Navigating by mental maps

Using vast computing power and evolutionary theory, researchers are gaining insights into the brain's complex circuitry, writes Geoff Watts.

If one bit of the brain is damaged by a stroke, injuring another bit can only make things worse, right? Not necessarily, according to Malcolm Young of Newcastle University's department of psychology. "It may be counterintuitive to believe that damaging a bit of undamaged brain is going to make anything better, but it can. And this is predictable from its connectivity."

That is, from the brain wiring: the circuitry of its component nerve cells. Trying to untangle this network has preoccupied scientists for more than 50 years. Young believes that progress will rely increasingly on what are still relative newcomers to neuroscience: know-how in computing and evolutionary theory.

The complexity is awesome. One of Young's collaborators, Gully Burns of the University of Southern California, cites the vast number of contacts that nerve cells establish with each other. "Consider that there are 10 billion cells in the human brain and that each cell can make up to 10,000 connections. That means that there might be something like a million billion possible interactions between these neurones in the brain."

Colin Blakemore, professor of physiology at Oxford University, notes that it was a similar problem of quantity and complexity that drove biologists working on gene sequencing to seek the help of computer specialists in handling their data. The outcome was a new discipline: bioinformatics. Neuroscientists have now established neuroinformatics.

Even with computers it will never be possible to map the individual connections of the entire brain. But researchers such as Burns and Young are not trying to analyse the circuitry at that level. Our genes do not carry a full wiring blueprint. They cannot: there are only 100,000 of them to store all our inherited information. Biology has side-stepped the problem by relying on what Young calls "simplifying regularities". Particular clumps of brain tissue tend to do similar things and have similar patterns of connectivity: a stereotyping that eases the task of understanding what is going on. "It's still complicated, but it could be 100 billion times worse," Young says.

The realistic goal is to understand the circuitry of the brain at a higher level. Anatomists recognise about 500 regions in a mammalian brain, and each makes contact with up to 30 others. Attempts to trace the layout of this more limited number of linkages have been proceeding for decades and have already generated a prodigious amount of data. In the rat brain alone, Young says, the past two decades have seen a staggering 14,000 reports on the links between its various gross structures.

While in no doubt about the need for computers in neuroscience, Blakemore does point to a limitation in their use. "They can't make value judgements. Anatomy is notorious from that point of view. It involves matters of personal interpretation that are difficult to express in a database: 'Well, maybe this is anterograde transport, maybe it's retrograde - maybe it's just a dirty slide!'"

And even without this limitation, computer-assisted analysis of brain circuitry can go only so far. As Blakemore puts it: "You can't judge what's going on inside the nervous system by some kind of neutral metric of information. You have to match your ideas about the system to the particular needs of the animal."

Visual systems are not, as people have sometimes thought, general purpose processors, Young says. "Evolution has reached right into different species' wiring systems and organised them to do specific things. For example, a nocturnal predator such as a cat makes its living by eating small, fast-moving animals in the dead of night. It doesn't bother with colour vision. But it must have rapid eye-to-claw coordination. Most of the outputs from the cat's visual system go straight to the bits of its brain that make it move.

"If you think of a species like us, a diurnal forager, we evolved colour vision because it helps in daytime jobs such as identifying fruit from foliage. And we don't need such rapid eye-to-claw, or rather eye-to-hand, coordination." In short, brain wiring in different species reflects their different behavioural ecology.

The potential of this approach was recognised last year when the Newcastle researchers were given a Joint Infrastructure Fund award to develop their work on "visual neuro-ecology". This, Young says, links the three elements that are essential to understand any visual system.

"First, a study of the evolutionary selection pressures that forged that system. This gives you an insight into its design principles. Then you also need to study the properties of the stimuli being processed to produce adaptive behaviour. If you're a kestrel, for example, it's handy to have ultraviolet vision because your prey, rodents, excrete something that shows up in UV light. The third component, the 'neuro' bit, is trying to understand how neural systems actually mediate vision. If you do all three things together, you've got the most thorough description possible of a visual system."

Work of this kind has practical applications. Keyhole surgery is tricky because doctors have to guide their instruments without a full range of visual clues. Young thinks that with the new insights into vision it should be possible to devise a way of restoring a surgeon's sense of being within the space he is seeing - "difficult but practicable", he says.

And then there is the example mentioned at the beginning: overcoming damage by doing more of it. The brain has two systems for directing attention to various parts of the visual field: one in the cortex, the other beneath it. A stroke in, say, the right cortical region would prevent you attending to the left half of your visual field. It is well known that inflicting equivalent damage on the hitherto intact side can restore the balance - and so the lost function.

Counterintuitive or not, this outcome does make sense - but only when you have some grasp of the wiring of what is, by any measurement, the most complicated information processing system imaginable.