Abstract
Background. “Bubbles” is a relatively novel psychophysical technique (Gosselin & Schyns, 2001) that effectively isolates the information in a stimulus most critical for its perception and categorization. In this paradigm, observers make forced-choice discriminations on stimuli in which small, randomly selected regions are visible from within a heavily masked image. Performance increases when the visible region includes informative most salient to the task. After many trials, the data reveal a pattern of differential saliency of information across the stimulus. Although this technique has been used extensively with static images, particularly faces (Schyns et al., 2003), it has not yet been extended to the study of dynamic events. The current study measures the temporal saliency of visual features of point-light biological motion. Method. Point-light “bubbles” animations are constructed by smoothly inserting randomly-selected temporal intervals (667 ms) of a biological motion walker into a 3 sec sequence of motion-matched noise. Observers discriminate the presence or absence of biological motion on each trial. Performance is maintained near threshold using a double-interleaved staircase that manipulates the number masking noise dots. Trials are sorted on the basis of observer performance (hits and misses), and frequency distributions of the frames visible in those trials are generated. Results. Our results reveal a distribution of information saliency across the gait cycle of a point-light walker. Optimal sensitivity approximates a sinusoidal function with peaks corresponding to the temporal intervals with opponent motion (e.g. those in which arms and legs cross). Conclusion. These results highlight the importance of mid-level optic flow features and support the hypothesis that opponent motion is a critical component for detecting point-light biological motion (Casile & Giese, 2005). This is the first known implementation of “bubbles” with dynamic stimuli and illustrates a technique that can be extended to other instances of dynamic event perception.