In arising rational-expectations the economic from simple models extrapolation revolution of old, in agents or macroeconomics, error-correction had backward-looking and rules. in microeconomics Then expectations, came the arising from simple extrapolation or error-correction rules. Then came the rational-expectations revolution in macroeconomics, and in microeconomics the spread and increasing refinements of modern game theory. Agents were now highly sophisticated information processors, who could not be systematically fooled. This approach reigned for several decades until the pendulum swung back with the rise of behavioral economics and its emphasis on "heuristics and biases" (as in Tversky and Kahneman 1974). Overconfidence, confirmation bias, distorted prob-ability weights, and a host of other "wired-in" cognitive mistakes are now common assumptions in many areas of economics. Over the last decade or so, the pendulum has started to swing again toward some form of adaptiveness, or at least implicit purposefulness, in human cognition. In this paper, we provide a perspective into the main ideas and findings emerging from the growing literature on motivated beliefs and reasoning. This perspec-tive emphasizes that beliefs often fulfill important psychological and functional needs of the individual. Economically relevant examples include confidence in ones' abilities, moral self-esteem, hope and anxiety reduction, social identity, polit-ical ideology and religious faith. People thus hold certain beliefs in part because This content downloaded from 141.20.102.129 on Thu, 09 Feb 2017 09:54:03 UTC All use subject to http://about.jstor.org/terms 1 42 Journal of Economic Perspectives they attach value to them, as a result of some (usually implicit) tradeoff between accuracy and desirability. Such beliefs will therefore be resistant to many forms of evidence, with individuals displaying non-Bayesian behaviors such as not wanting to know, wishful thinking, and reality denial. At the same time, motivated beliefs will respond to the costs, benefits, and stakes involved in maintaining different self-views and world-views . These tradeoffs can be influenced by experimenters, allowing for empirical tests, and by a person's social and economic environment, leading to the possibility of self-sustaining "social cognitions."1 At an individual level, overconfidence is perhaps the most common manifes-tation of the motivated-beliefs phenomenon. There is considerable evidence of overoptimistic tendencies on the part of consumers, investors, and top corporate executives (as discussed in a "Symposium on Overconfidence" in the Fall 2015 issue of this journal). While excessive overconfidence is quite dangerous, moderate amounts can be valuable: hope and confidence feel better than anguish and uncer-tainty, and they often also enhance an individual's ability to act successfully on their own behalf and interact productively with others. Using data from the Survey of Consumer Finances, Puri and Robinson (2007) thus find that more optimistic indi-viduals work more, save more, expect to retire later, and are more likely to remarry after divorce. Alloy and Abrahamson (1979) and Korn et al. (2014) find that most psychologically "healthy" people display some degree of overoptimism and biased updating, while it is primarily depressed subjects who seem to be more objective. People thus finds themselves motivated (often unconsciously) to achieve "positive" beliefs, and this typically occurs through a fundamental asymmetry in the process by which beliefs are revised in the face of new evidence: individuals update suitably when facing good news, but fail to properly account for bad news (Eil and Rao 201 1 ; Möbius, Niederle, Niehaus, and Rosenblat 2011; Sharot and Garrett et al. 2016). Although goal-directed, self-deception can nonetheless end up hurting the individual: since it is an informational game that people play with themselves, the outcome may be highly inefficient -a form of self-trap. When motivated thinking becomes a social phenomenon, consequences can be even more severe. Collectively shared belief distortions may amplify each other (an issue we shall address), so that entire firms, institutions, and polities end up locked in denial of unpleasant realities and blind to major risks: unsustainable fiscal imbalances or labor market policies, climate change, collapse of housing or financial markets, and so on. Case and Shiller (2003) surveyed the expectations of homeowners during the real-estate bubbles of 1988 and 2003. In both cases, 90 percent of respondents thought housing prices in their city would "increase over the next several years," with an average expected gain for their own property of 9 to 15 percent per year over the next ten years. In