B.M. Spruijt and J.B.I. Rousseau

Department of Medical Pharmacology, Rudolf Magnus Institue for Neurosciences, Utrecht University, Utrecht, The Netherlands

Studies on animal behaviour are used to gain more insight in brain-behaviour relationships. Animal behaviour is often used as model for the development of drugs for human psycho-pathology such as anxiety, stress, depression, cognitive deficits, etc. In contrast to the complexity of human behaviour accompanying such disorders the animal models suffer from simplicity. One of the limiting factors of developing more adequate animal models is the complexity of behaviour for which, apart from our own eyes and observer abilities, no other tools are available. A reliable registration and analysis of complex motor movements related to time exceeds the abilities of a human observer.

This prompted us to develop automated observation and analysis instruments by using digitized video displays. Behaviour consists of postures, changes in postures and movements. Our first attempts were aimed at distinguishing simple postures related to grooming and resting. Only very simple postures could be processed. It appeared that movements in terms of x-y coordinates can be detected and quantified relatively easily as well. The resulting walking pattern can be characterized by a variety of aspects which are usually not easily quantified by eye, such as length of the path, shape, velocity, bouts of moving and not moving. We have applied the analysis of moving patterns in open field behaviour of rats in a variety of experiments. The swimming pattern of rats in the Morris maze has also been subjected to detection and analysis by a computerized system.

Most animals, but certainly rodents, do show a extensive repertoire of social behaviours. The walking patterns of two interacting animals are even more difficult to follow, but do contain a lot of information as the animals continuously affect each other's behaviour. We have developed an algorithm to assess the approach and avoidance behaviour of rats and to distinguish movements directed to each other from general exploratory movements. This procedure has been used to study effects of isolation early in life and treatment with opiates at specific time on the animal's motivation to seek social contact. The results demonstrated that these tools did confirm the more qualitative data obtained by a human observer.

Rats tend to establish a territory in which dominant and submissive animals can be distinguished. Especially intruders in such a territory are usually approached with aggression. This is known as the so-called resident intruder paradigm. The complexity of moving patterns of various animals which have different relationships with each, other exceeds the situation of two animals.

One of the major issues emerging from using complex algorithms is that the output of a computerized system is the result of complex calculations and setting a variety of variables; eventually the final data escape the intuitive estimation of their relation to the behaviour as we perceive it. By making complex situations, which are biologically most relevant, accessible for reliable scientific analysis a set of new problems are introduced. Every situation has to be evaluated and the appropriate settings of variables has to be determined. The user does not only need skills with respect to observing and interpreting animal behaviour, but also with respect to relating well-known behaviours to newly defined parameters, which represent movements or in the near future postures, not necessarily matching the behaviours one is familiar with. However, if we want to extend the range of relevant animal models and increase the reproducibility of behavioural studies this is an inevitable and inspiring way to go.

Paper presented at Measuring Behavior '96, International Workshop on Methods and Techniques in Behavioral Research, 16-18 October 1996, Utrecht, The Netherlands