On Intelligence

Author(s): Jeff Hawkins with Sandra Blakeslee Publishers: Holt Edition: 2004 ISBN: 0805078533   Click here to purchase All sales commission goes to charity   < Back to list

After graduating with a degree in electrical engineering, Jeff Hawkins, the author of On Intelligence, became interested in the workings of the brain. In 1980, while working for Intel, Hawkins wrote to Gordon Moore, the legendary founder of Intel, suggesting that he lead a research team devoted to "understanding how the brain works". In his letter he wrote: "I am confident we can figure this out." Intel rejected his proposal but the seeds of his fascination in developing a unifying theory of how the brain works had been sown.

As something of an aside to his brain research, Hawkins spent the next few years of his life establishing the Palm Computer company and led the development of devices such as the PalmPilot, the Graffiti writing language and the Treo smart phone. On the back of the success Hawkins enjoyed in business, in 2002 he established his own neuroscience research centre (The Redwood Center for Theoretical Neuroscience), which is dedicated to finding an overall theoretical understanding of the neocortex - the part of the human brain responsible for intelligence.

The reason for Hawkins' interest in the neocortex is that he believes much of the research into artificial intelligence in the past has been misdirected. In his view, scientists have been trying to replicate the logical processes of intelligence rather than trying to understand and mirror the way in which the brain actually works.

To explain this Hawkins uses an analogy of carrying blocks of stone. Imagine you had to carry one hundred stone blocks across a desert. You can only carry one block at a time and each trip across the desert requires a million steps.

To complete the task more quickly it would obviously help if you recruited another 99 people to carry stone blocks for you.

The problem is that no matter how many people help you with the task, the number of steps required to move the stones across the desert is still one hundred million.

This is the equivalent of how computer power has developed. By building parallel processors, tasks that once took an eternity to complete can now be completed in a fraction of a second. The result is that today's computers can process information up to 5 million times faster than a human brain. But despite this tremendous speed and power, computers still cannot mirror basic human skills such as understanding jokes, recognise facial expressions or catching a ball. The reason humans can achieve these enormously complex tasks is that we do not process information in the same way as a computer - instead of treating each task as a new experience in the same way a computer would, humans rely extensively on our memory of performing similar tasks in the past.

Hawkins therefore argues that the primary purpose of the brain is not to process information like a computer, its primary purpose is to store and retrieve memories. Thinking is therefore a form of pattern-matching; associating new information with stored information in order to produce an answer. This is why a human brain can still outperform even the fastest computers on the majority of tasks.

In the rest of the book, Hawkins explains his theory of how the neocortex manages this process of memory recall and pattern matching.

On Intelligence is an interesting book that provides a different perspective on the brain. Moreover, if Hawkins' theories are eventually proved right, this book will become recognised as something of a seminal work.


Published October 2009