Abstract
Reverse-correlation methods can provide both quantitative and intuitive information about the information used and mechanisms underlying a visual task. In this study, we use reverse-correlation to measure the spatiotemporal templates for both normal and amblyopic observers.
We measured contrast discrimination for a target bar embedded in spatio-temporal noise (11 noise bars, each 0.1 × 0.8 deg; 21 noise frames, each 10ms). Five normal and six amblyopic observers participated. The preferred eye of each normal observer and both eyes of each amblyopic observer were tested at fixation. The same measurements were repeated on normal observers with spatiotemporally filtered stimuli (2-d Gaussian filter) at fixation, and with unfiltered stimuli at 2.5 deg eccentricity in the lower visual field. We used reverse-correlation to calculate the classification images (templates).
The classification image for normal observers and non-amblyopic eyes has a positive peak at the spatio-temporal location of the target, with negative flankers 0.1∼0.3 deg on each side of the central peak. Along the time coordinate, it has a negative peak 30∼40ms following the onset of the target. In the amblyopic eye, however, the negative peak in the temporal profile is completely missing in almost every case, and the inhibitory flankers in space are either missing or scattered. When the stimuli are blurred by a 2-d Gaussian spatio-temporal filter, the classification image of normal observers is not qualitatively different from that with the unfiltered stimuli. But when the stimuli are viewed eccentrically, the classification image shows a lack of inhibition, similar to the amblyopic eyes.
Our results suggest that low level visual processing in amblyopia is abnormal mainly in that the inhibition, but not the excitation, is significantly reduced, both in space and in time. This deficit cannot be explained by “blur”, but is similar in nature to normal peripheral vision.