Plato is often regarded as a founding father of modern epistemology, the first to develop theories of knowledge and belief. Recent work has complicated this picture by showing that epistêmê, the superior kind in his epistemology, is strikingly different from contemporary epistemology's knowledge and perhaps has more in common with what we now call understanding. 1 What about the inferior kind in Plato's epistemology, however-doxa? As one can see from standard translations, explicit discussions, and tacit assumptions, it is widely held that here Plato really is talking about just what we talk about now: belief. The aim of this paper is to show that Plato's doxa is very different from belief. It is a special kind of phenomenon, something which has no clear counterpart in contemporary epistemology, but which Plato thought absolutely central both to epistemology and to ethics: cognition of what seems. 2 Three important notes before I begin. First, this is not a paper about the meaning of the word 'doxa' in Plato. I am not trying to capture every use of that word, nor am I confining my attention to contexts in which Plato uses that word. I aim to show that we can find in his writings a developed theory of a specific kind of cognition, something which he usually refers to as doxa-although he sometimes refers to it using other terms (such as 'pistis' at Gorgias 454d or Timaeus 29c), and although he occasionally uses 'doxa' to refer to other phenomena, including, arguably, belief (see section 9). Second, I will focus on Plato's account of doxa in the so-called middle dialogues, especially the Republic, Phaedo, and Timaeus, for it is here that I think the account most developed, although it has clear roots in the earlier dialogues, and survives in some form in the later. 3 Finally, although I want to show that Plato thinks of doxa very differently from how people nowadays think of belief, I will not try to show that doxa is not belief. I can imagine an interesting argument that the phenomenon I will describe is belief, conceived very differently from how we conceive it today. I do hope however to show that the differences are radical enough and fundamental enough that it is not fruitful to study Plato's doxa by beginning from the assumption that it is belief. We should not use our theories and intuitions about belief to guide or constrain our interpretations of doxa. If we do, we not only create problems for Plato's texts where there are none, we also miss out on features of his epistemology that are revealing, compelling, and very different from our own.