Disentangling visual and motor contributions to continuous psychophysics

Abstract

Continuous psychophysics measures visual function by asking observers to track a randomly moving target. Because continuous responses provide more information than traditional discrete judgements, a few minutes of tracking can estimate visual sensitivity, making the approach attractive for testing children and patients. Tracking performance is often summarized by cross-correlating the target and cursor velocities: the resulting cross-correlogram peak, lag and width index how strongly, quickly and sharply the observer responds to the target. Tracking, however, requires both seeing the target and moving the cursor towards it, so poor tracking may reflect impaired vision, impaired motor control, or both. Here, we test whether visual and motor contributions to tracking performance can be told apart. Participants tracked a target with a computer mouse while wearing glasses that simulate two levels of visual impairment, using either their dominant or their non-dominant hand. We find that both manipulations alter the cross-correlogram metrics, lowering the peak correlation and delaying and broadening the response, but in different proportions. Visual impairment mainly delays and broadens the response, and does so in proportion to each participant's loss of visual acuity. The non-dominant hand mainly lowers the peak correlation. Two signatures of motor control explain this difference: with the non-dominant hand, participants pause more often and for longer, and move more noisily between pauses, whereas visual impairment alters neither. Together, response lag, time paused and movement noise reveal whether a change in tracking is visual or motor in almost every participant. Our findings demonstrate that, given a baseline measurement, the same tracking data that measure vision can also reveal whether a change in performance reflects a change in what an individual sees or in how they move.