ATTITUDES AND SOCIAL COGNITION On the Genesis of Confidence Michael J. Gill Lehigh University William B. Swann, Jr. University of Texas at Austin David H. Silvera Universitetet i Troms0 Confidence in personality impressions is proposed to stem from the richness of people's mental representations of others. Representational richness produces confidence because it enhances the fluency with which people can make judgments, and it increases confidence even when it does not result in more accurate impressions. Results of 3 experiments support these propositions. A 4th experiment suggests that representational richness is increased by both pseudorelevant and relevant information, but not by irrelevant information. A 5th experiment suggests that representational richness has effects on confidence above and beyond the effects of metainformation (i.e., extracontent aspects of information). The implications of these findings for evaluating evidence of error in person perception and for reducing stereotyping and prejudice are discussed. "We grew up together—I know her better than I know my- self." "I've only talked to him twice, so I may be wrong about him." People often qualify reports of their impressions with statements such as these, suggesting that some of their beliefs about others are held with great conviction, whereas other be- liefs are held only tentatively. In this article, we investigate the origin of these differences in confidence. Understanding the origin of confidence is important because confidence seems to affect whether people translate their beliefs into behaviors. For example, a hiring committee may be in- trigued by a job candidate on skimming her vita but will hire her only after boosting its confidence through conversations with the candidate herself. Likewise, one would entrust a large sum of money to a broker only after becoming confident that the broker will not take the money and run. This tendency for confi- dently held beliefs to be a more potent impetus for behavior than tentatively held beliefs has been confirmed in several empirical Michael J. GUI, Department of Psychology, Lehigh University; Wil- liam B. Swann, Jr., Department of Psychology, University of Texas at Austin; David H. Silvera, Department of Psychology, Universitetet i Troms0, Tromse, Norway. This research was supported by a University of Texas continuing fellowship to Michael J. Gill. We thank Joel Blanchard, Cicely Bomar, Robert McCann, and Adriana Weisz for their assistance during the empir- ical phase of this work. Sincere thanks also go to Jennifer Bosson, Ryan Brown, Morris Goldsmith, Asher Koriat, Arie Kruglanski, and Vincent Yzerbyt for their thoughtful comments on an earlier version of this article. Correspondence concerning this article should be addressed to Mi- chael J. Gill, Department of Psychology, 17 Memorial Drive East, Lehigh University, Bethlehem, Pennsylvania 18015. Electronic mail may be sent to mjg6@lehigh.edu. studies (e.g., Berger & Mitchell, 1989; Fazio & Zanna, 1978; Pieters & Verplanken, 1995; Swann & Ely, 1984). Presumably, confidently held beliefs motivate action because people believe confidence stems from an accurate perception of reality. But does it? The earliest social psychological research to address the relation of confidence and accuracy was con- ducted by Oskamp (1965). His work suggested that clinicians' confidence was affected by factors that did not influence the accuracy of their beliefs. Recent research using everyday person • perceiver s has shown similar results. For example, Dunning, Griffin, Milojkovic, and Ross (1990) had college students at- tempt to predict the behavior of their roommates, strangers whom they interviewed, or complete strangers. In all cases, confidence exceeded accuracy, and there was only a modest correlation between the two. Similarly, Swann and Gill (1997) found marked dissociations between the confidence and accu- racy of dating partners' and roommates' global personality im- pressions. Moreover, Wells and Murray (1984) reported virtu- ally no relation between the confidence and accuracy of eyewit- ness testimony. Finally, the work of cognitive psychologists on such phenomena as feelings of knowing and judgments of learn- ing suggests that confidence-accuracy dissociations also occur in nonsocial domains (see Jacoby, Bjork, & Kelley, 1994, for a review). A consideration of the ubiquity of such confidence- accuracy dissociations raises the following question: If confi- dence does not arise from the possession of accurate beliefs, from where does it arise? REPRESENTATIONAL RICHNESS AND THE GENESIS OF CONFIDENCE We propose that the richness of mental representations plays an important role in producing confidence. Representational Journal of Peiwnality and Social Psychology, 1998. Vol. 75. No. 5, 1101-1114 Copyright 1998 by Ihe American Psychological Association, Inc. 0022-3514/98/K 00 1101