Quantifying the contribution of structure information in direction discrimination of scrambled walkers

Abstract

It has been shown that it is possible to reliably discriminate the direction of a point light walker even when all the points have been scrambled, provided the trajectories are preserved (Troje & Westhoff, 2006). We investigated the strength of this motion signal by pitting it against a competing motion signal in the opposing direction that did or did not contain form information. To achieve this we split motion-captured 15-point walkers randomly into two subsets of points that contained opposing direction information. In the structure present condition one subset of points maintained their original joint locations and trajectories, whilst the others were scrambled to new joint locations and had their trajectories switched so that they were moving in the opposite direction. In the structure absent condition both subsets of points were scrambled and given opposing trajectories. For the structure present condition we defined the unscrambled points to have the “signal” direction and determined how many unscrambled points were required before direction was reliably reported in this direction. For the structure absent condition we randomly assigned one set of points to have the “signal” direction and measured how many points were necessary before motion was reliably reported in this direction. We performed an experiment involving 12 observers that used an adaptive procedure to determine the thresholds for discriminating walker direction. Results showed that in the structure absent condition thresholds were significantly higher than those for the structure present condition. Contrasting these thresholds gives a measure of the relative weight of the structural information in determining walker direction. In addition to highlighting the importance of structural cues in direction discrimination of point light walkers, the techniques used to generate the displays may also prove useful in biological motion based neuro-imaging work.

Footnotes

 ESRC-MRC, Autism Speaks