Back

Perceiving surface colour requires attention

Goddard, E.; Paul Remician, K.

2025-10-16 neuroscience
10.1101/2025.10.14.682481 bioRxiv
Show abstract

Colour constancy refers to our ability to distinguish changes in surface properties from changes in the properties of light illuminating the scene. The apparent ease with which we can tell that objects do not change colour, for example, when moving from sunlight to the shade, belies the complexity of solving this ill-constrained problem. Although there is a substantial body of work testing which image cues might be used to accomplish this, there is surprisingly little known of how the brain performs this computation. Here, we tested a fundamental aspect of this perceptual process: whether it requires attention. We measured visual search times for both surface colour (requiring separation of surface and illuminant properties) and raw, colorimetric colour (which does not). We found a clear difference between the two: visual search for colorimetric colour was fast and near-parallel, while search for surface colour was slow and consistent with the serial deployment of attention. That is, search times suggest that the perceptual separation of surface and illuminant properties in colour constancy may require an attention-based process analogous to perceiving conjunctions of simple features in feature binding. Colour discrimination thresholds suggested that while colorimetric colour detection is fast and parallel, once attention was directed to these stimuli and perceptual scission occurs, colorimetric colour information was discarded by the visual system. These results offer important new insights into the sequence of processes the brain uses to accomplish colour constancy. Public Significance StatementThis work sheds new light on how the human visual system recognises the colour properties of different materials under varying lighting conditions. We find evidence that although material and lighting properties are perceptually separated automatically when attended, that this process requires attention.

Published in Journal of Vision (predicted rank #1) · training set

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.