Figure 5. Sampling allocation as measured by the gaze time on the first pattern.

Gaze time on the first pattern includes all re-fixations and is expressed as a proportion of the overall gaze time on both patterns. In other words, this variable corresponds to the proportion of available sampling time (excluding the initial saccade latency and the duration of any subsequent movements) spent on the pattern that was cued first. Data are averaged over eight observers and the error bars display within-subject standard error [23]. (a) Gaze time contingent on the coherence of the first and second patterns. (b) Gaze time contingent on the accuracy of the perceptual decision. (c) Gaze time contingent on the number of switches.