A computational frame to study social behaviour in animals

This paper presents new methods and procedures for studying collective behaviour in rats. The animals are assumed to be indistinguishable one from another and the behaviour of the group is represented analysed and interpreted in terms of the temporal evolution of a finite state probabilistic automaton. The automaton states are defined by measures on the clustering degree considered as a social response variable. The electronic system developed to carry out the cluster analysis and the automatic control of the social behaviour in the experimental environment includes a multimicroprocessor interacting with a 'social box' in which, together with classical sensors and effectors, a phototransistor based position sensor is included. Preliminary experiments show the discriminative power of the cluster automaton concerning sexual differences and emotivity, as well as the extensive of a basic mechanism of clustering as a collective response to stress. Pharmacologically, the new experimental medium proposed in this paper may be used to detect a new range of products affecting social but not individual behaviour. Also, well-known products, which in normal doses produce no detectable modification of individual behaviour, might have detectable effects on the collective level. Be this as it may, the experimental environment described constitutes a further experimental facility for the analysis and control of animal behaviour.