Note from Con Slobodchikoff: This is a guest post by Dr. Alexandra Horowitz, whose work on dog attention and theory of mind was featured in a post on this blog on October 8, 2008. Dr. Horowitz has written a book, "Inside Of A Dog," that will be published in September 2009 by Scribner. It is about what a dog's point of view might be like, using scientific results to help draw the picture.
It is notoriously difficult to test for theory of mind in animals. "Theory of mind" (ToM) refers to the attribution of mental states--particularly belief, desire, and knowledge--to others. Research shows that human children are not born with this understanding, but they develop it by age four. Many claim that ToM is wrapped up in what enables us humans to be (at times) so cooperative, so empathetic--in a word, so human. As a result, there is great interest in determining whether any non-human animal has a ToM.
The question of non-human ToM is vexed in a couple of ways. For one thing, tests to determine when children develop a ToM--for instance, the false-belief test, which asks if a child realizes that someone who's been out of the room doesn't know about any machinations that happened inside the room--require that the children use language to tell researchers what they know. So much for testing animals. For another, with the non-verbal tests that have been developed, the results with animals (mostly chimpanzees) have been mixed. Sometimes chimps act with seeming understanding of what others know; sometimes they act as though utterly oblivious.
It occurred to me that there is a problem with our concept of ToM. The implicit premise is that ToM is an "all or nothing", on-or-off ability, which pops "on" early in childhood, rather than developing in stages. This kind of conceptual definition reminds me of the Cartesian notion that an animate creature is either self-aware and conscious, or is merely an unthinking material form. Descartes led us to dualism, which we spent hundreds of years struggling out from under; but we retain his binary template of an animal being either insightful or mindless.
I propose, by contrast, that ToM is not best considered a unitary ability that one either has in spades, or one does not have at all. Instead, an individual may be able to realize that something connects all the behavior you see someone else doing--without specifically postulating a "mind" behind it. This step could be described as having a "rudimentary" theory of mind: more than acting instinctively, or as a behaviorist, but less than acting with the theoretical understanding characteristic of humans.
Formulated in terms of what we need to attribute to animals in order to explain their social behavior, we get the following hierarchy:
Theory of behavior: the understanding that certain behaviors lead to certain outcomes
Rudimentary theory of mind: the appreciation that there some overarching element which leads to seen behaviors
Theory of mind: extrapolating that what others' behaviors have in common is their root in a mind
Much of animal behavior requires only a "theory of behavior" to explain: an animal's action is based on simple associative conditioning, having learned that one event follows another. For instance, a dog who learns to sit at attention when a familiar person approaches him with a biscuit need not have posited anything about the mind of the biscuit-carrier; he need only have associated the fact of the biscuit being carried with his ultimate reward (biscuit-eating). The five-year-old child sitting waiting for some similarly valued reward from a parent has drawn the same association, surely--but also has a head full of conjectures about what the parent intends, thinks--in other words, what is on her mind.
The intermediate level allows for a middle ground between these poles. The dog who has learned not just to sit before the biscuit-carrier, but also to continue to adjust his sitting to always see her face; the dog who has generalized from the one biscuit-handling person to investigate whether other humans carry biscuits, but learns that not everyone can have biscuits extracted from them by sitting dutifully--these dogs have a more advanced theory of behavior. They are sensitive to salient elements of the behavior--the attention of the human, for instance--and use this knowledge to generalize to future contexts.
As Con has described in an earlier post, when I studied dogs' behaviors in dyadic rough-and-tumble play, I saw dogs behave in a way that I think is best explained by imputing to them a rudimentary theory of mind. They play-signaled--a request to play--only to dogs facing them, thus able to receive that request. When a dog paused in play, his play-partner used attention-getters to redirect him before signaling again. And, even more impressive, dogs used attention-getters matched to their partner's state of inattention: if a dog is looking at you, you needn't bite him on the rump to get his attention; a minor leap up will do. But if he is playing with another dog, the leap-up is ineffective; something on the order of a rump-bite is required.
What this demonstrates, I have suggested, is that the dogs were acting as though they understood that not only must you react to the presence of other dogs, not only must you communicate to the faces of other dogs--but you must also attend to an invisible feature of them--a feature that between humans we call "attention"--in order to communicate, and thus play together. Acting with sensitivity to attention is much more sophisticated than simply acting in response to seeing a stimulus: play-signaling whenever you see another dog, or using any attention-getter if you can't see a face. We simply cannot tell from the data whether the dogs actually have a theory of mind--a real understanding of the minds of others--but their behavior plainly shows us that they have more than just a theory of behavior.
I realize that many dog owners have no reluctance in asserting that their dogs have a theory of mind--and much more sophisticated cognition besides. "Dogs have the intelligence of three-year-olds", it's claimed. As discerning and sensitive companions as dogs are, I haven't seen the evidence yet. And I'd suggest that we owners, in turn, don't truly treat dogs as theorists of mind: we wouldn't dream of leaving a three-year-old alone with little to do for hours on end, the way we do our dogs; we ignore them when we're busy with other people or with work; we do nearly nothing to advance their intellect. Dogs are best served not by over-attributing to them but by acknowledging what they do know, and acting accordingly.