Chenxi Liao¹ (), Jacob Cheeseman², Filipp Schmidt², Roland W. Fleming², Bei Xiao¹; ¹American University, ²Justus Liebig University Giessen

Material perception is challenging to study because many intrinsic and external factors affect material appearances. How the visual system encodes complex materials (e.g., translucent materials) remains unknown. Here, we explore the representation of materials using visual adaptation—perceptual distortion after exposure to stimuli for a prolonged period. In each trial, participants were presented with a pair of images (spatially jittering to avoid retinal afterimages) of different materials to the left and right of the fixation (adaptor pair). After a total of 4 seconds, they were followed by a brief presentation (0.2 seconds) of a pair of test images (test pair). Tests were either identical or different materials sampled along a morphed sequence between the two adaptors. Participants judged whether the two test images were the same or different. Baseline judgments were collected without prior exposure to the adaptors. All stimuli were created by training generative neural networks on custom-made photographs of everyday translucent objects, like soaps and stones. We selected nine distinctive adaptor pairs with different perceptual similarity in material category, color, shape, and other material properties (e.g., translucency, smoothness). We morphed between adaptors (e.g., a soap and a stone) to sample sequences of fifteen test pairs equally around the midpoint of each morph sequence, spanning a range of variation in those perceptual dimensions. In comparison with the baseline, we found that adaptation enhanced sensitivity to subtle differences in similar stimuli, leading to more "different" judgments. Meanwhile, it reduced the discrimination for distinct stimuli, resulting in "same" judgments. Notably, the magnitude of the repulsive effect depended on the specific adaptor pair. Our findings show that adaptation can lead to powerful distortions of perceived material appearance, affecting multiple perceptual dimensions, and even changing the apparent category of materials.

Acknowledgements: Supported by NIH, award 1R15EY033512-01A1; DFG, project number 222641018–SFB/TRR 135 TP C1, the ERC Advanced Grant "STUFF" (project number ERC-2022-AdG-101098225); and the Research Cluster "The Adaptive Mind" funded by the Hessian Ministry for Higher Education, Research, Science and the Arts.