We have found that, in dealing with such a largebrained mammal, we must keep the working hypothesis in mind that "they are highly intelligent and are just as interested in communicating with us as we are with them." ... If we use any other hypothesis, we have no success whatsoever in dealing communicatively with them. - JCL

Dolphins' Complex Communication

In 1961 we set up a "dolphin telephone" between two tanks. These tanks were conically insulated and isolated from one another. A telephone was arranged electronically to be two-way, i.e., the dolphin in tank A could talk to the dolphin in tank B and, simultaneously, the dolphin in tank B could talk to the dolphin in tank A. With the frequency band of the telephone band wide open, the useful frequencies transmitted were from approximately 2000 cycles per second to 50,000 cycles per second. By the use of electronic filters, we could limit this band to any part of the above band. The two ends of the telephone operated under water; in tank A there was a transmitter and a receiver and in tank B there was a transmitter and a receiver, all four of which were under water. A dolphin was placed in tank A and another dolphin in tank B.

The dolphin in tank A could communicate only with the dolphin in tank B, and vice vera. Thus their conversation would of necessity be limited to one another. As soon as the telephone was turned on, the dolphins exchanged sounds.

In a previous publication we described how two dolphins exchange such sounds when placed in the same tank isolated from one another physically but allowed to communicate through the water with one another.(Lilly, J. C. and Miller, A. M., "Vocal exchanges between dolphins.'' Science 134: 1873-76 (1961)) In this previous study we showed that dolphins exchange sounds very politely. When one is talking, the other one keeps quiet. In addition, we showed that they exchanged not only whistles, but also exchanged trains of clicking sounds. We also showed that the two kinds of sonic exchanges do not correspond in time, i.e., they can be talking with whistles and talking with clicks trains, the whistles and the clicks completely out of phase with one another. They can be using the silence of the whistle exchange with a click exchange and filling the silences of click exchange with a whistle exchange, and thus each are polite in the same mode. Thus one pair of dolphins talking can sound like two pairs of dolphins talking, one pair exchanging clickings, the other pair exchanging whistles.

These observations led to further studies in which we demonstrated unequivocally that each dolphin has at least two communication emitters, both in the nose, i.e., below the blowhole, one on each side. A right and a left phonation apparatus is demonstrated in the dolphin's nasal passages. Thus a given dolphin can carry on a whistle conversation with his right side and a clicking conversation with his left side and do the two quite independently with the two halves of his brain. An analogous human activity may be thought of as follows: if we could whisper and carry on a whispered conversation and at the same time carry on a vocalized conversation using two different apparatuses. Since we do not have the two sides divided with midline structures, we do not have this advantage. The dolphin can control the two airflows separately and the two membranes' vibrations separately. A comparable human activity is the typist typing a manuscript and at the same time carrying on a conversation. Now let us return to the "telephone" experiment.

With the telephone between tanks A and B, the resulting sonic exchanges were found to be very polite, most of the time each dolphin maintained silence while the other spoke It was found that while the telephone was turned on, the dolphins would exchange sounds most of the time. If we shut off the telephone, either all the sounds ceased or one or both dolphins gave the simple repetitious personal whistle ("signature whistle") characteristic of a solitary dolphin isolated alone. With the telephone off, any sounds that were emitted were completely out of synchrony with the sound emitted by the other dolphin. Little or no "interlock" was detected between the sounds emitted by the two dolphins. In other words, when sounds did occur with the phone not working, the alternating character of the clicks and of the whistles was lost. Either there was frequent overlap or many emissions were met with silence.

The telephone was modified by adding filters to reduce the intensity of sound at certain frequencies. The dolphins tested the system briefly. If the telephone was satisfactory, i.e., no missing critical frequencies, they continued using it. If it was unsatisfactory, i.e., missing critical frequencies, they stopped using it. In the latter case, they tested the system at intervals. If, meanwhile, we had restored the missing critical frequencies the dolphins resumed their "conversations" over the system.

We soon found that we could not cut the frequencies much below 28,000 cycles per second at the high end nor cut the frequences much above 5000 cycles at the low end without losing the exchanges. The best performances were found with bands extending from about 2000 to about 80,000 cycles per second. Thus the exchange frequency bands correspond fairly closely with the produced frequency bands. In other words, dolphin conversation used a large portion of those sounds whose frequencies are emitted. In addition, the exchange frequency band and the produced frequency band correspond surprisingly well with the predicted bands corresponding to our speech wave lengths in air by the constant wave length hypothesis. In other words, we use the same wave lengths for speech in air as the dolphins use for their speech in water; the frequencies used are in the same ratio ours to theirs as the ratio of the velocities of sound in the two media, air versus water ( I to 4.5).

It is wise to review the question asked at the beginning of this chapter. How do we know that these two dolphins are exchanging intelligent information? May they not be singing a senseless round, making dolphin music, playing a vocal game, or just saying repetitious simple phrases over and over, or possibly humming reassuring sounds to one another?

We do know that they are not just repeating the same thing again and again. To observe this result, one records these exchanges on a tape recorder and slows them down four times. (Ideally, 4 1/2 times according to the constant wave length hypothesis, to reduce the frequencies to our equivalent speech band.) At the new speed, one has lowered their frequencies four times, and stretched out each of their emissions by a factor of four. Thus we lower their 32,000 cycles per second to 8000 cycles per second. And their 1200 cycles per second to 300 cycles per second. Our speech, similarly lengthened, without the frequency changes, is not easily understood, the method is not ideal but we have found it to be useful. (A later development in the Institute allows us to shift to all of the frequencies without lengthening or shortening the emissions. This is discussed elsewhere in this book.)

These recordings are used to listen and to measure the sounds and to find out if the patterns are changing or are merely repetitious (for our pattern perception system, trained as it is to human patterns, not the delphinic ones). Apparently much smaller-brained animals exchange repetitious cries. At least they sound repetitious to us. Frogs, birds, fish, insects, bats, monkeys have different cries for different emotional states. No one so far has detected whether or not these are used for any communication other than the emotional state of the sender, i.e., signaling danger, sexual activities, hunger, etc. A relation seems apparent between the number of different patterns used and the size of the brain of the creature using them. We might thus expect a very large number of patterns in the dolphin exchanges, at least as many as we use in our exchanges. The very small-brained birds and fish have very limited vocabularies, at least as the patterns are currently measured and counted.

In measuring the sonic patterns one basic problem is whether one is measuring aspects that are important to the sender and to the receiver in carrying the meaning. Similarly, it is difficult to choose what to measure in the dolphins' exchanges; we may choose variables not at all importent to dolphins and sacrifice the important variables. Therefore, our criteria for differentiating and hence counting the number of different patterns may be totally incorrect. It is necessary to proceed empirically but cautiously and realize the limitations of the methods of arbitrarily choosing patterns.

When listening to the slowed-down exchanges, one is impressed with the numbers of different sounds one hears the dolphins use. The most varied of the transmissions that we have recorded are between "old" dolphins, those with large bones, scarred skin, truncated or missing teeth, and such marks of age. These are the really sophisticated vocalizers. When a Tursiops truncatus has become old enough so that the ends of his teeth are worn down flat, he has accumulated a very large number of sonic patterns which he exchanges with similar dolphins. Youngsters four to five years old have a sonic complexity which does not come up to that of the older ones; but even with them the first striking impression is that the versatility and complexity are well developed, that there is very little monotonous repetition, that one has a hard time keeping up with the new patterns as they emerge. Only if the dolphins are badly and continuously frightened are the sounds emitted monotonous and repetitious.

The sounds the dolphins use in their exchanges are difficult to categorize. They are all difficult to describe in words. In my laboratory, we use the following nine large classes to describe the sounds in a somewhat arbitrary fashion: (1) sounds that are emitted under water ("hydrosounds") and sounds that are emitted in air ("air sounds"); (2) whistles; (3) slow click trains; (4) fast click trains; (5) sounds resembling elements of human speech, called "humanoid" sounds; (6) a group that is like mimicry of other sonic sources (fish, ducks, sea gulls, boat engines [inboard or outboard], insects, etc.); (7) a group that is usually associated with emotional behavior (barks, screeches, hammerings, etc. ); ( 8 ) various non-vocalizing sounds including sneezes, respiration sounds (slow and fast), borborygmi, flatulence, tail slaps, the water noises of swimming at the surface, jumping, etc.; (9) ultrasounds (for us) used in echo-recognition and echo navigation (EREN), sometimes miscalled "SONAR" (sound navigation and ranging) after the human artificial systems.

The steady, most frequent outputs during non-emotional exchanges between dolphins are under water ("hydrosounds") (Figures 4, 6). These sounds are mostly whistles and various complex patterns of clickings and short humanoid emissions.

Their most frequent outputs with us are emitted in air, apparently to accommodate to us in our medium. They lift the blowhole up in the air, open it, and make very loud sounds. Such sounds can be whistles, clickings, barks, wails, and various "humanoid" sounds. The barks and wails in air seem to be analogous to their emotion-tied exchanges with one another under water.

The radical shift, voluntarily executed by dolphins making the sound in air as opposed to water is in response to our consistent use of air sounds with them. If we talk back to them under water, they answer us under water. If we talk to them in air, they answer us in air* (Figures 3, 8).*(Lilly, J. C., "Vocal behavior of the bottlenose dolphin." Proc Am. Philos. Soc.. 106:520-29 (1962))

We use the following working hypotheses in our communication research with dolphins:

The airborne whistles and the airborne clicks are attempts to communicate with us as they do with one another, i.e., attempts to induce us to use their mode of communication. Their humanoid sounds in air are their approximations to our communication sounds as distorted by their hearing and by their phonation apparatus (Figures 5, 6, 12). With the humanoid sounds, dolphins are attempting to communicate with us in our mode of communication.

At first a dolphin in the presence of a human uses mainly air clicks and air whistles. There are at least two main requirements for the use of humanoids in air: (1) the dolphin must have heard much human speech and (2) he must have had a long period of close, kindly contacts with us.

Once a dolphin has started airborne sounds with one or more of us in close contact, he may induce other dolphins in the colony (not in such close contact) to use the new mode, apparently in dolphin-to-dolphin exchanges. This latter airborne mode is apparently rarely, if ever, used by dolphins in the wild.

We have found that, in dealing with such a largebrained mammal, we must keep the working hypothesis in mind that "they are highly intelligent and are just as interested in communicating with us as we are with them." (With the species Tursiops truncatus this is reasonable; it may not be reasonable with smaller dolphins.) If we use any other hypothesis, we have no success whatsoever in dealing communicatively with them.

This hypothesis seems to be necessary and even overriding to accomplish the kinds of communication we are accomplishing and attempting to expand. The proof, the incontrovertible truth, that they are interested in this communication is developing slowly and carefully in our laboratory.

FIGURE 3. The Voice of the Dolphin in Air:
This table shows the dolphin's mimicry of human speech (the word "hello"), albiet at a higher frequency.

Computer Analyses

In each of 58 frequency bands a computer counts the number of times each of the bands is used above a chosen threshold as occurs in several tape replays of the "hello" and of the dolphin's reply. The bands extend from 135 Hertz to 8000 Hertz at 135 Hertz intervals. In the sixth band from the bottom (at 810 Hz) the number of instances of use (N) was 512. The use of each of the other bands is linearly proportional to the length of the black bar. It is to be noted that the woman's voice used frequency bands in two separate regions one at low frequencies and one at middle frequencies. The dolphin's reply shifts the lower frequencies to higher ones and matches the group of higher frequencies.

If and when dolphins and we do establish communication on a highly abstract level, the proof will become obvious and incontestable. In this book I give some of the details of this developing picture and give the reasons why we, the ones who work with them, just rely for some time on our faith in their intelligence. This faith is in the working hypothesis that both we and they are intelligent enough to break the interspecies communication barrier between these very different minds.

Without such a faith and working hypothesis one makes bad mistakes in tactics and in strategy with the dolphins. If one assumes that they are stupid, they act in a stupid manner. This is partly because in the eye of the beholder, stupidity is seen everywhere, and partly because dolphins understand, catch on fast, and act the way one expects them to act. We have seen dolphins acting rather stupidly in care of persons who think of them as "overgrown stupid fish kept in an aquarium." These dolphins develop some delightful contrasts in new behavior when one of the "believers" shows up and attempts communication.

This is one of the basic difficulties in this new field. One must have an unusual amount of consciousness of faith in one's hypotheses in order to make progress.

In reality, this faith factor is basic to all fields of science. It is necessary to elicit consciousness of this factor in researchers. Most of the sciences have been able to "forget" this necessity; however, it is present and used. In physics, for example, one constantly has a model in mind of what is happening in the system under investigation, and has a kind of temporary faith in the model. This is how physical apparatuses are designed to test the various consequences of a hypothesis.

In this new scientific area, we use the approach of the theoretical physicist teamed up to a certain extent with that of an experimental physicist. We set up hypotheses and operate temporarily as if they were true. We interact in the system under investigation, with each of us programmed with the hypothesis marked "as if true." We then estimate our progress and see how well we operate. We judge our success (or failure) by our success (or failure) in finding new information, i.e., data that were not predicted by the previous workers, nor by those current researchers whose hypotheses differ from ours. I consider this to be a very important point.

from: The Mind Of The Dolphin: A Nonhuman Intelligence - chapter 3