"Kissine offers a new theory of speech acts which is philosophically sophisticated and builds on work in cognitive science, formal semantics, and linguistic typology. This highly readable, brilliant essay is a major contribution to the field."
Hauser, Marc D. and Mark Konishi, ed. (2003) The Design of Animal Communication, MIT Press. (Paperback reprint of 1999 hardback edition)
Announced at http://linguistlist.org/issues/14/14-1970.html
Anne Reboul, Institute for Cognitive Sciences, CNRS, Lyon
INTRODUCTION
This book is the result of a symposium on animal communication organized in honor of Peter Marler in 1997 at UC Davis. Papers deal with mechanisms, ontogeny and evolution of communication among such diverse species as frogs, birds, bees, crickets, bats, non-human and human primates. Though it is not an easy read, it still is quite accessible to non-specialists, including linguists. One of its interests for linguists is that it enlightens the specificities and commonalities of animal communicative systems with the human language, thus laying a basis of knowledge on which it is possible to assess what is specific to language. In other words, it should be relevant to any linguist with at least an intellectual interest in the complex question of language evolution. One should not be daunted by the sheer size of the book (701 pages including the index and TOC): the fascinating insights provided in the quite real complexity of animal communication systems as well as the physiological mechanisms which sustain them are more than enough to balance the effort involved.
SYNOPSIS
The book has three main subdivisions, each with a short introduction. These three parts are ''Mechanisms of communication'', ''Ontogeny of communication'', and ''Evolution of communication''.
The ''Preface'' to the book, written by Konishi, highlights Marler's contributions to ethology of communication. Konishi also writes the introduction to the first part, ''Mechanisms of communication'', insisting on the wide range of methods used by animals to communicate and introducing the papers which follows.
The first paper of the first part, by Kelley and Tobias, ''Vocal communication in Xenopus laevis'', studies vocal communication among one variety of South African clawed frog. Vocal communication is used both intersexually to promote reproduction and intrasexually among males for sexual competition and territorial defense. Thus vocal communication is different among the sexes, and these differences ''have been traced to the action of gonadal steroid hormones, which act during development to control the morphogenesis of the vocal organ and during adulthood to control how the system produces male- and female- specific songs'' (9). The vocal repertoire of this species is based on clicks produced by the larynx, with modulation of frequencies accounting for the different types of calls, with different functions. Sexual differences in calls are reflected in differences between the vocal organs of males and females as well as differences in the neural circuitry used in vocal production, reflecting the action of gonadal steroids. Though the production of calls is now well- understood, their perception is less well-known and deserves more attention.
The second paper, ''The Motor Basis of Vocal Performance in Songbirds'', by Suthers, deals with oscine birdsong and its diversity, made more surprising by the relative similarity of the vocal organ, the syrinx, among the different species. This diversity is due to the exploitation of two independent and laterally specialized sound sources in the syrinx, which gives oscine songs its versatility. More specifically, ''the two sides of the brain can deliver different motor programs to the left and the right sides of the syrinx, giving each side an ability to generate sound independently'' (40). This means that a note can be produced on one side only, on both sides or be switched from one side to the other during a syllable. Respiration during the song is typically made through both minibreaths (very short inspirations) and pulsatile expirations producing each sound. In some species there is a lateralization of syringeal function. Questions remaining concern which motor strategies a given species adopts, and whether they are immutable or subject to tutoring.
The third paper, by Nottebohm, deals with ''The anatomy and timing of vocal learning in birds''. Vocal learning in songbirds has been hypothesized to depend on a ''sensitive period'' during which learning would occur more easily. Vocal learning seems to depend in general on two steps, the acquisition of an auditory model, followed by the acquisition of the motor skill necessary to faithfully reproduce the model. This suggests that there could be more than one sensitive period. All oscine songbirds are vocal learners though it seems that there are two main categories among them: age-limited learners, in which vocal learning only occurs during the juvenile stage and ends with sexual maturity and open-ended learners who add to the song repertoire acquired as juveniles during adulthood.
Deafening has a different effect on age-limited learners who usually retain their songs, while open-ended learners generally loose them. Brain pathways seem however to be pretty similar among songbirds, with a distinction between two pathways, one developing early in ontogeny for the production of unlearned sounds (e.g. food-begging calls), the other developing at the time when song is necessary, is found only in vocal learners seem specialized for the control and performance of learned sounds, i.e. songs. There is a sexual difference in some species relative to the development of this second pathway, which is more developed in males than females. Without going into anatomical details, there are four feedback loops used in the acquisition and production of songs. Some of this brain circuitry is in place before the sensitive period, though other parts develop, both in volume and connectivity at the onset of song development, in which the typical sequence of food-begging calls, subsong, plastic song and song crystallization depends, at least in part, on the timing and order of such anatomical changes. In the same way, the end of the sensitive period may depend on the reduction of synapses in some part of the brain. In open-ended learners, brain parts relevant to the acquisition of new songs will change in volume, such changes being associated to the seasonal occurrence of song learning. However, even age- limited learners acquire song memories in adulthood and can discriminate between the songs of strangers. There is also evidence of brain lateralization, though a reversion of dominance seems possible in open-ended learners (but not in age-limited learners). Song is mainly a male behavior and the development of the song system has been shown to be due to male hormones in some species. Nottebohm also reviews social effects on song acquisition. What is more, hearing and producing song induces changes in gene expression relevant for the development of brain song learning circuitry.
Michelsen addresses ''The dance language of honeybees: recent findings and problems'' in the fourth chapter, coming back to one of the first systems of animal communication precisely described. The dance language of bees has two unusual features: it is used to communicate about a remote event (the location of a source of food) and it is an abstract code. The dance occurs in the hive, the dancing bee walking in a waggling way (the bee wags her body from side to side) on the comb and circling back, alternating her return path between left and right. The problems Michelsen addresses are the following: how can the follower bees (the ''adressees'') perceive the message in the darkness of the hive? Which components of the dance do the follower bees perceive as signals? Three hypotheses have been proposed relative to the first question: (tactile) perception of dance sounds through the vibrations of the comb, physical contact between the dancer and follower bees, perception of the air currents triggered by the movements of the dancer. Michelsen reviews all three hypotheses, relating them to the second problem, by examining whether the information transmitted in these three ways could be precise enough relative to the distance and location of the source of food. He concludes that none of these hypotheses is quite satisfactory and reviews one of his own experiment using a mechanical model of a dancing bee, producing airflows similar to those of dancing bees and succeeding in recruiting bees to the location indicated (though distance was not always successfully transmitted). The wagging run appears to be the major component for conveying distance. These results however are far from answering all the questions raised by the dance of bees and further experiments will be necessary.
The fifth chapter is dedicated to ''Processing species- specific calls by combination-sensitive neurons in an echolocating bat'' and is by Kanwal. In addition to echolocation signals, bats have ''a complex repertoire of communication sounds or 'calls' that rival in variety and complexity of spectral structure those emitted by the most vocal primate species'' (133). He quickly reviews call processing in human and non-human animals, concluding that ''neural specialization can be based on facilitation of the response to combinations of call components'' (136). Such specialized neurons have been found in the processing of echolocation signals in bats and are also involved in call processing. The calls of bats are produced from a simple syllables combined to produce a given calls. These combinations, however, are strictly constrained, as shown by the fact that only 15 out of a possible total of 342 disyllabic combinations are actually produced. Different species of bats have different dialects, indicating neural plasticity and audiovocal learning.
Perret authors chapter 6 on ''A cellular basis for reading minds from faces and actions'', centering on primates abilities in this domain, and more precisely on visual processing. In this instance, ''visual recognition involves both the analysis of the visual image and the associations between what we see, what we feel and what we do'' (159). He approaches the question through a study of ''the response properties of single cells in the temporal cortex of macaques'' (159).Three types of cells are relevant: cells which encode the visual appearance of body and face both static and mobile; cells which encode specific bodily or facial movements; cells which respond to particular face and body movements as goal-directed actions. The first type of cells show generalization in as much as they respond irrespectively of the species of the perceived animal. They subdivide in cells selectively responsive to different parts of the body and are sensitive to orientation and perspective. Generalization for the same posture is achieved by combining the outputs of view specific cells. There is evidence for hierarchical combinations (e.g. eyes >> head >> body). The sensitivity to orientation is necessary for gaze following and for directing attention to a target. The second type of cells include cells which react to certain facial expressions and it is well-known that some neuropathological conditions (e.g. relative to amygdala) can disrupt the recognition of particular emotional expressions (notably fear) in humans. This capacity is lateralized in humans. The third type of cells responds selectively to particular body movements, interpreted as goal-directed, through the relation of the movements to other aspects of the visual environment. However, ''visual information about motor acts is also encoded in particular premotor areas'' (174). Thus, some prefrontal cells code both the motor component and the visual appearance of specific movements. They seem to code the ''general concept'' of the action. A two-steps scenario could be proposed, with temporal cells recognizing the movement and alerting prefrontal motor cells. These prefrontal cells ''share some features of visual selectivity'' (174) with temporal cells, which can be interpreted as supporting the hypothesis that ''substrates of perception and action overlap, at least to some extent'' (174). These temporal populations of neurons seem to have homologues in the human brain. There is a fourth type of cells in the temporal area which seem to distinguish between self-generated and other-generated dimensions of visual stimuli, maybe relying (respectively) on predictability and unpredictability of change. In conclusion, Perret points out that this sophisticated visual system may be sufficient for a complex social life without needing an additional mind reading capacity.
Adolphs signs chapter 7 on ''Neural systems for recognizing emotions in humans''. The chapter is mostly dedicated to fear, as an emotion that has received the most attention in animal studies and, hence, offers a good ground for an evolutionary approach. There are three domains of study regarding emotion: knowledge about emotion, experience of emotion, and expression of emotion. The paper focuses on ''the retrieval of knowledge about an emotion, on the basis of a stimulus that denoted or signals the emotion'' (188). The lesion method is applied, i.e. the performance of brain damaged patients is compared to that of healthy subjects with a view to establishing correlations between specific brain regions and specific tasks. Such factors as age and IQ are controlled. The stimuli have been facial expressions. There are six basic emotional expressions, recognized easily by normal subjects across cultural differences (with some overlap, however) corresponding to six basic emotions, i.e. happiness, surprise, fear, anger, disgust and sadness. Valence and arousal are two attributes that play a role in the recognition of emotions. ''Emotion concerns the continual changes of body and brain states that occur in an organism as it interacts with its environment'' (192). Thus the neural structures involved in processing emotions are both structures that represent body states and structures that link perception to body states. One such structure is the amygdala though its role seems more complex than was once thought. However, it does appear to play a major role in fear-conditioning in animals. Adolphs reports experiments similar to fear-conditioning in animals, involving one patient with selective bilateral damage to the amygdala, with similar results. The same subject was unable to recognize or draw facial expression of fear and showed impaired conceptual knowledge of emotions. The suggestion is that ''the amygdala may connect percepts of sensory stimuli (...) with a variety of neural systems involved in response to and knowledge of such stimuli'' (198). This would mean that the amygdala is involved in encoding rather in retrieval, an interpretation supported by recent findings. More generally, there is evidence of brain lateralization for emotional cognition (with right hemisphere involved in negative emotions and left hemisphere in positive ones in the human brain), which also seems the case for non-human primates. As lateralization also occurs for language, a suggestion is that, as processes involved in communication, both language and emotion have to be processed fairly quickly, lateralization being a premium because it reduces physical distance between the neural systems involved in, respectively, language and emotion. Adolph ends his paper with the hypothesis that expression and recognition of social signals, including emotional ones, have coevolved.
The 8th paper comes back to songbirds, Ball writing on ''The neuroendocrine basis of seasonal changes in vocal behavior among songbirds''. In temperate-zone birds, behavioral changes occur over the course of the year, notably in song. Oscine songbirds produce song more frequently in late winter and early spring, through endocrine and neural mechanisms. This is of course linked to reproduction that occurs in favorable circumstances climate wise. The cues most prominently involved in triggering the reproductive period, and hence the singing period for males, are changes in photoperiod. Over a certain treshold, birds are photostimulated, a process inducing ''a physiological cascade that includes increases in gonadotropin secretion, gonadal growth, and a range of hormone-dependent reproductive processes, including behavior'' (215). This photosensitivity in birds decreases over time leading to photorefractoriness, ensuring that birds stop breeding at an appropriate time. Short days then dissipate photorefractoriness and the cycle can begin again. Side by side with this ''absolute'' photorefractoriness, some species exhibit ''relative'' photorefractoriness, continuing breeding as long as long days persist. ''There may be a difference between the neuroendocrine mechanisms underlying the two forms of photorefractoriness'' (217). In the brain song control circuit of song birds, there are cells possessing receptors for gonadal steroid hormones leading to changes in the volume of particular nuclei of the song circuit and controlling survival and migration of new neurons. However, testosterone is not equally effective at all times of the year and ''plasticity in the starling song system appears to require the coordination of the appropriate hormonal milieu with a permissive photoperiodic condition'' (237). Photosensitivity may also have effects independent of testosterone. This suggest ''photoperiodic differences in steroid receptors or in steroid metabolism'' (239).
Wingfield et al. sign the last chapter (9) of the first part, on ''Testosterone, aggression, and communication: ecological bases of endocrine phenomena''. Agonistic communication, as manifested, for instance, in territorial aggression, is an important part of communication and seems regulated by androgens. However, though aggressive behavior seems similar at different seasons, it is not clear that it is and neither is it obvious that the contexts and hormonal bases for it are similar. Wingfield et al. introduce Jacobs' finite-state machine theory to descrobe different life-history stages where aggression may occur. Each stage has environmental input signals and yields a fixed set of predictable outputs, this being regulated by neuroendocrine systems. Just as different environmental factors can yield superficially similar behaviors, there is no reason to think that internal mechanisms will be identical in each case. History-stages are pre-determined by environmental cues such as day length, temperature, etc. though some unpredictable factors may trigger ''facultative'' life- history stages. The authors' view is that, while testosterone may control aggression during the breeding season, other factors, or testosterone but with different mechanisms involved, may operate at other stages and in facultative stages. Thus the general picture of the biological control of aggression may be much more complex than has generally been thought.
The second part of the book, on the ''ontogeny of communication'' opens with an introduction by Hauser, which should be of interest to anyone interested in the nature/nurture debate. Hauser points out that the debate stems from a predilection for dichotomies and that a reasonable position is to posit an interaction between gene and experience. He contrasts possible ontogenic trajectories and their underlying causes, advocating the view that ''the genome sets up what is possible, allowing experience to guide the organism to a stylized phenotypic outcome'' (286). He then introduce the different papers in this second part.
The first paper of this second part, chapter 10, is signed by Peter Marler and speaks ''On innateness: are sparrow songs 'learned' or 'innate'''. Marler points out that the use of dichotomous pairs such as instinctive/acquired, learned/innate, learned/unlearned are more of a hindrance than an help in the study of behavioral development. Taking oscine songbirds as an example, he points out that though some bird calls and songs are innate, oscine birds are learner species with a specific specialized network of brain nuclei, which lacks both in the brain of non-singing birds AND in the brain of birds with innate songs. Learned songs can be very simple (as innate songs) or very complex. There are three stages in the learning of a song: memorizing a song, storing it and producing it. ''During the memorization phase, there are sensory constraints that take the form of learning preferences'' (296-297), and ''predispositions are also manifest in the motor phase'' (297). These last are pan-specific but other predispositions are species-specific and Marler concentrates on these. Experiments regarding the role of innateness and experience in song learning have concentrated on isolates of different species. Such birds produce simpler songs and seemed influenced in the same way by isolation, suggesting a major role for experience. However, songs from isolates of different species were still recognizably different, indicating species-specific innateness. ''So, we have a paradox'' (300). ''Neither conclusion is satisfactory'' (302). Marler then reviews experiments focusing on early-deafened songbirds, ending with just as contradictory results as isolates experiments. He suggests that ''auditory input is significant, not only for the memorization and production of learned songs, but also for the activation of latent central mechanisms that actually encode more information about normal species- specific song structure than is usually evident in the amorphous morphology of deaf songs'' (308). Marler then insists on the fact that ''the potential for varying degrees of what evolutionary biologists call phenotypic plasticity (...) with the same genotype giving rise to more than one phenotype in different environmental circumstances, is extremely widespread if not universal'' (313). Thus, variability is not an absolute indicator of learning, any more than homogeneity would be an absolute indicator of innateness. Indeed, ''to achieve a full understanding of the evolutionary process, neither genetic nor environmental contributions to this interactive process can ever be ignored, a position which becomes self-evident if we approach the nature-nurture problem in developmental terms rather than as an issue of phenotypic typology'' (314).
The 11th chapter, by Kroodsma, is devoted to ''Making ecological sense of song development by songbirds''. Kroodsma begins by noting that songs are used to manipulate conspecific individuals, of which there are two sorts, males and females. This means that there may be a divergent evolution inside a given species regarding intersexual and intrasexual songs. This is the case of male chestnut-sided warblers, who not only have strikingly different kinds of songs for addressing males and females conspecifics, but also have different song-learning programs. Intersexual songs, used for matting, are of four different types, of which one is usually favored by a given male, though his ability to sing the other types is preserved. Thus intersexual songs are fairly stereotyped and it seems that ''it is the female that selects for this stereotypy'' (322).
Intrasexual songs have a small repertoire of four or five songs, which are shared among neighbours, but which changes with geographic distances. Of these four ''neighborhood'' songs, a given male usually favors the one which is rarely used by his neighbors. Inter- and intrasexual songs are not learned in the same way, neither do their respective learning program obey the same constraints. Given the homogeneity of intersexual songs, they can be learned anywhere and are typically learned during the hatching year. Intrasexual songs, by contrast, must be learned just after migration ''when and where the male acquires both his breeding territory and his singing neighbors'' (323). The uniformity of intersexual songs suggest that they are learned from many males to average individual differences (and learning may be innately constrained), while the variety of intrasexual songs suggest that they must be learned in situ, from the neighbors (and learning may not be as strongly innately constrained). Kroodsma then reviews a few scenarios for achieving both uniformity and variability, highlighting genetic constraints, learning from a single individual or from multiple conspecifics, improvising, as well as other ecological and social factors. He concludes that ''strategies of development, such as whether songs are imitated or improvised, should also be influenced by the probability of interacting with the same individuals throughout life'' (337).
The 12th chapter, by Doupe and Solis, is ''Song- and order-selective auditory responses emerge in neurons of the songbird anterior forebrain during vocal learning''. As said above, there are two phases in song learning, the first one for the memorization of a song (where it is crucial that the bird listen to a tutor's song) and the sensorimotor phase where birds begin to produce song, but do not need to listen to a model anymore. ''These observations indicate that song learning must involve memorization of the tutor song during the critical period'' (343), i.e. during the first phase. However, during the sensorimotor phase, birds need acoustic feedback (birds deafened before the singing phase produce abnormal songs). Thus, both phases crucially depend on hearing. During song learning, in addition to auditory brain areas, the brain song system is involved. This is the case for the anterior forebrain pathway of the song system, which ''contains auditory neurons, which change dramatically during learning'' (345), establishing a spectral and temporal selectivity for the bird's own song. This suggests that ''the development of this selectivity is an experience- dependent process'' (352). During the sensorimotor phase of song learning, some of these neurons will show preference for the bird's own song, while others will show preference for tutor's song, and still others respond equally well to both. There is, however, a correlation between the preference for bird's own song over tutor's song and the similarity between these two songs. This ''suggests that a bird's neural selectivity is related to its particular stage or accuracy of song development'' (357).
In chapter 13, Mundinger examines the ''Genetics of canary song learning: innate mechanisms and other neurobiological considerations''. Canaries have the advantage ''that several genetically different strains were produced by artificial selection'' (369). Some strains were selected for their appearance while others were selected for their songs. In the first ones, the songs are still similar to those of wild canaries while there are important differences in the second ones. Among the second type of strains, a strain is partially deaf. The goal of Mundinger's paper is to unravel the effect of genetic differences on ''interstrain differences in song learning preferences and production'' (370). More specifically, the question addressed was: ''are there interstrain differences in learning wild canary patterns, and if so, how do they differ?'' (371). The method was to compare the learning performances of the different strains of canaries as well as crossbreed individuals in several successive experiments. The results plead in favor of interstrain differences in learning wild canary song patterns, though the differences do not appear to be due to syringeal constraints. Regarding genetics, the following conclusions were reached: ''(1) [the genetic system] is polygenic; (2) the genes are on several chromosomes; (3) one of these is the sex chromosome; and (4) other genes (...) are autosomal [i.e. not sex-linked]'' (376).
Seyfarth and Cheney devote chapter 14 to the ''Production, usage, and response in nonhuman primate vocal development''. Though vocal production in primates seems to be innate and hardwired, vocal usages (the appropriate use of calls) and responses to vocalization may not be. Indeed, this is where parallels with human language learning could be found. In vervet monkeys, there are three main types of vocalizations: alarm calls, social grunts and a ''wrr'' signal indicating the presence of another vervet group. While the production of alarm calls seems innate (infant calls are indistinguishable from adult calls), grunts and ''wrr'', though partially innate, do change with age. Alarm calls are used to signal the presence of five different types of predators and are often misused by infants who use them in the presence of non-predators, while preserving the overall categories (i.e. harmless warthogs will induce leopard calls, but not eagle or snake alarms). Thus part of calls usage is innate, but part is acquired. The same is true of grunts and ''wrr''. In the same way, the acquisition of adult responses to calls develops with age.
An interesting case is the acquisition of response to another species' calls. There, again, ''auditory experience seem[s] the crucial variable'' (396), given that there is no question of innateness here. In the second part of their paper, Seyfarth and Cheney describe a cross-fostering experiment with rhesus and Japanese macaques, two species with the same types of vocalizations, but who use these differently, i.e. ''normally raised one- and two-year-old rhesus and Japanese macaques display a striking species difference in call usage'' (399). ''Cross-fostered subjects generally adhered to their own rather than their adopted species' call usage'' (400), which suggests ''a substantial innate component to the development of vocal usage and that vocal usage is only partially affected by auditory or social experience'' (400). Nevertheless, cross-fostered juveniles did not respond preferentially to calls of their own over their adopted species and both juveniles and adults learned ''to recognize and respond selectively to individual idiosyncrasies in the calls of animals from another species'' (404). Thus, the three aspects of primate vocalizations, production, usage and response to calls are the result of three different developmental processes, each with its own causal mechanisms and rates. Though primate vocalizations differ in certain major ways from human language (absence of generative grammar, rigid development of production and usage), there are some similarities in the existence of different developmental trajectories for production, usage and response, as well as overgeneralization and the ability to associate a sound an a referent. This raises two questions: is association a general learning mechanism? And are the same cognitive processes involved in human and non-human primates?
The next paper, by Kuhl, centered on ''Speech language and the brain: innate preparation for learning'' takes us straight to human language acquisition. Perhaps more than any other cognitive system, language has been the object of a heated innate/acquired controversy. Kuhl addresses the question through tests of phonetic perception in infancy. Though an interactionist view of the relation between development and learning is now generally accepted, ''at issue (...) is exactly how the two systems interact, and particularly whether the interaction between development and learning is bidirectional'' (422). Regarding language acquisition, one of the puzzles is the explanation of universal and orderly transitions in infants. Regarding phonetic perception, a dramatic change occurs during the first year, with new-born infants distinguishing phonetic units of any language and one-year-olds failing to discriminate foreign-languages contrasts. The crucial period seems to be between 7 and 10 months. ''A similar transition occurs in speech production'' (424). These transitions are not purely developmental (linguistic input is necessary). The suggestion is that ''language input alters the brain's processing of the signal, resulting in the creation of complex mental maps. The mapping 'warps' underlying dimensions, altering perception in a way that highlights distinctive categories'' (424). This is supported by the ''perceptual magnet effect'', i.e. the existence of phonetic prototypes. It occurs prior to word learning and ''the change in phonetic perception (...) assists word learning, rather than the reverse'' (427). Kuhl outlines the ''native language magnet'' (NLM) model, which ''argues that early experience establishes a complex perceptual network through which language passes'' (432).
Kuhl then turns to the notion of ''critical/sensitive period'', or ''windows of opportunity'' for learning. The explanation for such a period in speech learning could be explained in NLM not by an alteration of learning mechanisms in themselves, but by the fact that ''the learned structure may interfere with the processing of information that does not conform to the learned pattern'' (433). The same stored representations would account for the reduction of infant phonetic perception as well as for the reduction of infant phonetic production on this interference account. The visuo-motor system and parental hyperarticulation in agreement with phonetic prototypes when addressing infants contribute to the process. Finally, though there is strong evidence of left hemisphere specialization for language in adults, speech does not seem as strongly lateralized in infants of less than four months.
This leads us to the third part of the book, on the ''Evolution of communication'', with an introduction by Hauser. He briefly reviews Darwinian natural selection theory, highlighting the difference between ''homologies'' (features shared by different species through inheritance) and ''homoplasies'' (features shared by different species through convergence). Homoplasies are important because they ''show how and why adaptive design characteristics spring up in quite distantly related species, because for any given problem, the solution set will be limited'' (451). He then introduces the genetic twist to natural selection, according to which ''selection favors gene replication'' (452). This leads to two consequences for communication: first the idea that ''the function of communication is to manipulate receivers for personal fitness gains'' (451), leading to the selection of receivers by their mind-reading abilities; the second is the problem ''of working out the veridicality of the signal'' (451), which is done through the ''handicap principle'' (''signals are honest if and only if they are costly to produce relative to current conditions and if the capacity to carry such costs is heritable'', 451). Hauser then introduces the following chapters.
Hopkins devotes the 16th chapter to ''Signal evolution in electric communication'', i.e. electrosensory modality and electric signals among fishes. Electroreception, rare among aquatic organisms, was widespread among ancestral aquatic vertebrates, was lost among modern fishes and reappeared ''independently in two distantly related groups'' (462). Hopkins describes these two groups, mormyriformes and gymnotiforms. The range of electric signals is limited (one meter or less), though they are not affected by reflection, refraction or reverberation and there is no alteration of the temporal structure. This suggests that the pulse waveform may be significant. There are species, sex and individual differences in electric organ discharge (EOD) and the diversity concerns three features: wave shape, wave duration and wave polarity. Sexually dimorphic traits are often selected through female preference, an expression of the handicap principle. Another effective factor of EOD diversity may be the density of population. The structure of the electric organ in different species, as well as genetic analyses, allows studies of the phylogenetic relationships between the two groups, enlightening the evolution of the electric organ as well as electroreception.
The 17th chapter, by Bass et al., proposes ''Complementary explanations for existing phenotypes in an acoustic communication system''. The goal of biology is giving explanation for the existence of individual phenotypes, which can be done at several levels: functional consequences, evolutionary origins, ontogenetic processes, mechanisms as well as ontogenetic processes and life history. These can be combined, yielding multiple explanations. ''An existing phenotype can be operationally defined by behavioral and structural characters that exist within an ecological environment'' (493). Complementary explanations ''are specified as mechanisms, life history, fitness, and evolutionary history'' (494) with three potential ranges of variation, i.e. behavioral or structural characters and ecological environment. ''Mechanisms either correlate or causally link characters with each other or with an ecological environment and are specified as structural-behavioral, behavioral-ecological, and ecological-structural'' (494). These are repeated along the temporal axis of time of an individual's entire life history, while evolutionary history would be represented a sequence of explanations in terms of past present and future interactions. The authors take as an example the existence of an acoustic communication mechanism in midshipman fish, influencing mate choice by females. They conclude that ''the final mate-preference of any single individual depends on all three mechanisms, and hence variation within the population can arise from variation in one or more characters or mechanisms'' (511).
Gerhardt signs the 18th chapter on ''Reproductive character displacement and other sources of selection on acoustic communication systems''. The focus of the paper is on ''some environmental factors that account, at least in part, for evolutionary change in communication systems'' (515). ''These external factors interact with sexual selection: senders whose signals most effectively propagate in a given environment (...) have an advantage over their sexual competitors'' (515). Gerhardt studies the calls of four related species of Australian frogs, among which call differences occur at very small geographic distances. His hypothesis is that ''after establishment of the isolated populations that gave rise to the four recognized species, their calls diverged gradually by random mutation and drift'' (517), subject to sexual selection. There are two ecological pressures on vocal communication: physical conditions may influence the propagation of the call; predators can detect the callers. These two factors may act as selective mechanisms on the nature of the call. Another constraint is that the call must be recognized as either conspecific or not. This means that frogs living in areas where two different species or more overlap should be more inclined to differences in calls. This is called ''reproductive character displacement''. Its relevance to selection is however in dispute though there are a few examples of it.
The next and 19th chapter, by Ryan and Rand, is still concerned with frogs, dealing with ''Phylogenetic inference and the evolution of communication in tungara frogs''. The tungara frog is a small central American species, with a variable, two-components, call. The wine initiates the call and may be followed by one or more chucks. Calls composed of a single wine are simple while combinations of a wine and chuck(s) are complex. ''Females are responsive to variation in call complexity'' (535). ''Males vary call complexity (...) in response to vocal stimulation from other males'' (538). There is a predation risk represented by a bat which ''respond[s] to call variation in a manner similar to the female'' (538). ''Thus the countervailing forces of sexual selection and bat predation appear to have been responsible for the evolution of the variably complex call of the tungara frog'' (538). In addition to preferring more complex calls, females prefer lower-frequency chucks, frequency being negatively correlated with male body size. The auditory organ of the female matches these behavioral preferences. There are two explanations for these findings: the coevolutionary hypothesis (''signals and receivers coevolve'' (540)); and the sensory exploitation hypothesis (''females had a preexisting preference for chucks that was exploited by males'' (541)). Evidence seems to support the second hypothesis.
Wittenbach and Hoy devote the 20th chapter to ''Categorical perception of behaviorally relevant stimuli by crickets''. There are two ways that animals may differentiate physically similar stimuli and class them in different response categories: the perception may guide the categorization process or the categorization may guide the perception process. The authors are concerned with evidence of the second strategy (called ''categorical perception'') in a Polynesian field cricket, i.e. an invertebrate. These insects have to discriminate between the calls of conspecific and the ecolocation calls of predatory bats. There are four operational criteria for caractegorical perception of sound: distinct categories with sharp boundaries; no discrimination inside a category; a peak in discrimination at category boundaries; close agreement between categorization performance and the predictions made on the basis of absolute categorization. The authors were able to show through a variety of experiments, including the more and more popular dishabituation paradigm, that crickets did indeed use categorical perception. Categorical perception seems widespread among a variety of animals (including humans). Its main benefits are speed and accuracy.
In the 21rst chapter, Searcy and Nowicki describe the ''Functions of song variation in song sparrows''. They address the question at three levels: within individuals, between individuals within populations, and between populations. They make two basic assumptions: ''signal variation must benefit the signaler (...); song variation at some levels may not be functional at all'' (577). The functions of song are to attract females and to defend the singer's territory. This entails the necessity of producing a song recognizable by conspecifics of both sexes. The features common to a species song thus ''define a volume in a 'signal space''' (578). Male singers have several song types which allow for variants, commonality or not of minimal units of production (MUPs) determining the degree of similarity between songs. ''Song sparrows regard song types as being more distinctive than variants'' (581). ''Male song sparrows use switching between song types to signal level of aggressive motivation'' (581) suggesting a first function for song variation. Repertoire size seems correlated to overall fitness, as well as with lifetime and annual reproductive success and years of territory tenure. Within-type variation remains a puzzle. Between-individual variation clearly favors individual recognition and may be advantageous for neighbor recognition among territorial males. Geographic variations seem the product of historical factors than due to any specific functionality.
Hauser writes chapter 22 on ''The evolution of a lopsided brain: asymmetries underlying facial and vocal expressions in primates''. Lateralization is well-known in humans, but understanding its evolution may be done comparatively, through the study of hemispheric asymmetries in nonhuman primates. Thus, Hauser concentrates on hemispheric asymmetries in processing and producing facial expressions and vocalizations in rhesus monkeys. In humans, the left hemisphere is dominant for language, while the right hemisphere is dominant for facial expressions. After reviewing some of the (controversial) literature on asymmetry in nonhuman primates, Hauser describes very elegant fieldwork. The conclusions are that ''it appears that at least some nonhuman primates evidence cerebral asymmetries during communication, either sending a signal or perceiving one'' (618), though infant do not show asymmetries. Hauser also outlines the difference between fieldwork and laboratory experiments, which are complementary: ''Specifically, the field studies have been designed to assess biases in sound localization, whereas the laboratory work assesses asymmetries in response specificity and discrimination'' (621).
Cheney and Seyfarth describe the ''Mechanisms underlying the vocalizations of nonhuman primates'' in the 23rd chapter. They begin by reminding the reader that ''at least some nonhuman primate vocalizations function to designate objects or events in the external world'' (629). In other words, they have a referential function, but this does not mean that they are words in the linguistic sense. For instance, words are judged as similar not only on the basis of their acoustic features but on the basis of meaning. The authors report habituation studies that indicate generalization by meaning in vervet monkeys. There is more however to meaning than this, as words are taken by humans ''not just as signs for things, but also as representation of the speaker's knowledge'' (632) and that, though the case for chimpanzees may be more complex, it seems clear that monkeys do not attribute mental states. Nevertheless, quite a lot of monkey calls have a social function: they ''initiate and facilitate social interactions'' (632). To test whether monkeys can be said to communicate in the Gricean way outlined above, Cheney and Seyfarth examine the use of social calls: are or are they not intended by their producers to influence the receivers' mental states? The conclusion seems to be largely negative: calls ''mirror the intent and state of the signaler, and they fail to take into account their audience's knowledge or ignorance'' (636). This seems true even in nonhuman primates. Thus, ''from the listener's perspective, (...) nonhuman primate vocalizations share many similarities with human semantic signals'' (640). ''From the signaler's perspective, however, there are striking discontinuities between nonhuman primate vocalizations and human language'' (640).
Matsuzawa signs the final chapter of the book on ''Communication and tool use in chimpanzees: cultural and social contexts''. He begins by noting that ''the cognitive functions of humans should be considered in the light of an evolutionary background, and thus as sharing some aspects with living nonhuman primates'' (645). There are three aspects to intelligence, defined as ''a way of modulating behavior to adapt to an everchanging environment'' (645): ''social intelligence, material intelligence and intelligence about the intelligence of other individuals'' (645). The first one has to do with establishment and maintenance of social relationships in the group, the second one with, e.g. tool manufacture and tool use, the last has been called ''theory of mind''. The chapter will be mainly devoted on tool use in specific groups of chimpanzees and on its cultural aspects. ''Communities [of chimpanzees] may differ from each other in a variety of ways, including communication (...), tool use, feeding habits, and so on'' (649). This means that ''chimpanzees must individually learn the unique cultural traditions of their community, thereby maintaining the continuity as well as the distinctiveness of the latter'' (649).
Matsuzawa focuses on tool use in the community of Bossou (Republic of Guinea, W-Africa), reporting field-experiments. He concludes that tool use is flexible, shows individual hand preference, that metatool use (using a tool on another tool to modify to achieve a goal) exists, that there are developmental change in tool use, that observational learning plays a role in the acquisition of skills, that transmission may not be vertical (from mother to offspring), that a rudimentary form of possession may exist. In addition, the field experiments evidenced tactical deception among chimpanzees, rudimentary manufacture of tools and comparison with human children showed that ''the developmental course of nut/stone manipulation and stone tool use was fundamentally similar in the two species'' (662). Finally, geographically close communities of chimpanzees show different tool use traditions, enforcing the notion of cultural transmission. This leads Matsuzawa to the general conclusion that studies of tool use in chimpanzees cannot be restricted to material intelligence but must take into account social intelligence in cultural transmission.
CRITICAL EVALUATION
The question of language evolution has come back to the fore in the past few years, though linguists have not in general been the most numerous protagonists in the ongoing debate. Several of the authors in the present book insist that no evolutionary question can be settled, or even discussed, in ignorance of what exists in comparable species. This may be especially the case for language for which there obviously are no fossils. No matter which way language appeared, whether it evolved or was a by product of other evolutions or the result of a saltationist process, no direct physical evidence will be found for it (though genetics presumably should play a role). The evidence is of an ethological nature and what should be compared are communicative behaviors of different kinds. This is of course all the more true for species known to be near, such as nonhuman primates. This book is a major tool for linguists who, without forgetting that their main aim is to investigate the structure and the cognitive mechanisms of language and language use, think that, though language is unique in the biological world, it nonetheless is a part of it and that its place in it should be properly investigated (see e.g. Hauser, Chomsky and Fitch 2002). In other words, the time may have come for linguists to take a walk on the wild side and this book is a good incentive to... just do it.
REFERENCES
Hauser, M, Chomsky, N. & Fitch, W. T. (2002). The language faculty: what it is, who has it, and how did it evolve? Science 298: 1569-1579.
NB: A good companion to the present book with papers by Pinker, Newmeyer and Bickerton is: Christiansen, M. H. & Kirby, S. (eds) (2003). Language Evolution, Oxford, Oxford University Press.
ABOUT THE REVIEWER:
Anne Reboul is a First Class Research Fellow at the French Center for Scientific Research (CNRS) in France. She has a Ph.D. in Linguistics (EHESS, Paris) and a Ph.D. in philosophy (University of Geneva, Switzerland). She has written some books, among which an Encyclopaedic
Dictionary of Pragmatics and quite a few papers in French and English, on pragmatics and/or philosophic subjects. She is currently involved in a research project on language evolution.
