Abstract
A period of stable fixation is used to sample the visual field in order to decide where to fixate next in the scene. We used a reverse correlation paradigm (Caspi, Beutter, & Eckstein, 2004) to elucidate when the critical signals driving the eye movement decision occur.
Observers were presented with two dynamically modulating Gaussian blobs. The luminance of the patches was resampled at 40 Hz from two, partly overlapping, Gaussian distributions that differed in their means. The task was to saccade to the patch that was, on average, of higher luminance. We analysed the average luminance noise (at both the target and the distractor location) preceding the oculomotor decisions.
Our findings can be summarised as follows:
Error saccades occurred when, during some temporal interval, the distractor luminance was high or the target luminance was low.
This temporal interval appeared to be time-locked to the first 100 ms after display onset, regardless of the latency of the eye movement.
For correct saccades to the target there was a small, but consistent relationship between the average luminance difference between the target and distractor during this interval, and saccade latency.
These results suggest that decisions within the oculomotor system are based on visual sampling in a restricted time window. This sampling window appears to be relatively constant and unrelated to the variability in saccade latency. We propose that the sampled signals feed into an oculomotor decision stage. The major sources of saccade latency variability appear to originate from this decision mechanism and a subsequent dead time during which the metrics of the upcoming saccade can no longer be altered.
Caspi, A., Beutter, B. R., & Eckstein, M. P. (2004). The time course of visual information accrual guiding eye movement decisions. Proceedings of the National Academy of Sciences, 101, 13086–13090.