Consequences of error production in a perceptual-motor task

In 2 experiments, U.S. college students engaged in a computerized version of a tracing task in which a cursor on a monitor screen could be guided through 16 segments of an 8-pointed, star-shaped track into targets at the vertices of the track by moving a joystick handle. The presence or absence of consequences for errors was manipulated. With restricted consequences, the cursor was confined to the track, freezing in position if the joystick handle was moved to an unsuitable position. With unrestricted consequences, the cursor always moved corresponding to the position of the joystick handle, even through and beyond the borders of the track. Time to reach each of the 16 successive targets was recorded on each of 4 training trials, as well as on 2 transfer trials in which consequences were unrestricted, but only the cursor and next target were visible (the contextual cue of the star-shaped track was absent). Times were longer on diagonal segments and when consequences were unrestricted, but there was no consistent carryover to the test trials. The results suggest that feedback, to serve an effective "guidance" function in skill learning, requires production of unrestricted feedback about errors. True guidance, however, prevents response errors, so consequences and feedback about errors are precluded.