Neural correlates of the LSD experience revealed by multimodal ...

Neural correlates of the LSD experience revealed by multimodal neuroimaging Robin L. Carhart-Harrisa,1, Suresh Muthukumaraswamyb,c,d, Leor Rosemana,e,2, Mendel Kaelena,2, Wouter Droogb, Kevin Murphyb, Enzo Tagliazucchif,g, Eduardo E. Schenberga,h,i, Timothy Nestj, Csaba Orbana,e, Robert Leeche, Luke T. Williamsa, Tim M. Williamsk, Mark Bolstridgea, Ben Sessaa,l, John McGoniglea, Martin I. Serenom, David Nicholsn, Peter J. Hellyere, Peter Hobdenb, John Evansb, Krish D. Singhb, Richard G. Wiseb, H. Valerie Currano, Amanda Feildingp, and David J. Nutta a Centre for Neuropsychopharmacology, Department of Medicine, Imperial College London, W12 0NN, London, United Kingdom; bDepartment of Psychology, Cardiff University Brain Research Imaging Centre, CF10 3AT, Cardiff, United Kingdom; cSchool of Pharmacy, University of Auckland, 1142 Auckland, New Zealand; dSchool of Psychology, University of Auckland, 1142 Auckland, New Zealand; eComputational, Cognitive and Clinical Neuroscience Laboratory, Department of Medicine, Imperial College London, W12 0NN, London, United Kingdom; fInstitute of Medical Psychology, Christian Albrechts University, 24118 Kiel, Germany; gBrain Imaging Center and Neurology Department, Goethe University, 60528 Frankfurt am Main, Germany; hDepartment of Psychiatry, Universidade Federal de São Paulo, 04038-020, São Paulo, Brazil; iInstituto Plantando Consciencia, 05.587-080, São Paulo, Brazil; jDepartment of Psychiatry, McGill University, H3A 1A1, Montréal, Canada; kDepartment of Psychiatry, University of Bristol, BS8 2BN, Bristol, United Kingdom; l Department of Neuroscience, Cardiff University, CF24 4HQ, Cardiff, United Kingdom; mBirkbeck-UCL Centre for Neuroimaging, WC1H 0AP, London, United Kingdom; nEschelman School of Pharmacy, University of North Carolina, Chapel Hill, NC 27514; oClinical Psychopharmacology Unit, University College London, WC1E 6BT, London, United Kingdom; and pThe Beckley Foundation, Beckley Park, OX3 9SY, Oxford, United Kingdom

Lysergic acid diethylamide (LSD) is the prototypical psychedelic drug, but its effects on the human brain have never been studied before with modern neuroimaging. Here, three complementary neuroimaging techniques: arterial spin labeling (ASL), blood oxygen leveldependent (BOLD) measures, and magnetoencephalography (MEG), implemented during resting state conditions, revealed marked changes in brain activity after LSD that correlated strongly with its characteristic psychological effects. Increased visual cortex cerebral blood flow (CBF), decreased visual cortex alpha power, and a greatly expanded primary visual cortex (V1) functional connectivity profile correlated strongly with ratings of visual hallucinations, implying that intrinsic brain activity exerts greater influence on visual processing in the psychedelic state, thereby defining its hallucinatory quality. LSD’s marked effects on the visual cortex did not significantly correlate with the drug’s other characteristic effects on consciousness, however. Rather, decreased connectivity between the parahippocampus and retrosplenial cortex (RSC) correlated strongly with ratings of “ego-dissolution” and “altered meaning,” implying the importance of this particular circuit for the maintenance of “self” or “ego” and its processing of “meaning.” Strong relationships were also found between the different imaging metrics, enabling firmer inferences to be made about their functional significance. This uniquely comprehensive examination of the LSD state represents an important advance in scientific research with psychedelic drugs at a time of growing interest in their scientific and therapeutic value. The present results contribute important new insights into the characteristic hallucinatory and consciousness-altering properties of psychedelics that inform on how they can model certain pathological states and potentially treat others. LSD

be mediated by serotonin 2A receptor (5-HT2AR) agonism (7). Previous neurophysiological research with LSD is limited to electroencephalography (EEG) studies in the 1950s and 1960s. These reported reductions in oscillatory power, predominantly in the lower-frequency bands, and an increase in the frequency of alpha rhythms (8). Broadband decreases in cortical oscillatory power have been observed in modern EEG and magnetoencephalography (MEG) studies with psilocybin (9, 10), with EEG and the dimethyltryptamine-containing brew “ayahuasca” (11), and with rodent brain local-field potential recordings and a range of different 5-HT2AR agonists (12–14). The effects of psychedelics (other than LSD) on human brain activity have also previously been investigated with positron emission tomography (PET) (15) and functional magnetic resonance imaging (fMRI) (16). fMRI studies with psilocybin revealed decreased cerebral blood flow (CBF) and blood oxygen leveldependent (BOLD) signal in connector hubs (16), decreased Significance Lysergic acid diethylamide (LSD), the prototypical “psychedelic,” may be unique among psychoactive substances. In the decades that followed its discovery, the magnitude of its effect on science, the arts, and society was unprecedented. LSD produces profound, sometimes life-changing experiences in microgram doses, making it a particularly powerful scientific tool. Here we sought to examine its effects on brain activity, using cutting-edge and complementary neuroimaging techniques in the first modern neuroimaging study of LSD. Results revealed marked changes in brain blood flow, electrical activity, and network communication patterns that correlated strongly with the drug’s hallucinatory and other consciousness-altering properties. These results have implications for the neurobiology of consciousness and for potential applications of LSD in psychological research.

| serotonin | consciousness | brain | psychedelic

L

ysergic acid diethylamide (LSD) is a potent serotonergic hallucinogen or “psychedelic” that alters consciousness in a profound and characteristic way. First synthesized in 1938, its extraordinary psychological properties were not discovered until 1943 (1). LSD would go on to have a major effect on psychology and psychiatry in the 1950s and 1960s; however, increasing recreational use and its influence on youth culture provoked the drug’s being made illegal in the late 1960s. As a consequence, human research with LSD has been on pause for half a century. However, inspired by a revival of research with other psychedelics, such as psilocybin and ayahuasca, a small number of new reports on the psychological effects of LSD have recently been published (2–6). LSD has a high affinity for a range of different neurotransmitter receptors, but its characteristic psychological effects are thought to

www.pnas.org/cgi/doi/10.1073/pnas.1518377113

Author contributions: R.L.C.-H., S.M., K.M., R.L., J.E., K.D.S., R.G.W., A.F., and D.J.N. designed research; R.L.C.-H., S.M., M.K., W.D., L.T.W., T.M.W., M.B., B.S., and P.H. performed research; C.O., R.L., J.M., M.I.S., D.N., P.J.H., and H.V.C. contributed new reagents/analytic tools; R.L.C.-H., S.M., L.R., M.K., K.M., E.T., E.E.S., T.N., and R.L. analyzed data; and R.L.C.-H., S.M., L.R., and D.J.N. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Freely available online through the PNAS open access option. 1

To whom correspondence should be addressed. Email: [email protected].

2

L.R. and M.K. contributed equally to this work.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1518377113/-/DCSupplemental.

PNAS Early Edition | 1 of 6

NEUROSCIENCE

Edited by Marcus E. Raichle, Washington University in St. Louis, St. Louis, MO, and approved March 1, 2016 (received for review September 17, 2015)

resting state functional connectivity (RSFC) in major resting state networks (RSNs) such as the default-mode network (DMN) (17), and the emergence of novel patterns of communication (18, 19), whereas increased cortical glucose metabolism was found with PET (15). Notably, the spatial locations of the PET-, fMRI-, EEG-, and MEG-measured effects of psychedelics are relatively consistent; for example, high-level cortical regions, such as the posterior cingulate cortex (PCC), and some of the principal effects of psilocybin revealed by fMRI (e.g., decreased DMN RSFC) were recently replicated by a separate team working with ayahuasca (20). Consistent with a prior hypothesis (17), these studies suggest that an “entropic” effect on cortical activity is a key characteristic of the psychedelic state. However, a putative excitation of hippocampal/parahippocampal gyri activity has also been observed with fMRI and psychedelics in humans (19) and animals (14). Moreover, depth EEG studies in the 1950s reported activations in medial temporal lobe regions during psychosis-like states under LSD and other psychedelics (21, 22). Further, patients with epilepsy with resection of the medial temporal lobes showed attenuated LSD effects postsurgery (23), and electrical stimulation of medial temporal lobe circuitry produces visual hallucinations of somewhat similar nature to those produced by psychedelics [e.g., distorted visual perception (24) and dreamlike “visions” (25)]. The present study sought to investigate the acute brain effects of LSD in healthy volunteers, using a comprehensive placebocontrolled neuroimaging design incorporating ASL, BOLD signal measures, and MEG resting state scans. It was predicted that major RSNs (e.g., the DMN) and hippocampal/parahippocampal gyri circuitry would be implicated in the drug’s mechanism of action. Twenty healthy participants attended two scanning days (LSD and placebo) at least 2 wk apart in a balanced-order, within-subjects design. Sessions included an fMRI followed by a MEG scan, each lasting 75 min. Data were acquired during eye-closed, task-free, “resting state” conditions. Drug/placebo were administered in solution and injected i.v. over the course of 2 min. Two resting state ASL scans totaling 16 min were completed 100 min after i.v. administration of LSD (75 μg in 10 mL saline) or placebo (10 mL saline), corresponding to the initial phase of the peak subjective effects of LSD (peak effects were reached ∼120–150 min postinfusion). Two resting state BOLD scans totaling 14 min were completed 135 min postinfusion, and two resting state MEG scans totaling 14 min were completed 225 min postinfusion. All analyses applied multiple comparison correction (SI Appendix) and two-tailed hypothesis testing unless particularly strong prior hypotheses were held. Results The intensity of LSD’s subjective effects was relatively stable for the ASL and BOLD scans but attenuated somewhat for the MEG (SI Appendix, Table S1). Participants carried out VAS-style ratings via button-press and a digital display screen presented after each scan (SI Appendix), and the 11-factor altered states of consciousness (ASC) questionnaire (26) was completed at the end of each dosing day (SI Appendix, Fig. S1). All participants reported eyesclosed visual hallucinations and other marked changes in consciousness under LSD. Data from 15 volunteers were suitable for the ASL and BOLD analyses (four females; mean age, 30.5 ± 8.0 y; SI Appendix). Differences in CBF in the two conditions were calculated using a whole-brain analysis (cluster-correction, P < 0.05). Greater CBF under LSD was observed in the visual cortex (Fig. 1), and the magnitude of these increases correlated positively with ratings of complex imagery on the ASC (r = 0.64; P = 0.01; Bonferroni corrected P = 0.04; SI Appendix, Fig. S9A). The unthresholded difference in CBF can be viewed in Neurovault (27) (neurovault. org/collections/FBVSAVDQ/). Seed-based RSFC analyses were also performed. A bilateral parahippocampal (PH) seed was chosen based on previous findings with psilocybin (19) and a primary visual cortex (V1) seed was chosen based on the characteristic visual perceptual effects of psychedelics. V1 was identified using a retinotopic-localizer paradigm (SI Appendix). Finally, the ventromedial PFC (vmPFC) was 2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1518377113

Fig. 1. Whole-brain cerebral blood flow maps for the placebo and LSD conditions, plus the difference map (cluster-corrected, P < 0.05; n = 15).

chosen because of our previous focus on this region in studies with psilocybin and MDMA (16, 28). Analyses revealed increased RSFC between V1 and a large number of cortical and subcortical brain regions (Fig. 2 and SI Appendix, Table S3), decreased RSFC between the PH and the retrosplenial cortex (RSC) and PCC, and increased RSFC between the PH and dorsal mPFC and right dorsolateral PFC (Fig. 3 and SI Appendix, Table S4). Increased RSFC between the vmPFC and the bilateral caudate and inferior frontal gyrus was also observed, as was decreased vmPFC-PCC RSFC (SI Appendix, Fig. S4 and SI Appendix, Table S5). All the relevant unthresholded maps can be viewed in Neurovault (27) (neurovault. org/collections/FBVSAVDQ/). Increased V1 RSFC (to the most significant regions shown in Fig. 2: P < 0.01; 5,000 permutations; SI Appendix, Fig. S8) correlated with VAS ratings of simple hallucinations (r = 0.62; P = 0.012; Bonferroni corrected P = 0.048; SI Appendix, Fig. S9B), as well as ASC ratings of elementary (r = 0.63; P = 0.012; Bonferroni corrected P = 0.048; SI Appendix, Fig. S9C) and complex (r = 0.74; P = 0.0016; Bonferroni corrected P = 0.006; SI Appendix, Fig. S9D) imagery. Decreased PH RSFC (to the significant regions shown in Fig. 3) correlated with VAS ratings of ego-dissolution (r = 0.73; P = 0.0018; SI Appendix, Fig. S9E) and “altered meaning” on the ASC (r = 0.82; P = 0.0002; Bonferroni corrected P = 0.002; SI Appendix, Fig. S9F). Importantly, some of these (hypothesized) correlations were phenomenology selective (SI Appendix, Table S7): increased visual cortex CBF and V1 RSFC correlated more strongly with the visual hallucinatory aspect of the drug experience than the altered meaning/ego-dissolution aspect, whereas the opposite was true for decreased PH RSFC. Changes in vmPFC RSFC did not correlate with any of the ratings. Carhart-Harris et al.

Fig. 2. Significant between-condition differences (orange = increases) in RSFC between the V1 seed region (purple) and the rest of the brain. Unthresholded maps can be viewed here: neurovault.org/collections/FBVSAVDQ/ (n = 15).

Next, the effect of LSD on brain network properties was investigated. Twelve functionally familiar RSNs were identified in a set of 20 spatially independent components derived from independent data (human connectome project; SI Appendix). These RSNs are as follows: a medial visual network, a lateral visual network (VisL), an occipital pole network (VisO), an auditory network (AUD), a sensorimotor network, the DMN, a parietal cortex network (PAR), the dorsal attention network, the salience network, a posterior opercular network (POP), the left frontoparietal network, and the right frontoparietal network (rFP). Four metrics were calculated for each RSN: within-RSN CBF, within-RSN RSFC or “integrity,” within-RSN BOLD signal variance, and between-RSN RSFC or “segregation.” Between-condition differences in the first three metrics are shown in Fig. 4A, and the between-RSN RSFC results are shown in Fig. 4B. Differences (increases) in CBF were restricted to the visual RSNs, whereas differences in variance and integrity (decreases) were much more pronounced and universal. According to previous research with psilocybin (17), it was predicted that decreased DMN integrity (or DMN “disintegration”) would correlate with ratings of ego-dissolution, and this hypothesis was supported (r = 0.49; P = 0.03; SI Appendix, Fig. S9G). Given the large number of possible permutations, additional correlational analyses were not performed; however, to test the selectivity of the relationship between DMN disintegration and ego-dissolution, correlations were calculated for ego-dissolution and the integrity of the other 11 RSNs, and none were significant (SI Appendix, Table S2). Disintegration of the visual RSNs did not correlate with ratings of visual hallucinations. See SI Appendix, Fig. S5, for brain images of the RSN integrity results. Between-RSN RSFC or RSN segregation was also markedly modulated by LSD. Decreased segregation (red squares with white Carhart-Harris et al.

Fig. 3. Significant between-condition differences in RSFC between the PH seed and the rest of the brain (orange = increases; blue = decreases). Unthresholded maps can be viewed here: neurovault.org/collections/FBVSAVDQ/ (n = 15).

PNAS Early Edition | 3 of 6

NEUROSCIENCE

asterisks in Fig. 4B, right matrix) was observed between eight RSN pairs (VisL–PAR, VisL–dorsal attention network, VisO–POP, AUD–PAR, AUD–rFP, DMN–salience network, PAR–POP, POP–rFP), with only one pair (VisO–rFP) showing increased segregation (blue square with white asterisk in Fig. 4B, right matrix). Contrary to a prior hypothesis, decreased RSN segregation (in the eight networks that showed this effect) did not correlate with ratings of ego-dissolution (r = 0.12; P > 0.05). Data from 14 volunteers were suitable for the MEG analyses (three females; mean age, 32.1 ± 8.3 y). Primary analyses focused on between-condition differences in frequency-specific oscillatory power, measured during eyes-closed rest. The relevant data (14 min of rest) were acquired ∼50 min after completion of the MRI protocol and filtered into the following frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–15 Hz), beta (15–30 Hz), low gamma (31– 49 Hz), and high gamma (51–99 Hz). Results revealed decreased oscillatory power under LSD in four frequency bands (Fig. 5A), with some suspected residual muscle artifact confounding the gamma results. For the lower-frequency bands (i.e., 1–30 Hz), the decreases reached significance in most of the sensors. To explore relationships between these outcomes and subjective measures, VAS ratings of ego-dissolution and visual hallucinations (simple and complex) were entered into regression analyses, using cluster permutation testing. Significant relationships were found between ego-dissolution and decreased delta (mean cluster, r = −0.54; P < 0.05) and alpha power (mean cluster, r = −0.29; P < 0.05) and between simple hallucinations and decreased alpha power (mean cluster, r = −0.61; P < 0.05) (Fig. 4B). Plotting the power spectrum independently for each condition for the significant alpha cluster (Fig. 5C), it is evident that the distribution of power is decreased across a broad frequency

in CBF, however [r = 0.1 (P > 0.05) and r = 0.33 (P > 0.05) for integrity and variance, respectively], nor head motion (SI Appendix), but they did correlate with the mean decrease in power (significant sensors) for the four displayed frequency bands [r = 0.79 (P = 0.001; SI Appendix, Fig. S6) and r = 0.76 (P = 0.002) for integrity and variance, respectively]. Mean decreases in RSN segregation (for the eight pairs that showed this effect) correlated with mean decreases in RSN integrity (mean of all 12 RSNs, r = 0.53; P = 0.02; SI Appendix, Fig. S6) and reduced oscillatory power (delta-beta combined, r = 0.67; P = 0.017; SI Appendix, Fig. S6), but not decreased RSN variance (r = 0.33, P > 0.05) nor increased CBF (r = 0.18; P > 0.05). Given the number of possible permutations, we chose not to explore beyond these relationships.

Fig. 4. (A) Mean percentage differences (+SEM) in CBF (red), integrity (blue), and signal variance (green) in 12 different RSNs under LSD relative to placebo (red asterisks indicate statistical significance, *P < 0.05; **P < 0.01, Bonferroni corrected). (B) Differences in between-RSN RSFC or RSN “segregation” under LSD vs placebo. Each square in the matrix represents the strength of functional connectivity (positive = red, negative = blue) between a pair of different RSNs (parameter estimate values). The matrix on the far right displays the betweencondition differences in covariance (t values): red = reduced segregation and blue = increased segregation under LSD. White asterisks represent significant differences (P < 0.05, FDR corrected; n = 15).

range under LSD, and the peak alpha rhythm is reduced in amplitude and of higher frequency (i.e., 10 Hz under placebo, 12 Hz under LSD; t = 4.21; P = 0.0009). Source modeling revealed that sources of the power decreases were relatively distributed throughout the brain (SI Appendix, Table S8), with significant effects in the PCC/precuneus (theta, alpha, and beta) and other high-level cortical regions (delta-beta). This study’s multimodal design enabled correlational analyses to be performed between the various (significant) imaging outcomes. This was done in a hypothesis-driven manner, and because the outcomes’ directions were already known, one-tailed tests were performed. Relationships were observed between the increases in CBF (localized to the visual cortex) and decreases in alpha power in posterior (occipital cortex) sensors (r = −0.59; P = 0.029; SI Appendix, Fig. S6) and between increases in V1 RSFC (to the most significant regions: P < 0.01; 5,000 permutations; SI Appendix, Fig. S8) and decreased posterior-sensor alpha power (r = −0.81; P = 0.0015; SI Appendix, Fig. S6), but there was only a trend-level relationship between increases in visual cortex CBF and increases in V1 RSFC (r = 0.43; P = 0.055). The mean change (decreases) in the integrity of the 12 RSNs correlated very strongly with the mean change (decreases) in their variance (r = 0.89; P = 4 × 10−6; SI Appendix, Fig. S6). Neither metric correlated with the mean change 4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1518377113

Discussion The present findings offer a comprehensive new perspective on the changes in brain activity characterizing the LSD state, enabling us to make confident new inferences about its functional neuroanatomy. Principal findings include increased visual cortex CBF, RSFC, and decreased alpha power, predicting the magnitude of visual hallucinations; and decreased DMN integrity, PH-RSC RSFC, and delta and alpha power (e.g., in the PCC), correlating with profound changes in consciousness, typified by ego-dissolution. More broadly, the results reinforce the view that resting state ASL, BOLD FC, and MEG measures can be used to inform on the neural correlates of the psychedelic state (9, 16). Importantly, strong relationships were found between the different imaging measures, particularly between changes in BOLD RSFC (e.g., network “disintegration” and “desegregation”) and decreases in oscillatory power, enabling us to make firmer inferences about their functional meaning. The present study sheds new light on the relationship between changes in spontaneous brain activity and psychedelic-induced visual hallucinations. Strong relationships were observed between increased V1 RSFC and decreased alpha power, as well as ratings of both simple and complex visual hallucinations. The latter result is consistent with previous findings with psilocybin (29). Importantly, a very strong relationship was also observed between increased V1 RSFC and decreased alpha power in occipital sensors, suggesting that as well as being commonly related to visual hallucinations, these physiological effects are closely interrelated. The increase in V1 RSFC under LSD is a particularly novel and striking finding and suggests that a far greater proportion of the brain contributes to visual processing in the LSD state than under normal conditions. This expansion of V1 RSFC may explain how normally discreet psychological functions (e.g., emotion, cognition, and indeed the other primary senses) can more readily “color” visual experience in the psychedelic state. Biologically informed modeling has suggested that instability within the primary visual cortex may facilitate the emergence of geometric hallucinations via self-organized patterns of neural excitation (30), and eyes-closed fMRI recordings during ayahuasca hallucinations suggest the visual cortex behaves “as if” there is external input when there is none (31) (see also ref. 29). The present findings of increased visual cortex CBF, expanded V1 RSFC, and decreased alpha power may be seen as consistent with the notion of “seeing with eyes-shut” under psychedelics, because they are all properties normally associated with visual stimulation (32, 33). Cortical alpha has been hypothesized to serve a general inhibitory function, filtering out “stimulus-irrelevant” information (34). Thus, reduced alpha power (9, 29, 35) could have disinhibitory consequences, facilitating the release of anarchic patterns of excitation that manifest spontaneously and experientially as visual hallucinations. This hypothesis is leant (indirect) support by two prior studies that found reduced spontaneous visual cortex alpha power under psilocybin alongside reduced evoked visual responses (9, 29). Further work, using higher-resolution brain imaging, machine learning techniques, dynamic measures of functional and effective connectivity, and improved “capture” of visual hallucinations (e.g., via button press or experience sampling), may help to develop this appealing model (e.g., see ref. 36). Carhart-Harris et al.

Fig. 5. MEG results. (A) Statistical analysis of planar gradiometer-configured MEG data comparing LSD with placebo in the eyes-closed condition. Blue indicates less power under LSD. Units are t-statistics. Significant sensor clusters are marked such that stars correspond to P < 0.01 and crosses to P < 0.05 (corrected). Source localization results are also displayed. (B) Significant correlations between changes (decreases) in oscillatory power and subjective phenomena. (C) Power spectra for the significant sensor cluster in B (simple hallucinations), with placebo data plotted in blue and LSD in red (n = 14).

The present data also inform on another fundamental question; namely, how do psychedelics alter brain function to (so profoundly) alter consciousness? Interestingly, although the effects of LSD on the visual system were pronounced, they did not significantly correlate with its more fundamental effects on consciousness. Instead, a specific relationship was found between DMN disintegration and ego-dissolution, supporting prior findings with psilocybin (17). Also consistent with previous psilocybin research (9), a significant relationship was found between decreased PCC alpha power and ego-dissolution. Moreover, an especially strong relationship was found between PH-RSC decoupling and ego-dissolution (see also ref. 10). Thus, in the same way the neurobiology of psychedelicinduced visual hallucinations can inform on the neurobiology of visual processing, so the neurobiology of psychedelic-induced egodissolution can inform on the neurobiology of the “self” or “ego” (37), and the present results extend our understanding in this regard, implying that the preservation of DMN integrity, PH-RSC communication, and regular oscillatory rhythms within the PCC may be important for the maintenance of one’s sense of self or ego. Linking these results to pathology, an especially strong relationship was found between PH-RSC decoupling and the “altered meaning” factor on the ASC. Interestingly, altered activity Carhart-Harris et al.

PNAS Early Edition | 5 of 6

NEUROSCIENCE

within the PH-RSC circuit under psilocybin has previously been found to correlate with the spiritual experience and insightfulness dimensions of the 11-factor ASC (10), and altered RSC/PCC activity has been found to correlate with ego-dissolution (9), suggesting modulation of this particular circuit may be an important feature of especially profound psychedelic experiences. The altered meaning factor of the ASC is composed of items such as “some unimportant things acquired a special meaning” and “things in my surroundings had a new or alien meaning” that are phenomenologically resonant with the notion of “aberrant salience” in schizophrenia research (38). Impaired reality testing as a corollary of impaired ego functioning may explain an association between ego-dissolution and altered meaning. Similarities between aspects of psychosis and the psychedelic state have long been debated, and one of the most influential hypotheses on the neurobiology of schizophrenia proposes a functional disconnect between certain brain structures in the disorder (39). In this context, it is intriguing to consider whether the PH-RSC circuit is involved in certain psychosis-related experiences (e.g., refs. 40 and 41). More specifically, it would be interesting to examine the integrity of the PHRSC connection in cases of endogenous psychoses in which phenomena such as altered meaning, ego-dissolution, and/or impaired reality-testing are observed. To our knowledge, these specific phenomena have never been formally investigated in imaging studies involving patients exhibiting endogenous psychoses, but studies on early psychosis and the at-risk mental state may be informative in this regard (e.g., ref. 40). When the present results are considered in relation to previous human neuroimaging studies with psychedelics, some general principles emerge. It seems increasingly evident that psychedelics reduce the stability and integrity of well-established brain networks (e.g., ref. 16) and simultaneously reduce the degree of separateness or segregation between them (e.g., ref. 42); that is, they induce network disintegration and desegregation. Importantly, these effects are consistent with the more general principle that cortical brain activity becomes more “entropic” under psychedelics (17). Furthermore, with the benefit of the present study’s multimodal imaging design, we can extend on these generic insights to postulate some more specific physiological properties of the psychedelic state and how these relate to some of its key psychological properties; namely, expanded V1 RSFC relates to the magnitude of visual hallucinations and decoupling of the PH-RSC circuit relates to the level of ego-dissolution, and perhaps also the profundity of a psychedelic experience more generally (also see refs. 9 and 10 in this regard). Before concluding, we should highlight some general limitations of the present study and address a discrepant finding in the field. Regarding limitations, a fully randomized, double-blind design is often considered the gold standard; however, experimental blinding is known to be ineffective in studies with conspicuous interventions. Thus, a single-blind, balanced-order design with an inert placebo (offering the simplest and “cleanest” possible control condition) was considered an effective compromise. Also, although the multimodal design of this study was an advantage, the experimental protocol was demanding for participants, and the different scan types (ASL, BOLD, and MEG) occurred separately in time. Simultaneous EEG-fMRI may therefore offer some advantageous in this regard. Another general limitation of imaging studies involving potent psychoactive drugs, is the issue of between-condition differences in head motion and related artifacts. In this study, we opted to use the most rigorous motion-correction strategies available (SI Appendix), despite motion levels being no higher than those seen in previous studies by our group (16). Regarding the discrepant finding, a previous psilocybin ASL study of ours revealed decreased CBF postpsilocybin (i.v.) during eyes-open rest (16), whereas the present i.v. LSD study found increased CBF localized to the visual cortex with eyes-closed rest. One must be cautious of proxy measures of neural activity (that lack temporal resolution), such as CBF or glucose metabolism, lest the relationship between these measures, and the underlying neural activity they are assumed to index, be confounded by extraneous factors, such as a direct vascular action of the drug

(43). For this reason, more direct measures of neural activity (e.g., EEG and MEG) and/or more dynamic fMRI measures (e.g., RSFC) should be considered more reliable indices of the functional brain effects of psychedelics, and it is notable in this regard that our previous MEG (9) and RSFC (16, 19, 42) findings with psilocybin are highly consistent with those observed here with LSD. Thus, rather than speculate on the above-mentioned discrepancy, it may be more progressive to highlight the advantages of EEG/MEG and dynamic fMRI and conclude that further work would be required to resolve discrepancies in the literature regarding the effects of psychedelics on metabolically related metrics that lack temporal resolution. Finally, as evidence supporting the therapeutic potential of psychedelics mounts (6, 44–46), so does our need to better understand how these drugs work on the brain. In many psychiatric disorders, the brain may be viewed as having become entrenched in pathology, such that core behaviors become automated and rigid. Consistent with their “entropic” effect on cortical activity (17), psychedelics may work to break down such disorders by dismantling the patterns of activity on which they rest. Future work is required to test this hypothesis and the others that have been 1. Hofmann A (1980) LSD: My Problem Child (McGraw-Hill, New York). 2. Carhart-Harris RL, et al. (2015) LSD enhances suggestibility in healthy volunteers. Psychopharmacology (Berl) 232(4):785–794. 3. Dolder PC, Schmid Y, Haschke M, Rentsch KM, Liechti ME (2015) Pharmacokinetics and Concentration-Effect Relationship of Oral LSD in Humans. Int J Neuropsychopharmacol 19(1):pyv072. 4. Kaelen M, et al. (2015) LSD enhances the emotional response to music. Psychopharmacology (Berl) 232(19):3607–3614. 5. Schmid Y, et al. (2015) Acute Effects of Lysergic Acid Diethylamide in Healthy Subjects. Biol Psychiatry 78(8):544–553. 6. Gasser P, et al. (2014) Safety and efficacy of lysergic acid diethylamide-assisted psychotherapy for anxiety associated with life-threatening diseases. J Nerv Ment Dis 202(7):513–520. 7. Halberstadt AL (2015) Recent advances in the neuropsychopharmacology of serotonergic hallucinogens. Behav Brain Res 277:99–120. 8. Fink M (1969) EEG and human psychopharmacology. Annu Rev Pharmacol 9:241–258. 9. Muthukumaraswamy SD, et al. (2013) Broadband cortical desynchronization underlies the human psychedelic state. J Neurosci 33(38):15171–15183. 10. Kometer M, Pokorny T, Seifritz E, Volleinweider FX (2015) Psilocybin-induced spiritual experiences and insightfulness are associated with synchronization of neuronal oscillations. Psychopharmacology (Berl) 232(19):3663–3676. 11. Riba J, et al. (2002) Topographic pharmaco-EEG mapping of the effects of the South American psychoactive beverage ayahuasca in healthy volunteers. Br J Clin Pharmacol 53(6):613–628. 12. Wood J, Kim Y, Moghaddam B (2012) Disruption of prefrontal cortex large scale neuronal activity by different classes of psychotomimetic drugs. J Neurosci 32(9):3022–3031. 13. Celada P, Puig MV, Díaz-Mataix L, Artigas F (2008) The hallucinogen DOI reduces lowfrequency oscillations in rat prefrontal cortex: Reversal by antipsychotic drugs. Biol Psychiatry 64(5):392–400. 14. Riga MS, Soria G, Tudela R, Artigas F, Celada P (2014) The natural hallucinogen 5-MeODMT, component of Ayahuasca, disrupts cortical function in rats: Reversal by antipsychotic drugs. Int J Neuropsychopharmacol 17(8):1269–1282. 15. Vollenweider FX, et al. (1997) Positron emission tomography and fluorodeoxyglucose studies of metabolic hyperfrontality and psychopathology in the psilocybin model of psychosis. Neuropsychopharmacology 16(5):357–372. 16. Carhart-Harris RL, et al. (2012) Neural correlates of the psychedelic state as determined by fMRI studies with psilocybin. Proc Natl Acad Sci USA 109(6):2138–2143. 17. Carhart-Harris RL, et al. (2014) The entropic brain: A theory of conscious states informed by neuroimaging research with psychedelic drugs. Front Hum Neurosci 8:20. 18. Petri G, et al. (2014) Homological scaffolds of brain functional networks. J R Soc Interface 11(101):20140873. 19. Tagliazucchi E, Carhart-Harris R, Leech R, Nutt D, Chialvo DR (2014) Enhanced repertoire of brain dynamical states during the psychedelic experience. Hum Brain Mapp 35:5442–5456. 20. Palhano-Fontes F, et al. (2015) The psychedelic state induced by ayahuasca modulates the activity and connectivity of the default mode network. PLoS One 10(2):e0118143. 21. Monroe RR, Heath RG (1961) Effects of lysergic acid and various derivatives on depth and cortical electrograms. J Neuropsychiatry 3:75–82. 22. Schwarz BE, Sem-Jacobsen CW, Petersen MC (1956) Effects of mescaline, LSD-25, and adrenochrome on depth electrograms in man. AMA Arch Neurol Psychiatry 75(6):579–587. 23. Serafetinides EA (1965) The EEG effects of LSD-25 in epileptic patients before and after temporal lobectomy. Psychopharmacology (Berl) 7(6):453–460. 24. Mégevand P, et al. (2014) Seeing scenes: Topographic visual hallucinations evoked by direct electrical stimulation of the parahippocampal place area. J Neurosci 34(16):5399–5405.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1518377113

presented here as part of a broader initiative to properly utilize these valuable scientific tools. Methods This study was approved by the National Research Ethics Service committee London-West London and was conducted in accordance with the revised declaration of Helsinki (2000), the International Committee on Harmonization Good Clinical Practice guidelines, and National Health Service Research Governance Framework. Imperial College London sponsored the research, which was conducted under a Home Office license for research with schedule 1 drugs. For more methods see SI Appendix, Methods. ACKNOWLEDGMENTS. We thank supporters of the Walacea.com crowdfunding campaign for helping secure the funds required to complete this study. This report presents independent research carried out at the National Institute of Health Research/Wellcome Trust Imperial Clinical Research Facility. This research received financial support from the Safra Foundation (which funds D.J. N. as the Edmond J. Safra Professor of Neuropsychopharmacology) and the Beckley Foundation (the study was conducted as part of the Beckley-Imperial research programme). R.L.C.-H. is supported by an Medical Research Council clinical development scheme grant. S.M. is supported by a Royal Society of New Zealand Rutherford Discovery Fellowship. K.M. is supported by a Wellcome Trust Fellowship (WT090199).

25. Vignal JP, Maillard L, McGonigal A, Chauvel P (2007) The dreamy state: Hallucinations of autobiographic memory evoked by temporal lobe stimulations and seizures. Brain 130(Pt 1):88–99. 26. Studerus E, Gamma A, Vollenweider FX (2010) Psychometric evaluation of the altered states of consciousness rating scale (OAV). PLoS One 5(8):e12412. 27. Gorgolewski KJ, et al. (2015) NeuroVault.org: A web-based repository for collecting and sharing unthresholded statistical maps of the human brain. Front Neuroinform 9:8. 28. Carhart-Harris RL, et al. (2015) The effects of acutely administered 3,4-methylenedioxymethamphetamine on spontaneous brain function in healthy volunteers measured with arterial spin labeling and blood oxygen level-dependent resting state functional connectivity. Biol Psychiatry 78(8):554–562. 29. Kometer M, Schmidt A, Jäncke L, Vollenweider FX (2013) Activation of serotonin 2A receptors underlies the psilocybin-induced effects on α oscillations, N170 visualevoked potentials, and visual hallucinations. J Neurosci 33(25):10544–10551. 30. Butler TC, et al. (2012) Evolutionary constraints on visual cortex architecture from the dynamics of hallucinations. Proc Natl Acad Sci USA 109(2):606–609. 31. de Araujo DB, et al. (2012) Seeing with the eyes shut: Neural basis of enhanced imagery following Ayahuasca ingestion. Hum Brain Mapp 33(11):2550–2560. 32. Tolias AS, et al. (2005) Mapping cortical activity elicited with electrical microstimulation using FMRI in the macaque. Neuron 48(6):901–911. 33. Cavonius CR, Estévez-Uscanga O (1974) Local suppression of alpha activity by pattern in half the visual field. Nature 251(5474):412–414. 34. Jensen O, Mazaheri A (2010) Shaping functional architecture by oscillatory alpha activity: Gating by inhibition. Front Hum Neurosci 4:186. 35. Shirahashi K (1960) Electroencephalographic study of mental disturbances experimentally induced by LSD25. Psychiatry Clin Neurosci 14(2):140–155. 36. Horikawa T, Tamaki M, Miyawaki Y, Kamitani Y (2013) Neural decoding of visual imagery during sleep. Science 340(6132):639–642. 37. Lebedev AV, et al. (2015) Finding the self by losing the self: Neural correlates of egodissolution under psilocybin. Hum Brain Mapp 36(8):3137–3153. 38. Kapur S (2003) Psychosis as a state of aberrant salience: A framework linking biology, phenomenology, and pharmacology in schizophrenia. Am J Psychiatry 160(1):13–23. 39. Friston KJ, Frith CD (1995) Schizophrenia: A disconnection syndrome? Clin Neurosci 3(2):89–97. 40. Seidman LJ, et al. (2014) Medial temporal lobe default mode functioning and hippocampal structure as vulnerability indicators for schizophrenia: A MRI study of nonpsychotic adolescent first-degree relatives. Schizophr Res 159(2-3):426–434. 41. Lui S, et al. (2015) Resting-state brain function in schizophrenia and psychotic bipolar probands and their first-degree relatives. Psychol Med 45(1):97–108. 42. Roseman L, Leech R, Feilding A, Nutt DJ, Carhart-Harris RL (2014) The effects of psilocybin and MDMA on between-network resting state functional connectivity in healthy volunteers. Front Hum Neurosci 8:204. 43. Dyer DC, Gant DW (1973) Vasoconstriction produced by hallucinogens on isolated human and sheep umbilical vasculature. J Pharmacol Exp Ther 184(2):366–375. 44. Grob CS, et al. (2011) Pilot study of psilocybin treatment for anxiety in patients with advanced-stage cancer. Arch Gen Psychiatry 68(1):71–78. 45. Bogenschutz MP, et al. (2015) Psilocybin-assisted treatment for alcohol dependence: A proof-of-concept study. J Psychopharmacol 29(3):289–299. 46. Johnson MW, Garcia-Romeu A, Cosimano MP, Griffiths RR (2014) Pilot study of the 5-HT2AR agonist psilocybin in the treatment of tobacco addiction. J Psychopharmacol 28(11):983–992.

Carhart-Harris et al.

Supplementary information (SI)

Fig. S1. The 11 factor ASC was completed at the end of scanning days and is presented here as a radar plot with mean total values (0-1) for the LSD (blue) and placebo conditions (gray). Ten of the 11 factors were rated significantly higher under LSD than placebo (p < 0.05/11, Bonferonni corrected), with “anxiety” as the exception. See Figures S2 and S3 for additional subjective ratings.

1

Fig. S2. Additional VAS-style ratings completed directly after MRI scanning ranked by intensity. All items were rated significantly higher under LSD with the exception of the bottom three (p < 0.05/21, Bonferonni corrected). A control item “I felt entirely normal” (not shown) was rated higher under placebo than LSD. Items are scored and displayed in a consistent way to the ASC (i.e. 0-1 or % max score).

2

Fig. S3. Additional VAS-style ratings completed directly after MEG scanning ranked by intensity. All items, except the bottom six, were rated significantly higher under LSD than placebo (p < 0.05/21, Bonferonni corrected). A control item “I felt entirely normal” (not shown) was rated higher under placebo than LSD. Items are scored and displayed in a consistent way to the ASC (i.e. 0-1 or % max score).

3

Fig. S4. vmPFC RSFC, z-stat maps (p < 0.05, cluster-corrected), seed in purple. The blue horizontal lines on the sagittal sections give the locations of the preceding axial slices. Bottom row = significant between-condition differences in RSFC between the vmPFC seed and the rest of the brain. Blue = decreases and orange = increases in vmPFC RSFC under LSD. All analyses used cluster correction, p < 0.05. Note: the left side of the brain is displayed on the left in all of the presented brain images. Unthresholded maps can be viewed in the following link http://neurovault.org/collections/FBVSAVDQ/, n = 15.

4

Fig. S5. Decreased RSN integrity under LSD for the 7 RSNs that showed this effect to a significant degree. Purple = RSN, light blue = decreased RSFC (p < 0.05, 5000 permutations). For the mean % decrease in integrity within each RSN, see Fig. 2 of the main paper (blue bar), n = 15. 5

Fig. S6. Relationships between imaging outcomes: A) Increased CBF in the visual cortex correlated with decreased alpha power (log (LSD/Placebo)) in occipital cortex sensors. B) Increased V1 RSFC correlated with decreased alpha power in occipital cortex sensors. C) Decreased RSN integrity (mean of 12 RSNs) correlated with decreased signal variance within these same RSNs (mean of 12 RSNs). D) Decreased RSN integrity (mean of 12 RSNs) 6

correlated with the mean decreases in oscillatory power in the 4 frequency bands ranging from delta (1-4Hz) to beta (13-30Hz). E) Decreased RSN integrity (mean of 12 RSNs) correlated with decreased RSN segregation (synonymous with increased between-RSN RSFC and increased RSN desegregation). Mean values for the 8 RSN pairs that showed this effect were used in this correlation. F) Decreased RSN segregation (mean of 8 RSN pairs) correlated with decreased power (mean of 4 bands, delta-beta, 1-30Hz). The small graphics on the axes are intended to assist comprehension of the relevant metrics and the arrows indicate the direction of the effect (i.e. a downward arrow indicates a decrease under LSD); n = 11 for a,b (MEG and fMRI), n = 15 for c,d,e,f (fMRI only).

7

Fig. S7. Correlation between inter-node Euclidian distance (mm) and FD-RSFC correlation (r) for both LSD (a) and placebo (b) after pre-processing. Nodes were defined using the Craddock atlas with 240 parcellations, excluding supplementary motor and motor areas. For each pair of nodes, RSFC was calculated with pearson’s r and transformed into z using fisher’s transformation. For each pair of nodes, a correlation across subjects was calculated between mean FD and RSFC (r) and transformed into z using fisher’s transformation. This correlation is plotted against the distance between nodes (mm). The correlations for LSD and placebo were r = -0.0009 (p = 0.089) and r = -0.025 (p < 0.001), respectively, suggesting that motion did not affect RSFC in a distant dependant manner after pre-processing. 8

Fig. S8. Increases of V1 RSFC to most significant regions. In order to correlate the increases of V1 RSFC with subjective ratings, we needed to retrieve one z value for all of the (significantly increased) regions. The significant areas with p < 0.05 are too widespread for this purpose, we therefore used a threshold of p < 0.01 (5000 permutations) to define the most significant regions and subsequently derived the mean z value across all of these regions. A binarized mask of these regions is presented in this figure.

9

Fig. S9. Correlations between fMRI results and subjective ratings; p values of A and B are bonferroni corrected by 4 (4 visual ‘hallucination’ items) and p value of C is corrected by 11 (11 ASC dimensions). 10

Table S1. Table displays mean values (possible range = 0-20, increments = 1) and positive standard errors for VAS ratings completed at 3 different time points post LSD and placebo injection. See “subjective ratings” below for more details regarding the items. Ratings were visually presented after each scan (on a projection screen visible from within the scanner) and completed via button press. All items, in all 3 modalities, were rated significantly higher under LSD than placebo (p < 0.05/6, Bonferonni corrected, methods). EC = eyes closed.

Table S2. Independent correlations between decreased integrity of 12 RSNs and the VAS item “I experienced a dissolution of my ‘self’ or ‘ego’”. Only decreased integrity within the DMN correlated significantly with ego-dissolution (see main paper).

11

Table S3. Regions showing increased V1 RSFC (p < 0.05, cluster corrected). All changes were in the direction of increased V1 RSFC under LSD. Note: for tables 2-4, in order to get more segregated clusters, the cluster threshold was set to Z>3 (unlike the figures in the paper, in which Z>2.3). Furthermore, only clusters that were bigger than an arbitrary 20 voxels are reported. The placebo and LSD columns report the z value of each condition separately in the same point as Max.

Table S4. Regions showing increased (positive z values) and decreased (negative z values) PH RSFC under LSD.

12

Table S5. Regions showing increased (positive z values) and decreased (negative z values) vmPFC RSFC under LSD.

13

Table S6. Correlations between imaging outcomes and motion (framewise displacement, FD). Very few outcomes correlated significantly with motion but those that did are emboldened and marked with an asterisk.

14

Table S7. Comparing correlations of imaging results with different ASC/VAS ratings: In the main text we describe the following significant correlations: V1 RSFC with complex imagery (ASC), V1 RSFC with elementary imagery (ASC), PH-RSC RSFC with altered meaning (ASC), increased CBF in visual areas with complex imagery (ASC), V1 RSFC with elementary hallucinations (VAS), and PH-RSC RSFC with ego-dissolution (VAS). In this table, these hypothesis-driven significant correlations are compared with correlations of the same results with different ASC or VAS scales (for which no-strong prior hypotheses were held). The term “selectivity” (used in the main article) is used to refer to a significant correlation (those displayed in Figs. 1-3 & S9) being greater in strength than a (spurious) non-hypothesised correlation. The z values represent the difference between correlations. The z values were calculated using a web-utility (1) that is based on the work of Steiger (1980) (2). Correlations related to the visual system and visual experiences are bordered in red and correlation related to the PH-RSC RSFC and ego-dissolution/altered meaning are bordered in blue. If these correlations are indeed “selective” we would expect them to differ significantly from those that fall outside of their borders and indeed this is largely the case. It can be seen from the table that the following correlations are highly selective: V1 RSFC with complex imagery (ASC), PH-RSC RSFC with altered meaning (ASC), increased CBF in visual areas with complex imagery (ASC), and V1 RSFC with elementary hallucinations (VAS). PH = parahippocampus; PCC = posterior cingulate cortex; RSC = retrosplenial cortex. * p 40 units consumed per week), or a medically significant condition rendering the volunteer unsuitable for the study.

Study setting and overview Screening took place at Imperial’s clinical research facility (ICRF) at the Hammersmith hospital campus. All study days were performed at Cardiff University Brain Research Imaging Centre (CUBRIC). Participants who were found eligible for the study attended two study days that were separated by at least 14 days. On one day, the participants received placebo, and on the other day they received LSD. The order of the conditions was balanced across participants, and participants were blind to this order but the researchers were not. On scanning days, volunteers arrived at the study centre at 8:00am. They were briefed in detail about the study day schedule, gave a urine test for recent drug use and pregnancy, and carried out a breathalyser test for recent alcohol use. A cannula was inserted into a vein in the antecubital fossa by a medical doctor and secured. The participants were encouraged to close their eyes and relax in a reclined position when the drug was administered. All participants received 75 µg of LSD, administered intravenously via a 10ml solution infused over a two 17

minute period, followed by an infusion of saline. The administration was followed by an acclimatization period of approximately 60 minutes, in which (for at least some of the time) participants were encouraged to relax and lie with their eyes closed inside a mock MRI scanner. This functioned to psychologically prepare the participants for being in the subsequent (potentially anxiogenic) MRI scanning environment. Participants reported noticing subjective drug effects between 5 to 15 minutes post-dosing, and these approached peak intensity between 60 to 90 minutes post-dosing. The duration of a subsequent plateau of drug effects varied among individuals but was generally maintained for approximately four hours post-dosing. MRI scanning started approximately 70 minutes postdosing, and lasted for approximately 60 minutes. This included a structural scan, arterial spin labelling (ASL) fMRI, and BOLD fMRI. After the MRI scanning, there was a break of approximately 35 minutes, after which MEG scanning was performed. Once the subjective effects of LSD had sufficiently subsided, the study psychiatrist assessed the participant’s suitability for discharge.

Scanning design and content The ASL and BOLD scanning consisted of three eyes-closed resting state scans, each lasting seven minutes. After each seven minute scan, VAS ratings were performed in the scanner via a response-box. The first and third scans were eyes-closed rest but the second scan also incorporated listening to some music. This component of the study will be reported in detail in a separate publication. Prior to each scan, participants were instructed via onscreen instructions to close their eyes and relax. Participants also performed a retinotopic localisation paradigm at the end of the scanning session. This component of the study will be reported in more detail in a separate publication. MEG scanning had a similar structure to the MRI, i.e. there were three eyes-closed restingstate scans with the second scan incorporating music listening, and there were three eyesopen resting-state scans, with the second incorporating the silent viewing of a movie. Again, the music, movie and eyes-open components of the study will be reported in detail in separate 18

publications. Finally, a mismatch negativity paradigm completed the protocol for the MEG scanning sessions, and this will be reported in another separate publication. In this paper, we report on the eyes-closed resting data which was collected with ASL, fMRI and MEG.

Subjective ratings In scanner, VAS ratings were obtained after each scan. The scales included items for intensity, simple imagery, complex imagery, positive mood and ego dissolution and emotional arousal (Table S1). Specifically, they were phrased as follows: 1) “Please rate the intensity of the drug effects during the last scan”, with a bottom anchor of “no effects”, a mid-point anchor of “moderately intense effects” and a top anchor of “extremely intense effects”; 2) “With eyes closed, I saw patterns and colours”, with a bottom anchor of “no more than usual” and a top anchor of “much more than usual”; 3) “With eyes closed, I saw complex visual imagery”, with the same anchors as item 2; 4) “How positive was your mood for the last scan?”, with the same anchors as item 2, plus a mid-point anchor of “somewhat more than usual”; 5) “I experienced a dissolving of my self or ego”, with the same anchors as item 2; and 6) “Please rate your general level of emotional arousal for the last scan”, with a bottom anchor of “not at all emotionally aroused”, a mid-point anchor of “moderately emotionally aroused” and a top anchor of “extremely emotionally aroused”. Since the ASC ratings referred to the peak drug effects and this coincided with the fMRI session (and not the MEG), ASC ratings were only included from the 15 participants who featured in the fMRI analyses (Fig. S1).

MRI Anatomical Scans Imaging was performed on a 3T GE HDx system. These were 3D fast spoiled gradient echo scans in an axial orientation, with field of view = 256 × 256 × 192 and matrix = 256 × 256 ×

19

192 to yield 1mm isotropic voxel resolution. TR/TE = 7.9/3.0ms; inversion time = 450ms; flip angle = 20°.

BOLD fMRI Data Acquisition Two BOLD-weighted fMRI data were acquired using a gradient echo planer imaging sequence, TR/TE = 2000/35ms, field-of-view = 220mm, 64 × 64 acquisition matrix, parallel acceleration factor = 2, 90° flip angle. Thirty five oblique axial slices were acquired in an interleaved fashion, each 3.4mm thick with zero slice gap (3.4mm isotropic voxels). The precise length of each of the two BOLD scans was 7:20 minutes.

BOLD Pre-processing Four different but complementary imaging software packages were used to analyse the fMRI data. Specifically, FMRIB Software Library (FSL) (3), AFNI (4), Freesurfer (5) and Advanced Normalization Tools (ANTS) (6) were used. One subject did not complete the BOLD scans due to anxiety and an expressed desire to exit the scanner and four others were discarded from the group analyses due to excessive head movement. Principally, motion was measured using frame-wise displacement (FD) (7). The criterion for exclusion was subjects with >15% scrubbed volumes when the scrubbing threshold is FD = 0.5. After discarding these subjects we reduced the threshold to FD = 0.4. The between-condition difference in mean FD for the 4 subjects that were discarded was 0.323±0.254 and for the 15 subjects that were used in the analysis the difference in mean FD was 0.046 ±0.032. The following preprocessing stages were performed: 1) removal of the first three volumes; 2) de-spiking (3dDespike, AFNI); 3) slice time correction (3dTshift, AFNI); 4) motion correction (3dvolreg, AFNI) by registering each volume to the volume most similar, in the least squares sense, to all others (in-house code); 5) brain extraction (BET, FSL); 6) rigid body registration to anatomical scans (twelve subjects with FSL’s BBR, one subject with Freesurfer’s bbregister and two subjects manually); 7) non-linear registration to 2mm MNI brain (Symmetric Normalization (SyN), ANTS); 8) scrubbing (8) - using an FD threshold of 0.4 20

(the mean percentage of volumes scrubbed for placebo and LSD was 0.4 ±0.8% and 1.7 ±2.3%, respectively). The maximum number of scrubbed volumes per scan was 7.1%) and scrubbed volumes were replaced with the mean of the surrounding volumes. Additional preprocessing steps included: 9) spatial smoothing (FWHM) of 6mm (3dBlurInMask, AFNI); 10) band-pass filtering between 0.01 to 0.08 Hz (3dFourier, AFNI); 11) linear and quadratic de-trending (3dDetrend, AFNI); 12) regressing out 9 nuisance regressors (all nuisance regressors were bandpassed filtered with the same filter as in step 10): out of these, 6 were motion-related (3 translations, 3 rotations) and 3 were anatomically-related (not smoothed). Specifically, the anatomical nuisance regressors were: 1) ventricles (Freesurfer, eroded in 2mm space), 2) draining veins (DV) (FSL’s CSF minus Freesurfer’s Ventricles, eroded in 1mm space) and 3) local white matter (WM) (FSL’s WM minus Freesurfer’s subcortical grey matter (GM) structures, eroded in 2mm space). Regarding local WM regression, AFNI’s 3dLocalstat was used to calculate the mean local WM time-series for each voxel, using a 25mm radius sphere centred on each voxel (9).

fMRI motion correction After discarding four subjects due to head motion, fifteen were left for the BOLD analysis. There was still a significant between-condition difference in motion for these subjects however (mean FD of placebo = 0.074 ±0.032, mean FD of LSD = 0.12 ±0.05, p = 0.0002). RSFC analysis is extremely sensitive to head motion (7) and therefore special consideration was given to the pre-processing pipeline to account for motion. This section goes into more detail about the pre-processing steps that were performed to reduce artefacts associated with motion as well as other non-neural sources of noise. De-spiking has been shown to improve motion-correction and create more accurate FD values (10) and low-pass filtering at 0.08 Hz has been shown to perform well in removing high frequency motion (11). Six motion regressors were used as covariates in linear regression. It was decided that using more than six (e.g., “Friston 24-parameter motion regression” (12)) would be redundant and may impinge on neural signal (13) (especially 21

when other rigorous processes such as scrubbing (8) and local WM were applied (9)) . Using anatomical regressors is also a common step to clean noise and ventricles, DV and local WM were used in the pipeline employed in the present analyses. Local WM regression has been suggested to perform better than global WM regression (10). It has previously been shown that head motion biases functional connectivity results in a distance-dependant manner (7). Therefore, as a quality control step, at the end of the preprocessing procedure, cloud plots were constructed to test for relationships between internode Euclidian distance and correlations between FD and RSFC across subjects. In cases in which motion is affecting the results, proximal nodes will have high FD-RSFC correlations and distal nodes will have low FD-RSFC correlations. This would result in a negative correlation between distance and FD-RSFC correlation. In the present dataset, the distance to FD-RSFC correlation was very close to zero for both the placebo and LSD conditions (Fig. S7), suggesting that the extensive pre-processing measures had successfully controlled for distance-related motion artefacts. The final quality control step was to correlate the results with mean FD across subjects (Table S6). Reassuringly, very few results correlated with mean motion (FD) and these were: vmPFC-PCC (r = -0.48, p = 0.035), V1-bilateral angular gyrus (r = 0.56, p=0.015). The significant correlation between changes in vmPFC-PCC RSFC and FD is also mentioned in (8) and (14); therefore, we decided not to elaborate on this result in the manuscript as it may have been an artifact of motion.

Seed-based RSFC Based on prior hypotheses, 3 seeds were chosen for these analyses: 1) the bilateral parahippocampus (PH), vmPFC and V1. The PH seed was constructed by combining the anterior and posterior parahippocampal gyrus from the Harvard-Oxford probabilistic atlas and thresholded at 50%. The vmPFC seed was the same as one previously used by our team in analyses of psilocybin fMRI data (15) and MDMA fMRI data (16). The V1 seed was localized for each subject using a modified retinotopic scan. Specifically, subjects were presented with a 4:24 min video that alternated between vertical and horizontal polar angles (8 cycles, resolution=1400 x 1050, visual angle =23 x 23°, TR/TE=2000/25ms, 3mm 22

isotropic voxels). Fourier analysis with two distinct conditions was performed on the placebo data to identify activity corresponding to the vertical and horizontal polar angles (17). V1 was identified manually for each subject (using an in-house program). The vertical meridian served as the border between V1 and V2. Mean time-series were derived for these seeds for each rest scan. The time series of V1 was derived from unsmoothed data because it was based on a functional localizer acquired in the subject’s native space. RSFC analysis was performed using FSL’s FEAT. Pre-whitening (FILM) was applied. A fixed-effects general linear modelling (GLM) was used to combine the results of both rest scans (2 x 7 mins) within a session. Subsequently, a higher level analysis was performed to compare placebo versus LSD conditions using a mixed-effects GLM (FLAME 1), cluster corrected (z>2.3, p)2 > (where denotes the temporal average). According to this definition, Var(X) = 0 only for a constant time series and larger fluctuations around the mean imply larger values of the variance.

Cerebral blood flow The ASL time series were motion corrected using 3dvolreg within AFNI. Brain extraction was performed after correcting for coil sensitivity profiles by spatially regressing the 3rdorder polynomial fit of the minimal contrast data from the ASL time series data. Masks of the lateral ventricles were determined from the structural scan and the average value of the calibration scan within these masks was defined as M0,CSF. The equilibrium magnetisation for arterial blood (M0,blood) was then calculated according to methods previously described (27). For each TI, tag and control time series were separately interpolated to the TR, subtracted and averaged. CBF and arterial arrival times were quantified by fitting a general kinetic model (28) to the resulting multi-TI data using a non-linear fitting routine and the calculated M0,blood. The CBF maps were registered to the BOLD data and transformed to standard space using the same ANTS transformations as described above.

25

Bonferroni correction for correlations For correlations between imaging outcomes and subjective ratings related to visual hallucinations, Bonferonni correction by a factor of 4 was applied (there were 4 items related to visual hallucinations - two from ASC and two from VAS). For imaging outcomes correlated with the VAS measure of ego-dissolution, no correction was applied, as no other VAS items were used in exploratory correlational analyses. For correlations between imaging outcomes and the different dimensions of the ASC, correction by a factor of 11 was applied, as these analyses were exploratory and there are 11 dimensions to the ASC.

MEG MEG recordings For the MEG recordings, participants lay in supine position. Participants’ pulse rate and blood oxygenation level were continually monitored throughout the experiment via a probe over their left hand index finger. Whole-head MEG recordings were made using a CTF 275channel radial gradiometer system sampled at 1200 Hz (0–300 Hz band-pass). An additional 29 reference channels were recorded for noise cancellation purposes and the primary sensors were analysed as synthetic third-order gradiometers. Four of the 275 channels were turned off due to excessive sensor noise. In addition to the MEG channels, we recorded participants’ ECG: horizontal and vertical electro-oculograms as well as electromyograms from bilateral frontalis and temporalis muscles. Participant compliance was also monitored via an eyetracking camera. Seven minutes of resting-state data were recorded in each block after which the VAS scales were completed. Continuous monitoring of participant head position was employed using three fiducial coils (nasion and pre-auricular points).

Data Preprocessing All MEG recordings were initially high-pass filtered at 1 Hz, and segmented into epochs of 2 s in length (210 epochs). Each epoch was then visually inspected, and those with gross 26

artifacts (e.g., head movements, jaw clenches) were removed from the analysis. An automated algorithm was used to remove further epochs contaminated with muscle artefacts. In this algorithm, a set of 30 gradiometer sensors were predefined at the edge of the MEG dewar, as these are most likely to be contaminated by muscle artefacts (29). Using Hanningwindowed fourier transformations, we calculated the mean spectral power for these sensors in the 105-145Hz frequency band for each epoch. If the resulting power averaged across these sensors exceeded 10 fT (29) then that epoch was eliminated from subsequent analysis. On the remaining epochs we then performed independent component analysis (ICA) as implemented in Fieldtrip/EEGLAB (30, 31) to identify and remove ocular, muscle and cardiac artifacts from the data. Any components that showed a correlation (r > .10) in the time domain with the EOG/EMG electrodes were automatically removed. Likewise, any components that showed correlations (r > .10) with similarly filtered EOG/EMG channels after being bandpass filtered in the range 105-145 Hz were removed. Visual inspection was also used to remove artifact components. All subsequent sensor space analysis was performed on the ICA cleaned datasets. Of the twenty participants, one was unable to complete both sessions and a further five were discarded due to the presence of excessive muscle artefacts that could not be satisfactorily removed by ICA, or due to excessive head motion.

Frequency analysis – sensor space Using the FieldTrip toolbox (31) we converted our MEG data to planar gradient configuration and then conducted a frequency analysis of the individual vector directions. Frequency analysis was conducted using Hanning windowed fast Fourier transforms between 1 and 30 Hz at 0.5 Hz frequency intervals and then the planar directions combined to give local maxima under the sensors. Analysis of sensor-level MEG data in a planar gradient (spatial-derivative) configuration has the advantage of easy interpretability, because field maps can be interpreted as having a source directly underneath field maxima (32). For the higher frequency bands (30-100 Hz) we employed frequency analysis using slepian multitapers with spectral smoothing +/- 3 Hz (33). For statistical analysis, we divided individual spectra into the following frequency bands: delta (1 - 4 Hz), theta (4- 8 Hz), alpha 27

(8 - 15 Hz), beta (15 - 30 Hz), low gamma (30 - 49 Hz), and high gamma (51 - 99 Hz) (29). A relatively high upper alpha frequency cut-off (15 Hz) was used as preliminary analyses revealed a striking peak shift in the alpha-band frequency (Fig. 5c). The differences between LSD and placebo were tested using permutation testing of t statistics (34, 35). The Type 1 error rate was controlled using cluster randomization analysis with an initial cluster-forming threshold of p = 0.05 repeated over 5000 permutations.

MEG Source Localization Automated segmentation and labeling was performed for each individual MRI using the Freesurfer software package (36). Leadfield matrices were then computed on the resultant meshes for each subject using the overlapping spheres method (37). This method models cortical spheres beneath each sensor with elementary current dipoles estimated on a grid perpendicular to the cortical surface. Each individual forward model comprised 10,000 vertices. Noise covariance matrices were computed from empty room recordings. Before calculating the source kernel an additional head-cleaning procedure was performed. Transient head movements greater than 5mm were discarded, while significant repositioning of the head during the scan was dealt with by slicing the recording into discreet units, computing the forward solution for each, and concatenating the resultant sources. Source time-series were obtained by calculating an unconstrained kernel using the dynamical statistical parametric mapping (dSPM) method implemented in the open-source software Brainstorm (38). dSPM is a normalized implementation of the generalized minimum-norm solution (MNE) (39), which has been optimized to resolve the MNE inverse solution’s characteristic bias toward the sensors. dSPM and other normalized minimum norm solutions have shown to be less susceptible to dipole localization error than standard or weighted MNE, and yield more accurate estimates of deep lateral and midline sources of interest such as insular, fusiform, cingulate, and parahippocampal gyri (40). Dipoles were assumed not to have fixed orientation, eliminating the necessity of artificial post-hoc smoothing. Following source computation, data were band-pass filtered into frequencies of interest and projected into standard MNI space. Time- course normalization was conducted for each subject by 28

subtracting the voxel mean from each voxel time-point and dividing by the standard deviation. Data were then exported to SPM for statistical comparison. A paired t-test was run for all frequency conditions, comparing the LSD vs placebo time-courses. Error corrections were performed using False discovery rate (FDR) procedure, thresholded at p = 0.05.

Author contributions R.C-H designed and led the study, oversaw recruitment, performed the research and wrote the paper. D.J.N. advised on the study’s design and implementation and edited the paper. A.F. was instrumental in initiating the research and edited the paper. M.K. helped design the study, recruit volunteers, analyse the MEG data and perform and coordinate the research. S.D.M. helped design the study, performed the MEG analyses and the research itself and wrote sections of the paper, including that on MEG. L.R., E.T. and K.M. performed the fMRI analyses and wrote sections of the paper. L.R., R.L, C.O., J.M. and E.T. oversaw the BOLD analyses. L.R., R.C-H and E.T. were instrumental in producing and arranging the fMRI figures. P.He. helped with the fMRI analyses. K.M oversaw the CBF analysis and L.R. helped with performing this analysis. S.D.M., T.C.N., E.E.S. and M.K. all contributed to the MEG analyses with S.D.M. overseeing its implementation. T.C.N. performed the sourcelocalisation analyses and oversaw this aspect of the MEG analyses. L.T.J.W., T.M.W., M.B., B.S. helped perform the research and care for the participants. T.M.W., M.B., B.S. administered the LSD and served as medical/psychiatric cover for the study. H.V.C. oversaw the storage of the LSD. D.E.N. advised on the stability of and storage conditions for the LSD. R.W., K.S. and J.E. advised on the MRI and MEG scanning design and oversaw CUBRIC’s hosting of the study. P.Ho. was the principal radiographer for the MRI and W.D. ran most of the MEG sessions. M.I.S. oversaw the retinotopic localisation and L.R. implemented this and performed the relevant analyses.

29

References

1. 2. 3. 4. 5. 6. 7. 8.

9. 10. 11.

12. 13. 14. 15. 16.

17. 18. 19. 20.

Lee IA & Preacher KJ (2013) Calculation for the test of the difference between two dependent correlations with one variable in common [Computer software]. Steiger JH (1980) Tests for comparing elements of a correlation matrix. Psychological Bulletin 87:245-251. Smith SM, et al. (2004) Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 23:S208-S219. Cox RW (1996) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical research 29(3):162-173. Dale AM, Fischl B, & Sereno MI (1999) Cortical surface-based analysis: I. Segmentation and surface reconstruction. Neuroimage 9(2):179-194. Avants BB, Tustison N, & Song G (2009) Advanced normalization tools (ANTS). Insight J 2:135. Power JD, et al. (2014) Methods to detect, characterize, and remove motion artifact in resting state fMRI. Neuroimage 84:320-341. Power JD, Barnes KA, Snyder AZ, Schlaggar BL, & Petersen SE (2012) Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 59(3):2142-2154. Jo HJ, Saad ZS, Simmons WK, Milbury LA, & Cox RW (2010) Mapping sources of correlation in resting state FMRI, with artifact detection and removal. Neuroimage 52(2):571-582. Jo HJ, et al. (2013) Effective preprocessing procedures virtually eliminate distancedependent motion artifacts in resting state FMRI. Journal of applied mathematics 2013. Satterthwaite TD, et al. (2013) An improved framework for confound regression and filtering for control of motion artifact in the preprocessing of resting-state functional connectivity data. Neuroimage 64:240-256. Friston KJ, Williams S, Howard R, Frackowiak RS, & Turner R (1996) Movement‐related effects in fMRI time‐series. Magnetic resonance in medicine 35(3):346-355. Bright MG & Murphy K (2015) is fMRI “noise” really noise? Resting state nuisance regressors remove variance with network structureI. NeuroImage 114:158-169. Van Dijk KR, Sabuncu MR, & Buckner RL. (2012) The influence of head motion on intrinsic functional connectivity MRI. Neuroimage 59(1):431-438. Carhart-Harris RL, et al. (2012) Functional Connectivity Measures After Psilocybin Inform a Novel Hypothesis of Early Psychosis. Schizophrenia bulletin. Carhart-Harris RL, et al. (2014) The effects of acutely administered MDMA on spontaneous brain function in healthy volunteers measured with arterial spin labelling and BOLD restingstate functional connectivity Biological Psychiatry. Sereno MI, et al. (1995) Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. Science 268(5212):889-893. Van Essen DC, et al. (2013) The WU-Minn human connectome project: an overview. Neuroimage 80:62-79. Glasser MF, et al. (2013) The minimal preprocessing pipelines for the Human Connectome Project. Neuroimage 80:105-124. Beckmann CF, Mackay CE, Filippini N, & Smith SM (2009) Group comparison of resting-state FMRI data using multi-subject ICA and dual regression. Neuroimage 47(Suppl 1):S148. 30

21.

22. 23. 24. 25.

26. 27.

28.

29.

30.

31.

32.

33. 34. 35. 36. 37.

Roseman L, Leech R, Nutt DJ, Feilding A, & Carhart-Harris RL (2014) The effects of psilocybin and MDMA on between-network resting state functional connectivity in healthy volunteers. Frontiers in human neuroscience 8. Friston KJ (2011) Functional and effective connectivity: a review. Brain connectivity 1(1):1336. Garrett DD, et al. (2013) Moment-to-moment brain signal variability: a next frontier in human brain mapping? Neuroscience & Biobehavioral Reviews 37(4):610-624. Garrett DD, Kovacevic N, McIntosh AR, & Grady CL (2010) Blood oxygen level-dependent signal variability is more than just noise. The Journal of Neuroscience 30(14):4914-4921. Tagliazucchi E, et al. (2013) Breakdown of long-range temporal dependence in default mode and attention networks during deep sleep. Proceedings of the National Academy of Sciences 110(38):15419-15424. Garrett DD, McIntosh AR, & Grady CL (2014) Brain signal variability is parametrically modifiable. Cerebral Cortex 24(11):2931-2940. Wong EC, Buxton RB, & Frank LR (1998) Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II). Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine 39(5):702-708. Buxton RB, et al. (1998) A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine 40(3):383396. Muthukumaraswamy SD, Shaw AD, Jackson, LE, Hall J, Moran R, & Saxena N (2015). Evidence that subanesthetic doses of ketamine cause sustained disruptions of NMDA and AMPAmediated frontoparietal connectivity in humans. The Journal of Neuroscience 35(33):1169411706. Delorme A & Makeig S (2004) EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of neuroscience methods 134(1):9-21. Oostenveld R, Fries P, Maris E, & Schoffelen JM (2011) FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Computational intelligence and neuroscience 2011:156869. Bastiaansen MC & Knosche TR (2000) Tangential derivative mapping of axial MEG applied to event-related desynchronization research. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 111(7):1300-1305. Mitra PP & Pesaran B (1999) Analysis of dynamic brain imaging data. Biophysical journal 76(2):691-708. Nichols TE & Holmes AP (2002) Nonparametric permutation tests for functional neuroimaging: a primer with examples. Human brain mapping 15(1):1-25. Maris E & Oostenveld R (2007) Nonparametric statistical testing of EEG- and MEG-data. Journal of neuroscience methods 164(1):177-190. Fischl B (2012) FreeSurfer. NeuroImage 62(2):774-781. Huang MX, Mosher JC, & Leahy RM (1999) A sensor-weighted overlapping-sphere head model and exhaustive head model comparison for MEG. Physics in medicine and biology 44(2):423-440.

31

38.

39. 40.

Tadel F, Baillet S, Mosher JC, Pantazis D, & Leahy RM (2011) Brainstorm: a user-friendly application for MEG/EEG analysis. Computational intelligence and neuroscience 2011:879716. Dale AM, et al. (2000) Dynamic statistical parametric mapping: combining fMRI and MEG for high-resolution imaging of cortical activity. Neuron 26(1):55-67. Hauk O, Wakeman DG, & Henson R (2011) Comparison of noise-normalized minimum norm estimates for MEG analysis using multiple resolution metrics. NeuroImage 54(3):1966-1974.

32