Complexity and the brain

Of all the sources of inspiration that computer scientists have drawn on, none can compare with the brain. It has long been a goal of those working in computer science to endow a machine with attributes resembling human intelligence, yet up to now all attempts have failed miserably. For while conventional computers are very good at performing tasks most people find difficult, such as arithmetic and algebra, people routinely deploy such skills as seeing and talking that even the most powerful machines cannot match.

But by mimicking the architecture of the brain the key to its emergent properties such as intelligence and consciousness neural network computers have managed to learn and adapt to their experiences of the wider world. Nature's methods of adaptation and optimisation, which refine the design of organisms through evolution, are now enacted by computers using genetic algorithms to tackle intractable problems. Like nature, such evolutionary computer programming techniques which are so powerful, novel, and successful at dealing with problems of great complexity - all have random elements within them. This randomness leads to innovation - the discovery of smart and unexpected solutions to very hard problems.

The symbiosis between science and the computer is making it feasible to begin to understand and simulate some of the human brain's remarkable capabilities. This organ comprises a million million cells, among which are 100 billion nerve cells - the very stuff of thoughts, emotions and the mind. The latter figure rivals the number of stars in the Milky Way. Yet, for the first time, scientists are now producing plausible models of certain aspects of brain function using artificial neural networks. Some models reproduce the rhythms of the brain, others aspects of its organisation. In so doing, they are lifting a corner of the veil that, since antiquity, has divided mind from matter.