Movement-Based Human Identity Recognition: Analysis and Hypotheses

Human observers can recognize the identity of a person solely based on how the person moves. Although numerous empirical studies have confirmed this observation, this ability's underlying mechanism still remains unclear. The current work leverages an open-access human movement database to explore aspects of human identity recognition via point-light displays. Methods such as dimensionality reduction techniques (e.g., principal component analysis) and graph theory were used to extrapolate the actors' dynamic identity signature based on high-dimensional movement data. Based on the analyses, several hypotheses were presented and their corresponding behavioral experiments were proposed.