Figure 5.

All 5 subjects learned. (a) On the ordinate are means and standard errors of launch error (LE) for the 1000 initial control trials. On the abscissa are the same measures for first 200 reversed trials (5 gray symbols for the 5 subjects) and for the last 1000 reversed trials (black symbols). Across all subjects, launch error improved significantly by t-test (p = 0.025) and in late reversed trials was not significantly different from control (p = 0.35). (b) Adjustment (AE) error also improved (p = 0.025) to control levels (p = 0.47). (c, d) Similarly for launch latency (LL) (p = 0.035 and 0.26) and adjustment latency (AL) (p = 0.004 and 0.43).