Accurately acting on moving stimuli requires a higher fidelity motion representation than perceptual discrimination

The human visual system evolved to process and interpret continuous visual input. However, visual signals in modern societies increasingly emanate from digital displays, which emulate continuous motion by rapidly presenting a series of discrete frames at some fixed temporal frequency. How does such non-continuous, sampled motion influence visual perception and action? We asked participants to complete a combined oculomotor and speed-discrimination task, where they, first, tracked the motion of a stimulus via smooth pursuit eye movements, and second, compared the velocity of the pursuit stimulus to a reference velocity. Motion quality was manipulated by presenting discrete frames at one of four temporal frequencies (120, 60, 30, 15 Hertz), covering a range of values corresponding to different motion qualia, from quasi-continuous (120 Hertz) to highly discontinuous (15 Hertz). We found that participants could reliably discriminate velocities, even at low refresh rates. Pursuit accuracy, however, declined with decreasing refresh rate and increasing target velocity: a perception-action dissociation. The decline in pursuit accuracy can be explained by a model, which estimated a proxy for motion quality from the spatiotemporal energy profile of stimuli. Using this estimate, our model successfully predicts pursuit accuracy for arbitrary combinations of target velocity and refresh rate. Our results highlight the dissociation between perception and action in digital displays with sampled motion; while we may be able to accurately discriminate motion at reduced frame rates, acting on moving stimuli requires a higher fidelity motion representation.