Using Self-View Television to Distinguish between Self-Examination and Social Behavior in the Bottlenose Dolphin (Tursiops truncatus)

Ken Marten and Suchi Psarakos

Dr. Ken Marten e-mail:

Kailua, Hawaii

Reprinted with permission from:
Consciousness and Cognition, Volume 4, Number 2, June 1995,
pp. 205-224, Orlando: Academic Press.
Academic Press Web Page: ""

In mirror mark tests dolphins twist, posture, and engage in open-mouth and head movements, often repetitive. Because postures and an open mouth are also dolphin social behaviors, we used self-view television as a manipulatable mirror to distinguish between self-examination and social behavior. Two dolphins were exposed to alternating real-time self-view ("mirror mode") and playback of same, to determine if they distinguished between them. The adult male engaged in elaborate open-mouth behaviors in mirror mode, but usually just watched when played back the same material. Mirror mode behavior was also compared to interacting with real dolphins (controls). Mark tests were conducted, as well as switches from front to side self-views to see if the dolphins turned. They presented marked areas to the self-view television, and turned. The results suggest self-examination over social behavior.


An animal demonstrates self-recognition if it treats its mirror image as a reflection of itself, and not as another member of its species (Gallup, 1970). Apes and monkeys exposed to mirrors and real-time television have been tested for self-recognition by examining their response, when fully awake, to a conspicuous mark placed on their forehead while they were under anesthesia. Chimpanzees, orangutans, and gorillas exhibit self-recognition by attempting to remove the mark (Gallup, 1970; Lethmate and Ducker, 1973; Suarez and Gallup, 1981; Parker, 1989; Patterson, 1984, 1991, 1994), whereas monkeys and lesser apes (gibbons) interact socially with their marked mirror image (Gallup, 1979, 1982; Patterson, personal communication). Data from a previous study suggest that dolphins, like the great apes, may also use a mirror for self-examination (Marten and Psarakos, 1994).
In a 2 1/2 year study of bottlenose dolphin interaction with mirrors we found very suggestive evidence for self-recognition in this species (Marten and Psarakos, 1994). The study consisted of mirror mark tests and controls with five individuals. In the mark tests the dolphins appeared to use a 4 ft. diameter mirror to examine a zinc oxide mark on their side by using various maneuvering and twisting postures. (Unlike dyes for primates, zinc oxide is a tactile stimulus as well as a visual one.)In the control tests, mirror behavior and interaction with a real stranger through an underwater barred gate were quite different, the most outstanding difference being that the dolphins attended to a real stranger for close to 100% of a several hour test period, whereas they attended to a mirror for less than 1% of the time.
The lack of a hand to touch the mark (or mirror) makes the mark test less definitive for dolphins than it is for primates. Additionally, many of the behaviors that dolphins use in front of a mirror resemble social behaviors as described by Pryor (1975), Overstrom (1983), Ostman (1985; 1991), and Norris (1991). In the mirror/no mark situation the dolphins often engaged in rhythmic and repetitive head, mouth, and body movements. Open-mouth behavior, head movements, and posturing may also occur when a dolphin is threatening another dolphin, or merely excited. So, if a dolphin exhibits unusual or extreme posturing in the mirror while looking at his mark, or makes head or mouth movements, it is possible that the dolphin is using the postures or behaviors to interact "socially" with what he perceives to be another dolphin in the mirror. However, a self-aware species might use some of its social repertoire while examining its mirror image.


Our research used various forms of manipulated video as a tool to differentiate between self-examination and social behavior, as well as to devise alternative, species-appropriate methods of testing for self-awareness (Gallup, 1979). In the first test, we hypothesize that if dolphins recognize themselves on television, their behavior should be different with a real-time self-view ("mirror mode") than it is when they are shown taped playback of the same mirror mode material. The test is designed to elicit one response if the dolphin mistakes its television image for another dolphin, and a different response if the dolphin differentiates between real-time self-view and playback. In the second series of tests we present control data on dolphin interactions with real strangers, for comparison with mirror mode behavior. The third series of tests compares marked dolphin behavior in mirror mode with playback mode. In the fourth series, the turn tests, we examine whether dolphins turn when the real-time display on their television is suddenly switched from the usual frontal view to a side view. (Gallup, 1979 and 1987, suggests an alternate turn test setup using a phony background probe.) None of the research was done with food rewards, so all results were voluntary actions on the part of the dolphins.

Television Tests: Real-Time Self-View Television
("Mirror Mode") vs. Playback


Video setup.
We used a Minolta S-VHS Series V-2000 video camera with a Sigma 0.6 wide angle lens. The camera was located to the dolphins' left of the television (Fig. 1). We also used one Panasonic AG-1960 S-VHS VCR for playback to the dolphins, and another AG-1960 for recording the camera data. The television was a 20" Sony Trinitron Color television.

[Figure 1: 50K GIF Schematic of Underwater Lab]

Figure 1. Two-tank complex (top view) and underwater viewing laboratory, Sea Life Park, Hawaii. The camera next to the 20" television in the small square window provided the frontal self-view on the television (mirror mode) for the television tests. Cameras in the concave windows provided the side views of the dolphins which were shown to them in the turn tests.

The video setup is referred to as "mirror mode" when a video camera is aimed at the dolphin and the real-time image is displayed on the dolphin's television, allowing the dolphin to look at itself as if in a mirror. When we later play back the videotape of a mirror mode session to the dolphins on their television ("playback mode"), the movements of the taped dolphin on the television no longer mirror those of the dolphin watching. The subject dolphins could not be separated, thus the playback videotape often had both dolphins in it and both dolphins could watch. As a consequence of this, during playback mode a given dolphin was not always looking at playback of itself.
Sample. Adult male Keola, 14 years old, and juvenile male Hot Rod, 6 years old, were chosen because they gave sustained interest to the television. They were exposed to alternating sessions of 10 minutes of mirror mode and 10 minutes of playback mode.
The dolphins had received exposure to television prior to their first television test, including real-time self-view, previously taped playback of themselves, taped playback of other dolphins in a different tank, and miscellaneous video such as abstract graphics, cartoons, and television programs. Keola and Hot Rod each had television exposure (all types) of approximately 38 hours. Total mirror mode and playback mode television exposure prior to the television tests was 1 hour and 10 minutes, and 1 hour and 30 minutes respectively. The dolphins had also been exposed to mirrors prior to their television tests. Keola and Hot Rod each had mirror exposure of approximately 260 hours.
Experimental protocol: alternating "mirror mode" and "playback mode". The tests for Keola and Hot Rod were conducted over two days, using alternating mirror mode and playback mode sessions of the following duration (in minutes and seconds): Day 1 had two mirror mode sessions of 13:43 and 8:40, and one playback mode session of 16:06; Day 2 had four mirror mode sessions of 4:57, 10:50, 9:24, and 9:59, and four playback mode sessions of 10:09, 5:49, 4:11, and 6:20. (The protocol had a third video mode, "other dolphins" - playback of dolphins videotaped in another tank, not analyzed here. The dolphins appeared to treat playback of themselves and "other dolphins" similarly.)
Quantitative analysis of television test behaviors. Behavior categories were established looking at behavior from both modes, emphasizing the intricate head and mouth movements dolphins use while watching television and looking at mirrors. The categories are listed on the x-axes of the histograms presented in the results section. Open-mouth behaviors without head movements were divided between "wide open" and "not very wide open" and lasting less than or more than two seconds. Similarly, head movements without open-mouth behaviors were divided between big head movements and small head movements and lasting less than or more than two seconds. "Mouth and Head" as presented here is an important composite category for television watching consisting of about 20 permutations of open-mouth behaviors simultaneous or in sequence with head movements. The classification of the mouth and head components of the behavior was the same as the "mouth without head" and "head without mouth" behaviors described above. "Body spins" were 360-degree turns on the long axis. During "Watching" no overt movements or other behavior occurred. "Bubbles heard" was included because the blowhole was sometimes above the top edge of the field of view of the video camera. "Excitement" consisted of mainly small, rapid movements with widened eyes. "Body twist" was a quick, writhing turn of less than 360-degrees, not necessarily on one axis. "Touch"-ing the window was usually done with the rostrum. "Swim-by, some interest" means the dolphin glanced over at the television while swimming by, but did not stop.
The behaviors were counted in each television cycle. The behaviors were not distributed normally over sessions, but occurred in bouts. That is, the occurrence of any given behavior was predictive of subsequent behavior. Because television mode trials are not statistically independent, it is inappropriate to use statistical tests for differences between the modes. Unfortunately, this makes the comparison less quantitative and objective, but it cannot be avoided. Future research should incorporate independent events in the tests so that statistical tests can be used. This is a true challenge when dealing with intelligent animals like dolphins, whose current behavior has such a great effect on future behavior.

Television Test Results

Qualitative: Keola and Hot Rod.
In general, when the television was on the dolphins approached it with intent, visibly slowing down to arrive and remain in front of it. Keola's behavior was very different in mirror mode compared to playback mode, and this was obvious even without a quantitative analysis. In mirror mode he engaged in long bouts of rhythmic head and wide open-mouth movements (Fig. 2A), whereas these were virtually absent in playback mode, where he often just watched (Fig. 2B). Once, when the monitor was in mirror mode and the dolphins were being fed, Hot Rod swam to the monitor, popped a fish out of his mouth so that it was clearly visible on the television, retracted the fish, and left (Fig. 3H). He never repeated the behavior in playback mode. (We have seen other dolphins do this in mirror mode, including one who waved the fish up and down in front of the self-view monitor.) Other than this, there did not seem to be obvious qualitative differences in Hot Rod's behavior in mirror mode vs. playback mode, so we rely here on the quantitative analysis.
Quantitative: Keola and Hot Rod (Fig. 4). The bars in Fig. 4 represent for Keola and Hot Rod the overall rate of occurrence of each behavior in each television mode, from all sessions. The first thing to notice is that most behaviors occur more in mirror mode than playback mode. The biggest difference in bar height is Keola's combined mouth and head movements. Keola's higher number of "Short watches" (two to five seconds) and "Swim-bys" in mirror mode show how much more interested he was in mirror mode than playback. Interestingly, long watches (greater than or equal to five seconds) characterized by just watching and not doing anything else were more prevalent during playback.
Combined head and mouth behaviors: Keola and Hot Rod. The many head and mouth behaviors which occurred simultaneously or in unbroken succession are all combined into one category in Fig. 4. This combined category contains about 20 permutations of specific combined head and mouth behaviors. For Keola these combined head and mouth behaviors occurred more than five times as much in mirror mode than in playback mode; for Hot Rod it was reversed, as he engaged in head and mouth movements twice as much in playback mode than mirror mode.
Keola's head/mouth behaviors were qualitatively different in the two television modes. He made 21 combined head/mouth behaviors in mirror mode (a rate of 22 per hour), whereas he made only three in playback mode (a rate of 4.2 per hour). Eleven of the 21 in mirror mode involved "wide or big" open-mouth movements; neither wide open-mouth movements nor long duration (> two seconds) mouth movements ever occurred with head movements in playback mode. Repetition and head bobbing were occasionally associated with the combined head-mouth movements in mirror mode but not playback mode.
Although there were differences in Hot Rod's behavior in mirror and playback modes, we did not find major qualitative differences as we did with Keola.

[Figure 2: 33K GIF Photo]

Figure 2. (A) Mirror mode. When we display his real-time self-view image on the television ("mirror mode"), adult male Keola engages in repetitive head and open-mouth movements. (B) Playback mode. When these same open-mouth movements are played back to Keola (inset) he just watches and does not make open-mouth movements in response. (The inset is all he sees; it fills the whole screen.)

[Figure 3: 50K GIF Photo]
[Figure 3 (more detail): 183 GIF Photo]

Figure 3. (A) Open-mouth movements in response to a real dolphin are frenzied biting. (B) Keola has maneuvered so that the mark on his right side shows on the monitor even though the camera is to his left. (C) Wide open mouth following the application of denture adhesive to the teeth. (D, E, and F) Frontal view (D) is changed to side view (E), then dolphin turns (F). (G and H) The tongue curl (G) and presentation of a fish after eating it (H) are examples of unique behavior in front of the self-view television.

[Figure 4A: 33K GIF Graph]

[Figure 4B: 33K GIF Graph]

Figure 4. Television Test: Behavior in "mirror mode" vs. "playback mode". The bar graphs represent the total counts over all sessions in each mode divided by the total time in that mode. See text for a description of these behaviors. (A) Keola, adult male (14 years). (B) Hot Rod, juvenile male (6 years).

Television Test Discussion

Difference between mirror mode and playback mode: head and mouth behaviors.
The combined head and mouth behaviors, short duration head movements, and mouth movements which Keola used so much in mirror mode, but much less in playback mode, could be self-examination behaviors. In mirror mode Keola tended to stay right in front of the television, opening his mouth wide and rhythmically moving it all around (Fig. 2A). He never did this during playback mode (Fig. 2B). It is likely he was either watching the results of his own movements, or maneuvering his head to examine the inside of his mouth. To test the idea that he might be examining the inside of his mouth, we put denture adhesive on his teeth during a self-view television session. Although he opened his mouth wider with the adhesive (Fig. 3C), wide open-mouth behavior did not occur significantly more with it than without it.
We worked with 20 year old wild-caught male Kamalii after Keola and Hot Rod. Kamalii's reaction to playback of himself after weeks of real-time self-view (see General Discussion, "Apparent Social Behavior"), is interesting to compare to Keola's. Like Keola, he behaved differently with playback than real-time self-view. But unlike Keola, instead of passively watching playback, Kamalii engaged in exaggerated forms of the behavior he did in self-view: twitching, undulating body fat, and head movements, as if he were trying to force playback to do what mirror mode had done. (By exaggerated we mean the movements had easily noticeable larger trajectories, and some got repeated more than in self-view. This is a qualitative observation and no reliability counts were done.) Each dolphin distinguished playback from mirror mode in his own way, suggesting self-examination over social behavior.
Alternative interpretations of differences in mirror mode compared to playback mode. One would expect the self-aware dolphin to behave differently in the two television modes: perhaps to utilize the television image for self-examination in mirror mode, and to observe in playback mode. If the dolphin perceives its mirror mode television image as another dolphin, then whether it is watching mirror mode or playback mode, the dolphin should act the same way -- possibly interact socially with the "television dolphin", or possibly just observe it, but behave the same in both modes. However, alternative interpretations of differences in mirror mode compared to playback mode exist: the movement of the "television dolphin" is correlated to the viewing dolphin's behavior in mirror mode but not in playback mode. This could encourage some form of increased social interaction in mirror mode, but not in playback mode, because the mirror mode television dolphin "reacts", but the playback mode television dolphin does not. A specific case of this which may also increase interaction is when the subject dolphin may be sustaining its mirror mode response because it thinks the television dolphin is mimicking it. If the subject opens his mouth, so does the mirror mode television dolphin, and so on. Such a mimic is more engaging compared to the playback mode television dolphin whose behavior does not respond to the watcher no matter what he does. Mimicry and synchrony are very much a part of a dolphin's social repertoire, both in the wild and in captivity. Another alternative interpretation is that the dolphin could react differently to the two television modes because it recognizes a contingency between its actions and those of an abstract image on the television, not because it "sees" another dolphin, or recognizes itself. Thus, neither more interest in mirror mode nor playing with contingency need indicate self-recognition. Compare this to the infant who enjoys controlling the movement of the image of his foot in a mirror, even when he cannot see his own foot directly, does not know that it is his own foot he is seeing, and does not recognize himself in the mirror.

Control Tests: Encounters with Real Strangers


First-time encounter through an underwater barred gate: one adult male and four adult female dolphins.
The control experiment which most closely resembles our test situations was presented in Marten and Psarakos (in press). Four adult female dolphins were in the large tank (Fig. 1) and an adult male, Kamalii, they had never met was put in the adjacent small tank. They were separated by an underwater gate with bars 15 cm apart (Fig. 3A).
First-time encounter with no gate: two adult males (Keola and Maka). Keola was moved from our research tank and put directly into a tank that contained an adult male, Maka (15-20 years), whom Keola had never met. In this new tank he could also see and hear two dolphins he had not seen for several years (an old adult male and an adult female) through a metal, underwater barred gate. We observed and videotaped the dolphins' behavior from the surface for the first two hours after Keola's arrival.
First-time encounter through an underwater barred gate: three adult male dolphins, one juvenile male dolphin, and two juvenile false killer whales. Two false killer whales, a male and a female, were introduced across an underwater barred gate to four bottlenose dolphins, three adult males and juvenile Hot Rod. Although the dolphins had previously lived with a false killer whale, they had never met these two. The male false killer whale was introduced alone, then approximately 15 minutes later the female joined him. We observed and videotaped the encounter from the surface for one hour. (False killer whales are very closely related to bottlenose dolphins. They have mated and had fertile offspring at our facility.)

Control Tests Results

First-time encounter through an underwater barred gate: one adult male dolphin, four adult female dolphins.
The dolphins attended to each other for the majority of the 40-minute test period. A narrow channel between the two tanks may have restricted the dolphins. Open mouth behavior occurred on both sides of the gate, and was intense and frenzied, much like rapid biting (Fig. 3A).
First-time encounter with no gate: two adult males (Keola and Maka). As soon as Keola was put in the tank, he and Maka swam together rapidly around the tank. They stayed close, even jumping out of the water together, for approximately the first five minutes. They spent much of the next two hours continuing to swim close together, slower than at first, and without jumps. Throughout this swimming and in most of the gate behavior described below, Keola repeatedly put his rostrum near Maka's genital area.
Maka engaged in most of the gate encounters with the two neighboring dolphins, with Keola positioned behind him in the rostral-genital position. Keola's first major interaction at the gate occurred fifteen minutes after arriving in the new tank. The interaction was violent, with fast movements and quick-moving open-mouth behavior; head movements were hard to see, but if present were not pronounced. This encounter lasted approximately 10-20 seconds. Keola's subsequent encounters were less violent, shorter, and did not appear to have open-mouth behavior. His orientation to the gate was always perpendicular. Keola interacted at the gate approximately 10-15 times in the two hours.
First-time encounter through an underwater barred gate: three adult male dolphins, one juvenile male dolphin, and two juvenile false killer whales. When the male false killer whale was put in the tank, all four dolphins immediately swam to the gate. The dolphins remained oriented perpendicular to the gate throughout the one-hour observation period, with only a few brief departures after 25 minutes. The male false killer whale spent several minutes at the gate, facing the dolphins. There was a period of prolonged, rapid, and loud airborne vocalizing during the confrontation across the gate. After the female false killer whale was introduced the two whales swam close together breathing in synchrony, and circled their tank for the remainder of the observation period. There were no more direct encounters between the whales and dolphins at the gate. The dolphins' orientation was always perpendicular to the gate, never parallel. There appeared to be little or no posturing. The dolphins blew bubbles at least four times, once as a bubble stream. One dolphin spun once on its long axis. There was occasional open-mouth behavior, but apparently without simultaneous head movement.

Control Tests Discussion

First-time encounter through an underwater barred gate: one adult male dolphin, four adult female dolphins.
In this test, the adult male Kamalii exhibited all the social behaviors found in real social encounters but absent in mirror-directed behavior: long individual attention; frenzied, intense movements, including rapid biting movements; perpendicular rather than parallel orientation to the gate; intense, loud vocalizations; and lack of cyclic movements of the body and head. The females exhibited similar behavior, somewhat mitigated by the narrow channel on their side of the gate.
First-time encounter with no gate: two adult males (Keola and Maka). It is clear from this example of a strange encounter that if two stranger dolphins are introduced and have physical access to each other, the ensuing interaction is fast-paced, physically close, and contains strong movements. Overall, Keola's behavior with Maka, and with old friends through the gate, bore little resemblance to his mirror mode behavior: he moved fast and frenzied with the real dolphins, whereas he acted calmer in front of the television. (Keola's gate behavior was most likely affected by the excitement of being in the tank with Maka.) It is not clear whether the difference in his behavior is attributable to situational differences (a "mirror dolphin's" behavior is limited: he can't swim around the subject or jump out of the water), or to Keola's awareness of the true difference between his mirror image and a real dolphin.
First-time encounter through an underwater barred gate: three adult male dolphins, one juvenile male dolphin, and two juvenile false killer whales. There were both similarities and differences between the dolphins' stranger encounter with false killer whales and the television-directed behavior we have witnessed in our research. Bubble blowing, open-mouth behavior, and body spins occurred in both the stranger encounter and both television modes. During the stranger encounter, all four dolphins were fixated at the gate, and other than for a second or two did not leave it. In mirror mode television, dolphins' visits are brief, and attention span is on the order of seconds or minutes rather than hours. Thus, if we postulate that the dolphin perceives its mirror mode image to be a stranger, the amount of attention it gives the stranger on the television bears little resemblance to the amount of attention it pays to a real stranger. And finally, although open-mouth behavior occurred in both the stranger encounter and in mirror mode, open-mouth behavior in mirror mode is often accompanied by head movements, either side-to-side, up-and-down, or in circles; these were not noticeable in the stranger encounter.

Television Mark Tests


The concept of a mirror mark test was combined with the television tests described above. If the marked dolphin is interested in the mark and aware of his relation to the image on the television, he should use mirror mode to observe the mark, not playback mode. Keola and Hot Rod had three mark tests of this type; we present data on one of Keola's tests, since his positive mirror mark test made him an interesting subject for comparing television results with mirror results. The time samples in this test are too small for statistical tests, so the results must be treated as tentative suggestions rather than conclusions.
In addition to his mirror mode image on television, a .8 m by .7 m size rectangular mirror was also available 3/4 m to his left. The mirror was a reflective, aluminum-coated polyester mylar film made by DuPont, which we applied to the window with soap and water. Keola was marked on the head with zinc oxide, 2-3 inches behind his right eye (Fig. 3B). The video setup was identical to that in the Mirror Mode/Playback Tests described above. The self-view camera was beside the television on the dolphins' left (Fig. 1).
For analysis, the questions we tried to answer were: "Does the dolphin increase the television viewing time of his marked area once he is marked? Does he favor it more in mirror mode compared to playback mode?". We tallied the amount of time Keola had his left and right sides facing the self-view camera, and thus visible to him on the television, during three 10 minute segments: 10 minutes of control data without a mark, and 10 minutes each of mirror and playback modes with a mark on his right side. For analysis of the unmarked (control) segment we chose a 10 minute sample with a high incidence of television interaction to give us a lot of data.
In all the television experiments presented here the location of the camera to the dolphins' left means that if the dolphins do not favor showing a side, the left side will automatically show more on the television - and this is reflected in the data below. The dolphins were marked on their right side specifically to see if they would counter this bias by maneuvering the marked side into view.

Television Mark Test Results

In the control data, Keola spent eight of the 10 minutes directly in front of the television looking at himself. During the first mirror/playback cycle after his mark was applied Keola was directly in front of the television 2.43 minutes out of 10 minutes in mirror mode, and 2.30 minutes out of 10 minutes in playback mode. In the 10 minutes of unmarked control data, the left (unmarked) side was "favored" 45% of the time and the right (marked) side 10% - the bias expected with the camera on the dolphins' left. In playback mode the left side was favored 68% of the time, the right side 8% - again as expected with the location of the camera. In mirror mode, however, although the left side was favored 51%, the right side increased to 25%. Exposure of the marked side in mirror mode (Fig. 3B) was thus three times what it was in playback mode and 2 1/2 times what it was in the unmarked control.

Television Mark Test Discussion

The mark on Keola's right side increased Keola's viewing time of that side. It did not affect how long Keola spent at the monitor in mirror mode compared to playback mode, but only his orientation when he was there (Fig. 3B), tripling viewing of the right side. In a television mark test not presented here, adult male Kamalii (marked with zinc oxide on his side) repeatedly pivoted his body back and forth while watching his image on the television. He appeared to be comparing his marked and unmarked sides. He did five cycles (10 full lateral presentations) in 20 seconds. He never did this unmarked. The tests with Keola involve small time samples, and though the results are consistent with self-examination, they fall far short of proving it. Future research will require larger samples and even new methods to be conclusive.

Turn Tests


Video equipment.
We used two Minolta S-VHS Series V-2000 video camera with wide angle lenses. The front view camera was located to the dolphins' left of the television. The side view camera was located in the bubble window to the dolphins' right. The television was a 20" Sony Trinitron Color television, set back roughly 8 cm from the window.
Sample. Babies Maui and Tinkerbell were given a frontal, real-time self-view on television (Fig. 3D), but while they were watching it we switched what was on the television to a real-time side view of themselves (Fig. 3E). We then waited for approximately five to ten seconds to see if the dolphin turned (Fig. 3F). Maui and Tinkerbell were either alone or with Maui's mother Puna during these experiments. Session 1 of these turning trials occurred in June 1991, session 2: three months later, session 3: two weeks later yet, and session 4: a month later.
Control data: counting turns in mirror mode without switching views. The baseline data to which we compared turn test results was collected by counting turns made by the subjects watching television in mirror mode, when no switching of view occurred. Control data was collected only when the dolphin was in front of the television looking at itself (as in Fig. 3D). For Maui and Tinkerbell, we examined a total of 100 minutes of mirror mode videotape, chosen from three days of mirror mode experiments. The first set was from five days before their first turn test, the second from two days before their second turn test, and the third from the day of their second turn test (prior to it).

Turn Test Results

In the control data Maui turned to the right once in a cumulative total of 2.08 minutes of close viewing (when he was in front of the television, looking at himself); Tinkerbell never turned in 1.6 minutes.
In test session 1, no turns occurred in eight trials. In session 2, Maui turned in two trials out of eight. The first time he turned away from the side-view camera, then toward it. The second time he turned twice toward the camera, then away from it. In session 3, Maui turned once and Tinkerbell twice in 13 trials, all toward the side-view camera. In session 4, Maui turned in six trials and Tinkerbell in three trials, out of 26. In the first turn of this session, Tinkerbell turned away from the camera. All subsequent turns by both dolphins were towards the camera. In one of his turns, Maui turned toward the camera twice.

Turn Test Discussion

The control data suggest that turns occur at a rate of approximately once per two minutes of close viewing in Maui, and infrequently if at all in Tinkerbell. If a trial lasts an average of approximately 5 seconds, we might thus expect Maui to turn "randomly" once per 24 trials, and Tinkerbell not at all. (These estimates are rough, as the actual amount of time we allowed for a turn to occur varied slightly from trial to trial.) Instead, after the first session when they did not seem to react to the switching, Maui turned nine times in 47 trials and Tinkerbell, in sessions three and four, turned five times in 39 trials. For Maui the turn rate in the tests was about five times what it was in the control data. Tinkerbell never turned in the control data, making her six turns during testing all the more significant.

General Discussion

Interpretation of Combined Results From Mirror Mark and Television Tests

The most important conclusions we can make come from combining the results of the mirror mark tests (Marten and Psarakos, 1994) with the television tests presented here. The television mark test and turn test, although interesting, can only be taken to be suggestive.
Keola, the only adult to have both mirror mark tests and television tests, behaved significantly differently in mirror mode as compared to playback mode (Figs. 2 and 4). His behavior strongly suggests that he did not perceive the television image as another dolphin in either mode and was not interacting with it "socially". When he was marked with zinc oxide, he almost immediately positioned himself within a foot of the mirror so that the mark could have been visible to him and engaged in rapid, extreme, postures for 13 seconds. It is very likely that he was examining his mark. The only reasonable alternative, that he might have been posturing socially to what he believed to be another dolphin in the mirror, is rendered not very likely by the television tests.
The juvenile Hot Rod had both mirror mark and television tests. His behavior during his mirror mark tests is hard to interpret. During his first mark test he remained relatively motionless with his marked side facing the mirror for 10 seconds. This lack of motion in front of the mirror is unusual for Hot Rod, whose name derives from his kinetic energy. In his television tests, we could not tell if he treated mirror mode differently than playback mode - he acted interested and excited in both television modes with behaviors like bubble blowing, head and mouth movements, and body spins. He did not, however, act agitated like we would expect him to in the presence of a strange dolphin, or give prolonged, undivided attention as he did during the control test with the false killer whales. Some behavioral differences between mirror mode and playback mode are, however, worth noting for Hot Rod (Fig. 4). He did "Body spins with closed mouth" in mirror mode, but never in playback mode. "Bubbles heard" were greater in mirror mode than in playback mode for Hot Rod, suggesting increased excitement and interest in mirror mode. These interested, but hard to interpret behaviors are similar to the results obtained by Marino et al. (in press) in similar tests with two juvenile males of approximately the same age (7 years old).

Self-View Television Behavior Compared To Encounters With Real Strangers.

The most marked differences between self-view television behavior and known social behavior are attention and intensity. Open-mouth and head movements are slow, relaxed, and rhythmic in front of the television, but rapid and agitated when interacting with a live, unknown dolphin. Some of the television-directed open-mouth behaviors exhibited during mirror mode are unique and do not occur in other contexts, such as Hot Rod's presentation of his fish and Kamalii's tongue curling (Fig. 3G and below). In general, the dolphin attends to a real stranger close to 100% of the test time, but less than 1% of the time when viewing itself on real-time self-view television. Behavior in front of the self-view television is usually calm and involves exploring the television image. Head movements and open-mouth behavior appear to be "contingency-checking" (deliberate movements to test effects on a contingent image such as a mirror reflection or real-time television image). In contrast, behavior with live, unknown dolphins is hyperactive and open mouth behavior is usually frenzied biting. Since television-directed behavior is generally different from social behavior with real dolphins, and includes what appears to be contingency-checking behavior, we conclude that it is unlikely the dolphins are mistaking their television images for other dolphins. This suggests that they recognize the images as themselves.

Apparent Social Behavior with Self-View Television.

Of the twelve dolphins exposed to self-view television, only one treated his television image the same way he interacted with real dolphins. This was Kamalii, who engaged in intense social behavior in the underwater barred gate control test (Fig. 3A). He had been with adult male dolphins for most of his 18 years at Sea Life Park. Two days after the control test he was exposed to a mirror for two hours. Seven weeks later he had self-view television for an hour and a half. He acted calmly with the mirror, apparently playing with bubbles and body motions (Marten and Psarakos, 1994). He was calm also with the self-view television, coming up with a new behavior: opening his mouth and curling his tongue (Fig. 3G).
After he had been established with his new female companions for six months (with no more television and no more mirror exposure in the meantime), he had 15 more self-view television sessions: 22 hours spread over 7 weeks. In the first session his reaction was violently aggressive: he rammed the television window with his head and engaged in intense, aggressive open-mouth behavior and vocalizations, with his body oriented perpendicular to the window. Over subsequent sessions, starting with session four, this apparent social behavior gradually changed to what appeared to be the most elaborate contingency-checking we've seen: twitching, undulating his body fat, spinning and jerking his body, blowing bubbles and biting them, regurgitating fish, and moving his tongue outside his mouth. Kamalii's apparently aggressive behavior with the television alerts us to the fact that the results of self-recognition tests can depend not only on the individual's mirror experience, but also on internal state, and/or recent social history.

Self-View Television Compared To A Mirror

All twelve of the dolphins we exposed to the mirror spent more time in front of it than the self-view television. However rhythmic head and body movements were more common and exaggerated in front of the television: more repetitive head and mouth movements, greater excursions of movement, and more cycles. The smaller size of the television image may elicit more pronounced forms of contingency-checking than the mirror does. In Kamalii's case of apparent social behavior with the television (but not the mirror), the small size of the image may have been recognizable as a dolphin but not as himself, and thus instigated a social reaction. The lack of social behavior by all other dolphins tested in front of both mirror and self-view television is in striking contrast to apes, who have an initial period of social behavior. It is likely that the reflections from underwater windows in the dolphins' tanks may provide them with enough exposure that they are not naive to their mirror reflections.
Development of contingency play. Early development of contingency recognition is another similarity between dolphins and humans. The earliest age we saw contingency play was 1 1/2 years, in Maui, the youngest individual we observed. The dolphin babies' greater interest in mirror mode television over video playback of themselves and others, however, is different from human babies, who give greater attention to non-contingent playback of others over either contingent or non-contingent television of self (Lewis and Brooks-Gunn, 1979). Human babies also commonly play "peek-a-boo" games with contingent television (getting in and out of range of the television camera), which did not occur with the dolphins. The other most common human baby behavior with contingent television is making faces -- analogous, in our opinion, to dolphin head and mouth movements. Overall, the results from the dolphins are suggestive of self-recognition in adults (but do not prove it), but are mixed for the babies.

Window Reflections As Possible Mirror-Like Experience, And Control Tests

Our mirror mark test results indicate that the dolphins are probably not naive about their reflection. Dolphins in tanks with windows are most likely already experienced with their reflections in the windows. We have documented this by photographing our reflections in the windows of the tank. There is a reflection of varied quality depending on time of day, although never as good as a mirror. To the extent that a dolphin is familiar with its reflection from the windows, its image in a mirror or on television is not entirely novel. If this is the case, then perhaps a control test for television or mirror testing of self-awareness should include recording the dolphins' response across a barrier to a known dolphin. There is no formal data on this topic, however when dolphins at Sea Life Park are separated by a barred gate after shows, they often become more aggressive through the gate than when they were together. In addition, in our control test with two adult males (Keola and Maka), Keola also interacted with familiar dolphins through an underwater barred gate. His interactions with them were frenzied and aggressive. This type of behavior contrasts sharply with his behavior during mirror mode television, or in front of a mirror.

The Role Of Perception, Motion, Vision, and Sound In Self-Recognition

The television tests are based on the premise that dolphins perceive television like people do: as a representation of reality. Herman et al. (1990) found that dolphins responded to artificial language command gestures on an 11.5 inch television. To test if dolphins treat images on their television as a representation of reality in the absence of food reward, we played some of the dolphins a videotape of their trainer feeding them. Keola, Hot Rod, and to a lesser extent Tinkerbell, appeared to try to "catch" the television fish by opening their mouths each time a fish was thrown. After doing this for about a minute they left the monitor and went to the small tank (where they were accustomed to being fed). In one instance when Hot Rod was watching the video it came to an end with the trainer packing up her fish buckets and leaving; Hot Rod rushed to the small tank. These results demonstrate that television may represent reality for dolphins.
Motion. Dolphins are predators and are attracted by motion. They often ignore something until it moves. The dolphins' first reaction to mirror mode is to move, or contingency-check. Primate researchers seem to think that their subject recognizes itself visually. In dolphins the role of movement could outweigh visual form, and contingency-checking may be necessary for self recognition.
Vision. Dolphins have good visual acuity (Herman et al., 1975). (See Dawson, 1980, Nachtigal, 1986, and Madsen and Herman, 1980 for reviews.) Herman et al. (1990) found that dolphins responded to human artificial language gestures, and even degraded versions of these, on an 11.5 inch television placed 40 cm from their underwater window. As for visual field, the dolphin's eyes are more laterally placed and directed than in primates, so sometimes the dolphin is seeing its target with one eye. It is possible that side-to-side head movements are actually examination of the target with alternating left-right visual fields, as Povinelli found in elephants (personal communication), rather than contingency-checking.
Sound. The dolphin's primary sense for complex processing is sound, and this should not be overlooked when exploring self-recognition in these animals. Since echolocation clicks are reflected efficiently by the air side of the underwater window, according to the dolphin's sonar there is no dolphin where the television is. However, the reflected sound could be interpreted by the dolphin as a vocalization coming from a dolphin located behind the window. Furthermore, sonar clicks, after they have reflected off the window, could reflect off the dolphin, back to the window, and then back to the dolphin, containing a sonar image of itself. Gish (1978), working with bottlenose dolphins in two tanks with an acoustic link, found similar behavior in the acoustic mode to what we have found in the visual mode: prolonged aggressive interchanges through a "fence", mimicry, and recognition of the difference between echo of self and playback.


Self-view television behavior differs from real social behavior (controls) and from reaction to playback of self, and is characterized by what appears to be contingency-checking behavior. These findings, plus the results of the mark and turn tests, are consistent with self-examination and not social behavior. The material presented here, especially when combined with the highly suggestive results from the mirror mark tests and their controls (Marten and Psarakos, 1994), makes a significant case for self-awareness in this species.


This project is a cooperative effort between Earthtrust and Sea Life Park. It was conceived by Don White and Dexter Cate. Don White's participation was valuable at all stages. We thank Sea Life Park curator Marlee Breese and trainers Stephanie Vlachos, Susan Rodgers, Roberta Horne, Keana Pugh, Rich Nunes, and Carol Chang. We received help and valuable advice from Lori Marino and Gordon Gallup. Don Griffin suggested the turn experiment. Dave Hack provided the tank drawing. Oceanic Institute provided electricity. The research was supported by Mollie Malone, Wendy Grace, the Boudjakdji Family, the Sarah Stewart Foundation, the Barbara Gauntlett Foundation, Outrigger Hotels.


Anderson, J. R. (1986). Mirror-mediated finding of hidden food by monkeys (Macaca tonkeana and M. fascicularis). Journal of Comparative Psychology, 100, 237-242.

Caldwell, M. C., & Caldwell, D. K. (1965). Individualized whistle contours in bottlenosed dolphins (Tursiops truncatus). Nature, 207, 434-435.

Dawson, W. W. (1980). The cetacean eye. In L. M. Herman (Ed.), Cetacean Behavior: Mechanisms and Functions (pp. 53-100). New York: John Wiley & Sons.

Gallup, G. G., Jr. (1970). Chimpanzees: self-recognition. Science, 167, 86-87.

Gallup, G. G., Jr. (1979). Self-awareness in primates. American Scientist, 67, 417-421.

Gallup, G. G., Jr. (1982). Self-awareness and the emergence of mind in primates. American Journal of Primatology, 2, 237- 248.

Gallup, G. G., Jr. (1987). Self-Awareness. In G. Mitchell & J. Erwin (Eds.), Comparative Primate Biology: Vol. 2B. Behavior, Cognition, and Motivation (pp. 3-16). New York: Alan R. Liss, Inc.

Gish, S. L. (1979). Quantitative analysis of two-way acoustic communication between captive Atlantic bottlenose dolphins (Tursiops truncatus Montague). Unpublished doctoral dissertation, University of California, Santa Cruz.

Herman, L. M., Peacock, M. F., Yunker, M. P., & Madsen, C. J. (1975). Bottlenosed dolphin: Double-slit pupil yields equivalent aerial and underwater diurnal acuity. Science, 189, 650-652.

Herman, L. M., Morrel-Samuels, P., & Pack, A. A. (1990). Bottlenosed dolphin and human recognition of veridical and degraded video displays of an artificial gestural language. Journal of Experimental Psychology, 119(2), 215-230.

Lethmate, J., & Ducker, G. (1973). Untersuchungen zum selbsterkennen in spiegel bei orang-utans and einigen anderen affenarten. Zeitschriftfur Tierpsychologie, 33, 248-269.

Lewis, M., & Brooks-Gunn, J. (1979). Social Cognition and the Acquisition of Self. New York: Plenum Press.

Madsen, C. J., & Herman, L. M. (1980). Social and ecological correlates of cetacean vision and visual appearance. In L. M. Herman (Ed.), Cetacean Behavior: Mechanisms and Functions (pp. 101-147). New York: John Wiley & Sons.

Marino, L., Reiss, D., & Gallup, G. G. (1994). Mirror self- recognition in bottlenose dolphins: Implications for comparative study of highly dissimilar species. In S. Parker, M. Boccia, & R. Mitchell Eds.), Self-Awareness in Animals and Humans: Developmental Perspectives (pp. 380-3 91). New York: Cambridge University Press.

Marten, K., & Psarakos, S. (1994). Evidence of self-awareness in the bottlenose dolphin (Tursiops truncatus). In S. Parker, M. Boccia, & R. Mitchell (Eds.), Self-Awareness in Animals and Humans: Developmental Perspectives (pp. 361-379). New York: Cambridge University Press.

Nachtigall, P. E. (1986). Vision, audition, and chemoreception in dolphins and other marine mammals. In R. J. Schusterman, J. A. Thomas, & F. G. Wood (Eds.), Dolphin Cognition and Behavior: A Comparative Approach (pp. 79-113). New Jersey: Lawrence Erlbaum Associates, Inc.

Norris, K. S. (1991). Dolphin days: life and times of the spinner dolphin. New York: W.W. Norton & Co.

Ostman, J. (1985). An ethnogram for dolphin social behavior, and observations on changes in aggressive and homosexual behavior among two subadult male bottlenose dolphins (Tursiops truncatus) in a captive colony. Unpublished masters thesis, San Francisco State University.

Ostman, J. (1991). Changes in aggressive and sexual behavior between two male bottlenose dolphins (Tursiops truncatus) in a captive colony. In K. Pryor & K. S. Norris (Eds.), Dolphin Societies: Discoveries and Puzzles (pp. 305-317). Berkeley: University of California Press.

Overstrom, N. (1983). Association between burst-pulse sounds and aggressive behavior in captive Atlantic bottlenose dolphins (Tursiops truncatus). Zoo Biology, 2, 93-103.

Parker, S. T. (1989). Self-recognition by gorillas at the San Francisco Zoo: A comparative developmental model for sensorimotor self-recognition in primates. Paper presented at the meeting of the American Society of Primatologists.

Patterson, F. (1984). Self-recognition by Gorilla gorilla gorilla. Journal of the Gorilla Foundation, 7(2), 2-3.

Patterson, F. (1991). Self-awareness in the gorilla Koko. Journal of the Gorilla Foundation, 14(2), 2-5.

Patterson, F., & Cohn, R. (in press). Self-recognition and self- awareness in the lowland gorilla. In S. Parker, M. Boccia, & R. Mitchell (Eds.), Self-Awareness in Animals and Humans: Developmental Perspectives. New York: Cambridge University Press.

Pryor, K. (1975). Lads before the wind: adventures in porpoise training. New York: Harper and Row.

Suarez, S. D., & Gallup, G. G., Jr. (1981). Self-recognition in chimpanzees and orangutans, but not gorillas. Journal of Human Evolution, 10, 173-188.

Received May 7, 1993

25 Kaneohe Bay Drive, Suite 205
Kailua, HI 96734 USA
(808) 254-2866
FAX: (815) 333-1158
E-mail: EarthTrust

Return to Project Delphis page