Hi Benj,
Just to be pedantic, I said only that neuronal evidence would be relevant, not that it is conclusive. Somebody could argue that despite neuronal indiscriminability, (M) is object-involving, but (Hal M) is not. Indeed, that's what I take Snowdon to be arguing -- though I disagree.
But, regarding this:
"My DR cheerfully grants the possibility of perfect hallucinations that
are neuronally qualitatively identical to cases of seeing. The reason for this is that DR is a position
about visual consciousness, rather than about the physical realization
of visual consciousness."
If your DR thinks it possible that visual consciousness is not supervenient on neuronal states, then s/he is committed to a fairly robust dualism not implied by DR itself. (Or have I misunderstood? Perhaps s/he takes an externalist/relationalist view of consciousness itself. That position is unacceptable to me for reasons that go beyond the topic of this thread. But see Block
2005)
I agree with your second point: "hallucinating may be subjectively
indiscriminable from seeing". I was assuming that taking an after-image for a distal light is subjectively indiscriminable from seeing a distal light. In such a case, "it is to the subject as if there is a light in the distance". Moreover, I agree with you that it is to the subject as if s/he has "landed a substantive demonstrative".
By contrast, imagine you are swimming under water in a clean clear empty pool, where you encounter a scene that contains light but no objects (other than your own body). In such a situation, you may have a visual experience of light, but with no objects, no demonstratives.
My own view (which I didn't push earlier) is that visual object-perception not only supports demonstrative thought but is, in itself, a state of consciousness that has an object-locating component -- and that this gives the state a demonstrative aspect. So it is not that I am "pushing" demonstratives from thought onto consciousness, but rather that I locate demonstratives in consciousness for independent reasons.