If we could talk to the animals...

What do a honey bee's dance and a monkey's screech have in common? Marc Hauser describes how open-minded research is throwing up new insights into how animals think and communicate

Have you ever had a dream that was so real you almost believed it? I have.It took place 20 years ago while I was living in a tent at the base of Mount Kilimanjaro, in Kenya. I was researching the vocal communication of the East African vervet monkey, a beautiful animal with olive fur, a black face and a long tail. Like Doctor Dolittle, I wanted to understand what the monkeys were talking about. My dream provided a hopeful solution. While following one vervet, I saw that she was carrying a book under her arm. A closer look revealed a title: The Book of Vervet Translations. I asked to look at the book, but she refused, claiming that it was strictly for vervets.

Frustrated, I awoke. I patiently waited for another dream, one with greater illuminative powers. No luck.

My dream captures a problem that everyone faces in trying to understand animal behaviour: we need a system that will enable us to translate what they are doing into what they are thinking. Over 20 years I have developed a simple idea: to explain what and how animals think, it is important to recognise that the brains of all animals are equipped with specialised abilities, which I refer to as "mental tools". These tools are highly efficient, designed by Darwinian selection to solve animals' most pressing problems: finding home, counting allies, recognising predators, communicating with friends, and cooperating with kin. Three tools are shared among all animals and comprise the universal toolkit: the capacity to recognise objects, to navigate through space and to count.

The natural world is filled with numerical challenges. When a parasitic cuckoo bird hatches, it not only boots out the other eggs or nestlings from the host species, but mimics the host's begging behaviour by yelling as loudly as three or four young calling in harmony; such vocal trickery causes the host to feed the cuckoo as if it was worth three or four of her own. When a lioness hears roaring nearby, she must assess how many foreigners are present and whether a challenge would be wise given how many pride-mates are around to support her.

But how precisely do animals count, and how do their brains do the maths?

Experiments with rhesus monkeys show that if you give a monkey two empty boxes and then place different quantities of food in each, they will choose two pieces over one, three over two, and four over three, but pick equally for five versus four. Without being trained, rhesus monkeys can discriminate four items from three, and the same is likely to be true of pet dogs, cats and budgerigars. Human infants also show a limitation of about four, and, as the historical record tells us, a similar quirk appears in our symbol systems for number: among such diverse cultures as Mayan, Chinese, Roman and Arabic, the symbols are logically consistent up to about four and then change.

What these observations tell us is that without words, animals can count, precisely, up to about four. Beyond that, animals and young children can represent large numbers only approximately. With training, some animals can do a bit better, and with education, all children can do a lot better. The number sense is a universal tool that is shared by all creatures. What has allowed humans to go beyond this system for number representation is our language and, in particular, our number words.

If number is an ancient tool, then communication must surely be ancient as well: bees dance, ants leave scent trails, electric fish send high-voltage signals, lizards do push-ups, kangaroo rats drum, wrens sing, bats echo-locate, dolphins whistle, gorillas belch, human babies cry.

What, however, are they communicating? Are their signals simply a reflection of their emotions - passions converted into sounds, smells and visual displays? Or do at least some of these signals go beyond the emotions, providing information about objects and events in the world?

To answer these questions, biologists have had to imagine what it is like to be such animals, putting themselves in their paws and claws and devising creative experiments to understand what all the noise is about.

Consider honey bees. When Karl von Frisch, the Nobel laureate in ethology,first cracked the code of the honey bee's dance, he described it as a language. What he meant was that the bee's dance was symbolic, a sign that stands for something - in this case, for the distance, direction and quality of food that a bee has discovered.

Moreover, and in contrast to what one might expect from an animal with a relatively simple nervous system, bees do not blindly follow just any information from a dancing bee. When the biologist James Gould trained foraging bees to find food in a boat placed in the middle of a lake, and then allowed them to return to the hive to indicate this new location, the hive bees rejected their instructions, since pollinating flowers are never located in the middle of a lake. Even bees can look sceptically at a dancing fool.

Some birds and primates have a communication system similar to that of bees, but it is used more widely. Domestic chickens have a food call and at least two alarm calls, one for aerial predators and one for ground predators. Vervets and rhesus monkeys have calls associated with social interactions, the discovery of food and the detection of at least five different predatory species (large cats, snakes, eagles, small carnivores and baboons).

To show that these calls refer to objects or events in the environment, scientists have designed creative playback experiments. In a classic experiment conducted by the ethologists Robert Seyfarth, Dorothy Cheney and Peter Marler, alarm calls broadcast from a speaker elicited responses appropriate to a specific predator. Vervets looked to the sky upon hearing an eagle alarm call, they ran up trees in response to a leopard alarm and they stood and scanned the ground below in response to a snake alarm call.

This kind of symbolic system is narrowly distributed in the animal kingdom. There is no evidence that any insect other than the honey bee has it, and it is apparently absent in all frogs, most birds and many primates,including chimpanzees.

The capacity to communicate symbolically is a specialised tool. Yet when we compare this capacity with that in humans, even young children, we see striking differences. Whereas all the symbolic signals produced by animals are restricted to a few contexts, humans have the capacity to refer to anything, either imaginary or real, and to objects and events encountered in the past, present and future. While a good ethologist can predict what an animal will communicate in a given context, even the best linguist will have difficulty predicting what a human will say when he or she sits down at a restaurant or queues for a pint at the pub. For, unlike humans, animals do not seem able to combine their vocalisations or gestures into strings that create new expressions with new meaning - they do not produce sentences.

To say that some animals have a capacity for symbolic communication while others do not is not an endorsement of an intellectual hierarchy with people at the top. On the contrary, the revolution that we are witnessing in studies of animal cognition is a testament to a new brand of open-mindedness, one that embraces both the similarities and differences in the ways animals think.

Marc D. Hauser is professor in the department of psychology, Harvard University and author of Wild Minds, to be published by Penguin in August, Pounds 18.99. Primate Neuroscience Laboratory, Harvard University: 0 w.wjh.harvard.edu/mnkylab/