Spatiotemporal dynamics of abstract concept processing : An MEG study

Abstract

Our current understanding of how linguistic concepts are represented and retrieved in the brain is largely based on studies using concrete language, and only few studies have focused on the neural correlates of abstract concepts. The role of the motor system, besides the classical language network, has been intensively discussed in action-related concrete concepts. To advance our understanding of spatiotemporal dynamics underlying abstract concept processing, our study investigated to what extent language and motor regions are engaged in the processing of abstract concepts vs. concrete concepts. We used concrete, metaphorical, and abstract phrases as stimuli, creating a graded continuum of abstractness. Neuromagnetic signals were recorded from 26 Chinese native speakers using a 306-channel whole-head magnetoencephalography (MEG) system. Cluster-based permutation F-tests were carried out on the amplitude of source waveform for individual language and motor regions of interest (ROIs) in the three consecutive time-windows (200-300, 300-400, and 400-500 ms). Results showed that, compared with concrete and metaphorical phrases, abstract phrases evoked significantly weaker activation in the left posterior part of superior temporal sulcus (STS) at 200-300 ms, and significantly stronger activation in the left anterior temporal pole (TP) at 300-400 ms. We found no significant differences in the involvement of motor ROIs across conditions. Our results suggest that concrete concept processing engages more the posterior STS in an earlier time window, while abstract concept processing relies more strongly on the anterior TP in a later time window. Results are discussed by revisiting the ATL (anterior temporal lobe)-hub hypothesis and the novel definition of concrete and abstract concepts.