Abstract
We have previously shown that, when deciding which of two target locations to fixate, the only visual information used is the initial 60–100ms after the stimulus is presented (Ludwig et al., 2005). This is true despite the optimal strategy being to continue accumulating evidence until the target has been chosen. In this study, using fixations to Gaussian blobs that are spatially rather than luminance jittered, we instead investigate the temporal properties of the system that decides where to fixate. Here instead of finding a single, stimulus onset locked system, we find strong evidence for two separate systems being involved in the decision over where to fixate. The first has essentially identical properties to the system that decide which location to look at: it is locked to the stimulus onset, and has an impulse response of about 65ms. The second and more important system, integrates, not relative to the stimulus onset, but to information around 135msec before the saccade is made. Using a second Posner type pre-cuing experiment, we found an effect of the pre-cue on the stimulus onset based system, but not on the saccade locked system. We interpret this pattern of results in terms of a two stage process of eye movement control: firstly, a high level system decides which the general region of interest is. Only after this decision is made, is a fast, low level system used to choose exactly where within this region to fixate. We identify the first mechanism with information passing through the Basal ganglia, and the second with more direct information coming to the superior colliculus via V1 and MT.
This work was funded by an EPSRC grant (REVERB).