Images of Mind

"Yes, Posner is one of the great psychologists. He was interested in things like reaction times. Let’s suppose I ask you to press a button on the left if a light comes up on the left, and on the right if the light comes up on the right, and to do so as quickly as possible. That’s called a choice reaction. And it takes me roughly 500 milliseconds. But, how long did it take me to make the choice itself? Well, in the 19th century Donders showed that you can compare that time with the time it takes you simply to press a button if a single light comes on, roughly 200 milliseconds. Now subtract the simple reaction time from the choice reaction time and you get an estimate of 300 milliseconds for the time it took you to make your mind up. So you have two conditions, and you subtract what you find for one from what you find for the other. It was on this basis that Posner, working with Marcus Raichle and Steve Petersen, devised a classic experiment using PET. They were interested in the language system. So in one condition they showed a word and the person simply looked at it. In another condition, they showed a word and the person had to repeat it. But they were not interested in vision; so, though they scanned the subjects in both conditions, they then removed everything there was in the scan for the looking condition. What there were left with was a scan that only showed the areas that are involved in repeating. Yes. Then they had a third condition. They showed a noun but rather than repeating it the subject had to say a verb that’s relevant. So, ‘cake,’ or, say, ‘drop’— Yes. It’s up to the person. So now they were interested in how you generate a verb that is associated with a noun. Of course, the subject had said something, but they were not interested in speaking since they already had a scan from when the subject repeated the noun. So they removed everything that was in that scan and what they were left with was a scan that only showed the areas that were involved in generation. And that method’s described in the book by Posner and Raichle that came out in 1994, as a Scientific American publication. It’s a very simple book, but the subtraction method has become fundamental to the analysis of data from functional brain imaging. So you can scan the brain at work and the subjects don’t end up with a dreadful headache and you don’t have to kill anyone to do it! But you can also look at what the different bits of the brain do, and compare those results with what you find when you record the activity of brain cells in animals, or—as is now being done—what you see when you scan animals. People are now also looking at the anatomical connections between brain areas because you can visualise these using scanning methods and compare them in human and animal brains. Yes, this happens so often in science. Crick was a physicist, Watson was a biologist. It’s true that Kahneman and his colleague Tversky were both psychologists, but Tversky was really a mathematician: he published a book of 1000 pages with dense maths. Anyone who’s done science knows that most of the ideas generated are actually generated in discussion with other people, often younger people, and it’s never clear who actually thought of a particular idea in the first place. Take the book that I wrote with Steve Wise, The Neurobiology of the Prefrontal Cortex . We Skyped once a week—he’s in America, I’m here—so we had regular discussions over two years. He knows more about some things than I do and vice versa, I’ve no notion where most of the ideas in the book came from. So I think it’s very relevant that advances are often made when two people with very different backgrounds come together, such as Posner who is a psychologist and Raichle who is a neurologist. Yes, there’s a limitation, you see, of imaging. I ask you to make a decision in the scanner, and using the subtraction technique, I find, let’s say, activity in the prefrontal cortex when you made that decision. The image shows a patch in which the cells are active. The patch is in the order of several millimetres. But there are millions of brain cells in that patch! Though the patch tells me where something’s happening, it doesn’t tell me how the brain cells do it. But the fundamental aim of neuroscience is not to ask where are things happening but how they are happening, that is what the mechanisms are. In recent years methods have been developed to record from individual brain cells in the human brain during surgery. You can do this while the patients are awake because there are no pain fibres in the brain. But there’s a real problem: you can record from cells, or groups of cells, but there are an estimated 86 billion cells in the human brain. So if you can only record from 20 cells, or 200 cells, you’re in real trouble. How are you going to work out how 86 billion cells work? You might think that what you need to do is get a computer, and try to teach it how to do the sorts of things that people do. People at Deep Mind in London are doing just that, for example teaching a computer how to play the game Go. But we want to know how the actual brain works. And Donald Hebb, in 1949, published a book that is fundamental to this enterprise, my fifth book."