Essay: Look who's talking

The creatures in the cartoon film 'Antz' aren't pure fantasy. New research suggests that ants are not only better workers than us - they can talk, too. By Donald Michie, professor of machine intelligence

The depiction of ants in the new animated film Antz could strike a biologist as peculiar. Real ant society, quite unlike that in the movie, is almost exclusively female, with just a few chronically layabout males. However, the film's suggestion that ants show sparks of cognition is by no means silly. Several laboratories around the world are investigating ants' perceptions, and particularly their capacity to navigate by memor-ising landmarks and plans. Now, research suggests that some ants routinely communicate complex information.

Anything resembling cognition in ants must rest on a tiny neuro-anatomy. There are 10,000 million neurones in the human cerebral cortex; the equivalent part of an ant's brain contains a mere 250,000 neurones.

But Edward Wilson, the world's leading scholar on ants, reminds us that a colony of 40,000 is, in effect, a "super-organism" with a total of about 10,000 million neurones supporting its co-ordinated actions. In other words, it is possible to regard the actions of an entire colony as similar to the neurochemical co-ordinations of tissues and organs that support the goal-seeking behaviour of a single vertebrate animal.

Ants use chemical signals to recruit large numbers rapidly, transport pupae, store food or repel attackers. Wilson remarks of this aspect of their evolution: "They learned how to throw an entire colony against a single attacker or into a single task." Such chemical communication can be compared to the human use of visuo-auditory channels - as in religious chants, advertising images and jingles, political slogans and martial music - to arouse and propagate moods and attitudes. The biologist Lewis Thomas wrote: "Ants are so much like human beings as to be an embarrassment. They farm fungi, raise aphids as livestock, launch armies into war, use chemical sprays to alarm and confuse enemies, capture slaves, engage in child labour, exchange information ceaselessly. They do everything but watch television."

There is one other thing to remember about ants. In most ant societies, there is no such thing as cultural transmission. Virtually everything must be encoded in the genes. Each colony starts afresh from eggs laid by a single impregnated queen. So anything learned by the previous generation is lost. The behaviours, habits and tricks which interlock to form and maintain the new colony have to be biologically inherited, either ready- formed, or as prototype behaviours to be refined in use.

In humans, the opposite is true. Only basic propensities are in the genes of a newborn baby. The rest are developed during the first two decades of life.

If, for some reason, the cultural transmission link were broken, in a flash we humans would be back to the pre-agricultural, pre-tool-making, pre-fire-using era, with all our arts and crafts and languages gone. But most ant species could carry on practising their intricate crop-raising, herding, weaving, civil engineering and foraging as though nothing had happened. For ants, cultural discontinuity occurs in every generation.

It may seem that our cultural continuity gives us a huge advantage over ants.They have never mastered fire, nor progressed from the use of simple tools to tool-making, let alone to machines. By the agricultural standards of Biblical times their fungus-farming and aphid-herding crafts may seem advanced, but they have been now been eclipsed by human agri-business.

Or have they? The farming methods of ants are at least sustainable. Human agri-business not only ruins environments but emerges from any energy accountancy as a net loss-maker. Moreover, recent evidence suggests that the crop-farming of ants may be far more sophisticated and adaptable than was thought.

Ants were farmers about 50 million years before humans were. They can't digest the cellulose in leaves - but fungi can. Some ant species collect leaves as fodder for edible fungi which they cultivate in their nests, and feed on. Cultivation methods have become extremely complex. Farmer ants will secrete antibiotics to control other fungi that might act as "weeds", and spread their waste to fertilise the crop.

It was once thought that the fungus that ants cultivate is a single type that they have propa- gated, essentially unchanged, from the distant past. Not so. Ulrich Mueller of Maryland and his colleagues genetically screened 862 different types of fungi taken from nests. These turned out to be highly diverse: it seems that ants are continuously domesticating new species. Even more impressive, DNA analysis of the fungi suggests that they regularly swap and share strains with neighbouring ant colonies. The crop farming of ants is practised entirely indoors, in specially constructed chambers.

Whereas neither Neanderthal nor Cro-Magnon man had any exposure to urban lifestyles - the forcing-house of intelligence - the evidence suggests that ants have lived in urban settings for close on 100 million years, developing and maintaining underground cities of specialised chambers and tunnels.

When we survey Mexico City, Tokyo, Los Angeles, we are amazed at what has been accomplished by humans. Yet Hoelldobler and Wilson's magnificent Bible for ant-lovers, The Ants, describes a supercolony of the ant Formica yessensis on the Ishikari Coast of Hokkaido. This "megalopolis" was reported to be composed of 306 million workers and a million queens living in 45,000 interconnected nests across a territory of 2.7 square kilometers.

Such enduring and intricately meshed levels of technical achievement outstrip by far anything of Cro-Magnon man. We hail as masterpieces the Cro-Magnon cave paintings in southern France and elsewhere, dating back some 20,000 years. Ant societies existed in something like their present form more than 70 million years ago. Beside this Cro-Magnon looks technologically primitive. Is this then some kind of intelligence, albeit of a different kind?

A collaboration between scientists at the universities of Oxford, Sussex, Bonn and Zurich has shown that when desert ants return from a foraging trip they navigate by integrating compass bearings and distances, which they continuously update in their heads. They combine the evidence of visual landmarks with a mental library of local directions, all within a framework which is consulted and updated over the journey. So, ants can also learn.

In one of many experiments conducted in Siberia by the scientists Boris Ryabko and Zhanna Reznikova, three species of wood-ant were presented with a two-way fork in an enclosed path, one of which led to their receiving an electric shock. After one shock, they avoided taking that path. In these species of wood-ant, past experience can be transmitted by specialist ant "scouts". The scout forms a huddle with team-mates of a different type called "foragers", and makes repeated contact using antennae, mouth parts and legs; the huddle can last from tens of seconds to a few minutes. A forager who has been allowed such contact after the scout has been subjected to the electro-shock experience subsequently shows a strong preference for the safer alternative when confronted with the same choice of path. It seems she has been tipped off by her patrol leader.

In a 12-year programme of work, Ryabko and Reznikova have also found evidence that ants can transmit more complex messages. Scouts who had located food in a maze returned to mobilise their foraging teams. They again engaged in contact sessions, at the end of which the scout was removed, to observe what her team would then do. Often the foragers proceeded to the exact spot in the maze where the food had been. Elaborate precautions were taken to prevent the use of odour clues from the forager teams.

Discussion now centres on whether the route through the maze is communicated as a "left-right" sequence of turns, or as a "compass-bearing and distance" message. Message-passing of this second kind has already been well studied in the case of the honey bee's "waggle dance", which she performs on returning to the hive from a find. By following the angle of her repeated figure- of-eight run and noting the frequency of waggle, her hive-mates are able to fly to the indicated location.

Ryabko and Reznikova's results did not fit this model, though. Ryabko, a renowned coding theorist, had the idea of timing the ant huddles, to see if that could throw light on the nature of the messages. In "left-right" code, the messages of, say, LLLLLL and RRRRRR are obviously simpler to absorb than LRRLRL or RLLLRL. And, indeed, such was apparent in the recorded ant results: the simpler the route, the briefer the huddle. This relation makes no particular sense when assuming a simple direction-and-distance code, but rather supports the idea that a sequence of turns is be- ing communicated.

Having visited Boris Ryabko and his wife and co-worker Zhanna Reznikova, and studied original records from their many years of work, I find the phenomena hard to dispute. But biologists are trained to hesitate long (and sometimes loud) before accepting any startling claims. Only once has a case of ant teams of specialist workers been reported before. It is related in Hoelldobler and Wilson's The Ants. Nigel Franks of Bath University studied a species of army ant that can rapidly pull together multi-skilled, multi-caste teams for transportation feats far beyond the capacity of any individual or small ant team. This observation, made 10 years ago, was the first ever of ants forming temporary teams of specialists to solve problems. The Ryabko and Reznikova scout-forager teams contrasted with this by having stable membership, a discovery which showed that certain ants can recognise each other individually - possibly by smell. This raised the question of why such a mechanism might have been evolved. What purpose could this serve in the wild? These ant species have a propensity to herd and milk aphids, rather as humans herd cattle. On discovery of a new aphid herd on a particular leaf, it might be handy if a scout could return to the nest and dispatch a skilled team to manage and milk the aphids for their honeydew, leaving herself free to continue roaming. This is Reznikova's view, based on close study of the ants in their woodland habitat.

Between the wars, Edward Tolman trained rats to run mazes, and then wondered whether they had simply built up stimulus-response associations, with perceptual clues triggering their limb actions. He tried flooding the mazes with water, obliteraring all similarities to the training conditions. The rats now swam the maze - correctly.

Were they using a cognitive map? They were certainly operating on a higher level of abstraction than if they were simply responding to stimuli. I would call this "borderline cognition", as I would the "left-right" sequence- encoding hypothesis in Siberian wood-ants. But in the ants' case there appears to be communication too, sometimes beyond the basic encodings of sequence of turns.

I return to a stumbling block in all this. Do scout-forager ant teams really persist as identifiable working groups for months at a time? To investigate this, ant scientists will mark ants individually. But, with one exception, only the Siberian laboratory's investigation of these aphid- farmer ants has found constant teams. My acquaintance with Zhanna Reznikova's marking and recording of the ants leaves me concluding that ants can and do form lasting membership teams.

The phenomenon of constant teams is not confined only to these three particular species out of the 10,000 or so that make up the ant world. In Moscow, nearly 20 years ago, Anatoli Zakharov reported the existence of such groups in the ant species Formica rufa.

In her laboratory, Zhanna Reznikova gave me an uncannily fast-fingered demonstration of her colour marking procedure. She deftly plucked an ant from the large open-box tray supporting one of her experimental nests. The tray's inner walls had been spread with Vaseline, which the ants cannot cross. During the swift but elaborate operation, the creature lay prone and quite still on Reznikova's finger, only the tips of its two forelegs secured by her thumb "Why does it keep so still?" I asked. "I don't know," she answered, beaming at me through her spectacles with love and pride. "Do you know, even without paint-spots I think I know them by their faces."

There is a nonsensical idea that to elicit such empathy a small creature must be a fellow-mammal. An essay by Edward Wilson, "In the Company of Ants", rebuts this: "The question I'm asked most often about ants is 'What do I do about the ones in my kitchen?' And my answer is always the same: 'Watch where you step.' Be careful of little lives. Feed them crumbs of coffeecake. They also like bits of tuna and whipped cream." !

Donald Michie is Professor Emeritus of Machine Intelligence at Edinburgh University