Peripheral vision and action recognition
Laura Fademrecht, and
Humans are social beings that need to interpret the behavior of others in order to get a sense of how they are feeling and to interact appropriately with them. Much of the action related visual information that impinges on the retina falls outside the fovea. This peripherally presented action information is critical for understanding for example crowd behavior. Relatively little is known about the human ability to recognize social actions in the visual periphery.
Our aim is to get first insights into the perception of social actions in the periphery. We investigate the influence of emotional valence and the viewers perspective on the recognition of actions in far periphery and foveal vision.
Figure 1. The wide screen display used in the experiments covering 230° of the visual field horizontally and 125° vertically.
A life-size stick figure avatar carried out one of six motion captured actions, displayed on a panoramic screen. During the experiment participants had to fixate a cross in the middle of the screen while the stick figure appeared at different positions (0°, 15°, 30°, 45° and 60° eccentricity either left or right) in the participants visual field. Participants either evaluated the emotional valence of the action or categorized them on a basic level of recognition or on a sub-ordinate level of recognition. In a second study the viewers perspective was varied from second-person perspective to a third-person perspective.
These studies showed that even at far eccentricities up to 60° participants were highly accurate in recognizing the actions of another person (Figure 2). A linear decline of accuracy with eccentricity was found in the sub-ordinate level task. In the basic level and the valence evaluation tasks accuracy did not decrease up to 45° eccentricity. A higher accuracy for positive than for negative emotional valence was visible in the sub-ordinate level task. The measured reaction times indicate that valence evaluation is superior to semantic categorization, leading to shorter reaction times than both semantic tasks. Surprisingly the third-person perspective led to shorter reaction times than the second-person perspective, presumably due to the choice of actions that were more visible in the third person perspective.
Figure 2. Reaction times as a function of recognition task and eccentricity.
The recognition of the emotional valence occurs faster than the semantic categorization of the action, which contradicts the results of Calvo et al. in scene categorization in the periphery . Furthermore, our results clearly show that under more ecological viewing conditions the recognition of another persons actions is possible even in far periphery using life-size stimuli.
1. Calvo, Avero, Nummenmaa (2011) primacy of emotional vs. semantic scene recognition in peripheral vision, Cognition and emotion 25 1358-1375.
PhD student at the Max Planck Institute for Biological Cybernetics, Germany
|2007 - 2012||
Studies of Technical Biology at the University of Stuttgart, Germany;
Grade: German Diplom (equivalent to a Master's degree)
|1998 - 2007||Bildungszentrum-Reutlingen -Nord, Germany; Abitur (University entrance qualification)|
Organizational Unit (Department, Group, Facility):
- Alumni of the Department Human Perception, Cognition & Action
- Alumni of the Group Recognition & Categorization
- Alumni of the Group Social & Spatial Cognition