>And, unfortunately, the article doesn't mention any of it. The obvious prediction is: people who don't dream, don't generalize.
Which could very well be true, with a small correction: people who don't dream don't generalize well - their generalization capabilities decline. Or course with people who don't sleep well, their whole mental faculties decline over time.
Even after a couple of days lost sleep, people have been shown to drop several IQ points in standardized tests... and even lower quality sleep can lead to psychosis and worse (which could very well be fueled by lack of generalization).
Plus this "cleaning" process is not the part we remember as a story/dream when we wake: dreaming includes the parts we don't remember, including the botched parts that never fully made it into full deep REM sleep and conscious dreaming. So, people who don't dream will still clean up some of the overfitting, just do a worse job at it...
Which could very well be true, with a small correction: people who don't dream don't generalize well - their generalization capabilities decline. Or course with people who don't sleep well, their whole mental faculties decline over time.
Even after a couple of days lost sleep, people have been shown to drop several IQ points in standardized tests... and even lower quality sleep can lead to psychosis and worse (which could very well be fueled by lack of generalization).
Plus this "cleaning" process is not the part we remember as a story/dream when we wake: dreaming includes the parts we don't remember, including the botched parts that never fully made it into full deep REM sleep and conscious dreaming. So, people who don't dream will still clean up some of the overfitting, just do a worse job at it...