Relevant findings

FIND: P3 pattern trained MVPA can classify P600 trials just as well as P3 trials (but not N400 or N170)

SUGGESTS: P3 and P600 substantially share neural patterns; P600 thus more likely instance of P3 and therefore domain-general ERP

FIND: Both P600 and P3 are not stimulus-onset but RT-aligned; Whereas N400 is stimulus-onset aligned (2014: auditory, 2015: visual)

SUGGESTS:

FIND: P600 on reversal anomalies (”The bathroom floor was mopping yesterday”) larger and (posteriorly) more broadly distributed in sentences following incongruent stroop trials than congruent stroop trials (but no effect on N400)

SUGGESTS:

FIND: Long regressions (two words or more back) accompanied by the P600, although sentences did not contain any gramm. violations; P600s typical onset was ~100ms after the beginning of the saccade

SUGGESTS: Since P600 (and regressions) occurred on trials without any violations (to detect), both must be more related to re-checking, WM;

Comprehension problems probably occur at the fixation before the regression (or even the one before that, as indicated by fixation durations);

as P600 onsets after the saccade onset, it must reflect WM or reanalysis [not detection]!

FIND: P600 elicited by picture-sentence mismatches (both intra- and extradimensional), accompanied by earlier negativity (possibly reflecting N400)

SUGGESTS: P600 reflects the conflict between two conceptual representations, in this case of that of the picture and that of the sentence. This conflict triggers reprocessing or reanalysis (to check whether the initial sentence processing has been correct)