Fig. 2.

Steps involved in extracting the ERP data. An example of removing the overlapping auditory component from a V|_{A(−100 ms)} stimulus is shown here. A similar procedure was used for the other audiovisual stimulus types. After EEG recording (a), condition-wise averages (b) of all the multisensory ERPs (separately for each time window), unisensory stimuli, and no-stim (NS) trials were computed. c Subtraction of the no-stim ERPs from the unisensory and multisensory ERPs resulted in an ERP that eliminated the between-trial overlap, but which was still composed of contributions from an auditory input and a visual input. d The ADJAR procedure (Woldorff 1993) was used to convolve the unisensory ERP (as an estimate for the overlapping activity) with the distribution of all immediately preceding events (evt dist) to obtain an overlap free estimate of the contribution of one single modality (e), plus some variations due to attention and/or multisensory interaction processes (e.g., “ERP(V) within a specific multisensory context”)