Transcript - March 2009
Gareth Mitchell: This is the official podcast of Imperial College London. And I’m Gareth Mitchell, a lecturer on our Science Communication and Science Media Production MSc her at Imperial and I also host the BBC’s Digital Planet technology program. As for this podcast, I’m in pursuit of happiness with the Business School professor who reckons he can measure how happy we all are. That’s in just a moment. Also this month, I’m sorry, Dave, I can’t do that. How the renegade robot in the movie 2001 inspires a deep conversation about what it means to be human.
Murray Shanahan: The astronauts interact with HAL just by talking to HAL. And HAL doesn’t have any kind of obvious body. It doesn’t move around in the same space as the astronauts that it’s talking to. And that kind of disembodied AI is not what I’m interested in.
GM: So find out what he is interested in. He’s one of our leading cognitive robotics experts. And if that sounds good then does it taste nice too? Or maybe it has a pleasing colour. For people with synaesthesia, like one of our genetics postdocs, listening to sounds really can be sensory overload.
Julian Asher: For me specifically a violin is what I think of as almost like a red wine colour or a claret. I also perceive a visual texture to the colour. It’s sort of a shiny, very liquid looking red. As opposed to when I hear a cello I’ll get something which is almost like a deep honey gold colour. It has a viscous feel to it which does remind me a bit of honey.
GM: And now we have the most accurate picture yet of the genetics behind synaesthesia. And in our news roundup money is on its way to Imperial from the Bill and Melinda Gates Foundation. Find out why. That’s all right here on the Imperial College podcast.
Professor Paul Dolan on why happiness is a serious subject
Alright, well let’s jump straight in then. And on a scale of one to ten how happy are you? Oh, and I’m about eight, probably, since you asked. But then that’s the trouble with trying to measure how happy people are or what makes them happy or fearful. Because just asking that actual question tells you how they feel at that moment but it doesn’t necessarily say very much about their overall attitude to things. Luckily though none of this seems to get Paul Dolan down. As Professor of Economics at our Business School he’s happily working away on all kinds of methods of accurately determining what I should more correctly term as subjective wellbeing.
And far from being an airy-fairy study of the frivolous gauging individual and collective attitudes to our wellbeing has all kinds of economic consequences and it all feeds into hardnosed policy. How does a government decide what its health spending priorities are if it can’t get a reasonable handle on how various conditions affect our wellbeing. When I popped into see Paul Dolan the other day my initiation into the study of happiness began with a quick grounding in what economics itself actually is.
Paul Dolan: Basically it’s about how we satisfy an unlimited set of wants with limited resources. That’s what economics is about. But one of the ways in which we do that is to think about what the impact of allocating resources in the way we do is. And it’s not just about financial impact. Because all of us care about money but we all care about other things too. We care about time with our family. And in standard economics, in the language of economics, I should say, that’s our utility. And part of that will be our income, part of that will be your income, part of it will be my marriage, part of it will be my social contact. That’s a very important determinate of individual wellbeing is how much social contact you have. For me that’s every bit as much of economics as money is.
You could say this is all a bit kind of soft and airy-fairy or something. It really is money that counts. And someone can take that view but they would still want to know about happiness and wellbeing because of the association from happiness and wellbeing to income. What impact does income have on an individual’s wellbeing but what effect does an individual’s wellbeing have on how much money they make and growth and GDP and all those things that standard economics cares about. So even if you are wedded to this view of economic growth and GDP you might still want to know something about what things affect that. And happiness is one of the biggest things that has an impact I think.
GM: So ultimately then in policy your work has been used in guiding government policy. So how exactly? How does what you do here at your desk at Imperial translate into what happens down the road there in Whitehall?
PD: Where wellbeing measures of a very general kind have been most predominately used, most often used, in UK policy and elsewhere is in allocating healthcare resources. So when we look at institutions like NICE, National Institute for Health and Clinical Excellence, is was called the National Institute of Clinical Excellence and they added the ‘H’ but still call it NICE or it would be something like NIHCE or something. The way in which NICE assess the value of the benefits is in what many people will now of heard of is called Quality Adjusted Life Years, QALYs. And so one QALY would be one year of life in full health or two years of life in .5 health and so on. And the question then is how you value the ‘Q’ in the QALY. And the way that NICE does that currently is to ask a representative sample of the general population to imagine what it would be like to be in certain states of health and illness.
GM: So if I understand correctly then if you’re in NICE’s position and you have to advise on allocating the healthcare budget then you’re looking partly at the clinical outcomes but also at this rather harder to pin down wellbeing aspect too?
PD: Yes. One of the interesting questions for us is the relationship between those traditionally measured, as you say easier to measure in many ways, clinical outcomes and the harder to measure but softer measures, if you like, impact upon an individual’s experiences.
GM: But you’re quite critical about how those decisions are being made then aren’t you? And you’ve written widely about it and published a lot of papers on this. So what’s the gist of your criticism of all this?
PD: My sense is that NICE is 90 per cent nice and 10 per cent nasty. Most of what it’s doing much of the time I would stand up for to the hilt. I mean these kind of allocation decisions have to be made, and they are being made. They’re being made all the time. You can’t spend the same pound twice. It’s bringing some transparency, consistency and openness to what are incredibly difficult rationing decisions ultimately. Most of what they do is absolutely fantastic. It’s just I have a serious concern about the way in which the ‘Q’ in the QALY is valued. And essentially what I think NICE is doing by calculating the ‘Q’ in the QALY in the way that it’s doing is that it’s accounting for the states that people fear most not the ones that they may experience worse. Real suffering, I would argue, would probably lead us into investing relatively more money in mental health conditions. Because we’re not very good at imagining, at knowing, what the experiences of a real person in that health state are.
GM: Can we nail this down a bit. Can you give us a few examples of how you make these kinds of decisions?
PD: Let’s take a couple of health states. Let’s say one is associated with some problems walking about and the other is associated with having some anxiety depression. Both are bad conditions; neither of which we would want to experience if we had the choice. But if we’re choosing where to allocate scare healthcare resources we may at sometime be faced with that sort of choice. If you ask people to imagine these two conditions, these two health states, having some problems walking about or having some mental health problem, by and large they think of them as being equally bad.
If you go to people in those health states, experiencing them, living them day to day, then some problems walking about after some time, not straightaway but after some time, is actually not that bad relative to having some anxiety depression. It starts off bad and it stays bad. You don’t get used to being depressed. If you’ve been depressed for a year it’s as bad on the Sunday morning as it was on the Sunday morning when you were first diagnosed. Having some problems walking about is still not great but after a year on the Sunday morning when you get up and walk around in so far as you can it’s not as bad as it was the first Sunday morning when you got up. So what we want to know is we want to know about the flow of experiences in these two health states. Not just the imagination of what it would be like for me to be in that state.
GM: And you’re saying the problem then is that resources are being allocated largely to do with what people’s anticipation of these states are going to be and what their fears are rather than the reality?
PD: Yeah. So I think the long and short of it is that the quality of life assessments that go into NICE’s algorithms are based upon the states that people fear most. If we ranked all of the health states according to fear we get a different set of ranking in terms of their impact on suffering. So fear leads us in one direction and real suffering may lead us elsewhere. They’re never going to be the only thing that will inform policy, I’d be stupid to think otherwise, but they should inform policy more than they do. And what you really want do, in so far as we can, is to reduce the real suffering out there. Who would argue with that? Who doesn’t want to reduce suffering? And I’d like to see healthcare allocated in a way that does more than it does at the moment to try to reduce the real suffering of real people.
GM: Paul Dolan in our Business School. And speaking about the human condition, how do you find out what makes us quintessentially human? Simple. Just build a robot. Well, actually that was a lie, well the bit about it being simple anyway. More on that in a moment. First though let’s have a quick heads-up on the stories that have been putting Imperial into the headlines recently.
Headlines from around the College
There’s new hope of a successful vaccine for one of the most deadly strains of meningitis, so say researchers at Imperial working with colleagues at Oxford University. Meningitis, which causes serious inflammation of the tissue around the brain and the spinal cord, can take either a viral or a bacterial form and it’s the latter that’s the most serious. Of the three strains there are now vaccines for bacterial meningitis A and C. In meningitis B though the difficulties is that the bacteria responsible for the infection successfully passes itself off as a friendly human cell and therefore dodges attack from the immune system.
But reporting in Nature the researchers say they found out how the bacteria pulls off this Houdini-like feat. They’ve identified the structure of a molecule that friendly cells in our body use to tell our immune system to leave them alone. But the meningitis bacteria also uses this molecule to fool the body’s defence system. So knowing this it might now be possible to develop a vaccine where you inject a bacterial protein which lacks this protective molecular cloak thereby priming the immune system to mount a full attack on the deadly bug.
And on a not unrelated subject, over seven million dollars of funding from the Bill and Melinda Gates Foundation is on its way to Imperial’s CD4 Initiative to help combat AIDS in developing countries. The Initiative was founded to develop a quick, simple and cheap test for HIV. At just $2 a shot the test rapidly counts the number of CD4(+) T cells in a patient. These killer cells are what the immune system uses to fight disease. And in HIV patients T cells are greatly reduced. A test that can gauge this accurately and quickly in invaluable as it speeds up the whole process of diagnosing HIV meaning that crucial antiretroviral drugs can be started as quickly as possible. Imperial’s CD4 Initiative is working on all this with academic and industrial partners in the UK and the United States.
And you can catch up with all the comings-and-goings at Imperial before most of the rest of the world does by checking our Press Office and Events website. And that’s at imperial.ac.uk/news.
Professor Murray Shanahan on cognitive robotics
So a minute or two ago we were talking about the human condition and things remain philosophical now. Whilst the economists in our Business School wrestle with how to keep us all happy and healthy across campus, here at South Kensington, a group in the Department of Computing is interested in an even more fundamental problem. What makes us human? And specifically how do we think and therefore what makes us intelligent? Welcome to the realm of cognitive robotics. And that’s the chosen field for Professor Murray Shanahan.
Murray Shanahan: Cognitive robotics is all about trying to endow robots with high level reasoning and thinking skills. So it’s really about trying to make our robots more intelligent. Making them smarter. And the current generation of factory robots that we have, for example, are really stupid in the way that they behave. They don’t need to do much in the way of reasoning about their actions or planning. You can get away with that in factory robotics and it’s very useful.
But if we ever want to build the kinds of robots that we see in science fiction, and if we ever wanted to use them, say, in a factory situation to have much more versatility and to do a wider variety of jobs, then you need a lot more intelligence than we have in today’s robots. So cognitive robotics is really about making clever robots rather than stupid ones.
GM: And I’m sure even those terms like clever or stupid are problematic for you because you come at this very much from a philosophical standpoint. So before you can even collaborate with engineers maybe and think about building clever robots you need to determine what clever or intelligent actually mean.
MS: Yeah, indeed. Well the words clever and intelligent there are not necessarily the ones that I would use professionally. So that’s a more populist way of expressing it really. And what I’m really fundamentally interested in is really the nature of cognition. The reason why I’m interested in robotics is because I think that cognition is fundamentally an embodied thing. In other words, our ability to think and reason is very much tied up with our bodily abilities to interact with the physical world which is full of objects that move and have shape and so on. So nature developed cognition for us in order to make us better at interacting with this physical and social world. So if we want to understand the nature of cognition better then we really need to understand it in an embodied context. In the context of something that moves around in the world; that interacts with objects. And so hence my interest in robotics.
GM: Does that mean for you that the endpoint really is trying to make sense of what makes us tick in a cognitive context? It’s not necessarily to build better robots for the sake of building better robots you just want to know how people like me and you and six billion other people actually tick?
MS: That’s in fact exactly what I’m interested in. I’m really interested in the fundamental scientific questions about ourselves and about the nature of our own brains. How our brains enable us to do the kinds of things that we do. The kinds of things that we think of as quintessentially human. At the same time there’s always the feeling that, of course, if we can understand cognition in this way, by building robots, then at the same time we’ll be building potentially very useful artefacts. That’s the way you justify what you’re doing to the funding agencies.
GM: That’s the bottom line. Nobody can escape from that, who have their own levels of cognition as we all know. So what does building a robot tell you that studying humans wouldn’t? I mean in other words why don’t you just get a load of humans into a lab, obviously in an ethical way, and ask them questions, set them tests, do brain scans on them? Why build robots to find this stuff out?
MS: Yeah, well that’s an excellent question and, of course, that kind of science is exactly the way people have studied brains for many, many years. For decades; centuries. What’s really the issue for me is that I feel that if somebody really had a fundamental theory that really explained how we worked, how our brains worked, then a kind of acid test for that theory would be that it would be able to help us build things that could behave as we do.
What I’m less interested in is the kind of disembodied artificial intelligence work in which you simply have something that is a box on your desk and a keyboard and a screen and you interact with it in that way. A good example of disembodied AI in science fiction, which is always a bit of a touch-point for this kind of work, is HAL. So HAL in 2001, the astronauts interact with HAL just by talking to HAL. And HAL doesn’t have any kind of obvious body. It doesn’t move around in the same space as the astronauts that it’s talking to. And that kind of disembodied AI is not what I’m interested in.
GM: People may be reassured by that after what HAL did to that crew of astronauts in 2001. But that is interesting though, isn’t it. So where are we at, at the moment? Do we have robotic entities that are already starting to answer some of these really important questions about cognition?
MS: Well I don’t think we do. And I like to compare it to the Apollo programme, trying to get people to walk on the Moon. So when Kennedy announced that they were going to have a program where people would walk on the Moon within 10 years they were at least at the position then when they had rockets. And although the problems and challenges that they faced, the engineering problems, were absolutely enormous, but nevertheless they knew what a rocket was. Within artificial intelligence and robotics I’m not sure we even have rockets yet. We don’t know what the fundamental building blocks that we need are to make the first machine that we could make then a bit smarter and a bit smarter and a bit smarter until we reached human level artificial intelligence.
GM: Having done the interview we’ve just moved into the next office because by a very good stroke of fortune here the robot that’s at the centre of a lot of your research is currently being manipulated by some of your students.
MS: That’s right. This robot is called an iCub. So this robot is a humanoid robot. So it’s a full-torso humanoid robot and it’s about the size of a three year old child.
GM: Complete with arms and legs and a torso and a head.
MS: Yes, a full torso with arms and legs. And the hands are fully jointed. We’ve got four fingers and a thumb. And the head has stereo vision. It has two eyes. The eyes can move independently from the head. And it’s really a very beautifully engineered bit of kit that we’re very, very fortunate to be able to play with. So our agenda is to actually control this robot to do various things, various childlike things, such as being able to manipulate objects in its environment, pick things up and play with them in various ways. It’s a bit ambitious, I have to say.
GM: Fascinating to see all this. This is a real Aladdin’s cave of different bits of robots and circuitry. What I like about this is it shows how you are bringing together these really deep and wide and important philosophical ideas with really solid robot engineering as well.
MS: Absolutely, yeah. I mean it’s pretty sophisticated technology we’ve got here. I don’t think there’s anymore sophisticated robot platform in the world than the one that we’re looking at right now. We have got some very sophisticated technology here. We’re using that sophisticated technology to address fundamental and ancient philosophical questions. And you can’t do better than that, can you?
GM: Now, there’s a bombshell on which to end. Well, thank you very much to you and your students and, of course, your robots here. I appreciate it.
MS: Okay, thank you.
GM: Murray Shanahan there.
Dr Julian Asher on synaesthesia and the genetics of obesity
Well, finally then if you listened to last month’s podcast you’ll remember a very happy professor Steve Bloom in our Faculty of Medicine who’d just sold his spin-out company to a major US pharmaceutical firm for £100,000,000.00. At the centre of the deal was a novel anti-obesity molecule that could be the basis of a new generation of fat-busting drugs. But Professor Bloom isn’t the only one in the Faculty interested in obesity. Just around the corner from his office, on our Hammersmith campus, is the Department of Genomic Medicine where Dr Julian Asher has co-authored a paper recently published in the journal Nature Genetics. The subject: The Genetic Basis to Obesity.
Well I dropped in on Julian to find out more but before we talked about obesity I was also keen to hear how he’s been looking to the genome for some clues about the causes of synaesthesia, the curious neurological condition where the senses get crosswired so you end up hearing colours or, in Julian’s case, seeing and almost feeling sounds. Well just before joining Imperial, where he’s now a postdoc research associate, Julian did his PhD on the subject at Cambridge University.
Julian Asher: Synaesthesia is a neurological disorder which is commonly described as a mixing of the senses. Usually there will be a trigger in one sensory modality which then will trigger abnormal perception in a second sensory modality. The people that I study actually when they hear a sound they will then see a colour as well as also perceiving the sound. So it’s a simultaneous perception.
GM: And you’re actually a synaesthete yourself so how does it manifest itself with you?
JA: I have what we think of as non-linguistic sound-colour synaesthesia. That means that I see colours in response to non-linguistic sounds. In other words, I won’t see a colour if I hear somebody speaking but if I hear a piece of music playing I’ll see a colour.
GM: Can you describe what that’s like? I mean, supposing you hear a violin, what colour would that conjure up?
JA: For me specifically a violin is what I think of as almost a red wine colour or a claret. I also perceive a visual texture to the colour. It’s sort of a shiny very liquid looking red. As opposed to when I hear a cello I’ll get something which is almost like a deep honey gold colour. It has a viscous feel to it which does remind me a bit of honey. And it’s also got a dull sheen as opposed to a bright, shiny liquid. What’s interesting in my particular case is that most synaesthetes who have non-linguistic sound-colour synaesthesia perceive different colours in response to pitch. Whereas there are very few of us who perceive different colours in response to timbre. And I’m one of the latter variety.
GM: So of course it’s a very mysterious condition, isn’t it, and you bring your interest to this not just as a synaesthete but as a geneticist. And in quite recently published research you’ve honed down like a genetic basis to synaesthesia so just talk us through some of that.
JA: The clinical part of the project which involved patient recruitment and also diagnostic testing took place in the Department of Psychiatry at the University of Cambridge working with Professor Simon Baron-Cohen who is one of the founders of the filed of modern synaesthesia research. And the genetic portion of the project took place at the Wellcome Trust Centre for Human Genetics at the University of Oxford under Professor Tony Monaco. And we found four regions which are linked to synaesthesia. There are regions on chromosomes 2, 5, 6 and 12. So it’s much more genetically complex than was originally believed. Which makes sense from a phenotypic standpoint because the phenomenon itself is so complex it’s hard to believe that all the different manifestations could be due to a single gene.
GM: Having looked at these chromosomes it seems as if they are also implicated, or regions nearby anyway on the chromosome, in conditions like autism, for instance, and I think even dyslexia. Does that mean they’re linked to synaesthesia or should we be cautious about how much we read into that?
JA: I’d always be cautious about how much you read into that because after all none of our people who participated in our study have been formally valuated for autism or dyslexia so I’ve no way of knowing whether or not they are autistic or dyslexic. We did try to rule out anybody with known neurological or psychiatric disorders simply to make sure that we were looking only at synaesthesia.
But that’s not saying there couldn’t be an un-diagnosed dyslexic somewhere in there. And what it does say is that the neurological mechanisms underlying the conditions could be similar in that you’re looking at, particularly with autism, increased connectivity in the brain and that also has been observed in synaesthetes. And it’s interesting that the region on chromosome 2, which is the most strongly linked region, has been connected quite strongly to autism.
I was involved in a case study of a young man who has both Asperger's Syndrome, which is the high functioning form of autism, and synaesthesia. He’s a memory savant. He’s memorised Pi to over 22,000 digits. And that’s commonly been associated with autism however what we found when we actually examined this person, explored his memory abilities in further depth, was that that’s actually driven by synaesthesia. So now we’re looking more closely at the savantism phenomenon and thinking maybe the reason that not all autistic people are savants is that they require this unique combination of not only the autistic aspect which enables them to focus intensively on something like the digits of Pi but the synaesthesia which actually plays a role in the recall. And we have decided to follow up on that. And the group at the University of Cambridge is now looking at a large number of people with Asperger’s Syndrome and actually testing them for synaesthesia to see whether or not there is actually an increased incidence of synaesthesia in that group.
GM: Where do you ultimately hope it will lead? I mean, to a cure? I don’t know if I can use that word but to a cure for synaesthesia? Or just a deeper understanding of the condition?
JA: Well there are a couple of directions we hope it could go in. It’s interesting that even synaesthetes who do actually have quite a bit of disability from their synaesthesia would not give it up, if you asked them, if they had the choice to. However, I do think one area where it could be very valuable is if we could develop a genetic test for it. Because one of the keys to enabling children with learning differences to succeed is early diagnosis.
Because synaesthetic children are usually average or above average in intelligence but they may need specific adaptations in the classroom to enable them to succeed. And it’s very important not to create a situation where they become accustomed to academic failure. Whereas in a properly adapted classroom environment they do very well. And there’s no reason for that not to happen.
GM: Just before we leave this wider topic of genetics I know that your own research has moved on and you’re interested in the genetic basis for obesity as well. Just tell us about some of that.
JA: Yes, I’m working in the laboratory of Professor Philip Froguel and we’re studying the genetics of obesity. There’s also people in the group who work on the genetics of diabetes. It’s com monly believed that everybody has two copies of every gene and we’re finding that that’s actually not true. That most people, actually virtually everybody, has some level of copy number variation. So you can have no copies of it. You can have one copy rather than two. You can have the normal two copies. You can have three or four or even more copies. And we’re looking at whether or not the number of copies you have of a genetic region plays a role in increasing your susceptibility to obesity.
GM: So I suppose intuitively you’d think the more copies you have of a particular genetic attribute the more likely it is to have some manifestation or it will be expressed in some way. Is that what you’re finding with obesity?
JA: That is logical. Assuming that all of the copies actually are expressed, you know, that you would then have higher levels of whatever the product of that, assuming that it’s a coding region, whatever the product of that region is.
GM: Does it mean that overall obesity remains a fairly misunderstood condition when it comes to the genetics behind it?
JA: I think in terms of the genetic basis of common obesity as opposed to the rare monogenic obesity syndrome it is an area where a lot of work still remains to be done. And what’s particularly interesting is one of the findings that our group has come out with, along with other groups, is that obesity is really under our endocrine disorder rather than a metabolic one. And it seems that the control of appetite is actually one of the key determinants in whether or not people become obese. So many of the genes which play a large role in obesity are actually expressed in the brain rather than in the GI tract and play a role in appetite and food intake regulation. It is definitely an area of interest, particularly in terms of future drug targets or treatments, particularly for the most obese people who find that diet and exercise are not sufficient to treat their problems.
GM: Julian Asher on the genetics of obesity and also synaesthesia. And that’s it for this month. Hope you’ve enjoyed this edition. If so, tell your friends about it and keep up to date with what we’re doing on this podcast by visiting our Facebook page. Just type “Imperial College podcast” into the search box in Facebook to find us. And you can follow me on Twitter too, if you fancy. I’m Gareth M. Oscar Buldum wrote this theme music. And of course this podcast is a collaboration between the Imperial College Press Office and the Science Communication Group. I’ll have more for you in April but until then have a great month and thanks for listening. Bye for now.