Abstract
Tri-polar concentric ring electrodes (tCREs) offer a promising avenue for collecting visually-evoked potentials (VEPs) while minimizing certain artifacts that are common to traditional electroencephalography (EEG). We previously showed that VEPs recorded using tripolar EEG (tEEG) were generally comparable to those collected (simultaneously) from an emulated standard EEG signal (eEEG), with some subtle differences in the latency of the N70 and P100 components. Here, we used multivariate decoding analysis to compare the nature of the information carried by each electrode type. We recorded VEPs to large (1°) or small (0.25°) contrast-reversal checkerboard patterns at 7 different retinal locations. A linear discriminant analysis (LDA) classifier was trained and tested at every ms time point (0-496 ms post-reversal), separately for tEEG and eEEG signals. Results from the decoding of large vs. small checkerboards, left vs. right hemifield, and four quadrant locations were qualitatively similar. The time course of decoding was similar across the two types of electrodes, with significant above-chance decoding observed between ~75-350 ms. Decoding of the eEEG signal tended to produce slightly earlier and higher peak classification compared to the tEEG signal. As expected, decreasing the number of trials included led to worse decoding performance; peak decoding performance was achieved with ~40 trials per condition for both types of data. Consistent with the observed VEPs, the time course of decoding from the tEEG electrodes appeared shifted later in time by ~25 ms. For some comparisons, decoding based on signals from both tEEG and eEEG channels produced significantly better decoding than tEEG or eEEG alone, suggesting that the tEEG and eEEG signals may be sensitive to slightly different but overlapping neural sources. These results further demonstrate the utility of tEEG collected from tCREs to quantify the informational content of visually-evoked potentials.