If you are using the premise that no one understands quantum mechanics to assert a conclusion bearing on human knowledge, then I am afraid that your argument is lost on me.

In other words, such a network is unable to distinguish between a veridical perception and an hallucination. Doing so would require the existence of another, independent network, whose task it is to learn to associate specific input patterns with specific patterns of activity of the first network’s hidden units. That system would then be able to identify cases where the latter exists in the absence of the former, and hence, to learn to distinguish between cases of veridical perception and cases of hallucination. Such internal monitoring is viewed here as constitutive of conscious experience: A mental state is a conscious mental state when the system that possesses this mental state is (at least non-conceptually) sensitive to its existence. Thus, and unlike what is assumed to be case in HOT Theory, meta-representations can be both subpersonal and non-conceptual.