Collective Intelligence in Social Insects


It wasn't so long ago that the waggledance of the honey bee, the nest-building of the social wasp, and the construction of the termite mound were considered a somewhat magical aspect of nature. How could these seemingly uncommunicative, certainly very simple creatures be responsible for such epic feats of organisation and creativity? Over the last fifty years biologists have unravelled many of the mysteries surrounding social insects, and the last decade has seen an explosion of research in fields variously referred to as Collective Intelligence, Swarm Intelligence and emergent behaviour. Even more recently the swarm paradigm has been applied to a broader range of studies, opening up new ways of thinking about theoretical biology, economics and philosophy. It turns out that not only might we, as multi-cellular organisms, be composed of swarms, but so could our societies, economies and perhaps even our minds. In this essay I will outline three of the most promising areas of social insect-inspired AI: ant-based search algorithms, Particle Swarm Optimisation and swarm robotics, and hopefully provide an insight into how these studies have grown out of a small niche of A-life research into an all-encompassing new way of thinking.

In the Beginning

Konrad Lorenz (1903-1989) is widely credited as being the father of ethology, the study of animal behaviour, with his early work on imprinting and instinctive behaviour, however it might be argued that an even earlier pioneer of the field was a South African, Eugène Marais (1872-1936). Marais was a brilliant man - poet, writer, lawyer, psychologist and naturalist. He made ground-breaking studies into societies of wild apes a full sixty years before any other. He also studied termites, known in his day as white ants, publishing articles as early as 1925. In 1937 a book, The Soul of the White Ant[1] was published posthumously in which he described in painstaking detail the resemblance between the processes at work within termite society to the workings of the human body. He regarded red and white soldiers as analogous to blood cells, the queen as the brain and the termites' mating flight in which individuals from separate termitaries leave to produce new colonies as exactly equivalent to the movement of sperm and ova.

Like many geniuses, Marais' life ended in tragedy. A spiralling drug addiction and depression was worsened when in 1927, a Belgian author, Maurice Maeterlinck (1862-1949) published The Life of the White Ant[2] which was largely plagiarised from Marais' articles. In 1936, Marais committed suicide. The full text of The Soul of the White Ant is available on the web, and is well worth looking at since the implications of his insights have yet to be fully understood.

Stigmergy: Invisible Writing

Although Marais had created a detailed document on termites, he was unaware of the mechanics of termite communication. How is it that a group of tiny, short-sighted, simple individuals are able to create the grand termite mounds, sometimes as high as six metres, familiar to inhabitants of dry countries? The answer to this question was first documented by the French biologist, Pierre-Paul Grassé in his 1959 study of termites[3]. Grassé noted that termites tended to follow very simple rules when constructing their nest:

Obviously, this does not tell the whole story but a key concept in the collective intelligence of social insects is revealed: the termites' actions are not coordinated from start to finish by any kind of purposive plan, but rather rely on how the termite's world appears at any given moment. The termite does not need global knowledge or any more memory than is necessary to complete the sub-task in hand, it just needs to invoke a simple behaviour dependent on the state of its immediate environment. Grassé termed this stigmergy, meaning 'incite to work', and the process has been observed not just in termites, but also in ants, bees, and wasps in a wide range of activities.

The application of stigmergy to computation is surprisingly straightforward. Instead of applying complex algorithms to static datasets, through studying social insects we can see that simple algorithms can often do just as well when allowed to make systematic changes to the data in question.

Self Organisation

Moving earth around is only one of many ways in which social insects communicate through their environment. Another famous example of stigmergy is pheromonal communication, whereby ants engaging in certain activities leave a chemical trail which is then followed by their colleagues.

This ability of ants to collectively find the shortest path to the best food source was studied by Jean-Louis Deneubourg[4], when he demonstrated how the Argentine ant was able to successfully choose the shortest of two paths to a food source. Deneubourg was initially interested in self organisation, a concept which until then had been the fare of chemists and physicists seeking to explain the natural order occurring in physical structures such as sand dunes and animal patterns.

A self organising (SO) system is any dynamic system from which order emerges entirely as a result of the properties of individual elements in the system, and not from external pressures. The classic example of SO is Bénard cellular convection, named after the French scientist who discovered it. A Bénard cell consists, very simply, of a layer of fluid which is heated from below. Under the right circumstances, a perfect vertical temperature gradient is set up within the fluid, causing the system to become ‘top heavy’, with warmer molecules at the bottom compelled to rise to the top. You might expect the liquid to simply bubbling away with no pattern or organisation, but instead an ordered system is formed. Millions of molecules self organise into a hexagonal pattern, something like a honeycomb, which enables the most efficient convection for energy-dissipation.

Deneubourg saw the potential for this concept, which by 1989 had turned into a sizeable research project amongst physicists, to be applied to biology. In his experiment, a group of ants are offered two branches leading to the same food source, one longer than the other. Initially, there is a 50% chance of an ant choosing either branch, but gradually more and more journeys are completed on the shorter branch than the longer one, causing a denser pheromone trail to be laid. This consequently tips the balance and the ants begin to concentrate on the shorter route, discarding the longer one. This is precisely the mechanism underpinning an ant colony's ability to efficiently exploit food sources in sequential order: strong trails will be established to the nearest source first, then when it is depleted and the ants lose interest, the trails leading to the next nearest source will build up

Remember you can visit the Message Store to discuss this essay. Comments are always welcome!. There are already replies in the thread, why not join in?