CHARACTERISING REPRESENTATIONS OF HUE AND SATURATION IN THE CORTEX USING INFORMATION DECODING

Abstract

One of the enduring questions in the field of colour vision research revolves around how features of colour appearance, such as hue and saturation are represented in the brain. While considerable progress has been made in understanding the transformation of physical colour signals during early processing stages, the mechanisms by which these signals are recombined in later processing, ultimately giving rise to our perceptual experience of colour, remain elusive. A promising avenue for capturing these representations involves decoding from EEG signals. We captured EEG signals in response to 8 evenly spaced isoluminant colours at two saturation levels, taken from a perceptually uniform CIE Lab colour space. Surprisingly, our main finding challenges the expectation that representations of colours that are perceptually more similar elicit more similar signals. Instead, our results show that isoluminant hues perceived as maximally dissimilar tend to evoke the most similar EEG signals, which is in line with predictions derived from opponent theory. Additionally, decoding performs better for lower saturation than for higher saturation hues, contrary to expectations that more distinct hue representations would be associated with highly saturated hues but in line with prediction of magnified differences in latency and amplitude of EEG signals between low-level opponent mechanisms at lower contrast levels. Finally, our findings also highlight the non- uniformity of the cortical representation space for isoluminant colour, indicating that neighbouring hues exhibit varying degrees of similarity. Put together, the results show that low-level features, such as contrast and cone-opponency, drive the EEG response to colour.