To see or not to see

Bill continues his musings on intentionality here.  First quoting himself, he says,

'Sees’ is often taken to be a so-called verb of success: if S sees x, then it follows that x exists. On this understanding of ‘sees’ one cannot see what doesn’t exist. Call this the existentially loaded sense of ‘sees’ and contrast it with the existentially neutral sense according to which ‘S sees x’ does not entail ‘X exists.’

I should add that I consider the existentially neutral sense of 'see' primary for the purposes of epistemology. For if visual perception is a source (along with tactile, auditory, etc. perception) of our knowledge of the existence of material things, then it seems obvious that the perception verbs must be taken in their existentially neutral senses. For existentially loaded uses of these verbs presuppose the mind-independent existence of material things. [my underlining]

I disagree that it seems obvious.  To me, the opposite seems obvious.  For if it's true that visual perception is a source of our knowledge of the existence of material objects, then surely this fact warrants the inference from

I see a cat,

to

There is a cat,

else it would seem that visual perception is not a source, etc.   In order to think about the puzzle of intentionality we have to assume that the senses give us knowledge of objects.  The puzzle is then how this comes about.  The assumption is implicit, for example, in the distinction between 'complete' and 'incomplete' objects.  If we can't assume this then it's no use Bill offering us a picture of his car and contrasting the seeing of the complete car with the incomplete seeing of the car.  For we are entitled to ask, What car?