On intelligence

Jeff Hawkins, the author of On intelligence and founder of Palm One and Handspring, is not just an expert in mobile computing. For a long time, Hawkins has been interested in understanding how the brain works. His hope is that a better understanding on human intelligence will make it possible to build intelligent machines.

This passion for understanding intelligence led Hawkins to found the Redwood Neuroscience Institute in 2002 to advance research the field and, eventually, to write this book (together with Sandra Blakeslee).

On intelligence proposes a unified model to describe how the human cortex (the site of our intelligence) works. Hawkins’ hope is that neuroscientists will be able to use this model to accelerate the advance of research in that field; computer scientists might be able to use the model to start experimenting with new approaches to artificial intelligence.

The main ideas behind this model are summarized by Hawkins thus:

The human cortex is particularly large and therefore has a massive moemory capacity. It is constantly predicting what you will see, hear and feel, mostly in ways you are unconscious of. These predictions are our thoughts, and, when combined with sensory input, they are our perceptions. I call this view of the brain the memory-prediction framework of intelligence. [...] To make predictions of future events, your neocortex has to store sequences of patterns. To recall the appropriate memories, it has to retrieve patterns by their similarity to past patterns (auto-associative recall). And, finally, memories have to be stored in an invariant form so that the knowledge of past events can be applied to new situations that are similar but not identical to the past.

In the last chapters of the book, Hawkins briefly addresses ethical concerns around building intelligent machines, suggests possible applications of these ideas to the field of artificial intelligence and offers some cautious predictions.

Overall, I found the book an interesting and informative read. I definitely recommend it for those that are interested in learning a bit more about how the human brain works, or in new possible paths of exploration for the field of AI.

On the negative side, I got a bit bored by the overly detailed 70-page-long description of the cortex. Surely this detailed description is more interesting for someone that has some background in neuroscience (which I don’t), and it is necessary to give Hawkins’ ideas some scientific validity, but I found it the least fun part of the book.

More on the book’s official site: http://www.onintelligence.org/.