The Effects of Cognitive Load on Covert Stereotype Propensity
In cognitive psychology people are often considered to be limited capacity processors and when their capacities are strained, they are susceptible to making mistakes and omissions. The literature has demonstrated that cognitive load, by means of task engagement, facilitates overt stereotyping by decreasing the capacity to self-monitor evidence of biases. The present study was designed to test whether covert stereotyping, where the participant is not aware of the stereotype being activated, is also facilitated by cognitive load. This study featured a 2 (Presentation: no task requirement vs. digit rehearsal task) X 2 (sex: male vs. female) X 2 (race: white vs. black) mixed factorial design with the task condition manipulated within subjects. In the experimental group, a number rehearsal task was used to administer cognitive load. Half of the participants were engaged in this task and the other half, the control group, were simply given the same number to observe, but they were not instructed to rehearse it. Both groups completed the Modern Racism Scale. Because the scale assesses stereotyping propensity under the guise of surveying political attitudes the scores on this scale are well accepted to show a reliably positive correlation with covert stereotyping propensity. The study found that there was not a statistically significant difference between the control and experimental groups. Gender, but not race, produced significant differences. The findings of this study suggest that cognitive load, through digit rehearsal, does not lead to an increased propensity for covert stereotyping.
A great deal of research supports the idea that both information processing demands (e.g. Bodenhausen & Wyer, 1985; Stangor and Duan, 1991) and information processing concerns (e.g. Erber & Fiske, 1984; Pendry & Macrae, 1999) can strongly influence a perceiver’s propensity to activate and apply stereotypes. Whether stereotyping, which is the formulation an oversimplified or biased opinion, is intentional or automatic it all too often results in the attribution of unfair negative characteristics to out-group members (Pendry & Macrae, 1999). These are normally the kind of negative associations that people do not want to share publicly, and so if they can, they will attempt to monitor these, disallowing them from their outward communications (Bodenhausen, 1990). Monitoring outward communications in this way requires cognitive processing resources, something that people have in limited supply.
A large body of research has shown that when processing resources are low due to the fact that they are distributed between different tasks, performance on each task suffers. Cognitive load (the detrimental effect that task engagement has on the performance of working memory) can affect many forms of parallel processing and can even impact the ability to monitor prejudice within verbal discourse (Bodenhausen & Wyer, 1985). This literature has suggested explicitly that most people continually utilize their processing resources to censor evidence of their own biases and that this is made even more difficult under the condition of cognitive load (Wyer & Martin, 1986). A deficit in mental resources caused by multiple task engagement has been shown to influence individuals to make more prejudiced responses by diminishing their capacity to inhibit biased personal thoughts and subsequently to revise biased interpersonal language (Fiske & Neuberg, 1990).
Research has shown that a variety of dissimilar methods for limiting a subject’s cognitive resources can result in the preferential recall for stereotype-consistent information. Methods that involve distracting subjects with an unrelated audio broadcast (Stangor & Duan, 1991), overwhelming them by expecting them to observe multiple groups items (Stangor & Duan, 1993), instructing them to engage in digit rehearsal tasks (Sherman, Lee, Bessenoff & Frost, 1998), using other forms of rehearsal tasks (Pendry & Macrae, 1999) and implementing a variety of other methods for administering cognitive load (Sherman & Frost, 2000) have all been shown to increase the propensity for stereotype formation. Each of the 5 studies cited above uses procedures to study stereotype formation that are obvious to the participants. Even the particular stereotype that they were trying to invoke can clearly be discerned and each study assumed that the majority of participants would have wanted to inhibit evidence of their prejudice but were less able to due to the diminishment of their processing resources.
Stangor and Duan (1991) found that by simply asking participants to attend to specific sounds during a radio broadcast, that they could create cognitive load during task engagement. Here cognitive load vastly decreased the ability of the participants to censor stereotype formation (p<.05). Two years later, Stangor and Duan (1993) required that participants attend to associations related to multiple out-groups, making it more difficult to determine which stereotypes were being tested. Again the experiment showed that cognitive load results in significant (p<.05) increases in the number of stereotype related biases. Sherman et al. (1998) found similar results five years later when distracting participants with a digit rehearsal task. This study, and their next one, (Sherman & Frost, 2000) showed that digit rehearsal caused their participants to become significantly (p<.05) less likely to inhibit evidence of bias despite the fact that they recognized the majority of stereotypes presented to them. Pendy and Macrae (1999) used either a series of letters or symbols for their rehearsal task and, interestingly, they found that a significant (p<.05) proportion of the participants in the experimental group reliably chose the biased response while completing their stereotype identification task. Subsequent studies have replicated these findings, lending support to the present research hypothesis. However, each of these studies relied on overt stereotype formation, where the stereotype presented in each task is easily identified and evidence of the subject’s identification of the stereotype is clearly indicated by their response to the tasks questions (Sherman & Frost, 2000). In other words, in these experiments, the subject was should have been relatively aware each time they gave a biased response.
This is the first experiment that uses a carefully constructed, covert survey instrument like the Modern Racism Scale in an experiment involving cognitive load. The Modern Racism Scale is designed to be a nonreactive measure of prejudice presented under the guise of surveying political attitudes (McConahay, Hardee & Batts, 1981). This guise puts those surveyed at ease and makes them less conscious about the biases that might be found in their responses. The use of such an instrument should test a different hypothesis because, unlike the procedures used in the aforementioned studies, it measures implicit as opposed to explicit stereotype formation.
This study was designed to determine how cognitive load, in the form of a number rehearsal task, effects information processing demands and relates to the propensity to activate stereotypes. Prejudice will be assessed by the numerical score on the Modern Racism Scale (McConahay, 1986). This controlled experiment required subjects to evaluate a questionnaire while performing a task designed to increase cognitive load. The task consisted of the memorization and constant rehearsal of a seven digit number throughout the experimental proceedings, a common procedure for inducing cognitive load (Sherman & Frost, 2000). While the subject rehearsed this number the experimenter administered the Modern Racism Scale, a survey where the subject finds it more difficult to identify the stereotypes that they unwittingly activating and committing to (Bodenhausen, 1990). The present researcher predicts that engagement in the digit rehearsal task might lead to a diminishment in processing resources but that because the stereotypes in this particular questionnaire are covert, cognitive load will not will not increase stereotyping propensity. If this hypothesized effect does occur it should be because when stereotype assessment is covert, cognitive processing resources cannot help a person to inhibit evidence of their mental biases.
Many people in modern society are known to rely on stereotypes and use them as cognitive heuristics to inform their schemas and to facilitate everyday information processing. Many group based stereotypes, though, have the potential to harm ingroup and outgroup members. By determining which types of processing modalities facilitate stereotyping, we can better understand how stereotypes are actualized within an individual, how stereotypes are perpetuated and possibly how prejudice, a culturally derisive phenomenon, can be better mitigated.
A total of 302 Pepperdine undergraduate students of ages 18 to 23 took part in this study. Each of the 134 male and 168 female students had graduated from high school and lived in the greater
area. 40 subjects were black, 262 subjects were white. Participants were recruited through the use
of flyers, and by word of mouth. Los Angeles
The study was an experimental, randomized controlled trial featuring a 2 (Presentation: no task requirement vs. digit rehearsal task) X 2 (sex: male vs. female) X 2 (race: white vs. black) mixed factorial design with the task condition manipulated within subjects.
The variables in studies like this one have been clearly operationally defined and this so should be relatively easy to replicate. One factor that may affect internal validity is the assumption that digit rehearsal increases stereotype propensity due to cognitive load. It is possible that frustration with task requirements increase negative affect and for this reason increase stereotype propensity. This is a very difficult confounding factor to control but the literature generally accepts the notion that digit rehearsal is a reliable task to administer cognitive load. Experimenters were instructed to present the task requirements to participants in a friendly, amiable manner in an effort to minimize negative affect.
The materials used included the McConahay (1986) Modern Racism Scale, which is administered to test for subtle forms of racism that are prevalent in the modern day US. This scale uses a response format that asks participants to indicate their agreement numerically; with numbers ranging from -2 (strongly disagree) to 0 (neither agree nor disagree) to +2 (strongly agree). A higher score on this scale indicates more blatant racism, which will give us our measure of stereotype formation.
A separate demographic questionnaire followed the scale. This questionnaire requested information about age, race, handedness, sex and number of years in college. This questionnaire did not request any information that would allow the participant to be identified, allowing complete anonymity.
Individuals from this sample of convenience chose to participate in this experiment after encountering flyers or verbal requests. The study began with the participant being welcomed into the office by the experimenter, escorted to a private room with a desk where they would remain for the duration of the study. Next, the participant was given a cover letter (Appendix B) and asked to fill out an informed consent form (Appendix C) that advised them about the risks associated with the study and required them to signature, indicating that they understood and agreed to the details of the experiment.
Before the administration of the questionnaires, one half of the subjects were randomly assigned to the task engagement group and given 25 seconds to rehearse an 8 digit number. The participants in the task condition were informed that upon completion of the experiment they will be required to reproduce this number. Previous research has consistently demonstrated that digit rehearsal tasks such as this one have debilitating effects on processing resources. That is, throughout the experimental task, they expend conscious resources rehearsing the 8-digit number and this influences their allocation of attentional resources to the experiment proper.
The control group was given 25 seconds to rehearse the same 8 digit number. Like the other group they were asked to memorize it but then were told that they would not be asked to reproduce the number upon completion of the experiment.
To ensure anonymity all subjects were isolated from the other participants and, after the necessary instructions had been given, from the experimenter as well. Written instructions informed subjects that the questionnaire that they were expected to respond to was created to help researchers better understand political attitudes.
Subjects were then required to complete the seven-item Modern Racism Scale. This scale is designed to measure subject’s racial stereotyping behavior in a covert, nonreactive fashion. Subjects responded to the questionnaire by indicating if they agree or disagree with specific race-oriented statements. The questionnaire took none of the participants more than 20 minutes to complete.
Before participants in the experimental group placed their questionnaires in their envelope, as instructed, they were visited by an experimenter who requested that they reproduce the number that they had been rehearsing. The experimenter then discarded any questionnaires taken by subjects that could not reproduce the entire series of numbers, these were not included in the statistical analyses. These questionnaires were discarded because they may represent a case where the participant did not carefully rehearse their number, did not experience cognitive load, and so did not fulfill the requirements of the experiment. In fact, 27 questionnaires were discarded for this very reason.
Subjects placed their responses in an unmarked envelope, and then dropped the envelope into the box that corresponded to their experimental group, both of which contained many envelopes. Finally subjects were debriefed and thanked for their participation.
The data collected were analyzed by computing the mean scores on the Modern Racism Scale and using them to perform a linear regression with score as the dependent variable and gender, race and experimental group as the independent variables. It was assumed that the independent variables might be correlated with one another, so a multiple regression analysis was also performed in order to find out if the predictor variables add independent information to the prediction equation. Differences between groups would only be interpreted as significant if they satisfy a p-value less than .05.
Supporting the hypothesis, there was no significant difference in scores between the control group (M= -.403 ,SD= 1.212) and the experimental group (M= -.423, SD=1.114) as measured by the participant’s scores on the Modern Racism Scale (p-value <.05). It seems that task engagement did not increase the propensity for a racist score (a higher score) on the racism scale. The averages for individual groups are shown in table 1 below. The means for each group were calculated by using the average score, from -2 to 2 for each item on the scale rather than from the total score for each participant.
Table 1: Mean Scores for Control and Experimental Groups
M SD n M SD n
Male -.561 2.42 70 -.580 2.07 64
Female -.286 1.03 83 -.305 1.19 85
Black -.413 2.13 18 -.391 2.22 22
White -.418 1.81 135 -.421 2.10 127
Total -.403 1.21 153 -.423 1.11 149
A linear regression was performed with score on the Modern Racism Scale as the dependent variable, and 3 predictor or independent variables: gender, race and experimental group. The regression equation from the data set was:
Predicted Score = -.726 + .133*Experimental Group + .160*Gender + .149*Race
It seems that only gender, with a significance level of .022, played a role as a significant factor (at a p level <.05) in racism score. Neither experimental group nor race was significant contributors to racism score. The regression coefficients, starting with an intercept constant of -.726, for each predictor were as follows: .133 for the experimental group, .160 for gender, and .149 for race.
The main hypothesis that task engagement will have no effect on score on the Modern Racism Scale was supported. This experiment suggests that sometimes situational characteristics may play a smaller role in stereotype formation than once thought by others such as Spears and Haslam (1997), Oakes and Turner (1990) and Medin (2000). Since the participants in the present study were not exposed to blatant, overt depictions of stereotypical behavior, they may not have formed stereotypes due to conscious, deliberative attempts to understand social groups, but instead have automatically activated the heuristic based stereotypes due to a paucity of processing resources. Why race did not produce significant results is not known and the paucity of research in this area leaves this question open to further investigation.
Because the scale that was used measures covert bias, and because cognitive load did not affect stereotyping, the current findings suggest that people do not engage in conscious self-monitoring when they take the Modern Racism Scale. Knowledge that some forms of stereotyping and racism can increase even when processing resources are at their fullest may be informative for social psychology researchers (Fiske and Taylor (1991). This may be an important issue and should motivate psychologists to determine exactly how and under what circumstances cognitive load can create subconscious biases.
Previous studies have emphasized the importance of analyzing the effects of cognitive load on stereotype formation from a variety of perspectives in order to determine how stereotypes are formed (Stangor & Duan, 1991; 1993). It is suggested that future studies increase the sample size of the experimental group to increase validity. It may also be important to ensure that the number of black participants and the number of male participants are increased in these samples because the present study had a relatively low number of both of theses groups. The use of a more ethnically diverse sample, along with other means of stereotype assessment should shed even more light on the issue. It is also recommended that future replications of this study design implement alternative forms of task engagement (besides digit rehearsal) to administer cognitive load. This will allow researchers to better understand how and why stereotypes are formulated outside of the experimental sphere, in everyday interpersonal interactions.
Bodenhausen, G. V. (1987). Social stereotypes and information-processing strategies: the impact of task complexity. Journal of Personality and Social Psychology, 52,871-880.
Bodenhausen, G. V. & Wyer, R. S. (1985). Effects of stereotypes on decision making and information processing strategies. Journal of Personality and Social Psychology, 48,267-282.
Erber, R. & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47, 709-726.
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 and Social Psychology, (Vol. 23, pp. 1-74).
: Academic Press. New
Fiske, S. T. & Taylor, S.E. (1991). Social Cognition.
McGraw Hill. New York
McConahay, J. B. (1986). Modern racism, ambivalence, and the modern racism scale. In J. F. Dovidio & S. L. Gaertner (eds.), Prejudice, discrimination and racism (pp. 91- 126).
: Academic. New York
McConahay, J. B., Hardee, B. B. & Batts, V. (1981). Has racism declined in
? It depends on who is
asking and what is asked. Journal of Conflict Resolution, 25, 563-579. America
Medin, D.L. (2000) Are there kinds of concepts? Annual Review of Psychology, 51,121-147.
Oakes, P.J. & Turner, J.C. (1990). Is limited information processing the cause of social stereotyping? European Review of Social Psychology, 11,111-135.
Pendry, L.F. & Macrae, C.N. (1999). Cognitive load and person memory: the role of perceived group variability. European Journal of Social Psychology, 29,925-942.
Sherman, J. & Frost,
(2000). On the
encoding of stereotype-relevant information under cognitive load. Personality and Social Psychology Bulletin, 26,
Spears, R., Oakes, P.J., Ellemers,
(eds.) (1997) The social
psychology of stereotyping and group life. N. & Haslam, S.A. Oxford,
UK and :
Blackwell. Cambridge, MA
Stangor, C. & Duan C. (1991). Effects of multiple task demands upon memory for information about social groups. Journal of Experimental Social Psychology, 27,357-378.
Stangor, C. & Duan C. (1993). Effects of task demand upon stereotype formation. Journal of Experimental Social Psychology, 29, 121-127.
Wyer, R. S. & Martin, L. L. (1986). Person memory: The role of traits, group stereotypes, and specific behaviors in the cognitive representation of persons. Journal of Personality and Social Psychology, 50, 661-675.