Abstract
When a Gabor patch is composed of a drifting grating within a stationary envelope, the Gabor appears to change position – a motion-induced position shift (MIPS, a.k.a curveball or double drift illusion). MIPS can be used to create dramatic motion illusions of global form (http://illusionoftheyear.com/2016/06/motion-integration-unleashed-new-tricks-for-an-old-dog/), but is tedious to study parametrically. Here, we used a motion-nulling variant of a tracking task, in which observers attempted to keep a global form, 16 Gabor patches in a ring, centered while the envelopes and carrier were perturbed by a 2D Brownian walk. In the 'zebra’ (control) condition, the carriers had a fixed phase relative to the envelop, so the carrier and envelopes moved as one. In the MIPS or 'cuttlefish condition, the carriers were driven in a horizontal random phase walk while the envelope positions were driven in an independent vertical walk; this produced a powerful illusion of the ring doing a 2D walk, despite moving only vertically. We varied the size of the global ring (spatial frequency scaled per V1 RF size). An advantage of the nulling task is that the observers keep the ring approximately centered, so ring diameter is also stimulus eccentricity. In the zebra condition, tracking performance decreased with eccentricity. In the cuttlefish condition, performance in the vertical (envelope) direction also decreased with eccentricity. In the horizontal direction however, tracking responses increased with eccentricity. Therefore, the subjects nulled the local phase walk (MIPS) as though it were global motion, especially in the periphery. More generally, our results indicate that this motion nulling paradigm is a powerful tool for studying motion perception in active observers while allowing control over where the stimulus is in the visual field. This paves the way for extending continuous tracking tasks into neurophysiology in awake, behaving non-human primates.