Abstract
Real-world environments are often highly cluttered, presenting a challenge to our visual system. In particular, our ability to recognize an object in the periphery can be severely impaired by the presence of neighboring distractors, especially when those distractors share similar visual features with the target. Although crowding has been found to suppress the amplitude of responses in the early visual cortex, it is less clear as to whether crowding might impair the quality of information represented in these early areas. In this study, we relied on high-field, high-resolution fMRI at 7 Tesla to investigate the cortical representation of letters and digits in crowded and uncrowded viewing conditions. A digital-clock format with seven possible line segments was used to present numbers (‘3’, ‘4’, ‘7’, and ‘9’ in this study) and letters (‘E’, ‘h’, ‘J’, and ‘P’). In the crowded condition, the target letter/digit presented was presented with flanking stimuli that shared the same luminance polarity, whereas in the uncrowded flanker condition, the flankers were presented with the opposite luminance polarity. Results from 3 participants showed suppressed fMRI responses for crowded as compared to uncrowded targets, consistent with previous fMRI studies. To examine the quality of cortical representations, we applied pattern classification to letter/digit responses in individual visual areas and evaluated decoding accuracy using leave-one-run-out cross-validation. We found that decoding accuracy was significantly impaired on crowding runs, in comparison to uncrowded flanker runs and single-target runs. Moreover, we observed reduced functional connectivity between visual areas in the crowding condition, consistent with a previous report of crowding affecting functional connectivity (Freeman et al., 2011). Our results provide new evidence suggesting that when a target item is presented with visually similar flankers, the early visual representation of the target is disrupted and becomes less distinct due to mechanisms of crowding.
Acknowledgement: This research was supported by NIH grant R01EY029278 to F.T.