Spectro-temporal neural dynamics during sentence completion

This magnetoencephalography (MEG) study aimed at characterizing the spectro-temporal dynamics of brain oscillatory activity elicited by sentence completion (SC). For that purpose, we adapted a version of the SC experimental paradigm typically used in functional magnetic resonance imaging to MEG investigation constraints. Twenty right-handed healthy young adults underwent MEG recordings while they were sequentially presented with short sentences divided in three parts: the first two giving context and the last requiring completion. MEG data were then analysed using a prior-free, non-parametric statistical approach with stringent control of the family-wise error rate. We identified three successive significant neural response patterns associated with distinct spatial and spectro-temporal characteristics: (i) an early (<300 ms) bioccipital 4-10-Hz event-related synchronization (ERS); (ii) an intermediate (at about 400 ms) 8-30-Hz event-related desynchronization (ERD) in an extended semantic network involving the ventral language stream as well as bilateral posterior nodes of the default mode network (DMN) in both hemispheres; (iii) a late (>800 ms) 8-30 Hz ERD involving the left dorsal language stream. Furthermore, the left component of the ventral language stream displayed prolonged ERD after 800 ms compared to the right which showed signs of inhibition in the form of ERS. Overall, this study elucidates the dynamics of the recruitment of the language network that accompany SC and the spectro-temporal signature of an extended semantic network. This MEG adaptation of an SC paradigm also paves the way for novel approaches in presurgical language mapping and may help to understand the neural underpinnings of the alterations of sentence completion in various neurologic disorders affecting language.