Using Machine Learning to Discover Latent Social Phenotypes in Free-Ranging Macaques
© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Investigating the biological bases of social phenotypes is challenging because social behavior is both high-dimensional and richly structured, and biological factors are more likely to influence complex patterns of behavior rather than any single behavior in isolation. The space of all possible patterns of interactions among behaviors is too large to investigate using conventional statistical methods. In order to quantitatively define social phenotypes from natural behavior, we developed a machine learning model to identify and measure patterns of behavior in naturalistic observational data, as well as their relationships to biological, environmental, and demographic sources of variation. We applied this model to extensive observations of natural behavior in free-ranging rhesus macaques, and identified behavioral states that appeared to capture periods of social isolation, competition over food, conflicts among groups, and affiliative coexistence. Phenotypes, represented as the rate of being in each state for a particular animal, were strongly and broadly influenced by dominance rank, sex, and social group membership. We also identified two states for which variation in rates had a substantial genetic component. We discuss how this model can be extended to identify the contributions to social phenotypes of particular genetic pathways.
The authors would like to thank John Pearson, Sam Larson, Ashley Walker, Joel Glick, Josue Negron, and the CPRC staff for their feedback and research support. This research supported by NIH grant 5R01-MH096875-02. The CPRC is supported by grant 8-P40 OD012217-25 from the National Center for Research Resources and the Office of Research Infrastructure Programs of the National Institutes of Health.
This is the final version of the article. Available from MDPI via the DOI in this record.
Vol. 7, No. 7, 91