When is communication necessary for animals

I argue that the key condition for the evolution of human language was the extreme interdependency that existed among unrelated individuals in the hunter-gatherer societies of our hominid ancestors. This extreme interdependency produced multiple prosocial adaptations for effective intragroup cooperation, which in partnership with advanced cognitive abilities, set the stage for the evolution of language. Research programs on animal communication systems in nature have proceeded essentially independently of research programs endeavoring to teach language to animals.

This is surprising in light of the early, well-known efforts to relate these two research streams, especially by Hockett and Marler These efforts spurred two questions.

First, can animals be taught human language, even a simplified version? Second, do the natural communication systems of any animals rise to the level of simple language? Research since then has indicated that these two questions may have different answers: I would suggest a provisional yes to the first, and a provisional no to the second. In this paper I address this paradox, and make some suggestions toward its resolution. My paper is divided into four parts.

Hockett concluded that some animal communication systems have some of these design features, but none of them have all the design features, especially the key ones. I narrow this definition by identifying four design features — semanticity, arbitrariness, learnability and cultural transmission, and productivity — as necessary for the system to be classified as a simple language.

Second, I discuss bird song, a case where several but not all of the key design features are present. I will focus on one specific case of a song-based communication system that is clearly complex and nuanced, but nevertheless lacks three key design features, semanticity, arbitrariness and productivity.

Third, I consider the debate, not yet fully concluded, over whether animal communication should be conceived of as a process of information transfer or as manipulation of receiver by the signaler. The debate is germane to our more specific question because it provides a clue as to why we find no simple languages among animals despite the apparent capacity for it in at least some of them.

Finally, I suggest that although there appear to be at least some animals with the cognitive capacity for a language-like communication system, none of them have a social system with extreme interdependency among individuals on the scale of that which existed in the hominid hunter-gatherer system. I argue that this extreme interdependency was a necessary condition for the evolution of human language.

In this section I consider the extent to which the most important design features of human language are found in animal communication systems.

Mutual Benefit or Manipulation? This behavior is primarily tactical, that is, the individual delivering the blow will directly benefit it if its opponent responds by backing down.

Hockett listed prevarication — the ability to transmit misinformation, i. To say that a communication system is semantic is to say that it uses signals to represent particular things or actions. A well-known example in animals are alarm signals given in response to different predators. We can say in such cases that each of these signals represents one of several different predators, or more precisely, the appearance on the scene of one of these predators.

For example, vervet monkeys have three different alarm calls for three different classes of predators: raptors, terrestrial mammals and snakes, predators which depend on an element of surprise to capture the monkey. Although the vervets use these same signals in other contexts e.

Indeed, it is not unusual for an animal to use a particular signal to mean different things in different contexts Smith, , similar to some words meaning totally different things within different sentences. Nevertheless, I will argue later in this paper that the semanticity of animal communication systems is limited: although some things are represented by animal signals, the number of things is generally small.

Attempts to catalog the number of different things signaled in animal communication systems typically top out at 25 or so vervet monkeys, Struhsaker, ; Japanese macaques, Green, ; review in Hauser, The limitation does not appear to be due to production constraints the ability to produce enough distinct signals or to recombine enough of them to enlarge the signal set or to perceptual-cognitive constraints. A distinctive feature of human language is that not only are words semantic, they are arbitrarily so.

These signals seem totally arbitrary with respect to what they signify, and in theory they could be interchanged without problems, so long as senders and receivers were both aware of the convention.

How about animal signals? Identity signals — indicating species or individual identity, and occasionally group or kinship — are perhaps the most common animals signals that unequivocally have the arbitrariness feature.

But many, perhaps most, animal signals are not arbitrary. I am unaware of any clear example where the reverse is true, where the more effective signal is the one that is less conspicuous, for example, a softer sound, a more subdued color, a less vigorous display.

Moreover, quiet song is typically different in other respects besides loudness, for example, having some elements seen only in quiet song, such as very high frequency elements. Other animal signals are simple extensions or slight modifications of tactical behaviors, e. For example, a threat signal in many mammals is the open mouth display, where the teeth, the canines notably, are prominently displayed.

Another common threat signal is the raising of the hair or feathers, making the animal appear larger. Again, while these actions are plausibly considered ritualized displays, they are not arbitrary signals. If they were, you would also find cases where animals threaten by closing their mouths, or by making themselves appear small. In short, animal signals functioning to impress an opponent or potential mating partner are usually inherently impressive, not arbitrarily selected to represent threat or desirability.

An upright animal, with its hair raised, its tail raised, and staring at its opponent inherently appears dominant, whereas one with a flattened, slinking body, hair down, tail down, and looking away from the opponent, inherently appears subordinate. Many epigamic signals — signals designed to attract a mate and induce her to mate — are bright, striking ornaments, often ones that function like supernormal stimuli e. Many epigamic signals are energetically expensive and highly skilled behaviors, such as the complex male courtship dances of wolf spiders and jumping spiders Hebets and Uetz, ; Elias et al.

The motor performance revealed in these sorts of displays likely reflect whole-organism performance relating to survival, and thus should be good indicators of individual signaler quality.

There is considerable evidence that females choose mates in nature based upon their evaluations of male motor performance reviewed in Byers et al. The relevant point here is that these signals are not arbitrary, but inherently reflect the trait signaled: signaler quality.

Even in the example par excellence of communication of information about the external world — the honeybee dance language — the signals are not quite so arbitrary as generally assumed. For example, if the dance is done outside the hive, where the sun is visible, the bee dances with respect to the actual position of the sun, rather than with respect to the vertical Gould, That is, outside the hive, the symbology is not truly arbitrary. Moreover, the distance to the target is represented by the duration of the straight run — the further the distance, the longer the run — so this is at least partially non-arbitrary as well.

But what would be considered an extra-linguistic feature for humans is often the primary message in animals. For example, the initial stage of a battle between two male red deer consists of a roaring contest Clutton-Brock and Albon, This vocal signaling duel does far more than simply establish that each animal is a male conspecific ready to defend or fight for the harem — this undoubtedly was perceived by both parties before the contest began — rather, how loud and how long an individual roars establishes how motivated and formidable he is, and is used by the receiver to decide whether to continue the fight or depart.

Similarly, the plumage ornaments and courtship dance of a male golden-collared manakin do far more than simply identify species and sex — that is simply the necessary first step — the brightness of the ornament and the skill of the dance determine whether the receiver, the female, will choose to mate with this particular male or continue her search for the best possible mate Stein and Uy, ; Barske et al.

In summary, although we have examples of animal signals that are totally arbitrary, many others — perhaps most? I would add that to date we have found nothing comparable to the many different human languages, which are a consequence of the arbitrariness feature.

We do find geographical dialects in animals e. Human language is both learned and taught. Most animal communication systems are neither. A well-known exception to this generalization are the learned vocal communication signals of several taxa, most notably the oscine passerines songbirds , hummingbirds and parrots among birds, and cetaceans and at least some bat species among mammals reviews in Janik, ; Knornschild, ; Nowicki and Searcy, Evidence for vocal learning and cultural transmission in some other birds and mammals as well Walcott et al.

Later in this paper I return to the best-studied example of vocal learning, song learning in songbirds. Where the communication signals are learned, we should expect to find dialects, geographical variation in the signals. The occurrence of dialects is one criterion for identifying the occurrence of learning and potentially evidence for the arbitrariness design feature. An example that may illustrate the arbitrary nature of dialects is the recently-discovered modification of the song in eastern white-throated sparrows to resemble the typical song of western white-throated sparrows.

Investigators have traced this change to eastern birds learning the western version of the song on the migration grounds, where individuals of the two populations mix Otter et al.

Although Otter et al. Perhaps even rarer in animal communication systems than learning is teaching. The commonly accepted criteria for demonstrating teaching in non-human animals are that 1 teachers should modify their behavior in the presence of the learner, 2 this change in behavior should result in no immediate benefit to the teacher, and 3 the learner should acquire a behavior quicker or better as a result Caro and Hauser, The sense in which I am using this term is captured by Hauser , p.

Indeed, semanticity representation and productivity are probably the two central features of human language: by combining basic phonetic units into larger meaningful units, and combining these units further via syntactical rules, we can say almost anything.

Animal communication systems are not productive in this sense, and this is the primary reason we do not refer to them as languages. I note that while there is some controversy in phonetics about exactly what are the units of productive combination, there is agreement that all natural languages including sign language are made up of meaningless atomic units that are combined into larger meaningful wholes Zuidema and de Boer, Instead of productivity, we could describe the communication system in terms of information capacity.

The information capacity of human language is essentially infinite, in the sense that, in theory, we can communicate virtually anything. Our motor, sensory and cognitive capacities obviously will reduce how much information actually gets transmitted and received.

But still, the fact is that we can transmit an enormous amount of information with language. Attempts to measure information capacity or information transmission in animals, on the other hand, have given rather modest results. Two estimates of the information about distance and direction in the honeybee dance language have given a high value of My group has estimated the information capacity of the call signature system that parents of the colonial cliff swallow use to find their offspring in their large breeding colonies Medvin et al.

We estimated the capacity as 8. The information capacity of human language of course is orders of magnitude larger than this. We certainly find the potential for productivity in bird song. Still, although the units are there, and although songbirds may possess the cognitive capacity to comprehend hierarchical structuring in vocal signals Gentner et al. As Hauser , p. As discussed in the next section, theories on the function of song repertoires abound, but they all agree that the different songs function simply to provide diversity, rather than to represent different things.

Table 1 summarizes the conclusions of this section. The natural communication systems of animals fall short of human language on a number of the key design features of language. They come closest on semanticity, where signals sometimes represent things in the external world or within the signaler, and the signals are sometimes truly arbitrary. However, more commonly animal signals are not arbitrary but inherently meaningful, e. Most animal communication signals and responses are neither learned nor culturally transmitted.

And, so far as we know, no animal communication has the sine qua non of language: productivity. Table 1. Key design features of communication systems after Hockett, , pruned and combined. The oscine passerines songbirds are one of the rare animal taxa in which individuals learn their vocal communication signals. It has long been noted that vocal learning in songbirds has many similarities to language learning in humans Marler, ; Doupe and Kuhl, These similarities include the following.

While notable differences exist among songbird species with regard to the normal progression of song learning Beecher and Brenowitz, , these six features are essentially true for all of the many songbirds that have been studied to date. Despite the notable parallels between bird song learning and human language learning, none of the many studies endeavoring to teach a version of human language to animals have focused on songbirds.

This is all the more surprising given the language learning shown by Alex the African Gray Parrot, a member of another avian taxon with vocal learning, the psittacines Pepperberg, , Moreover, songbirds have strong cognitive capacities, a highly-developed vocal production mechanism, and a vocabulary of basic sound units in their song that rivals or exceeds the basic sound units of human language.

There are even songbird species that can mimic human speech sounds e. On the face of it, all the requisites would seem to be there to support a simple language in a songbird.

For example, song sparrows typically have nine plus or minus two or so very different songs. Each of these songs is made up of 5 or 6 distinct elements, and the order of these elements is important Horning et al. Song sparrows are somewhere on the middle of the song repertoire complexity scale: many species have larger and even more complex song repertoires. The key point for this discussion is that song repertoires provide clear potential for productivity, as song sparrows and many other songbirds have as many or more distinct units in their vocal communication systems e.

The most popular hypothesis about song repertoires for north temperate zone songbirds — where only males sing — is that they are an epigamic signal produced by males to attract females and that larger repertoires are more attractive than smaller ones Catchpole, ; Searcy and Yasukawa, ; MacDougall-Shackleton, ; Collins, Focusing on just the well-studied song sparrow, the evidence for this hypothesis is mixed Searcy, ; Reid et al.

The handicap principle, discussed in the next section, would suggest that if large song repertoires are preferred, it is because they are an indicator of some aspect of male quality.

Reid et al. Anderson et al. However, they found no correlations between repertoire size or two other measures of song learning ability with an overall measure of learning ability based on five different learning tasks. I should note, however, that a correlation of vocal learning ability with both overall learning ability and mating success has been found in another songbird, the Satin Bowerbird, a vocal mimic: in this case the vocal learning ability is the ability of males to mimic the calls of other local bird species, both the number of species mimicked, and the accuracy of the mimicry Coleman et al.

According to another hypothesis, song repertoires play a role in territorial competition, which in north temperate zone songbirds, where only males sing, is largely male-male competition, but outside the north temperate zone where both sexes sing, is pair-pair competition e. There are several hypotheses as to how repertoires might work in the territorial competition context. As one piece of evidence, he points to a positive correlation between repertoire size and population density in marsh wren populations, and also to the finding that birds in denser populations cycle through their songs faster, again a behavior that should reduce habituation Kroodsma, In contrast, song sparrows sing their much smaller repertoires with eventual variety, i.

In western, resident populations of song sparrows, song repertoires may function primarily to provide a bird with songs matching all or most of his neighbors, and thus potential individualized replies to each one of them Beecher et al.

Although as this brief discussion indicates, the theoretical debate has not yet concluded, the take-away point is that none of these hypotheses view song repertoires as a form of semantic communication.

Rather they view repertoires as having a direct effect on the receiver dishabituation , or as permitting individualized replies to multiple neighbors, or as quantitative signals with inherent rather than semantic meaning, that is, more songs or more song syllables are simply more effective.

I should add that most single-song species appear to have the potential to develop song repertoires yet do not tap into this potential. For example, when examined over an entire population, indigo buntings have a repertoire of over a distinct song syllables, yet a given individual uses just 6—8 of these in the single song it develops Rice and Thompson, ; Thompson, ; Baker and Boylan, In this section I describe one such case: how song sparrows use the songs in their song repertoire to negotiate territorial disputes.

The general point I will make is that their communication system is surprisingly complex and versatile, despite being neither semantic nor productive. Although I will not attempt to generalize to all songbirds given the incredible diversity of the song communication systems seen in this group Beecher and Brenowitz, , I suspect that this conclusion — complexity without productivity — applies broadly to songbirds, and perhaps to all animals. Song sparrows have a territorial system like that found in many animals and typical of many songbirds.

An individual carves out a territory where the mated pair will nest and raise their young, doing most of their feeding on the territory. Suitable habitat is typically densely occupied by conspecifics, so territorial disputes can arise during both the establishment and maintenance stages.

Because in territorial animals, neighbors have no fences, neighbors need to renegotiate territory boundaries from time to time. Negotiation can progress into fighting but avoiding fighting may benefit both parties and this common interest favors reliable signaling. Song sparrows in western, resident populations use their repertoires in a complex way to carry out territory negotiations.

Although they will engage in serious fights, established neighbors use their signaling system to avoid fighting if possible.

Before fighting they typically give their high-level threat signals, wing waves and soft song Searcy and Beecher, ; Searcy et al. Because western song sparrows learn songs from their neighbors in the area to which they disperse after fledging, a bird typically shares some of his songs with each of his immediate neighbors. The set of songs he shares with one neighbor is typically different from the set he shares with another. A partial example is shown in Figure 1. For example, if we represent the different songs of a bird with different capital letters, and the shared songs of neighbors with the same capital letter, then Bird 1 might share his song types A, B, and C with his neighbor Bird 2, his song types C, D, and E with another neighbor, his song types E and F with a third neighbor, and finally G, H, and I with no neighbors e.

A typical territorial negotiation might occur as follows. Bird 1, aiming to establish this new boundary, moves to that point and sings at his neighbor. Typically the two birds would still be a considerable distance apart at this point and out of sight of one another territories are large and song is a long-distance signal.

Let us say bird 1 sings B. Finally, rather than type-matching or repertoire-matching, Bird 2 can de-escalate by singing one of his unshared types, e.

Singing an unshared type is better than not singing at all because it signals that although the singer is not engaging, he is on territory and has heard his neighbor; it is a signal likely used for example when the bird is busy feeding recently-fledged young. Figure 1. Partial song repertoires of two neighboring birds. Shared songs are shown in the top three rows, and four of their unshared songs in the bottom two rows they are arbitrarily paired. Frequency scale: 0—10 kHz. Songs are 2—3 s long.

The system while not in itself resolving anything, does give the neighbors time to defuse the situation or work out a compromise. Note, however, that the semantic content is limited. No particular song in the repertoire means a particular thing.

Songbirds check several of the design feature boxes and they would appear to have the potential to use their songs in a productive way, i. However, despite considerable debate concerning the function of song repertoires, the different repertoire hypotheses all agree on one point: that the function of the vocal diversity is diversity per se , not the transmission of different messages with different songs.

Perhaps even more surprising, many single-song species have large song syllable repertoires an individual could tap into, but instead each individual uses just several of these syllables to develop its single song.

No songbird rearranges its multiple song syllables into different songs that signal different things. I echo here the conclusion of Fitch and Jarvis , p. Songbirds may use their repertoires in subtle, nuanced ways, as with the song sparrow hierarchical signaling system I described above, but what the system achieves seems better described as the management of behavioral conflict than as an impressive transmission of information.

That is, the system may function well, but it does not function like a language. In this section I discuss the debate within the field about the fundamental nature of animal communication. I believe this debate has provided us with a key to understanding why we find no examples of a simple language among the many communication systems of non-human animals, and true language only in the human animal.

We can trace the real beginning of the field of animal communication to the classical ethologists e. The ethologists provided detailed descriptions of animal signaling systems in nature, developed theories about the underlying proximate causes e. On the question of the function of animal signaling systems, they took a group-selectionist perspective: the benefit that a signaling system provided went not to signaler or receiver per se , but to the species see Tinbergen, definition in Table 2.

The assumption of mutual benefit seemed natural in cases where sender and receiver have a strong common interest, e. But as investigators began considering the many cases where signaler and receiver have conflicting interests, such as in agonistic encounters over an indivisible resource, they began to question the mutual-benefit, information transmission view.

They asked two questions about such cases. First, do both parties have to benefit? The manipulation viewpoint was famously developed by Dawkins and Krebs who argued that rather than expecting signalers to signal honestly, we should expect them to manipulate the receiver to their own advantage, e. Since the Dawkins and Krebs paper, the debate has continued as to whether it is justified or productive to conceptualize animal signaling as an information transmission process in which both parties benefit.

Simplifying somewhat, I will distinguish between the Information Transmission and Manipulation approaches to animal communication. Strong arguments on the information side over this same period include Green and Marler , Smith , Bradbury and Vehrencamp , Searcy and Nowicki , Carazo and Font , Seyfarth et al.

Definitions from some of these sources are included in Table 2. In conceiving of signaling as manipulation, Dawkins and Krebs essentially treated the communication interaction like a zero-sum game.

This seems reasonable in cases like disputes over an indivisible resource a food item, a territory, and a mate , and also in epigamic selection, where a male tries to persuade a female to mate with him now rather than to continue searching for a possibly better male. Although the manipulation view was enlightening in many respects, as originally presented it had a serious weakness: it gave no agency to the receiver.

While it was sensible to expect signalers to signal for their own benefit, why should we expect receivers to be passive in these evolutionary scenarios, especially if being manipulated by the signaler is costly?

Indeed, receivers can do more than simply ignore signals that do not benefit them: they can require signals that do benefit them, even if those signals are costly to the sender. For example, in many species males must sing or call to attract a female for mating.

If the male does not vocalize, potential female receivers will simply not engage. Moreover, these vocal signals may attract predators, a cost borne by the signaler but not the receiver. Indeed, the most effective or most-preferred signals may be the most costly, e.

When a male adds chucks to his calls, he not only attracts more females, but also predators: frog-eating bats that home in specifically on the chucks. Similarly, a calling male field cricket attracts more females than does a silent male, but he also attracts more parasitoid flies, and louder calls attract both more females and more parasitoid flies Cade, In some populations the rate of fly parasitism is so high that males have lost the ability to sing Zuk et al.

When a territorial songbird is deprived of its voice, however, potential rivals show up and proceed to take over its territory e. The receiver manipulation view prompts us to consider how the receiver might demand a more honest signal.

There are two related possibilities. First, the receiver can selectively attend to signals that are inherently honest due to physical constraints. Davies and Halliday showed that playback of low-pitched calls was sufficient to discourage smaller males from entering into battle with an apparently larger male. This principle was first proposed by Zahavi , modified and formalized by Grafen , given the intuitively pleasing graphical formulation by Johnstone shown in Figure 2 , and is still being subjected to further modification and clarification e.

One of the clearest demonstrations of honesty in an epigamic signal was carried out by Petrie and her colleagues on that poster animal for epigamic signaling, the peacock. Further Reading Hauser, M.

The faculty of language. What is it, who has it, and how did it evolve? Science, , — Rendall, D. What do animal signals mean? Animal Behaviour, 78 , — Seyfarth, R. Signallers and receivers in animal communication. Annual Review of Psychology, 54 , — Zhanna Reznikova 1 Email author 1. For instance, horses will kick each other to ward off threats or when competing for a mate.

As kittens, cats will nuzzle their mothers to show affection. Many species of primates will clean each other to bond and show affection. For some species of animals, communication is about pheromones or chemical markings.

They will leave their own scents to mark their territories, ward off predators or attract a mate. The most well-known example is that of skunks "spraying" their signature scents when threatened. Cats will rub against objects to make their territories, leaving scent markers from their mouths.

Crystal Owens is the managing editor at a Northern Virginia newspaper with more than 10 years experience in journalism. She has worked as a reporter in Florida, Pennsylvania, South Carolina and Georgia, covering various topics from crime to politics to health care.