Abstract
Action observation is a visual function of a great importance from both the ethological and social point of view. Recently, a number of studies provided new insights into its functional organization as a three-level cortical network encompassing in human and non-human primates occipito-temporal, parietal and premotor regions. However, there is still no general framework which would allow to establish a relationship between neuronal activity in these areas and its behavioral correlates. Demonstrating that general psychophysical laws are also applicable to the visual processing of observed actions could provide us with such a framework. We reasoned that changing the amount of dynamic noise in action movies would produce behavioral responses in human subjects qualitatively similar to the classical psychometric curves. To test this hypothesis, we presented human subjects (n=4) with the movies (2 sec) in which they had to discriminate between the two different hand-actions (rolling and rotation) in a two-alternative forced-choice task. The movies were randomly presented in 5 different fronto-parallel positions and at 2 depths. On every frame in each movie, a certain percentage of random dot-pixel pairs separated by a distance randomly chosen from within a fixed interval were scrambled. By manipulating the percentage of scrambled dot pairs, we created 6 noise levels from 60% (i.e. 60% of dots on each frame were scrambled) to 100%. Our data indicate that the amount of noise in action movies attenuated the ability of our subjects to discriminate between the two actions tested, such that the observers’ performance could be described by the classical logistic regression. This psychometric curve suggests that action discrimination follows general rules described in classical visual psychophysics. One implication is that by changing the input strength in action movies can be used to manipulate the activity in single cells and neural populations.
Meeting abstract presented at VSS 2015