Abstract
Visual crowding, the interference in target recognition caused by surrounding similar objects, occurs not only in the periphery but also at the very center of gaze, where visual resolution is highest. It is unclear what factors contribute to crowding at this scale and to what extent cone density defines crowding thresholds. Here we investigated this using an Adaptive Optics Scanning Laser Ophthalmoscope (AOSLO) to stimulate the retina while bypassing optical limitations and maintaining the stimulus at a fixed retinal location despite the presence of fixational eye movements. Subjects (N=6) participated in a 4AFC digit identification task using Pelli’s font, specifically designed to study foveal crowding. The target, flanked by other digits along the horizontal meridian, was presented for 500 ms and maintained at the preferred retinal locus (PRL). Flankers were set at different distances using a method of constant stimuli. Stimuli sizes were set to three times the thresholds obtained using a QUEST procedure with unflanked (isolated) stimuli under retinal stabilization. Target width covered approximately 4 cones in each subject. Our findings revealed the typical crowding effect, with critical spacings (maximum interference due to flankers) ranging from 0.52 to 1.93 arcminutes, about 10-100 times smaller than what observed extrafoveally. We then identified the cones covered by the stimulus array and assessed the average cone size. The ratio between critical spacing measures to cone diameter was on average 1.78±1.07. Behavioral performance, when spacing matched half cone diameter, dropped by an average of 50% compared to isolated target presentation. Mislocalization errors, typical in extrafoveal crowding, where subjects report either flanker on incorrect trials, did not exceed chance level. These results suggest information pooling for foveal crowding involves at most two cones. Hence, critical spacing could potentially be inferred from the average cone diameter at the PRL.