Making Sense? Neuroscience and Behaviour
Dr Brian McCabe
20th October, 2013
Good evening. My name is Brian McCabe. I first set foot in this chapel before many of you were born and have distinct memories of events here going back decades. I spend much of my time trying to understand how I'm able to store information and recall it after all these years. The ability to remember things is easily taken for granted but is nevertheless truly remarkable. It is imperfectly understood; and there is every reason to suppose that a scientific approach, the tried, tested and self-correcting way of discovering things about the natural world, will deepen our understanding.
It's always a good idea to try to say why one is doing something. As far as I'm concerned, curiosity is a strong motivation. Moreover, I am as fond of a challenge as the next person. There is also a widespread feeling, which I share, that it is important to understand how the brain works: after all, it's an organ that distinguishes our species and which has been the determinant of human achievement. So, 'How does the brain work?' is a big question. It has been the experience of many scientists, that having such big questions in mind is a good thing, even though one may at first have no idea whatsoever how to answer them. Because then, if the opportunity arises, one may at least be in a position to recognise important data that could lead to a major discovery. Study of the brain also has practical importance. Some of the cruellest diseases are those affecting the nervous system. Moreover, these diseases can be very difficult to treat. Attempts to cure neurological disorders are not going to be particularly successful unless we have a better understanding of how nervous systems work. Despite all of these considerations, one should recognise that science can be double-edged: increased knowledge of how the brain works may make it easier to do bad things as well as good. I believe that the benefits outweigh the risks, but I'd be the first to admit that this is an act of faith.
In order to make headway, one needs to choose the right object of study. I work mainly on the domestic chick which, as well as being in plentiful supply, learns rapidly, particularly through imprinting. Imprinting is a process whereby the young animal learns the characteristics of its mother and demonstrates this by following her in preference to other individuals. Chicks can become imprinted to all sorts of things in the laboratory, including me for example, so we have a very powerful and experimentally tractable form of learning, which resembles processes seen in many animals, including humans. In fact, the chick has a large number of advantages for studying learning and memory. One of these is that its previous experience is very limited, so that any change that is specifically related to learning is more likely to stand out. What we learn during this service will undoubtedly be encoded as changes in our brains but these are likely to be tiny in comparison with all the changes that have accumulated from experiences throughout our lives. A newly-hatched chick, in contrast, has had very little previous experience of anything and therefore learning-related changes in its brain are likely to be easier to detect.
It's useful, as a working hypothesis, to think of the brain as a machine that can remember things. Without going into too much detail, my colleagues and I have found a region in the brain of the chick that has the properties of a memory system. Various biochemical changes were found in this region, which appeared to occur only when learning had occurred (we could measure this behaviourally). The change was not found with the same behaviour and
sensory processes in the absence of learning. I'll describe one way in which we followed up these findings. There is a brain region suspected of storing information. If it is doing so, one might predict that nerve cells in this region would change from being relatively unresponsive to a stimulus in the naive animal to a relatively responsive state after the animal has become imprinted to that same stimulus. So one does an experiment. Taking care not to hurt the animal, under anaesthetic, one implants tiny electrodes - fine wires - to record the electrical activity of individual nerve cells (a procedure, incidentally, licensed and very tightly controlled by the Home Office in the interests of animal welfare). The chick is then allowed to recover from the anaesthetic and one determines whether the nerve cells respond to one or more stimuli (we use various sorts of flashing lights). Typically, the nerve cells show little interest in the stimuli before the chick becomes imprinted. The chick is then exposed to one of the stimuli, the imprinting stimulus, for two hours. This is plenty of time to become strongly imprinted. We again measure the nerve cells' response to the stimuli. Typically, it's just the response to the imprinting stimulus that is changed: the cells become 'tuned' to that, by now familiar, stimulus. One can also label nerve cells that have been active over the previous hour or so. In this case one finds that, just in the brain region in question, the greater the strength of learning, the more cells become labelled. With this technique, it is also possible to determine whether the effect occurs in cells having a particular chemical signature. So we now have more evidence that we are dealing with a memory system and have come to know a good deal about how it behaves.
Science is very useful to us and it might not be terribly surprising if something similar has encouraged the evolution of at least some organisms. A life scientist need never be idle (this is your Director of Studies speaking) because we are always carrying around a laboratory - namely ourselves. What is more, we don't even need ethical approval. Could it be that something similar to the scientific method is employed by our nervous systems? When attempting a complex action for the first time, say throwing a dart at a dartboard, the result is likely to be pretty poor. But with practice it gets better until it seems relatively effortless and quite accurate. The first time, our motor system evidently uses some sort of hypothesis to plan and execute an action; call that action an experiment. During and after execution of the action, we receive sensory information about our environment, the results of the action and how accurate the action was: information from vision, skin sense organs, stretch receptors in muscle, little strain gages in tendons and so on; call that data. These data, presumably, are used to modify the hypothesis because the next time we try the action it is usually more accurate. And so on, until our action is as good as it is going to get. Our hypothesis has been modified by information about the world and the consequences of our actions. This information is limited by the quality of the sensory processes that have mediated it. Get a better pair of spectacles and the hypothesis can be refined according to the improved information that is now being received. The hypothesis may depend on a crude, highly censored representation of the world but is perfectly good enough for very complex behaviour. There are, I think, several interesting aspects of this process. First, it happens automatically, at least ultimately when we can perform the action relatively effortlessly. Second, information representing the outside world is encoded in what I suspect is quite an economical way. An enormous amount of data comes in at any one time and this is likely to be quite drastically filtered, extracting only those pieces of information that are useful. Finally, actions such as the one I described, in the general run of things, are useful in their own right, but at the same time may also be considered as experiments. This is a very efficient use of resources, from which we might learn something useful about how to organise systems that we employ to do things for us, such as robotic systems, for example.
A final consideration is the question of models. Scientists build models all the time. This is quite challenging in the case of many biological systems because of their complexity and the paucity of information available to identify the model that fits the data best. Nevertheless, as I speak there are supercomputers around the world chuntering away to make models of the brain. It should be said that there is considerable scepticism as to whether they will be successful. Who knows? It seems likely to me, anyway, that they will be of some use in revealing hitherto undiscovered aspects of nervous systems. But if a satisfactory model is to be devised, there is a major problem in finding the level of organisation at which such a model can be understood. With 100 billion nerve cells in the human brain (give or take), many with over 10,000 inputs, that, to put it mildly, is a challenge. Which brings me to issues at the heart of all scientific endeavours. Our ideas about the natural world are formulated in terms of the information we are able to collect. From these ideas arise models which, provided they are testable, can be proxies for the real thing until more suitable models turn up. It seems obvious but it may be worth saying that, if you are going to construct a model, stick like glue to the data. And for heaven’s sake, keep it simple.
References for those who are interested
Ramachandran, VS (2003) The Emerging Mind: The BBC Reith Lectures 2003, Profile Books, ISBN-10: 1861973039, ISBN-13: 978-1861973030.