Planning of movement parameters in a visuo-motor tracking task

Abstract

Monkeys trained to track a continuously moving visual target with a joystick do so by making a series of intermittent positional corrections rather than in a single smooth movement. The amplitude of each correction is highly correlated both with the error between the target and joystick positions, and with the velocity of the target, measured at movement onset. This velocity estimate is used to predict where the target will be by the end of each movement, and thus helps to set its amplitude correctly. To do this successfully, the monkey must know in advance how long his next movement will take. But, confusingly, the eventual duration of each movement is also highly correlated with its amplitude. So it appears that the monkeys need to simultaneously know the amplitude and duration of a movement, but cannot determine one without prior knowledge of the other. We have examined two possible solutions to this problem; only one agrees with our data. The monkeys seem to select the amplitude of their movements by scaling target velocity by a standard time constant which gives the additional distance the target will move. They then add this to the positional error estimated at or near to the start of each movement, to get the final movement amplitude. The velocity scaling value that gives the best fit to the observed amplitudes is very close to the average duration of all the monkeys movements. We therefore propose that the monkeys use a standard time constant for the purpose of calculating how far the target will move during each of their positional corrections. Each movement is then executed with a peak velocity proportional to its amplitude. Thus the durations of their movements vary but remain linearly related to amplitude; the slope of this relationship increases with target frequency.