[roceasta]

>switch to the low level dual of this problem [...] reinforcement learning.

But that is itself a theory about the problem of consciousness! So why not do both? There's no obvious short cut through the confusion, but discoveries in one field may guide questions in the other. In general I think it helps to have one's feet on the ground as well as one's head in the stars (for instance Newton ground his own lenses).

[eli_gottlieb]

>But that is itself a theory about the problem of consciousness!

No it's not! It's a computational theory of motivated action.

[roceasta]

What I mean is that the idea that reinforcement learning is a 'low level dual' of the problem of consciousness is also a theory about the problem of consciousness. It's part and parcel of philosophical topics that one can't get out of the game...