Figure 2.

Generalization in time for distinguishing between congruent and incongruent trials, which elicit both N400 and P600. (a) A TGM matrix showing the ability to distinguish malformed versus correct sentences. Training times are along the x-axis, test on the y-axis. Black contours indicate above chance performance. (b) The diagonal decoding from figure (a). (c) Decoding performance for time periods corresponding to the N400 (300–500 ms; green) and P600 (600–800 ms; blue), averaged over rows of the TGM matrix in (a). Shaded regions represent 95% bootstrapped confidence intervals. Horizontal lines indicate time points significantly above chance (thick lines p < 0.01; thin lines p < 0.05). These results support the hypotheses that the N400 and P600 are cascading processes. The TGM matrix shows that the information processing regimes underlying the processes are not interchangeable. Adapted from Heikel et al. [27].