2.1 Participants

fMRI Blood Oxygen Level Dependent (BOLD) and MEG data were collected from 20 participants, of which 15 (four females; mean age, 30.5 ± 8.0 years) were deemed suitable for analysis. One participant withdrew from the study due to excessive anxiety, while four others exhibited excessive head motion. For full inclusion criteria, please refer to Carhart-Harris, Kaelen, et al. (2016), Carhart-Harris, Muthukumaraswamy, et al. (2016). Ethical approval was granted by the National Research Ethics Service committee London–West London. The research complied with the revised declaration of Helsinki (2000), the International Committee on Harmonization Good Clinical Practice guidelines, and the National Health Service Research Governance Framework.

2.2 Study Design

Participants underwent two scanning sessions, one with placebo (PLCB) and one with LSD, separated by at least fourteen days. Participants were administered either PLCB or 75 μg of LSD intravenously via a 10 mL solution infused over two minutes. After a 60-min acclimatization period, subjects were led to an MRI scanner for Arterial Spin Labeling (ASL) and fMRI and then to magnetoencephalography (MEG) 165 min after drug administration. As stated above, the ASL data was not analyzed in this study. Both the fMRI and MEG recordings that were analyzed were eyes-closed resting-state scans. Eyes-open resting-state, eyes-closed music-listening, and video-watching recordings were also acquired but were not included in this analysis due to a primary focus on eyes-closed rest (Carhart-Harris, Kaelen, et al. 2016; Carhart-Harris, Muthukumaraswamy, et al. 2016).

2.3 Subjective Reports

Directly after each scan, participants were administered a Visual Analogue Scale (VAS), which asks a series of questions about their subjective experience. The six domains of questions were the Intensity of the Experience, Simple Imagery, Complex Imagery, Positive Mood, Ego dissolution, and Emotional Arousal. The average (± S.D.) changes from PLCB to LSD were 10.4 ± 4.8 for Emotional Arousal, 6.7 ± 6.0 for Positive Mood, and 5.7 ± 7.2 for Ego Dissolution. At the end of each scan day, participants were also given the 11-dimensional Altered States of Consciousness (ASC) questionnaire, which asked participants to retrospectively rate their subjective experience at the peak of the compound's effects.

2.4 fMRI

2.5 MEG

3.1 Delta Analysis (fMRI)

In the “Delta” analysis, we first measured the effects of LSD on seed-to-voxel resting-state functional connectivity (RSFC) between the combined, bilateral DLPFC seed (Figure 1) and the rest of the brain, irrespective of the correlation with subjective ratings. This analysis revealed that LSD increased connectivity between the seed and the two major hubs of the Default Mode Network (DMN), the medial prefrontal cortex and the posterior cingulate cortex (Figure 2). It also revealed a reduction in RSFC between this seed and cortical regions including the left angular gyrus, right supramarginal gyrus, left precuneus, and a region in the left DLPFC. As the DLPFC is part of the frontoparietal network (FPN), this result substantiates the hypothesis that psychedelics increase between-network connectivity between the FPN and the DMN (Carhart-Harris, Muthukumaraswamy, et al. 2016; Girn et al. 2023; Müller et al. 2018; Roseman et al. 2014).

We also measured the RSFC of the left and right DLPFC on their own. LSD significantly reduced the RSFC between the lDLPFC and the rest of the brain, especially in the sensorimotor cortex (Figure S1). (Note that global signal regression was not performed on the data.) LSD did not have a significant effect on the RSFC of the rDLPFC.

3.2 Covariate Analysis (fMRI)

We first examined correlations between the connectivity of the combined DLPFC (both left and right hemisphere) and subjective ratings of ego dissolution, emotional arousal, and positive mood. We found significant correlations between ego dissolution and two clusters with peaks at the left thalamus and the right fusiform gyrus (rFG) (Figure 3a). Given the novel nature of our original seeds, we sought to test the robustness of this correlation. Therefore, we conducted a confirmational analysis in which the thalamus and rFG were the seed regions. If the analysis returned a cluster within the DLPFC, it would verify the specificity of the DLPFC in ego dissolution. Indeed, this confirmational analysis revealed two main clusters, which contained not only the rDLPFC but also the right inferior frontal gyrus (rIFG) (Figure 3b). Scatter plots of the correlation between ego dissolution and RSFC are illustrated in Figure S3; the correlation is higher in the confirmational analysis, demonstrating the specificity of the rDLPFC cluster in mediating ego dissolution.

Emotional arousal significantly correlated with connectivity between the rDLPFC and the IPS (Figure 4a). The confirmational analysis reproduced a portion of the rDLPFC while also revealing new regions, namely the right and left anterior insula (rAI, lAI), dorsal anterior cingulate cortex (dACC), and middle temporal gyrus (MTG), that correlated with emotional arousal (Figure 4b). The rAI, lAI, and dACC are main nodes of the Salience Network. As shown in the scatter plots of Figure S4, the correlation between emotional arousal and RSFC is once again larger in the confirmational analysis, indicating that the rDLPFC cluster is more specific to regulating emotional arousal than the original rDLPFC seed.

We did not find any significant correlations between positive mood and the connectivity of the left, right, or combined DLPFC.

We also performed exploratory correlations between the three seeds and the 11-dimensional ASC taken at the end of each scanning day. There were two significant correlations before FDR. Reduced RSFC between the lDLPFC seed and the left Lingual Gyrus was correlated with anxiety (Figure S2a). Secondly, increased RSFC between the rDLPFC and the left Putamen correlated with Changed Meaning of Percepts (Figure S2b).

3.3 GC Analysis (MEG)

Within each network, we measured Granger Causality (GC), an estimate of directed functional connectivity, between each pair of constituent regions, conditioned on all other regions in the network. We computed both time-domain (broadband) and frequency-domain GC in each network.

Within each condition (PLCB and LSD), all pairwise-conditional time-domain GC values were significant for all four of the networks (p < 0.002 in all cases). All frequency-domain GC values were significant in all four of the frequency bands—theta, alpha, beta, and low gamma—in all of the networks. Between PLCB and LSD, there were no significant differences in time-domain GC for any pair of regions in any of the networks. However, frequency-domain GC did exhibit significant differences between conditions (Figure 5). In particular, LSD significantly increased theta-band GC from the thalamus to the rDLPFC (p = 0.0366) and beta-band GC from the IFG to the rDLPFC in the ego reverse network (p = 0.0366). There were no significant between-condition differences in frequency-domain GC for the ego forward, emotional forward, or emotional reverse network. Our results suggest that LSD specifically alters GC in a portion of the rDLPFC, i.e., the one in the ego reverse network, rather than the entire rDLPFC seed that was used in the ego forward network.

4.1 Lateralization of Function in the DLPFC

Our study was motivated by the evidence that TMS therapy and psychedelic-assisted therapy report similar therapeutic outcomes via two different treatment modalities. If both methodologies can reduce subjective ratings of depression, they may do so through the engagement of similar mechanisms. TMS for depression has evolved from a multitude of targets to a specific coordinate in the lDLPFC (Cash et al. 2021). We utilized a generalized coordinate and its mirror in the right hemisphere as the seeds for our RSFC analyses (Weigand et al. 2018). This data could then be correlated with subjective reports of the acute LSD experience, namely VAS ratings of ego dissolution, emotional arousal, and positive mood. Through our forward and reverse analyses, we were able to confirm the importance of the rDLPFC in the subjective experiences of emotional arousal and ego dissolution, two intertwined but separable experiences.

These results contribute to a relatively new understanding of the lateralization of function between the left and right DLPFC. As stated in the Introduction, this lateralization is exhibited in the different therapeutic effects associated with stimulating the left versus the right DLPFC with TMS. In particular, while stimulating the left DLPFC can reduce the severity of depression symptoms, stimulating the right DLPFC can inhibit mania. Our findings regarding the neural correlates of ego dissolution and emotional arousal are consistent with the lateralization of function in the DLPFC as revealed by TMS. We also found that the rDLPFC is particularly implicated in bottom-up shifts in network balance.

We propose that the function of primarily the right, and not the left, DLPFC shapes the experience of emotional arousal under psychedelics, which explains our finding that emotional arousal significantly correlates with the RSFC of only the rDLPFC. We contend that emotional arousal on LSD corresponds to a state phenomenologically similar to mania, particularly with regard to the variety of emotional valence. Emotional arousal captures both positive and negative changes in emotion on psychedelics; it is an unvalenced measure of subjective experience. This generalizable phenomenon is best understood as an umbrella term that can encompass multiple dimensions, which can be associated with positive or negative feelings. Secondly, one could construe increases in emotional arousal under LSD as consistent with an increase in perceived ‘salience’ (Friesen 2022; Wießner et al. 2023), thus invoking the “aberrant salience” model of mania (Sass and Pienkos 2013) and psychosis more generally (Kapur 2003).

On the other hand, we propose that ego dissolution recruits the functions of both the left and right DLPFC, hence ego dissolution is correlated with the RSFC of both hemispheres. Indeed, ego dissolution can be highly euphoric yet also resemble a state resembling manic, first-episode psychosis, where ‘self’ abnormalities are known to be common (Moe 2016; Sass and Pienkos 2013). Ego dissolution predicts increases in positive emotional reactivity to psychedelic use (Orłowski et al. 2022) and can also predict improved well-being. Note also that a “mystical” loss of boundaries between the self and the other can also characterize manic psychosis. People experiencing this state have described a “breach in the barriers of individuality” leading to a sense of spiritual communion with the universe (Custance 1952; Landis 1964; Parnas et al. 2005; Sass and Pienkos 2013). In some cases, this state can trigger beliefs of, for instance, telepathy, or other such so-called “narcissistic delusions” such as “delusions of reference”. The phenomenology of this state is very similar to that of ego dissolution on psychedelics, which can also sometimes give rise to claimed telepathic-like experiences (Johnstad 2020).

4.2 The DLPFC and Emotional Arousal

Our fMRI RSFC forward analysis found significant correlations between emotional arousal and connectivity between the rDLPFC and left IPS. The IPS is thought to be involved in regulating emotional arousal during threats. For instance, an fMRI and MEG study found global brain connectivity increases in the left IPS when analyzing threat detection (Balderston et al. 2017). Low frequency rTMS over either left or right IPS during threat of shock has been shown to reduce arousal (Balderston et al. 2020). These findings have led to a theory of connectivity in which the IPS mediates hyperarousal in anxiety disorders through DLPFC regulation (Brown et al. 2023).

The reverse analysis, in which the IPS cluster was used as the seed, returned a cluster in the rDLPFC. While this cluster was smaller than the original rDLPFC seed, it nevertheless confirmed that the rDLPFC is specifically involved in mediating emotional arousal on LSD. The outputs of the reverse analysis also contain clusters in the bilateral AI, dACC, and MTG. The lAI, rAI, and dACC are well-known hubs of the Salience Network (SN), which processes the emotional significance or salience of information. Previous work has shown that the SN acts as an intermediary between the central executive network (CEN) and the DMN. Some recent work also indicated a potential over-involvement of the SN in depression (Lynch et al. 2024). Hubs in these three networks show differences in information flow and dominance of signaling contingent on the task (Menon and Uddin 2010; Molnar-Szakacs and Uddin 2022). Aberrations in network transitions are theorized to be at the root of psychological illness (Menon 2011). Furthermore, as explained in Section 4.1, experiences of aberrant salience are a key aspect of emotional arousal on LSD; emotional sensitivity to the environment is heightened because everything appears to be extremely significant. Future work on this network may take into account the DMN for GC analysis or examine individual dyads within the network for other subjective categories like anxiety.

Our MEG analysis of these forward and reverse networks displayed statistically significant connections between all pairs of regions, across all frequency bands combined, under both placebo and LSD, which would suggest that regions form a genuine network in the brain. However, there were no statistically significant changes from placebo to LSD in emotional arousal in the frequency or time domain. Other spectral patterns, such as cross-freuqency coordination, may explain changes in emotional arousal on LSD (Clarke-Williams et al. 2024; Vinck et al. 2023).

4.3 The DLPFC and Ego Dissolution

The experience of ego dissolution, or “the loss of a sense of oneself,” is an essential aspect of the classical psychedelic experience (Millière 2017). Furthermore, there is a relationship between ego dissolution and statistically significant reductions in depression scores within adults, even in the long term (Carhart-Harris, Kaelen, et al. 2016; Roseman et al. 2018; Weiss et al. 2024). Hence, reliable biomarkers and mechanisms of the psychedelic experience of ego dissolution are worth further investigation. By identifying the network of regions that mediate the experience, we may be able to stimulate them with TMS, either independently of or in conjunction with a psychedelic treatment, to augment ego dissolution and thereby improve or better predict the therapeutic outcomes (Copa et al. 2024).

The psychedelic experience of ego dissolution has been correlated with brain activity in a wealth of previous research. A previous RSFC study, which was conducted on the same LSD data examined in this paper, revealed that global connectivity changes in the left and right angular gyrus and left and right insula correlated with ego dissolution (Tagliazucchi et al. 2016). Ego dissolution has mainly been attributed to connectivity changes within the DMN (Carhart-Harris, Muthukumaraswamy, et al. 2016; Lebedev et al. 2015) but also reduced interhemispheric connectivity within the salience network (Lebedev et al. 2015). This ‘disintegration’ of the high-level networks has been conceived as a ‘relaxation’ of top-down inhibitory control, which, according to the so-called ‘RElaxed Beliefs Under pSychedelics’ (REBUS) model, demonstrates a ‘flattening’ of the functional hierarchy in the brain and corresponding ‘relaxed beliefs’ or assumptions about the self and environment. In a variety of mental illnesses, including depression, beliefs or assumptions can become (pathologically) over-weighted; thus, their relaxation under (and potentially after) psychedelic experience may explain their potential therapeutic utility (Moe 2016; Sass and Pienkos 2013). Previous work has made significant strides toward establishing changes in functional cortical hierarchy under psychedelics (Girn et al. 2022; Luppi et al. 2021; Shinozuka et al. 2025).

This study correlated ego dissolution with a network of brain regions, rather than a single region or the whole brain. We began by selecting a novel seed that comprised the combined left and right DLPFC. We identified a significant correlation between ego dissolution and changes in RSFC between the combined DLPFC and the thalamus & FFA. A confirmational, or reverse analysis, in which the FFA & thalamus became the input seeds, yielded two main outputs: a smaller rDLPFC cluster and the rIFG.

These analyses, which were performed in fMRI, capture undirected connectivity. We then measured directed connectivity in an MEG dataset that was acquired on the same set of participants. In particular, we evaluated GC on the source-reconstructed timeseries of the regions in the forward and reverse networks. While many methods exist for capturing directed connectivity, GC has strong advantages in identifying time-delayed interactions, is robust to noisy data, and is flexible in working with the frequency domain. While we did not directly correlate GC with the ego dissolution VAS scales, this analysis nevertheless revealed the effect of LSD on the direction of information flow between the identified clusters across different frequency bands: theta, alpha, beta, and low gamma. While the effects of LSD on directed FC across the whole brain have been previously documented (Barnett et al. 2020), our analysis reveals changes in information flow specific to areas correlated with ego dissolution. Our findings are inconsistent with previous literature, which found that LSD exclusively decreased GC (Barnett et al. 2020). However, our estimates of GC were conditional on all other regions in the network, whereas previous analyses were either unconditional or conditional on timeseries that were averaged across multiple regions.

The reverse analysis revealed two statistically significant increases in GC: from the thalamus to the rDLPFC in the theta band and from the IFG to the rDLPFC in the beta band. The centroid of the thalamus seed is specifically located within the medial dorsal (MD) thalamus, which is structurally and functionally connected to the DLPFC. In fact, the primary source of input to the parvocellular MD is the DLPFC (Byne et al. 2009; Mitchell and Chakraborty 2013; Pergola et al. 2015). Increased bottom-up information flow from the thalamus to higher-order regions of the brain, such as the prefrontal cortex, has been proposed as a critical component of the neural mechanisms underlying the psychedelic experience (Avram et al. 2021; Onofrj et al. 2023; Vollenweider and Geyer 2001). The thalamus plays a role in gating sensory information to the cortex, screening out irrelevant stimuli (McCormick and Bal 1994). Aberrant connectivity from the thalamus to the cortex can lead to a “sensory overload,” inundating the cortex with excessive interoceptive and exteroceptive information (Vollenweider and Geyer 2001). Altered connectivity between the thalamus and the DLPFC is a robust biomarker of schizophrenia and psychosis, conditions in which patients assign too much meaning to irrelevant stimuli (Anticevic et al. 2014; Pergola et al. 2018; Steullet 2020; Welsh et al. 2010; Woodward et al. 2012). Similarly, some studies have shown that psychedelics exclusively increase functional connectivity between the thalamus and many regions of cortex (Bedford et al. 2023; Müller et al. 2017; Tagliazucchi et al. 2016), though other studies have found some decreases in thalamocortical connectivity (Avram et al. 2022; Gaddis et al. 2022; Preller et al. 2018, 2020). Psilocybin specifically alters activity in a region of the thalamus that overlaps most with the MD nucleus, which is consistent with our findings (Gaddis et al. 2022). A recent theory proposes that theta-band spiking in the thalamus shifts thalamocortical coupling to a dysrhythmic state that underlies altered states of consciousness, which is also aligned with our observations of elevated theta-band thalamocortical connectivity on LSD (Onofrj et al. 2023).

The DLPFC is generally involved in determining whether information is pertinent enough to enter and stay in working memory (Altamura et al. 2010; Friedman and Robbins 2022; Petrides 2000; Rassi et al. 2023; Rosero Pahi et al. 2020). Crucially, the ability of the PFC to maintain and assign relevance to neural representations is mediated by the MD thalamus (Bolkan et al. 2017; Marton et al. 2018; Mitchell and Chakraborty 2013; Parnaudeau et al. 2015; Rikhye et al. 2018; Schmitt et al. 2017) Rodent studies have demonstrated that MD neurons stabilize context-relevant representations in the PFC, thereby enabling them to be maintained within working memory, while suppressing context-irrelevant representations (Rikhye et al. 2018).

Thus, enhanced directed connectivity from the MD thalamus may cause information from typically deemed irrelevant contexts to interfere with representations in the PFC, leading to a “hyper-flexible” state in which the brain is less constrained by the demands of the present cognitive context. We speculate that, during self-reflection, heightened connectivity between the MD thalamus and the cortex may amplify neural representations that do not ordinarily pertain to the self. These amplifications could induce instability across this network including thalamic dependent internal sensory perception within the rIFG (Dobrushina et al. 2021) and distort functionality in any one of the other regions in the ego dissolution network, which have all been correlated with self-identification (Herwig et al. 2012; Kaplan et al. 2008; Ma and Han 2012). This broadening of relevance could explain the experience of the “loss of the sense of self” found in ego dissolution; on psychedelics the neural mechanisms of the self “over-represent” or assign too much relevance to external stimuli, yielding a sense of vast interconnectedness with the environment.

Furthermore, increased beta-band connectivity from the rIFG to the rDLPFC, as we observed, could be involved in integrating, or attempting to integrate, experiences of ego dissolution. In particular, the rIFG and the rDLPFC are both part of a network that integrates disconfirmatory evidence, i.e., present stimuli that violate a prior belief (Ehlis et al. 2024; Lavigne et al. 2015). In other words, when prior held beliefs about the self are challenged by conflicting evidence, the rIFG and the rDLPFC work together to integrate this information and thereby maintain or update the belief. This could be consistent not only with the rIFG's established role in integrating internal representations (Adolfi et al. 2017; Dobrushina et al. 2021), but also REBUS' hypothesis that psychedelics revise prior beliefs by altering top-down inhibition from higher-order regions like the rIFG and the rDLPFC.

The observed effects of LSD on directed connectivity are similar to the neural signatures of schizophrenia, which relate to a longstanding, but contentious, view that psychedelics are “psychotomimetics” or models of psychosis (Nichols and Walter 2021; Umbricht et al. 2003; Vollenweider et al. 1998). Schizophrenia and mania are associated with a number of structural deficits in the DLPFC, including cortical thinning, reduced gray matter volume, and lower fractional anisotropy of outgoing white matter tracts (Abé et al. 2023; Ellison-Wright and Bullmore 2009; van Haren et al. 2011). Furthermore, as stated above, aberrations in connectivity between the MD thalamus and the PFC are commonly observed in patients with schizophrenia, perhaps explaining observations that task-irrelevant representations interfere more with cognition in these populations (Wagner et al. 2013).

While some of the changes in brain activity and phenomenology on LSD are also found in individuals with schizophrenia, it is important to note that schizophrenia is a chronic condition, whereas the LSD trip is a transient experience. Also, disruptions to self-awareness in schizophrenia differ significantly from those of LSD. For instance, patients over-attribute agency to themselves and overestimate their causal power over the world, whereas people who take LSD tend not to do so (Hur et al. 2014). Furthermore, individuals affected by schizophrenia often exhibit long-standing emotional withdrawal and cognitive blunting, which are markedly different from the emotional arousal and cognitive flexibility experienced on LSD (Doss et al. 2021; Martínez et al. 2021; Oorschot et al. 2013).

In summary, we propose that amplified connectivity from the thalamus to the DLPFC may cause representations of the outside world to interfere with self-related cognition, thereby giving rise to the experience of ego dissolution. The same neural mechanisms may underpin delusional beliefs about the self in patients afflicted with schizophrenia and psychosis.

4.4 Limitations

This data has a small sample size; data from only 15 participants was included in the analysis. The convergent and significant results we found in both the fMRI and MEG analyses indicate that the observed effects are real. There is also a potential that this data is confounded by the physiological effects of LSD, namely its vasoconstrictive effects. However, while vasoconstriction alters the fMRI signal, it does not affect MEG activity in the same way (Özbay et al. 2019; Wilson et al. 2016). The significant increases in directed information flow in the MEG analysis suggest that vasoconstriction does not underlie the observed changes in RSFC.

The study design may not have given sufficient time for the effects of LSD to wash out; there was only a 14-day separation between the LSD and placebo sessions, and LSD can affect brain activity for up to a month afterwards (Barrett et al. 2020; McCulloch et al. 2022). However, the order of LSD and placebo was counterbalanced across participants. If the changes in RSFC depended on whether participants received LSD first or second, then we likely would not have obtained significant results. If a longer washout period had been given, we suspect that we would have seen even stronger changes in RSFC.

Because the data is resting-state, it is difficult to relate different patterns of brain activity to specific psychological behaviors or processes that are modulated via LSD. To provide stronger evidence, tasks must be deployed to examine changes in self-awareness in response to working memory demands, while acknowledging the risk of generalized deficit confounds in this regard.

We did not correlate the directed connectivity of the MEG data to ratings of ego dissolution because these ratings were not available for all of the included participants. However, because the undirected connectivity between the same regions did correlate with ego dissolution in the fMRI dataset, we are confident that the directed connectivity plays some role in shaping the experience of ego dissolution. That being said, whole-brain MEG analyses may have uncovered a broader set of brain regions that correlate with ego dissolution and emotional arousal.

Finally, accurate source reconstruction of MEG data to subcortical regions like the thalamus is generally challenging, especially in resting-state data (Krishnaswamy et al. 2017). fMRI analysis of subcortical activity may be more reliable, so dynamic causal modeling on the fMRI data, which can be used to estimate directed connectivity, may be worth exploring to assess the robustness of the MEG GC findings.