Book: How to Create a Mind (Author: Ray Kurzweil)

Will we ever understand how our brains work?

11:00 9 November 2012

Laura Spinney, contributor

Several projects are trying to reverse-engineer the brain. In How to Create a Mind, futurist Ray Kurzweil champions their cause

WHEN it comes to the human brain, many scientists believe that we are incapable of understanding how it works because we lack the tools and intelligence to measure its mind-blowing complexity. Others are starting to question that notion, and to subtly redefine the task. In How to Create a Mind, futurist Ray Kurzweil has ridden into battle for the challengers.

He starts by asking what it means to be complex. If you consider that a forest is made up of trees, each of whose branches is different, then you might conclude that the forest's complexity is impenetrable. But if you realise that the forest grows according to certain rules and contains repeating patterns, then the problem becomes tractable. You don't have to measure every last gnarly twig; you can make predictions instead. As Kurzweil says, you can reverse-engineer it.

All you need to get started are the rules, and an understanding of the basic neural building blocks whose configuration they guide. Neuroscientists noticed long ago that the structure of the brain's outer layer, the neocortex, is remarkably uniform, made up of repeating units, like transistors in a microchip. Estimates of what this basic unit might be vary - from a single neuron up to tens of thousands - but all are relevant as they are likely to fit into a hierarchy whose functions become increasingly complex and abstract as you move up it.

There are now a handful of efforts afoot to describe both the units and the rules, including theHuman Brain Project at the Swiss Federal Institute of Technology in Lausanne, in which, in the interests of full disclosure, I should say my husband is involved. What they have in common is the idea that the brain is an integrated system, with properties that emerge only when its elements are combined.

This means that traditional neuroscientific research, which seeks to explain increasingly specialised aspects of brain function in isolation, may have limited scope. The reductionist approach has to be supplemented by a constructivist one - putting the pieces back together again to explain the whole. Modern tools, including supercomputers and the mathematics of complexity, make that possible.

You can see why, on learning about this new line of research, Kurzweil felt he had to write this book. For years he has been talking about what he calls the law of accelerating returns, according to which both biological and technological evolution are speeding up. In his 2005 book, The Singularity is Near, he argued that we are approaching a point where humans and machines will merge, producing a leap in intelligence. Reverse-engineering the human brain could open the door to all sorts of significant innovations, such as the design of a computer that thinks more like us. This could be the springboard from which to make that leap.

There is good evidence that neither biological nor technological evolution happens in the smoothly exponential way he says they do, and even that the blistering pace of technological progress of the last few centuries might be slowing. On the whole, though, this book is a breath of fresh air. Scientists tend to be cautious in their predictions - they certainly were about the Human Genome Project, which was completed in 15 years against all expectations - and Kurzweil makes an argument for optimism.

There are plenty of challenges ahead, not least because reverse-engineering the brain is only the first step. The result of that process will then have to forward-engineer itself - that is, learn. There are ethical concerns about what this new intelligence will be capable of, but for now at least they seem to be dwarfed by its promise, not least in helping us to understand and treat the rising tide of brain disease.

It is debatable whether this will happen as early as the 2020s, as Kurzweil predicts. But if, as seems increasingly clear, this is where the science must go, does it matter if he is off by a decade or two?