SCIENCE AT THE SHINE DOME canberra 2 - 4 may 2007

Symposium: Development and evolution of higher cognition in animals

Friday, 4 May 2007

The Rutherford Memorial Lecture, The Royal Society – plenary address

Professor Sir Patrick Bateson
Professor of Ethology, Cambridge University, UK

Patrick Bateson is Professor of Ethology, the biological study of behaviour, at the University of Cambridge (1984-2005). He was Provost of King's College, Cambridge (1988 to 2003). He was formerly Director of the Sub-Department of Animal Behaviour at Cambridge and later Head of the Department of Zoology. He was Vice-Chairman of the Museums and Galleries Commission and in 2004 was elected President of the Zoological Society of London. He was elected a Fellow of the Royal Society of London in 1983 and was its Biological Secretary and Vice-President from 1998 to 2003. He was knighted in 2003. He is a member of Sigma Xi and a foreign member of the American Philosophical Society. His research is on the behavioural development of animals, and much of his scientific career has been concerned with bridging the gap between the studies of behaviour and those of underlying mechanisms, focusing on the process of imprinting in birds. He has also carried out research on behavioural development in mammals, particularly cats, and has supervised field projects on mammals in East Africa. He conducted a research project for the National Trust on the behavioural and physiological effects of hunting deer with hounds. He has written more than 260 scientific papers and book chapters on imprinting in birds, the development of play in cats, the development and evolution of behaviour, neural mechanisms of learning, and the conceptual and methodological issues in the study of behaviour and animal welfare. He has also written articles on co-operation, the ethics of using animals in research, and the hunting of red deer with hounds. He has edited 15 books and is co-author (with Paul Martin) of Measuring Behaviour: Cambridge University Press; and Design for a Life: How Behaviour Develops. London: Cape.

Cognition and instinct

(Click on image for a larger version)

Rutherford was, of course, a Kiwi – although I notice that the Australians rather like to adopt Kiwis and pretend that they are really Australians, which annoys the New Zealanders enormously!

Among the things which made Rutherford a great scientist was his insight into the nature of physics. He also exhibited a certain arrogance when he claimed that ‘Science is physics and stamp-collecting’.

(Click on image for a larger version)

This is another famous expression of Rutherford’s: ‘If your experiment needs statistics you ought to have done a better experiment.’ Biology would be pretty much dead if we had taken that seriously.

(Click on image for a larger version)

The issue that I want to talk about today is one which binds together the fundamentals of animal behaviour and human behaviour and what we think about as our intelligence or our cognition. When you look at animals, you see sometimes that they do very complicated things. A famous example of this comes from the work of Wolfgang Koehler on chimpanzees.

At the top of this slide you can see a banana hanging down, and the chimp is piling up boxes to get nearer to the banana. In the panel to the right you see that the chimp reaches the banana. And Koehler, in his work, also found that the chimps would use sticks, as in the bottom left panel, to get at the banana, and sometimes they would slot sticks together.

The issue was whether they had had an opportunity to practise this beforehand, but nonetheless it looked like something very interesting and intelligent.

(Click on image for a larger version)

The issue, then, is where this insight learning might have come from. W H Thorpe, who founded the laboratory where I spent a large part of my working life, attempted to define this as ‘the sudden production of a new adaptive response not arrived at by trial behaviour or the solution of a problem by the sudden adaptive reorganisation of experience’. Later in his book he goes on to suggest that insight, the ability to perceive these interrelations, might be instinctive.

This raises a whole barrel of worms which I want to spend a bit of time on.

The question is what is meant by instinct, and the answer to that is: lots and lots of different things.

(Click on image for a larger version)

One of the meanings which is usually applied to innate behaviour is that it is present at birth, or at a particular stage of development.

Another meaning is that it is not learned; that is probably the commonest meaning that is applied to it.

Another meaning is that it is in some senses genetic, it is highly heritable – you will find it being passed down from one generation to the next.

A fourth meaning is that it was adapted during evolution.

A fifth meaning is that it develops before it has a useful function. I shall give you an example of that in a moment, where a fully formed behaviour pattern is shown before it actually serves a useful function.

Yet another meaning is that it is shared by all members of a species or sex or age group.

A seventh meaning is that it is part of a behavioural system. That is to say, it is to do with hunting or with cleaning the body or one of the many other aspects of what animals and humans do, and that an instinctive system is a system which binds all the functionally related features together.

Yet another meaning, which is used much more in contemporary neuroscience, is that an instinct is controlled by a specialised neural module. It is specific to a particular domain and it is in a sense encapsulated from other kinds of information. A famous example of this would be the capacity that we have for recognising faces. Some people, who have a tumour in a particular bit of the brain or have had this bit of the brain destroyed by injury, cannot recognise faces. They can perform all sorts of other recognition but they can’t recognise faces. And of course it’s very disruptive to their social life!

Finally, another meaning which is used increasingly now, and comes from developmental biology, is that instinctive behaviour is developmentally robust – that is to say, it is unaffected by disruptive experience. It is very well canalised, to use a term that Waddington used.

The point about all these different meanings is that they don’t necessarily covary. If you get evidence for one, it doesn’t necessarily mean you get evidence for another. Anyway, I will give you a couple of examples of things that people have used for these terms.

(Click on image for a larger version)

This comes from work by Niko Tinbergen, who was a Nobel Prize winner with Konrad Lorenz and Karl von Frisch. Lorenz, who was originally trained as an anatomist, was very interested in the fact that you could take behaviour patterns from closely related species and build up a kind of taxonomy based on behaviour. This example from Tinbergen’s work was done on the gulls, and I have a very soft spot for this because the first research work that I did when still an undergraduate was on the Ivory Gull in the high Arctic.

What you see here are two different species of gulls, in the course of a threat display, showing very similar kinds of movements. Tinbergen, and Konrad Lorenz certainly too, would have regarded these commonalities as forms of instinctive behaviour.

(Click on image for a larger version)

Another striking example is the kind of grooming that you see in rodents. Very shortly after they are born, they will go through highly stereotyped grooming movements. Anybody who has had a pet mouse or hamster will have seen these very stereotyped grooming movements, but the rodents do it very early on. Even before they have any fur they will go through these grooming movements. This is an example of a behaviour pattern which is shown before it actually serves any function.

(Click on image for a larger version)

Now, when you compare different rodents you find that the duration of an elliptical grooming movement round the face is strongly related to the body mass of the animal. So a mouse, which is small, moves very quickly; a guinea pig, which is much larger, moves more slowly. Lord Rutherford might have thought that this was all a matter of physics, that it was all to do with mass and of course you would expect a larger animal to move more slowly. It’s not true, in fact. When you look at the infant rat, you find it does it at exactly the same rate as the adult rat, even though it is much, much smaller. And the same goes for the guinea pig, namely that the baby guinea pig does it at exactly the same rate as the adult. So it’s nothing to do with mass; it is to do with the organisation of the motor patterns.

It is another striking example of the kind of thing that people refer to as being instinctive.

(Click on image for a larger version)

I have done quite a lot of work on the process called behavioural imprinting. We worked on chicks which had hatched out in the previous 24 hours, putting them into wheels which would record how much they were approaching an object. We had a flashing light that rotated and flashed red – a highly conspicuous object.

When you put into this wheel a chick which is about 24 hours old – it has been in a dark incubator, hasn’t seen anything before – it is very strongly attracted to the flashing light and may, in the course of an hour, try to approach for as much as the equivalent of a kilometre. It is very, very highly motivated.

(Click on image for a larger version)

We also used in our experiments a stuffed jungle fowl, which is the ancestral species of the domestic fowl. This stuffed fowl just rotated but we found that it was actually more attractive – as I will show you in a moment – than the flashing light, so long as it had a head and a neck. If you took off the head and the neck, then it wasn’t any different from the flashing light, but the head and the neck made the fowl tremendously attractive to the chicks.

Giorgio Vallortigara will talk more about some of these predispositions that chicks have. For my purposes I just want to illustrate a point about imprinting, which is that you get an interesting interplay between these initial predispositions by the naïve animals and the learning process which takes place. They narrow down their preferences to the object they have been exposed to.

(Click on image for a larger version)

We can calculate a preference ratio, which is equal to the amount that they try to approach the jungle fowl, multiplied by 100, divided by the amount that they try to approach the jungle fowl plus the amount that they approach the red box. And you can then get an index. If it was 100 per cent – which it never is – it would indicate a very, very strong preference for the jungle fowl. If it was zero, it would indicate a very strong preference for the red box. And if it was 50 per cent, there would be no preference.

(Click on image for a larger version)

When we look at what happens when we take naïve animals and give them a choice – successive choices, not simultaneously – what we find as shown by the first bar on the slide is that it is about 60 per cent. In other words, they are showing that they prefer the jungle fowl a bit. (These are animals which have never seen anything before. They have been kept in the dark, and then they come out and they are given this choice.)

If we expose birds either to the jungle fowl or to the red box for an hour, then we get the result shown by the second bar. After 60 minutes exposure to the jungle fowl, the preference goes up, of course. They are showing a stronger preference for the familiar object – that relates to imprinting. And if they’re exposed to the red box, it goes the other way – they get a stronger preference for the red box. And the two developmental processes, the ones which determine the initial predisposition and the process which determines the learned preference, appear at least here to add together.

(Click on image for a larger version)

What you see is, effectively, a fairly broad range of things which initially attract the animals. And then if you imprint them with one example of that broad range, they narrow down their preference to that, so they have a strong preference for what they have seen.

Anyway, the point from this kind of work is to show that you can have something which initially doesn’t appear to be learned – it obviously has to develop but it doesn’t appear to be learned – and that then works together with the learning process.

(Click on image for a larger version)

So, if one asks the question, ‘Are there two types of behaviour?’ as some people do, and they say that experience affects acquired behaviour and genes affect instinct, the answer has to be that it doesn’t. That would not be a useful way of thinking about it.

(Click on image for a larger version)

People still feel that somehow they ought to be able to divide up the sources of variation in behaviour. There has been a big discussion in the developmental literature about whether or not one can partition the influence of the genes and the influence of experience. It is incorporated in a population concept called heritability. I’m afraid it’s a concept that is quite confusing, because it has at least four different meanings. One of them is the common-sense one that it is covariational, that your children tend to be like you in some respects or there will be inheritances of that kind. That is the commonplace notion of heritability.

Then there is one that comes from evolutionary biology. If the characteristics of the organism change rapidly in response to a strong evolutionary pressure, then that is called a selectional heritability. Then there is one which comes from population genetics, a narrow term which is that the additive variation due to the genes is high, or a broader one which is that all the variation due to the genes is high. I will illustrate this point with an uncontroversial example.

(Click on image for a larger version)

If you take something like human height, and compare the man in the middle of this photograph, who comes from northern Europe, with two African pygmies, you see there is an astonishing difference in height. I think nobody would argue that some of this must be due to genetic differences.

(Click on image for a larger version)

It is also the case that human height has been undergoing extraordinary change. If you take the UK adult population – people tend to measure height because it is easy to measure – you find that the adult height over a period of 100 years or so was going up at a rate of about 3 cm a decade, which was a really quite striking change. It has levelled off recently. Interestingly, in women, where there wasn’t such a big secular trend in the first 50 years over these periods of measurement, they are now starting to catch up. So female height changes, which were initially about a third of the male increase, are accelerating and we now start to see a lot of very tall young women around.

The point is that here we have an example where clearly differences between human populations can contribute to height, and also differences in nutritional quality are probably making a big difference. And so there has been an attempt to partition these things.

(Click on image for a larger version)

The way it’s usually done is to say that we have a high genotype (let’s say the northern European population) and we have a low genotype (say, African pygmies). We also know that if we vary the nutritional environment, this will have an effect on height. You can measure the statistical variance due to the genotypes and the total variation of the population. The notion of heritability is that it is the genetic variance divided by the total variance, and so that is the kind of ratio which you apparently get out of the data. And this ratio is used a lot, even in aspects of behaviour which are most controversial, namely, those to do with higher cognition. You quite often see people quoting heritability scores for these most complex features of our behaviour.

There are several things that are problematic about this, and I will mention them very briefly. Clearly, this ratio will depend on the range of environments you sampled, and the bigger the range of environments the lower the ratio. It will also depend on the population you used, and of course if you compared Swedes with African pygmies you will get a big genetic variation and that will bump up the heritability ratio.

(Click on image for a larger version)

But there is a much more fundamental point than that, which is that you may get a non-additive interaction between the sources of variation. In work on development some people believe that everything is additive, so that when two developmental processes combine you can always somehow pluck out the original components – in this case, the red and the white component. But there is an alternative view, which is actually, I think, what most developmental biologists would now favour, namely that you get some interaction and that instead of being able to pluck out the sources of variation from variations in the final products, you get, in effect, pink and, unlike this example has properties that could not have been easily predicted from the variation in original components..

(Click on image for a larger version)

To go back to the issue of genes and environments, in terms of how they might affect any calculation about heritability: if, for example, under very poor conditions all people are roughly alike, but under very good conditions they diverge very markedly, then there is no way in which you can calculate a sensible ratio which applies across all conditions.

The point I would just alert you to is that when people talk about heritability you should be very suspicious, because it assumes no interactions. We know there are interactions, and there is a whole range of examples of this that come from all round the animal kingdom.

(Click on image for a larger version)

What you see is the individual’s phenotype affecting which genes are being expressed and also affecting the kind of environment it moves into, and that kind of interplay between the individual and the environment, and of course its genes, is what we really need to study in development. Wonderful examples come from the variability that we find in natural populations.

(Click on image for a larger version)

Here is a little freshwater crustacean called Daphnia. If the mother Daphnia was exposed to the smell of a predator – the predator is a midge larva – then her young are very different from those young born to a mother in a predator-free environment. They are born with spikes on the head and on the tail. They have this elaborate helmet which makes it more difficult for their predators to eat them.

There is a cost to the Daphnia when it does this, which is that the females don’t produce so many eggs if they have had to produce all this armour. Nonetheless, if you are going to reproduce you must first survive, and so there is a trade-off between survival and reproduction here. Anyway, it is a wonderful example of how from one genotype you are getting very different kinds of phenotypes.

(Click on image for a larger version)

Essentially, these phenotypes are matched to the environment in which the individual is going to grow up – usually, but not always as you will see. This matching is essentially the way in which such things evolve. And it depends on an interaction between the organism’s genotype and its environment.

(Click on image for a larger version)

Here is another wonderful example, which we have known about for years. All these ants, which are involved in cutting leaves which they bring back to the nest, where the leaves are fermented, are sisters. And they all have very specialised jobs in collecting the leaves, bringing them to the nest and tending to them in the nest. The difference between them depends on the food they received. So again from the same genotype you get very different types of phenotype. (These are very closely related, because they are all sisters.)

So we are very alert now to this kind of thing, that you get this very strong interaction between how an individual is growing up and the kinds of environment to which it is being exposed.

(Click on image for a larger version)

Here’s another wonderful example, which comes from lovely work done at the Liggins Institute, in Auckland.

The mother of the rat pup on the left was fed as much as she wanted; she had food all the time. The mother of the rat pup on the right had 70 per cent rations. And so you can see there is a big effect on the little pup on the right.

All is fine, the little pups are very healthy and survive very well, so long as they are on rationed food. If they are given a rich diet then they become hyperphagic, they start eating enormously.

(Click on image for a larger version)

And they get like this – tremendously overweight. The other ones whose mothers were given as much food as they wanted don’t, and so it is very much dependent on what they were exposed to when they were young. This kind of example now is being used increasingly to look at what has been happening in humans, where again we know that the human fetus is influenced by the mother’s state.

(Click on image for a larger version)

In the case of that rat we can get bad cases, we can get a mismatch between the phenotype of the individual and the environment in which it ends up. If the little rat grows up in an affluent environment, it ends up tremendously obese, and we are starting to see explanations for human behaviour in exactly the same terms.

(Click on image for a larger version)

The fetus is sensitive to maternal condition, and if the mother is nutritionally at a low level she will have a small baby, just as in the case of the rats. If she is well fed she has a big baby.

Now, if the big baby grows up in a world of plenty – which I have characterised here by a hamper full of food – then the child develops into a healthy adult and a healthy elderly person. There is no problem about that for that individual. Of course, other things might hit her, but as far as her phenotype is concerned she is okay.

If a small baby is exposed to affluent conditions – and this is the really interesting public health point – they are enormously at risk from heart disease and diabetes. This is a very big problem now in countries like India and China, where there is a big new middle class who were born small but then grew up in very affluent conditions.

The interesting thing is that if the little baby grows up in lean conditions, it develops perfectly happily and enters a perfectly healthy old age. And if the big baby enters conditions where there is very little food around, as in a famine, it is enormously at risk from rickets, which has a very big reproductive cost, particularly for girls.

Clearly, there can be a mismatch between what was forecast by the mother and what happens to that child. But here again we see an extraordinary interaction between, in this case, the child’s genotype and the conditions that it was exposed to during pregnancy. This has enormous public health implications now which people are becoming very interested in.

(Click on image for a larger version)

Given all these things and given the dynamics of development, could we have some sort of sensible taxonomy of innateness or instinct? Or is it the case that all combinations of developmental processes can in fact be found in practice? In a long article by Matteo Mameli and me in Biology and Philosophy we discuss this.

(Click on image for a larger version)

We take all the various definitions of innateness and ask, ‘Supposing one was able to look at the frequency of occurrence of different types of behaviour, and also to produce a kind of innateness score’ – using those nine definitions which I presented at the beginning of the lecture – ‘could you find some bimodal distribution?’ It might look like the blue curve shown here.

If that were true, if you tended to get clustering at either end of that score, namely forms of behaviour that tended to have all the characteristics of innateness and forms that had none of them, then there would be some reason for thinking that the dichotomy was justified.

However, if, as I think is probably much more likely, you don’t find any dichotomy of that kind, that is the death of any kind of notion that you have instinct and non-instinct. It really hits the dichotomy on the head.

We don’t know fully about this, but my strong hunch is that the second case is the correct one.

(Click on image for a larger version)

It is probably the case that we have strong ideas about innateness and instinct which just seem to come from folk psychology. We take it in and we assume it must be right. It’s not really a scientific notion; it is a notion which we take in with our culture.

My feeling is that as scientists we should say what we mean, until the credentials of a particular concept are properly established. I think it is very important that we are clear about this, otherwise we generate a lot of confusion.

We have identified many different types of developmental processes. In the case of imprinting that I was talking about, the chicks’ predispositions develop in a different kind of way from the way in which their preferences are narrowed down by experience of the particular object. These are different processes, but they are combined. And so one can suspect that as you go round the animal kingdom, or as you look at human behaviour, what you will find is that processes can be combined in different ways according to the needs of the species in a particular context.

So, in the case of grooming the body, there isn’t much that the animal needs to learn about that, and this is probably as good an example as you can get of behaviour which has many of the characteristics of innateness. But then there will be many other things, like foraging, which involve a lot of experience and couldn’t be classified in the same way.

(Click on image for a larger version)

A wonderful comparative psychologist, Frank Beach, wrote an article in 1950 called ‘The descent of instinct’. He was perfectly aware that ‘descent’ could also be expressed as ‘de-scent’ – in other words, ‘taking the stink out of’ – and in conversation he would refer to this paper as ‘Taking the stink out of instinct’.

(Click on image for a larger version)

What we have seen, I think, is that if we take this term and we take ‘stinc’ out of it, get rid of it, we’re left with ‘in-t’. For some people, INT might mean idiotic nonsensical trash – just forget about it – but if you’re a biologist like me it’s integrative and it’s non-dichotomous and it’s transactive. It doesn’t really matter, but the point is that we need to have a cleaned-up notion of the concept.

(Click on image for a larger version)

All that being said, how does that relate to Thorpe’s notions about the nature of cognition?

(Click on image for a larger version)

This is a very interesting issue which goes back a long way. Darwin supposed that if animals learned things and the behaviour was of use to them, and they went on doing those things generation after generation, gradually the behaviour would become incorporated into the animal’s repertoire and it wouldn’t have to learn it. He didn’t have a mechanism for this; he just felt that this must be right.

Then a few years after the publication of The Origin of Species, a brilliant man called Douglas Spalding suggested a mechanism. This was in a paper in 1873.

(Click on image for a larger version)

Spalding asks the reader to suppose that Robinson Crusoe found some parrots on his island, and he started to teach the parrots a phrase. He wanted to get the parrots to greet him and say, ‘How do you do, sir?’ (This is not a real experiment. This is just Spalding imagining.)

(Click on image for a larger version)

Robinson Crusoe has more than one parrot, so he then breeds from parrots and they have various offspring. Some of them produce funny sounds – Scrrr, or Kaaa, or Peee – but some of them say ‘Sir’, spontaneously. He uses them for breeding.

(Click on image for a larger version)

This process goes on, and then in the second generation some of the parrot babies say ‘do, sir’, rather than simply ‘sir’. This is a very naïve view of genetics, I have to say, but don’t worry. Anyway Spalding’s idea is that by his selection and by degrees the members of this population of parrots expresses more and more of the phrase ‘How do you do, sir?’ And then, when Robinson Crusoe finally dies and the parrots are left undisturbed on the island, people later come to the island and are very surprised to hear these parrots saying, ‘How do you do, sir?’ because they have inherited the whole pattern. Spalding, elaborating this idea, suggested this might be a way in which things could get incorporated into spontaneously expressed behaviour without learning.

(Click on image for a larger version)

Three other people, all well-known biologists – Lloyd-Morgan, Baldwin and Osborn – simultaneously and probably independently, produced the same suggestion as that of Spalding in 1896. The proposed evolutionary process was called at the time ‘organic selection’. Later it became, I think rather improperly, known as the Baldwin effect; that was probably because Baldwin was a great self-promoter. Rather than calling it the Spalding effect, which perhaps would be the right thing to do, it’s better described as ‘the adaptability driver’ because the adaptability of the animal drives the evolution of the phenotypic character. That’s what I propose to call it. So do we have any examples of this kind of thing?

(Click on image for a larger version)

The idea is first of all that adaptability allows the animal to solve a problem which requires several steps for its solution. If that were the case, the sequences, which could be sequences of behaviour, would be improbable if their evolution depended on simultaneous multiple mutations. It is very unlikely that they will all occur at the same time. But if it is learned, then it can be slowly assimilated bit by bit.

If it is the case that Darwinian evolution reduces the time and energy required by descendants to solve that problem because some components are expressed spontaneously, then gradually you can start to move in the direction that Spalding first suggested. And the evolutionary process of building up spontaneous expression of all steps can take place gradually, because it is always buttressed by learning. I will give you an example of something that might be like this.

[A video was shown, with the following commentary.] This is the Galapagos woodpecker finch, and initially you will see it pecking at bark, not using any tools at this point. It is looking for insect larvae.

Next you will see it picking up a stick and poking it into a hole. The first step is particularly interesting, because it is mandibulating the sticks, finding one that is particularly easy to control, and now it is poking it into the hole and gradually prising out the larva inside.

Here is another example: again it chooses a stick and pokes it into this hole, and out comes a delicious item of food – a remarkable example of tool use.

The question is: could we see here an example of behaviour which is on the way to becoming more spontaneous?

(Click on image for a larger version)

We can imagine that in the first phase the woodpecker finch has to pick up a stick and poke the stick into a hole, and both of these things are learned. The evolutionary process could now go down one of two routes. It could continue to learn to pick up sticks and fashion them, but once it has got a stick it can poke that into a hole spontaneously; it doesn’t have to learn that. Or it could go the other way, and it could have a particular interest in sticks and spontaneously pick them up and fashion them, but it then still has to learn to poke them into holes. And then the final stage in the evolutionary process would be that both of those things are expressed spontaneously.

Now, the point about this is that it would be extremely unlikely for the whole thing to appear just de novo, by mutation, because both components have to be present to begin with. So you have to have that combination, held in place either by learning or by a successive building up, over evolutionary time, of the spontaneous behaviour patterns. And we have some evidence that this is the case.

(Click on image for a larger version)

Sabine Tebbich and her colleagues showed that when they reared young woodpecker finches in the laboratory, the birds were very interested in sticks. They had a predisposition to pick up sticks. They were very interested also, at a particular stage in development, in watching other birds doing it. So it looks as though some of this is already starting to become spontaneous in the woodpecker finch.

(Click on image for a larger version)

We have another wonderful example, and you will hear more about this from Russell Gray. In fact, Gavin Hunt and Russell’s work is so well known and this example is so famous that New Caledonia, where the New Caledonian crow lives, made a stamp depicting it. I just quote one of their papers – a recent one, but this work has been going on for a long time. It is again a fascinating example of the use of tools, which Russell will tell you more about, in getting larvae out of holes.

A group at Oxford have followed on from this work and have been doing some developmental studies, rather in the same way that Tebbich did with the woodpecker finch.

(Click on image for a larger version)

Again what they find is that these crows have an intense interest in the tools when they are young. So it looks again as if some of these aspects are becoming spontaneous. Some of them will have to be built up by experience, but at least you seem to see the process en route towards something which is more spontaneous. Of course, we don’t know the true evolutionary history here, because we can’t replay it, but it looks like a plausible example of the kind of thing that Spalding was proposing.

(Click on image for a larger version)

So I have a modest proposal, which is that in many of the examples of higher cognition, what we see here is the product of a process which starts out being learned and then, over the course of evolution, has become partially or wholly made spontaneous. And so the thought is that we might have modules, which themselves can be changed, of course, as the result of experience, which actually predispose us to solve certain kinds of problems more easily than would otherwise be the case. Even Rutherford might have had some of these modules to help him do his physics.

The second point, which is part of the same modest proposal and which ties in very well with the theme of this symposium, is that in order to understand these processes not only do we need to understand the evolution of them and have sensible ideas about how that might work, but also we need to know about development.

(Click on image for a larger version)

I will conclude with a joke about human nature. This is a debate between two experts on human nature, and one expert says, ‘There are two kinds of people: those who like dichotomies and those who don’t.’ And the second expert on human nature says, ‘Nonsense.’

Discussion

Question 1: I was interested to what extent we know that these sorts of behaviours are even still evolving. It could already have reached a steady state. Unless we look at these sorts of behaviours – stick manipulation and so on – from multiple generations over a very long time, we will never really know.

Patrick Bateson: The question is whether we really know that they are still evolving, and of course we don’t. We can’t track the evolutionary process.

I think they’re interesting, because they indicate that the animals seem to be part of the way along there, and of course whether they will go the whole way along there is another matter. Maybe they didn’t need to.

But the issue, I think, for all these sorts of cases where we might have the adaptability driver working is: if there is a cost to learning, in terms of time or energy, and if the process can be speeded up, then those individuals which do it more speedily will be at an advantage over those who do it more slowly – just as in the case of Spalding’s hypothetical parrots.

It is what is driving the process that is the issue. We could, of course, get some information about that from the current field work; I don’t know whether Russell Gray is going to talk about that. It raises very interesting ecological questions.

Question 2: I noticed that the New Caledonian stamp showed the bird having a preference for a stick with a hook on the end. Does that suggest that there is a three-step process, that the intermediate step is in fact the finch that you were talking about that takes a straight stick, and then there is an extra bit with a crooked stick?

Patrick Bateson: Russell Gray will talk more about this, and certainly as I understand it the crows in New Caledonia do differ quite a lot. There are populations where they have much simpler tools and there are populations where they have the more complex ones. I won’t say any more.

Question 3: I liked your module development, or modulisation, whatever one wishes to call it, as the original process which starts all of the complex behaviours going in the first place, and they get incorporated into the central nervous system in some way we don’t quite know about yet. Could you apply the same argument to human speech? After all, that is one of the major achievements of our species. Every more- or less-normal child, put into a speech environment, will start using grammatical language very early and there is a great explosion of words which can’t be explained in terms of any kind of conditioning or reinforcement theory as used to be argued. Chomsky argued that there are processes in the central nervous system which underpin this, the so-called speech areas.

Could you use your modulisation argument in explanation of the development of speech?

Patrick Bateson: Absolutely. I think it is a very interesting case, actually. It is extremely implausible that these things would evolve suddenly, just like that. Some people have argued that, but it seems very unlikely. So what could have driven it? I think you are right, that there is a notion of adaptability and of its being beneficial to individuals to be able to communicate in this kind of way, and then to be able to do it more and more efficiently over time. That then imposes a strong pressure to make it more spontaneous. So yes, I like the idea.

Interestingly, if the explanation is correct, it counters the dichotomous thinking which Noam Chomsky uses. Chomsky talks about these processes underlying language as all being innate, which many developmental psychologists don’t like. Nor do I. Nonetheless, he is right that children will do much of their language acquisition spontaneously. There are differences, of course, according to their experience, and this is where he overstated his case. But if you strip away the dichotomous thinking, then I think it’s a very helpful way of thinking about it.

Question 3 (cont.): So we would find precursors of it in other primates, as has been claimed among chimpanzees, for example?

Patrick Bateson: Sure.

Question 4: I know you can’t ‘replay the tape’ to look at the evolution of spontaneity, but would it be possible to select for spontaneity in chickens, for instance?

Patrick Bateson: Yes, I think it probably would. People are trying to do this, not in chickens but in Drosophila. There is a very interesting group in Switzerland who have been working specifically on this. So yes, and given now that Caenorhabditis elegans is also showing adaptability, C. elegans might be a very good model for examining this. And so people are thinking quite hard about such modelling now. You need a rapidly reproducing species to do this, of course, and I don’t think chickens would be a good prospect for it, but there are good prospects.

Question 5: In the developmental origins of health and adult disease the emphasis is very much now on epigenetics. We know that some genes are epigenetically altered in the level of expression that they will have in the adult, so that the behaviour to eat or the hormonal changes or nerve factors which cause them to eat, or to groom in the case of rats, have been permanently altered by some environmental condition that the mother encountered, either just before or just after birth, during development. Lots of people now are working on methylation of genes et cetera. Is this coming into your sort of field now?

Patrick Bateson: Very much so. It is a very hot area of research, and interestingly some of these effects are not just one-generation effects; some of them are two- or even three-generation effects. And some of them are paternal, which is a surprise. You can get things transmitted down the male line. A few years ago we simply wouldn’t have believed that, but now it really does seem to be the case. And we have a mechanism, as you say, so we can start to say that genes are actually being turned off – or turned on, as well, perhaps.

So yes, it is a very, very important area, and in these examples where it will actually pay individuals to carry information across generations, to smooth out the effects of short-term fluctuations, you would expect this kind of inheritance, and we seem to find it.

Question 6: I was thinking of parallels with evolution of ideas, which I don’t want to get into, but I had a question. To take your birds: if you have one that has inherited one characteristic and another has inherited another, is there the possibility of one then copying the characteristic he hasn’t inherited and then having an advantage? It’s slightly different from the two-mutation situation you described, I think. Is there any evidence that you can copy and then inherit that characteristic? I presume you can.

Patrick Bateson: Well, in the case of the woodpecker finches, which is one of the few that have been studied carefully, they certainly do copy. And, interestingly, they are sensitive to the actions of adults. In this way you could then speed up the evolutionary process, yes.

The interesting thing about the adaptability aspect of this – and it would certainly include this kind of cultural inheritance – is that the learning, however it takes place, will tend to provide a way of stabilising the whole process, so that by degrees you can actually add on more and more spontaneous elements as time goes on.

So yes, you’re right.

Question 7: I think in this context there is still a very interesting research agenda about the ecology of it, because whether it pays to submit something into spontaneity or to turn a cultural ability into a spontaneous property depends very much on the speed of change and the predictability of the environment the animals live in.

Patrick Bateson: I absolutely agree. I think that what we really need now – and we are getting a bit from the New Caledonian crow work – is looking exactly at the sort of ecological conditions under which these things take place.

You could argue, and people have argued, that if you lose plasticity in the course of evolution, that might carry a cost. My own feeling about that is that it will probably be true in some cases, but in the case of a mammal or a bird with a highly parallel nervous system you can lose a little bit of plasticity without its actually hurting anything else. So the issue is whether there is real advantage to losing plasticity in a particular domain, and I think there probably will be cases where we see that.

But you’re absolutely right – these are very early days in this kind of work and I think we need to do a much greater amount than we did in the past.

I want to emphasise that this was a highly disreputable idea. People just didn’t believe it for a long time, and great evolutionary biologists dismissed the idea of such things. Now that a lot of people are accepting it, I think it is leading to some very interesting work.

Question 8: I don’t want to convulse the audience, but surely much reproductive behaviour must be purely instinctive.

Patrick Bateson: It depends what you mean by that! I hope I didn’t give you the impression that I was suggesting that there was no such thing as spontaneously expressed behaviour. I’m not, but what I am suggesting is that there are very many different definitions of instinct, and they don’t all go together. We can have behaviour which is adapted during the course of evolution for a particular function but is nonetheless learned. So those two don’t necessarily go together. And I can give you lots and lots of other examples.

What I am really trying to emphasise is this: if you find something which looks really interesting, describe what you have found. Don’t just say, ‘Oh, it’s instinctive,’ because that will only cause muddle. I breed cats, partly for fun, and at about 30 days after birth when the kittens have started to take solid food they will move over to a tray of litter and suddenly scratch away, and for the first time in their lives they squat down and they pee. After that they cover up the little puddle with litter. When you see it for the first time – this kitten has never done it before – it’s very impressive, of course, and I’m not denying the importance of those observations. But what I am trying to get across is that we need to be very clear about what we are describing.

Question 9: It is an extremely interesting topic and I guess the big question is how modulisation could come about. It could come about by the wiring being set up developmentally, or it could come about by a predisposition to respond to a certain cue to set up the wiring pattern in a modular fashion. And in a lot of these cases, of course, there are critical periods in which that stimulation has to occur.

I am wondering: has it been looked at, where there are these twig interactions, as to whether they are related to visual cues, or is there a critical time period in which these animals have to be exposed to this ability to make these tools? Really, the concept of how a module comes into play is an exciting but somewhat desperately difficult thing to think about, as to how it gets there.

Patrick Bateson: Absolutely. To a certain extent the work on the woodpecker finch goes some way towards answering your question, because it does look as though they are particularly interested in what other animals do at a particular stage in development. So there is a kind of sensitive period in development when they’re most affected.

Tying this rather behavioural type of work that I am describing in with the cognitive neuroscience is going to be quite a long haul, I think. It is very, very interesting, because it raises issues about the evolution of language, which I think are terribly important. But getting good empirical data on the processes is going to be quite time-consuming, I think.

Question 9 (cont.): Certainly with the language development, the time in which the oral understanding of language impregnation into the speech pattern can occur is very critical as well.

Patrick Bateson: Absolutely.

Question 9 (cont.): And again the GABA systems seem to be predominantly in play. So we are nowhere down the line in this regard in this system, in terms of trying to understand the neuroscience of it.

Patrick Bateson: That’s right. I agree.

Question 10: Part of your talk was trying to clarify what instinct means but I wonder whether we’ve got more work down the track for us, because I heard two words being used quite a bit, ‘modularity’ and ‘spontaneity’, and it seems as if each of those might fall out to have quite different meanings. The word ‘modularity’, at least, has three or four different meanings in the literature. ‘Spontaneity’ might be another one of those words that have quite a few different meanings as well. So maybe there is going to be another talk like this down the track, where we try and ‘take the stink’ out of some other words. I wondered what you thought about that.

Patrick Bateson: I couldn’t agree with you more. I would go back to the point that I was trying to make, which is that one should describe what one has found, and not put a word onto it prematurely. You are absolutely right: modularity has lots of different meanings, and so might spontaneity.

I don’t want to throw the baby out with the bathwater, but I do want to have clarity.

Question 11: The earlier question prompted me to think about fish that have a social order with, say, one male and a group of females. The male dies, and one of the females becomes a male. Presumably the morphological changes are genetically engineered, but presumably the associated behaviours are also spontaneous?

Patrick Bateson: Well yes, you get a whole suite of characteristics which are plastic, in the sense that while it’s a female it behaves like a female and when the environmental influences trigger the change and it becomes a male, then a whole suite of new components are expressed, partly morphological, partly behavioural. So yes, in a sense it has to be there.

A metaphor I have used sometimes for this kind of plasticity is that of a jukebox, in the sense that the tunes are all there but it’s a question of which one is selected for performance. The pressing of the button on the jukebox isn’t actually instructing the jukebox in any elaborate sense; it is simply selecting something which is already inside. That might be helpful in thinking about some of these cases of plasticity.

Question 12: I enjoyed your brief comment on Robinson Crusoe. He had clearly read something about Mendel’s work in order to get at least one parrot to say ‘sir’. He could have done much better and done it much more quickly if he had read Skinner’s work on the reinforcement of behaviour. After all, he taught a pair of pigeons to play table tennis together.

Patrick Bateson: Sixty years too early!