Sam Clarke¹, Sami Yousif²; ¹University of Southern California, ²University of North Carolina, Chapel Hill

Visual adaptation has been described as a “powerful tool for dissecting vision by exposing the mechanisms that are adapting” (Webster, 2015; p. 547), an “intrinsic feature of visual processing” which “reaches the status of a universal law” (p. 548). Others have described it as “a largely unambiguous and uncontroversial way to identify visual processing” (Kominsky & Scholl, 2020; p. 3). In recent years, the scope of visual adaptation has expanded dramatically, to include not only things like color and motion but also higher-level visual features like size, number, causality, and variance. Here, we expand the reach of visual adaptation even further: We demonstrate robust, spatiotopic visual adaptation to arbitrarily assigned value. In our task, observers were introduced to displays of fake “coins” with values ranging from one to five. They were told that they will be completing a quantity discrimination task in which their goal is to select the side with more total value. Prior to making each judgment, however, participants adapted to separate displays of coins. In three separate experiments, we found robust evidence of value adaptation: When participants adapted to a high-value display, they were subsequently more likely to indicate that the contralateral side was greater in value. *Visual* adaptation to value seems not to be possible, however: Value is not — and could not be —a perceptual dimension. The choice that observers make depends on their knowledge of the coin values, which is arbitrarily and rapidly mapped onto the visual information. Yet according to the agreed-upon standards for what constitutes visual adaptation, we have documented robust evidence of value adaptation. These findings thus apply pressure to our understanding of visual adaptation as a phenomenon, specifically the (popular) argument that adaptation is somehow uniquely indicative of perceptual processing.