Obligatory coactive processing of color and luminance challenges strategic modulation by predictiveness

Navigating the world requires accurate categorization of objects around us, which often involves processing multiple sources of information. The predictiveness of a source plays an important role in accurate categorization. This study aims to investigate how the predictiveness of features modulates the processing strategies of two features that are generally considered more integral than separable: color and luminance. Participants categorized a set of visual stimuli, created by varying levels of color and luminance, into two categories defined by logical rules. The stimulus-category mapping was 100% in Experiment 1, but it was reduced to 95% in Experiment 2. In both experiments, the predictiveness of both features was equal. Lastly, in Experiment 3, we introduced unequal predictiveness such that color was more predictive for some participants, while luminance was more predictive for others. These manipulations were designed to test whether, as predicted by the strong version of the relative saliency hypothesis, even integral features such as color and luminance could be processed serially if one were made more predictive of the category. Across the three experiments, we employed both system factorial technology (SFT) and computational modeling to infer processing strategies in nonparametric and parametric manners, respectively. Although some variability existed at the individual subject level, both non-parametric and parametric modeling revealed robust evidence for coactive processing for the aggregated group data, regardless of the varied stimulus-category mapping and feature predictiveness. These findings suggest that the processing of color and luminance within an object involves obligatory coactive processing, thereby challenging the strategic adjustment relative saliency hypothesis.