Mark Rowlands asks whether mirrors can help us understand animal minds.
Many years ago I was driving along in an old and unpredictable Skoda, when I spied a wheel hurtling off through a field and into the western arm of Cleddau estuary. “Some poor bastard’s wheel has fallen off”, I thought to myself. I quickly realised I was the bastard in question. The police were a great help – heartening me with a compendium of Skoda jokes they had been saving up for just such an occasion. How do you double the value of a Skoda? Fill it up with petrol. Why does a Skoda have a rear window heater? So your hands don’t get cold when you push it – and so on and so forth. What – you might wonder – does this have to with the question of self-awareness in animals? The answer is: everything, my friends: everything. Getting to that answer, however, will take (roughly) the next 2500 words of your lives.
It is clear that most animals must be, in some sense, aware of themselves. After all, it is rather important for an animal to know whether another animal is eating it or eating something else. And so it must be able to distinguish things happening to its body from things happening to the body of something else. But this isn’t enough for self-awareness in the sense that many have in mind. Being eaten is, I should imagine, rather painful. Therefore, all the animal needs to be able to do to differentiate its own body from the body of others is to detect when it feels pain. It does not need to recognise its own body as its own body.
An animal that can pass the mirror self-recognition (MSR) test, on the other hand, does seem to be able to recognise its body as its body. A mark (typically dye) is placed on an animal subject, in a position where it is visible only in a mirror. The animal’s subsequent behaviour in front of a mirror is observed. If it appropriately engages with the mark – using the mirror to inspect it, for example – then it is deemed to recognise that the body reflected in the mirror is its own body. This seems to qualify as a form of self-awareness.
It is not entirely clear which animals pass this test. It is generally accepted that humans (over the age of 18-24 months), common chimpanzees, bonobos, and orangutans consistently pass the test. Animals that have been argued to pass the test include elephants, dolphins, and pigeons. I gather that manta rays have been patiently building a case. Gorillas, to be quite frank, have struggled with the test, although this is probably because they regard eye contact as an aggressive gesture, and so avoid looking at each other’s faces – including faces that stare back at them from a mirror.
Also unclear is precisely what one has to do to pass. Dogs, for example, are deemed to fail the test, on the grounds they show no interest in any marks placed on them. This highlights the first shortcoming of the test: it only applies to creatures that are suitably motivated. Dogs get crap on themselves all the time – sometimes quite literally, especially if you live next to cows – and just don’t seem to care. So, why should we expect them to show any interest? One thing they do care about much more than how they look in the mirror is urine. Marc Bekoff ran a very nice experiment along these lines – the yellow snow experiment. Marc recorded how long his dog, Jethro, spent sniffing urine deposited by dogs in the snow. To make sure he could not identify his own urine by remembering where he left it, Marc shovelled up Jethro’s yellow snow and moved it to new locations. Jethro spent significantly less time sniffing his own relocated urine than he did that of other dogs – suggesting he is able to differentiate between urine that is his from urine that is not. Does this show that Jethro has a concept of “mine” and, therefore, the related concept “me”? Unfortunately, no matter how imaginative one’s experiments are, there is always some deflationary killjoy out there who will try to negate it. Perhaps, for example, Jethro merely discriminates yellow snow that is “more interesting” from snow that is “less interesting” – he is driven to pay more attention to some patches of yellow snow over others without understanding why. I’ll leave the reader to determine the plausibility of this counter hypothesis.
There are further problems of interpretation of evidence. A dog will exhibit one of two reactions to its image in the mirror. The first: utter indifference. This, apparently, means that it has failed the test. The second: it attacks the mirror – the opposite, in effect, of utter indifference. This also means that it has failed the test. Come on now: you can’t have it both ways. If indifference equates with failure, how can it opposite also be a symptom of failure? Consider the dog I know best in the world: Hugo, the family German shepherd. Hugo sees his reflection in a full-length mirror every day, a circumstance to which he is casually indifferent. Knowing Hugo rather well, I can say, without a shadow of a doubt, that if he thought there was an unfamiliar dog in his house, casual indifference would not be the way he would choose to deal with it. So, if he does see a dog reflected in the mirror – and if he doesn’t, the test is inapplicable anyway – this seems pretty good evidence that he knows that he is that dog. Indifference, in fact, might be evidence of passing the test and not, as it is commonly thought, evidence of failure.
What I am going to talk about most, however, is a further problem with the mirror test. MSR is taken to be a sign of bodily self-awareness: the ability to recognise a body as one’s own. I shall try to convince you that mirror self-recognition is actually a sign of only one form of bodily self-awareness. There is another – equally or more important – and the mirror test has nothing to say about it.
The kind of awareness tested for in MSR is what we might call reflective bodily self-awareness. I am reflectively aware of my body when I have a thought, or other mental state, that is about it. If, on looking at my reflection in a mirror, I think, “Yep, that’s me”, then I have a thought about my body. The philosophical term for this sort of about-ness is intentionality. Thoughts, beliefs, desires, hopes, fears, and expectations are all examples of intentional states in the sense that they are about something. Whenever you think, you think something. Whenever you believe, you believe something, and so on. To think or believe something is to represent an object as falling under a certain mode of presentation. When I look in the mirror and recognise the reflection as me, then I represent an object (the body reflected in the mirror) as falling under a mode of presentation (identical with me). Intentionality has a tripartite structure involving an act (e.g. recognition), an object (e.g. a reflected body) and a mode of presentation or concept (e.g. identical with me). An animal that passes the mirror test is able to represent its body, in thought, as falling under a given mode of presentation: “Me!” It is, in this sense, reflectively aware of itself.
There is, however, another form of bodily self-awareness. This is where my earlier tale of the Skoda begins to earn its keep. The point of this tale: there was a wheel that was, in fact, mine but which I didn’t recognise as such. Let’s transplant this idea to the mirror test. Here is an example inspired by David Kaplan. I am standing in front of a mirror, but don’t realise this. Reflected in a mirror is a man whose pants are on fire. I think: some poor bastard’s pants are on fire. You can guess how the rest of the story goes.
I can see a body in a mirror, and this body can be my body, without my recognising it as such. There is always the possibility of misidentification of this body. Or, equivalently: the body is not immune to error through misidentification. Let us contrast this with another kind of self-awareness – awareness that is immune to this sort of error. My mental states are, typically, immune to error through misidentification (IEM). If I see my body in a mirror, there is always the possibility I might not realise whose body it is. “Who is that?” is a perfectly intelligible, if unlikely, question. But the same does not apply to my mental states. “Here is a pain. I wonder whose it is?” does not seem to make sense. “I wonder who is thinking this thought” also seems unintelligible. When I am in pain, or think, I can’t be mistaken regarding who has the pain or the thought.
The body I see in the mirror is not immune to error through misidentification. But there is another way of being aware of my body that makes it immune. This awareness of the body we might call, borrowing from Sartre, pre-reflective bodily self-awareness. This sort of awareness is part and parcel of having experiences at all. Suppose I see a book. I see it, precisely, as a book. But it is also true that I do not see all of the book, but only the part of it that is oriented toward me (e.g. the front). Nevertheless, I see it as a book not as a book-façade. This is because I have certain implicit expectations or anticipations concerning how experience will change in given circumstances. I anticipate, for example, that if the book were rotated, I would see first the spine and then the back cover of the book, and so on. If I saw the object as a book-façade, on the other hand, the series of expectations this would generate would be rather different. This is the difference between seeing an object as a book and seeing it as a book-façade.
The crucial point is that I am implicated in many of these anticipations. Many of the relevant anticipations involve either my doing something or something being done to the book relative to me. I anticipate that if I were to move a certain distance to the left, the appearance of the book would change in a certain manner. And it would change in a somewhat different way if I were to move the same distance to the right. I also anticipate that if the book were to move relative to me, then the appearances it presents would also change in a certain way. Thus, to see these appearances as the appearances of a book involves a kind of awareness of myself. It is I – one and the same I – that am implicated in these expectations: if I were to move to the right, if the book were to move relative to me, and so on. This awareness I have of myself is required for me to see these appearances as appearances of a book (rather than as appearances of something else).
This form of self-awareness does not conform to the model provided by reflective awareness. Bodily self-awareness is implicated in seeing the various appearances of the book as appearances of one and the same object. But none of this involves having thoughts about myself. And none of it requires the existence of any intentional act that takes me as its object.
Consider another example. Much perception is perception for action, and pre-reflective self-awareness is also built into this. Perception for action is perception of the world in terms of the various possibilities – whether positive or negative – it affords for action. Let us suppose I walk into a room and spot an empty chair. In doing so I may have noticed very little about the chair: its colour, shape, fabric and construction may all have escaped my notice. What I did notice was one salient fact: it is empty. And because it is empty it is what we might call sit-able: it affords sitting. It is the affordance I see and not the other properties of the chair. But the chair affords sitting only because I have a body of a certain sort. If I were twelve feet tall, or twelve inches tall, or had four legs, then the chair would not afford sitting. To be aware of the affordances of any object is, at the same time, to be aware of one’s body.
The body I see in the mirror is the body as object: an object of an intentional act – the act of seeing. I might not recognise the body I see in the mirror as mine. This body is not IEM. The body I am aware of when I see the book as a book, or the chair as sit-able, on the other hand, is IEM. There is no question of whose body this is, because there is no question of for whom the book appears as a book. To see a book as a book is to have various expectations about how your experience will change in given circumstances. And no question can arise as to whose expectations these are. There is similarly no question over to whom the chair affords sitting. “Here is an affordance, I wonder who it is an affordance for?” is not a question that makes sense. The body you are aware of when you experience an object as something is not the body as object but the lived body – leib rather than körper, as the phenomenologists put it.
In the field of animal cognition there is, I think, an emerging consensus that the mirror test does not do justice to the range of phenomena that might be grouped under the category of self-awareness. I have outlined what I think is an important form of self-awareness that the mirror test fails to recognise. I borrowed a label from Sartre – “pre-reflective” – to describe this neglected form of self-awareness. The label may be Sartrean, but the underlying idea is Kantian. He called this form of self-awareness transcendental apperception – and distinguished it from the empirical apperception or inner sense that corresponds (roughly) to reflective self-awareness. Of the two forms of self-awareness, Kant thought transcendental apperception is the more significant, for it alone can ground the unity of the self. To have an experience of an object as an enduring thing – in contrast to a disjointed series of appearances – is, at the same time, to be aware of your self as an enduring thing. Transcendental apperception, therefore, underlies the unity of a mental life. If animals have experiences of objects as unified, enduring things, then they also transcendentally apperceive themselves as unified, enduring things.
I don’t want to annoy all the Kantians out there, so let me be clear: Kant would hate what I’ve done with his idea. Hate it. In particular, he would deny the antecedent of the final sentence of the previous paragraph: he would deny that animals have experiences of objects as enduring things because they lack the conceptual abilities required to synthesise successive appearances into a unitary whole. Kant knew a lot about transcendental synthesis, but not much about animals. And given what we do now know about animals – given the plethora of object permanence testing we have conducted on them in the last few decades – this denial is looking increasingly forlorn. The time has come to steel ourselves for the obvious: animals are self-aware, unified subjects of a mental life. But, a final twist: a unified mental life is the sort of thing many people have in mind when they use the word “person”. If that is what a person is, then many animals will qualify as persons. But that is a subject for another time.
MARK ROWLANDS IS PROFESSOR OF PHILOSOPHY AT THE UNIVERSITY OF MIAMI, AND AUTHOR OF EIGHTEEN BOOKS, INCLUDING THE PHILOSOPHER AND THE WOLF AND CAN ANIMALS BE MORAL?