Fixational eye motion

Fixational drift in the primate visual system is a random, smooth motion that occurs between micro-saccades. Although the range of this motion is quite small, it is surprisingly large on the scale of foveal vision: for example, the angular motion of the image on the retina due to fixational drift, occurring between two successive spikes of a single retinal ganglion cell (RGC), is typically larger than its receptive field. If the brain infers the information encoded by retinal spikes by simply counting spikes from each retinal ganglion cell over a time window of a few tens of milliseconds, the motion would naively blur the inferred image, in similarity to the blurring of an image collected by a camera while the hands are shaking. Yet, our visual system achieves exquisite acuity, comparable (and sometimes exceeding) the resolution of single photoreceptors and RGCs in the fovea.

There is compelling experimental evidence that visual brain areas do not use an efferent copy of the fixation drift to compensate for the motion of the projected image on the retina, relative to the photoreceptor array. Thus, the brain must solve a statistical inference problem, of distinguishing between different hypotheses on the content of the image amid an unknown, dynamic trajectory. We investigate how visual areas in the brain may solve this problem, and what are the consequences of fixational drift for perception.

We are interested also in the neural mechanisms that underlie fixational eye motion.

Selected publications

Ben-Shushan N. (*), Shaham N. (*), Joshua M., Burak Y.
Fixational drift is driven by diffusive dynamics in central neural circuitry.
* Equal contributors.
Nature Communications 13, 1697 (2022). Link

Burak Y., Rokni U., Meister M., Sompolinsky H.
Bayesian model of dynamic image stabilization in the visual system.
Proceedings of the National Academy of Sciences, USA 107, 17645 (2011) LINK