Purchase this article with an account.
Claire L. Roether, Martin A. Giese; Integration of synergies in visual recognition of emotional human walking. Journal of Vision 2005;5(8):942. doi: 10.1167/5.8.942.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In the domain of motor control it has been hypothesized that the control of complex body movements might be organized in terms of synergies. A synergy is a smaller subset of degrees of freedom (or joints) that are jointly controlled. This raises the question if synergies are also important for the visual perception of body movements. We tested how the information from two synergies, arm and leg motion, are integrated during the perception of human walking with different emotions. In particular, we tested whether the information provided by different synergies is integrated by the visual system in a statistically optimal way.
Method: Using a 3D motion capture system we recorded neutral human gaits, and gaits with four different emotional affects (angry, happy, sad and fearful). By motion morphing (modifying a technique of Giese & Lappe, 2002, Vis. Res. 38:1847) between neutral gait and the individual emotional gaits we created stimuli containing different amounts of information about the emotion categories. We created 3 different stimulus classes by: (1) morphing all degrees of freedom at the same time; (2) morphing only the arm movements, presented in combination with a neutral leg movement; and (3) morphing only the leg movements, combined with a neutral arm movement. Stimuli were presented as point light walkers.
Results: All morphed stimuli look very natural, even if they are composed from neutral and emotional synergies. The amount of information carried by the two synergies varies from emotion to emotion. Consistent with earlier work, the recognizability of the emotions increases when the movements are caricatured in space-time. Recognizability also increases if only individual synergies are exaggerated. Present work focuses on fitting the data with Bayesian ideal observer models to study whether the integration of the information form the two synergies is accomplished in a statistically optimal way.
This PDF is available to Subscribers Only