Trial by trial effects in the antisaccade task.

The antisaccade task requires participants to inhibit the reflexive tendency to look at a sudden onset target and instead direct their gaze to the opposite hemifield. As such it provides a convenient tool with which to investigate the cognitive and neural systems that support goal-directed behaviour. Recent models of cognitive control suggest that antisaccade performance on a single trial should vary as a function of the outcome (correct antisaccade or erroneous prosaccade) of the previous trial. In addition, repetition priming effects suggest that the spatial location of the target on the previous trial may also influence current trial performance. Thus an analysis of contingency effects in antisaccade performance may provide new insights into the factors that influence the monitoring and modulation of the antisaccade task and other ongoing behaviours. Using a multilevel modelling analysis we explored previous trial effects on current trial performance in a large antisaccade dataset. We found (1) repetition priming effects following correct antisaccades; (2) contrary to models of cognitive control antisaccade error rates were increased on trials following an error, suggesting that failures to adequately maintain the task goal can persist across more than one trial; and (3) current trial latencies varied according to the previous trial outcome (correct antisaccade, slowly corrected error or rapidly corrected error). These results are discussed in terms of current models of antisaccade performance and cognitive control and further demonstrate the utility of multilevel modelling for analysing antisaccade data.