 Come now, vast God, to this high occasion, to this Nobel conference that praises its scholars, anoints its students, and crowns its honored guests. We praise you for the courage of lonely explorers into the realms of fact, and for the collegial fellowship which advances all scientific discovery. Save those whose life is spent in research from irresponsibility for the end products of their work, and make them sensitive to the whole meaning of what they do. Forbid that our cleverness should thrust us toward disaster, or our earnest experiments deliver humankind to unforeseen tragedy. Seeing our work as our calling, may we make it serve the vast intention of your love. Amen. On behalf of the faculty and staff of Gustavus Adolphus College, it's my pleasure once again to welcome you to the Nobel conference. This is Nobel 20. We are pleased that so many of you could join us today. I know that some are still looking for seats, and I think you will find them by coming up the side aisles or looking in the balcony. I know it's easier for me to see them than for you to see them, but there are still seats available. Our reservations for tickets indicate that we have with us today about 4,000 guests on campus. You come to us from a number of places. You represent nearly 160 high schools in our area. You represent 75 colleges and universities, and we have registrations for this conference from 12 states in the Midwest area. We meet today as we have in past years in what we have come to know here at Gustavus as Laundorina. Laundorina is part of a larger complex which is just now being completed, and it will be dedicated on the 21st of October. We will name it the Laund Center. We invite you to look around this facility as time permits. You will find some areas locked or blocked off as construction continues in preparation for the dedication, but you are welcome to visit any of the areas that are open. Events like this, of course, do not just happen by some spontaneous process. They are the result of a great deal of work and thought by many people. I can't begin to recognize the literally hundreds of people who contribute to the success of the Nobel conference year after year, but I want to identify just a few. In a little bit, you will hear from Professor Mike Shaftow, who has chaired the committee for Nobel 20. We thank Professor Shaftow and his committee. A special acknowledgement to Chaplain Richard Elvie, who brought us the invocation and who is the program director for the Nobel conference. Special recognition is also due to those who work in the area of public events at Gustavus, Elaine Brostrom, Dennis Paschke, and all of their staff. Dozens of other people in the food service and the grounds crew, members of the faculty, all have participated. And of course we want to express our very special appreciation to the Lund family, and I know Patricia is here. I think I saw her walk in just a bit ago. Patricia, to you and the other members of your family, the Lund family, who have so generously provided for this conference and all the other things that you have done for this college. I hope that you received a program as you came in, a program with this face on it, which should be familiar to you by the end of the conference. You'll see it in many places. There are lots of events and lots of activities listed in that program, which go beyond the lectures themselves. These include the concert this evening in Christchapel, the art show in the art gallery, the firing line programs, which will be conducted this evening at 8.30, and other events as well. We invite you as guests to look around the facilities, particularly Bernadotte Library, which is just to the south of this complex. You're invited to stop in there as time permits in Nobel Hall and other places. I would like to just identify the members of our panel this year at this time. I am going to ask them to stand. I will simply introduce them by name. At the end of that time, we can greet them with a round of applause. They will be introduced more fully as their opportunity comes to present their lectures. But at this time, I would like to introduce our six guests, Daniel Dennett, Gerald Edelman, Brenda Milner, Arthur Peacock, Roger Schenck, and Herbert Simon. Thank you for being with us. We have come together these days to give attention to some of the issues related to human cognition and also to consider the ways in which our cognitive processes may relate to machines, especially to the computer. We will be discussing these topics during the next two days, and I am now going to make what will be probably my total contribution to the thought on this important topic. As I share with you an anecdote that I heard just a few days ago, one of the issues that is of great interest to a number of people has to do with the language capability of man and how this may relate to computer simulation and the like. The story goes that a young man very interested in language and in computers developed a program which he believed would translate English into Russian. He brought it to one of the major software companies and presented it to them. They were interested. They said we will try it out. Our test here is to translate from English to Russian, and then we translate back from Russian to English to see if we have the same statement that we started with. They took as their test phrase the old adage, the spirit is willing but the flesh is weak. They put it into the computer. It was translated into Russian. It was then translated back to English, and they read the following output. The wine is all right, but the meat is under cooked. I don't know, Mike Shaftow, what you can do with that, but I'm sure you can do something to recover the integrity of this conference. As I call upon Professor Mike Shaftow, the Professor of Psychology at Gustavus and currently with the Office of Naval Research in Washington, who has directed the committee who brought this conference together, Professor Shaftow. Welcome to the 20th Annual Nobel Conference, how we know the inner frontiers of cognitive science. I have just one special announcement to make. We have some excellent displays of educational technology and artificial intelligence that will be open to the public today and tomorrow. These displays are in Alumni Hall upstairs in the old Student Union building. If you don't know where that is, it will be worth the effort to ask someone. I invite you to go up and look at these displays at your leisure any time today or tomorrow. I want to at this time express my thanks to Richard Fuller of the Physics Department and Mark Krueger of the Psychology Department, as well as to Mr. Arnold Reiden from the Gustavus Board of Trustees for their initiative in bringing these displays to campus for Nobel. Now I'll turn the microphone over to my longtime friend and colleague and great recruiter of talented students into the Neurosciences, Tim Robinson, who will introduce our first speaker. It's a great pleasure to introduce Dr. Gerald Edelman. Since the details of Dr. Edelman's existence are documented quite well in the conference program, I restrict my remarks to a brief description of his work as a researcher and theoretician. Dr. Edelman did his undergraduate work at our Senus College in Pennsylvania and later obtained a degree in medicine at the University of Pennsylvania. After practicing medicine just briefly in the United States Army, he returned to graduate studies at Rockefeller University where he worked in the laboratory of Henry Kunkel. In 1960, he received the PhD degree and set up his own laboratory at Rockefeller where he's worked ever since. Now these were exciting times for molecular biologists. The structure of the DNA molecule had just been discovered only a few years before and researchers were just beginning to realize that knowledge of the chemical structure of significant biological substances was indeed possible. Dr. Edelman sought to understand the chemical structure of the immunoglobulin molecule. He began with the assumption that this giant molecule, like many others, is composed of two or more chain structures held together by crosslinks. By developing a procedure which allowed him to sever these crosslinks, he was able to show that the antibody molecule was indeed composed of one pair of light chains and another pair of heavy chains. He and Rodney Porter of Oxford University were awarded the Nobel Prize for this discovery in 1972. It's interesting to note that Dr. Edelman is the only person that I know who was awarded the Nobel Prize for his doctoral dissertation. He also received this award at the relatively tender age of 43. In the Nobel Prize citation, he and Porter were credited with having introduced the critical procedures which opened the gates to broader understanding of the molecular basis of the immune system. Now, at this point, some of you may be wondering how it is that a Nobel Prize winning molecular biologist working in the field of immunology came to be invited to address a conference dealing with the topic of learning. As it turns out, Dr. Edelman has been interested in the workings of the brain for many years. He hinted at this interest when he, along with 22 other Nobel Prize laureates, attended the 1975 Nobel Conference entitled The Future of Science. One of the keynote speakers that year was the veteran brain researcher, Sir John Eccles, who described the last, the attempt to discover the working of the human brain as one of the last remaining frontiers of science. In summarizing his many years of research on the topic, he finally concluded by saying that in order to understand some functions of the human brain, that it was actually necessary to posit the existence of a metaphysical force. Needless to say, this using a so-called ghost in the machine to explain higher brain functions surprised many of his fellow Nobel laureates. One of the most interesting responses to this talk came from Dr. Edelman, who presented an analogy from the field of immunology. He pointed out that in many respects, the immune system could be mistaken for a cognitive or thinking system, since it seems to possess a memory and has the ability to recognize foreign bodies, but that in fact it works according to the principle of selection and that its apparent complexity was perfectly explicable in terms of normal evolutionary processes. He concluded by suggesting that this principle of selection may possibly be operating in the brain as well. Two years later, a small book written by Dr. Edelman in collaboration with Vernon Mound Castle called The Mindful Brain appeared, in which he clearly described his vision of the brain as what he terms a selective machine. This vision of the brain as an organ of almost immeasurable complexity, which nonetheless operates according to the principles which have developed over the course of our evolutionary history will be the subject of his talk today. Let's welcome Dr. Gerald Edelman. Thank you, Dr. Robinson. President Kendall, Reverend LV, Professor Shaftow, colleagues, distinguished guests, ladies and gentlemen. It is a particularly happy occasion for me to return to Gustavus Adolphus, and I remember the occasion alluded to with great pleasure, and it's a particular pleasure indeed to have the opportunity to talk amongst these distinguished colleagues about a very interesting subject. Indeed, my remarks shall be about the greatest inner frontier of present-day science, the nature of the human brain. What I want to do is convey to you some of the excitement those of us privileged to work in this field feel about it, but I should hasten to say that this is not going to be a scientific lecture, and indeed it cannot be. Given the pressures of the occasion, the complexity of the subject, the vastness of the audience, the smallness of the speaker, I think we have to enter into some convenient arrangements together, a kind of pact. And rather than detailing the legal terms of this pact, what I would like to do is exemplify what I have in mind by telling you a story of two Jewish tourists from New York who went to Israel for the first time, and on their first night of their sojourn decided they would go to a nightclub in Tel Aviv, and they heard a stand-up comedian telling one-liners in Hebrew, and one of the tourists fell off the chair laughing, and the other looked down and said, what are you laughing at? You don't even understand Hebrew. And he looked up and said, I trust these people. Here is what I hope to do. What I would like to do first is to illustrate the difference between what I shall call the physical order, that which scientists describe, and what we might call the sensory order, that necessary but not sufficient step that scientists need in order to interact with the world. Then because it's particularly germane to some of my latter remarks, I want to talk a bit about what is called in biology population thinking, and to contrast it briefly with previous modes of thinking and previous views of biology and of our nature. And then what I'd like to do is discuss two views of the brain that I show for brevity called the information processing model, and certainly the most prevalent model I think in both neuroscience and in cognitive science today. And then what I'd like to do is discuss the rather new population model and consider some of the evidence that tends to support it. And finally, if time permits, what I would like to do, having been enjoined by the organizers of this symposium to speculate or extend my remarks into domains in which I'm not particularly expert, I'd like to see what the consequences of the assumption that the population model is correct might be. And of course that is what makes this not a scientific lecture. If it were a scientific lecture, I would in fact search for ways of showing that the model is incorrect. And so that is my program, and I hope you'll bear with me because it has many complex parts, but I have the privilege of being followed by both eloquent and informed colleagues. You're going to hear from my fellow speakers about a variety of most extraordinary psychological subjects. Information, learning, memory, intentions, and intentionality. What I'd like to do today, and perhaps it is not a miss since it is prior to all of these subjects in some sense, is to consider a subject closer to the immediate confrontation of the brain and the environment. Namely, that way or the principles in which the brain is organized to carry out perception, the awareness of things that are present to sense, and the kinds of questions I want to consider are, how are sensory discriminations made? How in fact is past sensory experience represented in the brain? And how does generalization occur upon a meager sensory experience? Well, many of the things you're going to hear about by subsequent speakers will probably take perception for granted. But it is not a trivial subject, I hope, to persuade you. Indeed, is it in many ways a much more challenging problem than the problem of learning itself, hard as that is. I also perhaps ought to emphasize right at the outset that it's entirely feasible to do much of psychology without any concern whatsoever for brain mechanisms. But ultimately, if we're to avoid error, we must confront the thorny problem of how the brain works. Otherwise, I'm afraid, we're going to be lost in refinements of language but not necessarily in refinements of perceived facts. So, I don't see a pointer here, and if in lieu of a pointer I wave my hands around, you'll get the idea. To illustrate that possible risk, I might have the first slide. I see nobody on the road, said Alice. I only wish I had such eyes, the king remarked in a fretful tone, to be able to see nobody, and at that distance too. That, in fact, prefigures the predicament that some of us have when we try to consider the matters of sensory awareness and perception. To turn seriously for a moment to the first problem, namely the difference between the physical order and the sensory order, let me begin by pointing out the difference in an example between the world described by science and the world of perception, which requires looking at objects that is not necessarily veridical and depends enormously upon the context in which they are perceived. May I have the next slide, please? Well, if you were asked whether the lines in these figures were parallel, I think most of you would agree that the lines that are in the top figure of this so-called vunt herring illusion were curved inwards and that the lines below were curved outwards. And if I asked you how would you be absolutely sure about that, the scientifically-minded amongst you would probably measure the distance at right angles between the two lines and come up with the fact that, in fact, the lines are parallel. This is a rather trivial but I think telling example of the problem we face in confronting the sensory order. Everything that we see is not everything that is the case, and yet we need that information in order to construct a valid scientific description. I wanted to start with this example because I think it's extremely important to understand that what is common sense is not necessarily the way in which the sensory order works and indeed there are probably very good evolutionary reasons for this. But my main concern here is not with that subject of which this is a part known as psychophysics. Instead it is to try to develop a view of the brain in terms of its organization and to confront how, in fact, we can perceive such objects and particularly how upon presentation of a few examples of them we can generalize to a very large number of cases. Of course that will take for granted that there are elements of learning in the system. Perhaps in order to provide a basis for my further remarks, I should deviate at this point and tell you a little bit about population thinking because it will be necessary to our argument and I'm not sure all of you are familiar with that. That is the great development for which may I have the next slide. Darwin and his colleague across the seas Wallace was familiar and here is Darwin in his legubrious latter years. The central theory of biology founded by Darwin and Wallace over a hundred years ago and I think most people would admit ideologically the most significant scientific theory ever constructed. Before Darwin, thinking about the origin of biological order was under the sway of an idea that has been variously called the great chain of being, the scholar Naturi or for short, essentialism. Since Plato, nature was assumed to consist of classes or taxa defined by properties from the top down, fixed and in plenitude and in this view individual variation was a noisy inconvenience to be ignored or it was assumed to be a symptom of the fallibility of our earthly life and in any case the origin of the species was assumed by definition. Darwin's great contribution in creating population thinking was to understand that individuality was of the essence, that variance in a population was real and not just noise, indeed the basis for change and it was this basis upon which natural selection acted through the environment to select those individuals who whose adaptations were on the average greater, thus leading to their differential reproduction and just for my colleague up the booth I'm going to switch down here and you might take the slide off, thank you. I'd have the lights for a moment. Clearly this is a course grade intelligence test for speakers. Here on this first transparency I've tried to illustrate to you the basic idea of population thinking refurbished of course in modern terms. The origin of the variation within a population is mutations in genes in the genetic material and eventually through a most complex process this leads to a functioning animal form but with variation constituting over a whole group of consorting animals, a species or a population. Natural selection will then in fact pick certain individuals, the ones that are uncrossed for differential reproduction. It does not mean survival of the fittest, it means average survival of the most adapted over some period of time. And this is the basic notion of population thinking. I dare say the central idea in biology. What can we conclude from this? What we can conclude is that all of Darwin's presuppositions except for his genetics were correct. Variation within the population is not informed as to outcome, it is by chance. The environment is remorselessly independent and on the average the most adapted will survive. It is not a pleasant thought for certain benefactors of modern science that the less fit must eventually die. These are the basic premises of the theory of natural selection. Now I want to turn from the consideration of this global indeed all-encompassing theory of biology to a very specific example. The immune system. To indicate to you that while the evolutionary system works over aions of time over large numbers of years, selective systems can operate in somatic time, that is within an organism and during its lifetime. And the best founded modern example we have of that is a field as you've heard that I used to be in called immunology and the basic point I wish to make is that the prevalent theory in that field was quite different 20 years ago than the theory that is in fact I think everybody will agree proven today. Let me say a bit about what the immune system does. The immune system is a system in your body represented by molecules and cells in your blood capable of telling the difference between self and not self at the molecular level. It is clearly a non-cognitive system despite the attempts of certain Russian biologists to prove that it is fundamentally influenced by the brain. But it is a system of exquisite specificity. To give you a feeling for that specificity the immune system can recognize the difference in two huge protein molecules of one carbon chain just tilted a few angles away and tell it from all other things in a positively naming sense. Now how can that be? Given all the different compounds that organic chemists can construct that certainly never existed before in the evolution of the species how can it be that your body can positively distinguish self from not self in this discriminatory fashion. The theory that prevailed before so much modern knowledge accumulated had as its greatest proponent Linus Pauli and it was the theory of instruction in the immune system and it assumed that the foreign molecule shown here by a little benzene ring this ring here with some nitro group sticking on it the foreign molecule transferred information about its structure to a cavity in the antibody molecule the recognizing molecule and then removed itself much as you might make a cookie with a cookie cutter if you will to give the reciprocal image and that that folded crevice was in fact the informed recognizing site which would then recognize all further instances of this molecule. You can see why it's called the theory of instruction information was transferred about three-dimensional structure from the molecule to be recognized to that which would recognize it. Now that theory has been displaced this is not the occasion for me to say why and how instead what I'd like to do is tell you about the theory that now prevails of course all theories in science are pro tem but the evidence is overwhelming that in principle this theory of selection is correct and the idea is quite consensical it says that prior to confrontation with any foreign molecule your body has the capability of making a huge repertoire of different antibody molecules with different shapes at their binding sites are priori and then when the foreign molecule shown by this little benzene ring here comes in it pulls that repertoire and when it finds a shape that fits more or less well it amplifies that recognition by stimulating those particular cells for example number two, number five and number seven to divide and reproduce. They reproduce to form the asexual progeny of a single cell which is known as a clone and therefore the theory which is first formulated by John Burnett is called clonal selection. Now you can see very interesting properties of this system in the first place there's more than one way given the a priori nature of the system of recognizing a particular thing above any criteria threshold. The second thing you notice about it is it has the potential for memory consider for example that this particular group of cells stops here in this branch and the rest go on to some end producing a kind that would recognize the original antigen but now I have very many more of the original stem cells because they've divided. That can constitute for the lifetime and overlapping lifetimes of different recognition cells a memory and indeed is a crude description of what is now known as immunological memory. A memory so staggering that if you're exposed to yellow fever for example you will still be immune to it at age 80 if you were exposed to age 10. Well the main thing about this that I want to convey is that the theory of instruction turned out to be wrong that in the biological example of a non-cognitive recognizing system with exquisite specificity it turned out that the principle was one of somatic selection upon a huge set of variant cells which of course had fiddled their DNA their gene gene material in such a way as to make different kinds of locks if you will that are going to receive the keys of foreign molecules and in ignorance of what they were to confront. Now perfectly clearly you see the consequence of that assumption that the fit is not perfect but nonetheless the system by a series of elaborate feedbacks can accomplish as exquisite a recognition as you please. And so we come to this position that in evolution essentialism turns out to be wrong. Taxonomic classes as Darwin so beautifully pointed out are defined from the bottom up through selection upon variance. That is an over-evolutionary time. In immunity in backboneed animals instruction is wrong. It turns out that selection is from a repertoire of specialized cells that make huge numbers of variants. Indeed you can have as many as 10 to the 11th of these cells, different kinds and countless trillions of possibilities for these a priori antibody combining sites. And now the question before us is this. In its fundamental operations not at the level of information processing and language in its fundamental operations closer to the evolution of cells and cell groups themselves is it the fact that the brain is constituted according to a theory that could be described by a theory of instruction in information processes or does the brain operate by selection? Now in order to get at this problem I have to deviate again but let me summarize what I've said so far. What I've said so far is in two major examples in which an extraordinary adaptation takes place. So refined in both cases that the initial impression was one of instruction it has turned out that instruction and typological thinking and essentialism is incorrect. The second thing I want to point out is that both selective systems are similar but you must not make the mistake that their mechanisms are the same. The things they have in common is that you must have a very wide repertoire of variants a priori. You must have some effective possibility of scanning those variants and you must have a very high gain amplification of the selected examples that happen to fit and the consequence in a finite population is that some of the others must be suppressed or died. Well let's take this up with respect to the brain but before I do that I have to say a little bit about the brain. Dr. Melner will speak in much more sophisticated terms. Given my project I would like to go over some elemental features and you will hear from her about the elegant studies that have been done relating the structure of the brain to that fundamental process of memory. Right now what I want to point out is for your memory itself of this subject that the brain really is composed of globs and slabs. These globs and slabs are connected to sense receptors, my eyes, ears, etc. In my case, in my eyes, not recommendable and in muscles which must not be underestimated. In fact they very commonly are. And I want to say something about the order of this brain but not get into much tedious detail. The great problem of course is much of the assumption that psychology can proceed without a description of this kind can lead to asking one question to many. One is of the doctor who is accused of malpractice and of dispatching a patient and the brain of the alleged decedent is put in a jar as exhibit A. And the zealous lawyer for the defense of this hapless doctor is questioning another doctor on the stand and he said now Dr. Brown did you see the alleged decedent Mr. Smith in your practice and he said no sir. He said well did you carry out a post mortem or were you present at a necropsy? No sir. I want you to declare a mis-trial. I want you to declare a mis-trial. There is not even any evidence that the alleged decedent is alive or dead. And the doctor looked at the lawyer and he said I cannot assert whether he is alive or dead. All I can say is that that is his brain in the jar over there. For all I know he might be out practicing law someplace. Before I turn to some concrete examples of how the brain is constituted microscopically of cells. For example in the cerebral cortex alone the organ of which you shall hear from Dr. Milner in the size of a very generous table napkin you have ten billion neurons with one million billion connections. If you count one per second you'll finish 32 million years later. These neurons or cells are connected in a very peculiar arrangement. They have processes and connect process to cell body process to cell body in a structure first named by the great physiologist Sherrington the synapse. And we now know at that synapse as a result of electrical activity in the neuron a chemical transmitter of a varied kind is released stimulating the next cell and so on down the chain. So we have an electrochemical interaction going across a structure. Now the point is what is the structure? The temptation is to assume that the structure is a very orderly thing as indeed any neurologist who makes a diagnosis will tell you it is. But I assure you the evidence to the contrary notwithstanding it is not a heath kit. I want to show you the counter examples of variants in the nervous system. Now it is a very important thing to understand in selective theories that you can't talk about variants until you talk about remembered commonality or common structure. That of course is clear in evolution. If it is all mutation it can't be any selection. If it is all selection and no mutation the system can't go on either. And so I have to show you some examples which I hope my colleagues who are more neurologically adept and I will not take a miss. They are not to say that the nervous system is highly ordered in the sense that your faces are highly ordered. It is to say in addition it is highly variant. And so I would like to show you some slides of that. And here now I guess that isn't going to work. Maybe could I have the next slide please? Here is a picture of the brain of an owl monkey. It looks extraordinarily faded. I see why. The course great intelligence test. There. This is a picture, a cartoon if you will, of the brain of an owl monkey. And what I want to point out to you is that not as corrugated as ours fortunately for researchers. That this structure is organized into zones or regions which have been well known. And I don't want to go into the tedium of that just to point out that I'm going to discuss one of them a little later on. For example that one up there called 3B and that one over here called 1. Those intricate maps over there. In different owl monkeys you'll see by using electrode to look for these electrical activity of which I spoke and stimulating a hand for example that it is mapped in a curious way finger by finger and zone by zone. I'm going to talk about that. And other portions of the cortex are mapped to deal with the primary receptors for vision, for hearing and what have you. And there are indeed as you will hear much more elaborate centers. So I want to introduce you to the idea that it is not just a plethora of neurons but there is in fact an organization in terms of mapping the world, the three or four dimensional world through a two-dimensional sheet onto this cortex. On the next slide you will see some examples of the variants. If I would zoom down and look by cutting through this cortex at the kinds of cells you will see the first thing I will notice is there are at least six layers and in these layers starting from the top that is the outside surface going down towards the center of the brain I will see an extraordinary host of so-called neurons nerve cells of different shapes, connectivities and chemical responses. So there is an enormous variety of types of neurons. A matter of some controversy in fact. Some people are lumpers and say there are really only two types excitatory and inhibitory others like the great Ramani Kahal who first described these things in detail would argue that there were really extraordinary numbers at least 30 or 40 different types and perhaps even more in terms of just sheer morphology. The second point I want to make is that there are not only types. Could I have the next slide but that amongst one particular type of nerve cell, could I have the next one? There is an extraordinary variance. Never mind the details but this happens to be so-called contralateral descending movement detector of the locust. It is a neuron this one descending here it's a neuron which is going to do something to an extensive muscle just before the locust takes flight and brings on a plague of locusts. If you look at four different locusts you will see an extraordinary variance. In fact amongst 80% of a randomly chosen population despite the fact that their behaviors are more or less like you will see this kind of variance in the neuroanatomy. Well you might argue they are not genetically inbred they differ in their genetic instructions and so we go to a little organism the water flea or Daphnia which is parthenogenetic and female thank god it takes on the burden there is no male chauvinism there and if you look at certain visual neurons or cells in Daphnia you will see that the left and the right side are not alike and if you pick four examples of genetically inbred and identical animals you will see they are not alike as you can see in this picture and finally to respect the great Ramani Kahal here is a picture of the cerebellum of the rabbit from his great classic and you can see even in repetitive nervous structures there is an extraordinary amount of diversity. Well so much for that so what I have said is that the brain could have the lights the brain is composed of countless millions of these neurons in many types perhaps you could put on the next slide and then give me the lights and the final point is that besides varying in type and besides having variance within type this is the so called terminal arbor of one of the neurons it's an actual picture or tracing of a picture of one of the neurons that goes to that sensory the touch cortex that I showed you on that diagram area 3B it was done by Landry and Deshen in Canada and the most amazing thing about it is while the neuron starts off quite simple it really covers about one millimetre square in its arborizations and is overlapped by the other neurons in a dense tangle that might remind you of a genre ok now I can have the lights thank you well what I've tried to indicate to you is the enormous complexity of the nervous system and despite the commonality of its structure an enormous amount of individual variants right down to the level of the so called neurons themselves now I think we can take that slide off and turn to our next problem which is to consider on this basis of information how could we put this all together if we confront an environment with such a nervous system what happens well the conventional idea abstracted immensely here is that somehow a stimulus some particular occasion in the world interacts through neurons with this network and that that leads to a change in the chemistry of the synapses between the nerves changing the value of how much one signal will go versus another and that eventually will account for learning and memory this is no theory this is a metaphor here now um nonetheless the metaphor prevails and the question before us is how can we decompose it into possibly competing more specific models well I'm going to take the liberty of saying that even at the level of neuroscience not as you will hear from Dr. Shank and Simon and Dennett at the level of language neuroscience there is a tacit information processing model namely that the stimulus of the external world represents some kind of information that is to say describable in some way a little more abstract and universal than the prejudice of the observer and that that goes into the brain which has some kind of program genetic and acquired through that other learning program which in turn leads to learning and memory and the learning and memory leads to behavior of course I've left out the feedback loop that would cover that is the kind of model I think that prevails today in neuroscience and what I'd like to do in the next couple of minutes is point out to you that there is an alternative way of looking at the nervous system that confronts several problems that the information processing model has when it deals with a world that does not come neatly labeled and named and that model which we'll call the group selection model meaning selection of groups of neurons of the kind that I talked about that are highly variant and enormous in number in large repertoires that the stimulus comes in and indeed as you'll hear later it's not neatly tagged the brain confronts that in various filtered forms and a selection is made from a vast repertoire of variant neurons in each individual leading to amplification of the synapses of the adapted groups in an analogy with differential reproduction in evolution that is indeed instead of dividing as individuals do an evolution to form progeny these cells will selectively strengthen certain synapses and weaken others in such a way as to lead to a change in the population balance of these repertoires and the assertion is that this is capable of the process of generalization namely the ability upon confrontation with a small number of members of a rather large set being able to confront and identify new members that puts before us a very fundamental question which I'm sure Dr. Dennett knows much more about than I do both historically and in fact and that is what is the nature of the stimulus is it an essential class that is does the world come packaged as some of the essentialists felt the jungle was for tigers in labors is it a list in which singly necessary and jointly sufficient features will define an object a chair, a table, a leaf a particular niche a vein on the leaf what have you or is it in fact an arbitrary class something which the animal simply names for convenience with no necessary relationship to other members I think Dr. Dennett will recognize the ancient doctrines of realism and nominalism hiding in there somewhere but don't ask me or is in fact a signal as Ryle the philosopher has named it on the basis of Wittgenstein a polymorphous set let me or a polymorphous class let me say what I mean by that this is a little quiz that was given to smart Cambridge students in England took them about six hours to find out the difference between yes here and no there and they did it by most devious means as students will occasionally do and most of them didn't get it I want to show you that pigeons get it very nicely this is a so-called polymorphous set it's a disjunctive class and the feature that distinguishes or the discriminant that distinguishes yes here from no is the following at least two of round or dark or symmetric every object in this class fits that and not that these are singularly difficult classes to resolve but they appear by studies in modern categorization theory in both psychology perceptual and conceptual psychology and in fact in thinking about the problem to be closer to the description of real stimuli in the world and if the world of stimuli is a world of polymorphous sets how can an animal adaptively categorize and generalize now I've focused our problem and if I could have the next slide I will try to show you an instance by Professor Heronstein and Professor Cerella at Harvard of a most startling finding related to this world of stimuli Dr. Cerella presented pigeons under an operant conditioning mode in which they were rewarded for a particular kind of behavior that arose when they were presented with images of oak leaves as you see on the top in four presentations upon reward pigeons positively discriminated all oak leaves from all other kinds of leaves shown on the bottom and the class is fairly large well you might say pigeons live in the world of oak leaves evolution could easily have selected their brains for that although these days you know one wonders where pigeons live so these clever researchers then exposed the pigeons to pictures taken by a scuba diver, a fish now I think most of you will agree the pigeons do not generally evolve in a world of fish and the astonishing thing is upon operant reward the pigeons upon a few examples recognized with a high degree of accuracy all subsequent pictures of fish in a highly complex mode on the next slide you'll see one of Dr. Heronstein's paradigms maybe you can lower that a bit the key examples are upstairs trees, pictures of trees photochromes chosen at random from a large set recognized upon three or four examples and then every example of trees trees near, trees far, trees against water, trees against buildings but not things that might look like trees like this vein pattern, this celery stalk this central park lantern and what have you indeed we can take the slide off indeed what is startling about this set of findings is the robustness of the pigeon behavior, we can take that one off okay the robustness of showing slides indeed Dr. Heronstein has told me that he took a picture of a lady in Cambridge and it sounds more like a French art movie when you tell it this way in various contexts and then showed pictures of this lady in forests at 800 feet in various busy thoroughfares and the pigeon had no problem whatsoever recognizing her if he dressed her friend up in the same clothes in the same scenes the pigeon rejected her now you must not rush to the idea that this pigeon is in fact generalizing upon language or a descriptor of human beings or as my colleagues might say parsing for that nonetheless the pigeon is generalizing in an astonishing way and I assert that it is not very likely that by any means of conventional learning you could account alone for this particular behavior now let's come back to our theory the theory of group selection pretends or hopes to account for this by saying that in fact during animal development enormous repertoire of variant networks of neurons are formed in every single individual brain and then experience which must be considered to start in fact then in utero for us vertebrates and mammals experience leads to a selection of those groups that are adaptive and this has a resemblance to some of the ideas of Dr. Shankett a much higher and more abstract level more than one kind of neuronal group ipso facto just like the immune system will be successful for any particular output of performance and one kind can be used for more than one signal for two or more different signals well if I present a theory of this kind sufficiently bodilorized for this occasion I should at least present some evidence and indeed now evidence has accumulated the first kind of evidence I'd like to show you now I'd like that next slide please comes from our laboratory we've been interested in the problem of variance in the developing nervous system in the following way what we want to know is how the nervous system makes its connections in the first place and in order to do that we've isolated molecules that are responsible for the interaction of one growing nerve with another during the early parts of formation of the network these molecules are known as CAMS or cell adhesion molecules and the one you're looking at in a cartoon form up here is known as NCAM the neural cell adhesion molecule is found on the surface of every neuron in the body central and peripheral and it is a very curious molecule indeed it comes in three regions as you see colored differently each one of which is quite separate in three dimensional space the red region out here is responsible for binding to another molecule that's sticking out from another cell here's the cell you must imagine it to be as big as this building and then put about half a million of these molecules sticking out of its surface like that waiting to stick to another molecule the middle part of this molecule has an unusual structure which is charged we won't go into that on the next slide the mode by which these molecules bind here's a cell over here sticking a molecule out another cell sticking out an identical molecule they bind to each other that's been shown mechanistically the astonishing thing about these molecules is they switch on and off during development in an extraordinary way to define the ultimate address of particular neural circuits there are now three such molecules and now as of yesterday the gene for one of them has been isolated and the picture is one of an enormous dynamism of shutting the molecules on and off depending on the milieu constructing circuits and indeed if you perturb the molecules that define the address not by having a particular place pre-assigned but by the dynamics of the situation you get a scrambled nervous system that's been done inside the animal but on the next slide maybe I can show it to you outside the animal concentrate on this middle frame on your left and compare it to the one on the right the one on the left shows those round globs up there those are actual nerve cell bodies seen in an electron microscope projecting their nerves in an orderly pattern a coronal radiation of branches when we block the cell adhesion molecule which is responsible for that structure we get the spaghetti like structure you see up at the top of the right middle pattern each individual nerve fiber just goes off at random and makes a mess so here we have a principle not only of common structure and regulation but also of implicit and obligate variation in the formation of every nervous system the consequence to make a long story short is that no two nervous systems even those of twins can be alike well this is one kind of evidence required by this theory you will remember could you take that slide off a minute you remember that if that's the case it's formed and these molecules are not the only way but they're an obligate path the next thing I have to do is show you how experience could select groups and here we have thanks to the work of Professor Mertzenick and his colleagues in California University of California I'll have the next slide now an absolutely extraordinary example in the so-called somatosensory cortex you see that our old friend up there of the map now his experiment is rather simple I wonder if this is going to look absurd he sticks an electrode into the monkey skull and he puts it in that area 3B he then touches the finger and of course in a controlled and meticulous and in fact immaculate fashion that scientists would otherwise if you did otherwise they'd consider irresponsible and he makes a map he presses the finger and he sees which cell fires and that's the criterion for his map now here is what he finds no two monkeys have the same map although all maps represent the digits as you can see here they represent the digits here's that map blown up in a particular fashion well they're named over here digit 1, 2, 3, 4, 5 up in the right panel and then the palma areas etc and you see the hand over there but no two animals alike then he does an astonishing experiment he cuts one of the three nerves to the hand the so-called median nerve this one over here which supplies the so-called glabrous or smooth skin of these fingers from the thumb to the middle of the middle finger over here and he ties it so it can't grow back and then he records immediately and for six months thereafter and he sees the most astonishing thing and I welcome Dr. Milner's comments on that for the speech area what he sees is a kind of Darwinian competition the map immediately changes when the access to the world changes the dark areas representing the back of the hand begin to take over as you can see in the bottom map and even more astonishingly the map areas assigned to those nerves that have not been cut are immediately changed in their borders and then for six months after a kind of Darwinian struggle occurs where the remaining nerves without any evidence whatsoever of growth take over the remaining area after a little bit of a fight landing up with a map that's different in each case well I won't go into all of the details of Dr. Mertzenick's experiment but it is an experiment which exactly fulfills the idea that the substrate of the map are groups of these neurons tied together the inputs select certain of them and after competition and struggle some of them come up as a map and if you change the conditions a new map comes up well I can have it off now and I can turn finally to a consideration of how this theory might be put together in a consistent fashion what I've explained to you is certain evidence which is consistent with parts of it but the question is could you build a machine which would do such a thing an unprogrammed or a tomaton constructed according to such principles again time and space do not permit me to go into all of the details but what I would like to do is briefly describe a machine of this kind my colleague George Ricci and I and now others have been constructing simulating it of course in a computer the first thing you will notice if you put together what I said is that it won't do to try to categorize an unlabeled world by just your finger or your eye you need finger and eye and motion and a whole lot of other kinds of samples that are independent for this set of polymorphous classes one possible way in which you might think of this is for example if you saw the pattern A not the letter that your eye in fact would be able to register certain local characteristics like the bar and what have you of the ending whereas your hand might be able to trace the continuity of the figure and in your brain, thanks to evolution these maps and a number of other things a correlation is made between the registrations in one area of the cortex and in the other you need both local features and you need to if you will define the object it is an interesting thing that besides pigeons, 3 year old human babies are marvelous experimental objects for this Professor Elizabeth Spelke for example from the University of Pennsylvania has rather convincing studies to indicate that a 3 month old infant defines an object not by its texture, color, size adjacency to other objects but by its systematic relative movement with respect to an occluding object and it does of course not have to be the object as a scientist describes it well how could that happen the idea is that whatever the pattern is, some local features are abstracted some correlation defining the object is then put in, selection occurs upon groups which are mapped to each other and their correlation now stands for aspects of a class we happen to call that a classification couple it doesn't really matter but in this machine could I have the next slide which I shall not describe in detail you can see an example of the layout and then I'm almost finished this machine is called Darwin 2 its left side is called Darwin its right side is called Wallace the idea is that right or wrong we have to be reverential and the purpose of Darwin you will see on the next slide but for a moment let's keep this go back, sorry about that and syntactical things are could you go back one yes the purpose of Darwin is to look at individual features of something presented to this visual monster which has no program but only boundary conditions and then selections are made on its quote neuronal groups in such a way as to yield patterns quite independently it does a kind of search of space and traces outlines of things and varying orders that aren't important and makes another pattern of a link together can I have the next slide in this automaton which has at least a million connections in a computer representation Darwin's choice is to make an individual unique response to each object in a particular position in space but after strengthening its synapses as you will to respond stronger to that particular object even though it hasn't identified Wallace on the other hand gives a similar response to different stimuli that have common class characteristics like all the forms of A and what happened and finally because of the way the structures higher up our link the interaction of these representations gives a kind of associative recall can I have the next slide this next slide shows a look into Darwin and Wallace's brain if we look on the left side here we're looking at Darwin if we look at the lower order networks we're looking at for convenience it's response to a narrow letter A a broad letter A and a letter X but higher up that response has no figural identity with anything we show it and at the same time the Wallace side has identical responses to the two different letters A now we've tested this automaton it doesn't do magnificently well but 80% of the time with a great variety of different objects and without any instruction from the outside it does rather well well that we feel is not an adequate but a an initial confrontation with the problem of the self-consistency of any selective theory of brain function of course the thing that would excite scientists the most is finding out additional facts of the kind that I mentioned well I am almost done with Mark or two about the consequences of such models should they prove to be correct before I do that I think it may be useful to read to you from a work of a Harvard physicist named Percy Bridgman I don't know if he's fallen in disfavor but he is an exponent of a theory called operationalism he's not the originator and he says it seems that we are coming to an awareness of the existence and importance of our mental tools from the side of the sciences rather than from the side of the humanities the reason is not any reflection on the humanities but is a consequence of human frailty and the fact that the humanities are so much more complex and difficult than the sciences the most important intellectual task for the future is to acquire an understanding of human tools and so to modify our outlook and ideals is to take account of their limitations this task is not to be accomplished by any return to the insights of the past the insight that there is any problem here at all is devastatingly new in human history the sciences and the humanities find themselves facing the problem together it is too difficult and too pressing to permit the luxury of division of forces appreciation of the existence and nature of the problem is the first step toward the invention of new methods and outlooks that will be necessary to solve it it seems to me that the human race stands on the brink of a major breakthrough we have advanced to the point where we can put our hand on the hem of the curtain that separates us from an understanding of the nature of our minds is it conceivable that we will withdraw our hand and turn back through discouragement well with that let me summarize ladies and gentlemen in this lecture what I wanted to do is give you a feeling for some new thinking that is occurring in brain science you will of course hear more sophisticated ideas about much more sophisticated problems later on but I hope you will share my faith that this is a reasonable basis for those subsequent remarks I have assumed that only one direction is correct, population thinking but by no means is that a proven fact it is in fact however gaining increasing support and it has rich implications for human values and I would like to say something about that with all the risks entailed by extrapolation individuality as an all population thinking is its central feature categorization is the fundamental act not conventional learning and it must be adaptive the fundamental act in memory in such a system is recategorization not exact recall chronology is important and repertoire is important as I will show as I have said categorization is fundamental and adaptive memory is recategorization as I will show that will lead to the conclusion I believe that all perception is in fact an act of creation we all prefigured subtly however boring have it is and that all memory is an imaginative act the consequence I think is clear but it's debatable no ideas are priori authentic by any method and that includes scientific ideas and scientific theories for the theologians amongst us I think one consequence again debatable is that free will and free agency exist but are limited both chronology and repertoire are important we are indissolubly linked to the world by both natural selection and this kind of somatic process if it be real of group selection now you can make one statement too many may I have the lights and I think I have done so and I will finish since this has a certain theological base by telling you a story of a medieval traveler could we have the lights please who was an illuminator of manuscripts who passed from monastery to monastery just as I have gone from the Rockefeller to Gustavus and when he finished the monks thanked him could I have those thanks and said why walk why don't you take our horse he said well how will I return it the other monasteries a hundred kilometers away they said it's all right we have a hurt system and when he got on the horse they said but there's just one thing we forgot to tell you this horse is a religious horse if you want it to walk you say thank god if you want it to counter you say thank god thank god could I have the lights full if you wanted to gallop you say thank god thank god thank god what do you say if you want this horse to stop and he said you say amen well he got on the horse and he said thank god and walked through the forest he saw a clearing and said thank god thank god and it began to counter and he built up his enthusiasm and he said finally I see a clearing thank god thank god thank god and the horse took off the hair blew through his skull he was extraordinarily excited he said this is the way to go this is the only way an illuminator of manuscripts should travel went to his horror within about a hundred yards he saw an enormous crevasse of a thousand feet and he said what was that word they told me what was that word and finally said amen and he got right within one inch of the thing when the horse stopped he looked up and he said thank god since we're running a little bit late I think we'll adjourn at this time and reconvene at two o'clock this afternoon