To understand how decisions to invest in stocks are taken, economists need to elicit expectations regarding risk–return tradeoff. One of the few surveys which has elicited such expectations is the Survey of Economic Expectations in 1999–2001. Using the data from this survey, Dominitz and Manski find considerable heterogeneity across respondents that cannot be explained by simple models of expectations formation. Adapting a principle of dual reasoning borrowed from Kahneman, this paper classifies respondents according to their sensitivity to some pathologies. We find a substantial amount of unobserved heterogeneity between the least and the most sensitive respondents. We then sketch a model of expectations formation. Copyright © 2010 John Wiley & Sons, Ltd.