SCIENCE AT THE SHINE DOME canberra 2 - 4 may 2007

Symposium: Development and evolution of higher cognition in animals

Friday, 4 May 2007

Professor Gisela Kaplan
Centre for Neuroscience and Animal Behaviour, University of New England, Armidale, New South Wales

Gisela Kaplan holds a PhD from Monash University (Arts), and a PhD in Veterinary Science from the University of Queensland, Brisbane. She specialises in vocal behaviour and higher cognition of animals and has conducted research on a range of species. Publications include over 250 research articles, 18 books, and countless contributions to magazines, encyclopaedias, documentaries and radio programs on animal behaviour. Among book publications are several on orang-utans (1994,1999, 2000), and more recent book titles include: Songs Roars and Rituals. Communication in Mammals, Birds and Other Animals (2000); Birds. Their Habits and Skills (2001), Comparative Vertebrate Cognition (2003); Australian Magpie (2004) and Tawny Frogmouth (2007). Her books have won awards and high acclaim. Her research on Australian magpies is ongoing, currently investigating their song control system and the function of various song types. Prof. Gisela Kaplan is a life member of the International Primatological Society, member of the International Ornithology Committee, member of the Scientific Committee of Eurasian Ornithology and is a research professor in the Centre for Neuroscience and Animal Behaviour, Faculty of Science, University of New England.

Higher cognition and communication in apes and birds, with special reference to the vocal repertoire of Australian magpies (Gymnorhina tibicen)

As an introduction to what I will be speaking about today I will make some preliminary remarks. The title of my talk promised to include birds and apes in the presentation.

However, I want to exclude the great apes after my preliminary comments and for good reason. This is not because of a predilection only for birds. Indeed, I have worked with orang-utans for a considerable period of time, together with my colleague Professor Lesley Rogers. There are two remarks that I want to make in particular.

First, when we look at images of birds and of great apes, we very readily have an affinity with great apes. I would suppose and suggest to you that this affinity with great apes is established largely via the face. We pay great attention to the face, and when one looks into the eyes of an orang-utan – it has been claimed in romanticised literature again and again, and it actually happens – we feel we recognise a sentient being, a being to whom one can relate almost instantaneously. Indeed, if one examines this assumed connection very carefully, it becomes clear that, what one is actually relating to, is probably not so much the face as a whole but particularly the eyes and the mouth.

Our communication with birds via these signals (and how we view the species) is greatly hampered by the fact that the ‘mouth’ is the beak, and one of the immobile fixtures that cannot express happiness and sadness. Not only the beak but also the eyes are fixed; because of their lateral position the eyes do not tell us very much in terms of any assumed or real emotions to which we could relate. Eyes, in the case of birds, do not function for us as ‘windows to the soul’. Hence, our readiness to understand that there are birds that would be able to do very similar things, or are as complex cognitively, as great apes is usually intuitively diminished because the markers to which we, as a species, like to relate are missing in birds. It is difficult for us to believe that the yellow-tailed black cockatoo may be able to perform cognitively at the same level as a great ape, but research evidence shows that that may be the case. In fact, it is highly unusual to assign the word ‘face’ to a bird. I think the only reference I have ever seen to a bird’s head as having a face was by our speaker this morning, Professor Patrick Bateson, who actually did so in one of his papers.

The other point I wanted to make about primates in relation to cognition – and then I won’t mention them again – is that the moment we turn to vocal learning and we turn to communication, our predecessors, the great apes, are not able to tell us as much as birds can. As neuroscience has mapped out now for the last 30 or 40 years, vocal learning is a feature that is shared by humans, songbirds, some cetaceans and some specific non-songbirds, such as parrots and hummingbirds, but not by great apes.

Hence, great apes are not a good model for understanding everything about humans and, despite the fact that there may be a close, and even very close, relatedness to us – the DNA is always stressed – their vocal apparatus as well as their brain is not organised to do what birds and humans can do, namely learn vocalisations. As models for understanding the processes of vocal learning and auditory memory, songbirds are much more appropriate. As to what is meant by vocal learning: vocal learning, as used in neuroscience, has a very specific meaning. It refers to the need and ability of an individual to learn and memorise sounds and this ability to acquire sounds and retain the memory may even extend to mimicry of other species for which, demonstrably, there is no genetic basis, no pre-fixed template in that individual or that species.

I’ll give an example. If, let’s say, I asked anyone in the audience now to bark like a dog, you would probably be able to do so. This means that you can actually take the auditory information from a different species and reproduce it. And when you give a reasonably good rendering of a dog’s bark I, in turn, would be hard pressed to prove that the ability to bark like a dog was part of your genetic make-up.

As my topic is vocal behaviour and vocal learning I will only speak about birds from now on, and keep the great apes in parenthesis. I will briefly summarise the main features of song and song production before turning to the aspects of vocal development that might be considered under the rubric of complex cognition.

(Click on image for a larger version)

We have been told often enough by our colleagues in the northern hemisphere that high-latitude birds have seasonal brains. The brain is seasonal in the sense that, at the time of the breeding season, some male songbirds sing in order to attract a female. The science of song production has been built largely on such vocally dimorphic species living in high latitudes with short breeding seasons.

(Click on image for a larger version)

The problem with the theories that have been developed in neuroscience is that they leave out much of what avian species of the tropics and in latitudes south of the equator do vocally. Indeed, neotropical, tropical and Australian birds may fit into entirely different models of song and breeding behaviour than their high latitude counterparts. We need to be on our guard not to assume that functions and circumstances of song, and thus of vocal learning, can be accommodated in the same, now dominant, model of the male seasonal singer. Although in Australia we have male breeding songs in a number of avian species, we also have singers that not only sing all year round but males and females sing alike and do so demonstrably not for reproduction.

Perhaps one of the most important sets of differences between avian species of northern latitudes and southern latitudes is between competition and cooperative models associated perhaps with length of breeding season. There are more cooperative breeders in Australia and the tropics alone than in the rest of the world. In many southern hemisphere and tropical avian species there is little sexual dimorphism and there are many instances of shared brood raising and various forms of cooperative breeding, including long-term pair bonding and so forth.

I now want to turn to the Australian magpie as an exemplification of a southern hemisphere songbird, and one of the foremost songbirds of Australia, that does not fit the models applicable to high latitudes.

I would like to point out that male and female magpies both sing year round, the female magpie having a slightly larger song repertoire. Their main song, called warbling, has no function to aid mating. Indeed, their vocalisations decline during the breeding season. Knowledge of that has been based on my studies over the last 12 years and a substantial dataset.

(Click on image for a larger version)

It is obviously important to first understand how and why a songbird can produce a song at all. The primary song control system has been investigated in great detail. Particular nuclei in the brain make singing and vocal learning possible and in non-songbirds these are either vestigial or absent. The song control system in birds is thus specific to songbirds and consists of a range of nuclei that can process auditory input and provide signals for vocal output.

The most important of these is the high vocal centre, as shown in this image.

(Click on image for a larger version)

We conducted a preliminary study in our laboratory, with our colleague Dr Chao Deng and with Lesley Rogers and myself, on the actual presence of these song nuclei in magpies. Here, in this image, a slice is shown of the higher vocal centre in the magpie brain, that of an adult female and of a female juvenile. Near the top of the frame of an adult female magpie you see the density of cells and this marks the high vocal centre. It is present in females and it is present already in female juveniles, quite early on, as shown in the image below. So, clearly, female magpies have the right nuclei to produce song. The secondary song control system, consisting of the elements that can produce song, are equally important. In birds, these consist of the syrinx (located at the tracheal end of the two bronchi), the trachea, larynx, and beak.

[A video was shown, with the following commentary.] I want to show you what a vocal action by a magpie looks like, close up, when it sings. This is one of my hand-raised magpies and hence permitted uninterrupted filming during song. It is quite clear from this video that the movements in the area of the throat are the result of laryngeal muscle activity involved in song production. You can also see clearly that, during this singing sequence, the beak is almost closed and then opened only a fraction for some milliseconds. The beak is thus used for modifying the sound that is produced. The form of singing presented here is called warbling, an undulating pleasant tonal sound. I will also play another form of singing called carolling, in which the bird throws back its head and, in this case, also opens the beak widely. Sometimes carolling and warbling are performed together. We know the function of carolling very well, but the function of warbling is not well understood. The latter is sometimes referred to as subsong since it doesn’t develop into the crystallised song of seasonal singers.

I want now to talk about vocal development a little to place the improvised song practice of nestlings in a broader context.

(Click on image for a larger version)

The obvious aspects of vocal development are physiological and anatomical. We would expect that body length and body weight would develop in parallel and could affect sound production.

(Click on image for a larger version)

In bird beaks the gape that is initially very wide narrows over a period of time but the beak actually lengthens substantially. And it is not just the beak, but equally important for sound production is the change in the length of the trachea. As seen on this slide, the neck is quite short in a three-week-old nestling, and in a post-fledging period it lengthens quite considerably. So the sounds that the bird can produce at each stage of development are obviously going to be different (confined to higher frequencies in the nestling and extending to capabilities in the lower frequency range once the bird has matured). This is a simple mechanism that is relatively well understood.

(Click on image for a larger version)

Many of you who live in Australia know the sound of a magpie begging. It is easily identifiable because it is stereotyped and, as seen in this sonogram (in the top left panel), it is also audible over long distances, with frequencies up to 8 kHz, but the main energy tends to be between 1.5 and 4 kHz.

(Click on image for a larger version)

Interestingly, the begging call develops in association with the distance of the parents.

(Click on image for a larger version)

In the first week of life the nestling barely produces audible peeps or begging calls at all. And then it will start begging when a parent bird or helper is at the most 50 centimetres away – so the nestling doesn’t actually vocalise until an adult is visible, and that distance increases by the third week. By the fourth week, just before fledging, it is over one metre. Begging calls are thus a very nice, simple representation of stimulus-response: seeing the parent nearby produces the begging call.

[A sound recording was played.] I just want to play to you the sound of the peeps in the second week and the calls in the third and the fourth week which audibly exemplifies the physiological and anatomical changes.

(Click on image for a larger version)

When the young have left the nest, the begging vocalisation becomes extremely intense and the juvenile actually solicits a parent to approach and provide food.

(Click on image for a larger version)

The reason why I have shown you the slides of stereotyped call development is to make a contrast to other vocal developments that, in the third week post-hatching, also occur quite spontaneously. If you look at this pattern of sound here, recorded in the third week post-hatching (still in the nest), you can see that these are quite complex and varied. Importantly, these sounds already have most of the elements that are later part of that warbling song, or subsong, as it is maintained by male and female magpies throughout their whole lifetime. There have been some studies, and my own studies, suggesting that the individual can have as many as 740 to 780 different elements in this song type alone, and these elements are practised amazingly early in the life of a magpie.

They have complex harmonics, with beautiful structures already emerging, and enormous variability as shown at the bottom of the slide.

(Click on image for a larger version)

The interesting part is that, opposite to the use of begging calls, the practice of the warbling by the third week occurs only once the parents are away, not when the parents are actually there. So the begging call is only elicited when the offspring see the parent; while song practice of warbling occurs only when the parents are out of reach.

The y axis here indicates the distance from the nest, showing that this is at least 10 metres. In some cases there is an outlier, like the one at the top of the graph for week 4, at least 25 metres away. However, this is unusual and is likely to have occurred because the nest was positioned in such a way that the nestlings had a clear view for 25 metres from the nest before the parents moved out of sight. It was obviously important for the youngsters that, before they practised, the parents were out of sight and, presumably, out of earshot.

(Click on image for a larger version)

That raises a very interesting theoretical question. The literature has talked about song acquisition in birdsong in a number of ways.

For instance, there is instruction-based learning for young males that are tutored by their fathers. This comes from the very well-studied model of the zebra finch but applies also to other species in which only the male develops a crystallised breeding song. The point is that ‘song’ itself is now generally defined as units of sound that have a final form and are used for specific purposes, be this for attracting a female or defending a territory.

Then there is selection-based learning as a way to over-learn more syllables than needed and later discarding some and then gradually approximating the bird’s own song to the right elements; then there is social shaping, which has been investigated particularly in the cowbird. The cowbird parasitises other nests, so the youngsters never hear their own song until they are actually out of the nest. The young males trying to woo a female don’t know at all what to do. Apparently, experienced females keep ‘shaping’ their vocal behaviour, until they do the right thing.

But the magpie style of learning does not fit any of these models of song learning because not only does their subsong appear to be entirely improvised it is also not going to crystallise into a final form with a definitive function. There is no evidence at all that anybody is teaching or shaping their song, and the magpie nestlings start song practice spontaneously and by themselves. And unlike selection-based learning, they do not shed elements from juvenile song practice but continue to add syllables as they develop. Moreover, they add new elements as adults as well.

If there is any form of vocal learning in magpies that I have been able to find so far, it is found in a song type called warbling. Carolling has been identified as playing an important role in territorial defence. It is a relatively simple, short and noisy call sequence and appears to be reserved for adult members occupying a permanent territory.

(Click on image for a larger version)

A possible form of learning is by listening without reproducing the sounds at the time they are heard. This may be seen in this particular photo here that was sent to me. There is a juvenile standing in front, feathers fluffed, and there is an adult female in the background carolling. The juvenile bird in front is adopting the posture of carolling but does not produce a single sound – which, by the way, supports the Italian school of opera that argues that one actually learns and improves song by listening. In begging, a stereotyped call that was described in detail before, it appears that little learning occurs. In carolling, although a relatively stereotyped call, it may be possible that some learning needs to occur or it may be that adults simply suppress the vocal output for social reasons. In other words, in the three main forms of vocal production in magpies, there is no immediate evidence that the birds learn anything by instruction. Yet there is evidence that magpies are cognitively very complex birds. They cache food, they have long-term memories, and there are complex social interactions including long periods of interactive play, all facets that might hint at higher cognitive abilities.

(Click on image for a larger version)

In terms of cognitive development, we have argued strenuously in many different species that play behaviour is actually one of the main opportunities for learning, for the development of skills and in social interactions. Certainly magpies play intensively and they vocally interact with each other for months at a time. They also have a long period of dependency on their parents and usually do not reproduce until they are aged 5 years or older and the long gap between sexual maturity (after first moult at one year of age) and reproduction are many years in which experiences can be collected. These are circumstances that make good candidates for complex cognition.

(Click on image for a larger version)

If we take a timetable of vocal development – this is based on data of 35 hand-raised magpies and 35 wild magpies – we see that they have distress peeps in the first week, and some beginning of begging calls in the second but very, very faint.

As I said, warbling first appears in the third week post-hatching and continues to develop post-fledging. Then there are calls that are related to their social interaction, like threat/disapproval calls that appear post-fledging and, at the same time as the first social play is observed, another form of vocalisation, that of mimicry, appears in the repertoire, and later still, by 7 months a few timid attempts at alarm calling can be heard.

I now want to speak briefly about mimicry. In magpies, mimicry arises solely within the context of warbling. Whether or not there is a significant relationship between social play and mimicry is not known but these two phenomena are developmentally co-timed.

Mimicry has been treated very badly in the literature – we talk about parrots ‘parroting’ and apes ‘aping’. Yet there is something very important about mimicry that we should not overlook. For research into vocal learning, mimicry is an ideal case because vocal learning can be shown to occur. It is very clear that if a bird can mimic another species it must have had some auditory input, must have formed a memory of the sound heard and then been able to reproduce it. Unlike all the other forms of vocal development that I have shown in magpies that can be explained physiologically and as genetically predisposed, mimicry is a clear example of the influence of environment shaping vocal production. In magpies, as in other species in which integration of mimicry has been studied, the full reproduction of a sound that has been heard from another species (not just avian but even mammalian) can occur in very little time and then be a relatively accurate copy of the sound being mimicked.

(Click on image for a larger version)

Very clearly, in this case, the magpies are able to learn this in under one minute, and they reproduce the sound within 48 hours, usually. (We have examples of this time-frame in parrots.) The latency to reproducing the mimicked sound is different from that in humans, who can reproduce the sound instantaneously; in those birds that mimic, there seems to be a 48-hour delay, as I found in magpies.

(Click on image for a larger version)

Here is an example of mimicry by a magpie. At the top of this slide is the sonogram for a kookaburra duet; below is a magpie mimicry of it. I really want you to hear this and look at it, because the structure of those calls in these sonograms is nearly identical, and the energy in the calls is very similar even though the magpie seems to be struggling to reproduce the noisy characteristics of the kookaburra calls.

[A sound recording of calls was played.] First, the kookaburra duet. Now the magpie. [Other sound examples of mimicry of a horse, a barking owl and of human voice were also played.]

Alec Chisholm had already shown very clearly that there are many avian species in Australia that mimic. The best known is perhaps the lyrebird, a seasonal singer that uses song to attract a female while dancing on a mound. It mimics extensively, and when it does so, it adds little pieces on, like a piece of string, and once it has done so, its repertoire will then return to the beginning and replay the whole cycle in the same order again. This is called the Albert Cycle. The elements, including the mimicked sequences, will never change position again. Here we know the function of song, and presumably of the mimicked sequences, and that is to dazzle a female and persuade her to mate with him. When looking at sonograms of the lyrebird mimicry performance, it becomes apparent very quickly that the lyrebirds only very superficially copy the mimicked sounds. The male singers try to get away with murder, and they do. Structurally, the rendering of the mimicry is not half as accurate as that of the magpie. The magpie is meticulous and pays great attention to detail of the sound structure. The magpie certainly has the makings of a scientist; I would not say that the lyrebird has.

What could explain the difference in performance? First of all, the magpie does not use mimicry to dazzle a female. Males and females mimic alike. Why do they pay such attention to the details of the mimicked sound? Are there any cognitive dimensions? I have chosen to look more closely at the mimicry of magpies to see whether these reveal any specific attributes.

First let me show you the typical distribution of magpie mimicry within a warbling sequence. At first, it seems entirely random because the mimicry is interspersed into the bird’s own song. Each colour that you see here represents a different kind of mimicry that is interspersed into the bird’s own song.

The only distinct feature that I was able to identify is that mimicry rarely seems to begin a warbling sequence and never seems to appear at the end phrase.

But the interesting part in this mimicry – I have already talked about the lyrebird – is that the mimicry is not random. When I analysed what kind of sound was mimicked, I found that a hand-raised magpie, over months of testing and over months of recording, actually reproduced and practised mimicry of human speech to the extent of 73 per cent, and then, in various degrees, of other species that it heard.

The next point I wish to make, and this is really important, is that the magpie constantly practised mimicked sounds and clearly managed to improve its performance substantially over time. Here is an example of learning, not by spontaneous acquisition but also by constant practice.

The third point I need to make is that work by Goldstein and various other people has argued that the developmental timetable in birds, particularly in subsong, seems to be a foreshortened version of that of human language development. Early infant vocalisations have identified babbling periods. Subsong has always been called babbling (or warbling). How this human vocal development and subsong development can be equated with a measure of scientific rigour is still not clear and the literature, I am afraid to say, is still in its infancy on this, but I would suggest that we need to do much more work on the actual language development in the first four months of infants as well as that of magpies and other songbirds to see whether there are really structural similarities that underlie the similar ability of birds and human to learn vocalisations.

(Click on image for a larger version)

In summary, vocal learning is demonstrated in mimicry; there is no evidence that song and mimicry practice are motivated by reproduction; mimicry is selective rather than random; it shows plasticity of the brain; and perhaps – I should now go into speculation – it serves territorial characteristics and awareness, because my own particular research has found that magpies do not, in fact, randomly use mimicry or incorporate mimicry. When they mimicked sounds, they used sounds belonging to species that permanently reside in the magpie’s territory, never transitory sounds.

So it seems to me that this kind of vocal development that incorporates learning and practice and is selective may lead us to have to look even at mimicry in a different light. Mimicry is memory formation, concerns sounds that are not laid down within the species as repertoire and requires some recognition of musical and structural elements of sound. It seems that in learning of vocal repertoire, as in its many other facets, the magpie has highly advanced cognitive abilities and many of these may well be en par with great apes, but in its vocal learning the magpie has very special abilities.

Discussion

Question 1: Do magpies mimic other animal sounds in nature?

Gisela Kaplan: In fact, we know of parrots that their mimicked sounds are usually artefacts of captivity either because people teach them to speak or they have regular exposure to specific sounds, animate and inanimate. Magpies mimic always in the wild. The examples I have, come from across all of Australia and they have been tape-recorded, so we have evidence that they use mimicry regularly in the wild. They do not regularly use mimicry in captivity and possibly cannot be taught by the main caregiver. In fact, I have tried to teach magpies to speak, totally unsuccessfully. As I later learned through observational research, this was an unsuccessful experiment because magpie offspring do not ever copy their parents. So if you are trying to hand-raise them, it is likely that you will be the one person who will not be copied. Somebody who is only secondarily associated with the magpie, on the other hand, may be copied instantly. But such are the accidents of experimentation and observation. Sometimes one learns most from the unplanned things. So there you are. That is the first example we have that mimicry is used in such complex ways.

Question 2: The human child’s brain uses 60 per cent of the available oxygen for the first six to nine months after birth and then turns back to 20 per cent of its general body. Do we know anything about oxygen usage in bird brains? Relatively, as a percentage of the oxygen that the whole animal uses, is more used by the brain in these periods?

Gisela Kaplan: We have undertaken a study this year to check the bilateral flow and the oxygen use in producing song, so for the syrinx, which is the primary sound organ for the bird, that has been done for the first time. We are just analysing the results, so we cannot tell yet what actually happens. There have been some previous studies in other birds, but in terms of total oxygen use I don’t think there is a single study that has fully investigated this question. However, it is known that birds have an elaborate network of air sacs, and they actually use up more oxygen than humans and have a very clever way of using oxygen both in flight as well as in vocal production. So there is a lot of work that needs to be done.

Question 3: I am completely convinced, as I am sure we all are here, that magpies mimic. The evidence is quite clear and very impressive indeed and beautifully done. But I am not quite certain what the functional significance of it is. Magpies are not going to go on the stage and do a Joan Sutherland. What are they training for, as it were?

Gisela Kaplan: There is some evidence but it is sporadic. Mimicry takes a few seconds to perform, so to observe it in the wild is quite difficult and takes a good deal of patience to discover, but there have been instances where we have been able to observe that mothers ‘feed’ their youngsters with the information of some of the mimicry items of the territory and there may be some functional relevance to that.

But my first suggestion is that mimicry is important in terms of the study of learning, and that can be quite independent of knowing about its function and in this approach we can get away without enquiring into possible functions.

The second point I want to make is that we have thought of mimicry as a unitary item: mimicry is mimicry, and it is always the same. There seem to be forms of mimicry that are, in fact, accidents of copying, as I believe Peter Slater and Meredith West suggested. That is one possibility, and certainly there is evidence of that, but this may not be applicable to all forms of mimicry. One function of mimicry was investigated in starlings. It was shown that starlings mimic their predators, and it was thought that either knowing the sound of the predators would keep the predators away, because a predator was already there, or the youngsters would learn about predators via the teaching by adults. To everybody’s disappointment, the predators were never fooled by the mimicry.

Then there is a third form of mimicry, which I think has not been explored before but one on which I have started to do some work now, suspecting that mimicry may, in fact, have a very specific function in magpies. The suggestion I am making here is novel and rather untested, in that mimicry may be associated with auditory memory and complex cognitive processes. You see, the magpie is a territorial species, meaning it usually stays its entire life in one territory, where it raises its young, obtains its food and it mates with one individual, normally for life-- although there are cases of divorce in magpies, remarriages and all of that involving resettlements. If a bird stays all its life in one territory, then territory is extremely important. And if it can be shown that the mimicry relates only to those sounds that arise from other permanent residents within it, then this may be evidence that it is in fact serving a function.

It may be that magpies form an auditory map, similar to the way we form roadmaps on the basis of various landmarks or geometric orientations. Who is to say that one can’t make auditory memories and use these cognitively? Indeed, it may be vitally important for the bird to know what its territory holds. In other words, mimicry may be prompted by territoriality as a concept of which the bird needs to know about. Whether we should insist that there is a function of this kind in mimicry needs to be treated with some caution though. Obviously, it is slippery ground.