Increased thalamic resting state connectivity as a core driver of LSD-induced hallucinations

Hallucinogenic drugs produce distinctive alterations in perception, cognition and emotion, and understanding their neural correlates may illuminate mechanisms of altered consciousness. Prior models and empirical work have implicated the thalamus as an important site of action for hallucinogens: one influential idea is that hallucinogens disrupt thalamic gating of internal and external signals, permitting increased passage of information to cortex. Neuroimaging studies have reported mixed effects of hallucinogens on thalamic metabolism and blood flow, and some recent resting-state fMRI work has suggested increased thalamocortical functional connectivity after psilocybin or LSD, but these findings had not been investigated in detail. This study tested the acute effects of a single oral dose of LSD (100 µg) on thalamic resting-state functional connectivity (rFC) in healthy volunteers using a placebo-controlled, double-blind cross-over design. The investigators hypothesised that LSD would increase thalamocortical connectivity relative to placebo and that those connectivity changes would relate to subjective LSD-induced visual and auditory alterations measured with the 5D-ASC questionnaire. The work aims to clarify whether altered thalamocortical interactions are a core neural correlate of LSD-induced hallucinations and altered consciousness.

The study used a randomised, placebo-controlled, double-blind cross-over design with at least a 7-day washout between sessions. Placebo or 100 µg oral LSD was administered at 09:00 and resting-state MRI was acquired at 11:30 (about 2.5 hours post-dose, near expected peak effects). The trial was approved by the relevant ethics committee and registered on clinicaltrials.gov. Subjective effects were assessed 3 hours after administration using the five dimensions of altered states of consciousness (5D-ASC) scale; blood for plasma LSD quantification was taken at 2 and 3 hours post-dose. Twenty-four participants began the study; after quality control for head motion and image artefacts (exclusion criteria: cumulative translation >2 mm or rotation >2°), 20 participants (10 male, 10 female; mean age 32.4 ± 10.9 years) remained in the final sample. No participant tested positive for drugs at screening or test sessions, and no serious adverse events occurred. Images were acquired on a 3T Siemens Prisma with an interleaved T2*-weighted EPI sequence (35 slices, 3.5 mm isotropic resolution, TR 1.8 s, TE 28 ms), with 300 volumes collected per resting-state run. Participants kept their eyes closed and were asked not to fall asleep. Preprocessing used SPM12 and the CONN toolbox: volumes were slice-time corrected, realigned, co-registered to T1, normalised to MNI space and smoothed (5 mm FWHM). Noise correction included scrubbing (global signal threshold z>3, composite motion >0.5 mm), regression of six motion parameters, task effects, white matter and CSF signals, and band-pass filtering (0.008–0.09 Hz). The investigators checked and found no significant differences in head motion between LSD and placebo sessions. Three functional connectivity analyses focused on the thalamus: (i) ROI-to-ROI using left and right thalamus separately against 130 target ROIs covering the brain, (ii) ROI-to-voxel using a bilateral thalamic ROI against all brain voxels, and (iii) a global correlation analysis (a voxelwise network-centrality measure computed as the mean correlation of each voxel with all other voxels). Structural ROIs were taken from the Harvard–Oxford and the automated anatomical labelling atlases. Connectivity time courses used bivariate correlations; drug effects were tested with paired two-tailed t tests and one-sample t tests where appropriate. Multiple comparisons were controlled using false discovery rate (FDR) procedures for ROI pairs and cluster-size correction (cluster-forming threshold P < 0.05, FDR) for voxelwise analyses. To probe relationships with subjective effects and exposure, the investigators correlated individual LSD plasma levels (taken immediately before MRI), and 5D-ASC dimensions 'visionary restructuralization' and 'auditory alterations', with thalamic connectivity contrasts (LSD > placebo). Masks of regions meta-analytically linked to visual or auditory hallucinations were used to extract clusters for targeted ROI correlations; all correlation analyses used Pearson's r and were FDR-corrected. They also tested whether reported sleepiness (an item on the 5D-ASC covering the whole post-dose period) related to connectivity differences to rule out vigilance confounds. Statistical calculations used SPSS v23.

Plasma LSD concentrations were undetectable at baseline and averaged 1.3 ± 0.6 ng/ml at 2 hours (pre-MRI) and 1.1 ± 0.5 ng/ml at 3 hours. Relative to placebo, LSD produced significant increases across all lower-order scales of the 5D-ASC measured 3 hours after administration (all P < 0.01). ROI-to-ROI analysis showed widespread increases in thalamic rFC under LSD versus placebo (all reported significant at P < 0.05, FDR). Specifically, the left and right thalamus showed increased connectivity with 104 of 130 investigated target regions; only one decreased connection was reported (left thalamus to vermis 10). The ROI-to-voxel analysis with a bilateral thalamic seed revealed pronounced clusters of increased connectivity in several regions, notably in the occipital lobe. Global correlation analysis, a measure of network centrality, identified a large cluster of increased centrality (841 voxels, P < 0.000001, FDR) that included bilateral thalamus (266 and 262 voxels), left and right caudate (103 and 2 voxels) and right putamen (41 voxels). A second cluster (186 voxels, P < 0.000001, FDR) comprised left caudate, left putamen and left pallidum, indicating increased centrality of striatal structures as well. When relating connectivity to plasma levels and subjective ratings, plasma LSD concentration did not significantly correlate with thalamocortical rFC derived from the ROI-to-voxel analysis (r = 0.19, P = 0.42). Connectivity between the thalamus and the right fusiform gyrus correlated significantly with the 5D-ASC 'visionary restructuralization' score (r = 0.66, P = 0.008, FDR). Connectivity between the thalamus and the right insula was reported to correlate significantly with the 'auditory alterations' scale, although the extracted text does not provide the correlation coefficient. The sleepiness item on the 5D-ASC did not significantly correlate with any of the investigated rFC measures, arguing against a simple vigilance explanation for the connectivity changes.

Mueller and colleagues interpret their principal finding as robust, global increases in thalamocortical resting-state functional connectivity after 100 µg oral LSD compared with placebo. They highlight that subjective indices of sensory alteration—'visionary restructuralization' and 'auditory alterations'—were significantly associated with increased thalamic connectivity to regions implicated in visual and auditory hallucinations (right fusiform gyrus and right insula, respectively), supporting a link between thalamocortical dysregulation and the sensory phenomenology of LSD. The authors place their results in the context of prior, partly inconsistent literature on the thalamus in hallucinogen research. While earlier PET, SPECT and fMRI studies reported decreased, unchanged or increased thalamic measures, the present findings align with at least one recent fMRI study that observed increased global connectivity of thalamic subdivisions after LSD. Differences in analytic approach and statistical thresholds may account for some discrepancies between studies. Functionally, the investigators discuss how thalamocortical interactions support large-scale integration and synchronisation across cortical regions; they note that rsfMRI connectivity relates to anatomical thalamocortical organisation and that electrophysiological phenomena such as cross-frequency coupling (CFC) may mediate interactions between low-frequency oscillations and higher-frequency local activity. The authors suggest that increased thalamic centrality and altered thalamocortical synchrony could permit otherwise suppressed or filtered information to enter cortically mediated conscious experience, possibly explaining hallucinations, synaesthesia and dream-like states under LSD. They propose the thalamic reticular nucleus as a potential locus of action because it contains GABAergic neurons, expresses 5-HT2A receptors, and modulates thalamic oscillations. The observed increases in striatal centrality are interpreted as implicating the broader cortico–basal ganglia–thalamo–cortical circuitry in LSD's effects. The paper acknowledges several limitations noted by the investigators: absence of dose–response data, modest sample size (n = 20), unavoidable unblinding due to the obvious subjective effects of LSD, possible influence of the MRI environment on subjective experience, lack of a directed causality analysis of thalamocortical interactions, and restriction of correlational analyses to visual and auditory domains to limit multiple comparisons. They also note that the 5D-ASC scales used are not specific to hallucinations, that item-level measures of sleepiness covered the whole post-dose period rather than the MRI epoch specifically, and that thalamic hemispheric time courses are correlated so lateralised findings should be interpreted cautiously. Strengths highlighted by the authors include documentation of plasma LSD concentrations, absence of differential head motion between conditions, and a sex-balanced, largely hallucinogen-naïve sample. Overall, the investigators conclude that increased thalamocortical rFC is a prominent neural effect of LSD in healthy subjects and that these connectivity changes relate to subjective sensory alterations, supporting the idea that thalamic dysregulation is a core driver of LSD-induced hallucinations.