Spatio-temporal dynamics during word production : studies in magnetoencephalography
This thesis attempts to contribute to the magnetoencephalography [MEG] language production literature in two ways. (i) An appreciation and critique of the general method of approach used for researching the language network using MEG, with a particular focus on the distinction between serial and parallel models of word production, and (ii) the contribution of empirical evidence concerning the dynamics of the language network during picture naming, from an MEG investigation of picture naming. The review of the MEG literature uses recent models of the language production network to frame the timing of effects found in MEG language production papers published in the last 20 years. These functional/anatomical frames show that a large proportion of effects reported go undiscussed by authors, and that framework provided by Indefrey and Levelt’s serial model of speech production casts a general trend towards seriality in the data. However, this serial trend breaks down when the data are framed using the Heirarchichal State Feedback Model (Hickok, 2012), or an attribution of cognitive function to cortical area from Price’s 2012 review of the fMRI literature. The review finds evidence supportive of a parallel model present in the data, such as early effects in conceptually late functional areas, such as the inferior frontal gyrus [IFG] and motor areas, and late effects in visual areas. Additionally, it reports that event related potential [ERP] and time/frequency [T/F] analysis complementary sources of information (as also shown by Laaksonen, Kujala, Hultén, Liljeström, & Salmelin, (2012), having generally different temporal characteristics, with T/F effects having earlier onsets than ERP effects. This review also highlights the diversity of tasks, source modelling techniques and analysis methods, as well as incomplete reporting of results, and make suggestions for more standardised methods to improve the comparability and interpretability of studies in this field. An empirical study was carried out using MEG to probe the cortical dynamics of picture naming. Using visual and semantic manipulations, the study reports the timing of response in cortex active during picture naming. It presents an exploratory analysis of the whole brain without a priori constraints in time or space, and contrasts between levels of semantic complexity using local false discovery rate correction. The results show an occipital to temporofrontal progression in the cortical response, a corroboration of previous findings. There was also activity in in cingulate cortex, and bilateral temporal poles, and activity in the right insula, left IFG, bilateral temporoparietal junction, and pericentral sensorimotor cortex. Early between conditions differences (90ms post stimulus) are seen in bilateral BA8 and anterior cingulate. Differences in right anterior medial temporal cortex (207ms post stimulus) which processes semantic retrieval, and right temporo-parietal junction (233ms post stimulus), associated with self-monitoring and secondary motor function, suggest parallel processing of articulatory and semantic task elements. Late-onset between conditions differences in the right cuneus also suggest ongoing visual processing in our picture naming task after 500ms post-stimulus. We also find early-onset ( 100ms post-stimulus) between-conditions differences in novel frontal medial areas responding to visual and semantic manipulations. The between-conditions differences observed question the timing estimated for semantic and phonological processing suggested by current serial models of speech processing.