How and When Do Individuating Information and Social Category Information Influence Implicit Judgments of Individual Members of Known Social Groups? A Review

Abstract

The present review discusses the literature on how and when social category information and individuating information influence people’s implicit judgments of other individuals who belong to existing (i.e., known) social groups. After providing some foundational information, we discuss several key principles that emerge from this literature: (a) individuating information moderates stereotype-based biases in implicit (i.e., indirectly measured) person perception, (b) individuating information usually exerts small to no effects on attitude-based biases in implicit person perception, (c) individuating information influences explicit (i.e., directly measured) person perception more than implicit person perception, (d) social category information affects implicit person perception more than it affects explicit person perception, and (e) the ability of other variables to moderate the effects of individuating information on stereotype- and attitude-based biases in implicit person perception varies. Within the discussion of each of these key points, relevant research questions that remain unaddressed in the literature are presented. Finally, we discuss both theoretical and practical implications of the principles discussed in this review.

1. Introduction

Imagine a hiring manager reviewing an application for a job. After interviewing the candidate, the manager has access to the applicant’s social category information (e.g., the applicant’s race, gender, etc.) and qualifications. Which type of information will weigh more heavily in the hiring decision?

In this hypothetical scenario, the manager may rely on social category information and relevant biases, individuating information (i.e., any information known about an individual social target other than their social category information [1]), or some combination of both in evaluating the applicant. How and under what circumstances do social category information and individuating information influence implicit (i.e., indirectly measured [2]) judgments of individual members of known (i.e., existing) social groups? As exemplified above, this question has implications for decisions that influence consequential life outcomes.

In this review, we discuss research that has addressed this question in the domain of both stereotype- (i.e., beliefs about social groups and their individual members [3]) and attitude- (i.e., global positive or negative feelings toward targets, such as social groups and their individual members) based biases in implicit person perception (i.e., indirectly measured perceptions of individual people; person perception is utilized interchangeably with judgments of individuals in this review). However, we first provide a brief review of this question as it pertains to explicit (i.e., directly measured; [2]) person perception for two reasons: (a) the issue of when and how much people rely on individuating information versus social category information was addressed in research on explicit person perception long before implicit bias (i.e., biases in judgment that are measured indirectly) became an area of research, and (b) earlier work on explicit person perception helps elucidate the core themes that have subsequently been addressed in research on implicit person perception.

There are several related topics that are beyond the scope of the present review. First, we do not address implicit bias writ large. In addition, we do not address effects of individuating information on perceptions of social groups, nor do we discuss any methods of bias reduction other than reliance on individuating information. Moreover, we do not review studies examining effects of individuating information and social category information on implicit perceptions of members of novel (i.e., unknown prior to the study, and usually fictitious) social groups, as perceptions of such groups and their individual members are fundamentally different than those of members of existing (i.e., already-known) social groups (for a discussion, see [4]). Finally, we do not discuss sources in which individuating information did not suggest a clear pattern that could be hypothesized to shift attitude- or stereotype-based biases in implicit person perception in any particular direction.

2. Reliance on Social Category Information and Individuating Information in Explicit Person Perception

Early theories of explicit person perception posited that people primarily rely on social category information in explicit person perception [5,6]. However, a subsequent theory took the opposite perspective: that people primarily rely on individuating information in such perceptions [1].

The results of a large meta-analysis of the empirical evidence available at the time of the meta-analysis supported the latter proposition. The average effect size for the effect of highly diagnostic (i.e., highly relevant to the judgment at hand) individuating information on explicit person perception in the presence of social category information was very large—r = .71—compared with a far smaller average effect size of social category information even in the absence of individuating information: r = .25 [1].

3. Why Study This Topic Using Implicit Measures?

Unlike explicit biases (i.e., biases in judgment that are measured directly via self-report), implicit biases have a degree of unintentionality [7]. It is possible that this element of unintentionality leads to reduced potential for demand effects (particularly social desirability bias, which is participants’ desire to portray themselves in a socially favorable manner) to affect scores on implicit measures (i.e., measures that assess a target construct indirectly). Moreover, one meta-analysis found that explicit and implicit measures predict behaviors toward members of social outgroups uniquely and approximately equally [8]. Thus, it would be scientifically unjustified to presume that the patterns of reliance on social category biases versus individuating information found in studies of explicit person perception necessarily apply to implicit person perception. Consequently, we embarked on a program of research systematically investigating this question; the present paper reviews that work as well as the work reported in publications by others that has also addressed it.

Below, we first provide some foundational information. Next, we provide methodological details of our literature search and discuss key thematic points that emerged from this literature. Finally, we discuss theoretical and practical implications of the body of research as a whole.

4. Foundational Information

4.1. The Implicit Association Test

In most studies that have addressed the topic of this review, participants completed Implicit Association Tests (IATs [9]) to measure stereotype- or attitude-based biases in implicit judgments of individuals, or to measure individuating information effects on implicit person perception (the latter are defined below). The IAT is a response latency measure in which participants sort stimuli into categories (target and attribute categories) that are paired in stereotype-consistent vs. stereotype-inconsistent ways. It is assumed that faster response times indicate stronger automatic associations between whichever type of category pairing (stereotype-consistent or stereotype-inconsistent) is present during that particular trial. Thus, stereotype- or attitude-based biases in implicit judgments occur when people have faster response times to stereotype- or attitude-consistent than to stereotype- or attitude-inconsistent pairings. Operationally, each perceiver’s mean response latency for stereotype- or attitude-consistent trials is subtracted from their mean response latency for stereotype- or attitude-inconsistent trials, and this difference is divided by the standard deviation of all of that participant’s responses [10]. This results in D scores, which are IAT scores. Higher D scores indicate more bias in implicit judgments.

4.2. Interpreting IAT Scores in Person Perception

According to one common interpretation of the IAT (which is also common in the literature reviewed here, e.g., [11]), D scores that are significantly above zero indicate stereotype- or attitude-based biases in implicit judgments (cf. [12] for an alternative interpretation). Scores that inferentially equal zero (i.e., do not statistically significantly differ from zero) indicate that the perceiver demonstrates no stereotype- or attitude-based biases in implicit judgments in either direction. D scores significantly below zero indicate biases in implicit judgments that are inconsistent with stereotypes or attitudes.

In this review, at times, we discuss whether stereotype- or attitude-based biases in implicit judgments of individuals were consistent or inconsistent with the direction of the individuating information. We discuss these classifications next.

4.3. Consistency vs. Inconsistency of Stereotype- or Attitude-Based Biases with Individuating Information

We state that, sometimes, stereotype- or attitude-based biases in implicit person perception were consistent with the direction of the individuating information. This description is utilized in two situations. The first is when we state that such biases were reversed in direction by counterstereotypic or counterattitudinal individuating information. This occurs when we have evidence that D scores for the stereotype- or attitude-based bias in implicit person perception significantly differed from zero in one direction (either positive or negative) in the absence of individuating information, but that in the presence of highly diagnostic individuating information that portrays targets in the opposite direction of the initial stereotype- or attitude-based bias, the D scores for person perception judgments reverse (compared to their original direction) and are significantly different from zero. The following hypothetical example clearly illustrates this. Consider a D score of 0.50 indicating that people stereotype men as taller than women. If perceivers then receive information that Joan is 6 feet tall and John is 5 feet 6 inches tall, a person perception IAT may produce a D score of −0.50, indicating that they implicitly perceive Joan as taller than John. The D score is reversed in direction. This type of pattern (broadly speaking) is the first type that indicates an implicit judgment of individuals that is consistent with the direction of the individuating information.

Another pattern demonstrating that stereotype- or attitude-based biases in implicit judgments of individuals align with the direction of the individuating information is when such biases shift to inferentially equal zero in the presence of information suggesting the two targets are equal on the judgment-relevant attribute. Consider a variation on the hypothetical example wherein people stereotype men as taller than women. If a man and a woman were portrayed as equal in height, D scores that shifted to inferentially equal zero aligned with the direction of the individuating information (which showed no difference between the targets), as is the case with D scores that reverse in direction in the presence of counterstereotypic or counterattitudinal individuating information.

In contrast, sometimes, we state that the stereotype- or attitude-based biases in implicit person perception were inconsistent with the direction of the individuating information. This occurs in two circumstances. The first circumstance is when individuating information portrays targets as equal on the judgment-relevant attribute, but a significant stereotype- or attitude-based bias remains in implicit person perception, and this bias is in the direction of the original bias. To illustrate, consider another variant on our hypothetical example. If a man and a woman were portrayed as equal in height, but stereotype D scores shifted from 0.50 to 0.30 in the presence of this highly diagnostic individuating information, we would describe the stereotype- or attitude-based bias in implicit person perception as being inconsistent with the direction of the individuating information since the judgment shows a perception of a difference in height, even though there was none.

The second circumstance in which we describe biases as inconsistent with the direction of the individuating information is when individuating information is counterstereotypic or counterattitudinal, and D scores either shift to inferentially equal zero or remain inferentially above zero in the presence of this information. This can also be illustrated by our hypothetical example. If the woman were portrayed as six feet tall and the man as five feet six inches tall, and the person perception D score was inferentially equal to zero (rather than being, say, −0.50), we would also refer to the stereotype-based bias in implicit person perception as being inconsistent with the direction of the individuating information. The same would be true if the original D score shifted from 0.50 to 0.30 in the presence of this same individuating information. This is because the individuating information illustrates a stereotype-inconsistent difference between the targets, yet D scores show no difference in judgments of the two targets or a stereotype-consistent difference between targets (respectively).

Thus, when we discuss stereotype- or attitude-based biases in implicit judgments of individuals being inconsistent with the direction of counterstereotypic or counterattitudinal individuating information, D scores in the presence of such information either (a) shifted to inferentially equal zero or (b) were significantly above zero. For stereotype- or attitude-based biases in implicit judgments of individuals to be considered inconsistent with the direction of individuating information that portrayed the targets as equal on the relevant attribute, D scores were significantly above zero in the presence of this information. However, if any of these outcomes occurred and D scores were significantly different than they were in the absence of individuating information, we consider the outcomes to nonetheless indicate a reduction of bias because, even though perceivers’ implicit judgments did not completely align with the direction of the information that was provided to them, these judgments did significantly shift due to the individuating information.

4.4. The Evaluative Priming Task

The evaluative priming task (EPT; [13,14]) has also been utilized to measure attitude-based biases in implicit judgments of individuals e.g., [11], though it has been utilized in fewer studies than the IAT. In the EPT, participants are presented with a prime and then a target stimulus. Participants are tasked with categorizing the target stimulus by valence while minimizing response latency. The logic is that participants should categorize target stimuli as positive more quickly after viewing a prime that they evaluate favorably, and vice versa for negative targets. Typically, researchers compare participants’ response latencies on trials where primes and targets are congruent to those where primes and targets are incongruent, though exact scoring procedures vary (e.g., [15]).

5. Literature Search Method

In Google Scholar, we searched for individuating information implicit person perception and combed through the first 300 results (of 32,800) when results were sorted by relevance. Of these 300, 12 were included in the present review; the others were not relevant or met exclusion criteria (which, to reiterate, were papers that addressed implicit bias writ large, the effects of individuating information on perceptions of social groups, methods of bias reduction other than reliance on individuating information; studies in which individuating information did not suggest a clear pattern that would be hypothesized to shift implicit attitudes or stereotypes; and studies addressing reliance on social category information and individuating information in implicit perceptions of novel—as opposed to known—social groups). Once we had gone through the first 300 articles, the last 148 were completely irrelevant to our review, so we reasoned that it was unlikely that we would find further relevant papers and thus decided to stop our search at that point.

In addition, we performed the following four keyword searches in PsycInfo: individuat* AND automat*; individuat* AND implicit*; impression formation AND implicit*; and individuat* AND impression formation. The first of these searches yielded four results, and one was included in the present review; the others were not relevant or met exclusion criteria. The second search yielded 12 results, and two were included in the present review; the others were not relevant or met exclusion criteria. The third of these searches yielded 28 results and two were included in the present review; the others were not relevant or met exclusion criteria. The last of these searches yielded 15 results and two were included in the present review; the others were not relevant or met exclusion criteria.

We utilized two additional methods of searching the literature after performing the searches above. First, using Google Scholar, we looked through all of the papers that had cited each of the papers we had already identified. This yielded no additional papers to add to the review. Then, we looked at the reference sections of all of the papers that we had identified. This also yielded no additional papers to add to the review. Table S1 (in Supplemental Materials https://osf.io/24fqd/?view_only=8df8aaaa39e84f27af191ebf476d5fc8) presents a brief synopsis of the methods and results of implicit measures for all studies discussed in the present review.

After using the search methods detailed above, we organized the relevant papers by themes. We discuss key thematic points that emerged from these papers below.

6. Key Thematic Point 1: Individuating Information Moderates Stereotype-Based Biases in Implicit Person Perception

Stereotype-based biases in implicit person perception generally do shift on the basis of individuating information. Thus, the presence of such information moderates stereotype-based biases in implicit person perception. However, the nature of these shifts varies, as explained below.

6.1. Sometimes, Stereotype-Based Biases in Implicit Person Perception Are Consistent with the Direction of the Individuating Information

A series of studies found that stereotype-based biases in implicit judgments of individuals found in the absence of individuating information shifted to become consistent with the direction of the individuating information that perceivers learned, indicating that individuating information moderates stereotype-based biases in implicit judgments of individuals. These studies showed that such biases were either reversed in direction by counterstereotypic individuating information ([16] Study 2; [17] Studies 1 & 2; [18] Studies 2a, & 2b) or that stereotype-based biases in implicit judgments of individuals were shifted to inferentially equal zero by highly diagnostic individuating information suggesting that the targets are equal on the stereotyped attributes ([19], Study 3). These studies all share the same general paradigm, in which perceivers are presented with two individuals. The individuals’ social categories vary within subjects in this paradigm (e.g., presentation of one Black and one White individual). Perceivers learn either only the targets’ social category information (sometimes accompanied by photos of the individuals [19,20]) or learn the targets’ social category information and also highly diagnostic individuating information. The individuating information has taken the form of vignettes [16,17,18,19,20,21,22,23] or photo pairings (i.e., photos of the target paired with photos of objects [19]).

One additional study using this paradigm ([18], Study 1) found no significant stereotype-based bias in implicit judgments of individuals in the presence of only social category information. However, the D score was negative in the presence of counterstereotypic individuating information. Thus, while bias did not technically reverse in direction (since it was inferentially equal to zero rather than positive to begin with), reliance on individuating information followed the same broad pattern as the studies discussed above; i.e., the direction of implicit judgments of individuals was consistent with that of the individuating information and not with the implicit belief about the groups.

The stereotype-based biases in implicit judgments of individuals in these studies encompassed multiple target groups and various stereotyped attributes. These included the following: (a) Hispanic individuals are street-smart while White individuals are book-smart ([16] Study 2), (b) Black individuals are athletic while White individuals are intelligent ([17] Study 1), (c) Black individuals are unintelligent while White individuals are intelligent ([17] Study 2; [19] Study 3), and (d) gay men are flamboyant while heterosexual men are masculine [20].

6.2. Sometimes, Stereotype-Based Biases in Implicit Person Perception Are Reduced by Individuating Information, but Remain Inconsistent with the Direction of the Individuating Information

In contrast, some research has found that stereotype-based biases in implicit judgments of individuals were significantly reduced by highly diagnostic individuating information, but that D scores in the presence of the information were inconsistent with the direction of the individuating information. This research has manipulated the presence or absence of individuating information. In within-subjects designs, participants’ implicit judgments of individuals were assessed twice: once after reviewing only social category information, and once after reviewing the highly diagnostic individuating information [22]. In between-subject designs, one group of participants received only social category information and a second group received both social category and individuating information ([23]1 Study 1; [16] Studies 1, 3, & 4; [21]). All of these studies found a statistically significant stereotype-consistent D score in the presence of the individuating information that was significantly lower than the D score in the absence of individuating information. Such results support the notion that individuating information moderates stereotype-based biases in implicit person perception since D scores did significantly change in the presence of the individuating information in a direction consistent with the information, though it does not provide evidence that is as strong as that discussed above since the direction of the scores was not consistent with that of the individuating information.

The research showing this pattern has investigated several target groups and attributes, including the following: (a) Black individuals are unintelligent while White individuals are intelligent [23], (b) Hispanic individuals are Spanish-speaking while White individuals are English-speaking ([16] Study 1), (c) Hispanic individuals are street-smart while White individuals are book-smart ([21] Studies 1 & 3), (d) men are doctors while women are nurses ([22] Study 1), (e) men are scientists while women are artists ([22] Study 2), (f) women are empathetic while men are assertive ([16] Study 4; [21] Studies 2 & 4), and (g) to dress up, men wear tuxedos while women wear dresses ([16] Study 3).

6.3. Unaddressed Questions

Despite the growing body of evidence that individuating information moderates stereotype-based biases in implicit judgments of individuals, some questions remain unexplored. The most obvious, perhaps, is why highly diagnostic individuating information sometimes causes stereotype-based biases in implicit judgments of individuals to correspond with the direction of the individuating information, while other times it does not. Contending explanations include the following: (a) the types of “highly” diagnostic information studied to date may nonetheless vary in diagnosticity so that perhaps stereotype-based biases in implicit judgments of individuals do not correspond with the direction of the individuating information when it is less diagnostic, and (b) perhaps outcomes hinge on the amount of highly diagnostic information (with a greater amount more likely to cause implicit judgments to correspond with the direction of the individuating information); this has varied slightly between studies. Future research should address these questions to improve the scientific understanding of how highly diagnostic individuating information can mitigate stereotype-based biases in implicit judgments of individuals.

Another unanswered question concerns the extent to which these patterns apply to groups not yet studied. It is possible that these effects do not apply to all target groups. Many target groups still have not been examined in the context of this research question (e.g., many groups in the LGBTQ+ community, age groups, individuals who have problems with their mental health, SES-based groups, political groups). Future research should continue to explore the role of individuating information in moderating stereotype-based biases in implicit judgments of individuals with respect to the many stereotypes not yet studied.

In addition, although research has presented perceivers with stereotype-relevant individuating information in written format and in the format of photo pairings and assessed its effects on stereotype-based biases in implicit person perception, these information learning modalities lack ecological validity (ability to generalize to the real world). It will be important to develop more ecologically valid approaches to having participants learn stereotype-relevant individuating information (as Navon et al. [11] did when examining the effects of individuating information vs. social category information on attitudes in implicit person perception, discussed below).

Finally, all of the research reviewed above has measured implicit stereotype-based biases in implicit judgments of individuals immediately after individuating information was presented. Thus, the question of the longevity of individuating information effects (assuming that perceivers remember the individuating information) on stereotype-based biases in implicit person perception remains unaddressed and is an important question for future research.

7. Key Thematic Point 2: Individuating Information Usually Has Little to No Effect on Attitude-Based Biases in Implicit Person Perception

Based on extant empirical evidence, individuating information moderates social category biases to a lesser extent when the dependent measure is attitude-based biases in implicit judgments of individuals instead of stereotype-based biases in implicit judgments of individuals ([11,24,25]; cf. [18]). For instance, Navon et al. ([11]; Studies 1–23) examined reliance on individuating information in implicit attitude-based biases toward well-known targets (i.e., celebrities) operating on the assumption that, since these individuals are widely known, perceivers had accrued individuating information about them. Navon et al. chose targets for whom participants had previously reported that they liked the stigmatized group member more than the non-stigmatized group member on explicit measures (i.e., measures that assess a target construct via self-report; Studies 1–21) or for whom participants reported that they liked the two targets equally (Studies 22–23). Thus, they assumed that if D scores were positive (indicating a more positive implicit attitude toward the non-stigmatized target), social category information was causing this result, whereas if D scores were negative (indicating a more positive implicit attitude toward the stigmatized target), individuating information was causing this result. They found that, despite participants usually (with the previously noted exception of Studies 22–23) expressing more positive attitudes toward the stigmatized group member on explicit measures, on implicit measures, participants usually (i.e., in 13 out of 23 studies) showed significantly more positive attitudes toward the non-stigmatized group member. This pattern is consistent with the conclusion that target social category membership oftentimes continued to bias implicit attitudes towards the individual celebrity targets despite the presence of individuating information and shows inconsistency between the bias and the direction of the individuating information.

Other research ([24], Studies 3–4) examined effects of counterstereotypic economic political ideology information vs. counterstereotypic social political ideology information on attitude-based biases in implicit judgments of political outgroup members. Participants identified as Democrats or Republicans and received one of the two types of counterstereotypic individuating information or no individuating information about targets who belonged to the opposing political party. In the counterstereotypic information conditions, participants learned that targets held social or economic ideologies consistent with participants’ political parties. Results indicated no difference in the implicit attitude-based biases towards the individual targets in the presence versus absence of this individuating information, showing no moderation of implicit attitude-based biases by individuating information.

In another series of studies, McConnell et al. ([25] Studies 1–3) provided participants with positive or negative behavioral individuating information about members of stigmatized groups (i.e., individuals with higher weights, Black individuals, or unattractive individuals) or non-stigmatized groups (i.e., individuals with moderate weights, White individuals, or attractive and attractiveness-neutral individuals). Following this, some participants read neutral behavioral information about the targets, and other participants read information that contradicted the valence of the initially presented information. For members of non-stigmatized groups, attitude-based biases in implicit judgments of the targets reflected the valence of the initial behavioral information and then reversed in the presence of the contradictory information. However, for members of stigmatized groups, attitude-based biases in implicit judgments of the target were negative regardless of the valence of the initial or contradictory behavioral information. The same pattern of findings appeared in Study 4, where only the initial set of behavioral statements was provided (i.e., there was no second set of behavioral statements that was either contradictory in valence or neutral). Implicit judgments were biased against the target with the higher weight regardless of whether positive or ambiguous behavioral information had been presented about that target. Thus, in this program of research as a whole, results suggest that individuating information did not moderate the effects of attitude-based biases on implicit person perception.

However, other research has not necessarily yielded results that conform to this pattern. Rubinstein et al. [18] examined implicit attitude-based biases against individuals with higher weights and against individuals who practice Islam either in the presence of counterstereotypic individuating information or in the presence of only social category information. This research found no significant attitude-based bias against individuals with higher weights nor against individuals who practice Islam in the presence of only social category information. When counterstereotypic individuating information was presented, implicit attitude-based biases in judgments of these individuals shifted to become significantly negative, and thus consistent with the individuating information that was presented (although the individuating information was counterstereotypic rather than counterattitudinal, the valence of the information about the non-stigmatized group member was negative, and the valence of the information about the stigmatized group member was positive).

Unaddressed Questions

While Rubinstein et al. [18] did measure attitude- and stereotype-based biases in implicit person perception in the presence of the same individuating information, the attitude-based biases (and one of the stereotype-based biases) intended to be studied by that research were nonsignificant in the presence of only social category information (i.e., in the absence of individuating information). Thus, it is possible that these biases were easier to shift than initially significant attitude-based biases in implicit judgments of individuals. Future research should test whether implicit attitude-based biases in implicit judgments of individuals are more resistant to change in the presence of counterinformation than implicit stereotype-based biases in implicit judgments of individuals by measuring both of these types of biases in the presence of the same individuating information, but using implicit attitude-based biases in implicit person perception that are clear and strong in the absence of counterinformation.

8. Key Thematic Point 3: Individuating Information Affects Implicit Person Perception, but Not as Strongly as It Affects Explicit Person Perception

The studies reviewed above investigated the effects of individuating information on stereotype- and attitude-based biases in implicit person perception. However, these studies did not examine the magnitude of the individuating information effect on implicit person perception itself. This is important for both theoretical and applied reasons. Individuating information effects on implicit person perception (rather than on stereotype- and attitude-based biases on implicit person perception) quantify the difference that variation in individuating information itself makes in implicit judgments of individuals, thereby measuring the effect of individuating information on implicit person perception rather than its effects on attitude- or stereotype-based biases in implicit person perception. Such evidence is important because it quantifies exactly how sensitive implicit judgments of individuals are to individuating information regardless of their social category membership. Effects of individuating information on person perception go to the heart of theoretical and applied issues, such as “To what extent do people judge others on their merits?” and “How much do targets’ behaviors and accomplishments influence others’ judgments of them?” In this section, we compare the magnitude of individuating information effects on implicit measures to that on explicit measures within the same study (i.e., wherein experimental manipulations are constant across implicit and explicit measures), thereby providing the strongest evidence for the potential differential sensitivity of such measures to individuating information.

To measure individuating information effects in implicit person perception, three studies have employed paradigms in which two targets belonging to the same social category (e.g., two Black individuals) were presented. Thus, instead of varying the targets’ social group membership, the content of the individuating information varied across conditions. In the first such study ([23] Study 3), participants reviewed college applications from two fictitious Black college applicants or two fictitious White college applicants. One of these applicants was portrayed as an excellent applicant (high SAT score, high GPA, etc.) and the other was portrayed as a weak applicant (low SAT score, low GPA, etc.). The stereotyped attributes that were measured in the study were intelligent and unintelligent. Thus, D scores in the study measured participants’ perceptions of how much more intelligent the excellent applicant was compared to the weak applicant when these applicants belonged to the same race group. The original study did not report an analysis of D scores in which these scores were inferentially compared to zero. Reanalysis of the original data using this method found an individuating information effect that was significant and large in magnitude according to standard interpretive effect size guidelines [26]: M = 0.30, SD = 0.30, t(211) = 14.51, p < .001, d = 1.00 [95% CI: 0.83, 1.16].

In the same study, individuating information effect sizes for explicit measures were d = 2.33 [95% CI: 2.02, 2.63] for an IQ estimate variable for the strong vs. weak applicant, d = 3.45 [95% CI: 3.07, 3.84] for GPA predictions of the strong vs. weak applicant, and d = 3.69 [95% CI: 3.28, 4.10] for competence ratings of the strong vs. weak applicant. The 95% CIs for the Cohen’s d values for the IAT did not overlap with any of those for the explicit measures, and the direction of the difference suggested that individuating information had a stronger effect on explicit measures than it did on the implicit measure.

In the two other studies utilizing this paradigm ([19] Studies 1 & 2), participants were presented with highly diagnostic individuating information about the intelligence of either two White targets (Study 1) or two Black targets (Study 2). This information took the form of either vignettes or photo pairings (photos of each target paired with photos representing the information from the vignettes provided to other participants). In an analysis original to this review, aggregating across the two studies and the vignette vs. photo pairing conditions (which did not significantly differ from one another), the individuating information effect was significant and large in magnitude, M = 0.41, SD = 0.29, t(228) = 21.24, p < .001, d = 1.40 [95% CI: 1.22, 1.59].

In the same aggregated dataset, there were two explicit dependent measures that asked participants different questions. Individuating information effect sizes for these explicit measures were d = 1.24 [95% CI: 1.01, 1.47] for an IQ estimate variable for the strong vs. weak applicant and d = 2.03 [95% CI: 1.74, 2.32] for intelligence ratings of the strong vs. weak applicant. Thus, in comparisons of each of the two explicit measures with the implicit measure, 95% CIs for the Cohen’s d values for the IAT overlapped with those for explicit measures in one of two cases. The direction of the difference between those that did not overlap suggested that individuating information had a stronger effect on the explicit measure than it did on the implicit measure.

Another paradigm has also been used to approximate a test of the magnitude of individuating information effects in implicit person perception, though it is not a direct test of the magnitude of such effects. Navon et al. ([11] Studies 28 and 29) had participants implicitly evaluate either a pair of celebrity targets for which their self-reports indicated that they preferred the stigmatized target (who was either Black or older) over the non-stigmatized target (who was either White or younger; i.e., the pro-stigmatized condition), or for which their self-reports indicated that they preferred the non-stigmatized target over the stigmatized target (i.e., the pro-dominant condition). The assumption was that over time, perceivers had accrued positive individuating information for the more liked target, and negative individuating information for the less liked target since information about celebrities is widely available. Thus, the identification of participants’ explicit attitudes towards the targets approximated a form of individuating information.

In Navon et al.’s design, higher IAT scores were interpreted as a greater preference for the White or younger target over the Black or older target. These IAT scores were higher in the pro-dominant condition than they were in the pro-stigmatized condition. The results were interpreted as follows:

“Because the group membership was the same in both conditions, we attribute this difference to the individuating information about the targets. Therefore, these results suggest that the individuating information about the targets influenced automatic preference” (pp. 515–516).

The logic here was that if IAT scores indicated a more positive attitude toward the stigmatized group member in the pro-stigmatized condition than in the pro-dominant condition, this would be consistent with the notion that individuating information was influencing implicit attitudes. This is because the extent to which the positive vs. negative individuating information was presumably affecting D scores was being measured via the difference in D scores between these two conditions. Specifically, D scores that were lower in the pro-stigmatized condition than in the pro-dominant condition showed that the more positive individuating information about the stigmatized group member and more negative individuating information about the non-stigmatized group member was reducing D scores relative to the pro-dominant condition.

In another analysis original to the present review, aggregating across the two studies, the implicit attitude difference—i.e., the magnitude of the difference between the D scores in the two conditions—was d = 0.40, suggesting that the overall effect was small to medium. On the other hand, the magnitude of the difference between the explicit measure of liking in the two conditions was d = 1.44, which is a large effect. This suggests that individuating information affected explicit person perception more than it affected implicit person perception.

Moreover, Navon and Bar-Anan [27] varied the extremity of the valence of individuating information about targets by varying how many good behaviors the positively portrayed target performed and how many bad behaviors the negatively portrayed target performed and which target of the pair (belonging to different race, age, or gender groups) was portrayed positively (vs. negatively). This research measured implicit and explicit attitudes toward the targets about whom participants had learned. In Studies 1–3, the implicit measure was the IAT, while in Studies 4–6, it was the EPT. Scores on the dependent measures reflected the extent to which participants preferred the positively portrayed target over the negatively portrayed target (regardless of their race), thus reflecting implicit person perception (rather than stereotype- or attitude-based biases in implicit person perception). We performed three novel analyses on these data.

We utilized an aggregated dataset to determine the average sizes of the individuating information effects in Navon and Bar-Anan ([27] Studies 1–6). In analyses original to this review, the average effect of individuating information was β = .53 on the explicit measure of attitudes toward the targets, β = .12 on IAT scores in Studies 1–3, and β = .05 in the three EPT studies. Thus, the average effect size for individuating information effects on the explicit measure was large, whereas average effect sizes were small on the implicit measures. This suggests that individuating information had larger effects on explicit person perception than on implicit person perception.

In sum, we compared individuating information effect sizes obtained from implicit and explicit measures in response to the same individuating information manipulations (or approximations thereof) and found that in almost all cases, the effect size was larger for the explicit measure than it was for the implicit measure. The difference between findings regarding the size of individuating information effects for implicit and explicit person perception is broadly consistent with the Associative Propositional Evaluations model (APE model; [28,29]), which proposes that implicit and explicit evaluations (i.e., implicit and explicit attitudes) are governed by separate but interacting processes.

Limitations and Unaddressed Questions

One must be careful in the inferences that one draws from the data presented above because the two types of measures may differ in their sensitivity to the constructs they are intended to measure. Different effect sizes may occur because of truly different effects, but a nearly infinite array of artifacts can also produce different effect sizes, even if the underlying effects are similar. Many artifacts could, at least hypothetically, lead to weaker effects on implicit than explicit measures. Implicit measures may have more random noise than explicit measures (more measurement error; e.g., [30]). They also could be more influenced by factors irrelevant to effects of either social category information or individuating information (e.g., ability to determine a correct response, ability to overcome bias, and guessing bias [31]).

Of course, explicit measures are not immune to artifactual influences either. The artifactual contender most capable of explaining larger effects of individuating information on explicit than implicit person perception may be demand characteristics. If participants determine that there is a “correct” answer based on the large amount of individuating information provided that suggests that a certain trait characterizes an individual, experiments may yield large effects of individuating information that do not necessarily reflect participants’ personal judgments, but instead, their desire to answer “correctly” or to appear unbiased. However, it is so logically reasonable to reach an inference that, for example, a target is “intelligent” if he has a high GPA and high SAT scores, that it is often difficult to clearly distinguish between demand characteristics and actual inferences without additional procedures further complexifying conducting such research (and which may have their own unintended consequences that themselves can threaten the validity of the inferences drawn).

Furthermore, another possibility is that response times might be generally more difficult to change than self-reported evaluations of individuals. If this is the case, effect sizes of individuating information on response time measures may be smaller than their effects on self-reported evaluations of individuals. To clarify this, consider the example of driving a car. It is important for people to learn how to drive well, but “driving well” means many things, such as keeping within speed limits, following other cars at a safe distance, and reacting quickly to another driver who does something dangerous nearby (such as cutting one off on a highway). Only the last skill (reacting quickly to being cut off) involves reaction times. If it is easier to learn how to follow other drivers at a safe distance than to reduce one’s reaction time to being cut off, effects of driver training programs may be larger on driving at a safe distance than on reaction times.

If, in this situation, both following at a safe distance and reaction times are interpreted not as outcomes in their own rights, but as “indicators” of some underlying construct (e.g., “good driving”), then driver training programs would produce seemingly different effects on “good driving” depending on whether one assessed driving at a safe distance or reaction times as the main outcome. This is similar enough to how both self-reported evaluations and IAT and EPT scores are interpreted–-not as outcomes in their own rights but as measures of underlying psychological explicit versus implicit judgments. If we temporarily eschew interpreting such studies through the lens of global psychological inferences about constructs such as implicit or explicit (or automatic vs. controlled, etc.) judgments when describing these results and, instead, stick very closely to the studies and their outcome measures, then a narrow but clearly justifiable conclusion is this: the effects of the individuating information that have been examined in studies to date have had far larger effects on measures assessing self-reported judgments of individuals than on the response times to IAT and EPT tasks. This interpretation is consistent with a more global psychological inference that individuating information has larger effects on explicit than on implicit person perception. However, because many alternative explanations for this pattern have not been decisively ruled out, there is far more uncertainty around the validity of this more global conclusion than there is around a narrow conclusion focusing specifically on the effects of individuating information on self-report versus reaction time measures.

Nonetheless, even with these many methodological caveats, the differences found in the many studies reviewed regarding effects of individuating information on explicit versus implicit person perception have been quite large. Our judgment—and one that should be addressed in future research—is that they are sufficiently large to render it unlikely, though not impossible, that they can be explained away primarily through artifactual explanations. The possibility that artifacts can account for the differences does not constitute empirical evidence that they actually do account for those differences. We believe that the evidence reviewed herein means that the burden of proof has shifted; the findings of much larger effects of individuating information on explicit than implicit person perception (or, at least, measures of explicit and implicit person perception) have been sufficiently consistently found that it is at least tentatively reasonable to conclude that this is a common pattern. To justify contrary claims now requires either empirical evidence demonstrating stronger or equal effects on implicit (compared with explicit) person perception or empirical evidence demonstrating that artifacts can explain away the differences that have been so consistently found by different research teams using different methods.

Given that there is a wide array of potential artifactual accounts for the differential effects of individuating information, one priority for future research should be to assess them directly. Does controlling for social desirability (e.g., [32]) reduce effect sizes for individuating information? Does use of experimental procedures (such as the bogus pipeline [33]) reduce effects of individuating information? The Quad Model parses IAT scores into four parameters, three of which reflect artifacts and only one of which reflects automatic activation of concepts [31]. Would an application of the Quad Model find larger or smaller effects of individuating information on the automatic activation parameter? This can be tested directly in future research.

In addition, in Navon et al.’s [11] studies, it is clear that pro-dominant implicit bias in attitudes towards individual targets was reduced in the pro-stigmatized condition. The results do suggest that this was accomplished by individuating information. However, individuating information was neither (a) manipulated nor (b) identified observationally as a variable that could be used to predict implicit attitudes. Thus, determining the causal role of attitude-relevant individuating information in the results requires future research.

Moreover, research has not yet tested the magnitude of individuating information effects on implicit person perception for target groups other than race, gender, and age groups. Future research can address this question using targets belonging to other types of social categories (e.g., groups belonging to the LGBTQ+ community, weight-based groups) and targets belonging to other racial or ethnic groups.

In addition, this question (and all questions discussed in this review) has exclusively been examined using the IAT and the EPT as implicit measures. The validity of both measures is subject to debate (e.g., [12]). With regard to the EPT, there are a series of constructs other than evaluation that influence scores on the EPT (see [34], for a review), including (a) similarity effects due to more similarity between positive words than between negative words (which makes response times to positive words faster than those to negative words); (b) integrative priming effects (i.e., evaluative priming effects are present when—and only when—primes and targets can be compounded together in a meaningful way); and (c) congruity proportion effects (i.e., the greater cognitive facility of congruent trials is reduced, eliminated, or reversed when there are more incongruent trials than congruent trials). With regard to the IAT, (a) similarity effects also are present (there are stronger effects on the IAT when participants categorize positive rather than negative words), (b) the processes of response detection and overcoming bias also affect IAT scores, and (c) IAT effects are able to be strategically generated in the absence of an underlying attitude (see [34], for a review). Thus, future research should evaluate the generality of the main patterns reported in this section using implicit measures beyond the IAT and EPT.

9. Key Thematic Point 4: Social Category Information Causes Biases in Implicit Judgments of Individuals More than It Causes Parallel Biases in Explicit Judgments

Regardless of whether the dependent measure is stereotype- or attitude-based biases in implicit and explicit judgments of individuals, social category information generally influences implicit measures more than it influences explicit measures. In an analysis original to this review, we calculated the average effect size for D scores from 14 studies in which counterstereotypic individuating information was provided (aggregated data are available in Supplemental Materials). In those studies, D scores reflecting stereotype-based biases in implicit person perception that were measured after the presentation of that information were compared to D scores in the absence of individuating information ([16], Studies 1–4, [17], Studies 1–3, [18], Studies 1, 2a, & 2b, [20,21], Studies 1–4). Results from conditions in which counterstereotypic individuating information was presented revealed that D scores (an index of the social category information effect in the presence of such information) showed an average effect size of d = 0.08 [95% CI: 0.04, 0.13]. Results from conditions in which only social category information was provided revealed that D scores showed an average effect size of d = 0.71 [95% CI: 0.66, 0.76].

In these same 14 studies, the 25 effect sizes for social category information on explicit measures in the presence of counterstereotypic individuating information ranged from d = −2.39 to d = −0.63. It should be noted that although the absolute magnitude of the effect sizes generally showed large effects (88% of |ds| > 0.80), these effect sizes were negative, meaning that the explicit stereotype-based biases in implicit judgments of individuals had reversed in direction due to the counterstereotypic individuating information and aligned with that information. The lowest bound of any of the 95% CIs was −2.88 and the highest was −0.44. Thus, none of the 95% CIs overlapped with the overall one for the implicit measure in the presence of counterstereotypic individuating information, suggesting that (a) even in the presence of the same counterstereotypic information that reversed the direction of biases in explicit person perception, stereotype-based bias in implicit person perception remained slightly consistent with the initial stereotype across studies, and (b) the effect sizes for explicit and implicit measures significantly differed from one another.

Also in these same 14 studies, the 25 effect sizes for social category information on explicit measures in the absence of individuating information ranged from d = −0.13 to d = 1.85. Despite this wide range, 76% of these effect sizes were small. The 95% CIs for this 76% of the effect sizes did not overlap with the overall one for the implicit measure in the absence of individuating information; the lowest bound of these was −0.40 and the highest bound was 0.53. Of the remaining medium or large effect sizes whose 95% CIs did overlap with that for the implicit measure (lowest bound 0.45, highest bound 2.05), it is worth noting that all of these stereotypes were either observable stereotypes (i.e., those that can be learned with little to no inference on the part of perceivers—e.g., women wear dresses to dress up while men wear tuxedos—[17]) or stereotypes of gay men, that all such stereotypes were medium to large in magnitude, and that there were at least two medium or large effect sizes belonging to each of these two types of stereotypes. Given that the large majority of effect sizes were smaller for explicit measures than for implicit measures and that those that were not smaller were limited to certain types of stereotypes, we conclude that generally, the evidence from social category effects in the absence of counterstereotypic information supports the same conclusions as that obtained in the presence of counterstereotypic individuating information.

Moreover, in Navon et al.’s [11] analyses, across the first 17 studies, the average social category information effect on the IAT was d = −0.01, with an average effect of social category information on explicit measures of d = −0.63. Across two studies utilizing the EPT (Studies 18 and 19), effect sizes for the effect of social category information on explicit measures were d = −0.74 and d = −0.23, reflecting medium-to-strong or weak preference (respectively) for the Black target. In contrast, the effects of social category information on the EPT were d = 0.18 and d = 0.22, interpreted as reflecting weak preference for the White target. In two additional studies utilizing the IAT ([11], Studies 26 and 27), effect sizes of social category effects on explicit measures were d = −0.58 and d = −0.07, reflecting moderate preference for the Black target and no preference for the Black vs. the White target, respectively. IAT effect sizes for social category effects were d = 0.92 and d= 0.82, interpreted as reflecting strong preference for the White target. Thus, taken together, these data indicate that on explicit measures, participants tended to favor the member of the stigmatized group, whereas on implicit measures, they favored the member of the dominant group or showed no preference between the dominant and stigmatized group. This reinforces the conclusion that social category information has a larger influence on implicit measures than on explicit measures since biases against members of stigmatized groups were mostly reversed in direction on the explicit measures and thus aligned with the direction of the individuating information, whereas this was not the case for implicit measures.

Finally, in additional analyses novel to this review, Navon and Bar-Anan’s [27] six studies had an average social category information effect of β = −0.22 on explicit measures. For Studies 1–3 (which utilized the IAT), there was an average effect of social category information of β = 0.37 on implicit measures. For Studies 4–6 (which utilized the EPT), there was an average effect of social category information of β = 0.12 on implicit measures. This indicates that social category effects on explicit measures reversed in direction relative to what would be expected based on societal prejudices. Specifically, the positively portrayed member of the group that would be expected to be disfavored was evaluated more positively than the positively portrayed member of the group that would be expected to be favored to a small extent. On the other hand, social category effects on implicit measures favored the group that would be expected to be favored according to societal biases; the positively portrayed member of the group that would be expected to be favored was preferred over the positively portrayed member of the group that would be expected to be disfavored to a small-to-medium extent.

Unaddressed Questions

The phenomenon discussed above has presented itself using data from many studies, using a variety of target groups, and across both attitude- and stereotype-based biases in implicit person perception. Nonetheless, no studies have yet attempted to identify the processes underlying the discrepancy between social category effects on implicit vs. explicit measures, including but not restricted to ruling out the types of potential artifactual challenges to both implicit and explicit measures that we reviewed previously. Future research should address these issues.

10. Key Thematic Point 5: The Ability of Other Variables to Moderate Individuating Information Effects on Stereotype- and Attitude-Based Biases in Implicit Person Perception Varies

10.1. Successful Moderating Variables

Three variables have been found to moderate individuating information effects on stereotype- and attitude-based biases in implicit person perception: (a) the diagnosticity of the individuating information ([23] Study 1), (b) the observability of the stereotype (i.e., the extent to which perceivers must make inferences when learning the stereotype [16]), and (c) the presence vs. absence of visual cues to social category membership [11].

With regard to the diagnosticity of individuating information, one study ([23], Study 1) varied whether perceivers received no individuating information (i.e., only social category information), somewhat diagnostic individuating information, or highly diagnostic individuating information. The information comprised college applications of varying degrees of completion. Stereotype-based bias in implicit judgments of the individuals (here, the belief that White individuals are intelligent while Black individuals are unintelligent) linearly decreased as the diagnosticity of the individuating information increased ([23] Study 1). Thus, the diagnosticity of the individuating information moderated the biasing effects of stereotypes on implicit person perception (cf. [21], in which diagnosticity of individuating information did not moderate its effects on stereotype-based biases in implicit person perception).

In addition, stereotypes about observable characteristics (hence observable stereotypes; i.e., those that require minimal to no inference on the part of perceivers to be learned, e.g., behavior by group members) can be distinguished from stereotypes about unobservable characteristics (hence unobservable stereotypes; i.e., those that cannot be observed directly and involve some inference on the part of perceivers, e.g., personality trait stereotypes [16]). Highly diagnostic counterstereotypic individuating information shifted unobservable stereotype-based biases in implicit person perception in the direction of the individuating information more than they shifted observable stereotype-based biases in implicit person perception in that same direction [16]. Specifically, in the presence of counterstereotypic vignettes, stereotype-based biases in implicit judgments of the individuals about the unobservable characteristics that (a) White individuals are book-smart while Hispanic individuals are street-smart and (b) women are empathetic while men are assertive changed more in the direction of the individuating information than did stereotype-based biases in implicit judgments of the individuals about the observable characteristics that (a) White individuals are English-speaking while Hispanic individuals are Spanish-speaking, and (b) to dress up, women wear dresses while men wear tuxedos. Thus, the observability of the stereotype moderated individuating information effects on stereotype-based biases in implicit person perception.

In addition, two studies ([11] Studies 24 and 25) using the previously discussed paradigm, in which participants implicitly evaluated celebrity Black targets that participants had rated more favorably than celebrity White targets on self-report measures, manipulated whether names or photos of the celebrity targets were presented to participants. When only names were provided, D scores were nonsignificant, showing no implicit preference for either the Black or the White targets, whereas when photos were present, implicit attitude-based biases favored the White over the Black target. Thus, the presence of visual social category membership cues moderated the effects of individuating information on attitude-based biases in implicit person perception.

Finally, while research has not directly assessed this question, there is some evidence that gender stereotype-based biases in implicit person perception may be less malleable than race or ethnic stereotype-based biases in implicit person perception. The stereotype-based biases in implicit person perception that have shifted to correspond with the direction of the individuating information presented have mostly been related to race and none are about gender groups (these are listed earlier in this review; [16] Study 2; [17] Studies 1 & 2; [19,21] Studies 1 & 3). By contrast, when considering the stereotype-based biases in implicit person perception that have been reduced by counterstereotypic individuating information yet which still did not correspond with the direction of the individuating information, the majority of these are related to gender stereotypes ([16] Studies 3 & 4; [21] Studies 2 & 4; [22]).

However, this discrepancy in patterns of shifts in biases may have occurred because the gender stereotype-based biases in implicit person perception studied were oftentimes initially stronger than the race-based biases. Thus, the gender stereotype-based biases may have been more firmly entrenched. To assess whether gender stereotype-based biases in implicit person perception are generally less malleable than race stereotype-based biases in implicit person perception requires research that holds constant stereotype strength. Such research should then systematically compare the extent to which counterstereotypic individuating information reduces gender and race stereotype-based biases in implicit person perception. This is an important avenue for future research.

10.2. Non-Moderating Variables

Three variables have been found not to moderate individuating information effects on stereotype-based biases in implicit person perception. These are the mode of presentation of the individuating information [19], individual difference variables [21], and controllability of group membership [18].

As noted above, Rubinstein and Jussim [19] varied whether individuating information was presented in the form of vignettes or photo pairings. This manipulation was intended to compare propositional (i.e., rule-based) vs. associative cognitive processes (e.g., [28,29]). This research found that, regardless of whether D scores represented individuating information effects on implicit person perception or individuating information effects on stereotype-based biases in implicit person perception, the mode of presentation did not moderate the obtained effects. All effects were similar in magnitude, regardless of mode of presentation.

In addition, research has tested whether individual differences moderate individuating information effects on stereotype-based biases in implicit person perception [21]. Specifically, this research tested the hypotheses that counterstereotypic individuating information would be less effective at reducing stereotype-based biases in implicit person perception among those who (a) are low in Need for Cognition (the extent to which individuals engage in and enjoy thinking [35]), (b) are low in Openness (the extent to which an individual wants to expand their experiences and consciousness [36]), (c) are high in Need for Closure (the extent to which an individual dislikes ambiguity and seeks definitive answers [37]), (d) are high on Dogmatism (the extent to which an individual’s belief system is closed and centers around a sense of authority [38]), (e) are high in explicit prejudice toward the stigmatized target group, and (f) hold entity theories of personality (i.e., the belief that human characteristics are fixed; see [39] for a review). The rationale for these predictions was that since these characteristics all predict stronger outgroup biases (e.g., [40,41,42,43,44]), perhaps they would also decrease the effectiveness of individuating information at reducing stereotype-based biases in implicit judgments of individuals.

The studies tested whether these personality variables moderated the effects of counterstereotypic individuating information on the following stereotype-based biases in implicit person perception: the stereotype that White individuals are book-smart while Hispanic individuals are street-smart (Studies 1 and 3) and the stereotype that women are empathetic while men are assertive (Studies 2 and 4). Results showed no consistent evidence of moderating effects; individuating information effects were similar across individual differences.

Finally, a series of experiments tested whether manipulating the controllability vs. uncontrollability of group membership (i.e., whether the group membership is under the individual’s control or not [45]) moderates the effects of individuating information on stereotype- and attitude-based biases in implicit person perception [18]. Controllability is a continuum, with some group memberships being far easier to change than others. For example, political party membership and religion can be changed, while ethnicity cannot. Biases can sometimes be stronger when group membership is portrayed as controllable as opposed to uncontrollable (e.g., [46]). As previously discussed, individuating information should be less effective at shifting stronger (vs. weaker) biases. Therefore, it was hypothesized that individuating information effects on biases in implicit person perception would be smaller when group membership was portrayed as controllable compared with when it was portrayed as uncontrollable or compared with when no controllability information was provided.

The research tested these hypotheses in the context of both attitude- and stereotype-based implicit judgments of individuals. The stereotype-based biases tested in the studies that did not meet exclusion criteria for the present review were the following: (a) individuals with higher weights are lazy while those with lower weights are motivated, and (b) individuals who practice Islam are aggressive while individuals who practice Buddhism are peaceful. Across three studies, controllability of group membership did not moderate the effects of individuating information on attitude- or stereotype-based biases in implicit person perception. Individuating information effects were similar, regardless of the controllability of the group membership.

10.3. Unaddressed Questions

Past theories of reliance on individuating information and stereotypes in explicit person perception have discussed motivating factors in the extent to which perceivers individuate vs. rely on social category information in such perceptions [5]. One motivation proposed to increase reliance on individuating information in explicit person perception is task-oriented outcome dependency on the target. When perceivers have an interdependent dynamic with a target, it benefits them to be able to predict the target’s behavior to improve their odds of a successful outcome [47,48]. Thus, they are more likely to attend to and rely on individuating information in their social judgments. Empirical evidence has supported this prediction in the domain of explicit person perception (e.g., [49]). Thus, task-oriented outcome dependency may strengthen individuating information effects on biases in implicit person perception, as well. Future research should test this proposition.

11. Implications

11.1. Theoretical Implications

The present synthesis of existing research provides several contributions to the literature. This review provides the first summary comparison of the malleability of stereotype- versus attitude-based biases in implicit person perception. Pending additional research to address the many limitations identified throughout this review, it can be tentatively concluded that stereotype-based biases in implicit person perception are more malleable than are attitude-based biases in implicit person perception in the presence of individuating information. The relatively greater responsiveness of stereotype-based biases in implicit person perception to individuating information than attitude-based biases in implicit person perception to such information suggests that stereotype-based biases in implicit person perception may be more subject to fast-learning processes, while attitude-based biases in implicit person perception may be more subject to slow-learning processes. Thus, these two phenomena lend support to different theories. The aggregate results from studies in which stereotype-based biases in implicit person perception were the dependent measure support propositional models of evaluations (e.g., [50]) and the APE model [28,29]. Both models predict that implicit judgments should be responsive to counterinformation in the evaluative circumstances posed by the research reviewed above. On the other hand, the consistent finding that attitude-based biases in implicit person perception are mostly impervious to individuating information supports the Systems of Evaluations Model (SEM; [51,52]), which posits that implicit judgments are slow to update in the face of new information.

Another contribution of the present review has been to aggregate data from multiple programs of research that have addressed reliance on individuating information and social category information in implicit vs. explicit person perception. Doing so permits reaching more general conclusions about the most common patterns found in this area of research. Utilizing data from studies in which implicit and explicit measures were administered after the same manipulations, we found that individuating information effects on person perception were larger on explicit measures than on implicit measures. By contrast, social category information effects were larger on implicit measures than on explicit measures. These findings suggest that, despite the fact that implicit person perception is oftentimes sensitive to individuating information, implicit person perception and social category biases in implicit person perception are less responsive to individuating information than are explicit person perception and social category-based biases in explicit person perception. This is broadly consistent with the APE model [28,29], which proposes that implicit and explicit evaluations are separate yet interacting processes.

11.2. Practical Implications

Perceivers must make a plethora of consequential judgments when they have at hand both social category information and individuating information. Beyond the hypothetical hiring scenario presented in the Introduction, imagine a situation in which a Black individual applies for a mortgage. The individual’s race may activate anti-Black prejudice and societal stereotypes of Black Americans as poor (e.g., [53]). Thus, if the lender were to rely only on this social category information, they might be inclined to deny the individual’s application. However, this is not the only information that the lender has at hand; the lender also has access to the individual’s credit score, income, and other financial information that would ideally help them arrive at a more informed conclusion. However, will they actually rely on this information, or will they allow pernicious stereotypes and prejudice to continue to bias their judgments and continue this element of structural racism (denial of Black Americans’ mortgage applications, an element of redlining, e.g., [54])? The research reviewed above may be relevant to this societally impactful question. However, none of the studies reviewed herein have addressed the extent to which such judgments inform people’s actual decisions. Thus, another important direction for future research is to perform a systematic assessment of implicit and explicit reliance on social categories versus individuating information under naturally occurring conditions.

Imagine another situation in which a transgender individual visits a doctor hoping to obtain support for gender-affirming care. This doctor may be prejudiced against transpeople and also hold the societal stereotype that transgender individuals are confused [55,56] and thus be tempted not to provide this support. However, if this particular transgender individual has wanted to transition since early childhood and has no history of mental health problems, this may serve as counterstereotypic individuating information. Will the doctor apply their stereotypes and prejudices to the individual, or will they rely on the individuating information? The research reviewed above speaks to this question, which is relevant to the facts that (a) in 2024 alone, 701 legislative bills denying transpeople of their basic rights were proposed in the U.S., spread across 43 states [57], and (b) the United States has recently elected a President who ran on an anti-transgender platform and has thus seen a slew of anti-transgender Executive Orders issued by the President. However, as was the case with the mortgage application scenario, although implicit judgments are relevant to the decision, they are not the only factor determining the perceiver’s decision.

The three applied scenarios that have been outlined in the Introduction and Implications sections of this review are only a few decisions that must be made when perceivers have at hand social category information and individuating information. These decisions may be partially influenced by attitude- and stereotype-based biases in perceivers’ implicit judgments of the individuals, and these decisions have serious consequences for the overall life trajectory of the target of the social judgment. The research discussed in the present review may be relevant to countless other similar decisions, though we cannot know to what extent, if at all, until such research is actually conducted.

12. Conclusions

The present review synthesizes a modest but growing literature on the effects of social category information and individuating information on implicit judgments of individual members of known social groups. The aggregate conclusions that can be drawn from this body of research have both theoretical and practical implications. Findings in the domain of stereotype-based biases suggest reason to be hopeful that bias in implicit judgments of individual members of known social groups can be shifted by individuating information. Findings in the domain of attitude-based biases against individuals suggest that individuating information has limited effectiveness at shifting biases in implicit judgments of individual members of known social groups, and that further research is needed to identify effective means of shifting such biases.