Brain substates induced by DMT relate to sympathetic output and meaningfulness of the experience

Pasquini and colleagues situate this study in the context of two complementary aspects of selfhood: a bodily or interoceptive self that depends on integration of autonomic and visceral signals, and a narrative self that depends on medial temporal and midline parietal structures associated with autobiographical memory and self-referential thought (the default mode network, DMN). They note that serotonergic psychedelics such as N,N-dimethyltryptamine (DMT) produce a rapid, short-lasting but intense alteration of consciousness with pronounced sensory immersion and changes in autonomic physiology, particularly transient increases in sympathetic tone measured via heart rate. Earlier neuroimaging studies of psychedelics have reported global connectivity changes and DMN disruptions, but the authors argue that the fast dynamics of intravenous DMT make it a useful probe for studying temporally evolving brain states and linked peripheral physiology using dynamic analysis methods applied to simultaneous fMRI and electrocardiogram (EKG) recordings. The study therefore aims to identify time-resolved brain activity ‘‘substates’’ that differentiate DMT from placebo and to link those substates to autonomic indices and subjective reports. Specifically, the investigators apply dynamic similarity-matrix analyses and graph-theoretical community detection to resting-state fMRI acquired with concurrent EKG in a within-subject, placebo-controlled design, to characterise the natural progression of neural and cardiac signatures unfolding immediately before, during, and after intravenous DMT administration. The goal is to test whether distinct substates arise under DMT, how long they persist, and whether substate-specific regional activations or deactivations relate to auditory distortions, ego-dissolution, meaningfulness of the experience, and heart-rate changes.

Design and participants: The dataset derives from a single-blind, placebo-controlled, counterbalanced pharmacological imaging study. Twenty participants were initially enrolled and completed two visits separated by two weeks; after quality control for head motion the analysed sample comprised 14 healthy volunteers (4 female). The extraction does not clearly report the sample’s mean and standard deviation for age within the Methods text provided. Each visit included a resting-state scan and a separate ratings scan; this report focuses on the resting-state session. Intervention, timing and subjective ratings: Participants received intravenous injections of either 20 mg DMT fumarate in 10 mL saline or 10 mL sterile saline (placebo), administered over 30 s and flushed with saline over 15 s, in counterbalanced order across visits. Resting-state fMRI acquisition lasted 28 minutes with the injection given at the end of the eighth minute. Approximately 30 minutes after recovery, participants completed 25 visual-analogue subjective items (0–1 scale) probing intensity, primary sensory distortions (including auditory), disembodiment/dissociation, ego-dissolution and meaningfulness. An EEG cap was fitted and EEG data were recorded but not analysed for this study. Neuroimaging and EKG acquisition and preprocessing: Functional MRI was acquired on a 3T scanner; whole-brain structural images were also collected. The paper refers to more detailed acquisition parameters in a parent publication. Standard preprocessing steps were applied using AFNI, FSL and in-house code: despiking, slice timing correction, motion correction and registration to an MNI template, scrubbing using a framewise displacement threshold of 0.4 mm (participants with >20% scrubbed volumes were excluded), spatial smoothing (6 mm FWHM), band-pass filtering (0.01–0.08 Hz), detrending, and regression of motion and anatomical nuisance regressors. EKG signals were sampled at 5 kHz, synchronised to MRI, and processed using the hrvanalysis Python module to derive continuous heart-rate estimates from RR intervals. Dynamic analysis and community detection: Voxel-level BOLD data were summarised by extracting mean signals from 100 cortical parcels (Schaefer atlas) and 12 subcortical regions (AAL). For each condition (DMT, placebo), time-resolved similarity matrices were computed by Pearson correlation between whole-brain node activity patterns at different time points, yielding matrices that reflect temporal homogeneity of brain-wide activity. A group-averaged subtraction matrix (DMT minus placebo) was then thresholded across Pearson r values from 0.20 to 0.30 in 0.01 steps; for each threshold the Louvain community detection algorithm (resolution parameter G = 1, asymmetric negative-weight handling) was run to identify temporal modules or ‘‘brain activity substates.’n Selection and statistical testing: Brain substates were retained if they (i) differentiated DMT from placebo, (ii) occupied at least 10% of the scan duration, and (iii) were differentially occupied in pre- versus post-injection epochs (chi-square tests, p < 0.05). Individual nodal activation maps for each substate were estimated via general linear models producing b-maps, which were compared across conditions using paired t-tests with false discovery rate (FDR) correction (q < 0.05). For regions showing significant DMT–placebo differences, averaged BOLD time series were extracted. Subjective change scores (DMT minus placebo) on selected scales were correlated with regional activity changes using Pearson correlations (p < 0.05). Paired t-tests compared subjective ratings across conditions (threshold p < 0.005), and heart-rate averages during specific substates were compared across conditions and correlated with regional activity and meaningfulness scores. A median split on heart-rate change was used for exploratory subgroup comparisons. The optimal threshold for the dynamic analysis was chosen as r = 0.25 because it maximised the number of substates differentially occupied pre- versus post-injection.

Sample and preprocessing outcome: After motion-based scrubbing, 14 participants remained for analysis. Resting-state scans were 28 minutes with injection at 8 minutes, and concurrent EKG was available for heart-rate analyses. Identification of dynamic brain substates: Using the group-mean subtraction (DMT minus placebo) time-resolved similarity matrix and Louvain community detection at an optimal threshold of r = 0.25, the investigators identified three major brain activity substates meeting the occupancy and pre/post-injection criteria. One substate (State 2) predominated before injection and late in the scan, accounting for about 34% (≈2.7 min) of the pre-injection period and 10% (≈2.0 min) post-injection; it featured orbitofrontal activation and, under DMT only, subcortical and medial/lateral parietal activations with deactivation of primary sensory areas. The first post-injection substate (State 4) increased in occupancy after injection, representing 19% (≈3.8 min) of the post-injection period (vs 6% pre), and was marked by widespread primary sensory and cingulo-opercular activations, with prefrontal deactivations under both conditions and additional medial parietal deactivations under DMT. A subsequent post-injection substate (State 5) occurred immediately after State 4, occupying 18% (≈3.6 min) post-injection, and resembled State 4 but showed pronounced temporopolar/anterior temporal pole activations. Substate-specific DMT versus placebo contrasts: Voxelwise comparisons of individual substate b-maps revealed that only the first post-injection substate (State 4) showed significant region-level differences between DMT and placebo after FDR correction. State 4 under DMT exhibited increased activity in the right superior temporal lobe and decreased activity in the left hippocampus and bilateral medial parietal regions overlapping the precuneus and posterior cingulate cortex. Mean framewise displacement did not correlate significantly with these regional effects, reducing the likelihood that motion confounded the findings. Time-series difference plots (DMT minus placebo) showed sustained superior temporal hyperactivation and hippocampal/medial parietal hypoactivation in the minutes immediately after injection. Relations to subjective experience and heart rate: Subjective ratings acquired after recovery were higher under DMT than placebo for auditory distortions (t(13) = 3.42, p = 0.005), ego-dissolution (t(13) = 9.17, p < 0.0001) and meaningfulness (t(13) = 7.96, p < 0.0001). Right superior temporal hyperactivation in State 4 correlated positively with the DMT–placebo change in auditory distortion intensity (R(12) = 0.57, p = 0.03). Deactivations in the hippocampus and medial parietal cortex did not correlate with ego-dissolution scores (R(12) = 0.28, p = 0.34) but correlated negatively with change in meaningfulness (R(12) = -0.61, p = 0.02), such that stronger deactivation associated with greater reported meaningfulness. Cardiac findings: Consistent with prior work, mean heart rate increased during the early-to-peak phase of the DMT experience. Heart-rate increases during the identified post-injection substate correlated positively with hippocampal/medial parietal deactivation and negatively with meaningfulness; participants with the largest heart-rate increases showed weaker deactivations and lower meaningfulness ratings. A median split separated participants into lower versus higher heart-rate responders for exploratory comparisons, though detailed subgroup statistics are not fully reported in the extracted text.

Pasquini and colleagues interpret their findings as evidence that DMT induces a rapid sequence of dynamic brain substates that can be separated from placebo and linked to both autonomic physiology and subjective phenomenology. They emphasise that a DMN-associated substate predominates before injection and reappears as participants recover, consistent with normal waking self-referential processing. Two distinct post-injection substates were identified: an initial post-injection state dominated by sensory, attentional and interoceptive activations together with medial parietal and hippocampal deactivations, and a later state showing anterior temporal pole hyperactivity that the authors suggest may relate to recovery of higher-level semantic or meaning-related cognition as the peak effects subside. The discussion highlights two specific brain–experience relationships. First, increased right superior temporal activity during the early post-injection substate correlated with the intensity of auditory distortions, aligning with the superior temporal lobe’s role in auditory perception and prior reports of sensory cortex involvement under psychedelics. Second, deactivation of medial parietal and hippocampal regions—components of the DMN implicated in narrative selfhood and autobiographical memory—was associated with higher ratings of the meaningfulness of the experience, though not with ego-dissolution in these data. The authors note consistency between these observations and earlier psilocybin, LSD and ayahuasca studies reporting DMN perturbations and hippocampal decoupling under psychedelics. A further interpretation links autonomic regulation to subjective meaning: increased sympathetic output (indexed by heart rate) was observed during peak DMT effects, and participants with the largest heart-rate responses displayed weaker medial parietal/hippocampal deactivations and lower meaningfulness ratings. From this pattern the authors propose that optimal sympathetic regulation—or the ability to ‘‘let go’’—may facilitate the neural deactivations that support meaningful psychedelic experiences, a concept paralleling therapeutic constructs such as acceptance or surrender. Limitations acknowledged by the authors include the relatively small final sample size, which may have limited statistical power and the spatial extent of detectable effects; the inherent heterogeneity of psychedelic experiences shaped by set and setting; and the fact that statistical thresholds may have constrained the identification of more widespread differences. They recommend future repeated-measures studies across dosing and days to characterise inter- and intra-individual variability, and multimodal clinical research to test whether autonomic changes predict therapeutic outcomes. Finally, they suggest their dynamic, graph-theoretical approach may be a productive route to link transient neural substates with peripheral physiology and phenomenology during fast-acting psychedelic states.