SCIENCE AT THE SHINE DOME canberra 2 - 4 may 2007

Symposium: Development and evolution of higher cognition in animals

Friday, 4 May 2007

Professor Russell Gray
Department of Psychology, University of Auckland, New Zealand

Russell Gray is professor in the department of psychology at the University of Auckland. His research spans four areas united by a strong emphasis on evolutionary thinking and principles: language evolution, animal cognition, avian evolution and philosophy of biology. His work on language evolution has pioneered the application of phylogenetic methods to questions about human prehistory such as the settlement of the Pacific and the spread of Indo-European languages. The research on animal cognition uses New Caledonian crows as a model for examining debates about the links between tool manufacture, cognition and cultural evolution. The work on avian evolution uses phylogenetic methods to answer questions origin of groups like penguins and Pelecaniforms and the evolution of their behaviour. Gray’s research on the philosophy of biology has focused on the nature/nurture debate and the role developmental systems play in evolution. He has been awarded a James Cook Fellowship from the Royal Society of New Zealand and a Hood Fellowship from the University of Auckland.

Tool manufacture, cognition and culture in New Caledonian crows

We humans are a pretty smug species, really. We like to think that we are ‘special’, that we are unique amongst the natural world. We point to things like our language, our culture, our ability to reason, our technology, and our tools that enable us to transform the planet, and say, ‘Surely these things must separate us from mere animals.’ And yet we know – at least I hope we know, and all agree – that we are products of evolution, and therefore that we share with other species an evolutionary history.

When it comes to studies of the evolution of cognition, what often happens is that the way in which people reconcile this tension between an emphasis on human uniqueness and evolutionary continuity is to tell rather triumphal linear accounts of the evolution of cognition, leading up to humans as the pinnacle.

(Click on image for a larger version)

Yet, as Darwin reminded us, evolution is not a scale, it is not a ladder, it is not an escalator going onwards and upwards, but rather a tree. And it is not a tree like a pine tree, with a main trunk leading up to humans at the top, but rather a bush. On this slide you can see Darwin’s early drawing of the first evolutionary tree, and you can see that it is diversifying out in lots of different directions.

The Darwinian way to understand the evolution of complex cognition is therefore not just to focus on our close primate relatives, but also, as we have heard rather wonderfully here today, to focus on insects like bees and on birds, and to study how, along different evolutionary lineages, natural selection has moulded particular cognitive abilities.

So, to disrupt the primate-centredness of typical accounts of the evolution of complex cognition, I would like to talk about one particular species of bird – the remarkable New Caledonian crow, and reveal the way in which these crows prick some of the standard primate pretensions.

(Click on image for a larger version)

Here is a list of features often claimed to be unique to human tool technology. Anthropologists and archaeologists have argued that in human evolution we see the evolution of discrete tool types. They contrast that with what often happens in primates, where within one population you don’t have different types of tools being manufactured.

Similarly, only humans, it is sometimes claimed, can manufacture hooks and use them functionally. If you give chimpanzees tools that have hooks and tasks that require hooks, often they use the tools from the wrong end.

If you look at the evolution of human stone tool manufacture, you see that one of the major advances was the ability to impose fine three-dimensional form upon our tools – the crafting of tools. And it is often argued that only humans have this ability to impose fine three-dimensional shape.

Similarly, it is sometimes claimed that only humans are lateralised, are ‘handed’, when it comes to the manufacture and use of complex tools.

Finally, and perhaps most fundamentally, it is often argued that only humans are able to incrementally, cumulatively, evolve culture – while other species might have cultural traditions but they are not evolving because they can’t pass on incremental improvements in design.

So what I would like to do is briefly go over some of the research that my colleague Gavin Hunt and I have been doing that challenges these five claims about human uniqueness. Over the past eight years we have been focusing on a bird not far from here – the New Caledonian crow.

(Click on image for a larger version)

Gavin Hunt discovered that these crows make two quite distinct types of tools. One type of tool is made from sticks, and sometimes, and only in some parts of New Caledonia, they are able to craft fine hooks on the end of those tools. They also make a completely different type of tool – tools made from the barbed edges of pandanus leaves. What Gavin discovered, and what was not previously known, was that they will take pandanus bushes, cut in at the edge of a leaf as shown on this stamp depiction, tear along for a distance, cut in again, tear along again, cut in yet again, jump down to the other end of the leaf, cut in there, and join up the rips to make a tool that is pointed and serrated, using the fine barbs on the edge to extract insects and larvae. So they have two quite distinct tool traditions.

We have recently found that even within one population, and even within families, you can get striking individual differences. One bird will make the pandanus tools, another bird will make the stick tools. So they have discrete tool types.

(Click on image for a larger version)

There are quite distinct types of pandanus manufacture that have very specific geographic localities. I have already mentioned the stepped tools. In other parts of New Caledonia, the crows make these narrow pandanus tools and also these wide pandanus tools.

(Click on image for a larger version)

I talked about how chimps don’t appear to grasp the functionality of hooks. The New Caledonian crows spend quite a lot of time, in certain places, crafting quite elaborate hooks, and they seem to be able to do so in a flexible way. They seem to be able to use a range of different material and converge upon a similar functional product. (Nicky Clayton talked about why flexibility is important in making inferences about cognition.)

So here you can see hooked tools made from a fern stolon, a thorny vine and a twig, converging on a similar functional product from quite different raw materials.

(Click on image for a larger version)

It seems that they might have some grasp of the functionality of these hooks. Some work done by researchers at Oxford found that one bird was able to do something quite remarkable. An experiment required them to get a bucket with some food in it out of a plastic perspex tube. They had one tool that was straight and one tool that was bent. The male disappeared off with the bent tool, and so the female was left just with the straight tool. What she did spontaneously was to bend the wire – to manufacture a hooked tool. So with novel material she was able to flexibly modify the tool to solve the functional demands of the task.

So not only in the wild do the crows routinely make tools – one of the things about these birds is that tool manufacture is a way of life for them, making tools all the time, all day, every day – but in the lab, when faced with a novel task and novel material, they are able to flexibly manufacture an appropriate tool.

So they have discrete tool types, and they make hooks and seem to be able to grasp some aspects of the functionality of hooks.

(Click on image for a larger version)

What about crafting, the fine imposition of form? In one part of New Caledonia they make these very elaborate hooked tools, and a lot of work and imposition of form is required. Two and a half million years ago, our ancestors were basically banging rocks together and we might have been able to get the sharp edge to act as a very crude kind of tool. A step up from that was Acheulean bifaced tools, where there is a more definite imposition of three-dimensional form.

What these New Caledonian crows do when they manufacture hooked twig tools is, firstly, to chop off a branch. Secondly – and this requires quite a lot of work – they break below the fork to produce a kind of thick fork. Next they trim off the side branches, and then they put in a lot of work honing up the hook on the end of that tool. (I will show you a video of that in just a minute.)

(Click on image for a larger version)

You can see here the kind of functional products they end up with, these rather nice-looking hooked tools. Seeing is believing, in this kind of thing. I think some people were rather sceptical that these birds really did this.

[A video was shown, with the following commentary.] I will just show you some video that we made with a French film crew, documenting the manufacture of the stick tools and the pandanus tools.

Where this was filmed, on Maré, an island off the coast of New Caledonia, the birds only make wide tools. What they need to do when they are using them is to line up the fine barbs along the edge. Normally they would try and extract larvae or insects; in this case we have baited the holes with food that they get out.

The commentator suggests at the end of the video: ‘The crow remains the only animal we know of on the planet that can manufacture such a sophisticated tool. Surely it now deserves our respect for hooking our attention on a talent for toolmaking that mirrors our own.’ Perhaps we should be a little bit more sceptical than that, and I am going to describe some experiments which test some of the hypotheses that were rather boldly stated by the commentator.

Before I do, let’s address the question of laterality.

(Click on image for a larger version)

People have claimed that laterality is widespread throughout the animal kingdom, but if you look, say, at primate tool use, you find it is typically not systematically lateralised. There might be evolutionary advantages to using one side of your brain in a specialised way for complex tool use. It doesn’t occur in chimpanzees. What we have found with these New Caledonian crows is that in the most complex tools they make – the hooked pandanus tools – there appears to be systematic bias for the edge of the pandanus bushes that they use.

You can see here that this bush spirals around clockwise, so that from a given position it is easier to access one edge than the other. What we found, looking at both orientations – some plants spiral clockwise and some anticlockwise – was that in the case of clockwise-spiralling plants there was a very strong left-edge bias. It is not so strong and not significant when they spiral in the other direction, perhaps counteracting the natural tendencies. What we have provisionally suggested is that perhaps this reflects right-eye dominance in the brain, where the birds are orientating their right eye down to help to manufacture the complex tool.

So these birds manufacture distinct tool types, they tend to have individual specialisations in the type of tools they make, they can make tools with hooks, and they are lateralised in the way in which they manufacture the most complex tools.

(Click on image for a larger version)

What about cumulative cultural evolution? We now know that the ability to make tools is quite widespread throughout the animal kingdom, but typically it is relatively simple. In the Galapagos finches, for example, the birds break off cactus spines and small sticks, as Patrick Bateson showed in his talk. Even in chimpanzees it is relatively simple. There is very little modification of the raw material.

So people have argued that tool use and tool traditions are common but that cumulative cultural evolution, where increments in design are passed on and ratcheted, is what is fundamentally unique to humans. People like Tomasello have argued that that real cumulative evolution requires cognitive abilities like genuine imitation, which enables us to pass on design improvements and improvements in culture with high fidelity. We think we have potentially the best case for possible cumulative technological evolution in the animal kingdom.

(Click on image for a larger version)

You see here some examples of quotes from people summarising a lack of evidence of cumulative cultural evolution outside humans.

Tomasello says that ‘only human cultural traditions’ accumulate modifications over time. Boyd and Richardson say that cumulative cultural evolution ‘is limited to humans, songbirds, and perhaps chimpanzees’. And Tomasello used to say, ‘No cultural products exhibiting anything like the ratchet effect have ever been observed in the ontogenetically acquired behaviours or products of nonhuman animals.’ Here is our challenge to that claim.

(Click on image for a larger version)

One of the great things about the New Caledonian crow pandanus tool manufacture is that they leave an ‘artefact record’ on the edges of their leaves. The crows cut out their tools from the leaves of the pandanus bush, as shown on this slide, so it is possible to go along and study the counterparts to get a representation – if you like, a fossil record, an artefact record – of what the population has been doing.

(Click on image for a larger version)

These counterparts stay intact on the plant for about four years. They correspond very nicely, we found, to the tools that come from them. So there is an accurate artefact record out there on the vegetation.

(Click on image for a larger version)

What Gavin Hunt did was to tramp all over New Caledonia documenting variation in the pandanus tool manufacture across New Caledonia. I have to show you this picture, because people think that we go to New Caledonia and sit on the beaches all the time. The points indicated here, high on a mountain, were two of the places where Gavin collected some samples.

(Click on image for a larger version)

What did he find? He found that the wide tools that I showed you being made on Maré have a very limited distribution in the south of New Caledonia. The narrow tools are found slightly more broadly throughout New Caledonia, and the tapered tools, the stepped tools, have a different distribution again. We looked for potential ecological correlates of these different tool types, but were unable to find any. So what we argued is that the most plausible interpretation is that these differences in the manufacture of different types of pandanus tool reflect social traditions.

(Click on image for a larger version)

We argued that the most plausible way of explaining those social traditions is as a case of cumulative cultural evolution. What happened is that the birds will routinely rip pandanus to get into the base of the pandanus vegetation, where there are often insects and larvae. All they need to do is cut off one of those bits and they have a wide tool. But the problem with wide tools is that you can’t get them into narrow crevices.

So the first design improvement they made was to make narrow tools. But they are rather flexible and weak, so a better improvement is to make a tapered tool that is both tapered and pointed at the tip and also thicker and more robust at the end.

And the final design improvement is to add extra steps, and therefore to add extra stiffness in this kind of way.

So we argued that in these birds what we see perhaps is the best nonhuman example of cumulative cultural evolution.

(Click on image for a larger version)

It would be tempting just to leave it at that, but I think that would be rather poor and sloppy science, because what we have at this point is really a lot of observations of what has happened in the field. We haven’t probed the animals to see what is happening in their underlying cognition. So what I want to do now is to make perhaps some rather naïve proposals about what we might predict about these birds, and look and see, with our more recent work, how true these turn out to be.

It is the case that the close relatives of the New Caledonian crows don’t make tools at all. So perhaps you might speculate that there must be some underlying adaptive specialisation that enabled only the crows on New Caledonia to do this, and not related ones.

We might predict that maybe, unlike primates, these birds have a really good grasp of what is called the folk physics, the causal understanding of the physical properties of their tools.

We might also predict that, if there really has been cumulative cultural evolution, there should be genuine imitation in these species (and hence should be able to pass on improvements in tool design in a cumulative way).

And perhaps we could speculate that underlying these apparently impressive tasks might be some ability to reason, perhaps by analogy, to solve various tool-related tasks.

One of the joys of doing science is that sometimes your intuitions and hunches go wrong. Let’s see what happens.

What about adaptive specialisations? Well, perhaps there is something different about their brains. Rather crudely, we could ask: is it the case that these New Caledonian crows have bigger brains than their close relatives?

One of the standard ways of looking at that is to graph a logarithmic scale for body weight against a logarithmic scale for brain weight, but the crows are significantly larger in their brains than their body weight would predict. So perhaps this is some evidence of adaptive specialisation in terms of their brains. But really that is far too gross. I think it would be much more telling to do specific neuroanatomy work, looking for specific differences, and that is what we are engaged on.

(Click on image for a larger version)

What about folk physics? You saw the crows apparently elegantly and deftly using those barbs on the pandanus tools to pull out food that we placed in the logs. What a PhD student of mine, Jennifer Holzhaider, has been doing is some simple experiments testing whether they really do understand the functionality of these barbs and their hooklike properties.

She has done two experiments, in one of which she gave them a choice in the log with the barbs facing in the right orientation versus the barbs facing in the wrong orientation so these tools would not be functional. We expected the birds to flip the nonfunctional tools over. Much to our surprise, however, there was no difference in the frequency of flipping from one condition to the other. So they seemed in that experiment not really to be grasping the functionality of those barbs.

More simply, if we gave them a choice between tools with barbs and without barbs, once again they didn’t appear to strongly discriminate.

So the expectation that, because they had been able to use the barbs functionally in the field, they grasped the properties of them, does not seem to be correct.

From more recent work with Jenny what we have found is that, looking at the way in which they develop these skills, what they learn is the order of operations in which to make the tool so that they are holding it so the barbs will be facing in the right direction. In other words, what they have learned are the procedures required, rather than an understanding of the functionality of the barbs. In psychological jargon, they have procedural knowledge rather than causal understanding.

(Click on image for a larger version)

What about imitation? We have been studying the way in which the development of these pandanus tools occurs. Here you can see Jennifer with a 21-day-old bird, Yellow-Yellow. In the upper right picture you can see it about a year later, on Maré. Let’s look and see, because the Oxford group have suggested that a lot of the tool manufacture reflects an evolved disposition – it doesn’t require much learning to take place. What have we found?

(Click on image for a larger version)

Finally, what about reasoning? What are some of the cognitive abilities that underlie these tasks? How flexibly can the birds solve problems? Is there some very domain-specific ability linked to their specific tools, or can they generalise them, in ways that perhaps might require analogical reasoning?

One recent experiment that Alex Taylor, a PhD student of mine, has been doing is what we call a ‘metatool’ task, where there is some food down a tube and there is a tool out for use, but the tool is too small to get the food. To solve this task, what the crows need to do is to use the small tool to get the big tool, to get the food.

In this task, very strikingly, the first thing that six out of the seven birds did, rather than trying to get the food out with the small tool, was to take the small tool and attempt to extract the large tool. And four of them succeeded in doing that on the first trial, ruling out associative learning as a possible explanation.

[A video was shown.] I will just show you one bird on its first trial.

To turn to some conclusions: what would our close primate relatives make of the performance of their distant avian cousins? What I hope I have shown you is that, in the adventure of trying to understand the evolution of complex cognition, it might be useful if we veer occasionally off the primate-centred path; that there are lots of really fascinating and insightful things to be discovered down other paths; and that one of the joys of doing science is that we don’t always find out what we expected. Certainly in the science of animal cognition there is an awful lot more to be discovered.

(Click on image for a larger version)

Discussion

Question 1: Given that it is so hard to find birds, for instance, that do this trick, how are you going to show that the ability to do the trick has evolved? It is pretty difficult. If you have only got one crow species in the whole lineage of crows, or crow ancestors, that do it, it looks like spontaneous occurrence – there, and in the finches in the Galapagos, and whatever.

Russell Gray: There are two answers to that question. One thing is to perhaps look at their close relatives, and there are sporadic reports of things that could be interpreted as tool use in some of the relatives of the crows, under laboratory conditions. But you might want to think what are the precursors of tool use. One of the things that the Oxford group has argued is that the caching behaviour is a possible precursor. So one strategy would be to look at, say, a corvid phylogeny, to look at features of, say, social evolution and caching behaviour and associated cognitive abilities across that whole corvid phylogeny.

Sure, that kind of very spectacular tool use is going to just pop up in New Caledonian crows, but you might be able to track the evolution of the precursors of some of those kinds of abilities, and in that way get a more general understanding of the kinds of factors that are involved in order for that complex cognition to develop.

That would be to answer the question about biological evolution. To look at the cultural evolution – because I think there is a strong possibility of complex cultural evolution going on here – what we have is a banded population of these birds so we can systematically follow them over generations. We have some birds that have behavioural innovations. For instance, there is one bird that makes a pandanus tool that it bends over, getting an extra layer of stiffness. So we can follow the spread of innovations of that kind in the population, and track their potential cultural evolution, looking to see whether the offspring of that bird also make tools like that.

Question 2: I am quite interested in the lateralised behaviour that you reported there, and I was wondering if there was any relationship between the lateralised picking-off of these tools and the frequency of the left- versus right-whorling pandanus in the environment.

Russell Gray: We have looked at that, as an obvious potential confound. I think we found that it didn’t explain the kinds of differences that we saw – the biases were independent. We had a paper last year in the Proceedings of the Royal Society in which we looked at the degree of specialisation for the left edge versus the right edge of the leaves, across the different types of pandanus tool. We found that it is most pronounced only in the most complex tapered tools. And that is independent of the frequencies of the orientations. So I don’t think there is some simple ecological differences explanation for that result.

Question 3: This is a more general question to do with the human uniqueness you were talking about at the beginning. Is there a kind of continuous raising of the bar, that every time we say, ‘This is the thing that makes humans unique,’ we then find something out and we say, ‘Well, I didn’t mean quite that. I mean this plus you have to have that. It can’t just be procedural toolmaking; it’s got to be this other kind of thing’?

My question is this: to what extent is our understanding of what we think is smart behaviour occluding what might be quite complex cognitive behaviour which we might not think is smart? The kind of thing I have in mind is that it turns out that we can make computers that can play chess, but making a computer that can find a Coke can in a crowded room actually turns out to be a damn sight harder than we thought. I just wondered if there was actually lots of interesting behaviour out there which we are missing because we have already got this preconceived notion of what constitutes ‘smart’.

Russell Gray: That is one of the points I wished to make but didn’t really dwell on. I think that the human-centred, triumphal kind of narrative has been taking our attention away from a lot of the phenomena that really should be explained if we look at the evolution of complex cognition more generally across the vast tree or bush of life. So I totally and utterly agree. And it might be that in doing that, the standard human categories, where we differentiate between, say, reasoning versus simple associative conditioning, are not very useful. There might be phenomena that are not well characterised in either of those categories. So I absolutely agree.

I remember seeing Jane Goodall interviewed once on TV, when she said that Louis Leakey defined humans as the ‘toolmaker’. Then, when she found that the chimpanzees could use simple tools to fish for termites, he said, ‘Well, we now have to change our definition of what it means to be human.’ So then it moved to ‘only humans manufacture tools’, and obviously we have to modify that kind of claim too. I think lots of that redefinition goes on. I think if we had a more truly evolutionary understanding of the evolution of cognition we wouldn’t make that kind of mistake all the time.

Question 4: What other food sources do they have? There is a lot that has been written about ‘extractive foraging’, and perhaps there is an argument to be made that if the main food source happens to be hidden below, and the only way to get to it has to be learned, then the one that has solved the problem is going to survive. Where do you sit with that?

Russell Gray: It is a real puzzle that I have no glib answer for: how come that only on this little island, stuck out in the middle of the Pacific, this one particular species of corvid has done this? You could point to the relative aridity of some parts of New Caledonia and say maybe that means a lot of insects are not that easily available, but then there are lots of other dry places in the world – we might be standing in one – where there are corvids who don’t manufacture tools. So I don’t think that sort of simple, one-factor explanation really accounts for a lot.

These birds are using these tools all day, every day, all the time. I had some fun video of papaya that we sometimes put out for them; they often eat papaya in the gardens. And we have juvenile birds that, rather than just eat the papaya – you can just eat it – use their tools to get at the papaya to eat it. That just shows how ingrained that kind of behaviour is.

Question 5: It is well known, of course, that many female birds choose their mate as a result of visual clues – bright plumage and the like. You have told us several times that they use this special way of hooking food out all day and every day. Is it possible that the female birds choose mates on the basis of knowing which male birds are good at that sort of thing, and hence you have built it into the genome in some sort of a way?

Russell Gray: That is a very nice idea. The answer is that we don’t know. One of the things we are looking at is whether they will mate assortatively on the basis of what kind of tool their partner makes. What I can tell you is that it doesn’t seem to be that pandanus-tool-using birds only mate with pandanus-tool-using birds. And there doesn’t seem to be strong sex difference in the types of tools they make. But we need to get much more information. We are doing various paternity studies and pedigree studies, to look at whether there is any kind of assortative mating that might be going on, on the basis of tool type. We don’t really know yet.

Question 6: This is a small question. You drew up a ‘tree’ of tool evolution, and it ended up with the tool with the jagged edge. Then you showed how a young bird had problems in knowing which way round to hold its tool for use. Is it possible that the evolution is actually driven by the much easier memorisation of which end of your tool is the important one to grasp, so that you evolve an asymmetrical tool that has an active edge and one edge which you should actually hold on to?

Russell Gray: So there might be some cognitive bias?

Question 6 (cont.): No, that the tool is not only a tool to extract grubs but is a memory tool. It makes it easier for you to know how to handle it, perhaps.

Russell Gray: I had never thought of that possibility. That is a fascinating possibility. There is quite a lot of extra work, though, to make a multi-step tool instead of a wide tool. That is an interesting possibility, but I would have to think of a way of trying to test whether the memory demands influence tool type.

Question 6 (cont.): The test would be to see whether young birds find it easier to know the right way of holding the tool.

Russell Gray: That’s a fascinating idea. Thanks a lot for that.