In Our Time - Imagination and Consciousness

Episode Date: June 29, 2000

Melvyn Bragg and guests discuss the question of consciousness, our sense of self, and how we are able to imagine things when they are not there, which are problems that have troubled the great minds of philosophy for thousands of years. Consciousness has been linked to language, has been married to the mind and divorced from the body; it has been denied to animals, opposed to the subconscious and declared irreducible, but still it defies definition, and the debate rages on as to why we evolved it at all. But perhaps science will finally provide the answer. With Professor Gerald Edelman, Director of the Neurosciences, Institute in California and winner of the Nobel Prize for Physiology or Medicine in 1972; Igor Aleksander is Professor of Neural Engineering Systems, Imperial College, London; Margaret Boden, Professor of Philosophy and Psychology, University of Sussex.

This BBC podcast is supported by ads outside the UK. What makes people want to believe in aliens? I'm Tristan Redmond, one of the hosts of the Global Story podcast from the BBC. Donald Trump last week announced that he'd be releasing the US government's UFO files. So why the renewed interest in life out there? And what deeper spiritual meaning might people be searching for? Check out the global story. We are serious journalists on BBC.com.

or wherever you get your pods. Thanks for downloading the In Our Time podcast. For more details about In Our Time and for our terms of use, please go to BBC.co.com.uk forward slash radio four. I hope you enjoy the program. Hello, the question of consciousness, our sense of self and how we're able to imagine things when they're not there, are problems that have engaged the great minds of philosophy for thousands of years.

Consciousness has been linked to language, has been married to the mind and divorced from the body. It's been denied to animals, opposed to the subconscious and declared irreducible, but it still seems to defy definition, and the debate rages on as to why we evolved it at all. Perhaps science will finally provide the answer. Today I'm joined by the Nobel Prize winner,

the neuroscientist Gerald Edelman, who claims his new book, Consciousness How Matter Becomes Imagination, is the first ever explanation based upon scientific experiment. Also here is Igor Alexander, who's been studying artificial consciousness for more than 30 years. He's professor of neural engineering systems at Imperial College, an author of a new book also called How to Build a Mind.

And representing the world of philosophy, his professor Margaret Bowden, an expert in cognitive science at the University of Sussex. General Edelman, your work in neuroscience is focused on what you've called neural Darwinism. You've said you wanted to complete Darwin's program. Could you elaborate that for us, please? Well, yes, I adopted that term because I think Darwin made the most extreme. extraordinary advance in so-called population thinking, the idea that categories and species of animals come out of differences in populations of individuals under natural selection

or competitive constraint. And my personal belief is that since the brain doesn't appear to be a computer and therefore are subject to the laws of logic in a fundamental way, that we have to have some other principle and I think the principle is the same as Darwin enunciated, although the mechanisms are different, namely that in each brain there's an enormous variety and a very, very great difference in the neuronal or nerve populations, and that the way the brain works to get pattern is by selection, as Darwin suggested for species. Can you give us any examples of that, any examples of the way particular selections

take place? Sure. during development of the brain there is a genetic component which determines the framework there are a whole series of genes that have been discovered called hox genes and packs genes

and things of that kind that set up the initial framework for say a human brain but in a very short time another principle takes over which might be stated as neurons that fire together wire together

and that's not genetically determined that's determined by events that occur in each animal and that creates an enormous diversity in the connectivity of each individual brain, even the brains of identical twins. But when you say neurons that fire together, wire together, that's going to leave some listeners behind.

So could you just open that up before I go around the time? Well, neurons connect to other neurons by connections called synapsis. And it's well known that during development, for example, if two neighboring neurons in one's retina or in that fetus's retina fire, then the connectivity of the extensions of those neurons called axons

will be influenced by that firing such a way to establish them over competition and so they'll bind together in particular parts of the brain. Now that's not genetically determined, it's individually determined by the history of that animal. And we're talking about billions of neurons. Oh, yes, indeed. We are talking about, for example,

in the cerebral cortex, which, if unfolded, would be the size of a large table napkin and about as thick, we're talking about a 30 billion. neurons and one million billion connections. If you counted one per second, you just finished counting 32 million years later. I think there's a pause for thought there. Igor Alexander, what's your reaction to Gerald Edelman's drive in that direction?

Well, one of great agreement. Certainly it's been something that's interested me for a very long time how these populations of neurons actually represent. the world out there. And this, if they fire together, wire together, is what we'd really call learning, the basis of learning, which is something that doesn't happen

in conventional computation all that much. So this is what distinguishes the neural networks we have in our heads and some of the neural networks we can put together in the laboratory. And by following that through, quite a lot of interesting facts about consciousness can be studied on a computer in the laboratory. And yeah, there's still a lot of mysteries around.

We're not absolutely clear how this learning mechanism of firing together and wiring together actually does give us a sensation of a world out there. We're beginning to see how it would allow us to create simple patterns in the fire of our neurons, but there is something deep and mysterious that still needs to be followed up. And I think it's a job for neuroscientists and engineers to work together on, because there's a lot of system science that goes into explaining these things. One thing that daunts me when I was reading about this is the number involved. I mean, Gerald Edelman gave us some idea if you sign up a second,

it takes you 32 million years, you're still counting. What sort of control group can you get out of that? I mean, how can you deal with that mass? When we're talking about Darwin and population, as he dealt with relatively few finches, as it were. You're talking about billions and billions and billions and billions more neurons than there are particles in the universe by an enormous number.

How can you get a control on that, you go? Well, perhaps not more neurons than there are particles in the universe, but certainly... More non-neural circuits? A large number of circuits. More neural circuits. Yeah. I think you just must decide you're not going to be phased by these large numbers

because the behavior of neurons is a bulk property. They do interact with one another. And even if you study a much scaled down version of a neural net, you can grab hold of the principles that cause it to do anything that might be called intelligent. So, you know, the kind of neural nets we build are about half, the size of a bee's brain. And you say, well, okay, you're nowhere with that. But it's quite interesting that half the size of a

beast brain can do some remarkable things, which are scaled down versions of what happens in human brains. So don't worry about scaling. Margaret Bowden, do you feel that philosophy has been left behind in the study of consciousness by indications of what's been said by Alexander and Gerald Edelman? No, I mean, I understand. entirely agree with Eagle when he said that if we're going to understand consciousness,

then the neuroscientists and the engineers, systems engineers have to work together. I'd entirely agree with that. And I would say, and there's a third person in this trio, and that's the philosopher. Because, again, as Eagle said, there are some deep mysteries here, which I don't think are purely scientific mysteries. They're partly scientific mysteries, and they're partly conceptual. That's to say, philosophical mysteries. So as the science, well, I mean, the one he mentioned, I mean, the one I think that people think of above all,

how is it possible to get qualitative experience out of anything happening in the brain? Now, of course, we look for, and Jerry, for example, among other people, has found some very interesting and systematic, you know, correlations between the sorts of things that happen in the brain and some of the sorts of features of conscious experience. And that's obviously very important and absolutely crucial. But I think that at the moment we are not in a position where we can say that we understand what we mean by qualitative consciousness well enough to say that given that these things happen in the brain, then there must be qualitative. experience. I mean, I'm not saying that we'll never get to a point where we can say that,

and we'll only get to that point, partly from scientific study, but it does involve a philosophical advance too. And I think that, you know, maybe 100, maybe 500 years from now, we may well have a scientific theory of consciousness, but I think it'll be in some ways, you know, very different from what we think of now. What do you think about the philosophical history here? because we have Plato, Descartes, but something you refer to very emphatically is the American philosopher William James

and his description of consciousness. Could you bring philosophy into your summary of your position? Well, of course, William James, whom Jacques Bazin, the American scholar, called that adorable genius, was not only a philosopher, but also a psychologist, one of the founders of experimental psychology.

And I guess the best example of the phenomenological decision, scriptor of consciousness. He really did a brilliant job in his principles of psychology. And of course, one can't dismiss philosophy, although I'm not sure of this, but I think it was G.K. Chesterton who said, I thought to go into philosophy, but cheerfulness kept breaking in. And it's clear that from the time of the Eleatic philosophers and Plato and on that this has been a dominant subject, which finally goes around the field of epistemology, the theory of knowledge. And so, there is something in what Margaret says, that this fundamental personal experience, which

has been described rather brilliantly as usual by Bertrand Russell, British philosopher and mathematician, who said, you know, light comes in the eye. There's a physical process. It goes down the optic nerve. There's another physical process. It finally lands up whereupon all of a sudden the whole physical train of events seems to be accompanied by and succeeded by this sensation. and so utterly different that metaphysicians have spent all of their lives thinking up weird explanations for how it could take place.

Well, I have this to say about the whole subject that, of course, there is something there and something that won't be penetrated only by experiment. But I do think it's very important for people to understand that the role of science is not to recreate the world, but simply to describe it in formal terms. So, for example, if I have a theory of a hurricane and I have a beautiful computer simulation of a hurricane, and it does say 98% predictiveness, it's not a hurricane. So if someone asked me about my consciousness, which is intellectually tied to my body and the workings of my brain and my history, that I should, by my theory, even if it were totally predictive, generate in some zombie the notion of what green is, there's a real category era here. The philosopher John Locke, the great British philosopher once described the blind man

who said, I think I understand what crimson is, and someone said what, and I think it was something like, it's like the blaring of a trumpet. And so given the fact that it's tied to your body, it's clear that no scientific theory per se can generate an exterior perception in some other creature. You can't transfer the thing. It is intellectually tied to your history and body, and you will have those experiences. So I don't think it's the role of science proper to give a complete explanation of that. I think it's the philosophical and symbolic significance

that has to be explored. I think that it's the role of science and it's certainly the aim of science to explain that. It's quite different from generating it. I mean, you're quite right. I mean, a theory doesn't generate anything. It doesn't generate the phenomenon. But if you talk about a theory of hurricanes,

then the meteorologist is able to say, well, if this, that and the other physical process happens, these winds and these pressures and so on and so forth, then you must, you must, not just you will, but you must get certain other phenomena, namely the hurricane, on pain of self-contradiction in the explanation. And what I'm saying is that at the moment, not only do we not have this for conscious experience,

but I don't think we have any, certainly no clear ideas, maybe even not any very helpful ideas at the moment, about how we might do that. I'm not saying we'll never get it. We don't have it now. Well, this notion of causal compulsion that it must be the case is in fact something that I think scientists really avoid, even with the second law of thermodynamics.

You're always probabilistic about even your causal explanation. And I would say that, in fact, we are coming on some notion of what that involves. for example, it's well known that if you destroy a part of the brain called the mesencephalic reticular formation, that's the end of consciousness. You're in coma steadily. It's also in my book described how it might be that a memory, based on previous experience of category, could interact with what's coming in in terms of perception to give a scene. Now, the quality of that scene, as you feel it, green, red, and all of that,

comes into this philosophical notion of quali or qualia. And I think all conscious experience involves qualia. And what I think they are is higher order discriminations. And we can tell, I think, from heart theories and work, what makes the difference between one another. But the actual experience itself, I think, will elude anybody except somebody with a body. Ego.

I have a little more hope for science. A couple of examples. There's a brilliant paper by Thomas Nagel. which asks what's it like to be a bat. This is the third-person experience, which he points very clearly to, that you cannot have a scientific approach to. But what science does in very simple terms

is to work out what a bat needs to have in order to know it's a bat. We're never going to know, but how do we know that that bat has the right circuitry to know anything? And I think that is a scientific question, and perhaps it's an important one. A bat knows it's a bat enough to survive as a bat. And that may be very little in comparison to the sort of things we have to know to survive with our complexity.

The other point is that David Chalmers, an American philosopher, I guess, a current philosopher, has divided the problem into the easy problem and the hard problem. He says that anything that you can do by fiddling around with neurons and doing experiments on brains and so that's the easy side of the problem. And he describes what Margaret was saying as the hard side of the problem. Well, what happens in science is that the easy problem turns out to be enormously hard. And it requires a lot of scientific effort, and as the scientific effort progresses,

it impinges a little on the heart problem, and I suspect that at the end of all this, it's going to get rid of it altogether. One of the most enigmatic qualities of consciousness is its ability to be unified and fragmented at the same time. I mean, you are talking, Igor is talking. At the same time, there's no doubt you have 58 other thoughts and aware that this studio is painted this colour,

and your microphone is raised, all this is going. But I hope, the one thing you're thinking about is getting on with this conversation, but there's hundreds of other things going on. Now, is it possible to explain that process through the function of the brain's neural networks? That is, in fact, a central concern of our book.

The fact that consciousness has that apparently contradictory property of being unitary, in the sense that there's no way of decomposing your present awareness into this umbrella, and I think without creating yet another full scene, And James was very aware of this. And at the same time, though, from within, there are billions and possibly a countably infinite number of conscious states that you can have.

Think of all your classmates, the pictures, the movies you go to, just what's happening in this room. How do you put it together? And we believe my colleague and I, Giulio Ternone and I, that the approach to this is to have a formal analysis of complex systems, not complex computer systems. There's a full theory of that. of algorithmic complexity, as it's called,

but the brain is a complex system. And if you work through that, you will see that you can get integration as well as differentiation by these mathematical analyses. I won't bore you with the details, but it's a very general property of consciousness that must be explained,

and I think the way you have to do it is a bit the way the physicists approach thermodynamics. You have to get some measure of what it is and what it isn't. Ego. It's odd how I came to almost the same conclusion but from a completely different starting point. The starting point for me was,

why is it that almost 50, well certainly 50 years of something called artificial intelligence has done enormously smart things with conventional computers, but why is it that it never got to grips with just these issues of how within our brains we can have both the integration and the diversity at the same time? It turns out that the principles of operation of the brain are completely different from the principles of operation of a computer. But working backwards through the question, what is it that a brain-like system, never mind the real brain,

but something that we would recognize as a brain-like system needs to have in order to do just this thing, to think about a million things at the same time and also use most of its neurons in some integrated way to do it. And some of these answers actually lie in engineering and in control theory, and they've been around for a very long time. Not too fond of the word complexity.

I think it kind of throws you a bit when you say this is, it's the complexity that we need to understand. I actually think that we should get rid of that world. Well, in fact, I think it was Ravel who said about good music, Complex, but not complique. You don't want to have it complicated. I can define a complex system, I think, in a crude way to get the idea across.

It's a system in which the smaller parts are more or less independent, don't exchange much information, but as they get interactive with each other, more and more information is transmitted, and the thing goes from being sort of like a gas in its small parts like a crystal and it's large. And so, yeah, there doesn't seem to be another easily used word for this, but I think we shouldn't overdo the problem of what to use.

The fact is complexity is a result, not a cause. Where does philosophy come in on what's been said by the moment? Well, I don't know whether you'll call this remark a bit of philosophy or whether you'll call it a bit of science. I don't think there's a clear distinction between the two myself, actually. And I think that one very important concept for talking about the sort of complexity which we're talking about here is the notion of a virtual machine. You know, the notion in computer science, AI, systems engineering, of a virtual machine which broadly speaking means the organized system of very, very different sorts of functions arranged on many different hierarchical levels and interacting with each other in various, hopefully, specifiable ways, which is what the mind is.

And that's why I said earlier in this discussion that I thought we needed both the neuroscientists, I mean, that's obvious, me the neuroscientists, but also the systems engineers. I wasn't just being sort of, you know, nice to ego, because I think that we do need this notion of a virtual machine to help us think about just what these functions are. And I think that when we do understand qualitative experience, we will understand it as an aspect of this virtual machine. And I think we already understand certain other,

originally very, very puzzling and difficult problems about consciousness, I think have been explained and outlined in terms of this sort of concept. Can I come to machines now, Igor, you're involved in creating machines. You've talked about imagining a blue banana with red spots on. For instance, do you think that the machines you're working with will ever have the possibility of doing that? Well, they have this possibility now, and in fact, they're doing it now. Some of the experiments we've been doing recently are based very much on the ability of these virtual machines

that Maggie is absolutely right about. The word virtual means that you've got something to work with, which is a bit like the real thing, without hurting anybody. I can work on virtual brains. Tear them to bits, and it's not going to upset anyone. Now, one of the tests we have is whether they can imagine or not. But imagination has a very broad set of meanings. I'll tell you the way that we interpret it.

It's got to be able to imagine something that it hasn't been exposed to during some learning period. And this is usually induced by language. So if this thing's never seen a blue banana with red spots, It should be enough for the language to induce activity in the neural net that can reconstitute this internally in the system. And that's what we call imagination. Interestingly, there are systems you can put together where tweaking some of the interconnections between these reentrant parts of the system totally stops them from imagining.

So imagination does seem to be a product of some very subtle structure within both neural nets and our brains, I should think. Very Adelman. Well, I'm very stimulated by what my colleagues have said. Let me say, first of all, and I'll come to imagination in a moment, that I think what's needed to get at these complex issues is, first of all, a global brain theory. and I've prompted the notion of neural Darwinism for that. One of the essentials of neural Darwinism is what are called these reentering pathways, this particular kind of thing

that is not feedback in the engineering sense, but can be engineered, massively parallel connections between brain maps that are constantly doing some kind of higher-order correlation to give you space and time together because the brain is not governed by logic the way a computer is or by a central clock the way a computer is.

The second thing you need, I think, is the kind of thing Dr. Alexander's been discussing, namely you need the ability, and Margaret as well, the ability to simulate and examine in a virtual way these things, because they are enormously complex. You can't just sit in a chair and imagine them all. You have to see them worked out. And the third thing you need is an exact kind of experiment, which gives you a neural correlate of consciousness. Now, we have, in fact, for the last 16 years at the Neurosciences Institute, been building

machines or devices we like to call them, because I think a machine is really just a derivative of a computer. It has a definite program and a symbolic end, whereas in a brain it's a bit like the lady in the EM-Fauster novel, how do I know what I think until I see what I say? And it's not an effective procedure in a very precise way, but it leads to pattern recognition. Now these things, and we're up to Darwin Six, we've named them after the great man. Darwin Six now has eyes and ears and is mobile and is not told anything. And as Professor Alexander has done, we simulate the brain inside a computer.

And it goes and it does primary and secondary conditioning just like an animal. Of course, it is about as complicated as a medium-sized insect. And that brings us back to your comment about how many neurons does it take. Now, the third thing is this matter of experiment, and it was one of great delight that when we did an experiment on living human subjects when they were perceiving two different kinds of images, horizontal blue lines and vertical red lines,

alternatively through red and blue lenses, so-called binocular rivalry, and signaled when they were conscious and when they weren't. And we could measure this with a machine called a magnetometer, these minute currents in their brain, we found that when they became conscious, there was a huge increase in the activity of very specific areas of their brain, and furthermore, there was evidence of this re-entry,

correlation amongst massive populations and neurons firing together only when they were conscious. Coming back to a question I asked much earlier, Margaret, the idea of a philosopher sitting down and engaging, if one might use the term, with pure thought, and not getting involved with devices, machines, the sort of what we've been, Professor Alexander and Edelman have been, discussing. Is that possible anymore? Is that time gone? Unfortunately, it hasn't. I mean, there are still far too many philosophers around, and particularly talk about philosophers of mind, right,

who don't take the trouble to find out about the neuroscience, and who don't take the trouble to find out about the computer simulations, but who nonetheless are, in some cases, perfectly ready to dismiss them as being irrelevant. Now, I mean, I've got absolutely no patience with that sort of attitude. So, I mean, take Descartes, who set this problem for us in the first, first place. He wouldn't have been well, he would have been surprised, we're all surprised. He wouldn't have been

philosophically upset by any of the discoveries which Jerry and other neuroscientists have made. He was the one who was the first one to say every time some specific thing happens in your consciousness, then some specific thing is happening

in the brain. I, Descartes, don't know what it is. You scientists go out and find out. I actually think that Descartes would have had me burnt at the mistake, quite honestly, because part of his talk about mind was one that goes all the way back to Aristotle, and that is that there is something divine about mind that it's divinely entrusted to us so that we can not only recognize our own existence, but also recognize the existence of God. And going back to Aristotle, the fact that soul is the thing that does thinking and soul is only to do with human beings is still around when we talk. about machines and even the person driving a car while they're listening to this program will think oh well yes but there's something special about my mind something perhaps divine that all this talk about machines doesn't have anything

to do with you sorry please say what you want to say there is something special about that driver's mind it's unique to his history and unique to the history of the universe he is if you come at the numbers the way we did before you can see that clearly that you call it the soul is another issue. Descartes, of course, was interesting and I think wrong, but the wrong way great geniuses are

by prompting the question and clarifying it in saying that there were two kinds of things, extended things and things of thought, race extensor and race cogitans. And the first were accessible to physics and the other was not in a direct way. But that polarized our

thinking, and that's why he's sort of the greatest modern philosopher and why what Margaret says is going to turn out it to be true and I think it will disprove his assumption. That doesn't mean that by asking the right question, even if you have the wrong answer, that you haven't done something marvelous. That's what he's done.

Excuse me, you were going to ask me. I was going to say, you write in your new book, the workings of the brain more closely resemble the living ecology of a jungle than they do the activities of a computer or any machine we could possibly imagine. I'd like all three of you to address that. Do you want to kick off?

Yeah, I can unpack that one a little bit. Let's start with the computer. After all, it was in this country that Alan Turing gave the theorem, the general theorem that describes that. A computer involves algorithms or effective procedures, which are quite precisely described, in this case in binary arithmetic. It doesn't have to be that way.

But if you don't have a precise description, if there's a lot of ambiguity, you've got a problem. You've got to put in an error-correcting code. Logic simply doesn't tolerate. Even fancy logics don't tolerate that kind of smear. Well, the brain, as I said before when I was describing Neurodism, is unique in each case, not only in its structure, but also in its history. And we mustn't forget that it's in a body. Your body constituted a certain way within your species, et cetera, and the two interact in ways that can't be completely isolated from each other.

Well, when you put that all together, and you look at the metaphor of an evolutionary garden or a jungle, you see it has structure, but it also has uniqueness. And that uniqueness comes from variation, just as Darwin observed in Finches and whatever. And that variation, of course, sometimes is trivial and sometimes is absolutely essential. I think the main point is that one is capable in such a system of getting pattern recognition. And I personally believe that the way this will eventuate to back up what Margaret said is that ultimately we will do a synthesis. The way we'll really understand it is we will synthesize some structure which actually has these properties. Now, there's an ethical problem there, but I don't think one has to worry too much,

because it'll be much harder to make something that's like a human body. We will embody the principles of consciousness in something someday, and therefore synthetically understand this deep thing. That will take some time, but I fervently believe that, just as has happened in other matters, science will finally create a conscious entity. Can I come to your ego, and then, Mark, I know you're both posting to get in with ego, Alexander. These virtual systems we were talking about that have some sort of virtual mind and possibly work on virtual worlds.

All of that happens inside the computer. The computer is just a red herring as far as all that's concerned. It's just a substrate on which these complex systems run. But one of the problems with these things is that sometimes we don't know how. to design them. It's not the easiest thing for a designer of brains to sit down and say, I'll design this thing, so it has consciousness. One of the features of some of these things is that they have come into being through a process of evolution, artificial evolution at that, but we've had to use the concepts of evolution and emergence of activities from evolved systems, which is not part of the program as remit.

Yes, well that's part of what I was going to say. I think Jerry made a false antithesis there between computers and brains. I mean, of course, human brains in particular, are wonderfully, you know, deliverously unique and complex and so forth because of not only their different genetic endowment, but their different life experience, of course. But if you have a halfway reasonably interesting computer program and run it that can take in information from the outside world, etc., etc.

And you run it in a, you give it different so-called life histories. You will end up with systems that are interestingly different. I'm not suggesting they're going to be as complex as human beings. Of course not, but they will be unique. And as Igor said, you can do this by using evolutionary principles in the programme. I think that Jerry, when he's described computers in this conversation, has described very old-fashioned, limited approach to computation, which isn't actually what's being done in a lot of work that's going on now.

So I just think this is a false antithesis. I'll have to respond to that, won't I? I hope so. I think old-fashioned or not Turing's theorem applies to everything we call a computer today, and that means there's logic in it, and that it has to be precise. And the funny thing about brains, and of course you know, Professor Alexander, that I'm all for selectionistic ideas. I think what's interesting about brains

is that they can pattern recognize in the face of ambiguity and that computers can't. That's not a false antithesis. I'll lay down a challenge. The fact is that the power, for example, of our language is not in its clarity,

but it's in its ambiguity. I would ask Margaret if she were born simply would say first order predicate logic and algebra, would she imagine a Wallace Stevens poem? Certainly not, but then these sorts of programs that I've been talking about that are actually all the sorts of systems that are being built today,

never mind what's going to be built in 50 years' time, don't work with predicate logic and algebra, but that is my point. But they don't work with poetic metaphor either. No, but they don't create actions. Well, they work with ambiguity. They work with certain sorts of ambiguity. Even, you know, very simple neural network pattern recognisers can, you know, have capacities to recognize what you call smearing,

between different examples of a pattern, which the old-fashioned sorts of systems couldn't do. But they're ludicrously simple. Let me ask you this question. Let me ask you this question, Margaret, if I were to offer you, if you were to go hunting for birds in a swamp, a bit the way early Darwin was

when his father said he wouldn't amount to anything because all he did his hunt, supposing you were to go for birds and a swamp on a rainy day, and I offered you the American Air Force computer in a teacup and it was friendly and spoke English. Would you take that or would you take a dog? I take a dog, because I know enough about this stuff to know that we haven't achieved very much yet. But that's not to say that we haven't achieved more than what you're suggesting we have achieved.

Well, I don't think we have achieved very much, but it's very important to keep the distinction. Can I just bring in what might seem clumpingly obvious, but people listening to this programme could say something like, it's all the world to talk about this, but will machines ever be able to have free will, for instance, ego? The answer is never. I think there is a deep confusion that takes place between using words like free will, which have to do with human beings, and applying them to machines. If you apply them to machines, I'm sure you can find a machine where you could find a behavior

that you'll describe as free will, just proving that it can make arbitrary decisions, for example. But I think the way that the machine arrives at those things will be, as a result of being a machine, whereas a human being arrives at its free will by being a biological human being. There's a lot in common between the two, but the confusion leads you to believe that if the machine has enough of these things,

it'll suddenly become a human being. It'll cross the line. I think that's got to be made very, very clear. It can't do that. And the reason is clear. We'll not have what we call the phenotype of a human being, The human body is just as miraculous and maybe more so, if you include the brain, as any notion of the workings of the brain,

and it's absolutely essential to the brain that you have the form and shape of your body from cells all the way on up. So my comment would be, I don't like the use of the word machine, even for the kinds of things you and I work on. I'll call them devices for the moment, because I think a touring machine does encompass all known machines. Be that as it may. It would never occur to a machine, I think, to imagine,

or wonder whether a human being were a machine or not. What's your comment on this free-law? Well, I disagree with both of those. I don't think free will is a matter in general of making arbitrary choices, though we can sometimes make arbitrary choices if we choose. I think it's a matter of a certain sort of functional organisation, a certain sort of virtual machine,

that we have and the dogs clearly don't and that newborn babies clearly don't. And it's the absolutely no reason whatsoever why that shouldn't be a feature also of some hugely complex computer system. It seems to me you and I are going to swap jobs at the end of this discussion. I was using the example of making arbitrary decisions because one could point at a machine like that and say, look, it's got free will, whereas what it's doing is something simple.

I think the freedom of will has got to be interpreted in the sense of the freedom of, of being a human being, in which case it's irrelevant to a machine. Why people are interested in free will of human beings, because it's linked with moral responsibility. And in the sort of choices which involve moral responsibility, they're not made arbitrarily, or if they are, we regard this as a very bad way of making them.

I mean, they're made with deliberation, thought, comparison between different sorts of consequences and different sorts of moral principles, etc., etc., etc. Now, this, et cetera, et cetera, et cetera, is what I meant by talking about a very complex set of functions in this virtual machine. And it isn't a matter of arbitrary choice. It's something much more structured than that. And that's what freedom is, and that's why we value it. And I don't see any reason in principle, as I said, why one couldn't have that sort of functional structure in a non-human system.

It seems to me that it's hard to imagine, if you did. and have something like language, and thus what I call higher order consciousness, the consciousness of a self and the consciousness of a past and a future, and the ability to imagine situations of responsibility and value. And all of that, in our case, I think, comes from language or at least symbolic capability that most animals don't seem to show signs of. Unless you have that, the issue doesn't even emerge. So until you have a machine of a kind that will master the problem of language and meaning,

least symbolic formation, then I think the issue is moot. We can get at issues of primary consciousness before we address that sticking point. Science must be modest. I believe that the issue is this. If we start with higher-order consciousness and the kinds of things Margaret says, I don't think we're going to get very far

because it's going to be a long way before we master the problems of how language is generated and how discursive symbolism occurs in human beings or musical or artistic symbolism even mathematical symbolism. But we can attack this problem of primary consciousness, which I believe it's possible to recognize

even in animals like dogs and what have you who have homologous structures. So, you know, one step at a time. So this notion of leaping forward to the machine that has all these implicit things as well as free will because it can write a line of Shakespeare and if you challenge it, it'll write a line of Coleridge is perhaps asking an awful lot.

We don't have to go that fast. Eagle? I'm not sure that the demarcation between primary and higher-order consciousness is as clear as is implied. I'm clear, it's sharp. Or sharp, you mean, rather than clear. I think that the artificial will knock on the door of higher-order consciousness. It can't get all the way because going all the way does mean having the consciousness of that particular species. which is related to the survival values of that particular individual,

which for a machine must be different, even a virtual machine must be different. I agree. I agree with that. Margaret, do you have a final comment on that? I would just say that I think the really interesting question, both from the philosophical and the scientific point of view, is what are human beings like?

And what is human, if we're talking about freedom or consciousness, what is human freedom and consciousness like? and I don't really care whether or not one could make a machine which had the same sort of consciousness or the same sort of freedom. My interest is how we could use that sort of study along with neuroscience to understand ourselves better. Final word from Igor Alexander? That seems to me the ultimate aim of building any machine.

We don't want to build machines that are going to go around and do terminate as seven. We want machines that tell us about ourselves. and a philosophy that tells us about ourselves, something that needs to be repaired a bit. Well, thank you very much to Professor Margaret Bowden, to Igor Alexander, whose new book is called How to Build a Mind, and to Gerald Aedlpon,

whose new book is called Consciousness, How Matter Becomes Imagination. Thanks for listening. We'll be back in September. We hope you've enjoyed this Radio 4 podcast. You can find hundreds of other programs about history, science and philosophy, at BBC.com.uk

forward slash radio 4.