Hm, the comparison of sight to the other senses is not homologous. You say that what you see resembles what is seen (how could it not?). OK, a little odd, but fine. But then you say that the smell of grass does not resemble grass. But what does that even mean? It seems to mean: the smell of grass does not resemble the sight
of grass. But why the privileging of sight? After all, it doesn't seem like the reverse operation is admissable - why not say, 'the sight of grass does not resemble the smell of grass?'.
In other words, what you call 'resemblance' already takes sight as its privileged sense. But why? Why is 'the wall which produces the sensation' understood on the modality of sight - a conflation of a sensory modality with the sheer existence of the wall as such.
The very language of resemblance is odd too: the idea is that you have two terms, X and Y, where one can or cannot resemble the other. But in the case of sight, the issue of resemblance apparently does not apply, insofar as there is simply one term: 'that which is seen'. But for some reason - and the confusion here seems linguistic rather than substantial - two terms are
admitted (arbitrarily?) for the other senses, except, having conflated sight with existence, every other sensory modality is judged to fail to 'live up to' the 'resemblance' understood as 'what it looks like'. But what kind of problem is this? Seems to me like asking why a fish can't climb a tree, despite the fact that for some inexplicable reason the fish seems to do very well in water. But the problem here is not with the fish, but the question itself. But perhaps I'm missing something. If so, what?
Yet another consideration: from a phenomenological standpoint, this separation of sensory modalities is artificial from the get-go. The idea that things don't smell like they look, or feel like they sound is simply not true to experience, outside of some very narrow and artificial boundaries. To quote Alphonso Lingis:
"A thing is not a whole assembled by the central nervous system out of separate sensory data, nor is it a conceptual term posited by the mind and used to interpret the data being recorded on the separate senses. The sense organ focused on a pattern is a segment of the whole interconnected mass of the sensory nervous system. What we pick up with the eyes is already sensed by the whole sensitive substance of our body. When we see the yellow, it already looks homogeneous or pulpy, hard or soft, dense or vaporous, it already registers on our taste and smell; anything that looks like brown sugar will not taste like a lemon. To see it better and to see it as a thing is to position oneself before it and converge one's sensory surfaces upon it. It is the postural schema that comprehends things. To recognize a lemon is not to conceive the idea of a lemon on the occasion of certain sensory impressions; it is to know how to approach such a thing, how to handle it, so that its distinctive way of filling and bulging out space, its distinctive way of concentrating color and density and sourness there becomes clear and distinct" (Lingis, Sensation).
Or in yet other words: all sensing is synesthetic from the get-go, and the parcelling out of senses into discrete modalities is an artificial, analytic operation undertaken after the fact, on the basis of a rationalist confusion.