Serotonin 2A receptor signaling underlies LSD-induced alteration of the neural response to dynamic changes in music
This double-blind, placebo-controlled fMRI study (n=25) investigated how LSD alters the neural response to music in healthy adults. It finds that 5-HT2A receptor signalling is critical for tracking the dynamic tonal structure of music and mediating associated increases in emotionality and meaningfulness.
Authors
- Barrett, F. S.
- Herdener, M.
- Janata, P.
Published
Abstract
Classic psychedelic drugs (serotonin 2A, or 5HT2A, receptor agonists) have notable effects on music listening. In the current report, blood oxygen level-dependent (BOLD) signal was collected during music listening in 25 healthy adults after administration of placebo, lysergic acid diethylamide (LSD), and LSD pretreated with the 5HT2A antagonist ketanserin, to investigate the role of 5HT2A receptor signaling in the neural response to the time-varying tonal structure of music. Tonality-tracking analysis of BOLD data revealed that 5HT2A receptor signaling alters the neural response to music in brain regions supporting basic and higher-level musical and auditory processing, and areas involved in memory, emotion, and self-referential processing. This suggests a critical role of 5HT2A receptor signaling in supporting the neural tracking of dynamic tonal structure in music, as well as in supporting the associated increases in emotionality, connectedness, and meaningfulness in response to music that are commonly observed after the administration of LSD and other psychedelics. Together, these findings inform the neuropsychopharmacology of music perception and cognition, meaningful music listening experiences, and altered perception of music during psychedelic experiences.
Research Summary of 'Serotonin 2A receptor signaling underlies LSD-induced alteration of the neural response to dynamic changes in music'
Introduction
Barrett and colleagues situate the study in the context that classic psychedelics (psilocybin, LSD, DMT) act primarily at serotonin 2A (5HT2A) receptors and profoundly alter perception, cognition and emotion. Prior work shows both that 5HT2A signalling can change neuronal responses along the auditory pathway and that psychedelics alter subjective responses to music, increasing emotionality, imagery and personal relevance. The authors note that music engages both primary auditory processing and a wide set of domain-general networks involved in memory, emotion and self-referential processing, many of which densely express 5HT2A receptors, suggesting a neurobiological overlap that could explain altered music experience during psychedelic states. This study therefore set out to test whether 5HT2A receptor signalling underlies LSD-induced changes in the neural response to the time-varying tonal structure of music. Using a computational model of tonal centre dynamics mapped onto a toroidal representation (toroidal/toroidal surface basis functions decomposed via spherical harmonics), the investigators applied a ‘‘tonality-tracking’’ (TT) analysis to fMRI BOLD data acquired while participants listened to personally meaningful, neutral and meaningless music under placebo, LSD and LSD after pretreatment with the 5HT2A antagonist ketanserin. The goal was to identify brain regions in which neural tracking of dynamic tonal structure was modulated by LSD and by blockade of 5HT2A signalling.
Methods
This report is a secondary analysis of previously published primary fMRI data from a registered, double-blind, randomised, full cross-over study (ClinicalTrials.gov NCT02451072). Twenty-five healthy adults were recruited in Zürich; after exclusions for excessive motion one participant was removed and 24 participants remained in analyses. Screening excluded current or past psychiatric disorders, first-degree relatives with major psychiatric disorder, left-handedness, poor German language skills, cardiovascular disease, neurological disorders, substance dependence, prior adverse reaction to hallucinogens and MRI contraindications. Participants abstained from drugs, alcohol and caffeine for specified intervals and provided urine tests to verify compliance; eight participants reported previous hallucinogen experience. Each participant completed three sessions separated by two weeks in a counterbalanced order. Conditions were: placebo+placebo (Pla), placebo+LSD (100 μg oral LSD), and ketanserin (40 mg oral) pretreatment followed 60 minutes later by LSD (Ket + LSD). Pretreatment (placebo or ketanserin) was given at 08:30 and LSD/placebo 60 minutes later; the fMRI music paradigm was conducted 100 minutes after treatment, corresponding to the expected LSD peak. Ketanserin alone was not tested. During scanning participants heard 20-s excerpts from three stimulus categories: personally meaningful songs (six songs provided by each participant; each participant identified the most meaningful 20 s), neutral songs matched via Last.fm, and personally meaningless songs selected from brief excerpts of free jazz or folk music. Excerpts were normalised and presented in pseudo-randomised blocks. A trial comprised 20 s of music followed by 6 s for a meaningfulness rating (4-point scale) and a jittered fixation (7–11 s); participants performed three practice trials before drug administration and were instructed to keep their eyes closed during listening. Each condition comprised ten presentations (five different excerpts presented twice), for a total of 30 blocks and 17.5 minutes of scanning. Responses were collected via a button box and eye-tracking used to monitor compliance with eyes-closed instructions. MRI was acquired on a Philips Achieva 3.0-T with a sequence designed to reduce scanner noise. Functional EPI parameters included TR 2500 ms, TE 25 ms, 39 axial slices, 3 mm slice thickness, in-plane resolution ~2.75 × 2.75 mm; a high-resolution T1 was also acquired. Preprocessing in SPM12 included slice-time correction, realignment, normalization to MNI space and 8 mm smoothing. Head motion was assessed and ArtRepair used to interpolate volumes with gross motion in two participants; one participant exceeded motion thresholds and was excluded. Tonality-tracking (TT) regressors were generated using 34 toroidal surface basis functions (Janata Lab music toolbox) that describe the time-varying tonal centre of each excerpt. TT regressors were entered into models fit to the residuals of a base GLM that already modelled block effects (self-relevant, neutral, meaningless), behavioural response events and motion parameters. Two principal TT analyses were performed: one contrasting personally meaningful (“self”) versus other (neutral/meaningless) music using two sets of 34 TT regressors (68 regressors), and one comparing drug effects by regressing residuals on 34 TT regressors derived from all stimuli within each scanning session (102 regressors across sessions). An attempted model with separate TT regressors for each combination of drug and stimulus (204 regressors) yielded no significant results and was not reported further, likely due to insufficient degrees of freedom. Subject-level TT voxels were identified using Monte Carlo nonparametric simulation (P < 0.05). TT bias was quantified as the ratio of F-statistics describing variance explained by TT regressors between conditions, adjusted for the number of musical selections. Group-level TT clusters were identified using cluster-mass thresholding with family-wise error correction (P < 0.05, minimum cluster extent 20 voxels). Anatomical labelling used the SPM Anatomy Toolbox, Duvernoy atlas and WFU PickAtlas for Brodmann areas.
Results
Comparing personally meaningful versus nonmeaningful music across all drug conditions, significant TT (variance explained by tonal-change regressors) was observed in clusters spanning prefrontal, cingulate, insular, temporal, occipital and cerebellar cortices and in the thalamus. Regions showing a TT bias toward personally meaningful music included bilateral superior temporal cortices, anterior insula, anterior cingulate, right inferior frontal and angular gyri, left cerebellum, calcarine gyrus and thalamus. The strongest TT bias for meaningful music was in the left anterior insula, right inferior frontal gyrus and bilateral superior temporal gyri (Brodmann areas 41 and 42), where roughly twice the variance was explained by TT regressors for meaningful versus nonmeaningful stimuli. No regions showed greater TT for nonmeaningful music. For drug comparisons, brain activity in temporal, frontal, cingulate, insular, parietal, occipital and cerebellar cortices, as well as amygdala and thalamus, was significantly associated with TT regressors under placebo, LSD and Ket + LSD. Placebo versus LSD: several regions showed a TT bias toward stronger TT during LSD than placebo, including superior, middle and inferior frontal cortices (notably Brodmann areas 10 and 11 and inferior frontal gyrus pars orbitalis), temporal pole, right superior temporal gyrus, angular gyrus, amygdala and cerebellum. The greatest LSD-related increases in TT (up to 7-fold) were reported in cerebellar regions and inferior frontal gyrus subregions. Conversely, a TT bias toward placebo (stronger TT during placebo than LSD) was found in left anterior insula, right inferior frontal gyrus pars opercularis, precentral gyrus, posterior superior temporal gyrus and right calcarine (V1); the left anterior insula showed more than 3 times the variance explained during placebo than during LSD. LSD versus Ket + LSD: a pattern similar to placebo versus LSD emerged, with TT biased toward LSD in superior, middle and inferior frontal gyri (BAs 9, 10, 11; IFG pars orbitalis), temporal pole, right superior temporal gyrus, angular gyrus, amygdala and cerebellum. Additional LSD-biased TT was observed in right anterior cingulate and thalamus. Placebo versus Ket + LSD: the extracted text reports significant TT in temporal, frontal, cingulate, insular and occipital regions and left thalamus, and notes TT bias toward Ket + LSD in inferior frontal regions; however, the extraction truncates the full cluster list and does not clearly report all group-level clusters or their effect sizes for this contrast. Some region-specific patterns did not follow a simple 5HT2A account: TT in right insula was biased toward LSD compared with placebo but toward Ket + LSD when comparing Ket + LSD with LSD. Mid- and posterior cingulate TT was biased toward Ket + LSD versus both placebo and LSD. A left superior temporal cluster (Brodmann area 41, Heschl’s gyrus) showed greater TT during placebo than LSD but no difference between LSD and Ket + LSD. The left calcarine gyrus showed TT bias toward meaningful stimuli and toward placebo versus LSD, yet toward LSD versus Ket + LSD. The authors note that some of these patterns could reflect LSD’s complex pharmacology beyond 5HT2A or interactions with ketanserin that were not isolated because ketanserin alone was not tested.
Discussion
Barrett and colleagues interpret the results as evidence that LSD alters the neural coupling between music’s dynamic tonal structure and activity in a widespread set of brain regions supporting auditory processing, semantic and predictive processing, memory, emotion and self-referential functions. A role for 5HT2A receptor signalling is supported by attenuation of many LSD-associated TT increases after pretreatment with ketanserin, implicating 5HT2A mechanisms in the enhanced neural tracking of tonal dynamics under LSD. The authors highlight particular regional findings: enhanced TT under LSD in a right superior temporal subregion (BA21) that preferentially responds to lyrical and phonetic content, and in inferior frontal gyrus pars orbitalis, consistent with effects on higher-order, semantic or predictive sequence processing in music and language. Medial prefrontal regions linked to autobiographical memory also showed LSD-biased TT, aligning with previous reports that psychedelics enhance autobiographical recollection and suggesting a mechanism by which LSD might facilitate memory recall or alter cognitive biases in therapeutic contexts. Increased TT in the angular gyrus is interpreted as greater engagement of a cross-modal integration hub that can increase salience and emotional impact during personally meaningful music under LSD. Enhanced amygdala TT is noted as consistent with increased sensitivity to auditory stimuli during LSD. At the same time, the authors acknowledge complex and sometimes contradictory effects that are not readily attributed solely to 5HT2A signalling. Examples include insular and cingulate TT biases that differed across contrasts and early auditory and visual cortex effects that did not consistently track the 5HT2A pattern. Given LSD’s affinity for a range of receptors (dopaminergic, adrenergic and multiple serotonin receptors), these inconsistencies could reflect contributions from non-5HT2A receptors or interactive effects. The study design could not assess ketanserin alone, which limits attribution of effects specifically to blockade of 5HT2A receptors. Overall, the investigators propose that 5HT2A-dependent alterations in tonality-tracking across higher-order association cortices and limbic regions could explain the commonly reported increases in emotionality, connectedness and meaningfulness of music during psychedelic experiences. They situate these findings as relevant for understanding the neurochemical basis of music perception and for the role of music as psychological support in psychedelic research and therapy, while noting that some observed effects remain to be clarified by future work that can disentangle receptor-specific actions and address design limitations.
Conclusion
The study concludes that 5HT2A receptor signalling modulates the neural coupling between patterns of change in musical tonal structure and activity in brain areas supporting an integrated musical experience. LSD altered tonality-tracking in multiple domain-general regions, and many of these effects were attenuated by pretreatment with the 5HT2A antagonist ketanserin, implicating 5HT2A mechanisms in the increased emotionality, connectedness and meaningfulness evoked by music under LSD. The authors suggest these findings inform the neuropharmacology of music perception and have implications for the therapeutic use of music during psychedelic-assisted interventions.
View full paper sections
INTRODUCTION
Classic psychedelics, including psilocybin, lysergic acid diethylamide (LSD), and dimethyltryptamine (DMT), are potent compounds that have their primary receptor mechanism of action at serotonin 2A (5HT 2A ) receptor sites. 5HT 2A receptors are widely distributed throughout the neocortex. Accordingly, psychedelic drugs, including psilocybin, DMT, and LSD, have substantial effects on perception, cognition, and emotional experience (reviewed in. Psychedelic drugs also have notable effects on the perception of music. This is not surprising, as 5HT 2A signaling has been shown to alter neuronal responses to auditory stimuli from the cochlear nucleusalong the precortical primary auditory sensory pathwaythrough to primary auditory cortex. Psychedelics, however, do not simply alter the perception of sensory stimuli such as music. LSD has been shown to increase positive mood during music listeningas well as music-induced imagery and communication of related brain regions). LSD has also been shown to increase the personal relevance of both meaningless and meaningful music, and alter functioning of brain areas involved in processing the meaningfulness of stimuli. This follows from the neurobiology, as the neural response to music involves both the primary auditory pathway and a wide range of domain-general brain networks including those involved in memory, emotions, self-referential processing, and visualization. Many of these brain regions densely express 5HT 2A receptors and show marked alterations in activity and/or connectivity during the acute effects of psychedelics). Thus, there is extensive neurobiological overlap in the brain regions that are impacted by psychedelics and the brain regions that may be recruited during music listening. Research and clinical methods have taken advantage of altered experience of music during psychedelic experiences. Music has played a key role in the conduct of psychedelic therapy and research for many decades, with the expectation that supportive music may facilitate a meaningful experience. Current best practices for safe conduct of a psychedelic session include the use of music, with the goal of providing psychological support. Neurochemical effects of music listening on stress, immunity, and social affiliation have been demonstrated (reviewed by, and music listening has specifically been shown to lead to dopamine release and drive reward circuitry. While these effects of music on neurochemical processes are important, there are very few empirical studies) that have investigated effects in the opposite direction; namely, the more general role of neuropharmacology in supporting or altering music perception and cognition. There is evidence suggesting commonalities among pieces of music that are optimally supportive during peak experiences with psychedelics, supporting the notion that there are structural principles to the relationship between music listening and psychedelic experiences. These principles may have emerged from a shared neuropharmacological basis of psychedelic experience and music perception and cognition. Music is a complex stimulus that varies in time in a number of dimensions that range from lower-level acoustic features (such as loudness, frequency spectra, or simple tonal features such as pitch height) to higher-level cognitive schema that represent relationships between events that define rhythm, meter, and tonality. Tonality refers to the system in Western tonal music of major and minor keys, in which notes and chords change over time and fulfill or violate expectancies to create a sense of tension and resolution. Tonality has been well-defined as an important cognitive schema for shaping expectations during music listening. Change over time in the tonal center of a musical selection (i.e., which key is implied by the music at a given time) can be collectively described and computationally modeled as changes in the pattern of activation on a toroidal surface. The toroidal surface model was initially derived from multidimensional scaling analysis of subjective ratings of the perceptual "fit" of probe tones after a tonal center was established in each of 24 major and minor keys, where distance between tonal centers on the 4-dimensional surface of a torus directly reflects perceptual distance (the inverse of perceptual "fit") between notes, chords, and keys. The model was shown to reflect not only the perception of tonal distance when listening to musical stimuli, but also the organization of tonal structure as understood in music theory). Thus, modeling tonal space on a toroidal surface simultaneously represents concepts in music theory, cognitive psychology, and the pitch statistics of western music. The tonal center of a piece of music at any given point in time can be computationally derived by integrating the tonal information in that piece of music over a window of time preceding the point of interest, and this tonal center can subsequently be reflected on a torus. As the melodies and harmonies of a piece of music unfold in time, the sense of tonal center of that piece of music also unfolds in time. Change over time in tonal center can be calculated using a sliding window, and across a piece of music, a timecourse of change in tonal center can be calculated and reflected on a torus. This time-varying pattern of information on the torus thus represents the dynamic tonal structure of a piece of music. The rate of change in that timecourse is determined by both the stability of the distributions of tonal information in the music as well as the duration of the sliding window over which one integrates the tonal information. Using spherical harmonic analysis, this timecourse of toroidal (4-dimensional) representation can be decomposed into a series of 34 spatially orthogonal patterns with associated weight vectors, and these weight vectors can be entered into a design matrix that can be used to regress fMRI blood oxygen level-dependent (BOLD) activity that was measured while an individual was listening to a piece of music). This analysis approach has been applied previously, and spherical harmonic regressors describing the change in tonal center over time were shown to explain variance in brain activity measured while volunteers listened to melodies that moved systematically through all 24 major and minor keys in tonal space. Change in tonal center (and toroidal space) has been associated with variance in BOLD signal in different domain-general brain regions depending on the psychological context of the music listening experience, such as the experience of music-evoked nostalgiaand autobiographical memories. This type of stimulus/brain coupling has been labeled "tonality-tracking" (TT). The current report applies TT analysis of BOLD signal collected while participants listened to both personally meaningful and nonmeaningful music after the administration of placebo, LSD, and LSD pretreated with the 5HT 2A antagonist ketanserin, to investigate the role of 5HT 2A receptor signaling in the neural response to the time-varying tonal structure of music.
METHODS
The following is a secondary analysis of data published elsewhere). The same stimuli and primary fMRI data that were reported previouslyare used in the current report. Participants in this study completed additional ratings and questionnaires, which are reported in the primary publication. The description below refers only to those measures and procedures investigated in the current report. The study was registered at ClinicalTrials.gov (NCT02451072).
PARTICIPANTS
Twenty-five participants were recruited through advertisements placed in local universities in Zürich, Switzerland. Interested persons attended a screening visit before inclusion in the study. Participants were screened with a short hearing test, medical history, physical examination, blood analysis, and electrocardiography, and had normal or corrected-to-normal vision. The Mini-International Neuropsychiatric Interview, the DSM-IV self-rating questionnaire for Axis-II personality disorders, and the Hopkins Symptom Checklistwere used to exclude individuals with present or previous psychiatric disorders or a history of major psychiatric disorders in first-degree relatives. Left-handedness, poor knowledge of the German language, cardiovascular disease, history of head injury or neurological disorder, history of alcohol or illicit drug dependence, a previous significant adverse reaction to a hallucinogenic drug, and any contraindication for magnetic resonance imaging, including claustrophobia were further exclusion criteria. Participants were asked to abstain from the use of any prescription or illicit drugs for a minimum of 2 weeks prior to the first test day and for the duration of the entire study, and to abstain from drinking alcohol for at least 24 h prior to each test day. Participants were required to abstain from smoking for at least 60 min before MRI assessment and from drinking caffeine at any point during the test day. Urine tests and self-report questionnaires were used to verify the absence of drug and alcohol use on the screening visit and each experimental visit before drug administration. Urine tests were also used to exclude pregnancy. One participant was excluded from final analysis due to head motion during scanning (see MR Data Acquisition and Preprocessing below). Of the remaining 24 participants, 8 reported previous experience with hallucinogens. The Swiss Federal Office of Public Health, Bern, Switzerland, authorized the use of LSD in humans, and the Cantonal Ethics Committee of Zurich approved the study. All participants provided written informed consent statements in accordance with the declaration of Helsinki before participation in the study.
STUDY DESIGN
In a double-blind, randomized, full cross-over design, participants received either: (1) placebo+placebo (Pla) condition: treatment with placebo (179 mg Mannitol and 1 mg Aerosil po) after pretreatment with placebo (179 mg Mannitol and 1 mg Aerosil po); (2) placebo + LSD (LSD) condition: treatment with LSD (100 μg po) after pretreatment with placebo (179 mg Mannitol and 1 mg Aerosil po); or (3) ketanserin+LSD (Ket + LSD) condition: treatment with LSD (100 μg po) after pretreatment with the 5-HT2A antagonist ketanserin (40 mg po). These 3 conditions were performed on 3 different occasions each separated by 2 weeks. Pretreatment with placebo or ketanserin was administered at 8:30 AM and occurred 60 min before treatment with placebo or LSD. The fMRI music paradigm was conducted 100 min after treatment with placebo or LSD, during the expected peak of subjective effects of LSD.
MUSIC PARADIGM
During the music paradigm, participants listened to an equal number of personally meaningful songs (songs provided by the participant), neutral songs (matched to the meaningful songs), and personally meaningless songs while BOLD fMRI signal was acquired. For personally meaningful songs, participants were asked to provide 6 songs that were personally meaningful to them, and identify the most meaningful 20 s of each song. Participants then completed a pre-task questionnaire (PTQ) after listening to each identified 20 s musical excerpt. The PTQ consisted of the following questions for each song: "How personally meaningful is this song for you?" (1: not at all, 4: very much); "How strongly does this song give you the chills?" (1: not at all, 4: very strongly); and "How do you feel when hearing this song?" (1: sad, 9: happy). Participants also answered the question "How meaningful is music to you in general?" (1: not at all meaningful, 9: very meaningful). The song least meaningful to the participant was identified using the response to the first item of the PTQ ("How personally meaningful is this song for you?") and used for the practice session conducted at each testing session before substance administration. If more than 1 song scored lowest on this item, the mean score of the other 2 items was used to identify the least meaningful song. To select the meaningless music played during fMRI acquisition, participants were presented with 4 music excerpts of 20-s duration. Two excerpts were classified as free jazz music and 2 as traditional folk music. In each category, 1 excerpt included vocals and 1 was instrumental. Categories were presented in alternating order. For each excerpt, the participants answered the following 4 questions: 1: "How personally meaningful is this song for you?, 2: "How emotionally touched do you feel by this song?, 3: "How pleasant do you consider this song?, and 4: "How connected do you feel to this song? All questions were answered on a 4-point scale ranging from 1 (not at all) to 4 (very much). Music excerpts were presented and responses registered using Presentation (version 17.0; Neurobehavioral Systems). Stimuli from the category with the lowest average mean rating on question 1 was presented during practice trials and fMRI acquisitions on test days. If both categories were rated equally, the category with the lowest mean rating across all 4 questions was presented. Folk music was presented for 3 participants and free jazz music for 21 participants. The music excerpts presented during practice trials and fMRI acquisitions on test days were different from those used at the screening visit. Six neutral music excerpts were matched to the 6 songs provided by the participant using the "search for similar music" function of music aggregator website, Last.fm (www.last.fm). Results were sorted according to the number of listeners and the song with the lowest number of listeners was chosen to reduce the chance that participants were familiar with the song. A list showing the artists and song titles was presented to the participants on the first test day before substance administration to make sure they did not know the songs and the songs Music, LSD, and 5-HT2A Barrett et al. | 3941 had no special meaning or personal relevance to them. The music excerpts were created by selecting a random 20 s period for the neutral and meaningless songs, and the identified most meaningful 20 s period for the provided songs, using Audacity 2.1.2 (www.audacityteam.org). All music excerpts were normalized to a maximum amplitude of -1 dB. On each test day participants performed 3 practice trials before drug administration to familiarize themselves with the task. The 3 trials consisted of the least meaningful music excerpt from the participant's provided songs, the respective matched neutral music excerpt, and one meaningless song from the participant's meaningless music category. The songs presented during the practice session were not presented during fMRI data acquisition. During fMRI data acquisition and practice sessions, a trial block consisted of music presentation for 20 s, followed by a period of 6 s in which participants provided a meaningfulness rating. A fixation cross was presented between blocks with a jittered duration of 7-11 s (mean, 9 s). The musical stimuli were presented in pseudo-randomized order using Presentation (version 17.0; Neurobehavioral Systems). Five different music excerpts were chosen for each condition and each musical excerpt was presented twice, for a total of 10 musical excerpt presentations for each condition. There were 30 blocks in total (3 conditions × 10 blocks per condition). The total duration of the paradigm was 17.5 min. Participants listened to the excerpts of music through MR-compatible in-ear headphones (MR Confon) additionally shielded by soundproof circumaural headphones. Responses (from the right hand) were collected using a 4-button response box (Current Designs,). Visual cues were presented with binoculars (NordicNeuroLab VisualSystem, Bergen, Norway,). Participants were asked to close their eyes during music presentation. Compliance to this instruction was monitored online using eye tracking (NordicNeuroLab VisualSystem,). All participants followed the instructions. A black screen was presented during music presentation. Subsequent to the music, participants heard the spoken word "rating", indicating that they should open their eyes and rate the meaningfulness of the music excerpt on a 4-point scale ranging from one (not at all meaningful) to 4 (very meaningful). During the rating, the text "Meaningful?" was displayed at the top of the screen and the 4 response options ("Not at all", "A little", "Moderate", and "Very") were displayed at the bottom of the screen.
MR DATA ACQUISITION AND PREPROCESSING
MR data were acquired on a Philips Achieva 3.0-T whole-body scanner (Best, The Netherlands). The sequence was specifically designed to produce quiet and constant scanner noise using the "SofTone" parameter. A 32-channel receive head coil and MultiTransmit parallel radio frequency transmission was used. Images were acquired using a whole-brain gradient-echo planar imaging sequence (repetition time, 2500 ms; echo time, 25 ms; slice thickness, 3 mm; 39 axial slices; no slice gap; field of view, 220 × 220 mm 2 ; in-plane resolution, 2.75 × 2.75 mm; sensitivity-encoding reduction factor, 2.0). High-resolution anatomical images (voxel size, 0.7 × 0.7 × 0.7 mm) were also acquired using a standard T1-weighted 3D magnetization-prepared rapid gradient-echo (MP-RAGE) sequence. Images were analyzed using SPM12 (www.fil.ion.ucl.ac.uk). Preprocessing consisted of slice time correction, realignment, spatial normalization to the standard echo planar imaging template of the Montreal Neurological Institute (MNI), and spatial smoothing using a Gaussian kernel of 8-mm full-width half-maximum to meet the statistical requirements of the general linear model. Head motion was assessed using the output of the realignment step of preprocessing. One participant was excluded due to having gross motion (>3 mm) in more than 20% of measured volumes. Two additional participants were identified who exhibited gross motion in <10% of measured volumes, and for these individuals, the ArtRepair toolbox) was used to interpolate volumes exhibiting gross motion.
TT ANALYSIS
Custom MATLAB scripts (the Janata Lab music toolbox []) were used to generate 34 toroidal surface basis functions that describe the timevarying tonal structure of each musical stimulus. These surface basis functions were used as TT regressors in subsequent TT analyses, following previously described methods. TT models were fit to the residuals of a general linear model (the "base" model) that described the experimental design of the study, the results of which have been reported elsewhere). The base model consisted of a set of block regressors indicating when self-relevant, neutral, and meaningless stimuli were presented, an event-related regressor that indicated when in-scanner behavioral responses were made, and motion parameters calculated from the realignment stage of preprocessing. Two models were fit to the residuals of the base model: 1 for analysis of music conditions, and 1 for analysis of drug effects. For analysis of music conditions, the residuals of the base model were regressed on a design matrix consisting of a set of 34 TT regressors that were derived from musical stimuli that were personally meaningful, or self-relevant ("self"), and a second set of 34 TT regressors that were derived from musical stimuli that were either personally meaningless or neutral ("other"; 68 regressors total). F-contrasts were calculated to estimate the variance explained by each set of regressors during each scanning session. For analysis of drug effects, the residuals of the base model were regressed on 34 TT regressors that were derived from all musical stimuli that were presented in a given scanning session for each drug condition (102 TT regressors total). F-contrasts were then calculated to estimate the variance explained by each set of regressors for each scanning session (after placebo, LSD, or Ket + LSD). An additional model was fit including separate TT regressors for each combination of drug and stimulus condition, yielding a design matrix with 204 regressors. This model was fit in order to directly test the interaction of drug condition and stimulus condition on TT within the brain, however this model yielded no significant results, likely due to exceeding the empirical degrees of freedom within the dataset. Thus, this model is not reported further. Monte Carlo simulation was used at the subject-level for each model to identify TT voxels, which are defined as voxels in which a significant amount of variance (nonparametric simulation thresholded at P < 0.05) was predicted by the entire set of TT regressors for a given scanning session. Preference (bias) of a voxel for TT during a given condition was assessed by calculating the ratio of the F-statistics describing the variance explained by tonality regressors in each given condition in each subject-level model, adjusted for the number of musical selections in each given experimental condition. TT bias was calculated separately for each pair of drug conditions (placebo vs. LSD, Ket + LSD vs. LSD, and placebo vs. Ket + LSD) and for musical condition (self vs. other). Cluster mass thresholding) was used as previously describedto identify brain areas that showed TT at the group level for each pair of drug conditions (placebo vs. LSD, placebo vs. Ket + LSD, LSD vs. Ket + LSD) and for musical stimulus condition (self vs. other), thresholding at P < 0.05 and a minimum cluster extent of 20 voxels, correcting for multiple comparisons and family-wise error rate. Average TT bias across participants was calculated for each experimental condition in each significant group-level TT cluster. Anatomical labels for TT clusters were identified using the SPM Anatomy Toolboxand the Duvernoy Atlas. Brodmann areas that were overlapping with TT clusters were identified using the WFU PickAtlas ().
COMPARING TT BETWEEN MEANINGFUL AND NONMEANINGFUL MUSIC LISTENING CONDITIONS, ACROSS ALL DRUG CONDITIONS
TT is determined by assessing the amount of variance in a given voxel or brain region that is explained by set of regressors that describe the change in tonal toroidal space over time for a piece of music. TT bias, or greater TT in one condition compared with another, is determined by assessing the ratio of variance explained in a given brain region by TT regressors in one condition (for instance, personally meaningful music) compared with the variance explained by TT regressors in another condition (for instance, nonmeaningful, or "other", music). TT bias has been shown in a number of brain regions involved in memory, emotion, and language while individuals listened to personally meaningful music that was autobiographically salientor evoked nostalgia, compared to when individuals were listening to stimuli that was not autobiographically relevant or nostalgic, respectively. In the current data, brain activity in a series of clusters in the prefrontal, cingulate, insula, temporal, occipital, and cerebellar cortex, as well as in the thalamus, was significantly associated with the time-varying tonal structure of music (Supplementary Table). Bias for TT (i.e., stronger TT) during personally meaningful music was observed in bilateral superior temporal cortices (Fig.), anterior insula, and anterior cingulate (Fig.), as well as right inferior frontal and angular gyri (Fig.), and left cerebellum, calcarine gyrus (Fig.), and thalamus (Fig.). TT bias was greatest in the left anterior insula, right inferior frontal gyrus, and bilateral superior temporal gyri (in Brodmann areas 41 and 42), with approximately twice the amount of variance explained in each region by TT regressors for personally meaningful music than by TT regressors for nonmeaningful music. No regions demonstrated TT bias toward nonmeaningful music.
COMPARING TT BETWEEN DRUG CONDITIONS
In order to test the hypothesis that LSD and 5HT 2A receptor signaling would alter the neural response to the time-varying tonal structure of music, the ratio of variance explained by TT regressors during each drug condition was compared in a pair-wise fashion (placebo vs. LSD, LSD vs. Ket + LSD, and placebo vs. Ket + LSD). Placebo vs. LSD: Brain activity in temporal (Fig./D), frontal (Fig./C/E), cingulate (Fig.), insular, parietal (Fig.), occipital (Fig.), and cerebellar cortex, as well as the amygdala and thalamus (Fig.), was significantly associated with TT regressors during placebo and LSD conditions (Supplementary Table). A significant TT bias toward stronger TT during LSD than during placebo was identified in superior, middle, and inferior frontal cortex (including Brodmann areas 10 and 11, and inferior frontal gyrus pars orbitalis), temporal pole, right superior temporal gyrus, angular gyrus, amygdala, and cerebellum. TT bias towards LSD was greatest in regions of the cerebellum (Supplementaryand), inferior frontal gyrus (Supplementaryandand) with up to 7 times the amount of variance explained during LSD than during placebo. A significant TT bias toward stronger TT during placebo than LSD was identified in left anterior insula, right inferior frontal gyrus (pars opercularis), precentral gyrus, posterior superior temporal gyrus, and right calcarine gyrus (area V1). TT bias toward placebo was greatest in the left anterior insula (Supplementary Table), where more than 3 times the amount of variance was explained during placebo than during LSD. LSD vs. Ket + LSD: Brain activity in temporal (Fig./D), frontal (Fig./C/E), cingulate (Fig.), insular, parietal (Fig.), occipital (Fig.), and cerebellar cortex, as well as the amygdala and thalamus (Fig.), was significantly associated with TT regressors during LSD and Ket + LSD (Supplementary Table). Similar to the comparison of placebo and LSD drug conditions, significant TT bias toward stronger TT during LSD than during Ket + LSD was identified in superior, middle, and inferior frontal cortex (including Brodmann areas 9, 10, and 11, and inferior frontal gyrus pars orbitalis), temporal pole, right superior temporal gyrus, angular gyrus, amygdala, and cerebellum. TT bias toward LSD was also observed in the right anterior cingulate and thalamus. TT bias toward LSD was greatest in regions of the cerebellum (Supplementary Placebo vs. Ket + LSD: Brain activity in temporal (Fig.), frontal (Fig.), cingulate (Fig.), insular, and occipital (Fig.) cortex regions as well as left thalamus (Fig.) was significantly associated with TT regressors during placebo and Ket + LSD (Supplementary Table). TT bias toward Ket + LSD was observed in bilateral right inferior frontal gyrus (Supplementary
DISCUSSION
The present study utilized music, pharmacological intervention, and a computational model of the time-varying tonal structure of music to investigate the role of LSD in altering the neural response while listening to personally meaningful and nonmeaningful music. Overall, LSD alters the neural response to music in a number of brain regions that have been shown to support varying aspects of subjective experience during music listening. A role for 5HT 2A receptor signaling in biasing the neural response to music was shown by comparing effects of placebo to effects of LSD alone and after pretreatment with ketanserin (a 5HT 2A antagonist).
LSD, AUDITORY PROCESSING, SELF-RELEVANCE, AND THE NEURAL RESPONSE TO MUSIC
Greater TT bias for LSD compared with placebo and compared with Ket + LSD was expressed in a wide range of brain regions, including a subregion of the right superior temporal gyrus (Brodmann area 21) that was previously shown to be activated by a pleasant melody that meandered systematically through all of tonal space. This brain region has been shown to preferentially respond to lyrical content in music over melodic content) and sentence content more strongly than white noise and musical instrument sounds, or melody content. This brain region was also shown to respond to phonetic elements of speech, and responded to spectral more strongly than temporal changes in speech content. Phonetic and spectral changes in speech are important for vocalization, or singing of lyrics, in music. While a number of other auditory processing regions on the bilateral superior temporal gyri (including Brodmann areas 22, 41, and 42) also displayed TT in the current report, TT within these regions was not biased by drug condition. This suggests a specific effect of LSD in a brain region responsive to lyrical content in music. Altered response to lyrical content may be either a mechanism or a product of the effects of LSD on increased meaning while listening to music. Brain regions responsive to language and tonality that have previously been shown to exhibit TT, including bilateral inferior frontal gyrus pars orbitalis, also demonstrated) depicts a sagittal brain slice containing a TT cluster. Shading within each cluster indicates the TT bias for the voxels in the cluster. TT bias is the ratio of variance explained by TT regressors for personally meaningful stimuli to variance explained by the TT regressors for other stimuli (corrected for the number of stimuli in each category). A TT bias value of 1 indicates voxels that tracked personally meaningful and other stimuli equally well; greater than 1 indicates voxels that tracked personally meaningful stimuli more strongly than other stimuli. TT clusters are circled in green, and the distribution of TT bias for all voxels in that cluster is indicated in the associated histogram. Each histogram refers to a cluster listed in Supplementary Table. The number in the lower-left-hand corner of each sagittal slice indicates the Montreal Neurologic Institute template coordinate (in mm) of that slice in the x dimension. TT bias for LSD compared with placebo and compared to Ket + LSD. Inferior frontal gyrus activity has been linked to processing structured sequences along an rostral-caudal abstraction gradient, where the anterior regions including pars orbitalis are involved in processing more abstract sequences (rather than the more concrete sequences processed by caudal regions). This is consistent with literature that implicates the pars orbitalis structure in predictive semantic processing of both language and music. Left inferior frontal gyrus has typically been shown to respond to speech (Ardila, Bernal, Rosselli 2016), while bilateral inferior frontal gyrus has been shown to respond to music. TT within the inferior frontal gyrus has shown bias toward greater TT during music-evoked nostalgia) and music-evoked autobiographical memories). Thus, LSD seems to particularly influence neural function in higher-order association cortices that respond to pitch, semantics, and memory. This might explain the increased salience of music that is anecdotally reported after the administration of classic psychedelics, including LSD and psilocybin. Medial prefrontal brain regions that have been implicated in core autobiographical memory networks) and specifically shown to support music-evoked autobiographical memory) also demonstrated TT bias toward LSD compared to both placebo and Ket + LSD. These brain regions have been shown to closely track tonal structureand exhibit TT bias toward music that evoked autobiographical memories. This is consistent with previous a report that psychedelics enhance autobiographical recollection, and supports a potential benefit of LSD during psychotherapy either to support the recall of memories or to reverse negative cognitive biases. The angular gyrus also demonstrated TT bias both for LSD compared with placebo and Ket + LSD, and for personally meaningful stimuli compared with other stimuli. While the precise functional role of the angular gyrus in a given context may depend on the regions that are co-active with it at a given time, a domain-general role of the angular gyrus may be to act as a cross-modal hub that integrates sensory information with top-down predictions in order to direct attention. The angular gyrus, while not previously shown to exhibit TT, has previously shown increased activity when participants listened to musical stimuli that were pleasing, familiar, and autobiographically salient. Increased involvement of the angular gyrus in tracking the tonal structure of music during both personally meaningful stimuli and LSD, along with increased TT in brain regions supporting auditory processing greater than 1 (plotted on the warm gradient) indicates voxels that tracked stimuli more strongly during placebo than during LSD; less than 1 (plotted on the cool gradient) indicates voxels that tracked stimuli more strongly during LSD than during placebo. Each histogram refers to a cluster listed in Supplementary Table. and autobiographical memory recall, may account for the increased salience and emotional impact of musical stimulias well as the increased self-relevance of musical stimuliduring the effects of LSD. TT bias has been observed in the current report in brain areas that respond to both music and speech (including superior temporal gyrus and inferior frontal gyrus) and also in higher-level cognitive brain regions associated with processing tonality (e.g., inferior frontal gyrus) and memory and emotion (e.g., medial prefrontal cortex and angular gyrus). Enhanced TT in the amygdala is also notable, and may be consistent with an overall increase in sensitivity to auditory stimuli produced by LSD, in line withwho reported lower thresholds to auditory stimuli after the intake of LSD. Overall, these findings suggest that LSD serves to support a deeper or more integrated experience of music, which could explain the wide range of emotional and cognitive effects that are encountered, especially those experienced in response to music, after the administration of LSD and other serotonergic hallucinogens. Though trending toward TT bias during LSD compared with Ket + LSD, the calcarine sulcus exhibited TT bias toward placebo compared with LSD, and for personally meaningful compared to other musical stimuli. This is consistent with TT bias in extrastriate visual areas that was previously demonstrated for autobiographically salient musical stimuli, notably during eyes-closed music listening conditions. While there is evidence for increased visual cortex connectivity during the effects of LSD, and clear evidence for a specific auditory-to-visual synesthesic effect of classic psychedelics (Luke and Terhune 2013), this effect was not observed with TT. Given that synesthesia is not necessarily encountered during every psychedelic experience, and given that synesthesia can be a very idiosyncratic experience, our failure to find TT in primary visual brain regions could be due to insufficient frequency or homogeneity of synesthetic experience in our sample. It may also suggest that, rather than being dependent upon the content of the physical properties of an auditory stimulus, the induction of synesthesia may depend on aspects of psychedelic experience that were not well-controlled or characterized within the current study. The Role of 5HT 2A Signaling in the Neural Response to the Time-Varying Tonal Structure of Music The primary receptor mechanism of action of classic psychedelics, including LSD, is the 5HT 2A receptor. Within the current study, LSD was given alone after pretreatment with placebo (LSD) or pretreatment with the 5HT 2A receptor antagonist ketanserin (Ket + LSD). TT was biased toward LSD in greater than 1 (plotted on the cool gradient) indicates voxels that tracked stimuli more strongly during LSD than during Ket + LSD; less than 1 (plotted on the green gradient) indicates voxels that tracked stimuli more strongly during Ket + LSD than during LSD. Each histogram refers to a cluster listed in Supplementary Table. superior, middle, and inferior frontal gyri, temporal pole and superior temporal gyrus, angular gyrus, and the amygdala when compared to both placebo and Ket + LSD. Given that these effects were observed in LSD when comparing to both the absence of 5HT 2A receptor agonism (in the placebo condition) and after blockade of the 5HT 2A receptor with ketanserin (in the Ket + LSD condition), these findings indicate a role of 5HT 2A signaling in the neural response to the time-varying tonal structure of music at multiple neural levels, including midbrain and primary auditory brain regions up through higherlevel cognitive brain regions.
THE UNKNOWN ROLE OF OTHER RECEPTOR SIGNALING MECHANISMS IN THE NEURAL RESPONSE TO THE TIME-VARYING TONAL STRUCTURE OF MUSIC
While ketanserin and LSD manipulations can help us to understand the role of 5HT 2A receptor signaling in TT, some curious effects were found in the current data that cannot be attributed simply to 5HT 2A receptor signaling. TT clusters in the right insula exhibited TT towards personally meaningful stimuli and also exhibited TT bias toward LSD when compared to placebo, but exhibited TT bias toward Ket + LSD when compared with LSD. TT in the mid-cingulate and posterior-cingulate cortex was also biased toward Ket + LSD when compared both to placebo and LSD. A TT cluster in the left superior temporal gyrus (Supplementary Table, Cluster 18; Brodmann area 41) exhibited greater TT during placebo compared to LSD, but did not display bias in TT between LSD and Ket + LSD conditions. This brain region is primarily within the transverse temporal gyrus/ Heschl's gyrus, and is thought to code basic auditory sensory information, including frequency content (tonotopically mapped from the cochlear nerve), intensity, and duration of sounds. Finally, the left calcarine gyrus, which exhibited TT bias towards personally meaningful stimuli, exhibited TT bias towards placebo compared to LSD, and exhibited TT bias toward LSD when compared with Ket + LSD. Given the complex pharmacology of LSD (with affinity for D2, D3/4, alpha 2, and a series of 5HT receptors), these findings suggest the potential involvement of receptors other than 5HT 2A in the neural response to music in the insula, the cingulate cortex, in visual processing regions, and in early auditory cortex regions. Due to the design of the study, the effect of ketanserin alone could not be investigated independently. Ketanserin alone has been shown to reduce neural response to fearful emotional stimuli in the medial orbitofrontal cortex and amygdala. Ketanserin was also shown to reduce the neural response in frontopolar cortex to negative outcomes associated with low-risk choices and associated with large missed rewards, and increase the response to low-risk negative outcomes in risk-taking individuals. While medial and lateral frontal regions have been previously implicated in TT, TT in these regions was not shown to be biased toward Ketanserin + LSD in the current findings. Thus, evidence would suggest that the findings observed in the current report in the Ketanserin + LSD condition are not likely due to effects expected from ketanserin alone.
CONCLUSION
The current study sheds light on the neuropharmacology and biology of music listening. LSD and ketanserin were used to demonstrate that 5HT 2A signaling may alter the coupling of activity in brain areas that support an overall musical experience with patterns of change in the tonal structure of that music. Alteration of the neural response to the tonal structure of music in a number of domain-general brain regions, primarily through 5HT 2A receptor signaling, could explain the increases in emotionality, connectedness, and meaningfulness that are observed in response to music after the administration of LSD and other serotonergic hallucinogens. This is particularly important since music listening is used to provide psychological support during research and therapy sessions with psychedelics. The current findings increase our knowledge about the neurochemical underpinnings of music listening-an experience with many health-relevant implications such as stress reduction and social bonding) that we seek and encounter without fail in our every-day lives.
Study Details
- Study Typeindividual
- Populationhumans
- Characteristicsplacebo controlleddouble blindrandomizedbrain measures
- Journal
- Compounds