I'll begin by reviewing a wide range of methods across cognitive science, which demonstrate that knowledge (rather than belief) is the more basic way of representing others' minds. I'll give particular focus to a growing series of studies that show (1) people can accurately evaluate others’ knowledge more quickly than they can evaluate their beliefs, (2) this difference extends to participants with Autism Spectrum Disorder and cannot be explained by pragmatic differences in knowledge and belief ascriptions, (3) this difference occurs cross-linguistically and cannot accounted for by differences in word frequency, (4) this difference also generalizes to the larger class of factive and non-factive attitudes (to which knowledge and belief respectively belong), and finally (5) using fMRI data, I'll show that the neural response that occurs when making evaluations of others’ beliefs is absent when making similar evaluations of knowledge. Together, these studies demonstrate that human adults can attribute or deny knowledge states without first evaluating belief states. At a broad level, I'll argue that these findings lend support to the view that knowledge representations are a basic and distinct way in which we understand others’ minds. I'll end by stepping back and outlining what I take to be the essential difference in format between knowledge and belief representations.