TODAY'S NEUROSCIENCE, TOMORROW'S HISTORY
A Video Archive Project
Professor Sir Michael Rutter
Interviewed by Richard Thomas
Supported by the Wellcome Trust, Grant no: 080160/Z/06/Z to Dr Tilli Tansey, Wellcome Trust Centre for the History of Medicine, UCL, and Professor Leslie Iversen, Department of Pharmacology,
Interview Transcript
Early training and influences, the Maudsley Hospital, London
At the time I came to the Maudsley, all but fifty years ago, most of us had trained in something else before we came into psychiatry and I was advised to do that. And I did, and trained in neurology and general medicine, and as it happened, also a period in paediatrics, although that was sort of almost accidental. As it turned out, that was a very useful experience to have had. So the context was different. The ethos here was definitely of questioning -- that was the order of the day. Aubrey Lewis, who dominated British psychiatry at that time and certainly dominated here, loved intellectual argument; he loved questioning. It had to be done really well but he was not a one for accepting theory or accepting authority. It didn’t matter what the prestige was of the person who said it, the question was, ‘What was the evidence, and how would you know?’
And that was a model that I adopted, like most of the people in the social psychiatry unit. So I had contact with him during my clinical training but then also in the research unit. In the research unit, Neil O’Connor and Beate Hermelin were the two experimental psychologists concerned with autism but who brought an experimental way of thinking, which made a big impact on me. I never worked directly with them but they certainly were mentors. Jack Tizard was also a mentor -- another psychologist -- and he brought me in on the Isle of Wight study and was very generous in giving me the responsibility for taking it forward; another iconoclast, but also somebody who … to whom it was very important to use research to deal with practical matters and policy issues. And he did that in a very innovative way, and that made an impact.
Well, Aubrey Lewis had decided that I’d do child psychiatry although I actually wasn’t very keen on the idea at the time, but he was right and I was wrong in that he did recognise that it: (a) played to my strengths, and (b) the research I’d already done was relevant to that. And he made two stipulations, one of which surprised me, and one of which I welcomed. The one I welcomed was that I had to learn about child development. The one that surprised me was that I should not be trained in child psychiatry at all because he said that firstly, the training was indifferent quality, which was correct at that time; but secondly, he thought that it was likely to inhibit creative thought, and he wanted me to develop the subject as a somebody who’d bring originality to it.
The need to question your own research
One of the things in science is that you need to question your own findings, your own ideas, ruthlessly, and to put it in a flip way, what you want to do, if you’re being ambitious, is to not just change your mind but you’re the person who shows that it was necessary rather than your competitors. And you show what can be done to build on that.
So, yes, an example, would be that in my Maternal Deprivation Reassessed, I said that the evidence showed that there was, with a few minor exceptions, very little in the way of longterm effects of early experiences that were independent of later experiences. Now factually, that is still correct but what I had overlooked is that that was the wrong way of putting it, because it supposes that early environments are independent of later environments, and the evidence from our own research as well as others, indicates that’s often not the case; that early experiences shape later experiences and that you do get important long-term effects, not because of independent effects, although we are aware now that those are more common than we thought, but because of chain effects -- one thing leading to another. So, yes, I did change my mind on that. I’m sure I was right to change my mind on that.
Epidemiological Psychiatry defined – the study of Romanian adoptees
Well, epidemiology is the study of the distribution of either disorders or traits in the population, and it can be used in two sorts of ways, both of which we used on the Isle of Wight. One is giving you information for the planning of services, in other words, a very practical role for epidemiology but the other is using the evidence on associations between risk factors, and in our case, psychopathology, as a way of getting ideas on causal mechanisms. And so I was keen from the outset to be able to use epidemiology in this sort of way, aware of course that observational data, you’ve got to be questioning of as to whether the inference of causation is justified or not.3
All people in behavioural sciences learn early on that you can’t infer cause from a correlation or an association because it may be a coincidence, it may be due to selection into the groups that are associated -- a whole variety of reasons. But you can get much closer to a causal inference, provided that you have an experimental approach. Now sometimes you can actually use experiments but of course in mental disorder, many of the risk factors you can’t manipulate experimentally. It would not be ethical or practical. So one of the things that over the years, that I’ve been very much involved with, is finding ways of using natural experiments as a way of testing the possibility that it is involved in a causal pathway. Of course with multi-factorial disorders -- disorders due not to a single cause but a lot of things coming together -- there isn’t a basic cause, but on the other hand there certainly are factors that make a huge … that play a huge part in the causal pathway.
For example, the study of Romanian adoptees – children who’d suffered profound deprivation in Romania. There are two sorts of ways in which one could test the causal inference. The first is, if the environment changes, does their behaviour and their development change? And the answer was clearly yes. There was a dramatic catch-up – not complete – but very substantial. So that made it clear that something about the conditions before they came to this country, i.e. in the depriving institutions, had caused their problems. But then, you’ve got another way, which is that some of them had deficits; they didn’t fully recover. They had problems of various kinds. So then the question is can you show systematic relationships between their pre-adoption experience and their outcomes, and what we have found is, yes, strong effects. And, moreover, effects that are as powerful at age eleven as they were at age six as they were at age four. So in this way, you have a fairly powerful inference that in some way these are involving causal pathways and so the skill, if you like, both the art and the science of using epidemiology in this way, is thinking of circumstances that will test.
Autism and the MMR vaccine
To give a very different example: there was a lot of hoo-hah in 1998 of a paper that claimed that the vaccine MMR caused the epidemic of autism. Well, it occurred to me that one way of testing that was to see if there were countries that had stopped using MMR where others had continued using it, and it turned out Japan had, for reasons that were actually unconnected with the measles part of the vaccine. And so the question is: what happened to the rate of autism in Japan after MMR was withdrawn? And what our findings showed is, that the rate continued going on up -- that the withdrawal of a supposed risk factor had not made any difference and if anything, it had got worse, not better. So here was a way of testing, and in this case, disproving an environmental mechanism. In other cases -- as with the institutional4 deprivation -- of supporting an environmental mechanism. So there are lots of examples of that kind, and that has become very much something that has remained an abiding interest and something I’ve tried to take forward.
Maternal Deprivation – relationships and attachments
Maternal deprivation was initially seen by Bowlby in terms of separation experiences, and one of the things that came out of the review that I did was that although that may be a part of it, actually it isn’t the key thing; that one needs to think about deprivation in terms of either a disruption of a relationship or an interference with the functioning of that relationship and not simply with separation. And so one of the things we did early on was to look at whether the effects on mental … the risk of mental disorder were the same for parental divorce and parental death. And interestingly, the effects are much stronger on divorce than on death. So death is, as it were -- a much more permanent separation obviously -- but did not involve the conflict and discord that seems to be the key driving factor.
I agreed with Bowlby more than I disagreed but I did disagree on several key things. The first was the notion that separation was the key thing, and he came actually round to my view on that, partly as a result of his own study of children admitted to TB sanatoria and who were separated. The other was the focus on there being just one attachment, whereas I think the epidemiological evidence indicates that most children form multiple attachments. Now, it's not to say that the main attachment isn’t particularly important but it is to say that the norm is several, and it is probably adaptive to have more than one. There’s a wonderful book looking at this called Kith, Kin and Hired Hands which is making the point that throughout history and across all societies, it is very unusual not to have people other than the mother involved in the care of the children. Sometimes done this way, and sometimes that way. And I also differed in that it seemed to me psychoanalytic theory was completely hopeless in dealing with this. Bowlby initially gave a lot of credit to psychoanalysis but his theories on attachment were greeted with extreme hostility of a personal kind and later, in his … I think probably his last book in 1988, he said that psychoanalysis was never more wrong than in its theory of development. So he and I were actually pretty close together. He laid emphasis on the value of psychoanalysis in forcing one to think about feelings and I’d agree with that, but he agreed with me in the end, that the specifics of that were just totally wrong.5
Lost relationships – coping, resilience and genetic factors
People nowadays don’t use the terminology ‘deprivation’ very much but I think the findings have stood the test of time, and that dealing with acute stress responses and with anti-social behaviour reflect probably two rather different mechanisms, so that because we are social animals, we respond with distress to loss. This is true in infancy and it's true in extreme old age, that relationships are very important to us. So that the effects of either rejection or death of a loved one is a stress, and we now know that this is a pretty general finding and we have some beginning understanding of some of the biology that’s involved in that.
There are a range of strategies that have been used in research to try and look at this, and to take three different examples: one is that one needs to think about it in terms of coping. That’s to say, you have physiological coping and you have psychological coping, so that the neuroendocrine response to parachute jumping is quite different in experienced jumpers than in people making their first parachute jump. They have adapted and their bodily systems have adapted so it is no longer the same stress it was first time. and there are plenty of other examples of a similar kind of thing.
Now that leads on to the second mechanism or related mechanism, which is that most of the human resilience research has focused on avoiding stress and adversity or diminishing its impact in some ways. But if one thinks about this in a biological sense, that’s the wrong way to think about it, so that if you want to protect people against infections, you don’t put them in a cocoon and stop them ever having contact with bacteria and viruses -- you expose them. But you expose them in ways that they can cope with. That may happen in terms of natural immunity or, of course, it may happen through vaccination. So the psychological equivalent is to say: what could we do to enable children to cope successfully with hazards? Because challenges, stress – that’s part of growing up and you have to learn to cope, and the only way you learn is through exposure.
The third mechanism is where the genetics comes in, where we know that the response to the acute stresses -- and the Dunedin work with Caspi and Moffitt, for example, shows that genetic factors play a role in influencing susceptibility to the environments. So one needs then to be looking at the genetic pathways that are involved in this, either increasing risk or increasing protection according to which end you’re looking at.
So resilience is a real phenomenon but the ways of looking at it immediately bring you into biological research and they bring you into – well, I’ve mentioned three different sort of mechanisms but there are of course more. What we know much less about is the mediating role of neuroendocrine factors; that there are neuroendocrine effects of a very important kind is not in doubt but do they account for the behavioural effect? We don’t know. They could, but that is research still to be completed.
Autism – a brain disorder with important genetic factors
The prevailing view in the sixties was that autism was an unusually early variety of schizophrenia and that it was due to poor parenting and other psychogenic causes. And the research that I’d been involved with cast doubt on that in indicating that the differences from schizophrenia were immense, and that it seemed very unlikely that it was anything to do with schizophrenia. But also the research had cast doubt on the notion that this was a psychogenic disorder, and our earlier follow-up study had provided really basically the first evidence that it was a neurodevelopmental disorder.
The follow-up study was important, I think, because we had excluded the children who had got a known neurological disorder but in spite of that, about a quarter of them developed epileptic attacks, epileptic seizures, in late adolescence. And that was, in a way, the first clear indication that there was a brain basis for this supposedly psychogenic disorder, and our research also showed that cognitive deficits played an important role, and language functioning similarly played an important role. So that what I was part of at that time was a paradigm shift from autism being a psychogenic disorder to autism being a brain disorder, and that this began very early. It's got nothing to do with schizophrenia, and a little bit after that we also did the first twin study, Susan Folstein and I.
And incidentally, that would be another example where I changed my mind because I wrote a paper in the mid-sixties concluding – along, I have to say, with some very distinguished geneticists, that it was unlikely that genetic factors played any major role in autism. And my reasoning, like theirs, was on the basis that the rate of autism in siblings, in brothers and sisters, was quite low -- below five per cent. I’d no sooner got this wretched paper published than I realised that was a stupid inference. The facts are correct, but the real focus should not be on the absolute low rate but the extremely high relative rate. So yes, five per cent is very low, but compared to what at that time was estimated as the four per ten thousand in the general population, that suggested not only that genetic factors were important but they were hugely important. And the twin study, although based on very small numbers, showed that was the case. It also involved, what at the time proved quite controversial, was the notion that the genetic liability extended beyond the handicapping disorder, and geneticists were very reluctant to accept that. It is now of course mainstream. So it was a very exciting time of having findings that, as it were, brought in the biology and brought it in in a way which is testable and could lead on to understanding mechanisms.
Autism – tests for diagnosis and measurement reveal a wider view of the disorder
At the time I entered child psychiatry, one really had no idea what to do with these kids, and indeed, didn’t know much about what to ask the parents. So one of the things that my colleagues and I did was develop a set of instruments for interviewing parents on the one hand, and observing the young people on the other, so that with the observation, what we set about doing was to devise a set of social presses i.e. situations in which there was an expectation for social interaction and for communication, in order to see how the young people responded to that. So it was a sort of experimental or quasi-experimental approach to this, and those have now become sort of gold standard instruments used in autism research.
So the interplay between the research and clinical work, I’ve always had, always enjoyed, always learned from, and that has tied in with longitudinal studies, so one of the real interests is following these kids into adult life. And indeed, I’ve got somebody who the family want to see me about because I first saw him as a child but in three years time he will be retiring at the age of sixty and they want to consider the implications for – he’s actually very independent, but he needs support. So dealing with those sides of the family, I’ve also enjoyed doing. So this is where the psychosocial and the biology, the clinical and the research all come together.
All measurement instruments, of course, are driven by the concepts of the day. How could they be otherwise? But you want to develop measures that are also sensitive to the unexpected, and with the Autism Diagnostic Interview we tried to do that. It was developed in relation to a somewhat narrower view of autism than would now prevail but our own findings with the Autism Diagnostic Interview, and the research that used it, made it clear that the boundaries went much wider. And our genetic research similarly. I was mentioning with the first twin study, one of the things that people resisted was the notion that the genetic liability went beyond the traditional diagnosis of autism. It was either much broader or it was dimensional, and we still don’t know quite which is which, but it's certainly a lot broader than the traditional handicapping disorder. So the genetic evidence and the epidemiological evidence clearly pointed to the need for a broader concept, and that’s no longer controversial. What is tricky though is knowing where the boundary lies, so clearly autism does not account for all social problems. I mean, we are social animals and that means that it’s likely that almost any mental disorder will impinge on social functioning to some degree or other. So, the challenge now, which we’re trying to engage with, is how can you tell which ones are associated with autism and which one’s aren’t? We don’t have firm answers on that although we certainly have some leads.
Autism – degrees of severity
Well, the notion of a two-hit mechanism so far as autism is concerned, arises from the fact that we know that in the families there’s an increased rate of what has come to be called the broader phenotype, meaning autistic-like abnormalities of a much milder kind. Now, they are very like autism in all sorts of ways but they differ in two ways: one is that the broader phenotype is not associated with epilepsy and it is not associated with mental retardation. And so one of the questions is: given the evidence that the broader phenotype is due to the same genetic liability as autism, why do some people have the full picture and some only these milder problems? Could there be… well, let’s take two possibilities: one is, it’s simply the level of genetic liability. If you have more of it you get the full thing; if you have less of it, you get the milder thing, or is it a two-hit mechanism in which the liability extends very broadly but something carries you over that threshold to develop full autism? Now, this is not a notion that is by any means restricted to autism because there’s exactly the same debate in schizophrenia. So that we know that the so-called prodromata of schizophrenia are much commoner than actual schizophrenia -- involving delusions, hallucinations, thought disorder and so on. And so the question is, we know that the liability to both is genetic in large part and that it seems to be the same, but what carries people over into the full syndrome? And is that some new stimulus of a kind that hasn’t been recognised? There are various possibilities that are being looked at and we don’t know what the answer is as yet.
It's proved really frustrating not to have an answer on the genes as such. We have loads of leads but so far nobody has got good evidence that the actual -- well, it won’t be the actual -- but one of the genes has been identified. It's puzzling because it is a strongly genetically influenced disorder. We have good measures; we have good strategies, so why on earth has it proved so difficult? And there are a variety of reasons that could be put forward. One is that it is the rule in medicine that there is a genetic heterogeneity; that’s to say the particular pattern of genes that is responsible in one person isn’t the same as in another. So that makes it much more difficult to tie it down, and it may also be that you are reliant on an interaction with some environmental risk factor. Because we haven’t got a good handle on that, that’s made it more difficult.