Parafoveal Preview Processing during Sentence Reading: Integrated Neural and Behavioral Evidence

Behavioral and neural measures diverge sharply in the estimates they provide about the time course of word recognition during reading, with eye-tracking during reading indicating that word recognition affects oculomotor control less than 250 ms after a word is fixated while EEG methods involving rapid visual serial presentation (RSVP) of single words only provide reliable evidence of word recognition in the range of 330-500 ms after a word is seen. Recent EEG research has studied neural correlates to word recognition under more natural circumstances than RSVP by obtaining fixation-related potentials (FRPs), time-locked to the onset of the fixation on a word, during the reading of word lists in which parafoveal preview of upcoming words was manipulated. This experiment studied FPRs and eye movements while participants read sentences for meaning, a natural task in which words are fixated for less time than during list reading. Gaze-contingent display techniques were used to alter the information available in the parafovea, so that preview of an upcoming target word could be lexically deviant (an unrelated word) or physically deviant (visually degraded). These two types of preview had qualitatively different effects on both eye-movement and EEG measures. Both types of preview caused increases in fixation durations compared to identical (valid) preview that were similar on average size, but the effect of lexically-deviant preview increased across the quantiles of the fixation-time distributions while the effect of physically-deviant preview was nearly constant over the distribution. Lexically-deviant preview caused a large N400 in FRPs to the target word, an effect that was not observed until after the effect of the manipulation on fixation durations. In contrast, physically-deviant preview caused a large P300, an effect that overlapped in time with the effect of the manipulation on fixation durations. These results identify two processes related to parafoveal-preview processing. The first involves reconciliation of divergent information from preview and direct fixation, where the index of neural processing (the N400) does not identify the mechanisms that underlie the behavioral consequence of the altered preview. The second involves a general process of resetting attention where the behavioral effect of the altered preview may be produced by the neural mechanisms identified by the P300.