Evidence of a limited visuo-motor memory used in programming wrist movements.

Human subjects can pre-program movements on the basis of visual cues. Experience in a particular task leads to the storage of appropriate control parameters which are used in programming subsequent movements, via a short-term motor memory. The form, duration and usage of this memory are, however, uncertain. Repetitive wrist flexion and extension movements were measured in four subjects. Three were neurologically normal men; the fourth subject had a peripheral large-fibre sensory neuropathy, depriving him of proprioceptive information about wrist movement. Subjects made alternating 45 degrees wrist movements between two visual targets; visual feedback of wrist position was provided for the first part of each trial. After 10 s of tracking, the subjects paused for an interval of 0-24 s before resuming tracking without visual feedback of wrist position. The positional accuracy of subsequent movements was analysed with respect to pause interval. Movement accuracy was reduced by the removal of visual feedback in all four subjects: movements after the pause interval were less accurate than those before the pause. Errors also accumulated within each sequence of movements made without visual feedback. Analysis of the first movement in each trial after the pause indicated a clear relationship between movement accuracy and pause interval. In all four subjects, movement accuracy decayed with longer pause intervals. In the deafferented subject, manipulation of the visual inputs (requiring visual fixation, rather than normal pursuit of the target; or direct viewing of the hand instead of viewing a cursor on a computer screen) affected the relationship between pause interval and subsequent movement accuracy. We propose that the memory used when producing these movements is a short-lasting visuo-motor signal, lasting a few seconds, which is derived from visual knowledge of previous movements, rather than a memory of a particular motor output. This visuo-motor signal is used to scale the amplitude of subsequent wrist movements. The brevity of the visuo-motor memory and the resultant inaccuracy of this deafferented subject and of our neurologically normal subjects implies that human feedforward control of the amplitude and position of wrist movements is severely limited.