Most of the things I will say below are not based on solid arguments at the present stage; I just want to outline my larger project for future research.
Philosophy of perception in the Anglo-Saxon tradition nowadays is generally shaped by philosophy of language. This is hardly surprising, since after the linguistic turn no branch in analytic philosophy (I like this label, but I am not going to defend this now) can be isolated from the linguistic approach. I tend to think, contra the currency, that this approach has serious limits when it comes to perceptual contents. It has been over-emphasized, and thereby over-reacted.
Most of us believe that there are rational relations between perceptions and beliefs, and we try to make good sense of this fact (notable exceptions include Donald Davidson and Robert Brandom, who downplay the very idea of perception). Rational relations require certain structures, and many analytic philosophers attempt to start with language and apply the structures to perception, since the latter seems to be much more elusive. Many of them think that if we do not regard perceptual contents as propositional, it would be impossible to account for their rational relations to beliefs. If perceptual contents are propositional, presumably they have to be conceptual, since the constituents of propositions have to be concepts.
But conceptualism seems phenomenologically off key, so nowadays very few philosophers hold this opinion. Bill Brewer changed his mind recently, and John McDowell is also not as resolute as before. In renouncing propositionalism of perception, it is difficult to understand his remarks that perceptions are "intuitional," albeit still "conceptual in form." It is not that this position is not defensible, but at least we can see that linguistic-driven philosophy of perception may go too far.
Nowadays more analytic philosophers of perception are phenomenologically-driven. This approach has its roots in British empiricism and the Phenomenological tradition, to be sure, but still it is partially due to the unsatisfactoriness of the linguistic-driven approach. The problem, however, is that it would be so difficult to cash out the rational relations between perceptions and beliefs without attributing propositional structures to perceptual contents. So a great chasm has been generated. The phenomenologically-driven philosophers generally do not take relevant thoughts of Davidson and Brandom seriously, and vise versa. The L-driven philosophers regard their approach as the only way to avoid something like the Myth of the Given, and the the P-driven philosophers regard their approach as the only way to do justice to perceptual experience. So in his 2006 paper, Tim Crane boldly asserts that his "interface philosophy" in Putnam's sense will not give arise any metaphysical or epistemological anxieties. This might be true, but we cannot know this without arguments.
There are a lot more to be said about this divide, but I shall stop for now and shift to something more positive. Let me start with David Chalmers' recent effort to apply his two-dimensional framework to perceptual contents. It assigns three different kinds of content to experiences: Fregean, Russellian, and Edenic. Details aside, Chalmers' picture has the merit that it is both heavily linguistic- and phenomenologically-driven: the 2D framework is linguistic enough, and the reason he adds Edenic content in the case of perception is that he thinks Fregean content itself is not phenomenologically adequate enough. Now, I tend to think that Chalmers' picture is still too linguistic. It relies crucially on both Fregean content and Rusellian content, and presumably they are propositional in nature. This is not clear, for Chalmers emphasizes at the end of the 2006 paper that his Fregean and Russellian contents need not be conceptual, and I take this to be also a reservation about propositionality. Still, Chalmers' analyses of perceptual contents are based on his view on linguistic contents, and he tries to make the whole picture phenomenologically more adequate by adding something else.
I have no good argument to raise against his view now, but I think it sounds more right to say that perceptual contents are not propositional at all. Consider an argument paralleled to the one often used by the L-driven philosophers: since perceptual contents are both spatial and temporal, in order to secure the rational relations between perceptions and beliefs, the contents of belief have to be conceived as both spatial and temporal in character. But this does not sound right. If not, however, why should we accept the argument that since beliefs are propositional, perceptual contents must be conceived as propositional as well in order to account for the rational relations? As analytic philosophers, we are too used to the idea that contents are propositional, so when applying to perceptions, it might sound weird, but still somewhat acceptable. But if so, how do we accommodate the parallel argument for the conclusion that the contents of belief are spatial and temporal?
I am more willing to say that perceptual contents are Carnapian, roughly in the sense of Aufbau. I know this idea has generated various problems, and most philosophers today think it only leads to a dead end. Again, I do not have good things to offer for now (But notice that Chalmers himself also elaborates this Carnapian-style project in his Locke Lectures). But I hope the parallel argument outlined above can motivate a more phenomenologically-driven picture. It is true that British empiricists were too naive in relevant regards, and the Phenomenological tradition seems to lack enough attentions and apparatus concerning linguistic structures, but still to attribute propositional structures to perceptions seems to be as wild as to attribute spatial/temporal structures to beliefs, or so I shall urge.