Identification of variability in behavioural response types in mice — ASN Events

Identification of variability in behavioural response types in mice (#642)

Marloes H. van der Goot 1 , Hein A. van Lith 1 , Frauke Ohl 1
  1. Utrecht University, Utrecht, UTRECHT, Netherlands

Individual differences in response in experimental behavioural research are frequently written off as unfortunate noise. We here suggest that part of this variation may provide useful information on variation in adaptive response types to challenging situations.

Retrospect analyses of behavioural data in laboratory mice suggest that it is possible to identify subgroups within species or even strains/stocks that follow a similar response pattern over time in the same task. By applying statistical approaches such as hierarchical clustering and multiple nonlinear regression we aim to provide a tool statistically to analyze individual response patterns over time and as such provide a new avenue for the analysis of inter-individual variation.

Preliminary results on behaviour in two mouse inbred strains (BALB/cJ, N=16; 129P3J, N=16), tested repeatedly over twenty trials in an initially unknown environment [the modified Hole Board setup], revealed two main response patterns: one that showed adaptation to this stressful situation over time and one that showed sensitization as the trials progressed. Interestingly, by analyzing individual variation, we found that these two response types were displayed by individuals of both strains of mice, suggesting that these patterns appeared to surpass inbred strain differences in response.

Employing this approach within the field of laboratory animal research not only provides a new take on the large interest in individual differences and variation, but it may also offer a method that allows for addressing the biological function of variation in behavioural responses within groups of animals: For example the stability of a social group may be related to its composition in terms of behavioural response types. We are currently applying this approach to larger datasets to elaborate on this finding and to optimize the statistical methods.