To appear in Advances in Experimental Social Psychology 1 Reflection and Reflexion: A Social Cognitive Neuroscience Approach to Attributional Inference 1 Matthew D. Lieberman Ruth Gaunt University of California, Los Angeles Bar-Ilan University Daniel T. Gilbert Yaacov Trope Harvard University New York University 1 This chapter was supported by grants from the National Science Foundation (BCS-0074562) and the James S. McDonnell Foundation (JSMF 99-25 CN-QUA.05). We gratefully acknowledge Kevin Kim for technical assistance and Naomi Eisenberger for helpful comments on previous drafts. Correspondence concerning this chapter should be addressed to Matthew Lieberman, Department of Psychology, University of California, Los Angeles, CA 90095- 1563; email: lieber@ucla.edu. "Knowledge may give weight, but accomplishments give lustre, and many more people see than weigh." Lord Chesterfield, Letters, May 8, 1750 Lord Chesterfield gave his son, Philip, a great deal of advice—most of it having to do with manipulating other people to one’s own ends—and that advice has survived for nearly three centuries because it is at once cynical, distasteful, and generally correct. One of the many things that Lord Chesterfield understood about people is that they form impressions of others based on what they see and what they think, and that under many circumstances, the former tends to outweigh the latter simply because seeing is so much easier than thinking. The first generation of social psychologists recognized this too. Solomon Asch observed that “impressions form with remarkable rapidity and great ease” (1946, p. 258), Gustav Ichheiser suggested that “conscious interpretations operate on the basis of an image of personality which was already performed by the unconscious mechanisms” (1949, p. 19), and Fritz Heider noted that “these conclusions become the recorded reality for us, so much so that most typically they are not experienced as interpretations at all” (1958, p. 82). These observations foretold a central assumption of modern dual-process models of attribution (Trope, 1986; Gilbert, Pelham, & Krull, 1988), namely, that people’s inferences about the enduring characteristics of others are produced by the complex interaction of automatic and controlled psychological processes. Whereas the first generation of attribution models described the logic by which such inferences are made (Jones & Davis, 1965; Kelley, 1967), dual- process models describe the sequence and operating characteristics of the mental processes that produce those inferences. These models have proved capable of explaining old findings and predicting new phenomena, and as such, have been the standard bearers of attribution theory for nearly fifteen years. Dual-process models were part of social psychology’s response to the cognitive revolution. But revolutions come and go, and while the dust from the cognitive revolution has long since settled, another revolution appears now to be underway. In the last decade, emerging technologies have allowed us to begin to peer deep into the living brain, thus providing us with a unique opportunity to tie phenomenology and cognitive process to its neural substrates. In this chapter, we will try to make use of this opportunity by taking a “social cognitive neuroscience approach” to attribution theory (Adolphs, 1999; Klein & Kihlstrom, 1998; Lieberman, 2000; Ochsner & Lieberman, 2001). We begin by briefly sketching the major dual-process models of attribution and pointing out some of their points of convergence and some of their limitations. We will then describe a new model that focuses on the phenomenological, cognitive, and neural processes of attribution by defining the structure and functions of two systems, which we call the reflexive system (or X-system) and the reflective system (or C- system).