Interesting theory

Love the first part where the theory is explained.
As an”AI practitioner, the last parts had some interesting ideas.

Merci. Votre vote a été pris en compte.

Vous avez donné votre avis sur cette évaluation !

Something really new

This book is really three books: the first is brilliant, the second is interesting, and the third is plain loopy.

The Thousand brain theory forms part one of the book, and puts forward a theory for the structure and organisation of the neo-cortex which is arresting and inspiring. This largest and most modern and distinctive outer cover of the human brain consists of 150,000 1mmx2.5mm columns of ‘spaghetti’ which are almost indistinguishable from each other. This is odd because it implies that most activity (seeing, smelling, listening or thinking) is done by very similar structures and neurones. Vernon Mountcastle discovered this back in the 70s, and little pieces of the jigsaw (the sizes and arrangements of neurones) have been added since. The question is what is going on in the columns. Hawkins and his team reckon that some primitive organism created a brain structure that could create a map of its immediate environment - food here, light over there - and then locate and move itself around that reference frame. Obviously very useful. Then at a later stage the branch that became mammals starting replicating numerous copies of that module to eventually create the pocket-spring mattress that you are using to read these words. We use versions of reference frames and location to map out the features of a coffee cup and the locations of all our sensors (e.g. fingers); we use our experience of mapping out coffee cups to create reference frames for abstract concepts such as democracy. OK, there maybe a few gaps and heroic jumps, but the idea that we are living in a model of the world created by our cortex does strike true. When we try to draw things, we draw the model, not what our eyes see. Indeed, art lessons are a constant fight between what we know is there (white teeth) and what eyes actually see (a dark shadow under a lip). Our brain is flexible because all the neo-cortex does is create models from the inputs it receives. So whether it is examining a stapler in modern Britain, or an animal on the prehistoric savannah, a car, smartphone or behaviour of its spouse, it just observes and models and then makes predictions. The predictions from many columns cumulate their activity by ‘voting’ to create a consensus prediction. Your cortex is at the command of your old brain, and then follows the actions with the predicted outcomes that fit with the old brain’s ‘desires’. This might sound a bit light on details, but it does start to pull on the fibres of some theory that could unravel thought and consciousness, and I found it more convincing and satisfying than most other discussions I have read. It reminds me of how Steven Pinker analyses thought and language (Stuff of Thought) where he shows that abstract language derives from concrete language, ‘hold that thought; look here; let’s push off that task until tomorrow’. Abstract thought evolves from more concrete thoughts, but uses the same structures.

Part two compares this model of intelligence to that of current AI, and shows that currently AI is much more artificial than intelligent. Not a surprise. We all know that Google translate does not have a clue WHAT it is translating, but that a human translating a book must have a good understanding not just of the sense but the meaning and tone of the material, to do a good job. Jokes abound. Still, Hawkins is right to make the points, and he draws a very interesting line between intelligence and consciousness, and the brain he has just decorticated and the ‘old brain’ (that is probably even more subtle and complicated and multi-part) that gives us our actual desires and drives (as well as controlling breathing and all that stuff).

The third part of the book is weird and loopy. Hawkins seems to believe that we are important on a cosmic scale because we have, for example, measured the size of the earth, estimated the probable age of our universe and calculated pi at some length. We need to preserve this ‘knowledge’ for some future intelligent species who might eventually evolve after we are extinct, and spend most of their spare time searching for our remains. Well, why on earth would they be interested in the size of our planet? Or ex-planet? Why would they not be able to calculate the ratio of the diamètre to the circumference of a circle for themselves - and would they do it in base 10??? Ditto for the age of the universe. It is a fact that they may or may not be able to work out for themselves, but they won’t necessarily want it in billions of years, they use neither billions (again, rather base ten, finger counting) nor years, based on the time it takes our silly old planet to get round our silly old sun. The tyranny of the 300 page book. Still, well worth the credit.

Narration. Professional, very good job. Nice short chapters for auditing.

Merci. Votre vote a été pris en compte.

Vous avez donné votre avis sur cette évaluation !

1 personne a trouvé cela utile