This paper offers a simple but general model of egocentric perspective-taking by postulating a link between the extent to which people project their beliefs onto others and to which they anticipate each others' projection. We provide direct evidence for this link and explore various predictions of such projective thinking. When people learn whether others oppose or support a norm in settings where dissent is costly, they infer antagonistic preferences. The more costly dissent is, the more average opinion concludes that the norm is genuinely popular. In trade, informed traders bluff too little, uninformed ones are cursed, and the predictions closely match the experimental evidence.