Top-down knowledge can affect perception when the input is ambiguous

Poster Presentation 26.442: Saturday, May 18, 2024, 2:45 – 6:45 pm, Pavilion
Session: Color, Light and Materials: Art, cognition

Michael Cohen1,2 (), Andrew Rios1, Esther Min1, Rosa Lafer-Sousa3; 1Amherst College, 2MIT, 3National Institute of Mental Health, NIH

Does what we see depend on what we know? Many findings suggest that top-down factors such as emotions, desires, and categorical knowledge affect perception and can change an item’s appearance. However, a variety of methodological flaws, replication failures, and extremely small effect sizes render these findings inconclusive, leading to tremendous debate in the literature. Here, we searched for an uncontroversial example of top-down factors altering the appearance of items using a novel variant of the memory color effect. The memory color effect is a phenomenon in which familiar grey items look like their canonical color (e.g., a grey banana looks faintly yellow, Hansen et al., 2006). Although it has been thoroughly studied, the memory color effect remains extremely controversial, with many researchers doubting its very existence (Zeimbekis, 2013; Gross et al., 2014; Valenti & Firestone, 2019). Here, we asked, can the memory color effect provide us with clear evidence of top-down factors affecting perception? We theorized that under globally ambiguous viewing conditions, categorical knowledge may impact perception since that knowledge will be used to disambiguate underdetermined sensory input. Thus, we focused on one specific globally ambiguous viewing condition: very dim light. Specifically, we showed participants images of highly familiar grey items printed on poster paper under very dim lighting (i.e., mesopic), which created a lighting condition in which discerning all the attributes of the item is difficult. As predicted, we found an extremely strong memory color effect for familiar items under dim light, but not under bright, unambiguous light. No such effect was found for unfamiliar items. Moreover, this effect could be subjectively appreciated by observers who reported that the item just “looks” colorful, even though it is not. These results demonstrate that in certain situations, top-down factors can directly affect perceptual experience and appreciably alter how items appear.