Rapid short-term reorganization in the language network

Abstract

The adaptive potential of the language network to compensate for lesions remains elusive. We show that perturbation of a semantic region in the healthy brain induced suppression of activity in a large semantic network and upregulation of neighbouring phonological areas. After perturbation, the disrupted area increased its inhibitory influence on another semantic key node. The inhibitory influence predicted the individual delay in response speed, indicating that inhibition at remote nodes is functionally relevant. Individual disruption predicted the upregulation of semantic activity in phonological regions. In contrast, perturbation over a phonological region suppressed activity in the network and disrupted behaviour without inducing upregulation. The beneficial contribution of a neighbouring network might thus depend on the level of functional disruption and may be interpreted to reflect a differential compensatory potential of distinct language networks. These results might reveal generic mechanisms of plasticity in cognitive networks and inform models of language reorganization.DOI: http://dx.doi.org/10.7554/eLife.25964.001

Differential contributions of left-hemispheric language regions to basic semantic composition

“…Moreover, after longer-lasting disruption of left AG, this area had an inhibitory influence on the left aIFG during semantic word decisions, further substantiating the strong interaction between both regions (Hartwigsen et al, 2017). Crucially, our results provide novel supporting evidence for the notion of a semantic network involving AG and aIFG not only at the single word, but also at the basic combinatorial level.…”

Neural correlates of intonation and lexical tone in tonal and non‐tonal language speakers

“…Tone compared to intonation processing involved bilateral AG/pMTG, SPL and PCC (Figure c), likely reflecting semantic processing. AG and pMTG have been consistently identified when participants are engaged in task‐related semantic decisions and lexical semantic processing (Binder, Desai, Graves, & Conant, ; Hartwigsen et al, ; Noonan, Jefferies, Visser, & Lambon Ralph, ). The specific involvement of AG/pMTG for tone processing fits well with this account since tonal pitch in Mandarin is employed to contrast lexical semantic meanings (Chao, ).…”

Multivariate single-subject analysis of short-term reorganization in the language network

“…However, there is increasing evidence that group-level analyses may obscure important parts of the signal response (Margulies, 2017;Poldrack, 2017). Here, we re-analyzed data from our recent study (Hartwigsen et al, 2017) that investigated functional reorganization in the language network. We show that across-voxel pattern-learning approaches are useful to isolate plastic changes in neural networks underlying cognitive functions at the individual subject level.…”

“…However, current knowledge of plasticity in language networks is insufficient (Turkeltaub, 2015). Combining focal perturbations induced by transcranial magnetic stimulation (TMS) and fMRI in the healthy brain, we recently provided insight into the reorganization potential of language networks (Hartwigsen et al, 2017). Causal impairment of a key node for the processing of the meaning of words (semantic processing) in left angular gyrus (AG) entailed decreased semantic activity in a large network, including the targeted area.…”

“…This analysis revealed areas that carried information about the presence versus absence of our TMS intervention. To provide principled guidance to our data-driven analyses, we based our searchlight approach on a topographical map of the left inferior/superior parietal lobe that was derived from neural activity information of two experimental task contrasts devoted to contrast semantic activity versus rest (showing increased activity in left AG) or phonological activity versus rest (showing increased activity in left SMG/SPL) in our original study (Hartwigsen et al, 2017). Both contrasts were derived by thresholding at a lenient threshold of p < .01 uncorrected, and the respective clusters were merged into a map that informed our searchlight approach.…”