If one rejects the reality of consciousness as Moore understands it, there remains the need to replace his account of perception with a better one. How shall one understand propositions of the form "x sees y" without bringing consciousness into the picture? And since this form suggests that seeing is a relation, is there some alternative analysis which preserves its relational structure? Here is a suggestion of D.M. Armstrong's: "Perception...is the acquiring knowledge of, or inclination to believe in, particular facts about the physical world, by means of our senses."¹ this is one way of developing Descartes' view that sensing is a kind of thinking. We see something when we are having veridical visual sensations; a visual sensation is just the acquiring of a belief or an inclination to believe; it is veridical when the belief is true; that is what veridicality consists of. When a veridical sensation is acquired by means of the stimulation of the visual system by light reflected by or emitted from an object, then we see it. Since the sensation is the acquiring of a true belief caused by the object of that belief, we have knowledge as well. Since seeing includes a causal relation between the perceiver and the object, the relational structure of seeing is preserved. Seeing something in particular and sense perception in general are interpreted in this account as cognitive achievements. For this reason, I shall label this type of view the cognitive theory of perception. The version of the cognitive theory that we are now considering is the one which assimilates sensing to believing.