Abstract
Certain key stimuli such as animals or human faces can trigger very fast saccades from as early as 100 or 120 ms after image onset (Crouzet et al., J Vis, 2010). Can such fast saccades be seen to other more conventional visual stimuli? Here we directly compared the speed and accuracy of saccades to targets defined by either color or orientation. We presented four circular grating patches at equal distances from fixation for 400 ms. There were four different colors (red, green, yellow and blue), and four orientations (vertical, horizontal, left and right oblique) combined randomly. For each of 8 blocks of trials, participants were instructed to target either a particular color, or a particular orientation. Surprisingly, with targets defined by orientation, performance was essentially at chance. In contrast, when the target was a particular color, performance was very good, and even very fast saccades initiated at latencies as short as 120–130 ms were biased towards the target color. We suggest that this major difference between color and orientation could reflect whether top-down task-related modulation can influence the initial part of the neuronal response. In the case of orientation, although clearly millions of neurons in V1 will code orientation explicitly, if their initial responses are unmodulated by the task, all the orientations would initially be equivalent. In contrast, if top-down effects can modulate the initial responses of color-selective neurons early on in the visual pathways, this could increase effective salience of the target color and allow fast saccades to be triggered. Intriguingly, while performance was particularly good for red, yellow and blue targets, accuracy for green targets was about 30% lower and reliable saccades were not seen until 30–40 ms later. Further experiments are currently being performed to try and determine the reasons for this poorer performance for green.