Abstract
Introduction. When confronted with multiple moving objects the visual system can process them in two stages: an initial stage in which a limited number of signals are processed in parallel (i.e. simultaneously) followed by a sequential stage. We previously demonstrated that during the simultaneous stage, observers are capable of simultaneously extracting direction information, which is bound to its corresponding element, from up to 3 signals (Vis Res, In press). When the number of elements presented exceeds this capacity, a subset of up to 3 can be automatically selected and processed. Here we investigate which element characteristics influence bottom-up saliency, driving this automatic selection process. Method. This was achieved by briefly presenting observers with 8 localized motion signals, each consisting of a group of four dots arranged into a square formation, then post-cueing one of the signals and measuring the accuracy at which observers identified its direction during the presentation. By systematically varying characteristics of the elements presented and examining performance as a function of these, we determined patterns of selection based on the following features: spatial location, direction, speed, and contrast. Results. The results show that when presented with a heterogeneous population of motion signals exceeding the limit of simultaneous processing, observers consistently demonstrated significant selection bias based on these characteristics. Elements presented at greater eccentricity, moving in cardinal directions, moving relatively faster, and of higher contrast are more likely to be selected than their counterparts, i.e. low eccentricity, oblique directions, slower, and low contrast. Conclusions. This indicates that during simultaneous motion processing, the automatic selection of signals is driven by variations in bottom-up saliency resulting from relative differences in both motion-specific characteristics, i.e. direction and speed, and other characteristics, i.e. spatial location and contrast.
Meeting abstract presented at VSS 2014