In spite of its intuitive appeal and widespread use, the interpretation of this task has been controversial. After all, in some sense many animals recognize themselves. Most animals, except for some dogs, do not chase their own tails. Various animals adjust their appearance to camouflage themselves within the environment. Bobtail squid even generate light from their underside, effectively canceling out their own shadow. Many mammals, including dogs, use urine to mark their territory and must be able to distinguish their own smell from that of others. So why do most animals fail to pass the mirror test, and what does it mean that great apes and humans can pass it?
As usual, there are advocates of rich and of lean interpretations. On the richer side, Gallup argues that to pass the task one has to become the object of one’s own attention: one has to be self-aware. The term “self-awareness” refers to much more than knowing what one looks like (even for people who are very concerned about their appearance). It implies awareness of where you come from and where you are going, what you are good and bad at, your personality, your likes and dislikes, your values, and so on and so forth. Can we infer this knowledge from the mirror test? Gallup suggests that we can and argues that passing the test indicates a capacity for inner reflection, thinking about self, and even awareness of one’s own mortality. And in fact, children’s passing of the task has been found to be associated with the emergence of self-conscious emotions, such as embarrassment, as well as with the use of personal pronouns. A killjoy skeptic, however, might wonder if any form of mental reflection, other than as a play on the word “reflection,” is necessarily involved. Can you use a mirror to shave or clean yourself without engaging in deep inner reflection?
Then there are lean accounts. The comparative psychologist Celia Heyes is skeptical of rich interpretations and argues instead that all one needs to pass the task is an ability to distinguish feedback from other types of sensory input, and thus she questions if anything special is measured by the mirror self-recognition task. In her view, any animal that manages to avoid bumping into things, or that avoids biting itself in a fight, has demonstrated such ability. Heyes’s perspective, however, fails to explain why only a few species can pass the task (or why young infants fail it) when in other contexts they evidently distinguish feedback from other input.
Between these very lean and very rich proposals lie interpretations that try to claim the middle ground. For instance, cognitive psychologist Ulric Neisser argued that only when infants notice that their facial appearance matters to other people do they become interested in learning about their own faces through mirrors. The mark test may therefore indicate a change in children’s (and apes’) attention to faces.
The developmental psychologist Josef Perner posits that the test measures a more general ability to mentally juggle two different ideas about the same thing. Just as in pretend play, where children simultaneously hold in mind, and can compare, two distinct notions of the same object (e.g., a banana as a fruit and as a telephone), here they need to consider and compare their reflection and their expectation of what they look like to discover that something is new—such as that novel mark.
How could empirical tests distinguish between these interpretations? My colleagues and I developed versions of the mark test with children that have provided some clues. First of all we surreptitiously marked our participants’ legs, rather than their heads. The children sat in a highchair with a tray that prevented direct view of their legs. A mirror was placed in front such that they could see their own legs but not their upper body. Children performed exactly as they do in the classic task, reaching for the mark on their leg at the same age as they do for a mark on their face. This suggests that there is nothing specific to cognitions about faces, as Neisser suggested.
This set us up for the main manipulation. We sewed baggy tracksuit pants to the highchair and slipped a new group of participants in them without allowing them to see the pants directly, as the tray blocked their view. When children were then presented with the mirror and marked on the now tracksuited leg, most children failed. They did not seem to recognize that it was their own legs they saw in the mirror. However, when we allowed other children to directly see these baggy pants for just thirty seconds before affixing the tray to block their view, they performed very well—as well as children did in the first study or in the classic task. This suggests that young children have a mental expectation of what they look like and that this expectation is rapidly updatable. The thirty seconds of prior exposure was enough.
Thus it appears that the classic mirror mark test measures more than the lean accounts proposed,13 though it need not necessarily demand the higher cognitive capacities that the rich accounts conjecture.14 The results suggest that passing indicates subjects have a mental expectation of what they look like—they recognize themselves and examine violations of expectations, like that peculiar mark.
A moderate account, such as Perner’s view, appears to be most strongly supported by the current evidence. His explanation proposes nothing specific about the self but puts forth a more general capacity to mentally entertain and relate multiple models of the world, in this case the perception of one’s image and the expectation of what one ought to look like. As noted, toddlers begin to recognize themselves in mirrors around the same time as they begin to pass stage 6 object permanence tasks and to pretend play. Though on the surface these skills are dissimilar, all three involve a capacity to consider more than what is directly available to the senses: an expectation of what one ought to look like, inference of the transfer of a hidden target object, and the imaginary pretend identities of objects or actions.15 The comparative evidence suggests that great apes share this basic capacity to mentally go beyond the here and now, although the extent to which they can do this may well be limited in various ways.
AS THESE EXAMPLES ILLUSTRATE, SYSTEMATIC study can establish what animals can do. There will remain some uncertainty, and alternative explanations will be put forth, but with appropriate controls, replications, and careful comparisons we can become increasingly confident about the mental capacities of animals. Evidently, establishing what animals can do is difficult, but determining what they cannot do, what is uniquely human, can seem even more challenging. How can we be certain that a mental trait is not present in animals? When animal behaviors appear to display the trait in question, we face the difficulties associated with the aforementioned rich and lean interpretations. But even when there exist no obvious candidates for animal competence, we cannot simply conclude that the trait in question is uniquely human. Absence of evidence is not the same as evidence of absence. It remains possible, for instance, that we have not looked carefully enough.
When we say that no other animal can, for example, learn to play chess, we are making a claim known as a “universal negative.” In principle, it should only take a single conclusive case to the contrary, such as one octopus that can give you a halfway decent game of chess, to reject the claim. To absolutely prove the truth of the claim, on the other hand, you would in theory have to test every living animal to show that absolutely none of them can do it. Even then, you would have to presuppose that there exists a foolproof way of ascertaining that a particular animal does not have the capacity. After all, the octopus genius may, for instance, have been tired or otherwise not motivated when we tested it. Failure on a task is difficult to interpret in general. There are typically a number of reasons why the subject might not have performed well, aside from not having the capacity in question, and so negative results are seldom published. Should we give up and accept that we can never know for certain that some animal might not have the capacity after all?
The situation is not that dire. Though proving the absence of a characteristic is troublesome, we can, of course, draw conclusions after reasonable attempts. Once we have tested, say, thirty octopuses, and none of them can pass the task, it is reasonable to assume (for now) that other octopuses also cannot do it. In fact we agree on universal negative clai
ms regularly. Until proven otherwise, we are happy to say the dodo is extinct. Until proven otherwise, we might be equally happy to accept that only humans can learn to play chess. Although we cannot directly prove the absence of mental characteristics, we can become increasingly confident about such claims when people have tried to prove the idea wrong but failed. The more people have explored the world without coming across a living dodo, the more likely it is that they are extinct.
To bolster claims of absence, then, we need to give animals the opportunity to falsify the claim and demonstrate competence. Consider the question of mirror self-recognition again. Given that humans and great apes show competence, small apes, the next closely related primates, are of special interest for our understanding of the origin of this trait. Three previous small-scale studies on gibbon self-recognition had produced equivocal results. There was no clear evidence that they could recognize themselves, but the researchers remained open to the possibility. So Emma Collier-Baker and I set out to test more individuals, as well as more diverse small apes than the other studies.
Over the course of a two-year project at zoos in Australia and the United States we tested seventeen gibbons (seven siamangs, three dwarf gibbons, and seven crested gibbons). Each animal was first exposed to mirrors for over five hours. Next we conducted motivation checks in which we offered each gibbon some cake icing of a color that we later used to mark their face. All of them eagerly consumed the icing. We then surreptitiously smeared some icing on one of their limbs. When the subjects later discovered the icing on their arm or leg, they all immediately examined and then consumed every last bit of it. Gibbons generally engage in less self-grooming than great apes, but this shows that they are clearly motivated to retrieve icing from their bodies. They should therefore also be keen to inspect a mark of the same color as the icing on their head, if they could discover it via the mirror. Yet none of our subjects passed the mark test. Instead, most of them looked or reached behind the mirror, in what looks to most human observers to be a search for that apparent other gibbon with a mark on its head.
FIGURE 3.3.
White-cheeked crested gibbons inspecting a mirror (photo Emma Collier-Baker).
Because failure may be due to a variety of reasons, we conducted a battery of subsequent tests to see if the gibbons would pass eventually. We repeated the test, this time putting actual icing on their heads at the considerable risk of producing false positives—that is, the apes might have smelled the icing on their head rather than inferred the location using the mirror. We jumped up and down behind them to emphasize the nature of the mirror, marked them with large stickers, and so forth—all without success. Perhaps most tellingly, when we smeared icing on the mirror surface itself, the apes would scrape or lick off every last bit of icing; yet they ignored the big blob of icing on their own head that was clearly visible in the mirror. We thus concluded that we now had more than absence of evidence: these results amount to evidence of absence. Until there are new data to suggest otherwise, it is appropriate to conclude that small apes, like monkeys, do not recognize themselves in mirrors. Systematic research can inform us not only about competences but also about the limits of animal minds.
People are typically more excited about a finding that shows that an animal has succeeded in doing something we did not think it could do than about a report that notes that an animal could not do a task. It is also generally more difficult to get negative findings published than positive results (though it can be done). There are good reasons for this, given that failures are often difficult to interpret, as they may be caused by any number of factors. But we need to consider that a persistent flow of published positive findings and a lack of acknowledged negative results may seriously skew our perception of animal capacities.
BY ESTABLISHING THE PRESENCE AND absence of mental traits in various animals, we can create a better understanding of the evolution of mind. The distribution of a trait across related species can shed light on when and on what branch or branches of the family tree the trait is most likely to have evolved.
In general, species may share similar traits for two distinct reasons: convergent evolution (leading to analogous structures) and common descent (leading to homologous structures). The wings of birds and insects, for example, both solve the problem of flight. A closer look at the structure of these types of wings, however, shows that they are quite different: they are independent solutions to a similar adaptive problem. There is no reason to believe that these two types of wings evolved from one original body plan and that all the nonwinged animals between insects and birds on the tree of life somehow lost that ancestral trait. Such convergent evolution has occurred many times. The wings of bats, for instance, are analogous to those of both insects and birds. Cases of convergent evolution tell us about the selection pressures that may have driven the evolution of certain characteristics. On the other hand, homologous traits, traits shared because of common descent, tell us about their origins on the tree of life. Bird species have similar wings because of common descent. Some birds, such as penguins, have lost parts of the trait and cannot fly, and others, such as kiwis, have barely any wings at all, but they are all descended from a feathery winged ancestor. Biologists come to this conclusion by comparing the likelihood of potential explanations of the current distribution of the trait. Again, parsimony is important: the simplest explanation, the one that requires the least number of assumptions, is the most likely explanation. It is less parsimonious to propose that each bird species independently invented a feathery solution to flying than to assume they inherited the trait by common descent from a single ancestor that evolved this solution.
Given that chimpanzees, gorillas, and orangutans have repeatedly demonstrated mirror self-recognition, we can ask how such potential may have evolved on this part of the tree of life. If this trait evolved independently, it would entail that it emerged at least four times: once in the ancestors of each of the genera of great apes and humans. However, if today’s great apes and humans share this capacity because of common descent, we would have to assume only one change event in the past: that over fourteen million years ago, before the line leading to orangutans split off, the common ancestor to great apes acquired the trait and passed it on to all its descendants. Homology, in this case, makes fewer assumptions and is therefore a more parsimonious explanation than convergent evolution.
Given that small apes and monkeys do not appear to have the capacity to recognize themselves in mirrors, we can narrow the emergence of this trait in primates down further. The great ape ancestor probably evolved the trait after the split from the line that led to modern gibbons some eighteen million years ago (see Figure 3.4). Based on current evidence, then, the potential for mirror self-recognition evolved between eighteen million and fourteen million years ago in the shared ancestor of hominids. We can make these assertions without ever having to lay eyes on a fossil of the ancestor that first evolved it. We do not know what this creature looked like, but it is likely to have known what it looked like.
FIGURE 3.4.
The likely origin of the hominid capacity for visual self-recognition.
Reasoning by homology is a powerful tool. It allows us to make inferences about the minds of long-extinct ancestral species.16 Given what we have reviewed, the common great ape ancestor and its descendants probably could reason about things that were not directly available to the senses. As we establish the capacities and limits of more species, we can develop a clearer picture of how mind evolved on the various branches of the tree of life. This method, of course, only works for traits that are shared between species. To establish the evolution of traits that are uniquely human we need to examine the fossil record—I will defer discussion of that topic until Chapter 11.
We are now ready to examine what mental traits make up the gap. What are the essential mental capacities that set us apart and led to the explosion of new behaviors that characterize the human condition? When I ask my students this questio
n, their most common answer is language. This is also the most prominent answer in the literature. So that is where we shall begin.
1There are many reasons why researchers may have such leanings. For instance, if you are dedicated enough to work, and perhaps live, with an animal for many years, it should not be surprising that you would become attached and prefer to see successes rather than failures. I have seen a prolific researcher cry over a sick animal and proclaim its life worth more than that of a human. Others may be attracted to the role of the detached, hard-nosed defender of parsimony in a world too easily swayed by romantic, wishful thinking.
2The opposite is also possible. That is, there may be a smaller gap than many people think, and yet animal minds can be explained in lean ways. This case has been argued on the basis that we may overestimate our own mental faculties and that much of human behavior may be based on very lean associative mechanisms that we share with other animals.
3The title of this book may raise suspicion that I might be biased toward putting the bar high to exaggerate differences between animals and humans. I am known in the discipline (at least to some) for proposing that a key advantage of humans can be found in our capacity to mentally travel in time (see Chapter 5). However, I have no particular interest in establishing or defending an overly wide—or overly narrow—gap. I have advanced some killjoy explanations as alternatives to romantic claims in the literature, but I have also made cases for richer abilities in primates than were previously known. For example, my colleagues and I reported the first evidence to suggest that chimpanzees can notice when a human copies them—something to keep in mind the next time you ape an ape at the zoo.
The Gap Page 7