LSD modulates music-induced imagery via changes in parahippocampal connectivity

Kaelen and colleagues situate their study in a long history of humans using psychedelics in combination with music and note that modern clinical interest in psychedelic-assisted psychotherapy has re-emerged. Earlier work suggests that music can act synergistically with psychedelics to enhance emotionality, mental imagery and access to personal memories, and prior neuroimaging studies implicate serotonin 2A receptor–mediated effects on high-level association cortices and medial temporal lobe (MTL) structures during psychedelic states. The parahippocampal cortex (PHC) is highlighted as a key MTL hub involved in scene imagination, autobiographical memory and music-evoked emotion, and is known to show altered functioning under psychedelics; direct stimulation of PHC can produce visual hallucinations and autobiographical imagery, motivating examination of PHC connectivity as a mechanistic substrate for music-evoked imagery under LSD. This study set out to test the hypothesis that LSD, in combination with music listening, modulates PHC functional connectivity and that such changes in connectivity would relate to enhancements in eyes-closed visual imagery and spontaneous autobiographical recollection. Using a balanced-order, placebo-controlled functional magnetic resonance imaging (fMRI) design, the investigators assessed seed-based PHC connectivity and applied Dynamic Causal Modelling (DCM) to probe the directionality of information flow between the PHC and visual cortex (VC), relating effective connectivity changes to in-scanner and post-scan subjective ratings of imagery and memory.

The study received ethical and regulatory approvals and was conducted under appropriate national licences. Twenty healthy volunteers (16 males, 4 females) were recruited and screened for physical and psychiatric health; key exclusion criteria included age under 21, personal history of diagnosed psychiatric illness, immediate family history of psychosis, absence of prior classic psychedelic experience, recent drug use, pregnancy and medically significant conditions. Screening included ECG, routine blood tests and urine drug/pregnancy tests. All participants provided written informed consent. Eligible volunteers attended two study days at a single imaging centre, separated by at least 14 days, receiving intravenously administered LSD on one day and placebo on the other in a balanced order; participants were kept blind to order but the researchers were not. The extracted text reports an LSD dose of 75 mg administered intravenously over two minutes followed by saline, with an acclimatisation period of about 60 minutes and BOLD fMRI commencing approximately 120 minutes post-dosing. Each fMRI session comprised three eyes-closed resting-state scans of seven minutes each; the music-listening scan was always the middle run. Two seven-minute music excerpts (A and B) from Robert Rich and Lisa Moskow were used, balanced for emotional potency and presented in counterbalanced order. After each run participants rated simple and complex visual imagery on continuous visual analogue scales (0–20); a post-scan questionnaire included the item "I saw scenes from my past". Imaging was performed on a 3T GE system. Structural and functional acquisition parameters are described, and preprocessing combined AFNI, FSL, Freesurfer and ANTS. Steps included removal of initial volumes, de-spiking, slice-timing and motion correction, brain extraction, registration to MNI space, motion scrubbing (FD threshold 0.4 after exclusions), spatial smoothing, band-pass filtering (0.01–0.08 Hz) and regression of nine nuisance regressors (six motion parameters and three anatomically defined signals from ventricles/CSF/white matter). Several participants were excluded: one for early termination due to anxiety, three for technical sound problems and four for excessive head motion, leaving twelve participants for group analyses; the text reports mean frame-wise displacement values and a significant between-condition difference in mean FD (placebo vs LSD). Seed-based functional connectivity analyses used a bilateral PHC region from the Harvard atlas to extract time series, modelled with a general linear model per run and then carried through fixed-effects within-subject contrasts (music versus no-music) for each drug condition; higher-level mixed-effects (FLAME1) assessed the interaction of music and LSD on PHC connectivity (cluster threshold z>2.3, p<0.05). For effective connectivity, DCM (SPM12b) was applied to the concatenated six runs (placebo no-music, placebo music, placebo no-music, LSD no-music, LSD music, LSD no-music). Time-series were taken from bilateral PHC and an occipital VC cluster identified by the PHC connectivity analysis. Models included bilinear between-node connections, stochastic driving inputs (resting-state fluctuations) and three modulatory inputs: main drug effect, main music effect and their interaction. A full model was inverted and post hoc Bayesian model optimisation identified the best-supported model; coupling parameters were then examined. Finally, Spearman correlations tested whether the DCM-derived interaction effect magnitude related to increases in eyes-closed imagery and to reports of seeing scenes from the past, with Bonferroni correction applied for multiple tests.

Participant data from twelve subjects (2 female, mean age 33±9 years, range 22–47) entered the group analyses; all had prior classic psychedelic experience and self-reported lifetime use of various substances (means and ranges provided in the extracted text). The extracted text reports that twenty participants were initially recruited but that several were excluded as detailed in Methods. Subjective effects: A paired t-test reported a significant increase in personal memory recollection under LSD for the post-scan item "I saw scenes from my past" (t = 1.9, df = 19, p = 0.04). The extracted text does not clearly reconcile the reported degrees of freedom (df = 19) with the final analysis sample size (n = 12). For in-scanner ratings, a two-way ANOVA on simple hallucinations showed a significant main effect of drug (F = 42.2, df = 18, p < 0.001) but no main effect of music or interaction. For complex hallucinations there was a significant drug effect (F = 24.7, df = 18, p < 0.001), a trend-level effect of music (F = 10.0, df = 18, p = 0.09) and no significant interaction. Drug effects on simple and complex ratings survived Bonferroni correction for two tests (adjusted alpha = 0.025). Seed-based functional connectivity: The PHC seed analysis revealed a positive interaction between music and LSD for the contrast [LSD (music vs no music) versus placebo (music vs no music)], showing increased coupling between the PHC and two principal clusters: bilateral visual cortex and the left inferior frontal gyrus. No decreases in PHC connectivity were observed for this contrast. A two-way repeated-measures ANOVA on z-scores extracted from the VC cluster showed no main effects of drug (F = 0.49, df = 11, p = 0.5) or music (F = 2.27, df = 11, p = 0.16) but a significant drug × music interaction (F = 17.04, df = 11, p = 0.002). Dynamic Causal Modelling: Post-hoc Bayesian model optimisation selected an optimal model with a Bayes factor 344 higher than the next best architecture, interpreted as strong evidence for that model. The winning model included main effects of drug and music modulating intrinsic connections of both PHC and VC, and an interaction effect specifically modulating the extrinsic connection from PHC to VC. Group-average posterior estimates (reported in the extracted text) were approximately -0.41±0.05 Hz for the drug effect on PHC, -0.42±0.06 Hz for music effect on PHC, -0.38±0.06 Hz for drug effect on VC, and -0.40±0.06 Hz for music effect on VC. The group average posterior for the interaction effect on PHC→VC was +0.02±0.04 Hz. Correlations with subjective effects: A significant positive correlation was found between the DCM-derived interaction effect (LSD×music) on PHC→VC effective connectivity and increases in in-scanner ratings of complex visual imagery (Spearman's ρ = 0.71, p = 0.01; Pearson r = 0.65, p = 0.03). A trend-level positive correlation emerged between the interaction effect and the post-scan item "I saw scenes from my past" (Spearman's ρ = 0.67, p = 0.02; Pearson r = 0.69, p = 0.01). These correlations were assessed with Bonferroni-adjusted alpha = 0.0167; the complex imagery correlation meets this corrected threshold while the autobiographical recollection result is reported as trend-level under correction.

Kaelen and colleagues interpret their findings as evidence that increased effective connectivity from the PHC to the visual cortex during music listening under LSD is associated with enhanced eyes-closed visual imagery, particularly complex imagery with autobiographical content. They link this to the established role of the PHC in encoding and retrieving context-related memory content and in music-evoked personal memories, and note prior reports that PHC–VC coupling is strengthened during construction of imagined scenery and during direct PHC stimulation that evokes complex visual hallucinations and memories. The authors propose a mechanistic account in which the observed increase in PHC→VC information flow corresponds to augmented top-down priors from high-level MTL regions being conferred onto lower-level visual processing, thereby allowing internally generated, high-level scene content to predominate in the absence of external input. They further discuss how psychedelics might facilitate this by dysregulating top-down inhibitory control over the PHC provided by posterior midline structures (retrosplenial cortex, posterior cingulate cortex) and medial prefrontal cortex—regions that express high levels of serotonin 2A receptors and have been shown to be affected by psychedelics. Reduced functional connectivity between PHC and retrosplenial cortex under psychedelics, reported elsewhere, is cited as consistent with this account. The discussion also situates the results in the historical and phenomenological literature on sound-evoked visual experiences under LSD and notes that while such reports resemble synaesthesia, they may differ in consistency and mechanism; nonetheless, the present findings provide a plausible neural mechanism for audio-visual imagery elicited by music under LSD. Regarding therapeutic implications, the authors suggest that music may deepen psychedelic experiences by evoking emotion and personally meaningful imagery, and that PHC–VC circuitry could be a target for understanding how music and psychedelics interact in therapeutic settings; however, they emphasise that further research is required to test therapeutic value directly. Several limitations are acknowledged. The authors note that no significant LSD×music interaction emerged on in-scanner subjective ratings despite connectivity evidence for such an interaction, and suggest this could reflect individual differences in subjective response, appraisal of the music, dislike of the chosen genre, or masking by MRI scanner noise. They recommend future studies that vary musical genre and emotional valence. Finally, the authors caution against extrapolating these findings to patient populations and call for further work to assess relevance for psychedelic-assisted psychotherapy.

The study reports a positive interaction between LSD and music on PHC functional and effective connectivity, specifically an increase in PHC→visual cortex coupling that correlated with enhanced eyes-closed visual imagery of a complex and autobiographical character. These results advance understanding of circuitry involved in visual imagery and propose a mechanism for how music and LSD may act synergistically to enhance internally generated imagery. The authors conclude that while these findings provide an initial mechanistic explanation relevant to music use in psychedelic-assisted psychotherapy, substantial further research is needed to determine clinical utility and to extend findings to patient populations.