Observation is the heart of any study of an animal society. We begin with description; and until we have a general picture of social behavior and organization, we have no firm foundation for experimental work (Scott, 1955, 1956). Such a descriptive study inevitably leads to the conclusion that the members of an animal society communicate with each other. Crows give cries that communicate alarm or the presence of enemies (Frings and Frings, 1957), and bees are able to inform each other regarding the location of food sources (Von Frisch, 1950).
Even when animals have no special means of communication, they respond to each others behavior. In the most general sense, communication includes any stimulus arising from one animal and eliciting a response in another. Since social behavior can be defined as behavior which is either stimulated by or has an effect on another member of the same species, this implies that communication is an essential part of social organization and, indeed, that every animal society must include some form or forms of communication.
KINDS OF COMMUNICATION
The most convenient way of classifying modes of communication is by the sense organs through which information is received. This is somewhat artificial, since in most cases an animal responds to a combination of stimuli rather than to stimuli received by one sense organ alone. A porpoise, when it is out of touch and sight of others, may respond only to auditory stimuli, but, when at close quarters, may react to tactile and visual stimuli as well. This is even more obvious in the social interactions of birds, where calls are often combined with visual displays of feathers or unusual postures. Nevertheless, most animals tend to emphasize one avenue of communication more strongly than others, and some examples and general characteristics of each are described below.
This occurs in at least three general forms. One of these is an active movement such as the pecking of one hen directed toward another, or one baboon running toward another, the whole being known as a behavior pattern. Second, movements may result in fixed postures. The assumption of a posture appropriate to a female about to engage in coitus is a common signal in primates, where it is known as “presenting.” In birds, postures frequently reveal certain markings and are known as displays. In the red-winged blackbird, the threat display results in the exposure of the colored wing bars. Movement may also result in a more or less fixed facial expression, such as the snarling of wolves and dogs, and the different facial expressions of primates.
This is a very widespread form of communication among air-breathing animals and is also found in the water-living vertebrates. Many fishes make auditory signals, and porpoises use high-pitched noises as a form of sonar for locating underwater objects and obstacles as well as for communication between individuals. Auditory stimuli are generated in a large variety of ways, the vocal cords of mammals being only one of many. Insects produce noises through the vibration of their wings, and more specific signals are produced by the stridulating organs of crickets and other Orthoptera. There is a wide range of frequency of vibratory stimuli, from very low-pitched sounds to the high-frequency noises produced by bats and porpoises, which go as high as sixty to eighty kilohertz and are completely inaudible to the human ear.
The distribution of chemical stimuli in either air or water is achieved through diffusion and currents. Consequently, these stimuli are much less reliable than either visual or auditory stimuli in conveying information, and their range is much more restricted. They are correspondingly difficult to detect by observation, as the human observer must usually remain at a distance from the animals he is watching and cannot detect the same chemical stimuli, even if his sense organs are equal to the task. Nevertheless, such stimuli perform important communicative functions, from the well-known examples of sexual attractants in insects to the distribution of odorous urine and feces by mammals.
These are much more restricted in their use than any of the above, but they are highly important in many forms of social behavior, especially sexual behavior and the care of the young in mammals. Even in animals having an external skeleton, such as the insects, tactile stimulation may be commonly used. Stroking movements of the antennae are a prominent part of sexual behavior in blister beetles (Selander, 1966).
Relationship of Communication and Ecology
The kinds of stimuli which are important in communication depend a great deal upon the habitat and mode of life of the particular species. Communication has been studied extensively in birds. These animals are primarily adapted for aerial existence, and the great majority of them are diurnal in their habits. In such rapidly moving animals olfactory stimuli have little importance, whereas visual stimuli in the form of movements and displays are very prominent. Auditory stimuli are almost equally important, and the vocal powers of birds are at least as well developed as those of mammals. Since birds are covered with feathers whose function depends upon their precise arrangement, tactile stimuli have relatively little importance.
Mammals, while primarily terrestrial animals, have evolved forms like whales that live an almost fish-like existence in the ocean, and others like bats which lead an aerial existence similar to that of birds. Mammals include both nocturnal and diurnal forms, and the nocturnal land-living species make much use of olfaction in their communication with each other. Nocturnal and diurnal animals can occur in related forms. Among primates, the higher apes and monkeys are diurnal and make much use of visual signals, and the same group includes the completely nocturnal tarsiers. In contrast to birds, tactile communication through the hair and vibrissae is important in many species.
Communication in reptiles has been little studied except in lizards. Many of these animals are diurnal and terrestrial, and they exhibit visual displays of form and color which are much like those of birds. Amphibia, on the other hand, are chiefly nocturnal in their habits and make much use of vocal communication.
Visual communication may be very important in certain fishes, espedaily those living in shallow clear water; however, other species living at greater depths or in turbid water must depend on other modes of communication, the most efficient of which are auditory or vibratory stimuli. Shark repellents developed from shark tissues indicate that chemical signals may also be received. Many species of ocean fish make noises, but most fresh-water fish seem to be quiet, the principal exceptions belonging to the carp family (Moulton, 1963). Auditory communication would be useless in a noisy mountain stream, and it is possible that vocalization would too easily attract the attention of predators in clear shallow water.
Among arthropods visual signals are reduced in importance, not because of the habits of the animals, but because the compound eye is less efficient than the vertebrate one. Much of the behavior and coloration of insects that results in visual stimulation has been evolved, not in relation to social communication, but as a protection against prédation by vertebrates. In nocturnal insects like moths, odorous signals are important, and in the grasshoppers and crickets auditory signals are highly developed.
Modes of communication are, therefore, related not only to ecology but to the basic morphology and physiology of the species under study. This brings up the problem of the limitations of human sense organs. The human senses may detect a great deal less than an animal’s, as is the case with the supersonic cries produced by many mammals and the kinds of odors which can be detected by many others. On the other hand, the human eye may detect more visual detail than is apparent to many insects, with their fixed-focus, compound eyes, and more color detail than the many mammals whose eyes detect only black and white stimuli. As a result, the observational study of animal communication is full of problems and pitfalls.
EVOLUTION OF SIGNALING SYSTEMS
As pointed out above, communication in the broadest sense may result from any stimulus produced by another animal. In addition, many animals have evolved means of producing stimuli which have the major function of signaling. These may include many of the above kinds of stimuli. Some of the most interesting cases are those in which behavior patterns have changed their function from some other adaptive function to that of signaling. Tinbergen (1964) in his review of the evolution of signaling behavior cites the example of self-preening in ducks which has, through evolutionary changes, become part of the courtship behavior of males. In addition, certain movements which originally had a direct function have been converted into “ritual” movements which have only a communicative one. For example, the “claw-waving” of fiddler crabs probably originated as an actual attack with the big claw.
We may, therefore, make a distinction between behavior which is communicative in a general sense, comprehending almost any activity, and that which has a specialized signaling function, such as bird song. These specialized signals often become very conspicuous, even to a human observer, indicating that they have an important adaptive function in the life of the species.
PROBLEMS OF RECORDING
The basic problem here is one of translating other forms of communication into human verbal symbols. This is often quite difficult, and in many cases recording is better done with devices which reproduce the original signals and make a permanent record of them without translation.
One of the best methods for general preliminary observation of animal behavior is to write a verbal description of the behavior as soon as possible after it is observed. In some cases the behavior takes place so rapidly that writing is impossible, and the verbal description can be taken down only with a tape recorder for later transcription. Both methods have the disadvantage of producing an enormous bulk of data in a form which is difficult to analyze. It is to be hoped that eventually computer methods will be devised for handling verbal descriptive data.
Once preliminary observations have been made, moving pictures can be taken of the most important behavioral sequences. Here again, long periods of recording can lead to an enormous bulk of material. The moving picture has two advantages. One is that the same scene can be played over and over again. In any real situation, behavior is never repeated exactly, and one’s attention always tends to be drawn to rare and novel aspects of the behavior observed, which are often the least important because they are so seldom repeated. Since the same scene can be replayed, it is possible to pick out details of behavior which ordinarily would escape the eye. The other advantage is that a moving picture can be used to communicate findings to an audience in a form which resembles the original behavior more closely than verbal descriptions.
The moving picture has certain limitations, one of which is that it lacks the perspective of a real scene. Another is that the focus is fixed, and it is not possible to switch from far to near focus at will once the picture has been made. Still another drawback is that the individual frames are usually blurred and conceal detail rather than reveal it. On the other hand, the moving picture can be used to slow down or speed up action and thus reveal details which are not visible to the unaided eye. For example, Banks (1962) has used the motionpicture technique and frame-by-frame analysis to demonstrate that in the fighting of mice there are no reliable visual signals preceding attacks.
Another photographic technique is that of the still camera. Still photographs are excellent for recording postures and facial expressions, provided they can be taken at exactly the right moment, since in many cases the posture is held for only a short time. Behavior patterns come out badly in photographs, as a good photograph has the effect of stopping motion and thus making the animal appear stuffed. It is often easier to convey a sense of motion with a drawing or series of drawings. The drawing also has the advantage of emphasizing the important aspects of the signaling movement, and many workers use simple outline drawings.
One of the least satisfactory methods of recording auditory signals is an attempt to render these by human phonetic symbols. The usual result is something that can be recognized only by a person who has actually heard the original sound. Musical notation also has limited usefulness, since most animal sounds are much too complex to be rendered as musical notes, often depart from any known musical scale, and frequently are simply noises.
The best solution of the problem is the tape recorder, which enables the sounds to be played back again and again for study. Significant portions of the tape can be selected for analysis on the sonagraph, which gives a general visual picture of the sound, with frequency graphed against time. Loudness is to some extent rendered by variation in the darkness of the graph, but it can be measured only on a relative basis, since the loudness of the recording depends on the closeness of the microphone to the subject and the setting of the recorder.
Tape recordings have limitations, particularly with respect to highfrequency sounds. The ability of a recorder to duplicate high-pitched sounds depends directly on tape speed, and to record the high-frequency sounds made by bats and porpoises an enormous amount of tape must be used. Sonographs of low-pitched noises indicate that even the sounds which are in the human auditory range have overtones which extend upward indefinitely and which are cut off by the ordinary tape recorder. Sonagraphs have been particularly useful with bird and insect sounds, where they can be used to study individual and species differences in various sorts of stereotyped calls and songs. Although widely used to analyze human speech defects, they have not yet been used extensively with the sounds produced by other mammals.
The “voice print” type of sonographic analysis renders the sounds in a topographical fashion rather than as a graph and has proved useful for identifying individual differences in human vocalization. It has not yet been used extensively with animal sounds but may have some possibilities.
As stated above, these signals are difficult to detect by straight observation, because ordinarily the observer must be far enough away from the animal that he cannot smell or taste the same things that the animal is experiencing, nor in most cases would he wish to do so. Usually, it is only possible to infer what is going on. When an animal noses another, he might be sniffing, although he could also be poking or feeling the other animal; and if he licks the other animal, presumably he is tasting, although this might be simply a case of grooming. Where such signals are suspected, the only way to test them is by simple experiments conducted as a part of observation.
The same sorts of difficulties are met in describing behavior involving touching another animal. The most that an observer can do is to record the behavior as accurately as possible and not come to firm conclusions without additional evidence. For example, a mouse touching another with its vibrissae is presumably receiving and delivering tactile stimuli, but one cannot be sure of this until the vibrissae have been removed and a change noted in the behavior of both the animals. Similarly, one mouse may appear to be grooming another’s fur, but actually he may be pulling it out.
PROBLEMS OF VERIFICATION
Many observers are quick to jump at a possible adaptive function as an explanation of an observed behavior pattern. For example, in a herd of American bison or buffalo that is grazing near a wooded area, one often sees an animal butting and thrashing at a small tree with his head. The observer might say, since the tree is marked and scarred, that the animal is marking a territorial boundary. There are, however, several other possible hypotheses concerning the adaptive function of this behavior. One is that the buffalo is attacking the tree, and that its behavior is a mock fight. Still another is simply that the buffalo’s head itches, and the behavior relieves the itching.
The first question to ask with respect to a possible signaling function is whether the behavior of other animals is affected by the supposed signal, in this case the marked tree. This is something that can be tested by observation, and repeated observations will either verify or deny the hypothesis. In fact, other buffalo do not seem to notice the marked tree, and the observational evidence is thus against communication as an explanatory function of the buffalo’s behavior.
Another example is the tail-rattling of male mice. In certain circumstances these animals rapidly vibrate their tails back and forth so that a rattling noise is produced when they touch the walls of a cage. This behavior occurs before one animal starts a fight with another, and it might be inferred that its function is that of a threat. However, observational evidence shows no obvious reaction to this noise by other mice, and it is possible that it represents only a release of internal tension. It may be that the mice are responding with internal emotional arousal, but verification of this hypothesis would depend on experiments using methods for recording physiological changes.
Certain experimental methods of testing the existence of a signaling function of behavior have been well worked out. In studying visual signals, models can be set up, their form and movements modified in various ways, and the effects observed upon animals even in the field. This technique has been used chiefly by ethologists to describe the nature of effective stimuli in connection with the releaser theory (Tinbergen, 1951), but it also has application to the study of communication.
In the same way tape recordings of auditory signals can be played back to animals, either in the field or laboratory, and the effects noted. Recently, Falls and Brooks (1965) have done this with bird song, and their results confirm the warning function of these sounds. The problems are somewhat different in mammals, and I shall illustrate these with some experiments on vocal communication in the dog.
Barking in Dogs
Pet dogs obviously use barking in a variety of ways, such as attracting the attention of their masters and signaling their desires to have doors opened. From Murie’s (1944) descriptions of the barking of wolves, this occurs chiefly when some predator or strange animal approaches the den. We put forward the hypothesis that the major and primitive function of barking is an alarm signal. This was confirmed by observations on dogs reared in groups in large fields apart from people. In these animals barking appeared whenever a strange person was sighted by them.
Barking has certain acoustic properties which are apparent to the human ear. One is that these sounds are quite loud and carry for long distances, and the other is that they are easily localized. One can listen to a barking dog and pinpoint his location within a degree or two, even when the dog is not visible, or one can follow the dog’s progress as he runs down a lane behind a hedgerow. Sonagraphs of barking (Figs. 1 to 3) reveal that for any individual dog (except those in the hound breeds), barking is usually a series of short, sharp sounds of uniform loudness and pitch repeated with very short intervals between them. These properties make it very easy to discriminate between sounds given from slightly different directions, and indeed, dog barking has acoustic properties similar to the sounds used by bats in echolocation.
We were interested by Scheins (1963) report that young turkeys give a precise behavioral response to the alarm cries of the mother or to any similar noise, and we wondered whether puppies would give a similar reaction. I suggested to Dr. Dale Morris that he design an experiment to test this hypothesis, and he prepared two tapes of dog sounds in reverse order. These included barks by the puppies’ mother and by a strange dog. As controls we included two other sounds, a pure musical tone, which would be very unlike a bark, and another sound which had the same general acoustic properties as barking, that of a board repeatedly struck on a table. On one day each week we played the tapes to the puppies, from the time they were four until they were thirteen wiks of age and observed their responses to each noise. The tapes were played four times, once with the mother present and once with the mother absent from the nursery room, and at two levels of loudness. The puppies became more responsive as they grew older, making almost twice as many responses in the second half of the experiment as in the first. The responses were classified into those which appeared to be fearful and those which resembled investigative behavior. Only about one-sixth of the responses fell into the fearful category, the usual response being one of alerting and searching. The mothers presence or absence made no difference, nor did the two levels of loudness.
The responses to the different stimuli are summarized in Table 1. The distribution departs significantly from that which might be expected by pure chance (p = 0.01). However, the numbers of responses made to the bark of the mother and to that of a stranger are very close, and the average of these is not significantly different from that to the sound of a board hitting a table. If anything, more responses were given to this artificial sound than to sounds made by other dogs. The greatest number of responses was produced by another artificial sound, the pure tone. This is the one sort of taped noise not commonly heard by the puppies. Indeed, the number of reactions appears to vary with the unfamiliarity of the signal.
This result can be interpreted in the following way. In the dog colony the puppies often heard barks of all sorts and were habituated to them. This habituation would extend less completely to sounds which merely resembled barks, so that only a completely unusual sound would produce a clear-cut increase in the number of responses. It seems likely that in the puppy there is no specific behavioral response to a particular alarm signal; rather, there is a general alerting and startle response to any unusual loud noise.
From observations of mothers and their puppies it appeared that mothers were more responsive to the cries of their offspring than the reverse. Distress vocalization in young puppies can be elicited easily and immediately by placing a puppy alone in a strange place. We made recordings of these sounds and analyzed them on the sonograph. In contrast to barking, the distress vocalizations made by any one puppy were highly variable in form and pitch, being a mixture of yelps, howls, and barks in no particular order. At first, we thought that these sounds might give a sort of ventriloquial effect, such as that produced by many bird songs, where variations in pitch and loudness make it difficult to locate the bird. With the help of Dr. Joseph Church, we played recordings of these vocalizations to human subjects in the open air. Any loudspeaker has a directional effect, and we overcame this by pointing the speaker vertically. However, human subjects were able to locate puppy sounds with great accuracy, provided they were loud enough to be heard distinctly, and we abandoned this hypothesis.
The variable nature of these sounds, nevertheless represents an interesting problem. Another hypothesis is that if distress vocalization was monotonous and invariable, the mother dog would habituate to it rapidly and pay no attention to the sound after a few repetitions. The continual changing of the pitch and form of the sound makes habituation difficult, at least to the human listener. This hypothesis could be tested by taking one sound by itself and playing it over and over again and then comparing the result with that obtained by playing a normal series of variable distress vocalizations. This experiment has not yet been done.
We did, however, play the tapes back to the mothers. If the puppies had been removed from the home pen, the mothers responded very strongly, attempting to escape and move toward the source of the sound. On the other hand, if the tapes were played to the mothers while their puppies were with them, they merely gave one glance at the puppies and paid no further attention. Thus, there appears to be no built-in response to the sound of distress vocalization other than the one of alerting the mothers to pay momentary attention to their puppies.
In their communication systems, dogs thus seem to have many of the capacities exhibited by human beings in their use of language. Dogs can make a variety of noises, and they can learn to use these instrumentally to obtain certain goals. They do not have responses that are automatically stimulated by certain sounds under all circumstances; rather, they respond to the vocal signals of other animals in terms of the general situation in which they are given. If the signals convey no information, dogs pay little attention to them. In contrast to people, dogs lack the capacity to use sounds as abstract symbols and, for the most part, lack the capacity to make sounds except when emotionally stimulated.
SOME UNSOLVED PROBLEMS
I have mentioned above the difficulties connected with the study of olfactory and tactile communication. For this reason, most studies of animal communication have been confined to visual and auditory signals, with the result that much of the communication of mammals has been entirely neglected. There are signs that this situation is improving. I understand that it is now technically possible to design electronic devices which will respond to specific gaseous molecules in extremely small quantities, so that it may be possible for human observers in the future to detect odors which are beyond the human range and to obtain records of odors at a distance and over considerable periods of time (Milleron, 1964).
Furthermore, experiments with pregnancy block and similar reproductive phenomena with mice indicate that certain odors act as pheromones, i.e., signals having a direct physiological effect which later modifies behavior (Bronson and Marsden, 1964). The experiments of Godfrey (1958) using Y-shaped tubes through which a current of air can be passed demonstrate that the odors produced by different species of voles have significance in species recognition.
Another problem is that of the hasty assumptions of observers concerning adaptive function, so often made without either consideration of alternative hypotheses or attempts at verification. As I have indicated, it is possible in many cases to make repeated observations as natural conditions vary and thus, in a preliminary manner, to test alternative hypotheses. The game of assigning adaptive significance to either structure or behavior is one of the rare areas still open to the free imaginations of scientists. However, in the study of communication we must have verification, and experimental tests of adaptive function are possible even in the field.
Finally, observation needs to be accorded its proper and very important place in the application of the scientific method. The study of communication in any animal species should begin with observation of behavior under natural or seminatural conditions in order to get some idea of the significance and importance of communication and also to assess the general range of capacities of the species. Furthermore, observation generates ideas in a way that experiments under highly controlled conditions do not. The function of experimentation is to test and refine theories, but major theories themselves originate from careful observation and systematic classification of the data so assembled.
Banks, E. M., 1962. A time and motion study of prefighting behavior in mice. J. Genet. Psychol., 101:165-183.
Bronson, F. H., and H. M. Marsden, 1964. Male-induced synchrony of estrus in deermice. Gen. Comp. Endocrinol, 4:634-637.
Falls, J. B., and R. J. Brooks, 1965. Studies of intraspecific individual recognition by song in birds. Am. Zoologist, 5:225 (abstract).
Frings, H., and M. Frings, 1957. Recorded calls of the eastern crow as attractants and repellants. J. Wild. Mgmt., 21:91.
Godfrey, J., 1958. The origin of sexual isolation between bank voles. Proc. Roy. Phys. Soc. Edinburgh, 27:47-55.
Milleron, N., 1964. Private communication.
Moulton, J. M., 1963. Acoustic behavior of fishes. In: R.-G. Busnel (Ed.), Acoustic Behavior of Animah. Amsterdam, Elsevier.
Murie, A., 1944. The Wolves of Mount McKinley. U.S.D.I. Fauna Series, No. 5. Washington, Government Printing Office.
Schein, M., 1963. Private communication.
Scott, J. P., 1955. The place of observation in biological and psychological seience. Am. Psychologist, 10:61-64.
——, 1956. The analysis of social organization in animals. Ecology, 37:213-221.
Seiander, R., 1966. Patterns of sexual behavior in pyrotine blister beetles (Meloidae). Am. Zoologist, 6:322.
Tinbergen, N., 1951. The Study of Instinct. London, Oxford Univ. Press.
——, 1964. The evolution of signalling devices. In: W. Etkin (Ed.), Social Behavior and Organization among Vertebrates. Chicago, Univ. Chicago Press.
Von Frisch, K., 1950. Bees, Their Vision, Chemical Senses and Language. Ithaca, N.Y., Cornell Univ. Press.