His thesis boils down to the idea that intelligence is about processing patterns over time, and predicting what patterns will come next. He says that all our sensory inputs, whether vision, hearing, taste, etc. all are converted to patterns of impulses. A single moment of impulse is typically not too interesting; it’s the changes in the patterns over time that are of interest. In the neocortex of the brain, the structure appears to be six layers, where patterns are detected and increasingly cross-verified. He claims that we are continually processing the inputs we get, to either confirm the patterns we are expecting or to find that something new is happening and adjusting. The patterns aren’t hard-wired into the brain; they are learned from experience. (My apologies if I’m oversimplifying way too much!).
Side note: Hawkins' idea of the changes relates to Gregory Bateson's notion of difference.
This all seems sensible enough to me. He goes on to claim that we can build machines that will be able to do this type of pattern processing; though he’s less interested in the attempt to build physical ‘robot’ type machines that physically move in ways that we move. I find this more likely than Ray Kurzweil’s claims that we will be downloading humans in the next 40 years.