Abstract
Humans can initiate ultrafast saccades towards face as early as 100ms post-stimulus onset (Crouzet & Thorpe, 2010, J Vis). However,other object classes, such as cars or animals, have slower mean saccadic reaction times and lower overall saccadic accuracy in comparison to faces. Based on the theory of Spike Timing Dependent Plasticity, which predicts that local input statistics can drive the learning of representations in the visual hierarchy, we hypothesized that an RSVP-based training paradigm, in which a large number of different stimuli could be presented per unit of time, might lead to stimulus-driven plasticity in lower visual areas which would improve saccadic speed and accuracy towards the trained object class. To test this hypothesis, subjects were shown 12hz RSVP streams of cluttered images containing randomly embedded 2° cars at 8 different positions, each at 9° eccentricity. Following 10 minutes of RSVP training, saccade latencies to car targets showed a strong and selective decrease, from a mean of 216ms pre-training to a mean of 167ms post-training (p< 0.005). This speed-up in car localization was not the result of a speed-accuracy tradeoff, as accuracy on car detection likewise increased (from 55% to 66%). We also simultaneously recorded EEG while the subject performed the localization task, and found that post-training, target location for cars could be predicted starting at 81ms after stimulus onset. These results suggest that it is possible to rapidly train neural representations in early visual areas which effectuate shortcuts in the visual hierarchy.
Meeting abstract presented at VSS 2015