Neuroscientists can now predict what color you're seeing (www.psypost.org)

🤖 AI Summary
Researchers at the University of Tübingen and the Max Planck Institute used fMRI, retinotopic mapping and shared response modeling to show that color evokes reproducible brain patterns across people. Participants viewed black-and-white moving checkerboards (retinotopic mapping) and later concentric rings in three colors (red, green, yellow) at two luminance levels. The team aligned individuals’ neural responses into a shared space using only the black-and-white data, then trained a classifier on some subjects and successfully predicted which color another subject was seeing—above chance. Cross-subject decoding was strongest in early and mid-level visual areas (V1, V2, hV4, LO1) and depended on where a color appeared in the visual field, indicating systematic, region-specific color biases tied to retinotopic organization. Technically, the paper demonstrates that retinotopic alignment alone can transfer color information between brains, supporting the idea of a common neural code linking spatial location and chromatic responses. That finding is important for neuroscience and ML-based brain decoding because it enables cross-subject models without color-specific alignment and suggests robustness of population-level color representations. Limitations include testing only three colors (no blue), simple lab stimuli, and no claim about subjective qualia; future work should broaden color space, test naturalistic scenes, and probe how experience modulates these shared maps.
Loading comments...
loading comments...