Ants

One of the most studied social insects is the ant. As for the other social insects, such as the bee or termite, there are at least two scales in the system: at the level of the individual ant, the behaviour is simple (though sometimes apparently random) while at the level of the colony one sees cooperative phenomena which is self-organised. That is, without a preconceived global plan or internal supervision, ants are able to construct elaborate nests and forage for food apparently using only simple local interactions among individuals. In the experiment described in Ref.[11], two identical food sources were placed equidistant from an ant nest. The food supplies were constantly replenished so that they remained the same. The distribution of ants between the two sources was then analysed and compared with theory. It was known that when an ant finds food, it lays a chemical (pheromone) trail back to the nest. That trail can be used by the same ant or others to guide them back to the food source. Thus ants leaving the nest are likely to follow the trails created by earlier ants, leading to self-reinforcement. Thus theoretically one would expect that the long term behaviour of the foraging ants would be largely determined by the actions of the first few, implying that the proportion of ants visiting the two food sources would remain stable. However in reality the proportion of ants visiting the two sites fluctuates, showing not only expected small variations but also occasional large and rapid swings. A better theoretical model to explain the ant's behaviour is the following [11]: An ant leaving the nest follows one of three rules,

1. It can revisit the food source it last encountered,
2. It can be recruited by a returning ant to visit the other source,
3. It can act independently and visit the other source.

Thus the model acknowledges an element of randomness in the behaviour of the ants. In actuality this randomness is an asset as it enables the colony to adapt to changing situations, for example in exploiting new food sources that might become available. In other words, the ant colony processes its options in parallel, continually exploring better solutions. Furthermore there is also redundancy in the system as many ants perform the same function and can back up their activities if the need arises. The redundancy is obviously crucial for the survivability of the colony. Ormerod (see Refs.[11, 12] uses the three rules above to explain not only the observed behaviour of ants but also of several social systems, such as economics (see also Ref.[15]). In the economics context, rule two above implies that individuals do not always act independently but can be influenced by others, while rule three implies that individuals can change their minds. Such behaviour of individuals is not incoporated in older economic models but is taken into account in newer agent-based simulations which mirror more realistically the non-equilibrium world. The remarkable collective intelligence of ants has also inspired computer scientists to develop better algorithms, such as those for rerouting traffic on a busy telephone network, using ideas abstracted from ant behaviour: See exercise (8).