INTRODUCTION: We observed that during mid-adolescence, mice show reduced sensitivity to social reward. Importantly, this effect is not attributable to decreased social investigation or to generalized deficit in reward processing. Quantification of social interactions is complex and susceptible to confounding factors. Human observers may introduce subjective bias, while automated tracking software carries the risk of over- or underfitting the behaviour classification criteria.

AIM(S): Our aim was to establish whether automatic analysis of social contact seeking produces results comparable to those obtained through manual scoring.

METHOD(S): Social contact seeking was assessed in a custom experimental cage (48 x 12 cm) that allowed interaction between two mice separated by a transparent, perforated divider. In automatic analysis using EthoVision, a social interaction was defined as a moment when the center point of the detected subject mouse (black pixel cluster) was present in a predefined zone adjacent to the divider. In manual scoring (BORIS), social investigation was defined as the active pursuit of social contact, with precisely defined onset and offset of the behaviour.

RESULTS: Both methods revealed no age-related differences in total social investigation time, indicating that automated scoring reliably detected behavioural patterns. However, automated analysis registered higher absolute values across all age groups. The results of the two methods were positively correlated, indicating that despite quantitative differences, the overall behavioural pattern was consistent across scoring approaches.

CONCLUSIONS: Human scoring of social behaviour remains the more precise and ethologically relevant approach. However, automated analysis produces results relatively consistent with manual scoring, and resolves the issue of blinded scoring, and is significantly more time-efficient.

FINANCIAL SUPPORT: This work was supported by the Polish National Science Centre (grant number 2016/21/B/NZ4/00198)