The IAT is sensitive to the perceived accuracy of newly learned associations

Three experiments tested whether the Implicit Association Test (IAT) is sensitive to the perceived accuracy of newly learned associations. In experiment 1, participants learned to associate positive or negative attributes with two novel groups. Participants in one condition were told that the attributes accurately described the groups; in a second condition, prior to learning, they were made aware that the attributes were randomly assigned to the groups. Participants were given an IAT and an explicit measure testing attitudes towards the two groups. When the participants were told that the attributes were accurate, their IAT performance and explicit measure responses indicated a preference for the more positively described group but when the attributes were known to be arbitrary, preferences were reduced according to both measures. Experiment 2 replicated these results and demonstrated that the associations were learned even in the random condition. Experiment 3 included a condition that placed “not” before each attribute, which demonstrated that people can incorporate a negative modifier into a learned association. Explicit attitudes and the IAT showed reversed preferences in this negation condition. These experiments imply that the IAT is sensitive to the perceived accuracy of learned associations. Copyright © 2011 John Wiley & Sons, Ltd. The study of attitudes is one of the central concerns of social psychology. An attitude is defined as an evaluative expression of an object or person. Attitudes have traditionally been thought of as consciously recognized constructs characterized by beliefs, behavioral intentions, and evaluations of attitude objects (Eagly & Chaiken, 1998). In recent years, researchers have developed dual attitude models. These models suggest that there is a sharp separation between the conscious and unconscious components of attitude. According to these theories, there are two distinct types of attitudes: Implicit attitudes, which exist outside of awareness, are activated automatically, require conscious effort to suppress, and are difficult to change. In contrast, explicit attitudes are constructed on the spot using whatever relevant information is consciously available and therefore require psychological effort to be activated and maintained (e.g., Wilson, Lindsey, & Schooler, 2000). Implicit attitude theories have led to the development of implicit measures. These measures are useful assessment tools because people are less able to hide their attitudes on an implicit measurement task. One of the most widely used implicit measures is the Implicit Association Test (IAT) introduced by Greenwald, McGhee, and Schwartz (1998). Demonstrating the popularity of the IAT, a recent meta-analysis by Greenwald, Poehlman, Uhlmann, and Banaji (2009) assessed the predictive validity of the IAT across 122 research reports. The IAT compares attitudes towards two groups by measuring the *Correspondence to: David E. Huber at the Department of Psychology, University E-mail: [email protected] Copyright © 2011 John Wiley & Sons, Ltd. association between the groups and positive and negative evaluations. When taking the IAT, participants must categorize four groups of items: positive adjectives, negative adjectives, and two distinct groups (e.g., Black faces and White faces). The IAT measures the attitude preferences towards the two groups by comparing reaction times between two blocks of trials. In one block, called the compatible block, the response to the preferred group uses the same response key as responses to positive adjectives. In the incompatible block, the response to the preferred group uses the same response key as responses to negative adjectives. People tend to respond more slowly in the incompatible block compared with the compatible block. This slowdown is called the IAT effect, and it can be measured by subtracting the average reaction time of the compatible block from the average reaction time of the incompatible block. This difference is used to determine the relative evaluation of the two groups (i.e., which group is preferred or, alternatively, disliked the least), although researchers have developed more advanced scoring algorithms designed to filter out extraneous factors that might influence a simple difference score (e.g., Greenwald, Nosek, & Banaji, 2003). It has been proposed that the IAT effect results from greater response conflict in the incompatible block compared with the compatible block, resulting in slower reaction times during the incompatible block to avoid errors (Greenwald et al., 2003). Response conflict arises from the associations held by the participants. In the incompatible block, the favored group of California, San Diego, 9500 Gillman Dr., La Jolla, CA 92093–0109, USA. Received 15 December 2010, Accepted 24 May 2011 190 Eric Siegel et al. shares a response key with the negative adjectives, and thus a stimulus from the favored group lends itself to two competing responses: one reflecting the group membership response and the other reflecting the association between that group and a positive evaluation response. In contrast, for the compatible block, the two responses elicited by a stimulus from the favored group indicate the same response key, and so there is no response conflict. A number of alternative explanations have been proposed for the process underlying the IAT effect. For instance, researchers have found that IAT performance is affected by the salience of the tested groups, indicating that the IAT effect partially reflects different degrees of familiarity for each group rather than associations between the groups and valence (Brendl, Markman, & Messner, 2001; Rothermund & Wentura, 2004). Other studies have found evidence that the IAT is sensitive to the associations between groups and cultural, rather than personal, constructs (Fazio, Han, & Olson, 2006; Olson & Fazio, 2004). Beyond these studies that question the type of association underlying the IAT, other studies have found that the magnitude of an individual’s IAT score is partially determined by individual differences in cognitive ability (e.g., Blanton & Jaccard, 2006; Klauer, Schmitz, Teige-Mocigemba, & Voss, 2010). Although these alternative explanations are disputed (e.g., Greenwald, Nosek, & Sriram, 2006), proponents on both sides of the debate agree that the direction of the IAT effect (i.e., whether the difference score is positive or negative, indicating a preference for one group or the other) is due to differences in the latent associations attached to each group, regardless of whether these associations are cultural, personal, or a sense of familiarity. Critically, it is now understood that these associations must be activated during the IAT to produce response conflict, and it has been found that mindset manipulations can change the IAT effect (Han, Czellar, Olson, & Fazio, 2010). Putting aside the manner in which these associations relate to attitudes, we seek to gain a better understanding of the associations that underlie an IAT effect. Because the IAT is typically used to assess previously learned associations (i.e., participants enter the lab with a lifetime of experience regarding the groups tested in the IAT), little is known about the types of learning that do or do not create the associations necessary to produce response conflict while taking the IAT. In the current study, we ask whether newly learned associations give rise to an IAT effect even if those associations are known to be inaccurate or arbitrary. In other words, we ask whether the IAT is sensitive to the perceived accuracy of newly learned associations. Karpinski and Hilton (2001) performed one of the earliest studies examining the influence of newly learned associations on the IAT. They argued that the IAT reflects the information people are exposed to rather than how they feel about that information: According to the environmental association model of the IAT, a high score on a White/Black IAT, for example, should not be seen as indicating that the individual has more favorable evaluations of Whites compared with Blacks. Instead, the score may simply indicate that the individual has been exposed to a larger number of Copyright © 2011 John Wiley & Sons, Ltd. positive-White and negative-Black associations than negativeWhite and positive-Black associations (Karpinski & Hilton,