A separate repeated measures ANOVA on the P3 peak latencies of the first target with Task as within-subjects factor did not reveal any significant difference between the tasks (F<1).
The difference between the tasks may be related to the horizontal/vertical geometry of the stimuli, which makes the landscape and line bisection tasks more prone to universal effects related to cerebral dominance.
The IC-F found in our present study had a sizeable contribution to the N140, but its contribution did not show any significant difference between the tasks.
Another difference between the tasks is that texture discrimination often needs sleep to improve performance [2], [3], [5] [9] while performance in a bisection task improves already within a session [20], [23].
A mixed model analysis on the slopes and the intercepts of the best fitting lines (three fixed factors: Stimulus, Task, and Direction) proved the difference between the tasks (slopes: F 1,13) = 238.56, p<0.01; intercepts: F 1,13) = 182.39, p<0.01), whereas the Direction factor did not significantly affect the subjects' performance.
The reason for this inconsistency is unclear, but one possible reason is a difference between the tasks for psychomotor slowness.
Another minor difference between the tasks was that the whole-brain MIL task results emerged from a parametric model of prediction error based on the fit of a reinforcement-learning model to choice data, whereas the MID task results were based on a simple comparison of gains to nongains (which does not depend on assumptions about subject choice behavior).
These deviations reflected the differences between the task and the activity, which is standard and known in ergonomics, but also between the planned nominal technical operation and the real operation including different situation recovery events, faults and failures (technical, human, organisational, etc).
There are two differences between the task switch in FR→PR and MS→RO that underlie why we did not include it in the conjunction analysis.
