(1A) there’s a message from S to Motor Area M that says to produce voices;
(1B) there’s a message from S to R that updates R on what S is doing (and in particular, it tells (R) that (1A) is happening right now);
(2) Motor Area M “does voices” (I’m hazy on the details), and some sensory consequence of those voices make their way back to R.
So then the auditory hallucination in my model would be if (1A) and (2) happen, but (1B) doesn’t happen.
Generally, I don’t think this story is very sensitive to timing. I think the nature of a hallucinated voice is that it feels exogenous not just for a tiny fraction of a second between the (2) signal and the (1B) signal arriving at R, but rather it continues to feel exogenous for many seconds.
The model would be:
(1A) there’s a message from S to Motor Area M that says to produce voices;
(1B) there’s a message from S to R that updates R on what S is doing (and in particular, it tells (R) that (1A) is happening right now);
(2) Motor Area M “does voices” (I’m hazy on the details), and some sensory consequence of those voices make their way back to R.
So then the auditory hallucination in my model would be if (1A) and (2) happen, but (1B) doesn’t happen.
Generally, I don’t think this story is very sensitive to timing. I think the nature of a hallucinated voice is that it feels exogenous not just for a tiny fraction of a second between the (2) signal and the (1B) signal arriving at R, but rather it continues to feel exogenous for many seconds.