Abstract
We have previously reported a jitter aftereffect that seems an error of compensation for image slip on the retina (Murakami & Cavanagh, 1998, 2001). After adaptation to dynamic noise in an annulus, a static noise pattern covered the disk and annulus regions; the disk portion appeared to jitter coherently. Dynamic-noise adaptation was interpreted to desensitize motion detectors in the annulus, creating differential-motion codes between the disk and annulus, when both regions actually suffer from the same retinal slip due to small eye movements. Here I present a new version of similar jitter illusion, which one can easily get on-line: that requires no adaptation, lasts forever, and allows quantitative assessment of motion deficits in the annulus.
On a uniform background of mean luminance, two random-dot fields (Julesz-type: 50% black, 50% white) were presented concentrically. The border between the disk and annulus was blurred by a sigmoidal contrast modulator. The disk was shown constantly, whereas the contrast of the annulus was modulated temporally, e.g., at 9.4 Hz (80 ms on-duty, 27 ms off) for the best condition. A compelling illusion was seen: the physically static random-dot pattern in the disk appeared to jitter coherently in random directions, whereas the annulus did not. When the eye was moved either by providing a pursuit target or by inducing reflexive movements, the illusory motion was directionally consistent with expected retinal slip. Matching experiments not only quantitatively confirmed these observations, but also estimated the extent of illusory jitter to be comparable to small eye movements. I propose that given the same retinal slip in the disk and annulus, the flicker elicits noisy directional responses of motion-energy units, degrading velocity estimation in the annulus. A subsequent process for retinal-slip compensation may use these artificially weakened signals as a frame of reference, resulting in undercompensation for velocity in the disk.