The YouTube link below is of Ray Kurzweil speaking at an MIT deep learning course (MIT Professor Lex Fridman’s artificial general intelligence course). Kurzwiel is a futurist and Google Director of Engineering. The way in which Kurzweil explains how humans learn and think (e.g., sequences of sequences in the neocortex), and how his brain research provided hints at how to build better and “deeper” neural networks… Grab yourself a bag of popcorn and watch this 50 minute video, just amazing stuff here!
Bonus for those of you familiar with Markov chains: Kurzweil suspects that we learn via hidden Markov chains in neocortex module hierarchies. If this sounds like jibberish (which it did for me until I watched these videos), Kurzweil explains it easily and eloquently.