Organization of Behavior: A Neuropsychological Theory

"Yes, it was a theoretical book, because at that time we knew very little about how the brain worked. But he had two ideas that have become absolutely central to our understanding of how must do so. The first idea was a suggestion as to how the brain learns. When Donald Hebb wrote his book, electrophysiologists and anatomists had shown the following. Brain cell A has a cell body and a long process or axon; and so-called ‘action potentials’ are propagated along this axon. But the axon of cell A doesn’t actually touch cell B ; instead there’s a gap between them. We call this the synaptic gap. The terminal of the axon of cell A influences cell B by releasing packets of chemicals which are taken up at so-called post-synaptic sites on cell B . But what happens during learning? Well, Hebb suggested that there must be changes at the synapse. He further proposed that the more frequently cell B fires at the same time as cell A , the more likely it will fire in future when cell A is active. This idea led to the term the ‘Hebbian synapse’. “David was the only genius of my generation that I have known” That idea was taken up by Giles Brindley, a physiologist from Cambridge who then went to the Institute of Psychiatry in London. And he had a PhD student called David Marr, a mathematician, who joined him there. In his thesis, David Marr produced theories as to how three structures of the brain might work: the cerebellum, the hippocampus, and the neocortex. These papers have had a phenomenal influence. David was the only genius of my generation that I have known. He was also a mentor. He once told me at tea that I should read less and think more. David Marr then went to Cambridge to work with Sydney Brenner and Francis Crick, but he was then poached by Marvin Minsky to go to MIT. While at MIT, David Marr helped found the field of computational neuroscience. Tragically, he died when he was 35. If you take Marr’s theory of the cerebellum, the fundamental idea is that it supports motor learning and that it can do so because of Hebbian synapses. But he died before anybody could succeed in testing the basic prediction which was that there were modifiable synapses in the cerebellum. The Japanese neuroscientist Masao Ito was able to confirm this prediction, but only after David Marr had died. This was ten years after the publication of Marr’s original paper on the cerebellum in 1969. And this was 20 years after the publication of Hebb’s book. Marr then went on to argue that they must be also be modifiable synapses in the hippocampus and the cortex. And it was around that time that a mechanism was found called ‘long term potentiation’. Understanding the chemical mechanisms for this has become fundamental to understanding how learning occurs. “Though he proved neither, these ideas have turned out to be incredibly powerful in understanding the brain” The other idea that Hebb had was what he called ‘cell assemblies’. The idea was that if I see object A , then this assembly of cells fire whereas If I see object B this assembly of cells fire. Many cells in assembly A will also be part of assembly B . So cell assemblies are arranged in a series of Venn diagram. When it became possible to record from individual cells, we started thinking that we can understand how the brain works in terms of the firing of individual cells. But, of course, It’s not that one cell fires when you see something: whole groups fire. So now we realize that the brain works in terms of so-called ‘population coding’. It is whole populations of cells are coding for something. So the reason why Hebb is so critical, is that—though he proved neither—these two ideas have turned out to be incredibly powerful in understanding how the brain actually works. This year the neuroscientist Lucia Vaina—David Marr’s wife—and I have edited a book in which a group of computational neuroscientists take up the original ideas of David Marr. They ask how we think the various operations the brain performs are actually implemented, given the cells that are there and the connections that are there? That must be the final aim of neuroscience. Neuroscience is coming of age. I think that science is phenomenally romantic. Or… I don’t know if ‘romantic’ is the right word. I mean exciting, because any day you might be looking in your data wherever the data is, and you might find things that totally change your mind. You might find something that you didn’t expect in the slightest. People sometimes think that all scientists do is have hypotheses and test them. But a huge amount of it is seeing things that you never expected to see. Astrophysicists are constantly being horrified by new phenomena that they didn’t know of! It’s so exciting. To paraphrase Shakespeare, there are things in heaven and earth we never dreamt of. Poets may dream of them, but scientists have the excitement of finding them. Yes, it’s true that we know very little about the brain. It’s the most complicated thing there is to understand. But at least things are on the move. For many years neuroscience was dull because nothing much was happening. And then suddenly, in the nineties, it took off. Because of brain imaging in particular it’s suddenly become a fast moving field. If I think about philosophy—philosophers have been wondering about dualism ever since Descartes, and they’ve got nowhere. But if you talk to a young graduate student or post-doc in neuroscience today, they’re not interested in what happened a few years ago because things are moving so fast. And then they can find something new tomorrow. Exciting. Yes, I do. Everybody knows that if we use terms, we need to be clear about what they mean. But philosophers worry about things like the fact that we can’t prove that there’s an outside world’— scepticism. They worry that because something has happened regularly in the past doesn’t mean that it is bound to continue to do so in the future – the problem of induction. But there are no solutions. Of course I shouldn’t say that philosophers have made no advances. They’ve made some technical advances. But if I compare that with the rate at which science can advance, then there’s no comparison. If you want to study consciousness, philosophers have got nowhere. They ask questions like ‘how do I know that when I see green, this is what you call green?’ and they’ve gone on worrying about that for an awfully long time, and they probably won’t be able to answer it. But the problem of consciousness, is one that actually is open to empirical investigation, and people are studying it in many ways, including people who’ve worked in my lab. For example, you can give propofol, an anaesthetic, and study what’s happening in the brain as people lose awareness of pain, sounds, and so on. You can look at what happens when people are or are not aware of things that they see. In other words the empirical study of consciousness has moved fairly quickly. The philosophical study of consciousness is static. Peter Medawar, a great biologist, wrote a short book called The Art of the Soluble . His point was that there is an art in finding problems in science that are tractable. I’m not interested in the question of whether I can prove whether the outside world exists. Nobody can. The great thing about science is that it takes all sorts and there are a multitude of problems out there—and they can be solved."