The Arithmetics of Mutual Help
Martin A. Nowak, Robert M. May, and Karl Sigmund
Scientific American, June 1995, Vol. 272, No. 6, pp. 76-81.
Computer simulations of agent (i.e. intentional entities) dynamics reveal that the emergence of stable
cooperative behavior may be reliably expected if the
agents interact repeatedly; are able to identify one another; and recall what transpired in past meetings. Additionally, the agents concerned must not discount future benefits too much in preference for immediate gains. That is to say, the agents must not be too temporally myopic.
The behavioral constraints mentioned work well for organisms of sufficient information processing sophistication. However, what may be considered cooperative behavior has also been discerned among interactants that cannot even be classified as agents (e.g. lower invertebrates) because, for example, they cannot recognize one another or
remember past interactions.
For these cases, what seems to be responsible, the computer simulations suggest, is a fixed
geometry of
interactions such that the interactants only interact with immediate neighbors. There is then no need to identify an interactant or to remember the results of past interactions with them. Indeed, the interactant need not even be alive: Chains of catalysts that, through their chemical interrelations, collectively reproduce themselves would then be the primordial, prebiotic example of cooperative behavior.
Beyond enabling cooperative behavior to spontaneously emerge, a fixed geometry of interactions also fosters a diversity of behaviors: It allows both cooperative and selfish behavior to co-exist. In different contexts, this may explains how it is that populations of hosts and parasites or prey and predators are able to co-exist despite the intrinsic instability of interactions between them.
More summaries about the The Arithmetics of Mutual Help