Cognitive influences on predictive saccadic tracking

We have investigated the effects of mental set on predictive capabilities using a saccade square-wave tracking paradigm with ten normal subjects, comparing three amplitudes (10, 20, and 40°) and five inter-stimulus intervals (ISIs) (400, 500, 625, 1000, and 2000 ms). Subjects were instructed simply to "follow the lights" (passive, reflexive instruction) or explicitly "move your eyes in time with the lights" (active, volitional instruction). Saccades were defined as reflexive (latency>100 ms), predictive (-200 ms<latency<100 ms), or anticipatory (latency<-200 ms). We also calculated arrival time (saccade latency+saccade duration). Instructions had a striking effect on predictive performance. The effects were greatest with the longest ISIs (1000 and 2000 ms) and the largest target displacement (40°). With the active instruction there were more predictive and anticipatory saccades and with the passive instruction more reflexive saccades. Furthermore, with the active instruction subjects could take into account the duration of the impending saccade so that the eyes would arrive closer to the appearance of the target no matter what the amplitude of the required saccade. In sum, cognitive set, as determined by the specific instructions given to the subject, has a striking effect on predictive saccade behavior, which has important implications for interpreting physiological and imaging correlates of predictive behavior in normals and in patients with neurological disease.