Neuroimaging Correlates of Distinct Phenomenological Subtypes of Auditory Verbal Hallucinations in Schizophrenia: A Critical Review of fMRI and Structural Connectivity Evidence
Sushil Kumar Sompur Vasanthkumar
Preprints.org November 6, 2026 preprint DOI: 10.20944/preprints202611.0642.v1 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Critical review |
|---|---|
| Population | People with schizophrenia-spectrum disorders experiencing auditory verbal hallucinations, as studied in functional MRI and structural/diffusion connectivity research |
| Key findings | Argues that three decades of neuroimaging have failed to converge on a single neural substrate for auditory verbal hallucinations partly because phenomenological heterogeneity is unmodelled. State studies converge on speech-production and speech-perception systems, with left inferior frontal and right middle temporal activation preceding onset by 6–9 seconds, while trait studies converge on reduced left superior and middle temporal engagement. Arcuate fasciculus findings diverge by metric and segment, and direct imaging comparisons within phenomenologically defined subtypes remain almost entirely absent. |
Abstract
Auditory verbal hallucinations (AVH) affect most people with schizophrenia and persist in a substantial minority despite adequate antipsychotic treatment, including clozapine. Three decades of neuroimaging have nonetheless failed to converge on a single neural substrate. This critical review tests the hypothesis that this inconsistency partly reflects unmodelled phenomenological heterogeneity. We synthesised functional MRI and structural/diffusion connectivity studies of AVH in schizophrenia-spectrum disorders, separated hallucination-state from trait designs, and mapped each finding onto empirically derived phenomenological subtypes using three explicit evidence tiers. State studies converge on speech-production and speech-perception systems, with left inferior frontal and right middle temporal activation preceding onset by 6–9 s, whereas trait studies converge on reduced left superior and middle temporal engagement. Arcuate fasciculus findings diverge by metric and segment: anisotropy and magnetisation transfer increase in whole-tract studies, decrease in segment-resolved analyses, and tract length rises with severity. Provisional subtype-specific signals emerge: parahippocampal deactivation before memory-replay experiences; right inferior frontal lateralisation tracking negative valence in low-complexity, commanding voices; reduced temporal-to-frontal effective connectivity in inner-speech-type hallucinations; and associative rather than primary auditory generators in deafferentation-related hallucinosis. Direct imaging comparisons within phenomenologically defined subtypes remain almost entirely absent. We propose a subtype–substrate framework and a research programme combining structured phenomenology, within-subject state and trait acquisition, segment-resolved diffusion modelling and stratified neuromodulation as the most tractable route to resolving the field’s inconsistencies and to mechanism-targeted treatment.