When the method makes a difference: Antagonistic effects on “automatic evaluations” as a function of task characteristics of the measure

Twenty-Five Years of Research Using Implicit Measures

The year 2020 marks the 25th anniversary of two seminal publications that have set the foundation for an exponentially growing body of research using implicit measures: Fazio, Jackson, Dunton, and Williams's (1995) work using evaluative priming to measure racial attitudes, and Greenwald and Banaji's (1995) review of implicit social cognition research that served as the basis for the development of the Implicit Association Test (IAT). The current article provides an overview of (1) two conceptual roots that continue to shape interpretations of implicit measures; (2) conflicting interpretations of the term implicit; (3) different kinds of dissociations between implicit and explicit measures; (4) theoretical developments inspired by these dissociations; and (5) research that used implicit measures to address domain-specific and applied questions. We conclude with a discussion of challenges and open questions that remain to be addressed, offering guidance for the next generation of research using implicit measures.

Automatic Antecedents of Discrimination

Abstract. In increasingly diverse societies, discrimination against social groups and their members continues to be a public and political concern. Research has addressed three basic cognitive processes that precede discrimination: categorization, stereotype/prejudice activation, and stereotype/prejudice application, suggesting that these processes occur in an automatic fashion. However, there are multiple components of automaticity, including unawareness, efficiency, unintentionality, and uncontrollability. Most of the previous research implies that these components of automaticity converge with respect to cognitive antecedents of discrimination. Here, we review evidence on the distinct components of automaticity in order to assess whether (a) categorization, (b) stereotype/prejudice activation, and (c) stereotype/prejudice application occur (1) without awareness, (2) efficiently, and (3) goal-independently. We highlight evidence indicating convergence or divergence of the automaticity components during each of the processing stages. This analysis provides readers with an up-to-date review that helps to evaluate whether a multi-component approach to automaticity is of additional benefit in aggregating knowledge about the cognitive antecedents of discrimination. We discuss open issues and avenues for future research.

Six Lessons for a Cogent Science of Implicit Bias and Its Criticism

Skepticism about the explanatory value of implicit bias in understanding social discrimination has grown considerably. The current article argues that both the dominant narrative about implicit bias as well as extant criticism are based on a selective focus on particular findings that fails to consider the broader literature on attitudes and implicit measures. To provide a basis to move forward, the current article discusses six lessons for a cogent science of implicit bias: (a) There is no evidence that people are unaware of the mental contents underlying their implicit biases; (b) conceptual correspondence is essential for interpretations of dissociations between implicit and explicit bias; (c) there is no basis to expect strong unconditional relations between implicit bias and behavior; (d) implicit bias is less (not more) stable over time than explicit bias; (e) context matters fundamentally for the outcomes obtained with implicit-bias measures; and (f) implicit measurement scores do not provide process-pure reflections of bias. The six lessons provide guidance for research that aims to provide more compelling evidence for the properties of implicit bias. At the same time, they suggest that extant criticism does not justify the conclusion that implicit bias is irrelevant for the understanding of social discrimination.

Taking Priming to Task

Abstract. The current research examined potential moderators of gender and racial stereotype priming in sequential priming paradigms. Results from five experiments suggest that stereotype priming effects are more consistent in tasks that elicit both semantic priming and response competition (i.e., response priming paradigms) rather than tasks that evoke semantic priming alone (i.e., semantic priming paradigms). Recommendations for future stereotype priming research and the implication of these results for the proper interpretation of stereotype priming effects are discussed.

At the Boundaries of Misattribution

Priming effects in the Affect Misattribution Procedure (AMP) have been explained by a misattribution of prime-related affect to neutral targets. However, the measure has been criticized for being susceptible to intentional use of prime-features in judgments of the targets. To isolate the contribution of unintentional processes, the present research expanded on the finding that positive affect can be misattributed to familiarity (i.e., positivity-familiarity effect). To the extent that prime-valence is deemed irrelevant for judgments of target-familiarity, positivity-familiarity effects in the AMP could potentially rule out intentional use of the primes. Seven experiments collectively suggest that prime-valence influences judgments of target-familiarity in the AMP, but only when the task context does not suggest a normatively accurate response to the familiarity-judgment task. Relations of positivity-familiarity effects to self-reported use of prime-valence revealed mixed results regarding the role of intentional processes. Implications for the AMP and misattribution effects are discussed.

Threat in the Company of Men: Ensemble Perception and Threat Evaluations of Groups Varying in Sex Ratio

Everyday, we visually perceive people not only in isolation but also in groups. Yet, visual person perception research typically focuses on inferences made about isolated individuals. By integrating social vision and visual ensemble coding, we present novel evidence that (a) perceivers rapidly (500 ms) extract a group’s ratio of men to women and (b) both explicit judgments of threat and indirect evaluative priming of threat increase as the ratio of men to women in a group increases. Furthermore, participants’ estimates of the number of men, and not perceived men’s coalition, mediate the relationship between the ratio of men to women and threat judgments. These findings demonstrate the remarkable efficiency of perceiving a group’s sex ratio and downstream evaluative inferences made from these percepts. Overall, this work advances person perception research into the novel domain of people perception, revealing how the visually perceived sex ratio of groups impacts social judgments.

Effects of Chocolate Deprivation on Implicit and Explicit Evaluation of Chocolate in High and Low Trait Chocolate Cravers

Diet failures are often attributed to an increase in cravings for attractive foods. However, accumulating evidence shows that food cravings actually decrease during energy-restricting weight-loss interventions. The current study aimed at elucidating possible mechanisms that may explain how and under which circumstances food cravings in- or decrease during dieting. Specifically, decreases in food cravings during weight-loss diets may be due to effects of energy restriction (homeostatic changes) and to effects of avoiding specific foods (hedonic changes). Thus, we used a selective, hedonic deprivation (i.e., restricting intake of a specific food in the absence of an energy deficit) that precludes homeostatic changes due to energy restriction. Furthermore, interindividual differences in food craving experiences might affect why some individuals are more prone to experience cravings during dieting than others. Thus, we investigated whether a selective deprivation of chocolate would in- or decrease craving and implicit preference for chocolate as a function of trait-level differences in chocolate craving. Participants with high and low trait chocolate craving (HC, LC) refrained from consuming chocolate for 2 weeks but otherwise maintained their usual food intake. Both groups underwent laboratory assessments before and after deprivation, each including explicit (i.e., state chocolate craving) and implicit measures (i.e., Single Category Implicit Association Test, SC-IAT; Affect Misattribution Procedure, AMP). Results showed that hedonic deprivation increased state chocolate craving in HCs only. HCs also showed more positive implicit attitudes toward chocolate than LCs on the SC-IAT and the AMP irrespective of deprivation. Results help to disambiguate previous studies on the effects of dieting on food cravings. Specifically, while previous studies showed that energy-restricting diets appear to decrease food cravings, the current study showed that a selective, hedonic deprivation in the absence of an energy deficit increases food cravings. However, this effect can only be observed for individuals with high trait craving levels. Thus, if attractive foods are strictly avoided through a selective deprivation, HCs are at risk to experience craving bouts in the absence of an energy deficit. As implicit preference was unaffected by chocolate deprivation, strong implicit preference for chocolate likely characterize a stable mechanism that drives consumption in HCs.

Propositional Versus Dual-Process Accounts of Evaluative Conditioning

Evaluative conditioning (EC) is defined as the change in the evaluation of a conditioned stimulus (CS) due to its pairing with a valenced unconditioned stimulus (US). Expanding on the debate between dual-process and propositional accounts, two studies investigated the relative effectiveness of counter-conditioning and counter-instructions in reversing EC effects on implicit and explicit evaluations. After conditioned evaluations were acquired via CS-US pairings, participants were either (1) presented with repeated CS-US pairings of the opposite valence or (2) given instructions that the CSs will be paired with USs of the opposite valence. Although both procedures reversed previously conditioned explicit evaluations, only directly experienced CS-US pairings reversed previously conditioned implicit evaluations. The findings question the functional equivalence of counter-conditioning and counter-instructions hypothesized by single-process propositional accounts. Yet, they support dual-process accounts, suggesting that associative and propositional processes jointly contribute to EC effects.

Effects in the Affect Misattribution Procedure Are Modulated by Feature-Specific Attention Allocation

Abstract. We examined whether automatic stimulus evaluation as measured by the Affect Misattribution Procedure (AMP) is moderated by the degree to which attention is assigned to the evaluative stimulus dimension (i.e., feature-specific attention allocation, FSAA). In two experiments, one group of participants completed a standard AMP while attending to evaluative stimulus information. A second group of participants completed the AMP while attending to non-evaluative stimulus information. In line with earlier work, larger AMP effects were observed when participants were encouraged to attend to evaluative stimulus information than when they were not. These observations support the idea that the impact of FSAA on measures of automatic stimulus evaluation results from a genuine change in the degree of automatic stimulus evaluation rather than a change in the degree to which automatic stimulus evaluation is picked up by these measures.

On the relationship between the indirectly measured attitude towards beer and beer consumption: the role of attitude accessibility

Although some studies have demonstrated that the indirectly measured attitude towards alcohol is related to alcohol use, this relationship has not always been confirmed. In the current study, we attempted to shed light on this issue by investigating whether the predictive validity of an indirect attitude measure is dependent upon attitude accessibility. In a sample of 88 students, the picture-picture naming task, an adaptation of the affective priming paradigm, was used to measure the automatically activated attitude towards beer. Attitude accessibility was measured using a speeded evaluative categorization task. Behavioral measures were the amount of beer poured and drunk during a bogus taste test and the choice between a bottle of beer or water at the end of the experiment. In line with our hypothesis, the indirectly measured attitude towards beer predicted behavior during the taste test only when it was highly accessible. In contrast, this attitude was related to choice behavior irrespective of attitude accessibility. This study confirms that indirect attitude measures can be valuable predictors of alcohol-related behavior, but that it is sometimes necessary to take attitude accessibility into account.

Can the Formation of Conditioned Attitudes Be Intentionally Controlled?

Evaluative conditioning (EC) is defined as the change in the evaluation of a conditioned stimulus (CS) due to its pairing with a valenced unconditioned stimulus (US). Counter to views that EC is the product of automatic learning processes, recent research has revealed various characteristics of nonautomatic processing in EC. The current research investigated whether the formation of conditioned attitudes can be intentionally controlled. Whereas EC effects on self-reported evaluations were reduced (enhanced) when participants were instructed to prevent (promote) the influence of CS-US pairings, EC effects on an evaluative priming measure remained unaffected by control instructions. Moreover, although EC effects on self-reported evaluations varied as a function of evaluative priming effects and recollective memory for CS-US pairings, motivation to control the influence of CS-US pairings qualified only the predictive relation of recollective memory. The results highlight functionally distinct contributions of uncontrollable encoding-related processes and controllable expression-related processes to EC effects.

What Drives Priming Effects in the Affect Misattribution Procedure?

The affect misattribution procedure (AMP) is one of the most promising implicit measures to date, showing high reliability and large effect sizes. The current research tested three potential sources of priming effects in the AMP: affective feelings, semantic concepts, and prepotent motor responses. Ruling out prepotent motor responses as a driving force, priming effects on evaluative and semantic target responses occurred regardless of whether the key assignment in the task was fixed or random. Moreover, priming effects emerged for affect-eliciting primes in the absence of semantic knowledge about the primes. Finally, priming effects were independent of the order in which primes and targets were presented, suggesting that AMP effects are driven by misattribution rather than biased perceptions of the targets. Taken together, these results support accounts that attribute priming effects in the AMP to a general misattribution mechanism that can operate on either affective feelings or semantic concepts.

Intention invention and the affect misattribution procedure: reply to Bar-Anan and Nosek (2012)

A recent study of the affect misattribution procedure (AMP) found that participants who retrospectively reported that they intentionally rated the primes showed larger effect sizes and higher reliability. The study concluded that the AMP's validity depends on intentionally rating the primes. We evaluated this conclusion in three experiments. First, larger effect sizes and higher reliability were associated with (incoherent) retrospective reports of both (a) intentionally rating the primes and (b) being unintentionally influenced by the primes. A second experiment manipulated intentions to rate the primes versus targets and found that this manipulation produced systematically different effects. Experiment 3 found that giving participants an option to "pass" when they felt they were influenced by primes did not reduce priming. Experimental manipulations, rather than retrospective self-reports, suggested that participants make post hoc confabulations to explain their responses. There was no evidence that validity in the AMP depends on intentionally rating primes.

An Inkblot for the Implicit Assessment of Personality: The Semantic Misattribution Procedure

Misattributions people make about their own reaction to ambiguous stimuli can be used to measure personality self–concepts implicitly. On the basis of a semantic misattribution priming paradigm [semantic misattribution procedure (SMP)], we assessed the implicit personality self–concept related to three dimensions included in the Big–Five model: conscientiousness, neuroticism, and extraversion. Across three studies (N1 = 98, N2 = 140, and N3 = 135), the SMP was robustly related, in the expected direction, to individual differences in self–reported personality questionnaires and managed to predict both self–reported and objectively measured behaviours. The main advantage of SMP over classical explicit measures of personality is its higher resistance to social desirability tendencies, although its psychometric properties are somewhat lower than those pertaining to explicit measures of personality. Finally, comparisons of our results with studies that used other implicit measures of personality self–concept indicate that the SMP has higher criterion validity. Copyright © 2012 John Wiley & Sons, Ltd.

No fear, no panic: probing negation as a means for emotion regulation

This electroencephalographic study investigated if negating one's emotion results in paradoxical effects or leads to effective emotional downregulation. Healthy participants were asked to downregulate their emotions to happy and fearful faces by using negated emotional cue words (e.g., no fun, no fear). Cue words were congruent with the emotion depicted in the face and presented prior to each face. Stimuli were presented in blocks of happy and fearful faces. Blocks of passive stimulus viewing served as control condition. Active regulation reduced amplitudes of early event-related brain potentials (early posterior negativity, but not N170) and the late positive potential for fearful faces. A fronto-central negativity peaking at about 250 ms after target face onset showed larger amplitude modulations during downregulation of fearful and happy faces. Behaviorally, negating was more associated with reappraisal than with suppression. Our results suggest that in an emotional context, negation processing could be quite effective for emotional downregulation but that its effects depend on the type of the negated emotion (pleasant vs unpleasant). Results are discussed in the context of dual process models of cognition and emotion regulation.

The Role of Self-Involvement in Shifting IAT Effects

Explicit measures can be affected by self-involvement in processing of a message (Johnson & Eagly, 1989). Here, we show that self-involvement in a counter-stereotypical message also influences implicit measures such as the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998). In our study, racial attitudes changed only after reading a counter-stereotypical scenario in which participants were asked to imagine themselves as victims of an assault as opposed to simply imagine an assault to a person. This shift did not depend on evaluative instructions and it was transient as it was no longer present after 1 week. These results suggest that the self-involvement might be an important factor in shifting implicit measures.

An inkblot for sexual preference: a semantic variant of the Affect Misattribution Procedure

A newly developed Semantic Misattribution Procedure (SMP), a semantic variant of the Affect Misattribution Procedure (AMP), was used in three studies as an indirect measure of sexual interest. Using a known-group approach, homosexual men (Studies 1 and 2), heterosexual men (Studies 1 to 3) and heterosexual women (Study 3) were asked to guess the meaning of briefly presented Chinese ideographs as "sexual" or "not sexual". The ideographs were preceded by briefly presented primes depicting male and female individuals of varying sexual maturity. As hypothesised, the frequency of "sexual" responses increased after priming with pictures of individuals of the preferred sex and increasing sexual maturation. The SMP showed satisfactory reliability and convergent validity as indicated by correlations with direct and two indirect measures of sexual interest. In two further studies, the hypothesised pattern was replicated whereas a standard AMP with the identical prime stimuli did not produce this result. The potential usefulness of semantic variants of the AMP is discussed.

Negation as a means for emotion regulation? Startle reflex modulation during processing of negated emotional words

This study investigated startle reflex modulation in 33 healthy student participants during the processing of negated emotional items. To build upon previous research, our particular interest was to find out whether processing of negated emotional items modulates emotional responding in line with the logical meaning of the negated expression, or instead leads to paradox emotional effects that point in the direction opposite the one logically implied by the negation. Startle reflex modulation was assessed during silent reading of pleasant and unpleasant nouns. The nouns were either paired with the possessive pronoun my or with the negation word no. The startle eyeblink amplitude was enhanced during processing of the unpleasant pronoun-noun phrases and attenuated during processing of the pleasant phrases. Negation attenuated the startle eyeblink for negated unpleasant nouns and enhanced it for negated pleasant nouns. In line with this finding, negation decreased arousal ratings for unpleasant nouns and reversed the valence ratings for pleasant nouns. Our results are the first to show an effect of negation on both peripheral physiological and subjective indices of affective responding. Our results suggest that negation may be an effective strategy for spontaneous down-regulation of emotional responses to unpleasant, but not to pleasant, stimuli.

There is a fire burning in my heart: the role of causal attribution in affect transfer

The role of causal attribution in affect transfer of primes was addressed by examining the consequences of explicit evaluation of primes within the framework of the affect misattribution procedure (AMP; Payne, Cheng, Govorun, & Stewart, 2005). We reasoned that affect transfer occurs when primed affect remains diffuse and not bound to a specific object, hence capable of freely colouring subsequent evaluations of ambiguous objects. Accordingly, we propose that when people explicitly evaluate the prime, affect is clearly bound to the prime and becomes less capable of influencing subsequent judgements. Supporting this notion, affect transfer in the AMP was observed when participants ignored the primes, thereby keeping the primed affect relatively unbound. However, this effect disappeared when participants explicitly evaluated the primes before target stimuli were presented. Implications of these findings in determining how and when affect arising from one object carries over to another is discussed.

Are We Puppets on a String? Comparing the Impact of Contingency and Validity on Implicit and Explicit Evaluations

Research has demonstrated that implicit and explicit evaluations of the same object can diverge. Explanations of such dissociations frequently appeal to dual-process theories, such that implicit evaluations are assumed to reflect object-valence contingencies independent of their perceived validity, whereas explicit evaluations reflect the perceived validity of object-valence contingencies. Although there is evidence supporting these assumptions, it remains unclear if dissociations can arise in situations in which object-valence contingencies are judged to be true or false during the learning of these contingencies. Challenging dual-process accounts that propose a simultaneous operation of two parallel learning mechanisms, results from three experiments showed that the perceived validity of evaluative information about social targets qualified both explicit and implicit evaluations when validity information was available immediately after the encoding of the valence information; however, delaying the presentation of validity information reduced its qualifying impact for implicit, but not explicit, evaluations.

Dangerous Enough: Moderating Racial Bias with Contextual Threat Cues

Research shows that participants shoot armed Blacks more frequently and quickly than armed Whites, but make don't-shoot responses more frequently and quickly for unarmed Whites than unarmed Blacks. We argue that this bias reflects the perception of threat - specifically, threat associated with Black males. Other danger cues (not just race) may create a similar predisposition to shoot, and if these cues promote shooting when the target is White, they should attenuate racial bias. We embedded targets in threatening andsafe backgrounds. Racial bias was evident in safe contexts but disappeared when context signaled danger, and this reduction was largely due to an increased tendency to shoot White targets.

Being on the Lookout for Validity

Recently, Sriram and Greenwald (2009) introduced a new IAT-like measure, the Brief Implicit Association Test (BIAT). Because the BIAT is a new development, empirical evidence for its validity is yet scarce. This comment focuses on two possible approaches to validation research on the BIAT: (1) a pragmatic correlational approach and (2) an experimental approach aiming at causal understanding of the BIAT task. We argue that both approaches provide valuable and mutually complementing evidence, but only experimental research can conclusively show that the to-be-measured constructs causally influence BIAT scores. Because such a causal analysis is at the core of the validity problem, research on the BIAT should reduce the asymmetry in favor of correlational validation that emerged in traditional IAT research.

Base rate effects on the IAT

We investigated the influence of stimulus base rates on the Implicit Association Test (IAT). Using an East/West-German attitude-IAT, we demonstrated that both overall response speed and differential response speed underlying IAT effects depend on the relative frequencies of the stimulus categories. First, when those stimuli that are more common in reality also occurred more frequently in the stimulus list, response speed generally increased. Second, IAT effects increased when congruent blocks profited from the compatibility of frequency-based response biases (i.e., frequent target stimuli and frequent valence stimuli mapped onto the same response key), whereas IAT effects decreased when incongruent trial blocks profited from response compatibility. These findings demonstrate that the stimulus context moderates the magnitude of the IAT effect. Simultaneously, they highlight the need to explore the extent to which implicit measures reflect properties of the task or the environment rather than attributes of test-takers.

Fast and Fragile

Numerous studies suggest that processing verbal materials containing negations slows down cognition and makes it more error-prone. This suggests that processing negations affords relatively nonautomatic processes. The present research studied the role of two automaticity features (processing speed and resource dependency) for negation processing. In three experiments, we tested the impact of verbal negations on affective priming effects in the Affect Misattribution Paradigm. Going beyond previous work, the results indicate that negations can be processed unintentionally and quickly (Experiments 1 and 2). In Experiment 3, negations failed to qualify affective priming effects when participants’ working memory was taxed by memorizing an eight-digit number. In sum, the experiments suggest that negations can be processed unintentionally, very quickly, but that they rely on working-memory resources.