It's entirely possible the way our human brains work may very well end up reducible to a sequence of very very complicated matrix operations.

If that's the case, there doesn't need to be anything new; fundamentally, linear regressions may be sufficient.