Role of the 5-HT2A Receptor in Self- and Other-Initiated Social Interaction in Lysergic Acid Diethylamide-Induced States: A Pharmacological fMRI Study

The paper frames the coherent sense of self—particularly the “minimal self” comprising ownership, agency and embodiment—as fundamental to conscious experience and closely tied to social cognition. Disturbances of basic self-experience occur across psychiatric disorders and can impair social interaction. The authors argue that pharmacological neuroimaging offers a way to probe neurotransmitter systems that underlie self-processing and social cognition by producing transient, experimentally controlled alterations of self-experience. Katrin and colleagues set out to test how stimulation of the serotonin 2A receptor (5-HT2A R) by lysergic acid diethylamide (LSD) affects self- and other-initiated social interaction. Using a double-blind, placebo-controlled, crossover design with and without pretreatment by the selective 5-HT2A antagonist ketanserin, they employed a real-time gaze-based interaction task during functional MRI to probe self-versus other-initiated interaction and joint versus non-joint attention. The central hypotheses were that LSD would loosen self-boundaries and reduce differentiation between self and other during social interaction, would alter joint-attention processing in regions implicated in self- and social cognition (for example the precuneus/posterior cingulate cortex and medial prefrontal cortex), and that ketanserin would at least partially block these effects.

Healthy volunteers were recruited and screened for medical and psychiatric suitability. After exclusions for technical failure, 24 participants (18 males, 6 females; mean age 25.42 years, SD 3.69, range 20–34) completed the study. Standard interviews and questionnaires were used to exclude present or past psychiatric disorders and first-degree family history of major psychiatric illness. Participants abstained from drugs and alcohol before testing; urine screens confirmed compliance. Adverse events were minor and transient. A double-blind, randomised, counterbalanced, three-condition crossover design was used. On separate test days 2 weeks apart subjects received (1) placebo pretreatment + placebo (Pla), (2) placebo pretreatment + LSD (100 µg, p.o.; LSD), or (3) ketanserin (40 mg, p.o.) pretreatment + LSD (Ket+LSD). Pretreatment occurred 60 minutes before LSD/placebo; the interactive task was performed 310 minutes after treatment. Subjective effects were measured retrospectively with the 5D-ASC (a 94-item questionnaire scored on validated subscales) administered 720 minutes after treatment; mood was assessed with the Positive and Negative Affect Schedule (PANAS) before pretreatment and 720 minutes after treatment with retrospective ratings for peak effects. The experimental task probed social interaction using an anthropomorphic virtual avatar in real time with eye-tracking. Trials required participants to establish eye contact and either lead (self-initiated) or follow (other-initiated) gaze to one of two objects, generating joint-attention (JA) or non-joint-attention (NJA) outcomes. Five conditions were presented (self-joint, self-nonjoint, other-joint, other-nonjoint, baseline) in 5 s blocks with 2 s instruction screens; each condition was repeated 20 times for 100 trials, presented pseudorandomly in two runs. Two behavioural latency measures were extracted: time to establish eye contact and time to establish joint attention. Errors were trials without successful fixation. Gaze was recorded with an MRI-compatible eye-tracking system (60 Hz); fixations and saccades were detected online and used to make the avatar's gaze contingent. MRI data were acquired at 3 T with an EPI sequence (TR 2500 ms, TE 27 ms, 3 mm slices, 45 slices) and a high-resolution T1. Preprocessing used SPM12: slice timing, realignment, normalization to MNI space and 8 mm smoothing. First-level models treated the five experimental conditions as 5 s blocks convolved with a canonical haemodynamic response; motion regressors and a high-pass filter were included. Contrasts of interest were self>other, JA>NJA, self-initiated JA>self-initiated NJA, and other-initiated JA>other-initiated NJA. Second-level random-effects paired t tests compared drug sessions. Analyses used small-volume correction (SVC) focused on four a priori ROIs (left precuneus/PCC, left medial prefrontal cortex, right middle temporal gyrus, left middle temporal gyrus) defined as 10 mm spheres around previously reported MNI coordinates; peak-level familywise error (FWE) correction at p<0.05 was applied within these ROIs. Behavioural and subjective measures were analysed with repeated-measures ANOVAs (treatment as within-subject factor, with additional within factors where appropriate). Significant effects were followed by Bonferroni-corrected pairwise comparisons. Correlational analyses tested relationships between LSD-induced BOLD changes (first eigenvariates extracted from the ROIs) and 5D-ASC subscales particularly relevant to self-experience ("changed meaning of percepts" and "experience of unity"); Pearson correlations were reported and Spearman used for replication, and subjective scores were also used as covariates in SPM analyses.

Subjective experience: Repeated-measures ANOVA on 5D-ASC scores showed strong treatment and scale effects and a treatment×scale interaction (treatment: F(2,46)=77.40, p<0.001; scale: F(10,230)=16.65, p<0.001; interaction: F(20,460)=15.06, p<0.001). Bonferroni-corrected comparisons indicated increased ratings across almost all 5D-ASC scales in the LSD condition relative to both placebo and Ket+LSD (all p<0.05) except for spiritual experience and anxiety (p>0.20). Placebo and Ket+LSD did not differ on any scale (p>0.90). PANAS analyses showed a higher positive than negative affect overall, a treatment effect (F(2,46)=6.10, p<0.01) and interactions with time; after drug administration positive affect was higher in the LSD condition than Pla and Ket+LSD (p<0.05), and negative affect was higher in LSD than Pla (p<0.05). fMRI — main behavioural contrasts: For the self>other contrast, the Pla versus LSD comparison showed greater BOLD signal in the left posterior cingulate cortex (PCC) in the Pla condition compared with LSD (peak x=−3, y=−46, z=28; cluster size 40; T=3.50; p<0.05 FWE after SVC). The self-joint>self-nonjoint contrast revealed greater BOLD in the medial prefrontal cortex for Pla versus LSD (x=3, y=59, z=13; cluster size 19; T=3.66; p<0.05 FWE after SVC). No contrasts produced greater activation for LSD>Pla. Ketanserin pretreatment blocked LSD effects: Comparing Ket+LSD versus LSD, the self>other contrast showed higher BOLD after Ket+LSD in left PCC (x=−3, y=−46, z=28; cluster size 30; T=3.31) and right middle temporal gyrus (x=51, y=−55, z=22; cluster size 42; T=3.50) (p<0.05 FWE after SVC). For other-joint>other-nonjoint, Ket+LSD produced greater BOLD in left PCC (x=−9, y=−49, z=31; cluster size 67; T=3.94) and left middle temporal gyrus (x=−48, y=−67, z=28; cluster size 7; T=3.35) (p<0.05 FWE after SVC). No suprathreshold voxels were found for LSD>Ket+LSD. Ket+LSD did not differ significantly from placebo in any contrast. Behavioural performance: The number of error trials did not differ between conditions (F(2,46)=2.36, p>0.1); means (SD) were Pla 11.71 (11.80), LSD 16.89 (13.31), Ket+LSD 10.29 (12.39). Latency to establish eye contact did not differ (F(2,44)=1.30, p>0.2; Pla 0.44 s (0.17), LSD 0.50 s (0.18), Ket+LSD 0.43 s (0.14)). Latency to establish joint attention showed a significant treatment effect (F(2,44)=6.90, p<0.01): LSD produced longer latency (mean 1.82 s, SD 0.27) than Pla (1.65 s, SD 0.17) and Ket+LSD (1.67 s, SD 0.17) (all pairwise p<0.05). Correlations with subjective experience: The LSD-induced decrease in PCC BOLD (self>other contrast) correlated positively with the 5D-ASC scale "changed meaning of percepts": Pla versus LSD change r=0.44, p<0.03; Ket+LSD versus LSD change r=0.47, p<0.02. Spearman replication reproduced these results except the Ket+LSD–LSD relationship did not reach significance in one test. Mood changes did not correlate significantly with BOLD changes (all p>0.1). Using 5D-ASC scores as covariates in SPM produced FWE-corrected SVC correlations between "changed meaning of percepts" and LSD-induced PCC decreases (peak voxels reported within the ROI, p<0.05 FWE).

Katrin and colleagues interpret their results as showing that LSD reduces neural differentiation between self- and other-initiated social interaction, particularly in the posterior cingulate cortex and angular gyrus (temporal cortex). These regions are linked to self-referential processing, the default-mode network and aspects of autobiographical and experiential self-reflection; the authors relate their findings to prior reports of retrosplenial and angular gyrus changes under psychedelics and to the broader literature on PCC involvement in self-related processes. The observed LSD-induced reduction in PCC response during self-versus-other processing correlated with the 5D-ASC subscale "changed meaning of percepts", which the authors argue reflects altered relationships between observer and environment and supports the notion that altered self-experience underlies altered social processing. Effects on joint attention were subtler than effects on self-processing but included increased latency to establish joint attention under LSD and reduced medial prefrontal cortical response for self-joint>self-nonjoint contrasts under LSD. The medial prefrontal cortex is discussed in the context of joint attention, mentalising and monitoring one's own emotions during interaction. Importantly, pretreatment with ketanserin normalised the LSD-induced reductions in self–other differentiation and blocked the increased latency for joint attention; Ket+LSD did not differ from placebo. From these findings the authors conclude that 5-HT2A receptor activation is a key mechanism mediating LSD's effects on self-processing and certain aspects of social cognition. The authors acknowledge limitations: other receptor systems engaged by LSD (for example dopamine receptors) were not tested, so contributions from non-5-HT2A targets cannot be excluded. They note the unequal gender distribution (18 males, 6 females) as a constraint on generalisability. The absence of between-condition differences in error rate argues against gross attentional confounds, but the authors caution that LSD effects on joint attention were more modest than on self-processing. Finally, they suggest clinical implications suggested by their data: 5-HT2A antagonists might be of interest in conditions featuring incoherent self-experience such as schizophrenia, whereas 5-HT2A agonists could have potential in disorders characterised by excessive self-focus such as depression, and they recommend future research to explore receptor-specific interventions for social deficits in patient populations.