The Dolphin in the Mirror

by Diana Reiss

  We thought it was a beautiful experiment and were quite proud of it. But then we realized it wasn't infallible as a demonstration of mirror self-recognition. It would be hard to explain away why the animal would go to the mirror after the suction cup was positioned on his head, but a skeptic could reasonably argue that the dolphin had learned: If I go to the mirror, I will see a suction cup, white or black, and then I will pull the appropriate lever. That line of reasoning did not necessarily imply self-recognition; it might mean that the individual was merely following a set of simple rules: see a white suction cup, pull the white lever; see a black one, pull the black lever. The dolphin in the mirror could easily be interpreted by the dolphin as a separate animal, and the lever-pulling would still be the same. So we had to give up on that. It would have been an easier experiment to run than the mark test, but, alas, it wouldn't do. So it was back to the mark test for us, with all its difficulties for no-handed subjects.

  ***

  We embarked on the mirror self-recognition tests in January 1998, setting the bar for success high. First, we predicted that if the dolphin knew it was his face in the mirror, he would not react as if he were seeing another dolphin—he would not show social behaviors. We also predicted that when marked after a feeding session, the dolphin would move to the mirror more quickly to check himself out than when he was not marked after a feeding session. But that alone was insufficient. The dolphin would also have to show himself highly motivated to inspect that part of his body that had been marked, clearly orienting that part of the body to the mirror as soon as he arrived.

  We planned to videotape all the sessions and then score fifteen-minute segments comparing the behaviors the dolphins exhibited in the pre-mark and post-mark conditions. Four independent scorers in two independent pairs would rate the behaviors, second by second, as we watched the videotape segments. We coded behaviors in several different ways. Mark-directed meant that the dolphin positioned himself directly in front of the reflective surface and then oriented the body so that the mark could be seen in the reflection. Self-directed meant the dolphin was inspecting parts of his own body or watching his own behavior in the mirror (for example, close-eye viewing, looking inside his wide-open mouth, repetitive head circling, viewing his own genitals). Social meant the kind of behavior these dolphins would display in the presence of another familiar or unfamiliar dolphin, such as aggressive jaw-clapping and charging, or affiliative responses and vocalizations (whistling and squawking). Finally there was exploratory, meaning that the animal seemed to be inspecting the mirror itself, looking behind it or over it, or pushing on it.

  As raters, we were unaware of whether each dolphin had been marked in any particular video segment being observed. We were "blind" to the condition we were rating. And we insisted that a particular instance of a behavior be included in further analysis if and only if there was 100 percent agreement between the two rating teams as to the behavior's classification. The coding was actually somewhat more complex than I've outlined, and it included not only the specific behavior at hand but also the time of onset of the behavior, time of termination, orientation of the body, and several other measures.

  We also included several controls in the study. First, before we actually put a mark on either animal, we did sham marking. This involved taking the nontoxic dye out of the marker, replacing it with water, and then touching the dolphin on a part of his body he couldn't normally see as if we had truly marked him. We also planned to do the sham-marking procedure after the animals had experienced true marking. We hadn't done these controls in the earlier study at Marine World, nor had we been anywhere near as meticulous in analyzing post-marking behaviors.

  The excruciating experimental details I've presented here are normally the stuff found in the methods section in academic papers, not in books for general readers. But our hypercompulsive and demanding experimental criteria were necessary because of what we were up against. If we announced that dolphins had the same cognitive capacities in the realm of awareness that humans and great apes and no other minds in nature had, then we had to back that up with iron-clad evidence. If we claimed that the primate-centered view of the origins of self-awareness in humans was incorrect, we had to have incontrovertible results. Extraordinary claims such as this one required extraordinary evidence. Gordon Gallup often says of his original work with the chimps that he felt he had demonstrated mirror self-recognition in the self-directed behaviors he observed, but he needed some more objective measure. That was the mark test. Lori and I felt the same about our study, but we had to work extra carefully with our non-handed dolphins.

  As part of our extreme caution we decided to use only Presley as our principal experimental subject for the paper we planned to write. One of Tab's eyes was slightly cloudy. Lori and I talked about it and agreed that it would be prudent not to include him in the full mark test. We reasoned that Tab might fail the mark test because his vision might be impaired and not necessarily because he didn't recognize himself in the mirror. It was safer to leave him out, we decided. It later proved to be an unfortunate choice.

  We carried out the first phase of the study, which involved getting base-line data on the dolphins' normal behavior in the rectangular indoor pool, before the mirror was installed. Then we exposed the dolphins to a very small, three-and-a-half-by-one-and-a-half-foot mirror (much smaller than the mirror described earlier) affixed vertically on the exterior wall of the pool, facing them. When the mirror was present, which was about half the time, the dolphins were definitely interested in it and in what it revealed. After a few initial bouts of touching the mirror, they began to position their bodies vertically, tightly aligned within the mirror's boundaries, as if fitting themselves into the reflective frame. Then each dolphin, alone at the mirror, began to do some very different, non-dolphinlike behaviors that were strikingly reminiscent of the antics of Groucho and Harpo Marx in the famous mirror scene in Duck Soup—except those comedians were crafting pretend reflections, with the straight man trying to catch the "reflection" in a mistake.

  Like the Marx Brothers, the dolphins displayed a rich repertoire of some pretty bizarre orientations and highly repetitive movements—but only when they faced the mirror. For example, in calm silence, one of them would hang vertically within the narrow sliver of the mirror and repeatedly rock his body from side to side. Or he would move his head in wide arcs and circular movements close to the mirror, repeatedly and exaggeratedly nodding from left to right or up and down. Each would hang upside down at the mirror and peer into his wide-open mouth. The dolphins would do sequences of repeated flips. Like Groucho, and like our close ape cousins, the dolphins appeared to be testing the contingencies of their own behavior at the mirror. What happens when the dolphin circles his head? What does the other dolphin do? In mirror self-recognition studies, this is called the contingency-testing phase, a critical hurdle to pass on the way to self-recognition. It does not guarantee the next step of self-recognition, but it seems to be a necessary step for individuals who make that leap.

  Presley and Tab did reach that milestone. They went on to show clear signs of self-directed behavior. For example, they would look at the insides of their mouths, often holding their mouths open wide in prolonged gapes; they gazed at their eyes, holding their eyes right near the mirror surface; they blew varieties of bubbles at the mirror; they twisted and turned their bodies in bizarre postures (unlike their normal behavior in the absence of the mirror); and they brought their toys to the mirror and engaged in toy play there. The self-directed behavior at the mirror was a clear indicator that the dolphin understood that the external image in front of him, that dolphin in the mirror, was himself. In a young child, self-directed behavior is often reported as close inspection of eyes, mouth, genitals, or other body parts that go unseen without a mirror. Children might pick their noses, eat, or play in front of a mirror. If you mark each child in a group of nineteen-month-olds with a small spot of rouge on their foreheads or elsewhere on the fac
e, about half of them will be able to look in the mirror and then point to the location of the spot on his or her body. Not all children of that age pass the mark test, although many show self-directed behavior, which is evidence in itself that they recognize the faces in the mirror are their own.

  Both Presley and Tab were well used to being handled, tickled, and rubbed, and they often solicited it. They didn't seem to mind being marked when we did the initial sham marks. We didn't expect them to race to the mirror or any other reflective surface immediately afterward, and they didn't. We figured they would just think we were tickling or rubbing them. When the mirror was present, which was about half the time, they were definitely interested in it, observing themselves in various ways, but they made no attempt to examine the part of the body we had touched with the water-filled marker.

  By October we felt we were ready to begin the true mark section of the study with Presley. The first time we marked him was quite revealing. As in the previous sessions, the dolphins were exposed to the mirror for thirty minutes, then called to their feeding areas to be fed and marked. We then gave them the release signal signifying the end of the feeding sessions, letting the dolphins know they could leave the area. This time, we carefully made a triangular mark on the side of Presley's head using a black, nontoxic marker. When we gave the release signal, he left station and swam around the pool playfully, as usual. When at last he swam by the mirror, however, he did a classic double take. He immediately turned around, went back to the mirror, and proceeded quite deliberately to orient himself in a way that would make the mark quite visible to him. We were elated.

  We piled up statistically significant numbers of trials, and the results were outstanding, even more clear-cut than either Lori or I had dared hope. We were able to show that when marked, Presley swam right to the mirror and positioned his body immediately to expose the marked area to the mirror. We marked him on different parts of his body that he could not see without the mirror, and he always positioned himself accordingly. Orienting to the mark in his first moments at the mirror was strong evidence that he was using the mirror as a tool to view his mark. Presley spent substantially longer time in this behavior when he was marked than when he was not marked, more than twenty times longer. When he was sham-marked, he would swim over and inspect himself, but with nothing unusual to see, he would swim away much sooner. We had many qualitative observations too that made the same point: these animals knew who they were seeing in the mirror.

  The dolphins' interest in the marks were never casual. They scrutinized their marks from many different angles, often for up to a minute. Presley demonstrated keen curiosity about the mark. We did most of these sessions in the two outdoor joined pools; the marking station was at the far end of the larger pool, and the mirror was around the corner from the entrance to the smaller pool. After being marked, Presley had to swim the length of the large pool, enter the smaller pool, and take a sharp left turn in order to use the mirror. He routinely did this very rapidly, sometimes in fewer than ten seconds, and never longer than a minute and a half.

  In the summer, with the dolphins in their outdoor pools, we marked Presley under his chin on his neck area for the first time. We prepared to give the release signal, but before we could move an inch he darted away from his feeding station, made a fast beeline to the mirror in the adjoining pool, and assiduously examined the marked area. He arrived at the mirror so quickly that I could hardly keep him in my viewfinder as I videotaped his behavior. This time he oriented in a very different manner. He came close to the mirror and stretched his neck up repeatedly, lifting his head, exposing the marked area, and looking into the mirror. This self-examination went on for several minutes. Presley seemed very calm and very interested. Then he slowly backed away from the mirror and began to do the spin dance that I described earlier. If you want to see your entire body in a small mirror, you have to move away from it. He understood the rules. His behavior provided strong evidence that he was aware that what he saw before him was an external representation of himself.

  We finally moved into what we called the late sham-marking phase, in which we again went through the motions of marking but with a water-filled marker. Presley had learned that after being touched by the marker, he was marked. As we expected, he raced to the mirror to inspect the spot. But he lost interest soon after seeing his usual, unmarked image. Presley spent about a fifth of the time in self-directed behavior inspecting himself when the mark was false and invisible.

  On one occasion, though, Presley seemed determined to find a mark, even though there wasn't one. We sham-marked him on his right pectoral fin while he was in the rectangular pool and there was no mirror present. He immediately went to a corner of the pool that had the most reflectivity (due to a dark wall directly behind it) and spent more than half a minute in a sequence of a dozen dorsal-to-ventral flips in that corner area, each time bringing the sham-marked spot into close view. You could almost hear him thinking, I'm sure they put a mark somewhere here.

  ***

  I had seen several videos of chimpanzees undergoing the mirror self-recognition test and read many accounts of them, and it was obvious that Presley was at least as motivated as the chimps, if not more, to examine the marks in detail. The chimps usually touched the mark a few times, sometimes sniffing or tasting their fingers, but they fairly quickly lost interest. Not Presley, who was a mark self-examiner par excellence. Another difference between chimps and Presley was in the response to another individual's mark. While one chimp usually expressed interest in another individual's mark, neither Presley nor Tab paid any attention to the mark on his companion. Dolphins do not groom one another, as chimps do, so perhaps they have less interest in changes to another's appearance.

  It's fair to say that anyone who observed Presley and Tab during this study would have been very quickly convinced that these animals were self-aware. I know that when I give public talks and show the videos of the dolphins after they've been marked, the audience sees what we saw. The tapes are compelling evidence that dolphins know there is a "me" there. But we needed quantitative evidence in support of that blindingly obvious qualitative conclusion. We probably could have published our results more than a year before we actually did. We knew we had it, but we also knew we had to convince everybody. So our two teams of raters spent more than a year grinding through the procedures I just described: detailed analysis of second-by-second behaviors, coding them by parallel independent teams. Not to put too fine a point on it, it was grueling, but there was no escaping it.

  During the spring of 2000, Lori and I went through several drafts of the paper describing our results, and we sent copies to prominent colleagues in the field. We received very helpful comments and a lot of encouragement. We finally submitted what we felt was a finely honed manuscript to the journal Nature early in May. The British journal Nature is one of the most prestigious of scientific journals, and it prides itself on publishing breakthrough research across the spectrum of scientific disciplines. That was our rationale for sending it there: we felt that our work was a really big breakthrough. "This is a pivotal study that clearly demonstrates that the emergence of [self-awareness] is not a byproduct of factors specific to primates," we wrote in our cover letter, addressing the assumption that had been supported by a prodigious literature during the previous three decades. "These findings further advance our understanding of factors that may contribute to the emergence of advanced cognitive abilities in diverse species as well as providing us with a greater appreciation of the cognitive capacities of the [big-brained] dolphin."

  We had high hopes for both a quick turnaround of the manuscript and its acceptance. We were right only about the speed. Within a couple of weeks I received a big brown envelope with Nature's logo on it containing a firm letter of rejection and copies of reviews from four anonymous referees. We were completely deflated.

  There's a joke among academic scientists about "the third reviewer," one that recently made the rounds on
YouTube. Essentially, the idea is that you get two terrific reviews of your baby—the submitted manuscript—but then a third reviewer trashes your paper, and you have to rework it or do additional experiments. Most journals do indeed send manuscripts out to three reviewers. In our case, Nature sent it to four, and it was the fourth reviewer who killed us.

  The first three reviewers were very positive, also offering helpful comments for improving the accessibility of the paper. One of the reviewers had scrutinized our statistics. This person judged them unexceptional, which is statistics-speak for "well within the norm." That referee also said that the effects we were testing with the statistics were so clear as to almost obviate the need for statistical analysis. Two of the referees commented on our small sample size of one—we had included only one individual, Presley. But researchers familiar with the field of animal cognition know that there is a long tradition of single-subject studies, such as those by Irene Pepperberg (with Alex, an African Grey parrot), Herbert Terrace, Allen and Beatrix Gardner, Duane Rumbaugh and Sue Savage Rumbaugh (all with chimpanzees), and many others. In scientific epistemology, the demonstration of a particular capacity in an individual of a species not previously known to possess that capacity is known as existence proof. It is the first step down the path of a particular discovery, and in our case we were confident that other bottlenose dolphins would eventually pass the mark test, just as we believed Presley had.

  And then there was the fourth referee. This individual's tone was completely different, and the message was unequivocal: our paper should not appear in any scientific journal. The referee's criticism of our statistics was simply incorrect, but he or she had objections to other aspects of our study that struck us as emotional and knee-jerk reactions, clearly outside the norm of a professional scientific review. Lori and I were so convinced of the force of our scientific case that we took the unusual step of calling the editor at Nature to find out what we might do to get the decision reversed. He was very polite, very nice (he was British, after all), but equally firm. No, he could not publish the manuscript as it was. Perhaps if we had data from a second dolphin, he offered, he might be prevailed upon to reconsider.