Will Deng¹ (), Kara Federmeier¹, Diane Beck¹; ¹University of Illinois Urbana-Champaign

Image memorability, the likelihood that a person will remember an image, is an intrinsic property of the image that is distinct from many other visual and cognitive features. Research thus far has not identified any particular visual features that sufficiently explain this intrinsic memorability. We investigated the perceptual component of memorability by postulating a connection between memorability and statistical regularity, which refers to how well images match learned visual patterns. Statistical regularity affects detection time for images, such that stimuli with higher statistical regularity can be detected with shorter presentation. Therefore, we probed if memorability affects how quickly images are detected: we presented participants with high and low memorability images in an intact/scrambled task wherein they indicated whether they saw an intact image or noise, and we estimated the presentation duration necessary for participants to reach 70.7% accuracy. With two sets of stimuli, we observed and replicated that more memorable images have shorter detection thresholds than less memorable images, suggesting that more memorable images are more readily perceived. We further explored the perceptual processing evoked by memorable images using event-related potentials (ERPs) time-locked to stimulus onset. In a continuous recognition paradigm with delayed repetition, participants indicated whether they saw a new or repeated image. We focused our analyses on the N300 and N400 components, associated with high-level visual processing and long-term semantic memory access respectively. More memorable images evoked less negative N300 and N400 amplitudes, suggesting both facilitated perceptual and semantic processing. Overall, our results support the idea that memorability is an intermediate property related to both ease of perception and access to semantic memory, consistent with the idea that certain stimuli may better match stored templates used for image perception and recognition.