Behavioral consequences of perceptual decision-making in oculomotor brain structures

Abstract

Neurons in several oculomotor brain areas (such as the lateral intraparietal area and superior colliculus) carry a multitude of signals. Even within a single task, individual neurons may exhibit responses related to visual/sensory events, decision formation, and saccade execution. Little is known about whether these signals interact and if such interactions have functional consequences. To test for interactions between decision-making and saccade execution signals, we used a behavioral dual-task paradigm that measured saccade reaction time (SRT) to a decision-irrelevant target while observers conducted a motion direction discrimination task. Observers (n=5) judged the direction (leftward versus rightward) of a random dot motion stimulus (of variable coherence; speed: 5 °/s). At the offset of the dots, a target appeared either left or right side of the fixation point (displaced by 20°). Observers were instructed to make an eye movement to the target as quickly as possible. After the eye movement, the direction of preceding dot motion was reported with a button press. We tested whether the SRT to the target was affected by the direction and strength of motion stimuli presented during the direction-discrimination task. Indeed, SRTs were faster on congruent trials (when the direction of motion was same as the saccade direction) compared to incongruent trials (F1,4=26.23, p<0.01). Furthermore, SRTs progressively decreased with increased coherence (F4,16=5.45, p<0.05). Thus, simple visually-guided saccades were affected by the ongoing buildup of activity related to decision formation, in a manner that was direction-selective and parametrically affected by motion strength. Our results are not due to the mere presence of motion stimuli because the dependencies on coherence and directional congruency were eliminated in a single-task (saccade only) control experiment. Revealing such interactions between decision making and saccade executions suggests that these two types of signals cannot be completely demixed in neural circuitry. Meeting abstract presented at VSS 2014