Monday, April 6,  12:00 – 1:30 pm, in 489 Minor Hall

Graduate Student Seminar

presented by

Kavitha Ratnam (Roorda Lab)

Fixational Eye Movements Improve  Visual Performance at the Cone Photoreceptor Sampling Limit

Even during fixation, the eye is constantly in motion. Fixational eye movements (FEM) are miniature ocular jitter that move an otherwise static image across the retinal cone mosaic. FEM were classically postulated to improve vision by preventing image fading and bringing retinal images back to the preferred retinal locus of fixation. However, the ‘dynamic theory’ of visual acuity proposes that FEM may enhance spatial resolution by moving an image over multiple cone photoreceptors whose local activations are averaged into a sharper signal. Although earlier studies did not find evidence supporting this theory, recent work has shown that FEM enhance discrimination of high spatial frequencies up to 10 cycles per degree (Rucci et al., 2007).  The effects of FEM, however, have previously not been studied at the limits of visual resolution— that is, at the cone mosaic sampling limit—due to technical limitations in delivering stimuli undistorted by the eye’s imperfect optics at the scale of individual photoreceptor cells. Adaptive optics, a technique used to overcome ocular blur, can be utilized to deliver dynamic stimuli to the retina. Combining AOSLO stimulus delivery with accurate retinal image stabilization, near-diffraction limited stimuli can be delivered either retinally unstabilized or stabilized to observe the effects of FEM at the retina’s sampling limit. Here we show that discrimination of image detail that are at or below this limit are enhanced by the retinal image motion caused by fixational eye movements, affording the visual system a form of ‘super-resolution’ that surpasses the photoreceptor sampling resolution.


Wesley Chaney, PhD Candidate (Whitney Lab)

 Spatial attention reduces correlated noise in the fMRI response

Spatial attention modulates sensory neural activity, enhancing the representation of relevant stimuli and leading to enhanced performance in a variety of tasks. However, the exact mechanisms by which attention modulates early sensory activity and how this impacts the fMRI BOLD response are not yet fully understood. Previous studies recording from single units in area V4 have demonstrated that attention reduces correlations in the noise of simultaneously observed neurons (Cohen and Maunsell, 2009; Mitchell, 2009). We tested whether analogous reductions of correlated noise also occur at the population level using the fMRI BOLD signal. We examined the effects of attention on the representation of four Gabor stimuli presented simultaneously in each quadrant of the visual field at jittered eccentricities. Subjects attended for contrast decrements in the Gabors in one visual field (either upper or lower in alternating runs) while ignoring the Gabors in the other visual field. After regressing out stimulus driven activity in V1 and V2 in a General Linear Model analysis, we analyzed pairwise correlations of the residual timecourses in voxels representing either the attended or unattended visual fields. We found that attention to either the upper or lower visual field reduces correlation of these timecourses. This reduction is opposite to what would be predicted from reduced noise within individual voxels or an increase in stimulus driven activity due to attention and it cannot be explained by vasculature differences or local scanner artifacts as each region is attended or ignored for an equal number of runs. This reduction in correlated noise within attended locations allows for more accurate signal estimation at the population level, and may facilitate readout by higher level processes that pool over information in early visual cortex. We also show that decorrelation is associated with improved position discrimination in early visual cortex.

← Back to Oxyopia Archive