What is a theory of perception? First of all a theory of perception must be a scientific theory which explains how a certain stimulus (described in physical terms) becomes a certain perceptual content (described in psychological terms). From this point of view it seems to me that we are very far from having a reasonable theory of perception, since, for instance, we have no answer to a question such as “how the wavelength 8000 Angstrom becomes the perception of red color” . We know partially how a train of photons impinges on our retina causing certain chemical processes, which produce a certain electrical signal, transmitted to the visual cortex. We know as well partially how the visual cortex works, but we have no idea of how a train of photons with wavelength 8000 Angstrom becomes red and one with 5000 becomes yellow.
Moreover there is a second question of normative character a theory of perception must answer: when a perception is truthful and when it is, as it where, “falseful”. Here there are at least two possible answers, the one of representationalism and the one of naive realism. The former maintains that each perception is a representation thrown by the subject while causally interacting with a body. This representation always has a content. A content is identified through its truthfulmess conditions. For instance the perception of a lamp is circular and this content is satisfied if the lamp is actually circular.
In my opinion this point of view is not compatible with our contemporary scientific world view, because we have no idea of how the causal nexus between the object of perception and our representation works. On the other side may be representationalism is compatible with our partial knowledge of perceptual mechanism. Indeed my representation of a rose as red is satisfied whether the rose I am seeing is actually red. But the latter “red” is not a perceptual content, but the stimulus defined in physical terms. In other words my representation of the rose as red is truthful if and only if the rose reflects photons with 8000 Angstrom of wavelength.
In the other perspective, i.e. naive realism, perceptions are all partially truthful, since the object of perception is part of the perceptual content. Here there is a clear-cut distinction between hallucinations on one side and genuine perception on the other. The latter comprehends illusions as well. Even illusions are partially truthful: for instance the broken oar under water, though actually it is not broken, nonetheless it is an oar. From this point of view perception is not a causal nexus, but a simple contact between subject and object. For the latter reason naive realism does not need to explain the strange production of qualia from physical stimuli. On the other hand this point of view is not compatible with the naturalistic thesis according to which red is not a quale, but a train of photons with 8000 Angstrom wavelength So naive realism has its problems as well.
Nevertheless there is an empirical fact, which confirms naive realism and disconfirms representationalism. The latter presupposes that hallucination and genuine perception are from a certain point of view indistinguishable. Representationalists affirms that hallucinations seems to the subject an actual perception. Only the world decides whether our perception is truthful or falseful. But from an empirical point of view this is not true. Whoever has had hallucination knows that they are phenomenologically different from genuine perceptions, since if one pays correctly attention it is clear that they come from our subjectivity. Even dreams are phenomenologically different form perception. Mental hilliness and naivete are indeed the incapability of making this distinction.
To sum up, from this point of view naive realism is better than representationalism, but the former still has many problems to solve in order to become a viable theory of perception.