Dr Daniel Glaser
Dr Daniel Glaser is Head of Special Projects in public engagement at the Wellcome Trust. He comes from a neuroscience background, was the first ‘scientist in residence’ at the Institute of Contemporary Arts (ICA) and is behind ‘Science Gallery’, which will soon open in London.
What intelligence really is about, is about the relationships between the brain and the world.
On becoming interested in science
My earliest focused memory of being interested in science is going to the Christmas lectures at the Royal Institution and I remember a crackling fire, a coal fire, and going with my brother and trying to be at the front of the stands there so that I could get picked to do the experiments on the telly. But that sense…I remember Heinz Wolff doing a brilliant experiment about chicken bones, demonstrating that they were both elastic and brittle, and he cooked one and then snapped it and he put another one in some kind of solvent and it kind of bent like a comedy bone in a cartoon and it started off…it was the same pair of bones to begin with, and just thinking, wow, there are these compelling demonstrations that can illustrate underlying stuff that’s all around us but really make a point in a kind of compelling, visually exciting mode.
On focusing on neuroscience
The route was somewhat tortuous, I did maths and English and physics at A-levels and matriculation and at university I started off doing pure maths at Cambridge, did that for 2 years, got bored – learnt to juggle, actually, which meant that my second year exams weren’t as good as they might have been – and switched to English Literature. I then did cognitive science for a year at Sussex, but became a neurobiologist and really I was interested in brains and thinking and I took this digression through artificial intelligence, trying to make computers that think, but in the end I converged on trying to understand natural intelligence, the thing that distinguishes us from lower animals perhaps.
We know lots of things and the list of things that we know is growing ever longer and the rate at which it’s growing is increasing, and probably you could argue that, for neuroscience, the rate at which we’re learning things is growing faster than almost any other area, with the possible exception of genomics. There are big initiatives happening in Europe and America now which are sort of trying to map out the whole of the brain and it’s called the Connectome Project and trying to learn everything we can about a brain. Having said that, our fundamental understanding of the nature of intelligence, what really sets us apart from, you know, apes or cats or dogs, I think hasn’t really advanced all that much.
Martin Rees, an old friend of mine, always says that the brain of a single rat is more complicated than the physics of the universe, and he would know; I’m not sure if I do, but I think the complexity is definitely part of the problem. And in particular, and I think this is gonna be a problem with these big projects, to write down everything we can about the brain, it’s really hard to know what difference makes a difference.
What I mean by that is, neurons are very complicated and there are lots of them and they have lots and lots of bits, neurons do, that connect many thousands or tens of thousands of other neurons and these connections between neurons and the properties of each individual neuron are changing all the time. At the same time there are these really big global properties of the brain like consciousness, it seems that, you know, you are either awake or you’re not, you’re aware of something or you’re not. These global properties…it’s very difficult to bridge the gap between these properties of the entire brain of a human and the low level stuff. I think that kind of explanatory gap, as we call it, between these bottom up accounts of the machinery and these top down accounts of behaviour is part of the reason why it’s so hard to grasp.
On defining intelligence
I think the Turing test, as you’ve mentioned, is a very good example; why is it a good example? Because what intelligence really is about, is about the relationships between the brain and the world. Intelligence is not a property of the brain which you can describe considering the brain in a vat, or considering it in an isolated machine. My own belief is that no matter how much we learn about the functioning of the brain as a machine, as a biological machine, we’ll still never get to the essence of intelligence because it’s about the relationship between the brain and the world and the brain and other brains. And the Turing test where, in the original version – and most versions now are similar – typing on a keyboard these one sentence answers, very much like the experiment that I constructed for my computer science O-level back in the day, allows you to use your own intelligence to assess the intelligence of others. Now, that’s circular: by that definition you need to be an intelligent machine to decide what an intelligent machine is, but I think that circularity is probably gonna be with us for quite some time. I think the definition is based on how other brains interact with the brain in question.
I’ve always been interested in questions and answers and the generation of narrative and of stories and I did – slightly annoyingly and precociously – an O-level qualification in computer science when I was 13. And the system that I built for the exam was a question answering system. You would say, ‘Who is the President of the United States?’ and it would answer…I can’t even remember who it was then, perhaps Ronald Reagan, or even earlier. And the system itself knew nothing, it knew nothing about language, it knew nothing about the world. But all it did was it looked for key words in the question like “President” and “United States” and it just matched the key words in this list of answers it had and it spat it out, the answer that matched most closely.
And what was weird was that it was possible to have Q&As with this system which appeared to be very intelligent; its knowledge appeared unlimited, it could answer questions convincingly. And what this did for me was to make me understand that the appearance of intelligence and the appearance of knowledge can be very, very superficial. That you can’t actually deduce what’s going on deep in someone’s head from their conversational abilities. And that was a really important insight for me, because at the time, and perhaps unwillingly now, I still base a lot of what I think about people from how they act and how they speak. This was an early sense that that might not actually tell you that much about what’s going on inside their heads, if anything.
On whether consciousness can be proven
Well it’s not the one thing…there will be other properties of the brain which are also of a similar nature, because the brain evolves and grows, so the genotype and the phenotype, how we came to be the genetically coded machines that we are and how the brain of an individual human develops through gestation and childhood and into adulthood. That’s all about the connection between the individual and the world.
Now we’re not talking about the wild child fantasy people have wondered about all these years whether, you know, if you left a child in the forest, would it grow intelligence and language, and there’s no uncontroversial data on that, but if you think about…I think quite often about dependency, how long does it take before a child is independent, able to look after itself, it takes ages compared with lots of other species and the reason for that is our independence is grown out of experience. We’re sucking in stuff about the world, making bits of ourselves like it and our interactions with it and so what makes us us is built out of world stuff, not just wet stuff.
I’m not sure how ethics necessarily plays into it; I think what we do when we’re doing experiments is to be good scientists, which is to try as hard as we can to think about every possible other interpretation of what we’re seeing that would be opposed to the one we’re looking for. So, I mean, this is a sort of version of the Popper, Coal Popper??? thing that to progress in science, you shouldn’t try and prove your theory is right, you should try and prove it’s false. The version here goes, you get a response in, let’s say a brain scanner, from having somebody do a particular task, you say wow, fantastic, I’ve found the bit of the brain that’s doing that task …task in question might be having a firm conversation with a lover. So you get someone to do that and then you say, great, this bit of the brain is active, that must be the love area. At that point, you have to stop and say, OK, what are all the other possible explanations for what I’ve just seen on the scanner? So doing brain science is no different from any other kind of science, you’ve got to be your own worst enemy, you’ve got to try and undermine yourself at every possible opportunity and resist the temptation – which is all too tempting – to publish, the “love area”, the “hunger area”, the “doing interviews” area, in a journal with a nice shiny picture that people will find compelling. Doing good science is the same across all disciplines: you’ve got to try and get it right.
On promoting scientific thought
Well the first thing to say about Science Gallery is that we’re not trying to promote anything, what we’ve generated is a space where experts and non experts can interact on an equal footing and that’s what you would call engagement as opposed to dissemination: the idea that the conversation is owned and controlled by the audience. And that’s important for, I guess three reasons at least: one is it’s a democratic duty, much of the science that’s taking place is paid for out of taxes, so there’s a tax payer’s right; secondly, we have really good evidence now that better science happened when the scientists are open to scrutiny. So, actually, if you’re interested in doing the best science you can…heads down approach, not talking to anyone who doesn’t understand exactly what you understand, is probably not the right way forward. Having a bright nine year old scrutinise your experimental programme once every three years probably makes your science better.
And the third reason is because it’s delightful: people just love it, and this sort of form of democratic participation around important questions about science, when it’s the young people, and particularly 15-25 year olds, driving the story forward, it’s just one of the most satisfying forms of human endeavour.
On the democratic process of science
I think what I’m quite interested in at the moment is about certainty, and there’s a funny thing that happens to a lot of scientists when they start addressing “the public” – in other words, not their peers – they develop a kind of rhetoric…let me say we, we develop as scientists a kind of truth, “let me explain to you how the brain works, no, it’s not like that” and, especially in cases like GM or MMR which are controversial, these controversial bits of science encourage the mentality amongst scientists when they’re talking to non-scientists that they have to explain things and be experts in it. And this is fundamentally incredible, unbelievable, because the public aren’t stupid, they understand very well that science progresses by trying to diss each other’s science, soon as the camera’s off, scientists will go back to the lab, she’ll go back to her daily work of trying to undermine, like I said before, her own hypotheses, trying to disprove them by her own experiments, and writing reviews which explain why the work of her fellows is wrong as well. So scientists spend all their time criticising each other’s work in the science group and then when they go out to the public there’s a tendency to say, “let me tell you everything about the brain” and that’s really unhelpful. What you want to do is to allow people to understand how you address a question and to give them tools to cope with uncertainty about scientific results. If you give them those tools, you give them insight into how science really progresses; you’re much more likely to engage them in the pursuit of scientific truth and they can form their own judgement about what’s good science and what isn’t.