The efficiency of reading, to a crucial extent, results from the fact that visual word recognition is faster in predictive contexts (like sentences or texts). This observation is consistent with predictive coding models, but it is so far not sufficiently understood which aspects of the rich set of linguistic representations that is activated during reading contribute to this context-dependent facilitation. Candidate representations are visual, orthographic, phonological, and/or lexical-semantic in nature. Our study investigates which representations contribute to efficient word recognition in predictive context with a well-controlled repetition priming paradigm, including words and pseudowords (i.e., pronounceable nonwords), that was combined with magnetoencephalography (MEG) measurements in human participants (both sexes). We used high-powered linear mixed modeling to test the hypothesis that context-dependent facilitation relies on the pre-activation of linguistic representations prior to perceiving the expected stimulus. Behavioral data from 49 participants indicate that word predictability (i.e., context present vs. absent) facilitated orthographic and lexical-semantic, but not visual or phonological processes. MEG data from 38 participants show sustained activation of orthographic and lexical-semantic codes in the interval between prime and target, i.e., before processing the predicted stimulus. Also, we found a positive correlation between these pre-activation effects and brain responses elicited when processing the expected letter string. Pre-activation was more stable across time in words than in pseudowords. These results suggest that readers use orthographic and lexical-semantic representations to actively predict upcoming words and that this predictive process is modulated by the presence of prior knowledge (in words as opposed to pseudowords).
bioRxiv Subject Collection: Neuroscience