Cheep imitations of human thought

Despite many efforts to prove otherwise, the minds of animals are profoundly different from ours. We should stop trying to compare animals to us and start thinking about them as they are, argues Stephen Budiansky

Several times a year, readers of the world's major newspapers can expect to encounter a story that goes roughly as follows. Researchers have found that (chimpanzees, gorillas, dolphins, elephants, racoons) have an ability to (count, add, subtract, create sentences, solve problems, make tools, deceive others), an ability previously believed to be possessed only by humans.

This discovery, the article continues, proves that animals can think, thereby challenging the centuries-old belief that humans are unique among animals. "The artificial barriers that we have erected between humans and other supposedly 'lower' animals have fallen in the face of this new finding," an eminent expert can be relied on to say.

The most recent manifestation of this story came a few weeks ago, with a widely reported account of research showing that monkeys can count. The research was fascinating: monkeys trained to place sets containing one to four items in ascending order spontaneously extended that ability to sets containing five to nine items.

But the invariable gloss placed on these findings - that another barrier between man and animals had been breached - was, as always, terribly misleading. Not just misleading, but deeply ironic. For, in the name of championing animals and rescuing them from the evils of man's arrogant "anthropocentrism", those who pursue this line of logic adopt the most anthropocentric stance imaginable. To claim that distinctly human mental abilities such as language or mathematics have been demonstrated in rudimentary form in animals does not, as is so often claimed, knock man off his pedestal. Instead, such claims perpetuate the very idea they seek to challenge: that human minds are the gold standard against which all other minds should be measured.

There is a sameness about the minds of animals, but there is also a profound otherness. It is only human for us to read humanlike thoughts and emotions into the animals that surround us, especially the dogs and cats many of us share our lives with. It is harder for us to grasp the strange and even eerie otherness of their minds. Dogs live in a universe of odours and their trails; pigeons orient by the earth's magnetic pull; Clark's nutcracker, a bird that caches seeds for the winter, inhabits a mental world of thousands of memorised geometries. These are mental worlds literally beyond our ken.

We are visual animals, and it is telling that the animals we think of as intelligent are those with good eyesight. For example, rats were considered to be less bright than monkeys because they did less well on a test that involved learning to distinguish visual symbols. When rats were given a second try, this time with the task of distinguishing and remembering different smells, they passed the test with flying colours. We are ridiculously blind to the otherness of much of nature.

Those who champion the humanlike intellectual prowess of animals often do so as part of an explicit agenda to challenge man's claims to uniqueness. One prominent ape-language researcher, Sue Savage-Rumbaugh, writes that viewing ourselves as mentally apart from animals has "led us blindly to exploit the world of nature". Savage-Rumbaugh is best known for her work with Kanzi, a pygmy chimpanzee that has learned to use plastic "lexigrams", geometric symbols that stand for words. Like others who argue for the humanness of animal minds, Savage-Rumbaugh insists that those who would deny that animals use language just as humans do are reducing them to mere automatons, Skinnerian stimulus-response machines.B. F. Skinner, the pioneering animal behaviourist, did argue decades ago that animal intelligence was just mechanical, stimulus-response behaviour. There are few such radical behaviourists left now, and the Skinnerian model is today a straw man. Still, he gets the stuffing predictably beaten out of him with every new discovery about animal intelligence. "See," the animal champions say, "animals are not just bundles of instinct or pre-programmed robots; they are thinking, feeling creatures just like us."

Thinking, feeling, yes. But just like us? What this simple dichotomy ignores is that since the 1960s a new model to understanding animal thinking has been in the ascendant. This model, sometimes called the "cognitive" or "informational" perspective, is the key to understanding what makes animals at once very much like us and - at the same time - wholly different.

Animals clearly do think, by anyone's definition of that word. They take in information from their environment, process it and produce decisions. The amazing thing, however, is that they do this without using language. Dogs and horses, for instance, can find their way home by recognising landmarks along a familiar trail or by dead reckoning - orienting themselves in space by computing how far they have travelled and in what direction. Rats can learn to enter the fourth tunnel along a wall, even when the spacing between the entrances is varied. Blue jays can categorise pictures of different species of moths that would stump most humans.

Herbert Terrace, one of the scientists who carried out the recent monkey counting experiment, told me that, to him, studying how animals do these things without language is a more interesting question than trying to see if a chimpanzee can imitate rudimentary human mental feats. The ape-language experiments are the most notorious examples of the latter. Apes are trained over and over with carloads of sweets as rewards for constructing sentences in sign language or with lexigrams. In one of Savage-Rumbaugh's experiments, chimpanzees learned to push lexigram keys on a computer keyboard to write out sentences such as "please/machine/give/M&Ms". Unfortunately, even the star ape-language performers never have anything much more interesting than that to say. According to Savage-Rumbaugh's published data, 96 per cent of the "sentences" produced, even by Kanzi, are demands for immediate rewards such as food, toys or tickling. As Terrace points out, a pigeon can be taught to push a sequence of four coloured lights to obtain a food reward. Labelling a button with a symbol or calling it a word does not make it language when a chimpanzee pushes it.

To say that animals think is only to state the obvious. And to say that animals think is not to say that animals think like us. The continuity between animal and human thought is most strongly evident in the many computations that the human mind performs quite unconsciously. Without even being aware of it, we are able to perform feats of calculation that would put a computer to shame. We can accurately throw a football while running towards someone else who is running - try writing the ballistic equations for that feat. We can walk a familiar route and make all the correct turns without having to think about it - think in the sense of consciously working it out. Animals do these sorts of things, too, and some that are even more remarkable. (Sheep can recognise scores of individuals by their faces; most of us would be hard pressed to tell one sheep face from another.) It is a promising area of research to try to understand how minds are constructed to accomplish things that stump the most powerful computers in the world. We can build a computer that can beat the world chess champion; we cannot yet build a computer vision system that can guide a robot competently over uneven ground.

But alongside this continuity that exists between our minds and those of other animals, there is a profound discontinuity that consists principally in the human use of language. Many animals communicate, of course. Even insects do and, for that matter, plants. Human language, however, is on a completely different conceptual plane in one crucial respect, for it makes possible what philosophers call "meta-representation". Human language is not just a sort of one-for-one substitution code, in which a sound or symbol represents a thing. Rather, language makes possible thoughts about thoughts, ideas about ideas; it is what allows humans to think about whether they will go to the movies next weekend, read a book to learn what happened 200 years ago, fix an automobile engine, debate a moral issue or imagine the conversation they wish they could have had with Groucho Marx.

Animals, I can reliably state, cannot do any of these things. To play the game of trying to smash the pedestal of human uniqueness by showing that animals can, too (after a fashion), make tools, or create sentences or count, misses this fundamental point.

Human language is also intimately tied to one unique human quality, which is the ability to conceive of the existence of minds outside of one's own. Language-trained apes, again, use their "language" almost exclusively to demand immediate rewards. Human infants from the earliest age use language in a profoundly different fashion - they use words to elicit the joint attention of a fellow human. A baby may say "Plane!" or "Red!" to its mother not because it wants the plane or car but just for the pleasure of sharing its awareness with another mind.

Terrace notes that human infants indeed use words often just to show that they know that the word stands for an object. "In the Beginning Was the Name" was the apt title of a paper he wrote about this observation. Terrace's counting monkeys do think, but they are not thinking like this.

The key to appreciating animals for what they are is to study what they do well. That means understanding how they think without the benefit of symbolic language. That means understanding how pigeons navigate, how sheep recognise faces, how birds recall where they have cached seeds, how horses remember landmarks. The first step is to stop thinking about animals as compared to us, and to start thinking about animals as they are.

Stephen Budiansky is the author of If a Lion Could Talk: Animal Intelligence and the Evolution of Consciousness (Weidenfeld & Nicolson).