Blakeslee Sandra - On Intelligence

FROM THE INVENTOR OF THE PALMPILOT COMES A REVOLUTIONARY NEW THEORY OF INTELLIGENCE AND A BOLD VISION FOR THE FUTURE OF INTELLIGENT MACHINES.

Jeff Hawkins, who created the PalmPilot, the Treo smart phone, and other handheld devices, has reshaped our relationship to computers. Now, with this brilliant book about the true nature of human intelligence, Hawkins stands ready to revolutionize both neuroscience and computing in one stroke.

On Intelligence develops a powerful theory of how the human brain works, explaining why computers are not intelligent and how, based on this new theory, we can finally build intelligent machines. Previous attempts at replicating human intelligencethrough artificial intelligence and neural networkshave not succeeded. Their mistake, Hawkins argues, was in trying to emulate human behavior without first understanding what intelligence is.

The brain is not a computer, supplying by rote an output for each input it receives. Instead, it is a memory system that stores experiences in a way that reflects the true structure of the world, remembering sequences of events and their nested relationships and making predictions based on those memories. It is this memory-prediction system that forms the basis of intelligence, perception, creativity, and even consciousness. Intelligence is the capacity of the brain to predict the future by analogy to the past.

In an engaging style that will captivate audiences from the merely curious to the professional scientist, On Intelligence explains what intelligence is, how the brain works, and how this knowledge will finally make it possible for us to build intelligent machines, in silicon, that will not simply imitate but exceed our human ability in surprising ways.

Written with the acclaimed science writer Sandra Blakeslee, On Intelligence promises to completely transfigure the possibilities of the technology age. It is a groundbreaking book in neuroscience, psychology, and the quest to build intelligent machines.

JEFF HAWKINS is one of the most successful and highly regarded computer architects and entrepreneurs in Silicon Valley. Currently the chief technology officer at palmOne, he founded Palm Computing and Handspring and created the Redwood Neuroscience Institute to promote research on memory and cognition. He is a member of the National Academy of Engineering and serves on the scientific board of Cold Spring Harbor Laboratory. He lives in northern California.

SANDRA BLAKESLEE has been writing about science and medicine for the New York Times for over thirty years and is the coauthor of Phantoms in the Brain by V. S. Ramachandran and of Judith Wallersteins bestselling books on psychology and marriage. She lives in Santa Fe, New Mexico.

Whenever someone asks me, What do you do for a living? I never know what to say. For the truth is I do very little. But I have surrounded myself with people who seem to get a lot done. My contribution is to nudge them every once in a while, and when necessary try to redirect the team down a new path. To the extent that I have had success in my career, I owe it mostly to the hard work and intelligence of my colleagues.

I have had the chance to meet many scientists and almost all of them have taught me something and therefore they have all contributed to the ideas in this book. I thank them all but can mention only a few here. Bruno Olshausen, who has joint positions at the Redwood Neuroscience Institute (RNI) and the University of California at Davis, is a walking encyclopedia of neuroscience. He is constantly pointing out what I dont know and suggesting ways to rectify my ignorance, which is one of the most valuable things a person can do. Bill Softky, also at RNI, was the first person to teach me about reduction in time in the cortical hierarchy and the properties of thin dendrites. Rick Granger at the University of California at Irvine gave me insights into sequence memory and how the thalamus could play a role. Bob Knight at the University of California at Berkeley and Christof Koch at the California Institute of Technology have been instrumental in the formation of the Redwood Neuroscience Institute and many other science matters. All the staff at RNI have challenged me and forced me to refine my ideas; many of the proposals in this book were a direct result of meetings and discussions at RNI. Thank you all.

Donna Dubinsky and Ed Colligan have been my business partners for a dozen years. Through their hard work and assistance, I was able to be an entrepreneur while simultaneously working part-time on brain theory, an unusual arrangement. Donna used to say one of her objectives was to make our businesses successful so I could afford to spend more time on brain theory. This book wouldnt exist if it werent for Donna and Ed.

I couldnt have written On Intelligence without lots of help. Jim Levine, my agent, believed in this book even before I knew what I would write. Dont write a book without an agent like Jim. He introduced me to Sandra Blakeslee, my coauthor. I wanted this book to be readable by a wide audience and Sandy was essential for achieving that. I take all the blame for the difficult sections. Matthew Blakeslee, Sandys son and also a science writer, provided several examples used in the book and suggested the term memory-prediction framework. All the folks at Henry Holt have been great to work with. I would like to especially thank John Sterling, president and publisher of Henry Holt. I met John face-toface only once, and we talked on the phone only a few times. That is all it took for him to have a huge impact on the structure of this book. He immediately understood the issues I would face proposing a theory of intelligence and then he suggested how the book should be written and positioned.

I would like to thank my daughters, Anne and Kate, for not complaining when their dad spent many weekends at the computer keyboard. And, finally, I want to thank my wife, Janet. Being married to me cant be all that easy. I love her more than brains.

Every theory should lead to testable predictions, for experimental testing is the only sure way to determine the validity of a new idea. Fortunately the memory-prediction framework is grounded in biology and leads to several specific and novel predictions that can be tested. In this appendix I list predictions that can falsify and/or support the proposals made in this book. This material is somewhat more advanced than the material in and is definitely not required to understand the rest of the book. Several of the predictions can only be performed on awake animals or awake human subjects because the tests involve expectation and prediction of the onset of a stimulus. The predictions are not ranked in terms of importance.

Prediction 1

We should find cells in all areas of cortex, including primary sensory cortex, that show enhanced activity in anticipation of a sensory event, as opposed to in reaction to a sensory event.

For example, Tony Zadors lab at Cold Spring Harbor Laboratory has found cells in rat primary auditory cortex that fire precisely when the rat expects to hear a sound even when there is no sound (private correspondence). This should be a general property of cortex. We should find similar anticipatory activity in visual cortex and somatosensory cortex. Cells that fire in anticipation of a sensory input are the definition of prediction, a basic premise of the memory-prediction framework.

Prediction 2

The more spatially specific a prediction can be, the closer to primary sensory cortex we should find cells that become active in anticipation of an event.