Neural Correlates of Naturally Occurring Speech Errors during Picture Naming in Healthy Participants

Most of our knowledge about the neuroanatomy of speech errors comes from lesion-symptom mapping studies in people with aphasia and laboratory paradigms designed to elicit primarily phonological errors in healthy adults, with comparatively little evidence from naturally occurring speech errors. In this study, we analyzed perfusion fMRI data from 24 healthy participants during a picture naming task, classifying their responses into correct and different speech error types (e.g., semantic, phonological, omission errors). Total speech errors engaged a wide set of left-lateralized frontal, parietal, and temporal regions that were almost identical to those involved during the production of correct responses. We observed significant perfusion signal decreases in the left posterior middle temporal gyrus and inferior parietal lobule (angular gyrus) for semantic errors compared to correct trials matched on various psycholinguistic variables. In addition, the left dorsal caudate nucleus showed a significant perfusion signal decrease for omission (i.e., anomic) errors compared with matched correct trials. Surprisingly, we did not observe any significant perfusion signal changes in brain regions proposed to be associated with monitoring mechanisms during speech production (e.g., ACC, superior temporal gyrus). Overall, our findings provide evidence for distinct neural correlates of semantic and omission error types, with anomic speech errors likely resulting from failures to initiate articulatory-motor processes rather than semantic knowledge impairments as often reported for people with aphasia.

Face-selective responses in combined EEG/MEG recordings with fast periodic visual stimulation (FPVS)

Fast periodic visual stimulation (FPVS) allows the recording of objective brain responses of human face categorization (i.e., generalizable face-selective responses) with high signal-to-noise ratio. This approach has been successfully employed in a number of scalp electroencephalography (EEG) studies but has not been used with magnetoencephalography (MEG) yet, let alone with combined MEG/EEG recordings and distributed source estimation. Here, we presented various natural images of faces periodically (1.2 Hz) among natural images of objects (base frequency 6 Hz) whilst recording simultaneous EEG and MEG in 15 participants. Both measurement modalities showed face-selective responses at 1.2 Hz and harmonics across participants, with high and comparable signal-to-noise ratio (SNR) in about 3 min of stimulation. The correlation of face categorization responses between EEG and two MEG sensor types was lower than between the two MEG sensor types, indicating that the two sensor modalities provide independent information about the sources of face-selective responses. Face-selective EEG responses were right-lateralized as reported previously, and were numerically but non-significantly right-lateralized in MEG data. Distributed source estimation based on combined EEG/MEG signals confirmed a more bilateral face-selective response in visual brain regions located anteriorly to the common response to all stimuli at 6 Hz and harmonics. Conventional sensor and source space analyses of evoked responses in the time domain further corroborated this result. Our results demonstrate that FPVS in combination with simultaneously recorded EEG and MEG may serve as an efficient localizer paradigm for human face categorization.

Implicit, automatic semantic word categorisation in the left occipito-temporal cortex as revealed by fast periodic visual stimulation

Conceptual knowledge allows the categorisation of items according to their meaning beyond their physical similarities. This ability to respond to different stimuli (e.g., a leek, a cabbage, etc.) based on similar semantic representations (e.g., belonging to the vegetable category) is particularly important for language processing, because word meaning and the stimulus form are unrelated. The neural basis of this core human ability is debated and is complicated by the strong reliance of most neural measures on explicit tasks, involving many non-semantic processes. Here we establish an implicit method, i.e., fast periodic visual stimulation (FPVS) coupled with electroencephalography (EEG), to study neural conceptual categorisation processes with written word stimuli. Fourteen neurotypical participants were presented with different written words belonging to the same semantic category (e.g., different animals) alternating at 4 Hz rate. Words from a different semantic category (e.g., different cities) appeared every 4 stimuli (i.e., at 1 Hz). Following a few minutes of recording, objective electrophysiological responses at 1 Hz, highlighting the human brain’s ability to implicitly categorize stimuli belonging to distinct conceptual categories, were found over the left occipito-temporal region. Topographic differences were observed depending on whether the periodic change involved living items, associated with relatively more ventro-temporal activity as compared to non-living items associated with relatively more dorsal posterior activity. Overall, this study demonstrates the validity and high sensitivity of an implicit frequency-tagged marker of word-based semantic memory abilities.