Impressions and Levels of Prejudice By Omar
Azam
Social categorization is an activity that people engage in
everyday,
and it determines, to a large part, who these people will interact with,
who they will trust, and ultimately, who they will judge to be similar to
them. Like many other psychological processes, it seems that social
categorization
happens almost "automatically" (Devine, 1989). By this, what is
meant is that categorization is not a task that necessarily requires
active
attention. Perhaps because as humans we are so familiar with the stimuli
(other humans) and the goal (making a judgment about them), we find it a
relatively undaunting task to make judgments about other people. Such
judgments
include, for example, whether or not a target conforms to a trait or how
likely it is that the perceiver would aid the target. Research has shown
that, indeed, people can go about assessing others without them
explicitly
paying attention to the task.
Social categorization has been shown to rely to some extent on the use of
group stereotypes. The term "stereotype," as it was used by its
originator, Walter Lippman (1922), referred to supposed "little
pictures
we carry around in our heads." The term has evolved to have various
meanings, among these a popular and a social psychological one. To the
public,
stereotyping implies the use of dysfunctional categorization strategies
that are greatly influenced by motivational factors. Research in social
psychology has shown, however, that stereotyping is more ubiquitous than
was previously thought. Humans inevitably group people on the basis of
certain
traits, and although stereotypes are somewhat one-dimensional and
sometimes
lead to instantaneous negative appraisals, they are part and parcel of
the
human memory system. It is useful for human beings, when they are in
contact
with such a large variety of people and groups, to invent words that
describe
members of these groups. As such, stereotypes are part of the shared
societal
consciousness. The majority of people in the United States, for example,
could tell you what the stereotype of a homosexual male would be, for
example,
although there is no concrete research supporting the notion that those
who possess this knowledge are necessarily discriminatory or prejudiced.
Actual members of groups are more likely than not to be familiar with
stereotypes
of groups of which they are part. It seems self-evident that
stereotypical
knowledge of a well-known group, then, is accessible to most members of
a population.
Stereotypes are important from a behavioral standpoint only to the extent
in which they play a part in discriminatory behavior. To this end,
psychologists
have aimed to examine what role these mental constructs play on actual
behavior
towards target groups. Devine (1989) found that stereotypes do exist in
the minds of even low prejudiced people. Furthermore, they are highly
accessible
knowledge structures that can be automatically activated, even if they
are
not actually endorsed. Thus, since knowledge of group stereotypes are
present
in our consciousness, responding in a low prejudiced manner requires
inhibition
of these highly accessible schemata.
Psychologists, although they would have to agree on the notion that
stereotypes
do exist in our language, do not presently agree on some fundamental
aspects
of the functional nature of stereotypes and their close relatives,
prototypes.
"Prototypes" are central members of a category, and as with
nonsocial
categorization, research has supported the notion that stereotypes are
maximally
useful if they represent prototypical members of a category. When one is
asked to think of the stereotype of a "homosexual male," one
would
probably not call to mind a picture of a male who is Republican or who is
in the army. The picture that one would call to mind would be a cluster
of descriptive traits and other facets of the personality of a typical
person
in that group. The evidence that people do recall prototypical members of
a category better that marginal members is robust (Rips?). However, how
do people create these prototypical impressions of group members? Does it
happen on-line, when the person is asked to recall the group's
stereotype?
Or is it preformed and then regurgitated from crystallized impressions at
the time of the judgment task? More fundamental is the question of how
large
a role encountered exemplars from the category play in subsequent
categorization,
and if they do, then at what temporal point in the judgment making
context
do they exert any influence. Smith & Zarate (1992) and Linville et
al.
(1989) have found that exemplars are sufficient to produce a prototype
and
this mechanism can account furthermore for phenomena such as the Outgroup
Homogeneity Effect (OHE). Whereas such research has found that exemplar
processing models are sufficient, they do not prove that they are
necessary
or efficient. Pure exemplar models of stereotyping have been challenged
as inefficient (Hamilton & J.W. Sherman, 1994). Also, on-line
processing
would be difficult with groups that have large numbers of exemplars
(Hastie
& Park, 1986; Park & Hastie, 1987).
At the other end of the theoretical spectrum is the view that abstract
impressions
play a central role in producing group stereotypes, and that when a
categorization
task is encountered, exemplars will usually be important only insofar as
they influence the group stereotype. Pure abstraction models are not
supported
by evidence such as that of Mackie, Sherman, & Worth (1993), for
example,
that found that there is a positive correlation between information
recalled
about a group and judgments made about the group. Such evidence suggests
exemplar-based group knowledge. It is important to take note, however, of
the fact that there is ample evidence suggesting the opposite, i.e. that
there is a lack of positive correlation between information that
participants
can remember about group members and judgments they make about the group
(e.g. Park & Hastie, 1987; Judd & Park, 1988).
Discrepant findings such as these have led to dual process theories that
incorporate elements of both abstraction- and exemplar-based models
(Busemeyer
et al., 1984; Klein et al., 1992). Research on impression formation and
self-perception has supported such a mixed model of social perception
(Klein
et al., 1992). The conditions of the task have been found to determine to
what varying extent a perceiver will use abstract information or exemplar
knowledge to make a decision.
One variable that has been found to be important in this determination is
the amount of knowledge a perceiver has about the group. Sherman (in
press)
found evidence that the more information that is accumulated about a
group,
the more likely it is that an abstract impression will be used to answer
a trait judgment task such as, "Is Bob kind?" At low levels of
experience, group typicality knowledge is represented in terms of
specific
group members. But as experience increases, an abstract stereotype is
formed
that is stored and retrieved independently of knowledge about particular
exemplars.
Another variable that has been shown to play a role in determining which
mode of knowledge will be brought to bear on a trait judgment task is the
relevancy of the trait to the group stereotype (Sherman, in press). For
example, kindness is a dimension of importance to the stereotype of
priests:
they are typically kind. Intelligence, on the other hand, is an
irrelevant
dimension. When asked to judge whether a particular priest is kind,
abstract
information about priests in general may be preferentially used due to
the
highly accessible stereotype (Devine, 1989) of kind priests. This would
free the perceiver from the task of actually weighing exemplar knowledge
about the priest on-line. This would preserve attentional resources and
is consistent with the accepted trend (see S.J. Sherman, Judd, &
Park,
1989) that people, whenever possible, act as "cognitive
misers."
However, when asked if that priest were intelligent, exemplar knowledge
would be brought to bear since the stereotype of priests does not provide
information about intelligence to answer the question. Sherman found this
trend in his experiment with priest and engineer stereotypes. That
exemplars
are activated during one task (irrelevant task) and not the other
(consistent
task) provides further support for the idea that dependence upon
exemplars
depends on what task is being performed by the perceiver. A dual model
can
account for these findings whereas an either/or model cannot.
This tendency to automatically attribute stereotype-consistent traits to
targets is an alarming one since it flies in the face of logic about
rational
decision making. Even when subjects are presented with relatively small
amounts of information about an individual, as in Sherman's study (in
press),
they do not seem to activate exemplar information when asked to assess
whether
a target conforms to a stereotype-consistent trait. However, this
phenomenon
does make sense in the context of Devine's (1989) belief that stereotypes
are very tempting to the decision maker because of their easy
accessibility.
This reasoning would predict therefore that even low prejudice people who
do not embrace stereotype-consistent views, are nevertheless prone to
preferential
attention at the encoding stage and preferential retrieval at the
judgment
stage. Increased knowledge about a stereotype might not automatically
lead
to prejudicial behavior, but because a salient stereotype is at the
disposal
of the perceiver, the chances it may get used may increase. Even though
this tendency is not a consciously "motivated" prejudice, it is
due to motivational factors at the lower cognitive levels where encoding
and retrieval strategies come into play.
It has been shown, then, that amount of knowledge and relevancy of task
do influence the likelihood that exemplars will be activated in a
judgment
task about a target. An obvious question that has not been given much
attention
is, "Do clearly prejudicial people even use exemplars when rating
targets
on stereotype-consistent traits?" Sherman's study (in press) found
that questions with highly accessible answers ("Is this priest
kind?")
lead to relative nondependence on exemplars. If people with high levels
of prejudice were found to be using abstraction-based strategies and
low-prejudiced
people were found to be using exemplar strategies, then this would
provide
further support for a mixed model and would demonstrate another variable
that influences whether or not exemplars will be used in a categorization
task.
Previous research supports the notion that high and low prejudiced people
would use different strategies at the time of retrieval. Devine (1989)
argues
that whereas stereotypes are accessible and are activated almost
automatically,
to respond in a non-stereotypical way, such as low prejudiced people do,
requires inhibition of this stereotype in favor of less accessible, more
rational, belief-based strategies. This framework implies, then, that low
prejudiced people do have the same stereotype at disposal, but are more
careful in the application of the stereotypical knowledge. They would be
more likely to "check the evidence" by referring to the
exemplar
knowledge that is directly related to the task at hand. Low prejudiced
people,
therefore want to individuate, whereas high prejudiced people are content
with application of stereotypical knowledge.
Since along with retrieval strategies, encoding strategies are a relevant
issue in categorization tasks, attention should be turned to the
literature
on differential encoding strategies as a function of personal relevance
of information. First, it is important to note that attentional
allocation
is at the heart of encoding processes since it determines what will and
will not be stored in memory. Therefore, motivational factors (such as
prejudice)
are important to the memory system at the time of encoding only insofar
as they bias the type of information that is encoded and the salience of
that information in memory. At present, the schema-filter model provides
the dominant framework to explain how the memory system is influenced by
preconceived notions such as those held by people who are highly
prejudiced.
The schema-filter model asserts that the stronger a stereotype is, the
more
biased processing will be toward stereotype-confirming and away from
stereotype-disconfirming
information. Therefore, a person with high prejudice against homosexuals
will be more likely to fixate on stereotype-consistent exemplar
information.
If this is true, then highly prejudiced people should be more likely to
have salient impressions of a feminine (stereotype-consistent) gay man,
for example. Then at the time of recall, exemplar information should be
more salient to the person with high prejudice, which should in turn lead
to a greater likelihood to rely on exemplar information. Note that this
prediction runs counter to the earlier prediction based on the mixed
model,
such that low-prejudiced people, because of their desire to individuate,
should be more likely to rely on exemplars than people with high levels
of prejudice.
A way to accommodate both of these predictions is to examine what is
meant
by the phrase, "to rely on exemplar knowledge." People with
high
prejudice may have more salient exemplar knowledge, but may fail to use
it in a judgment task. And people with low prejudice may have weaker
impressions
of exemplar information, but may pay special attention to it at the time
of a judgment task. In this way, metacognitive strategies aimed at
reducing
cognitive dissonance may supersede the raw data (i.e. exemplar and
abstract
impressions) at the time of judgment. Therefore, a study to solve the
problem
of whether low or high prejudiced people have superior memory for
exemplars
is relevant to issues of representation, attention, and motivation.