If you look at the rescorla wagner model of classical conditioning, you see that there's this (lambda-Vtot) term in the deltaVx equation, where lambda is the strongest possible association for the unconditioned stimulus and Vtot is the sum of the association strengths for all conditioned stimulus. When the two terms are equal, the learning rate drops to zero. This keeps us from "learning" extraneous associations for a stimulus that has already been explained.
And from what I have seen with hypnosis, it looks like Vtot ~= lambda in general. In other words, I think we walk around thinking that we already know everything and that there is nothing left that needs to be explained, so there's nothing to learn. It fits with how people can just brush off unexpected events as "god dun it" or "it's just random" - and then go on with their day as if there's nothing strange happening. Hypnosis and mind hackery are all about creating a gap where you think you have something to learn - whether it's explaining the uncharted territory of "hypnosis" from a position of authority, or driving in a wedge with a pattern interrupt or confusion pattern.