Individual differences are an essential property of all living things, and personality provides a unique glimpse into the biology underlying behavioral variability. And yet, because of the lack of a systematic approach to personality, most works on animal personalities still end up examining a limited subset of subjectively chosen behavioral readouts. Here, we show how personality can be inferred directly and objectively from high-dimensional natural behavioral space. While this approach is not species-specific, we have demonstrated it on mice as it is one of the most common model animals. The mice were videoed over several days, and their behavior automatically analyzed in depth.
Altogether, the computer identified 60 separate behaviors such as approaching others, chasing or fleeing, sharing food or keeping others away from food, exploring, or hiding. We found the mice personalities by working backward from behavior and extracting the underlying traits that differ among individuals while being stable over time and across contexts. We validated that traits found this way (which we term identity domains) were stable across social context, do not change with age, explain the variability in performance in classical tests, and significantly correlates with gene expression in brain regions related to personality. Expanding this method to human behavior, by using location and physiological data from cellphones and smartwatches, revealed a highly structured personality space which resembles that of the mice. This method allows for better informed mechanistic investigations into the biology of individual differences, systematically comparing behaviors across species, as well as develop more personalized psychiatry.
- Oren Forkosh*, Stoyo Karamihalev*, Simone Roeh, Uri Alon, Sergey Anpilov, Chadi Touma, Markus Nussbaumer, Cornelia Flachskamm, Paul Kaplick, Yair Shemesh, Alon Chen. Identity domains capture individual differences from across the behavioral repertoire. Nature Neuroscience, 2019
- Shemesh, Yair*, Oren Forkosh*, Sergey Anpilov, Mathias Mahn, Yehezkel Sztainberg, Tamar Shlapobersky, Gili Ezra, Yael Kuperman, Sharon Haramati, Ofer Yizhar, Elad Schneidman, and Alon Chen. Ucn3 and CRF-R2 in the medial amygdala regulate complex social dynamics. Nature Neuroscience 2016
- Yair Shemesh*, Yehezkel Sztainberg*, Oren Forkosh*, Tamar Shlapobersky, Alon Chen, and Elad Schneidman. High-order social interactions in groups of mice. eLife 2013