W.H. Dudok van Heel, C. Kamminga and J.D. van der Toorn (1982)
An experiment in two-way communication in Orcinus orca L.
From: Aquatic Mammals 9(3): 69-82
Table of contents
Subject and training
Appendix I - Time dependent frequency stimuli
Appendix II - Considerations with respect to future research
During the first phases of the experiment it appeared that the orca could visually choose the correct object when an acoustical stimulus was presented. Moreover, she proved to be able to produce the correct acoustical signal at the sight of an object. As explained by Geschwind (1964) a connection between the visual and acoustical association areas in the brain must exist in order to be able to perform such tasks. There can be little doubt that this condition is realised in the orca. It means that the planning of future communication experiments with this species is considerably facilitated.
The next step was to teach the animal the meaning of two action words, "take" and "bring", in combination with the object signals. Within a short time she was able to respond correctly to the four combinations of signals. As usual as things go on unchanged, Gudrun became bored and tried to get rid of the objects. She discovered that her attempts were only successful if she gave the signal sequence "take fender" or "take dumbbell", interpreting the signal "take" as away from the speaker. She never asked to be given one of the objects, in which she was no longer interested. In order to make "give" more interesting, a new object signal "fish" was introduced. She learned the meaning of the signal amazingly fast. A combination of the signals "give" and "fish" could not be accomplished before the experiments had to be terminated. However, her initial reactions suggest that she would have eventually been able to produce the required signal message.
The outcome of this experiment forms a good launching pad from which further research on man-orca communication can be carried out.
The use of frequency-modulated signals, which were to a certain extent related to the animal's natural tonal vocalisations, enables more structural information to be contained in the same time span as constant frequency stimuli.
Undoubtedly, an experiment of this scope involves the help and interest of several people also outside the field of delphinid research. To single out a few of these we would like to mention Dr. E. Leeuwenberg and Dr. H. Buffart (University of Nijmegen, Netherlands) for their help on the coding of the stimulus contours, Prof. dr. Fl. Verheyen (University of Utrecht) and Prof Ir. Y. Boxma (DeIft University of Technology). The technical part of our experiment would not have been possible without the skilled assistance of Mr. B.M. van den Boom and the work of the student Mr. J.VI. Akkermans.
We thank Prof. R.I. Harrison, Mr. V.J.A. Manton, Mr. A.D.G. Dral and R. Paul Terry for their critical reading of the manuscript and their valuable suggestions.
We are grateful to Anglia Television (formerly A.T.V. now C.T.), in particular Mr Robin Brown, director of the documentary "The Talking Whale", which covers the experiment with Gudrun, for their interest and cooperation and not in the least for their donation to the Netherlands Foundation for Aquatic Mammal Research.
We finally thank this Foundation for the financial and logistic backing of our work.
Central Television (formerly ATV) Documentary: "The Talking Whale", directed by Robin Brown.
Intuitively we note that constant-frequency stimuli possess a minimum transportable amount of information with regard to a frequency-modulated pattern. Moreover, constant-frequency stimuli are not quite natural stimuli in information-processing experiments, although they are easily generated and well-established patterns in most psycho-physical experiments.
The information theoretical description of the stimuli used is based on the outcomes of the study of pattern dimensions as developed by Leeuwenberg and Buffart in the last decade. This pattern-coding theory enables one to calculate the structural information load of a pattern (the number of degrees of freedom), in contrast to the well-known selective information theory proposed by Shannon (1963).
The amount of structural information gives a measure for the - perceptual - complexity of a pattern, as short as possible. Without going into detail about theoretical backgrounds of coding theory - the interested reader is referred to the coding manual by Buffart and Leeuwenberg (1982)- we shall describe the information load of the various patterns used in our experiment.
The five types of stylised tonal patterns we used are represented in figures 2 A to E. (Click here to load figure 2 (Signals) in current window or click here to load figure 2 in a new window.)
If we take a closer look at the pattern indicated in figure 2A, representing the time-dependent frequency structure, the following parameters are included:
Summing up the independent changes that occur in the pattern over time; we arrive at 5 degrees of freedom, i.e. 5 parameters that might change our basic pattern. So we observe that the minimum information load in this configuration is equal to the number of degrees of freedom, that is I=5.
In exactly the same way we can describe the phase-inverted pattern as is presented in figure 2B. It contains the same structural information load, although the perceptive value of the information is not included: a pattern going from a low frequency via some path to a higher frequency gives another perceptive sensation then the inverted one.
We thus arrive for the signal B at a value of I=5 for the amount of structural information.
We now take figure 2C and proceed along the pattern in time to note the degrees of freedom. At first glance the three segments of this pattern cover figure 2A's evolution in time up to 1.0 sec. From then on there is not only the continuance in time for 0.3 sec but also a decreasing frequency present. After attaining the original starting point with regard to the frequency domain, the pattern ends after 0.1 sec without any change in the frequency.
Summing up the various points of change and the corresponding changes in time and frequency at these points, we arrive at an information load in this pattern of I=8.
If we were to take into account the fact that in the second part of the pattern, where we arrive at a frequency that has already been determined (indicating that a certain 'memory' in the perceptive experience is built-in), we could decrease the information load to I=7. However, as there is by no means evidence for such a phenomenon in the auditory perception of delphinids, we are inclined to insist on the higher information load of 8 degrees of freedom.
Proceeding along the same lines as for figure 2C in the case of signal C, we encounter no problems in defining a structural information load of 1=8 for the case of the phase-inverted pattern, signal D. By virtue of this reasoning, it is now clear that a constant-frequency stimulus as presented in figure 2E yields the minimum information load contained in such a pattern, namely 1=2 (see the JANUS research program of J.C. Lilly, 1977).
As far as concatenated patterns to form a message are concerned, we could simply add the amounts of information by using the additivity theorem of the individual patterns if we treat them to be independent of each other.
Thus, a message like the one used in our experiment in the form of A+C or B+D gives a total amount of structural information as I(A)+I(C) = 13 and similarly I(B)+I(D) = 13.
If Gudrun would become available for research again, which is not possible at present, it would be very interesting to continue the initial experiment in order to see what level she would be able to master in two-way communication. However, if at some stage she would not be able to improve her language capabilities the following argument should be taken into account.
A nightingale male chicken has at a very distinctive period in its early life to hear an adult male sing. If it does not and the critical period has passed, this male chicken will, once adult, not be able to sing the characteristic male nightingale song. In the same way children will have to learn and speak before the age of at best 8 years, otherwise the mastering of the human language to its full extent (i.e. expressing abstract thoughts) is impeded.
We have no prove whether an orca has such a critical learning period, neither at which age this period, if present, may end. However, the existence of this period is most likely.
Gudrun was about 5½ years of age at the time of the experiment. If her training would continue and a maximum level would be established, this would mean, if my view is correct, that there are two possibilities. Either Orcinus orca, i.e. this particular individual, is not intelligent enough to reach a more complicated level, or we were too late in Gudrun's life to give her a full chance to develop her innate capabilities. Therefore I am strongly in favour of repeating this experiment with the youngest possible animals.
Gudrun when caught was only 270 cm long, but completely weaned and she was no exception as we learned. This is an indication that the Icelandic population probably grows to a smaller length than the NE Pacific population, of which the juveniles are weaned at a length of at least 300 cm.
This paper was originally published in Aquatic Mammals 9(3): 69-82. In this online version, a picture of Gudrun and drawings of the objects used in the experiment were added. Consequently, the figure numbering has been adjusted. With the exception of some text markup changes, the original text has been left untouched. A Table of Contents has been added as well as the following short history of Gudrun.
Gudrun was caught in Iceland, in the Skeiðarársandur area (SE Iceland) in October 1976 in an operation led by Jón Gunnarson (see Sigurjónsson and Leatherwood, 1988) and was transported to the Dolfinarium in Harderwijk, the Netherlands. . She was caught together with Kim, a female that was transported to Marineland, France and Kenau, a female that was transported to Sea World in San Diego in May 1977 (Hoyt, 1990). She was named after the boat that was used in the capture operation, the M/V Guðrún. Although she occasionally had other killer whales for company (whales in transit to other parks) she spent most of her time in the company of a group of bottlenose dolphins. She was part of show performances. In 1987, the Dolfinarium decided to move Gudrun, officially on a breeding loan, to Sea World of Florida, in Orlando. At first she had to get used to being with other killer whales after years in the company of bottlenose dolphins only. As Sea World trainers put it: she was behaving as a dolphin and had to learn to behave as a killer whale. She adapted quite well to her new environment. Within a year she was integrated in the Sea World performances. In 1988, she became pregnant and had her first calf, Taima, the next year. A few years later she had another calf. Unfortunately, Gudrun died on February 25, 1996 of the complications of a miscarriage (her third pregnancy).
|Back to the Online Papers Page||Back to the Publications Page|