Abstract
Purpose: To study how attention influences low-level motion processing using the motion repulsion paradigm. Methods: 2 superimposed random-dot patterns (RDPs) drifted in different directions. Observers indicated the direction of one or both motions. Single-vector measures: the direction to be measured was cued in advance (attended) or prompted after the trial (divided attention). Double-vector measures: two independent pointers were used to indicate both directions at once (divided attention). Single-vector measures were also made of the unattended motion: subjects monitored the motion of one RDP to detect randomly inserted luminance changes but indicated the direction of the other, unattended motion. Noise was introduced by varying the RDP coherence level. Perceived direction of one-vector stimuli was also explored: one RDP drifted while the other was present but stationary. Results: When attending only to one motion, repulsion is greater for the other, unattended motion. Divided attention conditions produced greater motion repulsion than cued single-vector measures. Adding noise decreased the attentional influence on motion repulsion. One-vector stimuli too were systematically misperceived, being slightly attracted towards the cardinal axes. Conclusions: Low-level motion mechanisms presumed to underlie motion repulsion are influenced by attention. The attentional effects are broadly consistent with gain-based models of attention. The standard mutual inhibition model of motion repulsion, combined with a gain-based attentional model, can account for most of these results. The data concerning divided attention, however, do not fit this scheme. Also, the misperception of one-vector stimuli shows that motion repulsion is not entirely due to inhibitory interactions between direction-selective units.