On Intelligence

A few days ago I finished 'On Intelligence'. One of the most inspiring books I ever read. It's a very accessible, yet in depth, explanation of how the brain works. While many statements sound bold and unfounded, there is a scientific base to it, and most importantly a few falsifiable predictions are made so the theory can be scrutinized the way Carl Popper would approve it.

Jeff Hawkins explains how neural networks are no way near the way a real brain works. He also explains why the whole AI movement is on the wrong track and proposes a very interesting algorythm for the brain which integrates a lot of recent work on neurology. If you're interested in the algorythm: go read the book. I won't try to explain it in one post.

What I found stunning is that the algorythm Hawkins proposes explains a lot of patterns in human behaviour that are mentioned in other books I like. (McWhinney, Korzybski, Pirsig, etc.) It's like a giant puzzle coming together.

One of Hawkin's coworkers has modeled the algortyhm in a computer. It seems to work. I want to try that too. It seems the new LEGO NXT has a more powerful chip and programming software by the guys at National Instruments. Maybe I can build intelligent stuff with that...

My favorite quote from the book is "Consciousness is what it feels like to have a neocortex". Meaning that the hierarchical neocortical system that allows for abstraction and pattern recognition makes us experience what we call consciousness.