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
The experiment was divided into 11 phases. Each successive phase was based on the results obtained in the preceding phase and/or changes in the electronic equipment.
The sequence of signals presented during phases 1 to 4 and phase 8 was determined on the basis of a list of random numbers between 0 and 100. If two signals were used one after another to form a message, the action signal was given first and then the noun signal. Normally there were 8 runs per day in which about 15 signals or signal sequences were presented.
June 30, 1980 through July 9, 1980
Places used: Equipment Al, Hydrophone H1, dumbbell Ol and fender Ol'
The experiment was initiated in the following manner: first a "noun"-signal was given and afterwards the object belonging to that signal was held above the surface of the water for Gudrun to touch. Sometimes the object was held up without giving a signal, but then Gudrun was not allowed to touch the object. Each object had its own place, which later facilitated the control.
After Gudrun had learned to touch the object at the appropriate signal, it was checked to see whether she came toward the objects or toward the trainer. To that end, the objects were placed in the water at the moment that the signal was given and the trainer walked away. Once a signal was given, she was supposed to swim to the object and afterwards was rewarded accordingly with fish. In these cases, Gudrun went toward the object and not toward the trainer. In fact, she always swam toward the objects, even if no signal was given. After some time she began to be noticeably nervous, probably because she could not understand why she was rewarded one time and not the other.
From the fact that Gudrun always touched the objects which were offered and that she was clearly nervous we
concluded that she could not hear the signals or in any case not well enough. It was therefore decided to
discontinue this plan. The control situation in which Gudrun had to make a choice was not carried out anymore.
*) With all phases compare Fig. 1.
July 10, 1980 through July 20, 1980
Places used: Equipment Al, Hydrophone H1, Objects O2.
The advantage of this rearrangement was that Gudrun found herself more often near the hydrophone. The sequence was as follows: first a signal was given, and afterwards an object was held above the water. If the object presented belonged to the signal given, Gudrun was supposed to touch the object and she was then rewarded. If the object did not match the signal, Gudrun was not supposed to touch it. If in that case she did not touch the object, she was rewarded. If she wanted to touch the object anyway, it was removed and she was not given a reward.
We started with signal C, which meant that she could touch the dumbbell but not the fender. The second day she refused very clearly to touch the fender. She surfaced within ca. 1 metre of the fender and made a lot of noise. The next day we began with signal D. Now she was allowed to touch the fender but not the dumbbell. The first two times that the signal was given she didn't want to touch the fender. Afterwards, however, she began to touch it, initially hesitantly, but the hesitation quickly disappeared. Later that day, signal C was used once and the fender presented. She refused directly, in her characteristic manner, to touch this. Since July 13 series of both signals C and D were used, thus a combination of the dumbbell and the fender were offered.
The score was good: depending on her motivation 80 to 95% correct reactions. The speed with which she reacts to the signals is a good measure for her motivation. When she reacts very slowly, her score decreases, although at least 80%, and when she reacts very quickly her score is very high, about 95%.
We conclude that now Gudrun is able to hear the signals, she also reacts correctly. Because she clearly behaves differently when she doesn't want to touch an object than when she does, interpretation mistakes regarding her reactions can be ruled out. We may thus conclude that at this point Gudrun is capable of distinguishing between the two signals and further that she knows which object belongs to each signal.
July 21, 1980 through August 13, 1980
Places used: Equipment Al, Hydrophone H1, Objects O2.
The aim of this phase was to stimulate Gudrun to produce the signal herself if the artificial signal was delayed. Now Gudrun was expected to touch the object first and then wait under water for the appropriate signal, after which she was rewarded.
No registrations resembling the artificial signals were made. On August 1st her motivation clearly began to decline, which was made obvious by her continued attempts to eat the hydrophone. It appeared that Gudrun wasn't ready to produce the signals herself
August 15, 1980 through October 3, 1980
Places used: Equipment Al, Hydrophone H1, Objects O2 and O2'.
In this phase Gudrun, upon hearing a signal, had to make a choice.
To prevent her from having a preference in orientation, both objects were laid down at the edge of the pool, one to the left and one to the right of the hydrophone, and they were interchanged at random. The meaning was to get Gudrun, having heard the signal, to move in the direction of the object belonging to that signal.
Already at the end of the first round Gudrun began hesitantly to turn. Within a week a score of 95% correct reactions was obtained. But once again, her motivation began to decrease after some time. This was concluded among other things from the observation that she began purposely to make mistakes, then scoring less than 5%. After a 3-week break she was able to react correctly from the first run, but as the day wore on she began to get bored again.
From the fact that within a week Gudrun was able to react almost faultlessly to the signals we conclude that she could distinguish between them very well, and further that she learned very quickly. However, her motivation decreased as time went on, because no changes in the course of the experiment were made. The rapid speed of learning and the fact that a high score could be obtained points to the conclusion that she primarily made mistakes on purpose. Had she simply made a random choice the score would have been 50% instead of the 5% which she then actually scored.
October 6, 1980
Places used: Equipment Al, Hydrophone H2, Objects O3.
After the nouns we now turn our attention to the action words, the verbs. As preparation for the introduction of the verb signals a new object was introduced, namely a skippy ball. Because she was already trained to fetch objects that were thrown in the water, this phase was directed toward first presenting signal B and then throwing the object in the water. Already at the end of the day Gudrun had learned to wait for the signal. However, this phase was quickly terminated because the risk was too great that Gudrun would simply interpret the signal as a 'go-signal'.
October 7, 1980
Places used: Equipment Al, Hydrophone H2, Objects O3.
The method is the same as that in phase 3, namely touching objects and waiting for the signal.
During these runs Gudrun imitated both signal C and signal D several times. A few times, if the signal was deliberately delayed, Gudrun produced the correct signal herself. The verity of the reproduction was initially ascertained by ear, but a later analysis of the tapes confirmed that Gudrun's imitations very closely resembled the artificial signals. These registrations confirmed the expectations that Gudrun was able to produce the sounds herself. That she was able to do this correctly and even to use them intelligently appears from the fact that when she spontaneously gave a signal it was always the correct one.
October 8, 1980 through October 9, 1980
Places used: Equipment Al, Hydrophone H2, Objects O3.
Phase 7 was a continuation of phase 5. To avoid the risk of having signal B considered as a start signal, B was now combined with the object signals to form the message BC and BD. It was important to keep Gudrun under the water long enough so that she could hear both signals belonging to the message.
The fetching of the signalled objects presented no problems. It was more difficult to get Gudrun to stay under water long enough to hear both signals, since she tended to come directly to the surface. Because the stimulation of the use of signals was considered of the utmost importance at this point, this phase was discontinued.
October 10, 1980 through November 10, 1980
Places used: Equipment Al, Hydrophone H2, Objects O3.
The method was the same as in phase 3, thus touching objects and waiting for the signal.
Because it was now possible to listen to Gudrun's vocalisations under water we tried to stimulate her to make the signals herself. In the beginning she was given an extra reward if she made the sound under water. The time delay in presenting the signals was made longer and longer to stimulate her to make the signal herself at the proper moment. If she produced the signal she was royally rewarded. If not, an artificial signal was presented as a reminder.
On October 21st she imitated the signals four out of five times. Three days later she often produced the correct signal spontaneously when an object was shown. She never made the wrong signal. She produced signal D 80% of the times the fender was shown, but signal C only in 35% of the cases the dumbbell was shown. The reason for this preference for signal D is unknown.
We concluded that she was able to correlate the signals with the objects shown.
November 11, 1980 through January 26, 1981 (with a break from November 26, 1980 through December 22, 1980).
Places used: Equipment Al, Hydrophone H2, Objects O3 and O1, later Equipment A2, Hydrophone H3, Objects O3 and O1.
In this phase both verbs were introduced, that is, signal A for "take" and signal B for "bring" or "give". To introduce the new action word "take" Gudrun was first offered the signal combination AC or AD. She then received the object from one trainer and was obliged to bring it to a trainer waiting at the other side of the pool, and then to swim back without the object to the hydrophone on the other side of the pool, where she was rewarded.
After only one day the help of the extra trainer was unnecessary. Gudrun herself, after receiving the signal AC or AD, brought the object from O3 to O1, or from Ol to O3 after message BC or BD.
On November 24th she reacted correctly to both objects, even when they had floated to the middle of the pool. Upon hearing the signal combination AC or AD she brought the correct object away from the trainer and after BC or BD she brought the object closer to the trainer.
After a few days Gudrun obviously became bored. She became less and less cooperative and she tried to remove the objects from the pool. The objects were always returned, however, and the experiment continued. The aim was to make it clear that she could only have it her way if she "asked" for it by producing a signal sequence. On December 23rd Gudrun pushed the fender against the hydrophone and gave, above water, the signal AD, that is, the message "take the fender". The fender was taken away and she was of course rewarded. Later, she used both AC and AD, always in a meaningful context, by which is meant that she used the signal combinations only in situations where it was possible to react by taking the object concerned. There are 3 important points in Gudrun's use of signals:
January 27, 1981 through March 2, 1981
Places used: Equipment A3, Hydrophone H4, Objects O1.
The objective of this phase was to ensure that Gudrun did not associate the actions with fixed positions but that the direction was determined by the location of the trainer. It was intended to emphasize the idea that signal A was to be interpreted from the "speaker" and B as going toward the "speaker", independent of what position the "speaker" might take.
The training started by presenting as an initial test the message AC. Gudrun took the dumbbell, swam away with it and set it down at position P, the same point which she later used to respond to signal AD*. Even when the objects had drifted to the middle of the pool she reacted correctly to the signals. Also in this phase she time and again used AC and AD.
The fact that she had interpreted AC and AD as indications of direction had now been emphasized. It is clear that
points Ol and O3 themselves were not the determining factors for her movements, with the objects, but that the
most important point was the indication of the direction with respect to the speaker. We further note that as a
means of documentation at this point, films and tape recordings were made of Gudrun going through this and all
previous phases. (ATV/CT Documentary "The Talking Whale").
* Position P had not been used before during the experiment.
February 16, 1981 through March 2, 1981 (alternating with Phase 10 sessions)
Places used: Equipment A3, Hydrophone H4, Objects O1.
Because Gudrun has not yet used signal B, that is, she had not yet asked for something to be given her, a situation was created in which she would be stimulated to produce her own signal B. As she had already worked so often with the dumbbell and the fender, a new signal, signal E meaning a 'large fish' was introduced.
Three objects were used: the dumbbell, the fender and a large fish (mackerel or herring). It is important to note that these fish were used only for the training and not as a reward. Initially, Gudrun was expected, just as in previous phases, to touch the object and then wait for the signal that belonged to it: first the dumbbell, then the fender and finally the fish. In the first two cases, the object was given to Gudrun, released and then taken away, after which she was to wait for the signal. In the last case, however, the fish was extended to her, she held it in her mouth: and after she had given it back to the trainer and signal E was given she was rewarded with a number of fish. Later it was attempted to get Gudrun to perform the same actions with the fish that she had with the dumbbell and the fender. Therefore, as with signals BC and BD, the signal BE was presented and Gudrun was expected to take the fish to the trainer on the other side of the pool.
It was only necessary in the early phase of the training of signal E to hold the fish in order to prevent Gudrun to swallow it. After that early phase - which only lasted about 10 minutes - she returned the fish without being restrained by hand.
The speed at which she learned signal E was amazingly fast. Already the second time the signal was presented she imitated it perfectly under water and then later regularly above water as well. Later the same day she began spontaneously to produce signal E herself above water, for which she was consequently rewarded.
The speed at which she was able to learn signal E suggest that she found it very easy to learn. On February 19th the signal BE was introduced twice in an unsuccessful attempt to teach Gudrun to bring the fish to the trainer. She took the fish, turned in the appropriate direction, but apparently overcome by temptation, swallowed the fish.
Unfortunately, due to circumstances beyond our control, the experiment had to be stopped, on March 2nd.
The numerous times that Gudrun spontaneously produced signal E shows that she indeed correlated this signal with fish. She had never done this so quickly with other objects. Nevertheless, there is no reason to assume that this signal is any easier to reproduce than the others. This was a difficult experiment for Gudrun because she was expected not to swallow the fish immediately. That she first turned in the direction of the trainer indicates that she understood what was expected of her but that it cost her too much self-control to bring the task to its proper end.
We postulate that, had the experiment been continued, Gudrun would have eventually been able to apply her knowledge of the new signal E to form the message BE, that is to ask for the fish to be given to her. Note that she had already begun to use signal E to demand fish in the beginning of this phase. At this stage in the experiment, no care was taken to remove the trainer, which rnight seemingly imply that we allowed the "Clever Hans syndrome" to operate. However, the syndrome does not play a part at all in the crucial phase of the experiment. The animal was not asked to make a choice, which could be influenced because of the syndrome. On the contrary, Gudrun was given the option to take the initiative to act i.e. to communicate. To us the task to await, to absorb her question c.q. command and to act accordingly.
Previous page |
Next page |
Back to the Online Papers Page |
Back to the Publications Page |