(In Memoriam) Professor Emeritus
Department of Neurobiology, Physiology and Behavior
Animal communication; neuroethology
Even before I decided to earn a living as a plant ecologist, I had embarked on the study of the behavior of animals, especially birds. My first project was a survey of geographic variation in the song of a European bird, the chaffinch, in Britain , France , and the Azores , conducted in my spare time while I was doing vegetation surveys of potential nature reserves. I found local dialects on a similar geographical scale to those in human speech (4:1952). With a fellowship from the newly founded Nature Conservancy I went on to do a field study of the behavior of the chaffinch for my zoology PhD at the University of Cambridge, England.
The Cambridge Years
Under the mentorship of William Homan Thorpe in Zoology and Robert Hinde at the newly-founded Madingley Ornithological Field Station of Cambridge University, I was plunged into the methods and concepts of the emerging discipline of ethology. One set of studies focused on the ethology of aggression and agonistic behavior of birds kept in aviaries (7, 8:1955; 10, 13:1956; 16:1957; 58:1971; 79:1976; 94:1977; 98:1978), but my main interest was on the vocal behavior of birds.
At Cambridge , Thorpe made it clear that research on song was his special domain, so I decided to focus on the large and interesting call repertoire of the chaffinch. Making use of Thorpe’s newly acquired sound spectrograph, I did what I believe was the first functional analysis of the entire vocal repertoire of an animal, the chaffinch (11:1956). Robert Hinde filled the many aviaries of Madingley with cardueline finches for his studies of their courtship behavior and I was able to record their calls as well. Comparisons of various finch call repertoires provided the first evidence of evolutionary convergence in animal sound signals. In the first of many clashes with authority, which I attribute in retrospect more to youthful exuberance than intellectual acumen, the physical structure of signals proved to be, not arbitrary as Konrad Lorenz had asserted, but directly related to function, in this case to facilitate or hinder localization of the caller in space (6:1955; translated into German by Erwin Stresemann and published in the Journal für Ornithologie, 9:1956). Comparisons of the vocal repertoires of several of the bird species living together in deciduous woodland near Cambridge showed that the selective pressure for specific distinctiveness varies with signal function. I found specific distinctiveness to be high in loud, long range signals that play some part in reproductive isolation, but less extreme in soft, close-range signals, especially in alarm calls of animals that live together and are endangered by the same predators (15:1957; listed as a Citation Classic in Current Contents, 1985). Bird calls have been a major research focus for me ever since, and 50 years later I still find myself urging behavioral neurobiologists not to neglect bird calls in their burgeoning studies of the brain mechanisms underlying avian vocal behavior (269, 270, 271, 272:2004).
As a pre- and postdoctoral fellow Thorpe gave me free rein to explore his 78 rpm library of recorded sounds donated by the BBC. Just before moving to Berkeley , I prepared a review of some of what I had found in the priceless archival material I found there and in my own field studies. I also made an effort to integrate my new findings with the current literature on the structure, function, evolution and development of processes of communication, including olfactory, visual, and auditory signals, with some coverage of both vertebrates and invertebrates (17:1959). With the surge of ethological research in the forties and fifties the field developed rapidly. Looking back, I find that some of the thinking in that 1959 review anticipated ideas I would develop more fully 20 years later. For an international symposium in Berlin on recognition of complex acoustic signals organized by T.H. Bullock I prepared a more analytical account of signal structure in relation to function, and the development of animal communication (88, 89:1977). We expanded on some of the themes later in talking about Sarah Partan’s studies on multimodal communication (254:1999; 266:2002; 277:2005).
The Move to Berkeley
With the move in 1957 from my research fellowship at Jesus College , Cambridge University , England to my first faculty position at Berkeley , University of California , I began searching for tractable subjects for my own laboratory studies of song learning. A series of studies of song variation and development in several American and Mexican birds (18, 19, 20:1960; 22:1961; 27:1962) led me to the white-crowned sparrow, whose song dialects made it an ideal choice for intensive study (29:1962). Building on Thorpe’s chaffinch work, I showed that while the song dialects are indeed learned, and a young sparrow is eager and ready to learn a range of sounds, it nevertheless innately favors songs of its own species when given a choice. Studies of the interplay of learned and innate preferences were the first step in the eventual formulation of the notion of ‘instincts to learn’. I found that learning occurs during a sensitive period, somewhat flexible, but still relatively brief and early in life in this species; acquisition preceded production as in speech development (35:1964; reprinted in 97:1977; 41:1967; 57:1970; reprinted in 61:1972). These findings, cited in many textbooks, provided the basis for several comparisons of avian and human vocal learning (e.g. 55:1970; reprinted in 69:1973, 73:1974; 95:1977).
As I prepared my first course in animal behavior I began writing a much needed textbook, in collaboration with William J. Hamilton, an expert on animal orientation and navigation (39:1966). The book, translated later into German and Hungarian, dealt with mechanistic approaches to animal behavior, with material on both vertebrate and invertebrate animals; it took about 5 years to write. This was a new and major intellectual undertaking for me, but it turned out that my experiences in Cambridge had given me a good start. I benefited from a seminar organized in Jesus College by zoologist Thorpe and psycholinguist Oliver Zangwill. Among the participants were Horace Barlow, Donald Broadbent, John Crook, Richard Gregory, Robert Hinde, and Larry Weiskrantz. We all gave papers, and since I was immersed myself in the ideas of Lorenz and Tinbergen at the time I presented a reinterpretation of the ethological concept of selective responsiveness to external stimulation (sign stimuli; releasers; innate release mechanisms, etc.), emphasizing the multiple peripheral and central levels of sensory processing at which specificity of responsiveness can be imposed (23:1961).
Before the move to Berkeley I had spent time visiting libraries and laboratories in France and Germany , striving to become more familiar with different aspects of behavioral development (e.g. 34:1964), and social communication (e.g. 40:1967). The textbook writing received much help from Berkeley colleagues, especially Howard Bern and Donald Wilson (Zoology), Frank Beach, Gerry McClearn, and Edward Tolman (Psychology), Curt Stern (Genetics), and Sherwood Washburn and Alfred Kroeber (Anthropology). Berkeley was blessed with an outstanding cadre of graduate students at the time, including in my own laboratory, John Eisenberg, Mark Konishi, Fernando Nottebohm, and Thomas Struhsaker. Interactions with Frank Beach and his students gave me some familiarity with behavioral endocrinology, a theme I returned to later, led on by my colleague John Wingfield, in a study of the role of testosterone in song development. (163:1987; 177:1988). I collaborated with him and Carol Whaling on chapters for two editions of the monumental The Physiology of Reproduction (166:1988; 223:1994). Years later, in an invited contribution to the Society for Behavioral Endocrinology I found myself reflecting on the many historical links in between their discipline and the ethology of Lehrman, Lorenz, and Tinbergen (274:2005).
In the early sixties the subject of animal behavior diversified and grew explosively. As I had anticipated in 1966 in the introduction to Mechanisms of Animal Behavior, genetically-based approaches were beginning to open up new vistas for the study of the evolution of social behavior. This revolutionary development was already on the horizon with W.D. Hamilton's work on kin selection, a prospect I highlighted in my invited contribution to a 1969 symposium Donald Kennedy organized on the interface between organismal and population biology (53:1969). The innovations of behavioral ecology soon followed, and blossomed into a new sub-discipline. In the meantime, however the subject of animal communication remained almost terra incognita.
In 1964 a group of primatologists convened at the Center for Advanced Studies in the Behavioral Sciences at Stanford to bring together in book form all the new field data emerging on the behavior of non-human primates. They invited me, as the lone non-primatologist, to review social communication (36:1965). This opened up a new phase in my career. It led to fieldwork in Africa during a sabbatical year at Makerere University College in Uganda , in which I documented for the first time the nature and function under field conditions of the vocal repertoires of several forest-dwelling primate species (48:1969; 54:1970; 62:1972; 67:1973). These preliminary studies laid the background for future in depth studies by John Oates, Tom Struhsaker, and others. As I had found previously in birds (15:1957), alarm calls of cohabiting forest monkeys vulnerable to similar predators showed less species divergence than vocal signals concerned with territoriality and rallying of the social group (43:1968; 67:1973), a step forward in characterizing the relationship between signal structure and function. Working with Jane Goodall at Gombe National Park in Tanzania , I conducted the first sound spectrographic analysis of the vocal repertoire of wild chimpanzees (47, 49:1969; 56:1970; 86:1977). We prepared a sound movie on the vocalizations of chimpanzees, filmed by Hugo van Lawick. A quantitative analysis of the long-range vocal signal of pant-hooting in identified chimpanzees supported Goodall’s impression of consistent individual differences, sufficient to support both neighbor-stranger and individual recognition (77:1975). Comparing the chimpanzee data with other higher primates, especially the gorilla, some new ideas emerged on the nature and usage of different types of vocalizations in relation to social organization (81:1976; 86:1977). Of special interest was the frequent use by more sociable primates of highly graded sound signals, particularly at close range. The tendency to use graded signals, first noted by Hinde and Rowell in macaques, was especially marked in species like the red colobus. The use of highly variable signals invited comparison with the inter-graded sounds of speech and the importance of categorical perception in responding to them, a process discovered by Alvin Liberman (75:1975; 81:1976; 102:1979). A collaboration with William Stebbins at Michigan , a leading student of primate audition and his students, led to the discovery of categorical processing of graded coo calls used by Japanese macaques, previously documented in detail in Stephen Green's thesis research (105:1979). The Japanese macaque project also provided a remarkable case of lateralized cerebral processing, with right-ear, left hemisphere dominance, manifest with conspecific signals, but not with sounds of other species (141:1984; see also 118:1981; 130:1983). Some years later, Nelson and I demonstrated categorical perception in birds processing the sounds of their own song (179:1989; c.f. also 122:1982).
Song Learning at Rockefeller
After 9 years in the Berkeley faculty I had made a move to The Rockefeller University in New York . Again I had to find new subjects for song learning studies (63:1972; 78:1975). The first candidates were canaries bred for their song that we imported from Belgium . They were subjects in a range of studies. The songs of canaries raised in masking noise and then deafened were acoustically very primitive, but gradually improved when the noise was terminated, and there was clear evidence of social influences on song learning (70:1973; 83, 84:1977). These birds became pivotal in Nottebohm’s research on the neurobiology of song learning. I had invited him to move from Berkeley to Rockefeller and the New York Zoological Society as an assistant professor with me, along with primatologist Thomas Struhsaker. Intensive birdsong studies with many colleagues followed over the years, including Susan Peters, Don Kroodsma, Bill Searcy, Steve Nowicki, Ken Yasukawa, and John Wingfield, all conducted at the 1000 acre Field Station set up by the Rockefeller University for Donald Griffin, Fernando Nottebohm, and me in Upstate New York. Among many projects there, we raised nestlings of two local species, one with a complex song (the song sparrow), the other with a simpler one (the swamp sparrow). Brought into the same laboratory environment, both learned from tape recordings, but the way in which they did so was very different. It became clear that while there is a single, basic strategy underlying song learning, the details vary greatly from species to species. The underlying neural and behavioral mechanisms have a central core, but the details are in some degree both genetically labile on an evolutionary scale, and developmentally labile, with each species evolving its own optimal solution, hence the notion of species-specific ‘instincts to learn’.
We found an age-related sequence of memorization, storage, retrieval from memory, and finally vocal production, and this sequence became the model for future work (121:1982). There were sensitive periods in the first phase, when the readiness to memorize new songs peaked. The timing was different in the two species we studied (164:1987; 168:1988) and the potential for variation was reconfirmed later in a study in a third species, the white-crowned sparrow in California in two populations inhabiting different environments (231:1995). The sensitive period was curtailed in birds living in a more severe montane environment, compared with the more benign conditions on the coast. A controversy developed about the validity of song learning studies using tape recordings. To explore this we compared tutoring with tape recordings and live singing birds. A live tutor was much more potent but with regard to timing the results were very similar (168; see also 225:1994). The evidence of innate learning preferences for conspecific song was clear, even in birds raised from the egg in the laboratory. Presented with the same tutor tapes of mixed species songs, each naive sparrow selected songs of its own species to memorize and produce. We made use of artificial tutor songs in which phonological and syntactic features were independently varied to show that although they are close genetic relatives, song and swamp sparrows use different sets of physical features as the basis for their learning preferences. The physiological basis for this discriminative ability, playing a central role in ‘the instinct to learn’ of each species, still remains largely unexplored (but see 169:1988; 180:1989).
In a major undertaking Susan Peters and I traced the production side of the imitative process in detail from its earliest manifestations in subsong, which is amorphous and highly variable, through plastic song, when the first hints of crude imitations appear, to the stabilized patterns of crystallized song (125, 128:1982). We were fascinated to find that overproduction of song types occurs in plastic song, with the excess winnowed down to a small subset by a socially guided process of selection at the time of song crystallization (124:1982). If the singer is counter-singing with a territorial rival, the selection process may lead to a choice of themes that approximate a match with the rival’s songs – a kind of learning process. The notion of selective processes in song learning – a departure from customary instruction-based models - have since been found in late stages of vocal development in several sparrows (216:1993; 226:1994). They exemplify the potentially strong influence of social factors on song development.
Song learning studies have consistently evoked interest among students of speech acquisition in human infants (131:1983; 139:1984; 143:1985; 178:1989; 186:1990; 192, 195:1991), and there are notable parallels with selective responsiveness to speech sound categories in infancy, and sensitive periods for the development of speech and signing. An overview called “The instinct to learn” was reprinted in volumes on language acquisition and cognitive development (219, 220:1993) and on biology and cognition (193:1991).
The interpretation of learning preferences is often confounded by overlapping constraints on perception (input) and production (output) and this is a problem with many song learning studies. Nelson and I developed a song recognition assay for very young birds before they have begun to sing, based on call responses triggered when they hear song. Again young birds innately favored own-species over other-species songs. We found that pre-production song memorization had strong effects later on song responsiveness in both males and females (215:1993; 234:1995; 244, 246:1997). The growing list of acoustic features of own-species song to which naive sparrows are responsive (e.g. 261:2000) proves to be a long one, leading to a renewed interest in neuroselective concepts of song learning. In one model combining instruction and selection, innate foreknowledge about some of the natural constituents of species-specific song is assumed to be pre-encoded in brain circuitry that needs to be activated by auditory experience before it influences what the bird sings. The activation process is the basis for instruction about the particular arrangement, sub-typing and sequencing of elements in the tutor’s song, matching an individual’s personal experience (139:1984; 210:1992; 243:1997). The model has excited interest among theoreticians and neurobiologists, as well as linguists inclined to favor a significant role for innate knowledge in language development (see commentary in 230:1995).
In studies of the meaning of animal signals and the relationship between animal communication and human language semiotic theory proved to have some potential (11:1956) as a corrective to over-rich introspection-based interpretations of signal meaning, as applicable to animals as to humans (25:1961; see also 203:1992). There is a widespread belief that animal signals are based on affect, and that only humans possess symbolic signals. One objective distinction between affective and symbolic signals applicable to animals is referential specificity; assuming that the referents for affective signals, if they exist, are relatively generalized and unspecific. But some animal signals, especially some alarm calls, have a high degree of referential specificity sufficient for them to qualify as symbolic (90:1977; 99:1978).
Thomas Struhsaker discovered predator-specific alarm calls in the vervet monkey in Africa while he was a student in Berkeley . Later, at Rockefeller, I invited postdocs Dorothy Cheney and Robert Seyfarth to conduct playbacks of tape-recorded calls in the field. Whether the predator was a snake, an eagle, or a leopard, the distinctive alarm calls elicited specific responses, functionally appropriate for each one (109, 110:1980). The referential specificity of these calls, greater than any previously described, blurred one of the distinctions between affective and symbolic signaling. Many cases of referentially specific alarm and food calls have since been described (e.g. 161:1987; 212, 214:1993), providing illustrations from birds and primates of ‘honest’ communication about food quality and quantity. It is interesting from a cognitive point of view that they are sometimes used deceptively (150:1986; 167:1988; 224:1994; 235:1996) and Hauser discovered that monkeys were punished when deception was detected by companions (217, 218:1993).
Searching for insights into the question of intentions to communicate, we explored effects of having someone to communicate with on signal production in birds. We found that in the presence of a predator or food, calling is not completely impulsive, but is socially controllable, depending on the caller’s circumstances and the intended recipient (189:1990). For example a bird that sees a hawk above is more likely to alarm call if a companion is present than when it is alone. This is true even though the caller’s other behavioral responses may not be distinguishable from those of a solitary individual (154:1986; 170:1988). With food calls, a solitary bird that discovers a preferred food calls more than to a less-favored food, but in both cases calls less when alone than with companions (151:1986; 224:1994). Some calls proved to be immune to audience effects, such as mobbing calls; these are given by birds to ground predators, but are addressed more to the predator than to companions (235:1996). So audience effects do indeed provide one operational window on the intention to communicate.
The Instinct to Learn
I originally thought that culturally transmitted song dialects must be virtually free of genetic constraints, but it has gradually become clear that each bird approaches song development with its own set of innate learning predispositions. Birds raised in identical environments reacted very differently, displaying contrasts that must ultimately be genetically based. Some songbirds develop small song repertoires, some large ones. Some learn avidly, and speedily, with precision, in response to minimal stimulation (206:1992); others are reluctant to learn and more resistant to environmentally induced change. Even the most eager mimics display learning preferences when given a choice, often favoring own-species songs, or own-species imitations of others. Some remain faithful to their tutors but others improvise and invent copiously, as we found in the red-winged blackbird (63:1972). In some birds females learn to sing as well as males, and there are always individual differences in accent, complexity, and repertoire size.
While differences in experiential history always have a strong influence on behavioral development, genetic differences are also crucial (252, 253:1999), hence the idea of ‘instincts to learn’ (156:1987). The conclusion that the genome is as important as experience is reinforced by comparisons of human and animal communication. As Chomsky has long argued (230:1995), key aspects of language are uniquely human. No animal speaks naturally in meaningful sentences, and even language-trained chimps are only minimally competent (255:1999). ‘Phonological syntax’ (diverse recombinations of elements with essentially the same meaning) is widespread in animals, especially in birdsong. However natural cases of ‘lexical syntax’ (sentence construction) have yet to be described in animals (247:1998; 250:1999), although there are primordial, as in John Mitani’s gibbon calls (e.g. 181:1989). The distinctively human suite of abilities required for language includes the capacity to learn new vocalizations, to link them to specific meanings, and to assemble them into meaningful sentences: birdsong only qualifies for the first of these (82:1976; 247:1998). Some calls of birds and mammals are possible candidates for the second (270, 272:2004), but the third seems to be uniquely human. Language aside, it is likely that comparative studies of the vocal behavior of animals will shed more light on the origins of music (247:1998; 259:2000; 263:2001).
Throughout the history of behavioral science issues of nature and nurture have been the subject of endless debate. Many students of behavior have focused exclusively on experience as the dominant factor in shaping behavioral ontogeny (257:1999). In recent years research on birdsong has provided something of a counterbalance by acknowledging the role of nature as well as nurture, but further progress in understanding mechanisms of learning awaits application of the methods and concepts of molecular and developmental genetics to behavior. A new era is dawning with the first complete genomic mapping of a songbird soon to become a reality, but the challenge for the next generation will be considerable. Researchers will need to be properly reductionistic as geneticists, and yet broad minded enough as students of behavioral mechanisms to grapple with the larger questions of nature and nurture and the evolution of behavior. But with the appropriate blend of holistic and reductionistic approaches researchers can for the first time deal properly with those questions of nature and nurture that have so long bedeviled behavioral science.
- 1948 BS University College, University of London
- 1952 PhD (Botany) University College, University of London
- 1954 PhD (Zoology) University of Cambridge
Palleroni, A., M. Hauser & P. Marler. 2005. Do responses of galliform birds vary adaptively with predator size? Animal Cognition 8:200-210.
Partan, S.R. & P. Marler. 2005. Issues in the classification of multimodal communication signals. American Naturalist 166: 231-245.
Palleroni, A., C.T. Miller, M. Hauser, & P. Marler. 2005. Prey plumage adaptation against falcon attack. Nature 434: 973-974.
Nelson, D.A. & P. Marler. 2005. Do bird nestmates learn the same songs? Animal Behaviour 69: 1007-1010.
Marler, P. 2005. Ethology and the origins of behavioral endocrinology. Hormones and Behavior 47: 493-502.
Marler, P. 2004. Science and birdsong: The good old days. In: Nature’s Music: The Science of Birdsong, P. Marler & H. Slabbekoorn (eds.). Elsevier Academic Press, San Diego , CA , pp. 1-38.
Marler P. 1999. How much does a human environment humanize a chimp. American Anthropologist. 101:432-436
Marler P. 1999. Origins of music and speech: Insights from animals. The Origins of Music. :Cambridge: MIT Press
Marler P and DF Sherry. 1999. The nature and nurture of developmental plasticity. Proceedings of the 22nd International Ornithological Congress. Durban South Africa: University of Natal Press:in press