"Classification trajectories" - ???

Noises from incorrect trials resemble observer's template? Can't just average velocities, so compute likely statistic Weighted average across incorrect trials