On Intelligence
by
Jeff Hawkins & Sandra Blakeslee
Order:
USA
Can
Times Books, 2004 (2004)
Hardcover, e-Book
Read an Excerpt
Reviewed by Hilary Williamson
J
eff Hawkins, inventor of the Palm Pilot, believes that the '
question of intelligence is the last great terrestrial frontier of science.
' In 2002, he started a brain theory research center, the Redwood Neuroscience Institute, '
dedicated to finding an overall theoretical understanding of the neocortex - the part of the human brain responsible for intelligence.
' Hawkins believes that the crux of intelligence is the brain's
predictive
ability, and he explains why in
On Intelligence
.
H
awkins essentially tells us that both neuroscience and artificial intelligence (AI) research can benefit from a stronger cross-disciplinary approach. He argues that AI work has been flawed by an emphasis on
behavior
. He introduces human brain function, including the '
layer of neural tissue that envelops most of the older parts of the brain
', the neocortex (containing around 30 billion neurons), responsible for most of what we consider intelligence. Indeed, Hawkins tells us '
Your neocortex is reading this book.
' It's arranged in functional regions, like '
an irregular patchwork quilt
', but - and Hawkins tells us that this discovery by Vernon Mountcastle in 1978 is very significant - it's '
remarkably uniform in appearance and structure.
' Hawkins calls Mountcastle's discovery (supported by recent findings on brain plasticity) '
the Rosetta stone of neuroscience
'. He then explains how information flow translates into both spatial and temporal patterns in the brain.
F
ascinating stuff, especially when the book's language waxes lyrical - in describing retinal input through time, Hawkins tells us that '
Vision is more like a song than a painting
', and he shares the notion that '
The world is an ocean of constantly changing patterns that come lapping and crashing into your brain.
' His explanation of the sequential nature of memory, and its involvement in problem solving, makes a lot of sense. Hawkins elaborates on four key factors in neocortex function - storage of sequences of patterns, in '
invariant form
' (how is not understood), in a hierarchy, and recalled '
auto-associatively
'. He then argues in well reasoned detail that takes up most of the book (and may daunt some readers) that '
making predictions is the essence of intelligence.
' Following this is a discussion in context of some of the big questions, such as '
Are animals intelligent?
', '
What is creativity?
' and '
What is consciousness?
' (don't try to read them before taking in the earlier chapters).
H
awkins concludes by discussing the difficulty of predictions, but nevertheless telling us that intelligent machines will be around in a matter of decade(s), that they won't be what fiction has led us to expect, and that they'll surpass human capabilities in areas that take advantage of speed, capacity, replicability, and sensory systems.
On Intelligence
is a significant book, of great interest to anyone who wants to understand what underlies human intelligence, to extrapolate how that knowledge might lead to the development of intelligent machines, and to speculate on the kinds of capabilities such AI will surprise us with.
Note: Opinions expressed in reviews and articles on this site are those of the author(s) and not necessarily those of BookLoons.
Find more NonFiction books on our
Shelves
or in our book
Reviews