Using Machine Learning to Discover Latent Social Phenotypes in Free-Ranging Macaques

Area of Science:

• Behavioral Ecology
• Computational Biology
• Genetics

Background:

• Social phenotypes are complex and high-dimensional, making them difficult to study using traditional methods.
• Understanding the biological underpinnings of social behavior requires analyzing patterns rather than isolated actions.

Purpose of the Study:

• To develop a machine learning model for quantitatively defining social phenotypes from naturalistic observational data.
• To identify and measure behavioral patterns and their relationships to biological, environmental, and demographic factors.

Main Methods:

• Applied a machine learning model to extensive naturalistic observations of free-ranging rhesus macaques.
• Identified distinct behavioral states representing social interactions like isolation, competition, conflict, and coexistence.

Main Results:

• Social phenotypes, defined by the rate of occurrence of identified behavioral states, were significantly influenced by dominance rank, sex, and social group.
• Two specific behavioral states showed a substantial genetic component in their variation.

Conclusions:

• The developed machine learning model effectively captures complex social phenotypes and their biological influences.
• This approach provides a framework for investigating the genetic architecture of social behavior and its evolutionary basis.