Abstract
Human scalp electroencephalography (EEG) can be used to measure electrical activity in the cortex and is usually thought to have low spatial resolution. Recently, research has shown that by combining activity across EEG sensors we can use multivariate analyses to decode information that includes more fine-grained spatial elements such as visual stimulus orientation, spatial location, spatial frequency, and motion direction. The decodability of other visual features, such as colour, remain less well explored. Although several attempts of decoding colours from brain activity have been successful in fMRI and more recently also in MEG, only few studies have evaluated colour decoding by means of EEG – and it often remains unclear to what extent such decoding reflects verbal labelling of distinct colours vs the parametrical and retinotopically-specific processing of colours in visual brain areas. In the current study, we presented participants simultaneously with two Gabor gratings, each with a unique colour and orientation and applied linear discriminant analysis to learn and successfully predict all four colour and orientation features. Our analyses show that both the orientation- and the colour-feature space are represented parametrically, rather than categorically, and preservation of characteristic contralateral decoding topographies (thus ruling out ‘verbal labelling’ as a potential explanation). We also show a robust increase in decoding performance when combining information across both EEG sensors and across time, suggesting the neural responses to different colours can be further distinguished through their unique temporal response profiles. We thus find reliable decoding of colour-specific processing in visual cortex from human scalp EEG, even when presented with multiple stimuli of different colours and at lateral locations. Unlike orientation decoding, colour decoding cannot capitalise on spatial information; therefore, it should be less biased by eye movements, opening up many new opportunities for future research.