Application of signal detection theory to error detection in ballistic motor skills.

Abstract

Signal detection procedures were used to measure subjects' ability to detect errors in performance as they tried to hit a target in a ballistic motor task, since current methods were found to have several important shortcomings. Two tasks were used; one involved moving a slide 24.1 cm in 150 msec (temporal target), and the other involved rolling a ball at a visible target (spatial target). Subjects in both tasks were poor at detecting errors. Variability in the perceptual processes, measured by signal detection proceudres, was found to be about twice as much as that in the motor processes, measured by calculating variable error. This finding was inconsistent with a strictly closed-loop model of motor skills. Signal detection theory was also able to fill a gap in present theories of motor learning by explaining why some of the subjects in this experiment were able to detect errors better in the absence of knowledge of results.