Control in living systems : an exploration of the cybernetic properties of interactive behaviour

Abstract

Many types of animal behaviour, especially seemingly complex social interactions, have been attributed to the existence of complex cognitive mechanisms, underpinned by stimulus-response (S-R) rules. Indeed, as speci c behaviours are analyzed in greater and greater detail, the increasing number of minor variations observed, even under tightly-controlled experimental conditions, seem to necessitate the operation of increasingly powerful computational devices. An alternate view, inspired by cybernetic theory, is that what is important is not the speci c behaviours used by animals, but the goal of the organism in a particular context. In this thesis, a closed-loop cybernetic methodology for understanding behaviour is developed and implemented. Evidence is presented that, not only do at least some behaviours of animals function like engineered control systems, but also that this type of architecture is widespread in phylogenetic terms, relatively robust to interference, and able to be arti cially reproduced. Implications for the study of the behaviour of all organisms are discussed.