Detection of low dose effects of psychopharmaca: Application of a semi-Marcov model to rhesus monkey behaviour

Behavioural testing of psycho-active drugs at lower doses is needed for medical applications and for the study of pharmacological brain mechanisms. Sensitive methods to detect effects on the time structure of behaviour are lacking. We propose a procedure based on a description within the framework of continuous time Marcov chain models. This class of models is generalized to account for constraints on the organisation of a motor pattern including the speed at which it is terminated. This leads to semi-Marcov models with transition rates that quickly increase from zero to a constant level. We describe and compare several models. As an application we consider amphetamine effects on infant rhesus monkeys. One of the models is preferred since, besides giving a good fit, it demonstrates consistent amphetamine effects that can be interpreted as short-term effects on motor constraints. Amphitamine also induced more mother-directed behaviour. We give maximum likelihood procedures for behaviour analysis based on the preferred model.