Interference effects in short-term motor memory

Abstract

Five experiments were conducted to determine the characteristics of motor short-term memory. The motor task used in all studies was the application of an isometric force to a control knob in either the upward or the downward direction, though with no actual movement of the control knob. Retention of the applied force was studied as a function of time and of other activities. Experiment I indicated that forgetting did not increase with time, for retention intervals of up to 60 sec. Experiment II found significant improvements in recall, and confirmed the previous observation of Experiment I of an overshooting response set during the recall trials. Experiment II also demonstrated a generally-increased decrement in recall when the retention interval was filled with backwards-counting as opposed to unfilled intervals, which suggested the importance of generalized muscle-tension states in motor memory. Experiment III compared the relative effectiveness of an interpolated counting task with an interpolated force-application task in generating interference using a standard RI design. Interference was noted in the form of a directional biasing. Smaller interpolated forces produced smaller errors than did larger interpolated forces. The results of Experiment IV provided further support for the directional-biasing mechanism, using a PI design. Experiment V investigated the effects of repetition on the forgetting function. The results indicated that with more repetition, performance at recall deteriorates. A consideration of the overshooting response set associated with the force task, in conjunction with a fading trace theory of memory reconciled the contradictory findings. Conclusions integrated the results of all five studies within a dual-factor theory of motor short-term memory. The factors were trace decay and trace interaction. Given a response set of overshooting, the type of anomalous results noted in the present paper are predictable by the two factors; given undershooting as in prior studies, typical forgetting functions are predictable.