This study investigated how automatically activated racial attitudes and motivation to control prejudiced reactions contribute to the impressions formed of targets whose photos varied by race, gender, and occupation. In earlier sessions, participants completed Dunton and Fazio's (1997) Motivation to Control Prejudiced Reactions scale, and underwent a priming procedure (Fazio, Jackson, Dunton, & Williams, 1995) that provided an unobtrusive estimate of their automatically activated racial attitudes. In the final session, participants provided trait ratings of a number of target photos. Automatically activated racial attitudes were related to the trait inferences participants made of Blacks compared to matched Whites. However, this effect was moderated by motivation to control prejudiced reactions. Among the more motivated, both those with more negative and those with more positive attitudes displayed evidence of correction for their attitudes. This discussion focuses on the ways in which different kinds of motivation to control prejudiced reactions may manifest differently.First impressions are often formed as a function of the category of which the social target is a member. The general lesson from research in impression formation is that categorical-based impressions are relatively effortless and occur under conditions of low motivation or ability, and individuation is effortful and occurs under conditions of high motivation or ability (Brewer, 1988;Devine & Monteith;1999;Fazio, 1990;Fiske, Lin, & Neuberg, 1999;Fiske & Neuberg, 1990;Kunda & Thagard, 1996; but see Gilbert & Hixon, 1991;Spencer, Fein, Wolfe, Hodgson, & Dunn, 1998). Most models of impression formation assume that stereotypes and other group level information are activated from memory on perception of a group member (e.g., Banaji & Hardin, 1996;Banaji, Harden, & Rothman, 1993;Brewer, 1988;Devine, 1989;Fiske et al., 1999). 1 More deliberate and careful processing, on the other hand, can occur as the target's power over the perceiver increases (Fiske et al., 1999), under expectations of future interaction with the target (Johnston, Hewstone, Pendry, & Frankish, 1994), increased accountability for one's judgments (Tetlock, 1992), fear of invalidity (Kruglanski, 1989), and with the presence of unexpected, difficult, or surprising information that begs an explanation (Kintsch, 1988;Kunda, 1990).However, there are some notable qualifications to automatic stereotype activation. Gilbert and Hixon (1991) showed that under conditions of cognitive load, stereotypes may not be activated at all (see also Spencer et al., 1998). Stereotype activation also depends on the goals of the perceiver (e.g., Macrae, Bodenhausen, Milne, Thorn, & Castelli, 1997 In the realm of racial prejudice, other models of avoiding prejudiced responses suggest that motivated individuals might "replace," or "override" prejudiced responses with more positive ones, instead of attempting to correct for the bias associated with their automatically activated attitudes (e.g., Devine & Monteith,...