Abstract
Smooth pursuit is highly accurate for two-dimensional target motions whose tangential speed decreases with increasing curvature according to the two-thirds power law, a characteristic shared with many biological motor actions (de’Sperati and Viviani, 1997). Nevertheless, the effect of changes in target direction on pursuit of constant velocity motions has been controversial. Subjects pursued a target that moved at constant speed (8 deg/s) starting from one of 8 positions along an imaginary circle, moving toward center, and returning to one of the 8 positions. Turning angles at the center ranged from 0 to180 deg. Start and end positions were selected randomly. On half the trials (predictable paths) a line marking the path was displayed. Turning angle had large effects. Pursuit speed began to decrease 500 ms before the turn, in anticipation of the change in direction. For predictable paths pursuit speed decreased steadily, reaching a minimum by 50–100 ms after the turn. Pursuit speed decreased at a slower rate for unpredictable paths until 100 ms after the turn when speed decreased sharply. Minimum pursuit speed was reached 100 ms after the turn for predictable, and 200 ms after for unpredictable paths. Minimum pursuit speed for both predictable and unpredictable paths varied linearly with turning angle, with the lowest value, 20% of target speed, found for the sharpest turns. These results show that pursuit of two-dimensional motions depends on the geometry of the motion path, with marked reductions in speed accompanying sharp changes in direction. Cues that disclosed the amount of direction change caused smooth pursuit to reduce speed earlier and more gradually. These results may reflect changes in the neural representation of direction over time, and serve to make properties of pursuit compatible with the naturally-occurring motions that are pursued during real-world tasks.