LSD’s effects on slow brain activity and local brain coordination linked to receptor levels

This study analyzed imaging data from the OpenNeuro Dataset ds003059 (Carhart‐Harris et al. ; Carhart‐Harris et al. ). The original sample included 20 subjects.

Once recruited, participants provided written informed consent and were screened for both physical and mental health conditions (Carhart‐Harris et al. ). Exclusion criteria included being under 21 years of age, a personal diagnosis of psychiatric illness, a family history of psychotic disorders, lack of prior exposure to classic psychedelics (e.g., LSD, mescaline, psilocybin/magic mushrooms, or DMT/ayahuasca), use of psychedelics within 6 weeks prior to the first scan, pregnancy, excessive alcohol consumption (more than 40 units per week), or any medical condition that would disqualify them from participating.

The original study received approval from the National Research Ethics Service Committee London‐West London and was conducted in accordance with the revised Declaration of Helsinki (2000), the International Council for Harmonisation's Good Clinical Practice guidelines, and the National Health Service Research Governance Framework. The research was sponsored by Imperial College London and performed under a Home Office license for work involving Schedule 1 drugs (Carhart‐Harris et al. ).

Participants deemed eligible for the study attended two separate fMRI sessions, spaced at least 14 days apart. On one day, they received a placebo (PBO), and on the other, they were administered LSD. The order of these conditions was counterbalanced among participants, who remained blind to the treatment sequence, though the researchers were aware of it. On the LSD imaging days, participants were given 75 μg of LSD intravenously, delivered via a 10‐mL solution infused over a 2‐min period, followed by a saline infusion. This dosage is considered a moderate, fully hallucinogenic dose, capable of producing marked alterations in the state of consciousness (Carhart‐Harris et al. ; Passie et al. ). Participants reported experiencing the initial effects of the drug between 5 and 15 min after dosing, with peak effects occurring around 60‐ to 90‐min postdosing. MRI scanning began approximately 70‐min postdosing and lasted roughly 1 h. More information is available in the original study (Carhart‐Harris et al. ).

Participants underwent three consecutive fMRI scans of 7:20 min each during each session: the first and third scans were resting‐state scans conducted with eyes closed, while the second scan was acquired while participants listened to music. For this study, only the resting‐state scans (no music) were included in the analysis (rest1 and rest2). Therefore, four scans were available for each subject: PBO‐rest1, PBO‐rest2, LSD‐rest1, and LSD‐rest2.

Imaging data available on OpenNeuro Dataset ds003059 were already preprocessed (Carhart‐Harris et al. ; Carhart‐Harris et al. ). More information about anatomical and data acquisition, preprocessing and motion correction procedures is reported in the(Sections e1–e3) and in the original paper (Carhart‐Harris et al. ; Carhart‐Harris et al. ).

Considering the existing literature on LSD (Canal and Murnane ; Halberstadt and Geyer ; Moreno et al. ; Passie et al. ; Perry and Perry ; Seeman et al. ) and the available receptor maps at the time of study conduction, the following spatial features were retrieved from: acetylcholine receptors (both nicotinic—AChN—and muscarinic—AChM1), acetylcholine transporter (VAChT), dopamine receptors (D1 and D2), dopamine transporter (DAT), glutamate receptors (mGlutR5 and NMDA), serotonin receptors (5‐HT1A, 5‐HT1B, 5‐HT2A, and 5‐HT4) and transporter (SERT), opioid receptors (morphine—MOR—and kappa—KOR), cannabinoid receptors (CB1), and GABA receptors (GABAa).

As above,was used to convert ALFF/ReHo‐shift maps to a common space of choice () (Buckner et al. ; Wu et al. ).was then used to estimate the vertex‐wise spatial correlation (Spearman rho) between the group contrast (now in) and the selected feature map (the single receptor/transporter density map) across the whole surface (Markello et al. ). To estimate statistical significance, null models were computed by rotating the reference images and deriving the null distribution of correlation coefficients (spatial nulls, = 1000) (Vázquez‐Rodríguez et al. ).‐values were adjusted for multiple comparisons using false discovery rate (FDR‐) correction (Benjamini and Hochberg ). As a sensitivity analysis, receptor‐enrichment procedures were repeated using mALFF‐shift maps.

As reported in the original study (Carhart‐Harris et al. ), one subject did not complete the fMRI scans due to anxiety, while four other subjects were discarded from the group analyses due to excessive head movement (see Section e2 in the). The final sample included 15 subjects (four females; mean age, 30.5 ± 8.0 years).

ALFF values demonstrated a statistically significant widespread decrease under LSD compared to PBO (i.e., a negative ALFF‐shift) in both rest1 and rest2 runs (Figure ). Conversely, no ALFF increases were observed. The regions exhibiting negative ALFF‐shift values in both runs (Figure ) included the following: bilateral primary visual cortex, cuneus, precuneus, retrosplenial cortex, right fusiform gyrus, supramarginal gyrus, posterior cingulate, dorsolateral prefrontal cortex (dlPFC), middle frontal gyrus, inferior frontal gyrus (IFG), right supplementary motor area (SMA), sensorimotor cortex (SMC), and posterior parietal cortex (PPC). See Table for region details and coordinates.

Similarly to ALFF, ReHo showed several regions with lower, but not higher, values in LSD compared to PBO conditions in both rest runs (see Figure ). The regions exhibiting negative ReHo‐shift values in both rest1 and rest2 runs (Figure ) included the following: bilateral supramarginal gyrus, bilateral IFG, SMA, fusiform gyrus, primary visual cortex, SMC, and dlPFC (where significant ALFF reductions in LSD were also present), left superior temporal gyrus, bilateral hippocampus, bilateral thalamus, amygdala, and bilateral cerebellum (changes exclusive to ReHo results). See Table for region details and coordinates. Results from mALFF‐shift analyses were highly consistent with those from ALFF, confirming widespread decreases under LSD. The only differences concerned minor variations in cluster extent and the emergence of some positive shifts, likely reflecting relative effects introduced by normalization (Figure ).

Post hoc power analyses are available in Table . The achieved power for ALFF‐shift analyses was 81.1% at rest1 and 81.5% at rest2; for ReHo‐shift analyses, achieved power was 81.5% at rest1 and 79.6% at rest2.

Vertex‐wise correlations between rest1 and rest2 runs showed that test–retest reliability was high for ALFF‐shift (rho = 0.80, = 0.001) and moderate for ReHo‐shift (rho = 0.46, = 0.001). Vertex‐wise correlations between ALFF‐shift and mALFF‐shift maps, as a sensitivity analysis, showed strong correlations both for rest1 ( = 0.96, < 0.001) and rest2 ( = 0.96, p < 0.001). See Figures andfor a graphical representation of results on cortical surfaces.

ALFF‐shift*ReHo‐shift conjunction maps (during rest1 and rest2) are visualized in Figure ; overlapping regions are listed in Table . Overlapping regions between ALFF‐shift and ReHo‐shift were localized bilaterally in the occipital lobe, frontoparietal cortex, and sensorimotor regions. Overall, ALFF was more sensitive to LSD‐induced changes in dorsal brain regions, whereas ReHo highlighted LSD‐induced changes in more ventral brain regions and the cerebellum. ALFF and ReHo results were moderately correlated (rest1: rho = 0.36, = 0.001; rest2: rho = 0.56, = 0.001; see Figure ).

Overall, our enrichment analyses revealed only significant positive correlations between ALFF/ReHo‐shift and receptor maps. ALFF‐shift and ReHo‐shift maps showed a significant positive correlation with D2 and 5‐HT1A receptor maps (negative ALFF/ReHo‐shift corresponding to lower receptor density) across both rest1 and rest2 scans (Figure , ALFF‐shift rest1: D2 FDR‐ = 0.009, 5‐HT1A FDR‐ = 0.009; ALFF‐shift rest2: D2 FDR‐ = 0.045, 5‐HT1A FDR‐ = 0.036; ReHo‐shift rest1: D2 FDR‐ = 0.002, 5‐HT1A FDR‐ = 0.002; ReHo‐shift rest2: D2 FDR‐ = 0.006, 5‐HT1A FDR‐ = 0.006). 5‐HT1A and D2 cortical maps are available in Figure . Furthermore, LSD‐induced effects, as measured by ReHo‐shift, were spatially aligned across both scans with 5‐HT4 receptors (rest1: FDR‐ = 0.002; rest2: FDR‐ = 0.006).

ALFF results were relatively stable across scans, with an alignment with DAT approaching significance in rest2—although not surviving correction for multiple comparisons (Figure ; rest2: = 0.005, FDR‐ = 0.258). The same analyses based on mALFF yielded consistent findings (Figure ). In contrast, ReHo findings demonstrated significant alignment with various neurochemical receptors across both scans (D2: rest1 FDR‐ = 0.043, rest2 FDR‐ = 0.006; 5‐HT1A: rest1 FDR‐ = 0.002, rest2 FDR‐ = 0.006; 5‐HT4: rest1 FDR‐ = 0.002, rest2 FDR‐ = 0.006; MOR: rest1 FDR‐ = 0.004, but rest2 = 0.024, FDR‐ = 0.086). However, ReHo‐shift during rest1 exhibited a spatial alignment with an even broader array of receptors in comparison to rest2 (AChM1: FDR‐ = 0.028; D1: FDR‐ = 0.002; mGluR5: FDR‐ = 0.002; NMDA: FDR‐ = 0.023; 5‐HT2A: FDR‐ = 0.023; 5‐HT6: FDR‐ = 0.002; KOR: FDR‐ = 0.002; CB1: FDR‐ = 0.002; GABAa: FDR‐ = 0.002) (Figure ).

Our study reports a brain‐wide reduction in ALFF across both sensory and associative regions, including the somatosensory and visual cortices, as well as regions belonging to the default mode network (DMN) (precuneus and posterior cingulate) (Baghdadi et al. ) and the frontoparietal network (FPN) (dlPFC, inferior and posterior parietal lobule, and right temporoparietal junction) (Vendetti and Bunge ). The biological and functional significance of ALFF remains unclear (Merola et al. ). In fact, ALFF cannot be considered a metabolic proxy for brain activity, as it correlates with cerebral blood volume but not with the metabolic rate of glucose, the metabolic rate of oxygen, or cerebral blood flow (Deng et al. ). By definition, a decrease in ALFF corresponds to reduced power in the low‐frequency fMRI range (Yang et al. ; Zang et al. ), suggesting that local neuronal activity is dominated by desynchronized microstates rather than stable, strong fluctuations. This change may be driven by an increased contribution of higher‐frequency oscillations, which normally have lower amplitude and reflect more dynamic states (Klar et al. ; Thompson and Fransson ; Zhang et al. ). Consistently, LSD has been shown to induce increases in cross‐frequency co‐activation and high‐frequency harmonics, supporting a shift toward more diverse and flexible brain dynamics (Atasoy et al. ; Luppi et al. ).

From an information theory perspective, "noisy," desynchronized patterns indicate higher entropy (Carhart‐Harris ; Carhart‐Harris et al. ). Thus, LSD‐induced ALFF reductions may reflect an increase in entropy, aligning with the EBH (Carhart‐Harris ; Carhart‐Harris et al. ). The EBH posits that typical waking consciousness operates near a critical zone, straddling order (low entropy) and disorder (high entropy). Studies have consistently reported LSD‐induced increases in brain entropy (Carhart‐Harris ; McCulloch et al. ), suggesting a shift closer to this critical zone, potentially enhancing information processing and explaining LSD‐related phenomena such as ego‐dissolution, sensory overload, and visual hallucinations (Atasoy et al. ; Lebedev et al. ; Luppi et al. ). Notably, Lebedev and colleagues reported an LSD‐induced increase in Sample Entropy across sensory and higher‐order networks (Lebedev et al. ), with a pattern that mirrors the ALFF reductions observed in our study. Future studies may investigate the relationship between ALFF and entropy to confirm their possible inverse association. In brief, reduced ALFF in LSD may reflect an amplification of high‐frequency harmonic activity, measurable as a relative reduction in the dominance of low‐frequency fluctuations. A smaller difference in power between low and high frequencies is a characteristic spectral feature of entropic increases (Damiani et al. ).

ReHo reductions were primarily observed within sensory areas such as the visual and sensorimotor cortices, as well as in subcortical regions including the thalamus and amygdala. Previous studies on the general population during cognitive tasks have documented that a brain‐wide reduction in ReHo is associated with increased between‐network connectivity and decreased within‐network connectivity (Damiani et al. ; Tommasin et al. ). Our results suggest the presence of similar dynamics under LSD. Previously, reductions in within‐network FC (referred to as "integrity") and increases in between‐network FC (i.e., decreased "segregation" or "modularity") were described during exposure to LSD—particularly in sensory regions—and were associated with visual hallucinations and ego‐dissolution (Bedford et al. ; Carhart‐Harris et al. ; Dai et al. ; Linguiti et al. ; Luppi et al. ; McCulloch et al. ; Moujaes et al. ; Müller et al. ; Tagliazucchi et al. ). Interestingly, LSD was found to exert differential effects on sensory versus associative areas; for instance, it increased brain‐wide connectivity in visual and sensorimotor cortices while reducing connectivity in associative areas (Preller et al. ). In line with these findings, Girn et al. () observed a reduced differentiation between sensory unimodal and association multimodal areas, revealing a flattening of the brain's functional hierarchy under LSD. Consistent with these previous observations, we found a specific LSD‐induced reduction of ReHo in sensory regions, which is indicative of a fragmentation of network integrity.

Conversely, during propofol‐induced anesthesia, increases in ReHo have been observed (Huang et al. ) alongside heightened network integrity (Schrouff et al. ). In other words, states of "diminished consciousness" (such as anesthesia) display increased segregation, integrity, and ReHo, patterns essentially opposite to those observed in psychedelic states—which are often characterized as "expanded" states of consciousness (Carhart‐Harris ; Carhart‐Harris et al. ).

Finally, as for ALFF, ReHo reductions may also be consistent with an increase in entropy (Trevino et al. ). Previous investigations have consistently reported that higher ReHo values are associated with simpler forms of information processing (Jiang et al. ; Jiang and Zuo ), which may be reflected by lower levels of entropy according to EBH. In this regard, one study directly demonstrated a negative correlation between entropy and ReHo in a sample of patients with Alzheimer's disease (Wang et al. ).

ALFF‐ and ReHo‐shift patterns showed a moderate vertex‐wise correlation during both rest1 and rest2. Consistent with previous fMRI studies that employed both ALFF and ReHo (An et al. ; Shen et al. ; Sun et al. ; Wang et al. ; Yuan et al. ; Yue et al. ), these measures were found to provide both congruent and complementary information regarding neural functional dynamics. Concerning their similarities, ALFF and ReHo converged towards reductions in somatosensory and visual regions. This finding suggests a disruption of the brain functional hierarchy, favoring brain‐wide integration of the visual and somatosensory networks at the expense of segregation.

However, LSD‐induced reconfiguration also revealed notable differences between ALFF and ReHo. Specifically, ALFF selectively decreased in associative regions belonging to the DMN and FPN, whereas ReHo selectively decreased in ventral and subcortical regions such as the thalamus, amygdala, and cerebellum. Moreover, we observed that ALFF‐shift exhibited greater stability between rest1 and rest2 compared to ReHo, with vertex‐wise correlations of 0.80 versus 0.46, respectively. Previous studies have similarly reported that ALFF generally demonstrates slightly higher test–retest reliability than ReHo (Chen et al. ; Golestani et al. ). It should be noted, however, that test–retest stability of fMRI metrics varies across different measures and conditions. For example, general anesthesia—a state of diminished consciousness—has been shown to improve test–retest reliability of various fMRI metrics, including ALFF and ReHo (Vedaei et al. ). Therefore, it can be hypothesized that LSD, which is associated with expanded or altered consciousness, might reduce test–retest reliability by enhancing neural variability.

Neurochemical enrichment analyses confirmed that, although not replicated across both runs, the known main target of LSD, namely, 5‐HT2A, was significantly and spatially associated with ReHo results. Our hypothesis that LSD‐induced effects on ALFF and ReHo are spatially aligned with the receptor maps beyond 5‐HT2A alone was also confirmed, as D2 and 5‐HT1A receptors were consistently associated with both these measures across both runs. In this regard, it is important to note that previous studies have shown that LSD may have comparable or higher binding to 5‐HT1A (Ki ~ 3 nM) and D2 receptors (Ki ~ 2.5 nM) than to 5‐HT2A (Ki ~ 4 nM) (Ray ; Rickli et al. ). Notably, all the significant spatial correlations were positive, indicating that the greater the ALFF and ReHo reductions—i.e., the primary LSD‐induced effect we observed—the lower the receptor density. Furthermore, the overlap between ALFF‐/ReHo‐shift and the cortical distribution maps of 5‐HT1A and D2 receptors was minimal (see Figures ,, and).

The full biological and functional significance of these results is difficult to posit, given the low sample size, the use of normative receptor profiles, and limitations imposed by receptor distributions (e.g., D2 receptors show low cortical distribution). Nevertheless, different hypotheses may be proposed. First, neurochemical enrichment analyses may reflect direct or indirect modulations rather than direct receptor binding. In fact, previous studies observed that FC reductions after MDMA intake were not linked to 5‐HT2A, as expected by direct binding potentials, but rather to 5‐HT1A (Dipasquale et al. ). Similarly, another study conducted on magnetoencephalography data rather than fMRI also showed no significant spatial association between LSD effects and 5‐HT2A receptors (Shinozuka et al. ). Secondly, a recent paper described that 5‐HT2A psychedelic agonism can alter neuronal and hemodynamic functioning (Padawer‐Curry et al. ). Current results may then be influenced by divergent neurovascular coupling, in line with ALFF reflecting more complex dynamics than simple neuronal activation/deactivation. Thirdly, neural populations rich in specific receptors such as 5‐HT1A and D2 may "prevent" LSD‐induced ALFF/ReHo reductions, in a similar manner to the "selective sparing" potential observed for similar receptors in other structural studies dealing with other neuropsychiatric disorders (Tarchi et al. ; Wiesman et al. ). Indeed, although LSD provokes silencing of serotonin receptors in the dorsal raphe through the activation of 5‐HT1A inhibitory autoreceptors (Aghajanian et al. ), 5‐HT1A agonists may induce burst firing in other brain regions, including those rich in noradrenergic and dopaminergic neurons (Díaz‐Mataix et al. ; Lejeune and Millan ; Piercey et al. ). In this regard, an early report showed that 5‐HT1A agonists may antagonize the effects of psychotomimetic drugs on dopaminergic neurons (Piercey et al. ), an observation that aligns with current results showing high 5‐HT1A receptor density as being associated with a lack of LSD‐induced ALFF/ReHo reductions.

Our results revealed that while ALFF remained relatively stable between rest1 and rest2, ReHo exhibited notable differences between these sessions. This suggests that ReHo may be more volatile and sensitive to short‐scale temporal dynamics under LSD, whereas ALFF might better reflect more stable characteristics of BOLD signals (see Section ).

A recent review by Linguiti et al. () highlighted key challenges in the field of neuroimaging studies on psychedelics, including considerable variability in study designs, reliance on small and often overlapping samples, and inconsistent control of Type I error. To improve the reliability of our results, we repeated analyses on both rest1 and rest2 runs. All our results demonstrated moderate to robust test–retest reliability.

Despite the particular attention paid to addressing head motion during the preprocessing stages, a small yet significant difference in FD mean values still remained between LSD and PBO conditions (see Section e2 in the). Head motion can introduce artifacts in both ALFF and ReHo computations, potentially affecting our LSD‐PBO results. Nevertheless, the test–retest stability of our findings supports the robustness of the ALFF/ReHo results.

A potential limitation of this study is that the second resting‐state scan followed a music‐listening session, which may have influenced neural activity independently of LSD. Previous studies on this dataset have used both runs (McCulloch et al. ), a choice that is particularly valuable given the relatively small sample size. However, those studies typically averaged the two runs before analysis (Luppi et al. ; Speth et al. ), which may inadvertently mix potential music‐related effects with drug effects. In contrast, we analyzed the two runs separately and focused our interpretation on results consistently replicated across both scans, thereby reducing the likelihood that our main findings are driven by music‐related effects.

As discussed above, this is the first study to employ both ReHo and ALFF to investigate acute psychedelic‐induced neural dynamics. The lack of previous findings for direct comparison with our results limits our conclusions, so as to avoid overly speculative hypotheses. Despite this, we have sought to support our inferences by discussing our results in the context of various previous fMRI analysis methodologies and by incorporating neurochemical enrichment analyses. Future studies could strengthen these findings by directly comparing local fMRI measures with other noninvasive techniques, such as EEG or MEG, to better characterize the temporal and spatial aspects of psychedelic brain dynamics.