Predictive plus online visual information optimizes temporal precision in interception.

Humans time their interceptive actions with remarkable precision. This daily-life performance is far too good to be explained by reported experimental perceptual estimates of when an object will arrive at the interception location. One option is that people use general principles to reduce variability such as integrating early estimates from predictive mechanisms with late estimates from online vision. Here we explore this possibility by presenting virtual balls that people had to catch and compared 3 conditions: early, late, and full vision of a parabolic path. If people integrate these different estimates, the precision of the timing under full vision should be higher than when only late vision is available. We tested this hypothesis and found a benefit for full vision, but only for those (steeper) trajectories in which early and late estimates are likely based on different cues. Overall, the integration of the different estimates of the impending interceptive event was optimal and can help explain the observed high temporal precision in many daily-life situations. Finally, by revealing the situations in which people do not take into account early predictions and rely on online visual information only we elucidate the theoretical controversy between predictive versus online control of timed actions.