Marr's theory of vision (1982) is often said to exemplify wide psychology. The claim rests primarily on Marr's appeal to a high level theory of computational functions and secondarily on the sort of representations he posits for computational processes (Burge 1986, 28). I agree that Marr's theory embodies an exemplary form of wide psychology; what is exemplary about it is his appeal to perceptual tasks. But I shall argue that the result of invoking task considerations is that we need not adhere to Marr's own restricted construal of proper width. Indeed, the larger conclusion I want to draw is that there is no single conception of width that has a special place in explanation.