Quantifying the Effect of Multiple Neurotransmitter Systems on Group-Level Animal Behaviour through Machine Learning

This research project contains four major novelties: (a) We aim at automatically assessing sets of high-dimensional data of behaviour at high spatial and temporal resolutions, allowing to break down the neurochemically induced pathology to the key building blocks of behaviour. (b) We envisage to extrapolate from the behavioural screening of individuals to the group level, addressing potential impacts of neurological and psychiatric disorders on group dynamics and social interactions. (c) This research programme integrates the most modern approaches from animal behaviour and ethology with those of neurotoxicology and translational medicine, by employing algorithms that arose of the most recent research efforts in machine learning. (d) The behaviour of fish shoals will be described in a biologically relevant context, increasing the ecological validity, i.e. we employ tasks that emphasise social aspects of group behaviour, such as social learning in foraging situations or exposing the group to novel situations. This research project is of significance for a variety of domains and finds missing and interesting links between these domains by integration of biological, psychological, medical (translational) sciences and machine learning, known from artificial intelligence research. The research aims at the next logical step of measuring and quantifying behavioural consequences of neurotoxic substances, i.e. the level of social interaction, the manifestation of abnormal behaviour in the group.