Fig. 5.

Classification of error correction patterns. Shown are patterns of overshoot error corrections when conditions changed from load to no load (similar patterns for undershoot error corrections when changing from nonloaded to loaded conditions). A: pattern 1. The error in the test (T) trial (white bar) is eliminated in the next trial (P1; light gray bar), and behavior is consistent in all subsequent control (C) trials (black bars) within the same block. B: pattern 2. The error in the T trial is partially corrected in the first posttest trial (P1) and eliminated in the next trial (P2; dark gray bar). Behavior is consistent in subsequent C trials within the same block. C: pattern 3. The error in the T trial is completely corrected by the second posttest trial (P2), but the behavior is inconsistent in subsequent trials (more than one error in C trials). D: pattern 4. No stable error correction behavior occurs in a block of trials. In A–D, the horizontal lines show the limits of the 6° wide target window.