Benedict Vassileiou | How to memorize a sentence: Structural connectivity, neural oscillations, and computational modeling

Language comprehension critically necessitates working-memory resources to temporarily maintain verbal information for later recall. In this talk, I will present two studies investigating the neural correlates of working memory using magnetic-resonance and electrophysiological techniques. The first study provides tractography evidence for the selective functional relevance of the left superior longitudinal fascicle (SLF) but not of the arcuate fascicle (AF) in the storage and rehearsal of verbal information. Specifically, the data suggest that extended anterior fanning of the left SLF towards its anterior cortical projection zones is the structural property most strongly associated with good working memory abilities. The second study provides electroencephalography evidence for a beneficial role of alpha-band desynchronization for the successful encoding of sentences into working memory. Strikingly, combining single-trial computational modeling of syntactic and lexical-semantic comprehension difficulty with single-trial reconstructions of source-level activation time courses, we show that alpha-band power fluctuations within the left-hemispheric language network track the syntactic structure of the to-be-remembered sentence, in turn predicting later retrieval success. Our second study suggests that a fronto-parietal network also underlies sentence encoding into working memory. Yet, verbal information alone is not sufficient for successful sentence encoding, as our computational modeling reveals. Future work is called for to investigate the role of the SLF and the AF for the encoding of syntactic structure and the associated alpha-band power fluctuations.

Poster