Review | In A Thousand Brains, Jeff Hawkins presents his theory of intelligence and what it could mean for AI
- Technologist and neuroscientist Jeff Hawkins suggests suggests the brain works on some sort of universal principle that enables it to adapt itself to any task
- He tiptoes round the edge of the pool of philosophical ideas, preferring to remain in the realm of claims that can be experimentally tested
A Thousand Brains: A New Theory of Intelligence
by Jeff Hawkins
Basic Books
“People often say the brain is the most complicated thing in the universe,” writes neuroscientist Jeff Hawkins. “They conclude from this that there will not be a simple explanation for how it works, or that perhaps we will never understand it.”
In A Thousand Brains his crisp explanations succeed in proving this deduction wrong.
Hawkins found universities reluctant to accept him as a PhD candidate with a project to find a satisfactory theoretical framework into which to put what was already known about the brain and thus explain consciousness and intelligence. So instead he went off to make a fortune in handheld computing, then used it to found his own neuroscience institute.
A Thousand Brains demonstrates a similar roll-up-your-sleeves-and-get-on-with-it approach to research, as well as how a new understanding of intelligence has been reached and what the implications of this might be for the future of the human race.
The key is the neocortex, a thick envelope of cells wrapped around the top of what he calls the old brain and that forms about 70 per cent of the brain’s total volume. This is the location of all forms of perception, analysis and rational behaviour, and is the seat of consciousness and abstract thought such as mathematics and philosophy. Its struggle for dominance with the more basic drives originating in the old brain is an underlying theme of the book.
Hawkins builds on earlier research that showed how the neocortex had grown by making copies of the same basic circuit – the semi-autonomous cortical columns of cells that form its cross-section. Each is connected to a tiny area of the retina, or particular patch of skin, for instance.
One mystery has been how all these nearly identical circuits are processing identical electric signals from our various senses and yet managing to turn them into our perceptions of different sights and sounds, for example. Connect the same cortical region to different senses and entirely different results are obtained.
Crucially this suggests the brain works on some sort of universal principle that enables it to adapt itself to any task – a function differentiating true intelligence from what we currently call artificial intelligence (AI).
How do we define intelligence in Hong Kong?
Just as Alan Turing’s breakthrough was to imagine an all-purpose computing machine that could solve a wide variety of problems, so Hawkins points out that current single-purpose versions of AI need to be superseded. True intelligence – the kind shown by humans – lies in constant adaptation to changing situations and an ability to tackle almost any task.
Since the cortical columns involved in language and other higher-level cognitive activity are essentially the same as those processing information from the senses, these must all involve the same basic function. This is making reference frames – millions of mini-maps of objects perceived, constructed from the motion of our senses around them.
He suggests that thinking is itself a form of moving, and so the same structures can apply the same reference frame system to abstract thought.
Hawkins’ theory answers many puzzles, including questions it would not have occurred to most of us to ask. How does the brain combine information from different senses into unified perception of a single object?
Why do Asians have bigger brains than Europeans or Africans?
He admits the theory is not complete, but claims it so elegantly fits the available data that the details will eventually be filled in. Darwin’s theory of evolution similarly had to wait for an understanding of genes and DNA to explain its mechanism.
His eventual aim is to make truly adaptable intelligent machines that can take on any task. Such machines may one day build colonies on Mars – something of an obsession with tech entrepreneurs, but which he regards as likely an essential part of preserving the human race’s neocortex-generated knowledge.
He tiptoes round the edge of the pool of philosophical ideas, preferring to remain in the realm of claims that can be experimentally tested, even when discussing why we should attempt to contact alien intelligences.
That as far as we know “the only thing in the universe that knows that the universe exists is the three-pound mass of cells floating in our heads” is an idea that clearly bothers him and to which he repeatedly returns. Yet it is he who has explained how the brain takes impressions of what is outside it, much as mud once took the footprints of dinosaurs, and that its self-awareness is simply a matter of biochemistry.
