Figure 7. A schematic of the forward-encoding approach applied to EEG activity.

(A) Participants viewed individual gratings at fixation, each with a specific orientation. (B) Neural activity evoked by each grating was measured over the entire scalp. (C) Evoked neural responses were convolved with canonical orientation-selective functions (grey lines in C) to determine coefficients for the different orientations (coloured dots and lines, which match the colours of the outlined gratings in A). These coefficients were then used to generate a regression matrix. (D) General linear modelling was used on a subset of training trials to generate weights for each channel. These weights were inverted and simultaneously applied to an independent test set of data to recover orientation selectivity in the EEG activity. As EEG activity has high temporal resolution, we can apply the procedure to many epochs following stimulus presentation to determine the temporal dynamics of orientation processing (see Figure 3).