I read Jeff Hawkins’ On Intelligence today. Fascinating. He gives what seems to be a coherent unifying theory. (Caveats: I don’t know any brain science, so I can’t tell if he’s right, if his facts are right, if he’s being fair, if he’s saying something old that just sounds new to me, etc.)

His theory has the virtue (and possibly the vice, but I can’t tell) of simplicity: The same process explains all the senses and all intelligence. As I understand it, our cortex has six layers, each of which stores patterns at different degrees of abstraction. Hawkins emphasizes the bidrectionality of the signaling: it’s not simply your optic nerve jangling the optic neurons which then get pattern-recognized; rather, the movement goes both ways, from the lower-order to the higher-order, and vice versa.

I’m not going to go any further into it because I probably already go it wrong. But here’s why I liked the book. First, it gets past the “brain is a computer” analogy. Instead, Hawkins says the brain is all memory, no CPU. It has no programs other than the most basic pattern-matching “algorithm” — and I don’t think that’s the right word…the ocean does a good job of sorting rocks on the beach without using any algorithms. Second, it gives a simple explanation that (if right) accommodates enormous complexity. Third, he gives us an experiment to try at home: Get a dummy hand, or, at worst, draw an outline of your hand on a piece of paper. Put it and your hand on a table, posed identically. Put up a divider so you can’t see your hand. Look at the dummy hand. Now have a friend touch both hands in identical ways simultaneously. Hawkins says that your brain will quickly get confused and “You will actually feel the sensations being applied to the dummy hand as if it were your own.” (p. 60-1.) The description is unclear so I don’t actually know what to expect, but I can’t wait to try it. (Despite that lapse, the book is generally well written, with much of the credit probably going to the tiny-type co-author, Sandra Blakeslee.)

I was bothered, however, by the use of mental language throughout. He talks about the cortex predicting, commanding, trying to interpret, etc. That felt sloppy to me, especially since he doesn’t always break those terms down into neural behavior. Also, towards the end he speculates on whether consciousness is something more than neural behavior and he concludes that “consciousness is simply what it feels like to have a neocortex.” (p. 196) That’s cute but it cheats: Expand “feels like” and we are back at the question that this phrase is supposed to answer. The the phenomenology of consciousness is outside the scope of his book is not a criticism.

Hawkins’ theory has some relevance to the book I’m writing because I want to know whether hierarchical thinking is basic to the brain or is something our culture has learned.

Hawkins postulates a physical correlate to the hierarchy of concepts: Functionally, the brain is a “branching hierarchy,” he says (44). As the sensory input of your face changes (you change expression, the light shifts, our vision twitches three times a second), the highest order neurons maintain a steady “That is a face” state. (His theory also builds ambiguity into the functioning of the brain: The higher-order neurons remember relationships, but not details. p. 75) If Hawkins is right, however, it’s hard for me to imagine what constitutes the highest level patterns. Do we have a bridge pattern sufficiently general that it recognizes both The Golden Gate and the bent blade of grass I watch the ants crossing? And that pattern sufficiently distinguishes bridges from the letter H? Do we have pre-existing patterns for all separable qualities? For orange? For metal? For constructed rather than natural? For things in San Francisco? For things you have to pay to cross? I suppose we must, but even though it’s all learned, having stored patterns for every which way we might understand our world feels so a priori. And how do we put these together? Or do we only take them apart? Most important, how badly am I misunderstanding Hawkins? I think quite badly. [Technorati tags: EverythingIsMiscellaneous JeffHawkins brain OnIntelligence]

Susan Crawford in April posted a terrific explanation of what’s new and exciting in the the Hawkins book. Run to it!

Categories: Uncategorized dw