To what extent can social structure result from evolutionary processes, as opposed to deliberately organised by a collection of intelligent individuals? That is, could social organisations (which are usefully identifiable as an entity in their own right) result from a mechanism which only relies on the processes of random variation and selective reproduction. Or, on the other hand, is it necessary for the individuals to have acquired (through evolution or design) sufficient cognitive abilities to deliberately organise themselves into such entities (using planning, reasoning, deliberate experiment, anticipatory learning or the like).
The question is important for the answer (w.r.t. any particular system) makes a difference. For example, in the former (evolutionary) case the abilities of the individuals can co-evolve with the social structures that they can (unthinkingly) sustain so that some of their features (e.g. a propensity towards cooperation or a distrust of strangers) could have evolved to reinforce or build upon the existing social structures and so boot-strap the complex society we inhabit. If this is not the case then we must understand the particular social structure as essentially intended institutions built subsequent to the abilities developed in other circumstances. This might be important in either understanding observed societies or producing/managing new ones (including artificial ones, such as Multi-agent systems).
In a sense this question is the complementary half of the Machiavellian Intelligence theses that our intelligence evolved (partly) because it provided an ability to manipulate social structures and situations to our (individual) advantage. It can be seen as a part of the, more general, Social Intelligence Hypothesis [2], which is that suggestion that our intelligence gives us evolutionary advantage via the social structures it enables. An example of this is the ability to imitate [5] which may allow groups of people to develop a culture of skills, rules and traditions that equip that group to inhabit specialist ecological niches [20].
Of course evidence has the primary role as far as observed social structures are concerned. For example, the apes can not sustain as sophisticated social structures as humans as so the present abilities of humans that go beyond those of apes must be crucial for their production and/or maintenance. However we do not have strong evidence about how our present abilities developed over biological evolution, there might have been a whole succession of structure then ability then structure etc. which resulted in the current position. Clearly new social structures and institutions have arisen over a historical timescale and thus it seems clear that some of our social structures rely upon our present abilities.
On the other hand, some social structures clearly do not require complex cognition by its individuals. Slime molds organise themselves (under the right conditions) into what can be meaningfully called a single entity (a fruiting structure). Since they, as individuals, have no (or only simple) information processing abilities, some social structure does not require complex cognition.
Thus the question should be rephrased into: which mechanisms might result in what social structures under which conditions?
Computational simulation/systems can help answer this question by establishing some possibilities in terms of the causal connection between mechanisms, conditions and resulting structures. That, is by starting to map out the possible triples of: (mechanism, conditions, results). This is not an easy task, for simulations can be deceptive in terms of the robustness of their results against “small” changes in their set-up and, given their intrinsic complexity, we can easily be mistaken as to our interpretation of how and why they produce the results they do.
This paper looks at one stream of computational experiments that start to address this project – they examine how a particular class of mechanism (tag-related mechanisms) can produce group-like structures for cooperative behaviour. In this sense this paper is a mini-survey and prospective concerning these models and their results – models and results in which I have played only a part, the main work being done by others including: David Hales, Rick Riolo and Emma Norling.
Accessible as: