Christopher DiMattina¹ (), R. Nathan Pipitone¹; ¹Florida Gulf Coast University

Images whose statistical properties deviate from those typical of natural images are often uncomfortable to view. One proposed explanation for this discomfort is that these unnatural images over-activate the primary visual cortex, a hypothesis known as the Cortical Hyper-metabolism Hypothesis (CHH). We tested whether the CHH could explain visual comfort ratings for different natural image categories and natural images with amplitude and phase spectrum manipulations. Observers rated the visual comfort of images from three different categories: 10 landscape images (LSC), 10 control textures from the Brodatz database (BDZ), and 10 trypophobic textures (TRY). Each image was presented in three versions: (1) its original form (or), (2) with its phase-spectrum randomized (ps), and (3) with its amplitude spectrum set to 1/f (1f). These same images were passed through biologically plausible population models of V1 neurons, from which we measured the population kurtosis and population activity as measures of coding efficiency. Restricting our analysis to non-trypophobic images, we found that consistent with the CHH, the landscape images were more efficiently coded by the V1 models and more comfortable to view than the Brodatz textures. Also consistent with the CHH, we found for both categories (LSC, BDZ) the original images were more efficiently coded by the V1 models than their phase-scrambled controls, and were also more comfortable to view. For the trypophobic imagery, we found a completely different pattern of results. Trypophobic images were much more uncomfortable in their original form than when phase-scrambled, despite the fact that the original images were more efficiently coded. This suggests that one cannot fully explain trypophobic visual discomfort in terms of the CHH and low-level statistical properties of images. We suggest that the visual discomfort elicited by trypophobic imagery may represent an over-generalization of a disgust response to visual patterns resembling skin disease.