thinking is dangerous

Stigmergy and social interaction

I've found the buzzword I've been looking for. I've also found the people who have been doing research in my area, and they are all in Northern Europe. I wonder if it's too late to move to Sweeden.

The term 'stigmergy' was created by Grasse in the late 50's, from the Greek stigmos meaning 'pricking' and ergon, meaning 'work'. He was studying ant and termite behavior, and ran headlong into the so-called "coordination paradox"

The concept of stigmergy provided an alternative theory for understanding the coordination paradox, i.e., the connection between the individual and the societal level: looking at the behaviour of a group of social insects,they seem to be cooperating in an organised, coordinated way, but looking at each individual, they seem to be working as if they were alone and not involved in any collective behaviour.

Grasse was looking for "a class of mechanisms that mediate animal-animal interactions", which was severely lacking from the scientific repertoire. The only tool available were analogies drawn to the functioning of an organism in terms of its individual organ systems, but this had no explanatory value, and in fact suffered from the same coordination issues. The alternative was to merely describe the individual agents with no respect to their interactions. This view was advocated by Rabaud, who was generally skeptical of holistic explanations.

The focus on individual behaviour had a tendency of oversimplifying the nature of social phenomena, and Rabaud claimed that the only cause of behaviour lies within an individual, and "if cooperation occurs it is only by chance and as a result of unexpected incidents" (Theraulaz & Bonabeau, p. 99). According to Rabaud each individual was doing its own work, without paying any attention to the work of others, and therefore they had no noticeable influence on each other. Rabaud considered collective work as merely a "juxtaposition of individual works", and that "common work is no more than a side effect of interattraction that gather individuals together" (ibid., p. 100).

However bad this view turns out for the human case, it was even worse for the apparently more simple case of ant and termite colonies. However, Rabaud's work was not entirely unhelpful, and in typical 50's behaviorist fashion his work relied on the central notion of interaction.

However, the work of Rabaud led to the introduction of two important concepts: interaction and interattraction. Interaction is the reciprocal action where one individual's action may influence and modify the behaviour of another individual. The term of interaction formed a bridge between the individual and the social level. Interattraction means that animals belonging to a social species are attracted in a specific way by other animals belonging to the same species. These ideas were further developed by Grasse, whose basic idea was that "sociality is not a trivial consequence that results from interattraction, but a biological characteristic deeply rooted in the ethological heritage of every species" (ibid., p. 101). The action of an individual can provide a stimulus for other individuals, who respond with another action, triggered by the previous action. In termite nest building, for example, the existence of an initial deposit of soil pellets stimulates workers to accumulate more material through a positive feedback mechanism, and each worker in turn creates new stimuli as a response to the stimulating structure. This allows complex structures, such as pillars and arches, to emerge without central coordination. Thus each individual, or the result of its work, can act as a direct source of stimuli for other individuals. In addition, this mechanism allows for an indirect coordination of individual activities as each individual's activities organise the environment "in such a way that stimulating structures are created; these structures can in turn direct and trigger a specific action from any other individual from the same species that comes into contact with them. Chemical trails that are produced by some ants species..., muleteer trail networks, and even dirt tracks and trail systems in man... result from interactions of this kind" (p. 102).

The mediating mechanism for social interaction, then, was not to be found in the individual but in the environment itself that is structured by the individual participants for group coordination. This view of ant behavior has become the standard view, but the source of this view is often under appreciated. In this way we solve the coordination paradox, through indirect communication.

The basic principle in stigmergy states that traces left and modifications made by individuals in their environment may feed back on them and others: activities are partly recorded in the physical environment, and this record is used to organise collective behaviour. As the examples show, various kinds of storage are used: chemical traces, building material, spatial distribution of elements, etc. Thus individuals do interact to achieve coordination at the societal level, but they interact through indirect communication, and therefore, looking at each individual, they do not seem to be engaged in coordinated, collective behaviour. In sum, stigmergic explanations of social insect behaviour consider the agents as simple creatures, simple in the sense that without deliberation they (re-) act or respond according to stimuli provided by other individuals and/or the environment.

I found this while looking through the evolutionary robotics literature for information on the distinction between proximal and distal explanations of functional organization. The paper I cite is by Susi & Ziemke (2001) entitled "Social Cognition, Artefacts, and Stigmergy" (PDF). More information can be found on this disappointingly low-tech website on stimergic systems, which references everything from Lingo to Google's patent.

'Stigmergy' is an ugly, awful word. It doesn't roll off the tongue, it must be scraped forcefully. But the concept itself is exactly what I have been looking for. I am now opening discussion for any suggestions on what would be a better term.
18:45 :: :: eripsa :: permalink