> Therefore, if people educated in the right way could hear voices I'd expect there still to be groups of people (members of some religious sect perhaps) who hear voices.
My (entirely made up) theory here is that it's mediated through stress and exhaustion. Even today there's something called "brief psychotic disorder" where during a period of extreme stress somebody can experience delusions, hallucinations, etc: https://en.wikipedia.org/wiki/Brief_psychotic_disorder
There are also a variety of religious traditions where modern people will hear the voice of god, enter trance states, channel spirits, etc. And we all know that sleep deprivation can produce all sorts of mental effects.
Given that civilization is mainly marked by improvements in life quality over time, it seems reasonable that the further back we go, the rougher people have it. So figure less sleep, worse nutrition, more desperation, more trauma, both direct and through the frequent death of children, friends, family.
So it seems reasonable to me that early on in civilization people were on average pretty bonkers. And social creatures that we are, that we'd be jointly bonkers; if our parents and siblings and and neighbors experience parts of themselves as the voice of particular gods, we would be likely to as well.
As our lives over generations get slowly better, the stress level drops. So too does the level of biologically driven psychosis. Which undermines the cultural elements that support and channel it, which might collapse relatively suddenly. (A reasonable modern parallel might be the European collapse in religious belief over recent generations.)
All pure speculation, of course. Jaynes's theory is certainly fun to think about, but we would have a hard time getting real experiments past an IRB.