Results Experiment 3: visual feedback.

Similar to Figure 3 except that participants received either unreliable or reliable visual feedback of their initial pointing position. A,B) shows the perceptual noise for the left/right discrimination of the errors and the motor noise respectively, before and after making secondary corrective movements. The median and 25% and 75% percentiles across participants for each condition are indicated by the bars and the error bars respectively (medians for after corrections approach infinity). The separate points indicate the results for individual participants (filled symbols indicate performance significantly above chance). The dashed line shows the average across participants for the standard deviation in initial random pointing error . C) Behavioral gain vs. the optimal gain for each participant for the two correction conditions with visual feedback. With more reliable visual feedback correction gains generally increase as predicted. D) Behavioral endpoint versus initial error ratio vs. the predicted error ratio. With more reliable visual feedback end-point variance is reduced the most.