Humour me, I'm a bit slow. It sounds like he's using the word "illusion" to mean "illusion", as in something which seems real but is fake. Does this not still entail a Cartesian theater and a consciousness watching it to be fooled by the illusion ?

But there's nothing in that passage to counter what I said. Illusion still needs some conscious entity to be fooled.

Besides, the idea that: "The underlying reality, however, what exists in itself and not just for us or for other creatures, is accurately represented only by the scientific image—ultimately in the language of physics, chemistry, molecular biology, and neurophysiology."

That's again a certain, centuries old, epistemology. And not a very good one. I would be better if he would have said "is best approximated" rather than "accurately represented".

I disagree. I have shown a random blob to an face-recognition algorithm. It "saw" a face were there was none. It was not conscious, but it was perfectly able to have illusions.