Not what I expected, but pleasantly so: I thought the book would describe "computing machines", but, instead, it proposes an overarching theory that states that prediction is the primary purpose of the cortex.

Theoretical evidence is given by Hawkins and seems to fit quite well with recent discoveries in neuroscience. In any case, he proposes ways to test his hypothesis and underlying support.

Finally, he suggest how this theory may lead to "intelligent" machines that can be taught patterns and sequences and then allowed to discover new patterns and predictions. The sensory input from which to discover patterns and make predictions does NOT have to be limited to our 5 "human-senses", but can consist of additional or different senses, such as infrared detection, weather sensor data input, and so forth.

Hence, this book can be treasured for both the insights it provides for a model of how the real human brain works, and how this model may lead to new intelligent machines, even if proven to be not completely correct. ( )

Hawkin's book is very interesting, especially for those interested in figuring out what the mind does when it thinks, and why it seems that the computational model of intelligences falls a bit short.

Hawkins's idea is this: intelligence cannot be reduced to computing (as that discounts novel ideas and creativity). It cannot be reduced, either, to intelligent behavior, as one can be intelligent without doing anything but thinking. Hawkins suspects that intelligence is much more about the ability to make predictions - to intake factual information, remember it, and use it in future instances to be able to predict situaitons.

As a teacher, I think that Hawkins makes a lot of sense, as his definition jibes with what we test on intelligence tests. We test factual retention (which is the first part of Hawkins theory), the ability to recognize patterns (the ability to recall relevent facts at appropriate times), and the ability to complete patterns (predict using relevant data).

Hawkins also talks a lot about how this faculty resides in the very thin neocortex which, coincidentally, we have much more of than other mammals. He does a good job at describing how the neocortex works and, while it is necessarily wattered down, is quite thorough nonetheless.

For those interested in figuring out what intelligence may consist of, and who are dissatisfied with current seemingly incomplete attempts, this is a very good book to read. ( )

This is very easy to read book to Jeffs ideas on AI.

Jeff does come across as a bit pompous and self assured at times and I am not so sure about the originality of his ideas as he is. He keeps claiming that others in the field have come up against dead ends because they did not understand the full implications of feed back in neural networks. Yet before I read the book I thought they worked as he broadly describes, and I had picked that knowledge up over the years by reading various bits and pieces. It makes me wonder if he is building a bit of a strawman to make his own position stronger.

Regardless it is a very interesting book. If I had the time I would get stuck into the Numenta software that his company has made and is now available. ( )

For the detailed, nit-picking review : http://www.perichoresis.com/onintelli...

In brief - lots of good stuff to chew on about what makes us smart, marred by Hawkins being a sloppy philosopher. ( )

This comprehensive theory of intelligence is easily one of the best popular science books I've ever read. If I were just starting college, reading this book would have radically changed my career goals. It may even now. ( )