Effective connectivity changes in LSD-induced altered states of consciousness in humans

Classic psychedelics produce profound alterations in mood, perception, cognition, and the sense of self, yet the neural mechanisms that give rise to these altered states remain incompletely understood. One influential account, the cortico‑striato‑thalamo‑cortical (CSTC) or thalamic‑filter model, proposes that psychedelics disrupt the thalamus’s capacity to gate sensory and sensorimotor information to the cortex, producing ‘‘sensory flooding’’ and cognitive disintegration. Neurotransmitter systems implicated in this model include serotonergic (notably 5‑HT2A), dopaminergic and glutamatergic pathways, but direct tests of the model’s predictions in humans have been lacking. Preller and colleagues set out to test the CSTC gating hypothesis by characterising directed (effective) connectivity among key CSTC nodes after acute administration of lysergic acid diethylamide (LSD), and after pretreatment with the selective 5‑HT2A antagonist ketanserin (Ket). Using spectral dynamic causal modelling (DCM) applied to resting‑state fMRI in a double‑blind, randomised, placebo‑controlled, crossover sample of healthy participants, the study tested two primary predictions: LSD would increase thalamus→cortex effective connectivity and reduce striatal (ventral striatum, VS) influence on the thalamus, and many of these LSD effects would depend on 5‑HT2A receptor stimulation.

Twenty‑five healthy volunteers (19 males, 6 females; mean age 25.24 years, SD 3.72, range 20–34) were screened to exclude psychiatric, neurological, cardiovascular, substance‑use and MRI contraindications. Participants provided informed consent and were asked to abstain from drugs, alcohol and caffeine for specified intervals before sessions; urine testing and other checks were used to verify compliance. A double‑blind, randomised, crossover design comprised three sessions per participant separated by two weeks: placebo pretreatment + placebo (Pla+Pla), placebo pretreatment + LSD 100 μg (Pla+LSD), and ketanserin 40 mg pretreatment + LSD 100 μg (Ket+LSD). Pretreatment occurred 60 minutes before LSD to allow ketanserin to reach peak plasma levels. Resting‑state fMRI scans (10 minutes) were acquired twice per session (approximately 75 min and 300 min after treatment); because main drug effects did not differ across sessions the analyses focussed on the earlier session. Subjective effects were assessed with the 5‑Dimensions Altered States of Consciousness (5D‑ASC) questionnaire administered 720 min after dosing. MRI data were acquired at 3T using standard EPI parameters and preprocessed with slice timing correction, realignment, spatial normalisation to MNI space and 6‑mm smoothing; motion did not exceed 3 mm for any participant. Regions of interest (ROIs) selected for effective connectivity analyses were the thalamus, ventral striatum (VS), posterior cingulate cortex (PCC), and superior temporal gyrus (Temp). Each ROI was defined as an 8‑mm radius sphere centred on MNI coordinates derived from prior task‑based contrasts in the same cohort (thalamus: −15, −8, 1; VS: 9, 8, −8; PCC: −3, −46, 31; Temp: −56, −54, 8). Nuisance regressors (six motion parameters, CSF and white matter signals) and a 128‑s high‑pass filter were applied before time‑series extraction. Spectral DCM (DCM12 within SPM12) was used to estimate fully connected models for each condition and session. To improve efficiency the investigators concatenated time series over subjects per region and condition, yielding cross‑spectral density (CSD) estimates; this step, however, precluded analyses of between‑subject correlations (for example, linking connectivity to subjective effects). Between‑condition effects were modelled using parametric empirical Bayes (PEB) with two orthogonal contrasts: Pla < [LSD + (Ket + LSD)] to assess drug effects, and (Ket + LSD) < LSD to assess 5‑HT2A blockade. Effects were reported only when posterior probability exceeded 0.95 based on Bayesian model comparison (free energy). The team also extracted signed connectivity parameters to determine putative excitatory or inhibitory influences. The authors note that the extracted ROI time series and statistics are deposited in Bitbucket.

Subjective effects assessed with the 5D‑ASC showed that ketanserin blocked LSD’s subjective effects: Ket+LSD did not differ significantly from placebo, while LSD alone produced robust subjective changes (detailed results reported elsewhere). For effective connectivity, the drug‑versus‑placebo contrast [Pla < (LSD + (Ket + LSD))] revealed multiple directed changes. Increased effective connectivity was observed from the thalamus to the ventral striatum and from the PCC to the ventral striatum. Decreases were reported for thalamus→Temp, VS→thalamus, VS→PCC and VS→Temp. Regarding intrinsic self‑connections, the PCC showed reduced self‑inhibition (interpreted as disinhibition or increased excitability) under drug. The ketanserin‑related contrast [(Ket + LSD) < LSD], isolating effects dependent on 5‑HT2A activation, showed increased effective connectivity from the thalamus to the VS and from the thalamus to the PCC, and from the VS to the Temp when ketanserin preceded LSD; decreased effective connectivity was observed from the PCC to the thalamus, and an increase in self‑inhibition of the Temp was found. Using signed parameter estimates, the analyses indicated that LSD increased thalamus→PCC effective connectivity while reducing PCC→thalamus coupling; these particular changes were blocked by ketanserin and thus attributed to 5‑HT2A receptor stimulation. The Temp region showed decreased excitability (increased self‑inhibition) and reduced thalamus→Temp coupling, also reported as 5‑HT2A‑dependent. By contrast, several LSD‑induced changes involving the ventral striatum were not blocked by ketanserin: decreased VS→thalamus and VS→PCC, decreased thalamus→Temp, increased PCC→VS, and reduced self‑inhibition of the PCC. The paper reports that only effects with posterior probability >0.95 were considered significant and that tables and supplementary material contain the detailed signed parameter estimates and variances.

The investigators interpret their findings as supporting key predictions of the CSTC thalamic‑filter model. In particular, LSD increased effective connectivity from the thalamus to cortical targets (notably the PCC) via a mechanism dependent on 5‑HT2A receptor stimulation, while LSD decreased the influence of the ventral striatum on the thalamus independent of 5‑HT2A blockade. Together, these changes are taken to indicate a partial ‘‘opening’’ of the thalamic filter that is selective across cortical targets rather than a global cortical inundation. Increased thalamus→PCC coupling and concurrent disinhibition of the PCC are discussed in relation to predictive coding accounts, where gating is implemented by modulating the precision of ascending prediction errors. The PCC’s role in arousal, awareness and the balance between internal and external cognition is highlighted, and the current results are related to prior reports of thalamo‑cortical and PCC‑related alterations after psychedelic administration, decreased alpha power and changes in default mode network connectivity. The authors suggest that decreased thalamus→Temp coupling and increased Temp self‑inhibition may bear on emotional and social information processing and might be relevant to psychedelics’ potential antidepressant effects, while noting that the present data are acute and cannot speak to long‑term therapeutic outcomes. Regarding pharmacology, the pattern that ketanserin blocked thalamus‑PCC effects but did not block several VS‑related changes leads the authors to hypothesise that non‑5‑HT2A mechanisms—potentially dopaminergic pathways acting on the striatum—may contribute to some LSD effects. They emphasise, however, that the present study cannot conclusively determine contributions from other receptor systems and recommend follow‑up studies, for example using dopamine antagonists or different LSD doses. The authors acknowledge several limitations: ROI selection was restricted to nodes implicated by the CSTC model and the current cohort, so distributed effects outside these regions were not tested; concatenating time series across subjects precluded subject‑level correlations between connectivity and subjective reports (an issue they plan to address using hierarchical PEB); and the absence of a ketanserin‑only (Ket+Pla) condition prevents characterising ketanserin’s standalone effects on connectivity. They also propose methodological extensions such as smaller, subregion‑level ROIs and inclusion of additional CSTC hubs like medial prefrontal cortex. Finally, the study is presented as a demonstration of spectral DCM combined with PEB for group‑level pharmacological resting‑state fMRI, with the authors arguing that the results advance mechanistic understanding of LSD’s action in the human brain and have implications for developing novel therapeutics and for theories of altered conscious states.