Dreams are vivid, emotional and often bizarre. Do they do something useful for memory and learning, and is their strangeness part of the point?
If dreams help the brain generalise, AI training could borrow the trick — and we'd understand sleep far better.
One proposal says strange dreams help the brain generalise, like noise in AI training. It makes testable predictions but hasn't been tested, and other sleep theories remain live.
A neutral summary kept current by super users, not a verdict.
Dreams supply corrupted, unusual input that stops the brain over-fitting to the day.
Hoel 2021 ('overfitted brain')
Dreams are a by-product of memory replay and consolidation, not a function in themselves.
Open
No decisions yet. When this room changes a flow or an evidence label, it is recorded here permanently.
Dhi: Room opened. A deciding test: does selectively disrupting dreaming harm generalisation more than memorisation?