A review of functional and structural neuroimaging studies to investigate the inner speech model of auditory verbal hallucinations in schizophrenia

Abstract

Although the pathophysiology of auditory verbal hallucinations remains uncertain, the inner speech model remains a prominent theory. A systematic review and meta-analyses of both functional and structural neuroimaging studies were performed to investigate the inner speech model. Of the 417 papers retrieved, 26 met the inclusion criteria. Meta-analyses found the left insula to be significantly active during auditory verbal hallucinations and to have a significantly reduced grey matter volume in hallucinators. Dysfunction of the left insula may contribute to the misattribution of inner speech due to its suggested roles in both inner speech production and the salience network. No significant activity was found at Broca's area or Heschl's gyrus during auditory verbal hallucinations. Furthermore, no structural abnormalities were found at these sites or in the arcuate fasciculi. Overall, evidence was found to both support and oppose the inner speech model. Further research should particularly include a systematic review of task-based trait studies with a focus on inner speech production and self-referential processing, and analyses of additional language-related white matter tracts.

Fig. 1. A flow diagram illustrating the study selection process [51].

VBM voxel-based morphometry, AVHs auditory verbal hallucinations, DTI diffusion tensor imaging.

Fig. 2. The location of the clusters of significant activity during auditory verbal hallucinations (AVHs), found by the meta-analyses of state studies.

The location of the significant clusters, in coronal (left), sagittal (middle) and axial (right) planes, are highlighted in red. a The significant cluster in the left hemisphere, centred in the inferior parietal lobule and also including the postcentral gyrus and the insula. b The significant cluster in the right anterior lobe of the cerebellum, centred in the culmen and also including the dentate. c The result of the second meta-analysis of state studies (which included the Diederen et al. [28] study rather than the Sommer et al. [27] study) finding a significant cluster in the left hemisphere, centred in the inferior parietal lobule and also including the postcentral gyrus and the insula.

Fig. 3. The result of the sensitivity analysis of voxel-based morphometry studies showing the location of the cluster of significant grey matter volume reduction in the left hemisphere, centred in the claustrum and also including the putamen and insula, in hallucinators compared to non-hallucinators.

The location of the significant cluster, in sagittal (left), coronal (middle) and axial (right) planes, are highlighted in red.

Fig. 4. A forest plot showing no significant difference between hallucinators and non-hallucinators for the fractional anisotropy of the left arcuate fasciculus.

SD standard deviation, IV inverse variance, CI confidence interval, FA fractional anisotropy.

Fig. 5. A forest plot showing no significant difference between hallucinators and non-hallucinators for the fractional anisotropy of the right arcuate fasciculus.

SD standard deviation, IV inverse variance, CI confidence interval, FA fractional anisotropy.