perceivers always fall
prey to the automatic activation of stereotypes whenever they encounter a
member of a stereotyped group. Devine (1989) argued that stereotype activa-
tion is automatic. She presented participants subliminally with the category
“Blacks” and some of its stereotype content (but no hostility-related words)
and found subsequent ratings of a target person to be more extreme on
hostility-related scales. These indications of stereotype activation were
observed independently of the participant’s level of prejudice. Devine’s con-
clusion was that upon being confronted with a social category, the attendant
stereotype gets activated automatically, and there is nothing the perceiver can
CATEGORIZATl ON AND STEREOTYPING 123

do to prevent this. This view is still quite prevalent in current thinking about
stereotypes, but it has not gone unchallenged.
Some empirical studies suggest that when confronted with a stimulus person
who clearly belongs to a specific social category, for example an Asian person,
the associated stereotypes may not necessarily become activated. Gilbert and
Hixon (1991) obtained results suggesting that cognitive load may prevent the
activation of stereotypes. In their first experiment, they had participants watch
a video of either an Asian or a Caucasian assistant. On the video, the assistant
turned cards with word fragments on them, which the participants were
instructed to complete (thus providing the dependent measure, i.e., the extent
to which the word fragment completions were stereotypic of Asians). Some
Downloaded by [Tufts University] at 13:18 08 October 2014

participants watched this video while they were cognitively busy (rehearsing
an eight-digit number), while others watched it under normal circumstances.
As it turned out, the cognitively busy participants who watched the video of
the Asian assistant did not activate the stereotype of Asians (which their non-
busy counterparts did, as shown by their stereotypic word completions). Thus,
it seems that stereotype activation does not automatically and unconditionally
follow upon being exposed to a member of the stereotyped group. Recently,
Spencer, Fein, Wolfe, Hodgson, and Dunn (in press) replicated this finding for
the stereotype of African-Americans. They also showed, however, that when
participants had just received negative self-relevant feedback (and therefore
might need to boost their self-esteem), the stereotype was activated despite
their cognitive busyness.This latter result suggests that although cognitive load
may normally prevent stereotype activation, there may exist overriding prior-
ities that make them pop up anyway.
Macrae, Bodenhausen. Milne. Thorn, and Castelli (1997) demonstrated that
stereotype activation is dependent on the goals with which information is
processed.They presented their participants with photographs of either female
undergraduate students or household objects. On some photographs, a white
dot appeared. Some participants received a non-semantic processing goal: they
were simply asked to report for each photograph whether it contained a white
dot or not. Others received a semantic processing goal: they were asked to
decide whether each photograph depicted an animate or an inanimate object.
After each photograph participants did a lexical decision task to measure
stereotype activation. From the results it appeared that participants who
processed the photographs semantically did activate the stereotype but the
others (participants in the dot condition) did not.
The results of the studies of Gilbert and Hixon (1991) and Macrae et al.
(1997) present a problem for theories claiming that stereotypes are activated
automatically and unconditionally upon meeting someone who belongs visibly
(or otherwise obviously) to a stereotyped group. At this stage, the underlying
dynamics are still obscure. It is conceivable that at some level of attention the
category (e.g., Asian, female), or defining features of it, are registered on-line,
124 A D VAN KNIPPENBERG A N D A P DIJKSTERHUIS

enabling participants to report later the correct category membership of the


stimuli. Despite this early category detection, however, it might be that the
subsequent activation of the associated stereotype is being prevented because
of lack of resources or incompatibility with the current processing goal. Alter-
natively, the social category activation itself may have become inhibited
(maybe immediately following early pre-conscious activation) because of com-
peting task priorities (e.g., rehearsing eight-digit numbers, spotting white dots),
as a result of which the associated stereotype gets inhibited too (i.e., secondary
inhibition, see Anderson & Spellman, 1995; Dijksterhuis & van Knippenberg.
1998). The results of Spencer et af. (in press) suggest that at least lack of
resources does not seem to be decisive, since in the same cognitive load con-
Downloaded by [Tufts University] at 13:18 08 October 2014

dition participants did activate the stereotype when they apparently needed
to do so. Thus, considering the present evidence, it seems that whether or not
stereotypes get activated may somehow depend more on task priorities (i.e..
processing goals) than on resources per se.
With regard to the studies described above, it is still unclear whether under
conditions of high cognitive load or incompatible processing goals, the social
category of the stimulus persons became activated to begin with. Both Gilbert
and Hixon (1991) and Macrae et al. (1997) observed that participants were
able to report the correct category later, but these memory data do not have
the same compelling quality as the word stem completions or the lexical de-
cision data showing the absence of stereotype activation. This inconclusive-
ness as to what really happened was also expressed by Gilbert and Hixon
(1991) in their discussion section. They wrote:

Our studies do not deny that [such] rapid categorizations can and often do
happen. Our busy subjects did not experience activation of racial stereotypes
about an Asian assistant to whom they were exposed, and although these sub-
jects were able to recall the assistant’s race after the experiment was over, they
apparently did not categorize her in terms of race during the experiment. And
why should they have? (Gilbert & Hixon, 1991, p. 515).

In the above studies the visual stimulus conveys category membership, i.e.,
information which in principle can be either picked up, or ignored or even
inhibited depending on cognitive load, visual attention, and processing
priorities. Consistent with ideas about competitive category selection dis-
cussed earlier, it may be suggested that competing processing goals (maybe in
combination with limited resources) may prevent a social category from being
activated, as a result of which the associated stereotype also does not become
activated.
Thus, in line with our functional approach to social categorization and
stereotyping, we see social category activation and stereotype activation as
conditionally automatic processes, that is, they tend to get activated when
whatever we are doing at the time requires the sort of information stereotypes
CATEGORIZATION A N D STEREOTYPING 125

offer (e.g.. knowledge about category members’ traits, stereotypical expecta-


tions). In other task situations, i.e., situations in which the stereotypic infor-
mation is utterly irrelevant, stereotype activation may be suppressed. This idea
is pointedly expressed in the following functional argument used by Macrae
et al. (1997). If we are in a supermarket we usually come across many people,
but we treat them mostly as objects. We have to navigate around them like we
have to navigate around posts. In such a situation, it would be dysfunctional
to activate stereotypes. since we are not interested in these people (as people)
anyway.
Now, it may well be that our analysis of situations in which stereotypes are
not activated fails to address the real problem underlying the question of the
Downloaded by [Tufts University] at 13:18 08 October 2014

inevitability of stereotype activation. What may be complicating the issue dis-


cussed is that the question of the inevitability of stereotype activation is prob-
ably asked in the context of its social undesirability. We merely argue here that
in situations in which perceivers have other things on their minds, the activa-
tion of a social category and the stereotypes associated with it may be pre-
vented or inhibited. In other situations, e.g., situations in which semantic
processing goals are in place, or perceivers are in the process of judging target
persons, it may well be that stereotype activation does inevitably occur
(although not necessarily in every perceiver). We suppose that much of the
discussion on the motivated abstinence from stereotype use pertains to the
domain of stereotype application.

Do We Always Apply Stereotypes?


Do we always use stereotypes when we possibly can? The answer is that we
do not necessarily use a stereotype merely because some social category is
available on the basis of which we could in principle stereotype. As Gilbert
and Hixon (1991) note, although activation of a stereotype may increase the
likelihood of its use. “it does not mandate such use, nor does it determine the
precise nature of its use” (p. 512).We can, within a functional framework, iden-
tify several circumstances in which stereotyping will not take place, even when
we know a target person’s social category membership.
First, under some circumstances stereotypes are not informative with regard
to what we want to know or want to do. Sometimes, for the purposes of our
anticipated interaction, the generally global information associated with the
category membership of the target person (i.e., the stereotype) may not be
very helpful. In these conditions perceivers may prefer to use other informa-
tion instead of the stereotype. Erber and Fiske (1984), for instance, showed
that when participants have to form an impression of a person, the stereo-
typicality of these impressions decreases when they are told that they have to
work with this person on an important task. Basically, if the perceiver needs
specific information that is not contained in the stereotype, e.g., related to the
126 AD VAN KNIPPENBERG A N D AP DIJKSTERHUIS

expectation that one has to work with that person, they may use stereotypes
to a lesser extent (see also Neuberg & Fiske, 1987; Fiske & Neuberg, 1990;
Pendry & Macrae, 1994).
Second. stereotypes are more useful under some circumstances than under
others. Bodenhausen and Wyer (1985; see also Bodenhausen, 1988; Boden-
hausen & Lichtenstein, 1987) conceive of stereotypes as judgemental heuris-
tics. As with heuristics and other mental shortcuts, they are useful, because
using them only requires few cognitive resources. As a consequence, stereo-
types are even more useful when for some reason we have limited resources
available. Therefore, when we are faced with a difficult judgement, or when
cognitive load is high, or when time pressure is high, or when for some
Downloaded by [Tufts University] at 13:18 08 October 2014

other reason we did not pay sufficient attention to relevant information, we


apply stereotypes to a larger extent than under normal circumstances
(Bodenhausen, 1988; Bodenhausen & Lichtenstein, 1987; Dijksterhuis &
van Knippenberg, 1995; Gilbert & Hixon, 1991, expt. 2; Macrae, Hewstone &
Griffiths, 1993; Macrae, Milne & Bodenhausen, 1994; Stangor & Duan, 1991;
D. van Knippenberg & A. van Knippenberg, 1994; van Knippenberg, Dijk-
sterhuis & Vermeulen, 1999).
Third, we can refrain from stereotype use because we want to (or because
other people want us to). The application of negative stereotypes has negative
consequences and the use of these stereotypes is often unfair. In recent years,
a lot of research has been devoted to the question of whether stereotype use
can be wilfully controlled. We do activate them, sure, but can we suppress
these activated stereotypes before they affect our judgements and our behav-
iour? Recently, this line of research has been thoroughly reviewed (Monteith,
Sherman, & Devine, 1998). In our view, the current state of affairs-admit-
tedly oversimplified-is that stereotypes can indeed be suppressed with some
success. Early experiments showed that stereotypes can be suppressed upon
the explicit instruction of an experimenter to do so (Macrae, Bodenhausen,
Milne, & Jetten, 1994; Monteith, 1993). Furthermore, some of these studies also
demonstrated potential costs of stereotype suppression. Macrae er al. (1994)
showed that stereotype suppression can lead to a rebound effect (cf. Wegner,
1994). rendering the stereotype highly accessible upon removal of the goal to
suppress (but see Monteith, Spicer, & Tooman, 1998). Later work, however,
gives rise to more optimism regarding the efficiency of stereotype suppression,
as several researchers have shown that the process of stereotype suppression
itself can be automatized (Kawakami, Dovidio, Moll, Hermsen, & Russin,
1999; Macrae. Bodenhausen, & Milne, 1998; Moskowitz. Gollwitzer, Wasel, &
Schaal, 1999). Automatized suppression makes the process of suppression
more efficient and possibly prevents rebound effects from occurring.
The relation between stereotype activation and application is not a one-to-
one relation. First, application is to some extent gradual: we apply stereotypes
but under some circumstances we apply them more than under other circum-
stances, depending among other things on how much we need them to cope
CATEGORIZATION AND STEREOTYPING 127

with the task at hand. Second, negative stereotypes are maybe not always
applied simply because people do not want their judgements and behaviour
to be (or appear) biased and prejudiced. The latter suggests that stereotype
application is not just a function of task requirements and resources, but also
of higher order goals (e.g., social desirability, social approval). It is possible
that, if one finds oneself frequently in situations in which stereotype suppres-
sion is desirable, the act of suppression may become automatized.

Individual Differences in Stereotype Activation and Application


It has been suggested that in the case of gender and race the automatic
Downloaded by [Tufts University] at 13:18 08 October 2014

activation of stereotypical knowledge would affect subsequent judgements,


even in those perceivers who have egalitarian beliefs and are not prejudiced
(Devine, 1989). However, several recent publications suggest that there are
prejudice-related individual differences in the activation and use of stereo-
types (Blair & Banaji, 1996: Dovidio, Kawakami, Johnson, Johnson, &
Howard, 1997;Lepore & Brown, 1997;Wittenbrink, Judd, & Park, 1997).While
in order to activate stereotypes, Devine (1989) used category labels as well as
derogatory labels and bits of stereotype content to activate the stereotype,
Lepore and Brown (1997) separated labels from content. They showed that
priming with only the category label induced the application of stereotypical
knowledge for prejudiced subjects, but not for participants low in prejudice.
In line with Devine (1989). they showed that both high- and low-prejudiced
participants knew the stereotype content, but apparently the category prime
only fostered stereotype use in social judgement by highly prejudiced subjects.
Similarly, using subliminal “Black” and “White” primes in a lexical decision
task, Wittenbrink, Judd, and Park (1 997) showed that automatic stereotype
activation correlated with explicit measures of prejudice.
Our present argument concerning spontaneous social category activation
not only supports the latter line of reasoning. but it also adds a new element.
Wittenbrink et af. (1997) and Lepore and Brown (1997) show that neither the
automatic activation of stereotypical knowledge nor its application to social
judgement is inevitable, that is, low-prejudice participants seem not (or less)
susceptible to these effects. As we argued in part I of this chapter, our present
claim goes a bit further to suggest that even the spontaneous cognitive acti-
vation of the social category in question may not occur (or may occur less) in
participants who do not have a priori beliefs about pertinent intercategory dif-
ferences (i.e., who do not endorse the stereotypes in question).

Determinants of Stereotype Content


There is a large amount of experimental evidence showing that the activation
of a social category enhances the cognitive accessibility of the stereotypical
traits that are associated with it (cf. Blair & Banaji, 1996; Devine, 1989;
128 A D VAN KNIPPENBERC A N D AP DIJKSTERHUIS

Dovidio, Evans, & Tyler, 1986; Dijksterhuis & van Knippenberg, 1996; Gaert-
ner & McLaughlin, 1983; Macrae, Stangor, & Milne. 1994). It is therefore some-
what surprising that little is known about why and how these social categories
have become associated with these particular stereotypical traits (but see
Diehl & Jonas, 1991; Ford & Stangor; 1992; Stangor & Lange, 1994). Why do
we think of the elderly as grey and slow? How come that upon hearing the
word “skinhead”, we immediately think of “aggressive”? In order to gain more
insight into the nature of stereotypes, we must analyse their function a little
more closely.
A stereotypical trait is informative about the nature of a group if it is dis-
tinctive, that is, if the group differs on that trait from comparison groups. Put
Downloaded by [Tufts University] at 13:18 08 October 2014

differently, a trait would have little value as a stereotype of a group if all or


most other groups had this trait to the same degree. So, perceived intercate-
gory difference must be characteristic of stereotypical traits (Stangor & Lange,
1994). One could emphasize the importance of intergroup distinctiveness in
specific comparisons (as one would be inclined to do in the case of natural
comparisons, e.g., men vs. women), but in most cases it seems quite adequate
to look at distinctiveness as a generalized index (McCauley & Stitt, 1978), that
is. traits may become stereotypical if the group involved scores higher than
most comparison groups-practically speaking, if the group scores clearly
above average. Such a generalized conception of distinctiveness has the atten-
dant advantage that such stereotypes may be useful across situations, that is,
they have a wider range of cross-situational validity (and/or applicability). We
shall indicate this distinctiveness feature of stereotypical traits as “high central
tendency”.
Stereotypes are for stereotyping, that is, they are supposed to be applicable
to group members. Upon meeting a specific exemplar from a social category,
the stereotype must be informative about what sort of person this is. There-
fore, a stereotypical trait of a group should not only have high central ten-
dency, but it should also fairly invariably apply to members of the group
involved. In brief, stereotypical traits should be characterized by high central
tendency and low variability (Verplanken, Jetten & van Knippenberg, 1996).
In other words, stereotypical traits are not only characterized by intercategory
difference (high central tendency) but also by intracategory similarity (low
variability).
One might also approach the question of the association of social categories
and traits from the perspective of associative learning. Classic (Pavlov, 1927)
as well as contemporary (Fisk & Schneider, 1984; Shiffrin & Dumais, 1981;
Shiffrin & Schneider, 1977; Schneider & Fisk, 1982) approaches to associative
learning suggest that the frequent co-occurrence of two stimuli (e.g., elderly
and slow) will establish an association between the two; the more frequent the
co-occurrence, the stronger the association. The more often you meet elderly
people who are slow, the more you learn that on average elderly people are
CATEGORIZATION A N D STEREOTYPING 129

slow (high central tendency), and the stronger the association between elderly
and the concept of slowness. In associative learning, apart from frequency of
association, consistency is important too. The observation of too many incon-
sistent exemplars of the category may effectively weaken the association-That
is, if from time to time we come across “unbelievably” quick elderly people,
our faith in the stereotype that elderly people are slow will be seriously under-
mined. So, also from an associative learning perspective, “stereotypical” is
what distinguishes the target group from comparison groups (high central
tendency) and this distinguishing feature occurs without too many exceptions
(low variability) (see e.g., Lambert, Rosenkoetter, Lickel & Wells, in press;
Ryan, Judd & Park, 1996).
Downloaded by [Tufts University] at 13:18 08 October 2014

We tested these ideas in two experiments (Dijksterhuis & van Knippenberg,


1999). In the first experiment, we gave participants information about the
members of “group K”. The information consisted of the test scores of several
members of group K on four dimensions. On two dimensions (e.g., on intelli-
gence and stability) they scored on average high, on two other dimensions they
scored low (e.g., friendliness and extraversion), while on half of these the
variance of the scores of group K members was high (e.g., on intelligence
and friendliness), and for the other two dimensions (e.g., stability and ex-
traversion) variance was low. This way central tendency and variability were
manipulated independently within subjects (of course, content of traits was
counterbalanced). Immediately after this information was presented, partici-
pants were administered a lexical decision task with various synonyms of the
manipulated traits as critical items (e.g., clever, smart, witty; none of these
was ever explicitly mentioned in providing the information about group K
members on the test scores). These lexical decisions were preceded by 30ms
prime containing the word “group K” which was masked (with XXXX) and
followed by the stimulus (either word or non-word) with a stimulus onset asyn-
chrony (SOA) of 250ms. Lexical decision latencies on the critical words were
used as dependent variables. The results are summarized in Table 3A below.
We found, as predicted, that upon priming with “group K” faster lexical de-
cisions were obtained with words pertaining to traits that combined high
central tendency and low variability compared to the remaining traits (with
low tendency or high variability or both). Thus, it appeared that traits associ-
ated with an artificial group (“group K”) are more accessible upon priming
with the category label when they differentiate the group from other groups
(i.e., high central tendency) as well as fairly invariably apply to the members
of the group (i.e., low variability).
In our second experiment (Dijksterhuis and van Knippenberg, 1999) the
same conceptual issue was addressed on the basis of existing stereotypes.
Across the many stereotypes that people have acquired, the basic features of
stereotypical traits should be high central tendency and low variability. In an
extensive pilot study, category-trait combinations were selected that varied in
130 AD VAN KNIPPENBERG AND AP DIJKSTERHUIS

Table 4.3 Lexical decision latencies (in ms) of stereotype trait words as a function of
central tendency and variability, with manipulated information about an artificial group
(3A), and with existing stereotypes and real groups (3B) (adapted from Dijksterhuis
& van Knippenberg, 1999)

3A: Artificial 3B: Existing


group stereotypes
Central tendency Central tendency
High Low High Low
Variability
High 496 48 1 495 492
Low 469 494 452 492
Downloaded by [Tufts University] at 13:18 08 October 2014

central tendency and variability. Participants were presented with trait words
and non-words in a lexical decision task preceded by various category primes,
according to 2 (high vs. low central tendency) x 2 (high vs. low variability)
within-subjects design. The results are shown in Table 3B. As predicted, it was
found that participants reacted significantly faster when a trait had high central
tendency and low variability for the primed category, compared to traits that
had either low central tendency or high variability or both. Thus, we also
showed with existing stereotypes that priming a category label enhanced the
accessibility of those traits that were characteristic of the group (high central
tendency) and that applied relatively invariably to the group members (low
variability).
These experiments thus show two things. First, with regard to the determi-
nants of stereotype content, it appears that as far as traits are concerned they
distinguish the group from other groups (i.e., stereotypical traits have high
central tendency) and they are quite informative as to what individual group
members are like (i.e., stereotypical traits have low within-group variability).
Secondly, stereotypes are quickly formed, and when measured immediately
after being induced, stereotypical traits already show enhanced accessibility
upon category priming. Apparently, people are capable of quickly learning
group traits that have high central tendency and low variability.
It should be noted that this experimental evidence for stereotype learning
(see also Sherman, 1996) only represents one of the ways in which perceivers
acquire stereotypical knowledge and stereotypical expectancies. It seems
likely that there are many other ways in which individuals in our culture come
to know what the relevant social categorizations and their associated attrib-
utes are. It may be argued that direct learning through personal experience
may actually constitute only a small part of how people normally acquire their
stereotypical knowledge. To us, the two studies were of particular interest
because they show some of the functional characteristics of strong (or easily
CATEGORIZATION AND STEREOTYPING 131

acquired) stereotypical trait associations (high central tendency, low variabil-


ity) rather than how they are learned.

Stereotype Activation and Application: Some Conclusions


We have raised the question of whether category activation always leads to
stereotype activation and application. There is experimental evidence that the
occurrence of both these processes is subject to limiting conditions.
First, it appears that upon being confronted with a member of a stereotyped
group, the associated stereotype may not become activated under processing
conditions that, for all practical purposes, do not require stereotypical knowl-
Downloaded by [Tufts University] at 13:18 08 October 2014

edge. For instance, it seems that the processing of information in the task one
is performing must be on a semantic level. In a situation in which the partici-
pant’s task is merely to “spot the dot” (Macrae et af.,1997), the activation of
(semantic) stereotypical knowledge does not seem to take place. In other
words, some tasks (e.g., navigating through the supermarket) do not require
stereotypical knowledge, and may even be hampered if such knowledge
became activated.
Secondly, there is some evidence that stereotype activation may become
blocked under very demanding task conditions, for example, when the per-
ceiver’s mind is kept busy doing other things (such as rehearsing eight-digit
numbers). Whether the lack of resources really excludes stereotype activation
is, however, doubtful, because under the same demanding task conditions,
stereotypes do become activated when the participant needs them for reasons
of self-esteem protection (Spencer et af.,in press). However, further research
is needed to clarify by what mechanism stereotype activation is susceptible to
such intricate variations in processing conditions. It seems that implicit pro-
cessing priorities may play a role in selective stereotype activation.
It is important to emphasize that the above limiting conditions of stereo-
type activation were demonstrated to exist in comparison with “normal”
control conditions in which stereotype activation did occur. So, normally we
do activate stereotypes upon category activation, but in some specifiable
conditions we don’t.
With regard to the application of stereotypes upon stereotype activation,
there are also some limitations. First, the sort of information contained in
the stereotype must be compatible with what you want to know or do. For
instance, the anticipation of having to work with someone on an important
task may incite people to look for more individuating information, that is, to
look for more than the stereotype can offer. Secondly, we may be motivated
not to stereotype, or attempt to correct our perceptions and judgements for
possible stereotype biases. Thirdly, in conjunction with these two limiting
conditions (when we need more or other information than the stereotype
can offer, or we want to abstain from stereotyping), we must have sufficient
132 A D VAN KNIPPENBERG A N D AP DIJKSTERHUIS

resources not to stereotype. In other words, high cognitive load, or some sort
of distraction, or complex tasks, may necessitate stereotype application, even
in situations in which we would rather not stereotype.
In line with specifications of what stereotypic beliefs essentially are for (the
provision of meaningful knowledge, i.e., knowledge about intercategory dif-
ferences and within-category similarities) we presented some evidence that
it was indeed these features of stereotypical traits (high central tendency
and low variability) that predicted, in combination, the accessibility of trait
words upon social category priming (Dijksterhuis & van Knippenberg, 1999).
“Stereotypical” knowledge is knowledge that is informative about the group,
and applicable to its members.
Downloaded by [Tufts University] at 13:18 08 October 2014

PART 111: THE SELF-PRESERVATION


OF STEREOTYPES
Our functional approach to categorization and stereotyping suggests that
we use specific categories because we expect them to co-vary with important
differences (that is, differences on whatever we find socially important).
However, what the perceiver encounters in terms of real data may simply not
meet these expectations. That is, the expected between-category differences
and within-category similarities may not be present in the stimulus pattern.
Would that not very soon frustrate the perceiver’s choice of categorization and
the use of the activated stereotypes? In other words, how do we deal with
“faulty” stereotypes?
The fact that stereotypically expected differences may not actually occur
does not necessarily disqualify the use of the social category in question. First,
the categories involved often have a history of effective use, in the sense that
they may have served the perceiver’s needs many times in the past. Therefore,
it is reasonable to use them again on “appropriate” occasions, even when from
time to time the observed data do not nicely fit the expected pattern. Second,
and more importantly, there are various sorts of biases that help to make the
categories look good even when they actually were inadequate. For instance,
van Knippenberg et af. (1994) found a tendency for name confusions to be
stereotype-confirming, that is statements that were inconsistent with the cat-
egory of the real source were often attributed to someone belonging to the
other category (to which they were consistent). In addition, there are biases
of a more active nature. Using an information search paradigm, Johnston
and Macrae (1994) found that participants preferentially selected questions
whose features matched the stereotype of the target group, that is, their infor-
mation seeking was biased in a stereotype-confirming direction. In general,
there may be stereotype-consistency biases in information selection,
perception and memory (Duncan, 1976; Dijksterhuis & van Knippenberg.
CATEGORIZATION A N D STEREOTYPING 133

1995,1996; Johnston, 1996; Johnston & Macrae. 1994; Sagar & Schofield, 1980:
van Knippenberg & Dijksterhuis, 1996; van Knippenberg, Dijksterhuis, & Ver-
meulen, 1999) that help to create the illusion of fit, even in situations in which
it is absent. Below, we will expand on several mechanisms that seem to operate
in the service of stereotype maintenance.

Stereotypes as Filtering Devices


Stereotypes seem to have a built-in defence system that manages to refute
much of the evidence raised against them. Stereotypes have a way of defend-
ing themselves (for a similar reasoning, see Dijksterhuis, 1996; Dijksterhuis
Downloaded by [Tufts University] at 13:18 08 October 2014

& van Knippenberg. 1996). Their defence begins with the way in which they
“frame the issue”: they offer an interpretational framework that suits them
well simply by making some concepts more accessible while preventing other
concepts from entering the stage. Stereotypes colour the interpretation of the
perceptual data.
We studied such accessibility effects of social category activation in three
experiments (Dijksterhuis and van Knippenberg. 1996). In the first experi-
ment, half of our participants were asked to think about soccer hooligans for
5 minutes and to write down anything that came to their mind regarding soccer
hooligans (cf. Macrae, Stangor, & Milne, 1994), while the remaining partici-
pants were not given such a task. In an ostensibly unrelated lexical decision
task, participants were presented with words and non-words that appeared one
by one on the computer screen. They had to decide for each stimulus as fast
as possible whether it was an existing word or not. Among the words appear-
ing on the screen were traits that were stereotypic for soccer hooligans (e.g.,
violent) and traits that were counter-stereotypic for soccer hooligans (e.g.,
friendly). It was found that, in comparison with a no-prime control condition.
social category activation made participants respond faster to stereotypic
traits and slower to counterstereotypic traits. A similar design was used in our
second experiment, except that instead of a lexical decision task we used word
puzzles.These puzzles consisted of three rows of letters presented on the com-
puter screen. The string of letters in each row was meaningless, but reading
from left to right the participant could detect an existing word when selecting
a particular pattern of switching up and down between the rows. As in the first
experiment, the participants were either primed (in this case they had to think
about the stereotype of college professors) or not primed prior to solving the
word puzzles. Within the consecutive series of puzzles on the screen, some
hidden words were stereotypic of professors, some were counterstereotypic,
and some were unrelated (irrelevant) to the primed category. The percentage
of correctly solved word puzzles within the time allowed (40s per puzzle) as
a function of priming and type of word are presented in Table 4.4.These data
show that priming with the category facilitated the detection of stereotypic
134 A D VAN KNIPPENBERG A N D AP DIJKSTERHUIS

Table 4.4 Percentages of correctly identified traits as


a function of priming (adapted from Dijksterhuis &
van Knippenberg, 1996)

Consistent Inconsistent Neutral


N o prime 71.3h 78.2b 73.0”
Prime 85.5’ 66.7a 76.9h
Means with different superscripts differ at p < 0.06.
Downloaded by [Tufts University] at 13:18 08 October 2014

trait words, and also inhibited the detection of counterstereotypic trait words
(both compared to the same words in the no prime control condition and com-
pared to the primed irrelevant words). The results of this second study were
replicated in a third study with the hooligan stereotype and slightly adapted
word puzzles. In sum, across three experiments we demonstrated that stereo-
typic traits become more accessible after priming with the social category
(i.e., facilitation effects; see also Blair & Banaji, 1996; Devine, 1989; Dovidio,
Evans, & Tyler, 1986; Macrae, Stangor, & Milne, 1994), while traits that are
counterstereotypic become less accessible (that is, inhibition of stereotype-
inconsistent traits upon social category priming).
These two simple cognitive effects of category activation, i.e., facilitation and
inhibition, have important consequences and constitute, in our view, a major
reason why stereotypes are so hard to change. The facilitated accessibility
of stereotypic traits and the inhibited accessibility of counterstereotypic traits
lead automatically to biased perception, biased inferences, biased judgements
and biased retrieval. Importantly, all these processes are biased in a stereo-
type-confirming manner. Such enhanced or reduced accessibility of trait con-
cepts can be conceived of as a “filter”, making the perception of stereotypic
information easier than counterstereotypic information, stereotypic inferences
and judgements easier than counterstereotypic inferences and judgements,
and the retrieval of stereotypic information easier than counterstereotypic
information. These consequences may quite effectively protect stereotypes
against change. We will briefly discuss some of these filtering effects.

Stereotype Maintenance through Biased Perception


The accessibility of a construct determines the ease with which construct-
relevant information is detected (Bruner, 1957; Higgins, 1989). For example,
the more accessible the trait “intelligent”, the easier it is to detect intelligent
behaviour. We need less intelligence-related “stimulus input” to perceive intel-
ligence when the concept of intelligence is highly accessible. Thus, stereotypes
influence perception because social category activation makes stereotypic con-
CATEGORIZATION A N D STEREOTYPING 135

structs more accessible and counterstereotypic constructs less accessible com-


pared to when the social category is not activated.
Macrae, Stangor, and Milne (1994) conducted a series of studies in which
they obtained evidence for the idea that category activation makes it easier to
detect stereotypic information. In one of their experiments, participants were
requested to think about child abusers for 5 minutes. In a second task, partic-
ipants were presented with words on a computer screen. Some of these words
were stereotypic traits (e.g., perverse) while other words were non-stereotypic
(e.g., arrogant). These words however, were degraded. They were masked by
a dot pattern. On successive trials, this dot pattern gradually decreased,
thereby slowly revealing the word. The results showed that participants
Downloaded by [Tufts University] at 13:18 08 October 2014

needed fewer trials to detect stereotypic words than to detect counterstereo-


typic words. That is, stereotypic words could be detected while they were
hidden behind a more dense dot pattern than non-stereotypic words. This
experiment demonstrates that upon category activation people need less
stimulus input to detect stereotypic information. Their results were replicated
in two other experiments in which different stereotypes were used and in
which ease of detection was measured in different ways.

Stereotype Maintenance through Biased Inferences and Judgements


If incoming information passes the first hurdle, that is, if it is actually detected
and processed, the next hurdle immediately comes in sight. Behaviourial infor-
mation is often ambiguous. Somebody crossing the Atlantic in an inflatable
rowing boat can be perceived as adventurous, but also as reckless. Higgins,
Rholes, and Jones (1977) showed that the accessibility of relevant traits
determines to a large extent which inference is made. If “adventurous” is
highly accessible, it is very likely that the behavior in question is perceived as
adventurous, while when “reckless” is more accessible, it becomes more likely
to infer that crossing the Atlantic in an inflatable rowing boat is indeed
reckless.
Now if accessibility determines inferences, category activation should affect
inferences as well, since category activation enhances (in the case of stereo-
typic traits) or decreases (in the case of counterstereotypic traits) accessibility.
This idea was tested and confirmed several times (Devine, 1989;Duncan, 1976;
Kunda & Sherman-Williams, 1983; Sagar & Schofield, 1980). Devine (1989),
for example, primed participants with stereotypic words of Blacks, without the
participants’ awareness. Some participants, however, received a more power-
ful priming procedure than others. Later, participants were asked to read a
passage about a boy named Donald and to make evaluative judgements about
Donald. The judgements were clearly biased by the activated stereotype. Par-
ticipants who received the powerful procedure rated Donald as more hostile
(in line with the stereotype of Blacks) than participants who received the less
136 AD VAN KNIPPENBERG A N D AP DIJKSTERHUIS

powerful priming procedure. In sum, judgements are biased in a stereotype-


confirming direction by the mere activation of these stereotypes. Just as in
the perceptual stage, stereotypes bolster themselves when we make
judgements.

Stereotype Maintenance through Biased Memory


Another way stereotype-confirming mechanisms may be manifested is in
retrieval processes in episodic memory after stereotype activation. The ques-
tion we asked ourselves some time ago is “How does a category label attached
to a person or a group affect our memories of earlier stored events concern-
Downloaded by [Tufts University] at 13:18 08 October 2014

ing this person or group?”


This question was first addressed in a study of Snyder and Uranowitz (1978).
Participants were given information about a woman called Betty K, about her
youth, high school years, her relationship with her parents, and so on. Later,
some participants were told that Betty K was currently living as a lesbian,some
others were told that she now had a heterosexual relationship, and still others
received no further information. One week later, participants were tested for
memory by means of a series of multiple choice questions. It appeared
from these data that participants’ memory displayed a stereotype-consistency
bias: participants tended to remember information that was in line with the
later activated (lesbian or heterosexual) stereotype, rather than information
that went against it. Snyder and Uranowitz interpreted these findings as
reflecting memory reconstructions: “. . . we do not remember an event by
activating to ‘replaying’ some fixed memory trace. Rather, we construct a
schematic representation of our past experience by piecing together bits and
pieces with new facts that we (knowingly or unknowingly) supply to flesh out
or augment our emerging knowledge of the past” (Snyder & Uranowitz. 1978,
p. 942).
In several experiments (see van Knippenberg & Dijksterhuis, 1996), we
expanded on these ideas. A typical experiment went as follows. Participants
were presented with behavioral information. Some information implied intel-
ligence, other behavior implied aggressiveness. After participants read this
information, we activated a stereotype among half the participants. This
stereotype (e.g., professors) made, in retrospect, some behavioral information
inconsistent while other behavioral information was made consistent. In a sub-
sequent recall task, it turned out that stereotype activation inhibited memory
for inconsistent information. In concrete terms, participants who were primed
with the stereotype of professors after they had read the behavioral informa-
tion were hardly able to recall any aggressive behaviors.
In sum, stereotype activation affects recall by inhibiting memory for incon-
sistent information. By means of this mechanism, stereotypes protect them-
CATEGORIZATION A N D STEREOTYPING 137

selves: mere stereotype activation makes counter-evidence inaccessible and


thereby ineffective. As a result, the owner of a stereotype does not feel the
need to change it.

Some Conclusions Concerning Stereotype Maintenance


Above we have presented some evidence suggesting that in various stages of
information processing, the activation and application of a stereotype sets off
the operation of various mechanisms that contribute to stereotype-confirming
perceptions, interpretations, and memories of persons and situations.
In line with our functional argument, all this serves a purpose, that is, to
Downloaded by [Tufts University] at 13:18 08 October 2014

keep intact knowledge structures that are extremely useful in understanding


social reality. Our cognitive equipment does not seem to allow that
stored long-term knowledge gets easily refuted. We assume that this state of
affairs is useful for the perceiver. It is not very likely that these stereotype-
confirming mechanisms would operate simply to lead us astray, to make us
arrive at wrong perceptions and faulty conclusions. The crucial reason for
the tenacity of well-established stereotypes is that they do a good job in
guiding our perceptions and actions in everyday life. Conversely, from the
same functional perspective, we must assume that if reality consistently
went against existing stereotypes, they would dissapear. That is, if stereotypi-
cally expected intercategory differences and intracategory similarities in social
reality were to change, sooner or later these stereotypes would fade away.
But, as it is, the regularities in social life do not change overnight. So, it
may be a useful feature of stereotypes that they have built-in resistance to
change.

CONCLUSIONS
In this chapter we have advocated a functional approach to social categoriza-
tion and stereotyping. The fundamental principle underlying our main argu-
ment is that people use social categorizations that are meaningful to them in
a given social situation. Much of the rest of our argument is derived from that
basic tenet. Meaningfulness means, here, that we use social categorizations that
optimally capture intercategory difference and within-category similarity
[note the correspondence with Turner’s (1987) concept of meta-contrast ratio]
or, in other words, meaningful categorizations co-vary with important stimu-
lus features.
Meaning thus defined may reside in the eyes of the beholder, that is, it may
be something that the perceiver projects onto perception of reality. Meaning
can also be discovered as co-variation in the data. This may potentially
138 AD VAN KNlPPENBERG AND AP DIJKSTERHUIS

constitute the genesis of a stereotype. If such an observation of co-variation is


repeated again and again, a long-lasting stereotype may develop.
In many social situations, the subjectively projected meaning of a social cat-
egorization (which is assumed to take place when there is normative fit) and
real co-variance tend to go together. Sometimes, however, the data may not
nicely fit the stereotypical expectations. On these occasions stereotypes may
be able to recruit a host of mechanisms that make their expectations seem to
come true despite the real data. In our discussion above, we have explicated
several of these mechanisms.
Theoretically, on a more abstract level, an interesting implication of these
mechanisms in category selection and subsequent biased data processing is
Downloaded by [Tufts University] at 13:18 08 October 2014

that one can conceive of spontaneous social categorization as a process in


which a specific (stereotypical) interpretation of reality has the advantage of
first strike. What you already believe to be true gets the first chance of being
proven right. And, because of built-in stereotype-confirming devices, they are
unlikely to be disproven. As we argued, this may be a useful feature of stereo-
types, as long as one is willing to assume that social reality is not likely to
change over night.
Throughout this article, we have advocated a functional approach to social
categorization and stereotyping. We are aware of the fact that our present line
of argument fits in with a long-standing tradition in social categorization and
stereotyping research (see e.g., Allport, 1954;Tajfel, 1969,1972,1978), and we
know that many contemporary theorists in the domain of stereotyping share
this functional point of view. We see social categorization and stereotyping as
processes that are useful and indispensable in normal social functioning. It
seems to us that this is the prevailing (although often implicit) point of view
of many researchers who work within the social cognition approach to stereo-
typing. Equally (and maybe more explicitly) the social identity and self-
categorization approach to intergroup relations and stereotyping endorse, as
a basic component of their theorizing, the idea that social categorization and
stereotyping processes operate first, and last, and always, in the service of a
veridical understanding of social reality and, therefore, of adequate social
behavior. We believe that both perspectives make their own unique contribu-
tion to our understanding of the basic functions of social categorization and
stereotyping.

ACKNOWLEDGEMENTS
The authors are grateful to Miles Hewstone, Wolfgang Stroebe, and three
anonymous reviewers for their helpful comments on an earlier version of this
chapter.
CATEGORIZATION A N D STEREOTYPING 139

REFERENCES
Allport, G. (1954). The Nature of Prejudice. Cambridge, MA: Addison-Wesley.
Anderson, M. C., & Spellman, B. A. (1995). On the status of inhibitory mechanisms
in cognition: Memory retrieval as a model case. Psychological Review, 102, 68-
100.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, effi-
ciency and control in social cognition. In R. S. Wyer & T. K. Srull (Eds), Handbook
of Social Cognition (2nd edn, pp. 1-40>.Hillsdale, NJ: Erlbaum.
Berger, P. L. (1966). Identity as a problem in the sociology of knowledge. European
Journal of Sociology, 7,105-15.
Blair, I. V., & Banaji, M. R. (1996). Automatic and controlled processes in gender
stereotyping. Journal of Personality and Social Psychology, 70,114-63.
Downloaded by [Tufts University] at 13:18 08 October 2014

Blanz, M. (1999). Accessibility and fit as determinants of the salience of social cate-
gorizations. European Journal of Social Psychology, 29,43-74.
Bodenhausen, G. V. (1988). Stereotypic biases in social decision making and memory:
Testing process models of stereotype use. Journal of Personality and Social Psychol-
ogy, 55,726-37.
Bodenhausen, G. V., & Lichtenstein, M. (1987). Social stereotypes and information
processing strategies:The impact of task complexity.Journal of Personality and Social
Psychology, 52.871-80.
Bodenhausen, G. V., & Macrae, C. N. (1998). Stereotype activation and inhibition. In
R. S. Wyer Jr (Ed.), Advances in Social Cogrrition,Vol.11. (in press).
Bodenhausen, G. V., & Wyer, R. S. (1985). Effects of stereotypes on decision making
and information processing strategies. Journal of Personality and Social Psychology,
48,267-82.
Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64,123-52.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled com-
ponents. Journal of Personality and Social Psychology, 56,s-18.
Diehl, M., & Jonas, K. (1991). Measures of national stereotypes as predictors of the
latencies of inductive versus deductive stereotypic judgments. European Journal of
Social Psychology, 21,317-30.
Dijksterhuis, A. (1996). Stereotypes and memory: The effects of a posteriori stereo-
types on recall of social information. Unpublished Doctoral Dissertation, University
of Nijmegen.
Dijksterhuis, A., & van Knippenberg, A. (1995). Memory for stereotype-consistent
and stereotype-inconsistent information as a function of processing pace. European
Journal of Social Psychology, 25,689-94.
Dijksterhuis, A., & van Knippenberg, A. (1996). The knife that cuts both ways: Facil-
itated and inhibited access to traits as a result of stereotype activation. Journal of
Experimental Social Psychology, 32,27148.
Dijksterhuis, A., & van Knippenberg, A. (1998). Inhibition, Aberdeen, and other
cloudy subjects. In R. S. Wyer Jr (Ed.), Advances in Social Cognition, 11,83-96.
Dijksterhuis, A., & van Knippenberg, A. (1999). On the parameters of associative
strength: central tendency and variability as determinants of stereotype accessibility.
Personality and Social Psychology Bulletin, 25,527-36.
Dovidio, J. F., Evans, N., &Tyler, R. B. (1986). Racial stereotypes: the contents of their
cognitive representations. Journal of Experimental Social Psychology, 22,22-37.
Dovidio, J. F., Kawakami, K., Johnson, C., Johnson, B., & H0ward.A. (1997). On the
nature of prejudice: Automatic and controlled processes. Journal of Experimental
Social Psychology, 33,510-40.
140 AD VAN KNIPPENBERG AND AP DIJKSTERHUIS

Duncan, B. (1976). Differential social perception and attribution of intergroup vio-


lence: Testing the lower limits of stereotyping Blacks. Journal of Personality and
Social Psychology, 34,590-8.
Ellemers, N., & van Knippenberg, A. (1997). Stereotyping in social context. In R.
Spears, P. J. Oakes. N. Ellemers, & S. A. Haslam (Eds), The Social Psychology of
Stereotyping and Group Life (pp. 208-335). Oxford: Basil Blackwell.
Erber, R., & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent
information. Journal of Personality and Social Psychology, 47,709-26.
Fisk, A. D., & Schneider, W. (1984). Memory as a function of attention, level of pro-
cessing and automatization. Journal of Experimental Psychology: Learning, Memory
and Cognition, 10,181-97.
Fiske, S. T. (1998). Stereotyping, prejudice and discrimination. In D. T. Gilbert. S. T.
Fiske, & G. Lindzey (Eds), The Handbook of Social Psychology (4th edn, Vol. 2, pp.
Downloaded by [Tufts University] at 13:18 08 October 2014

357-41 1).
Fiske, A. F’., Haslam, N., & Fiske, S. T. (1991). Confusing one person with another:
What errors reveal about the elementary forms of social relations. Journal of Per-
sonality and Social Psychology, 60,656-74.
Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from
category-based to individuating processes: Influences of information and motivation
on attention and interpretation. In M. P. Zanna (Ed.), Advances in Experimental
Social Psychology (Vol. 23, pp. 1-74). New York: Random House.
Ford, T. E., & Stangor, C. (1992). The role of diagnosticity in stereotype formation:
Perceiving group means and variances. Journal of Personality and Social Psychology,
63,356-67.
Frable, D. E. S., & Bem, S. L. (1985). “If you are gender schematic, all members of
the opposite sex look alike”. Journal of Personality and Social Psychology, 49,459-68.
Gaertner, S. L., & McLaughlin, J. P. (1983). Racial stereotypes: Associations and
ascriptions of positive and negative characteristics. Social Psychology Quarferly,46,
23-30.
Gernsbacher, h.1. A., & Faust, M. E. (1991). The mechanism of suppression: A com-
ponent of general comprehension skill. Journal of Experimental of Psychology:
Learning, Memory, and Cognition, 17,245-62.
Gilbert, D. T., & Hixon, J. G. (1991). The trouble of thinking: Activation and applica-
tion of stereotypic beliefts. Journal of Personality and Social Psychology, 60,509-17.
Hewstone, M., Hantzi, A., & Johnston, L. (1991). Social categorization and person
memory: The pervasiveness of race as an organizing principle. European Journal of
Social Psychology, 21,517-28.
Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectivity and suffer-
ing from unconscious sources. In J. s. Uleman & J. A. Bargh (Eds), Unintended
Thought. New York: Guilford.
Higgins, E. T., & Brendl, M. (1995). Accessibility and applicability: Some “activation
rules” influencing judgment. Journal of Experimental Social Psychology, 31,21843.
Higgins, E.T., King, G., & Mavin, G. (1982). Individual construct accessibility and sub-
jective impressions and recall. Journal of Personality and Social Psychology, 43,
35-47.
Higgins, E. T., Rholes, W. S., & Jones, C. R. (1977). Category accessibility and impres-
sion formation. Journal of Personality and Social Psychology, 13,141-54.
Hoffman, C., & Hurst, N. (1990). Gender stereotypes: Perception or rationalization?
Journal of Personality and Social Psychology, 58,197-208.
Johnston, L. (1996). Resisting change: Information-seeking and stereotype change.
European Journal of Social Psychology, 26,799-825.
CATEGORIZATION A N D STEREOTYPING 141

Johnston, L., & Macrae, C. N. (1994). Changing social stereotypes: The case of the
information seeker. European Journal of Social Psychology, 24,581-92.
Jost, J. T., & Banaji, M. R. (1994). The role of stereotyping in system-justification and
the production of false consciousness. British Journal of Social Psychology, 33,l-27.
Kawakami, K., Dovodio, J., Moll, J., Hermsen, S., & Russin,A. (1999). Just say no (to
stereotyping); Effects of negation training of stereotypes on stereotype activation
(unpublished manuscript).
Kelly, G. A. (1955). The Psychology of Personal Constructs. New York: Norton.
Kunda, Z., & Sherman-Williams, B. (1993). Stereotypes and the construal of
individuating information. Personality and Social Psychology Bulletin, 19,90-9.
Lambert, A. J., Rosenkoetter, L., Lickel, B., & Wells, J. The influence of group vari-
abilty and processing goals on the ease of making judgments about social categories.
Personality and Social Psychology Bulletin (in press).
Downloaded by [Tufts University] at 13:18 08 October 2014

Lepore, L., & Brown, R. (1997). Category and stereotype activation: Is prejudice
inevitable? Journal of Personality and Social Psychology, 72,27547.
Mackie, D. M., Hamilton, D. L., Susskind, J., & Rosselli, E (1996). Social psychologi-
cal foundations of stereotype formation. In C. N. Macrae, H. Stangor, & M.
Hewstone (Eds), Stereotypes and Stereotyping (pp. 41-78). New York: Guilford.
Macrae, C. N., Bodenhausen, G. V., & Milne, A. B. (1995). The dissection of selection
in person perception: Inhibitory processes in social stereotyping. Journal of Person-
ality and Social Psychology, 69,397407.
Macrae, C. N., Bodenhausen, G. V., & Milne, A. B. (1998). Saying no to unwanted
thoughts: Self-focus and the regulation of mental life. Journal of Personality and
Social Psychology, 74,578-89.
Macrae, C. N., Bodenhausen, G. V., Milne, A. B., & Jetten, J. (1994). Out of the mind
but back in sight: stereotypes on the rebound. Journal of Personality and Social
Psychology, 67,808-17.
Macrae, C. N., Bodenhausen, G. V., Milne, A. B., Thorn,T. M. J., & Castelli, L. (1997).
On the activation of social stereotypes:The moderating role of processing objectives.
Journal of Experimental Social Psychology, 33,471-89.
Macrae, C. N., Hewstone, M., & Griffiths, R. J. (1993). Processing load and memory
for stereotype-based information. European Journal of Social Psychology, 23,7747.
Macrae, C. N., Milne, A. B., & Bodenhausen, G. V, (1994). Stereotypes as energy-
saving devices: A peek inside the cognitive toolbox. Journal of Personality and Social
Psychology, 66,3747.
Macrae, C. N., Stangor, C., & Milne, A. B. (1994). Activating social stereotypes: A func-
tional analysis. Journal of Experimental Social Psychology, 30,370-89.
McCauley, C., & Stitt, C. L. (1978). An individual and quantitative measure of stereo-
types. Journal of Personality and Social Psychology, 36,929-40.
Monteith, M. J. (1993). Self-regulation of prejudiced responses: Implications for
progress in prejudice-reduction efforts. Journal of Personality and Social Psychology,
65,469-85.
Monteith, M. J., Sherman, J. W., & Devine, F? G. (1998). Suppression as a stereotype
control strategy. Personality and Social Psychology Review, 2,6342.
Monteith, M. J., Spicer, C. V., & Tooman, G. D. (1998). Consequences of stereotype
suppression: stereotypes on and not on the rebound. Journal of Experimental Social
Psychology, 34,355-77.
Moskowitz, G. B., Gollwitzer, P. M., Wasel, W., & Schaal, B. (1999). Preconscious
control of stereotype activation through chronic egalitarian goals. Journal of
Personality and Social Psychology, 77,167-84.
Neuberg, S. L., & Fiske, S.T. (1987). Motivational influences on impression formation:
142 AD VAN KNIPPENBERG A N D AP DIJKSTERHUIS

Outcome dependency, accuracy-driven attention, and individuating processes.


Journal of Personality and Social Psychology, 53,431-44.
Oakes, P. J. (1987). The salience of social categories. In J. C. Turner. M. A. Hogg, P.J.
Oakes. S. D. Reicher, & M. S. Wetherell (Eds), Rediscovering the Social Group: A
Self-categorization Theory (pp. 11741). Oxford: Basil Blackwell.
Pavlov, I. P. (1927). Conditioned Reflexes. New York: Dover.
Pendry, L. F., & Macrae, C. N. (1994). Stereotypes and mental life: The case of the
motivated but thwarted tactician. Journal of Experimental Social Psychology, 30,
303-25.
Perdue, C. W., Dovidio, J. F., Gurtman, M. B., & Tyler, R. B. (1990). “Us” and “Them”:
Social categorization and the process of intergroup bias. Journal of personality and
Social Psychology, 59,475-86.
Ryan, C. S.. Judd, C. M., & Park, B. (1996). Effects of racial stereotypes on judgments
of individuals: The moderating role of perceived group variability. Journal of Experi-
Downloaded by [Tufts University] at 13:18 08 October 2014

mental Social Psychology, 32,71-103.


Ryan, C. S.. Park, B., & Judd, C. M. (1996). Assessing stereotype accuracy: Implica-
tions for understanding the stereotype process. In C. N. Macrae, C. Stangor, & M.
Hewstone (Eds), Stereotypes and Stereotyping (pp. 121-57). New York: Guilford.
Sacks. 0. (1995). An Anthropologist on Mars. New York: Random House.
Sagar, H. A.. & Schofield, J. W. (1980). Racial and behavioral cues in black and white
children’s perception of ambiguously aggressive acts. Journal of Personality and
Social Psychology, 39,59&8.
Schneider, W., & Fish, A. D. (1982). Degree of consistent training: Improvements
in search performance and automatic process development. Perception and
Psychophysics. 31,160-8.
Secord, P. F., Bevan, W., & Katz, B. (1956). The negroe stereotype and perceptual
accentuation. Journal of Abnornial and Social Psychology, 59.78-83.
Sherman, J. W. (1996). Development and mental representation of stereotypes.
Journal of Personality and Social Psychology, 70, 112641.
Shiffrin, R. M.,& Dumais, S. T. (1981). The development of automatism. In J. R.
Anderson (Ed.), Cognitive Skills and their Acquisition (pp. 111-40). Hillsdale, NJ:
Erlbaum.
Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human informa-
tion processing: 11. Perceptual learning, automatic attending, and a general theory.
Psychological Review, 84,127-90.
Smith, E. R.,& Henry,S. (1996). An in-group becomes part of the self Response time
evidence. Personality and Social Psychology Bulletin, 22,635-42.
Snyder, M., & Uranowitz, S. W. (1978). Reconstructing the past: some cognitive con-
sequences of person perception. Journal of Personality and Social Psychology, 36,
941-50.
Spencer. S. J., Fein, S., Wolfe, C. T., Hodgson, H. L., & Dunn. M. A. Automatic acti-
vation of stereotypes: The role of self-image threat. Personality and Social Psychol-
ogy Bulletin (in press).
Stangor, C., & Duan, C. (1991). Effects of multiple task demands upon memory for
information about social groups. Journal of Experimental Social Psychology, 27,
357-78.
Stangor, C., & Lange, J. (1994). Mental representations of social groups: Advances in
understanding stereotypes and stereotyping. In M. F! Zanna (Ed.), Advances
in Experimental Social Psychology (Vol. 26, pp. 357-416). New York: Academic
Press.
Stangor. C., & Schaller, M. (1996). Stereotypes as individual and collective represen-
CATEGORIZATION A N D STEREOTYPING 143

tations. In C. N. Macrae, H. Stangor, & M. Hewstone, M. (Eds), Stereotypes and


Stereotyping (pp. 3 4 0 ) . New York: Guilford.
Stangor, C., Lynch, L., Duan, C., & Glass, B. (1992). Categorization of individuals on
the basis of multiple social features. Journal of Personality and Social Psychology,
62,207-18.
Tajfel, H. (1969). Cognitive aspects of prejudice. Journal of Social Issues, 25,79-97.
Tajfel, H. (1972). La categorisation sociale. In S. Moscovici (Ed.), Introduction a la
Psychologie Sociale (Vol. 1, pp. 272-302). Paris: Larousse.
Tajfel, H. (Ed.) (1978). Differentiation between Social Groups: Studies in the Social
Psychology of Intergroup Relations. London: Academic Press.
Tajfel, H. (1981). Human Groups of Social Categories. Cambridge: Cambridge
University Press.
Tajfel, H., & 'hrner, J. C. (1979). An integrative theory of intergroup conflict. In
W. G. Austin & S. Worchel (Eds), The Social Psychology of Intergroup Relations
Downloaded by [Tufts University] at 13:18 08 October 2014

(pp. 3348). Monterey, CA: BrookslCole.


Tajfel, H.. & 'hmer. J. C. (1986). The social identity theory of intergroup behavior.
In S. Worchel & W. G. Austin (Eds), The Social Psychology of Intergroup Relations
(pp. 7-24). Chicago: Nelson-Hall.
Tajfel, H., 8i Wilkes, A. L. (1963). Classification and quantitative judgement. British
Journal of Psychology, 54,101-14.
Taylor, S. E., & Falcone, H. (1982). Cognitive bases of stereotyping: The relationship
between categorization and prejudice. Personality and Social Psychology Bulletin, 8,
426-32.
Taylor, S. E., Fiske, S.T., Etcoff, N. L., & Ruderman, A. J. (1978). Categorical and con-
textual bases of person memory and stereotyping. Journal of Personality and Social
Psychology, 36,778-93.
Turner, J. C. (1987). A self-categorization theory. In J. C. Turner, M. A. Hogg, P.J.
Oakes, S. D. Reicher, & M. S. Wetherell (Eds), Rediscovering the Social Group: A
Self-categorization Theory (pp. 4247). Oxford: Basil Blackwell.
van Knippenberg, A. (1984). Intergroup differences in intergroup perceptions. In H.
Tajfel (Ed.), The Social Dimension: European Developments in Social Psychology
(pp. 560-78). Cambridge: Cambridge University Press.
van Knippenberg, A., & Dijksterhuis, A. (1996). A posteriori stereotype activation:
the preservation of stereotypes through memory distortion. Social Cognition, 14,
21-54.
van Knippenberg, A., & Ellemers, N. (1990). Social identity and intergroup differen-
tiation processes. European Review of Social Psychology, 1,137-70.
van Knippenberg, A., & Ellemers, N. (1993). Strategies in intergroup relations. In
M. A. Hogg & D. Abrams (Eds), Group Motivation: Social Psychological Perspec-
tives (pp. 17-32). London: Harvester Wheatsheaf.
van Knippenberg, D., & van Knippenberg, A. (1994). Social categorization, focus of
attention and judgements of group opinions. British Journal of Social Psychology,
33,477-89.
van Knippenberg, A., van Twuyver, M., & Pepels, J. (1994). Factors affecting social
categorization processes in memory. British Journal of Social Psychology, 33,419-31.
van Knippenberg, A.. Dijksterhuis, A., & Vermeulen, D. (1999). Judgement and
memory of a criminal act: The effects of stereotypes and cognitive load. European
Journal of Social Psychology, 29,191-201.
van Wuyver, M., & van Knippenberg, A. (1995). Social categorization as a function
of priming. European Journal of Social Psychology, 25,695-702.
van Twuyver, M., & van Knippenberg, A. (1998). Effect of group membership and
144 AD VAN KNIPPENBERG AND AP DIJKSTERHUIS

identification on categorization and subtyping in memory. European Journal of


Social Psychology, 28,531-54.
van Twuyver, M., & van Knippenberg, A. (1999). Social categorization as a function
of relative group size. British Journal of Social Psychology, 38,135-56.
Verplanken. B., Jetten. J., & van Knippenberg, A. (1996). Effects of stereotypicality
and perceived group variability on the use of attitudinal information in impression
formation. Personality and Social Psychology Bulletin, 22,960-71.
Wegner. D. (1994). Ironic processes of mental control. Psychological Review, 101,
34-52.
Wittenbrink, B., Judd, C. M.. & Park, B. (1997). Evidence for racial prejudice at the
implicit level and its relationship with questionnaire measures. Journal of Personal-
ity and Social Psychology, 72,262-74.
Wyer, R. S., & Srull, T. K. (1986). Human cognition in its social context. Psychologi-
Downloaded by [Tufts University] at 13:18 08 October 2014

cal Review. 93,322-59.

You might also like