Investigating Emotional Reactivity in Experienced Users of Psychedelics: a cross-sectional fMRI study

Classic psychedelics such as psilocybin, LSD and DMT are known to alter perception, mood and cognition, and growing experimental and clinical evidence suggests they can produce sustained changes in emotional reactivity that may underlie improvements in well-being, depression and anxiety. Previous controlled studies have reported acute and subacute reductions in responses to negative emotional stimuli, including decreased amygdala reactivity, and alterations in large-scale networks such as reduced Default Mode Network (DMN) and Salience Network (SN) activity. However, most of these data derive from tightly controlled therapeutic or laboratory contexts with screening, standardised dosing and psychological support, leaving open the question of whether naturalistic, less-controlled psychedelic use produces similar persistent neural and behavioural effects. Orłowski and colleagues set out to address this gap by comparing experienced naturalistic psychedelic users to matched non-users on behavioural and fMRI measures during an emotional face classification task. The preregistered, cross-sectional study tested whether regular users (operationalised as ≥10 lifetime psychedelic experiences) differ from non-users in neural reactivity to negative versus positive facial expressions, with the primary hypothesis that users would show attenuated neural responses to negatively valenced stimuli, reflecting persistent modulation of emotion-processing circuits.

The study was preregistered and approved by the institutional ethics committee. Recruitment began with an online Qualtrics survey completed by 2,573 respondents reached via university social media and harm-reduction organisations. The survey collected sociodemographics, screening measures for alcohol (AUDIT-C) and cannabis use (CUDIT-R items), lifetime and past-year use of classic psychedelics and other psychoactive substances, meditation practice, psychiatric/neurological diagnoses and medication use, and the Perth Emotional Reactivity Scale-Short Form (PERS-S). Respondents consented to be contacted for the fMRI phase. Eligibility criteria excluded individuals with current or past psychiatric or neurological diagnoses, substance dependence history, current psychoactive medication, MRI contraindications, and left-handedness. The Users group required ≥10 lifetime psychedelic experiences; the Non-Users group comprised participants who reported never having used classic psychedelics. Groups were matched on age, sex, education, place of residence, alcohol and cannabis use scores, lifetime/past-year use of other drug classes, and meditation history; matching was checked with appropriate statistical tests. Participants were instructed to abstain from psychedelics for at least 30 days before scanning. The preregistered target sample for the fMRI experiment was 70; 69 sessions were run (34 Users, 35 Non-Users) and data from two participants were excluded (one discontinued due to scanner distress, one for non-compliance), yielding a final analysed sample of 33 Users and 34 Non-Users. Stimuli comprised 104 colour photographs from the Warsaw Set of Emotional Facial Expression Pictures: 13 female and 13 male models each showing angry, fearful, happy and neutral expressions. The task consisted of 208 trials (52 per emotion) presented in two runs; on each trial a face was shown for 300 ms after a jittered fixation, followed by a 2,000 ms response screen where participants selected one of four emotion labels via a button pad. Button-emotion mappings were counterbalanced across participants. MRI data were acquired on a 3T Siemens Magnetom Prisma with a 64-channel head coil. Structural T1-weighted images and task-based multiband T2*-weighted EPI runs were collected; reverse-phase EPI images were acquired for susceptibility distortion correction. Functional preprocessing used fMRIPrep including motion correction, slice timing, co-registration, normalization to MNI space and extraction of confound regressors; further first-level processing in FSL FEAT included brain extraction, 4 mm smoothing and high-pass filtering. First-level GLMs modelled four stimulus regressors (angry, fearful, happy, neutral) plus regressors for correct, error and no-response trials; motion and mean CSF signal were included as nuisance regressors. Run-level contrasts were combined with fixed-effects for each participant. Two group-level approaches were used. An exploratory whole-brain analysis (not preregistered) was performed with AFNI 3dLME testing Emotional Condition (within-subject) × Group (between-subject) effects; key contrasts subtracted Neutral from each emotion and used a voxel Z-threshold of 2.5 with a minimum cluster size of 25 voxels. The preregistered ROI analysis used Harvard-Oxford atlas masks thresholded at 50% and included ACC, frontal medial cortex, frontal pole, middle frontal gyrus, bilateral amygdala, fusiform gyrus and parahippocampal gyrus; mean beta values were extracted and analysed in mixed-design ANOVAs with Group and Emotional condition factors, applying Greenhouse–Geisser correction and post hoc emmeans comparisons with Bonferroni correction when applicable. Behavioural reaction times were analysed with linear mixed-effects models and accuracy with binomial GLMMs, both modelling Group, Emotional Condition and their interaction plus a random intercept for participant; post hoc tests used Holm correction.

Behavioural analyses showed a robust main effect of Emotional Condition on reaction times (F(3,13356) = 526.76, p < 0.001); angry and fearful faces elicited the longest reaction times (angry M = 1.31 s, SD = 0.37; fearful M = 1.27 s, SD = 0.35) whereas happy faces were fastest (M = 1.08 s, SD = 0.28). There was no main effect of Group on reaction time, but a significant Group × Emotional Condition interaction emerged (F(3,13356) = 45.31, p < 0.001). Post hoc tests indicated that Users responded faster than Non-Users specifically for angry faces (Users M = 1.26 s, SD = 0.37; Non-Users M = 1.36 s, SD = 0.37; estimate = 0.10, SE = 0.04, p = 0.008); no other emotion showed significant between-group RT differences. Accuracy analyses likewise revealed a main effect of Emotional Condition (χ²(3) = 109.64, p < 0.001), with highest accuracy for happy faces (M = 98.3%, SD = 12.7%) and lowest for angry faces (M = 92.1%, SD = 27.1%). A main effect of Group indicated higher overall accuracy in Users (Users M = 95.9%, SD = 19.9%; Non-Users M = 94.8%, SD = 22.2%; χ²(1) = 4.47, p = 0.034). A Group × Emotional Condition interaction (χ²(3) = 9.79, p = 0.020) was driven by better performance by Users for angry faces (Users M = 93.8%, SD = 24.2%; Non-Users M = 90.4%, SD = 29.5%; estimate = -0.44, SE = 0.21, p = 0.03) and no between-group differences for fearful, happy or neutral faces. Whole-brain contrasts (emotion minus neutral) identified several group differences. For angry faces, Users showed lower activation than Non-Users in left insula, left supplementary motor area and bilateral inferior frontal gyri, while showing higher activation in the right inferior parietal lobule. Fearful faces elicited greater activation in the precuneus for Users relative to Non-Users. Happy faces produced a pattern of greater activation in Users across parietal and sensorimotor regions—including bilateral inferior parietal lobule/supramarginal gyrus, left superior parietal lobule, left cerebellum, right supplementary motor area and right superior parietal lobule—whereas the right inferior frontal gyrus showed lower happy-related activation in Users. The ROI analysis highlighted significant Emotion × Group interactions in two DMN-related regions. In the frontal medial cortex (mean betas predominantly negative), there was a main effect of Emotional condition (F = 6.18, p < 0.001) and an Emotion × Group interaction (F = 2.78, p = 0.048): Non-Users showed greater deactivation for Angry and Fearful versus Happy faces, whereas Users showed no significant differentiation between emotions; no significant between-group contrasts were observed within individual emotions. In the parahippocampal gyrus there was a main effect of Emotional condition (F = 8.38, p < 0.001) and an Emotion × Group interaction (F = 3.54, p = 0.017), with the Non-User group showing stronger deactivation for Angry and Fearful versus Happy (and Angry versus Neutral), while the User group again showed no significant emotion-specific differences. No significant group differences were detected in amygdala activation in the extracted text. The authors note fuller ROI and whole-brain statistics are reported in tables and supplementary materials.

The investigators interpret the behavioural and neuroimaging pattern as indicating attenuated processing of threat-related information in experienced naturalistic psychedelic users. Faster and more accurate classification of angry faces by Users is taken to reflect reduced interference from threat-related cues during task performance. Correspondingly, whole-brain results showing lower activation to angry faces in insula and salience-related regions are presented as consistent with prior controlled studies reporting reduced processing of negative emotional stimuli after psychedelic administration. At the same time, Users exhibited increased precuneus activation to fearful faces and enhanced parietal and sensorimotor responses to happy faces. The authors relate the latter pattern to prior reports of elevated neural sensitivity to positive emotions following psychedelics and to questionnaire findings of greater positive emotional reactivity in their earlier work; they suggest that fMRI may be more sensitive than EEG to broader network-level responses to positive valence. Importantly, ROI findings in the frontal medial cortex and parahippocampal gyrus—both DMN hubs—showed reduced differentiation between emotional categories in Users, which the authors propose could reflect a persistent alteration of DMN function associated with greater emotional resilience and reduced negative affect. The discussion acknowledges discrepancies with prior literature, notably the absence of group differences in amygdala reactivity, which several clinical studies have reported acutely and subacutely. Possible explanations offered include contextual differences between naturalistic and clinical use (dose, setting and support), the ≥30-day abstinence requirement in this sample which may have allowed acute/subacute effects to dissipate, and reduced sensitivity of the event-related fMRI design for small subcortical structures compared with block designs. The authors candidly note key limitations: the cross-sectional design precludes causal inference; residual confounding and expectancy effects cannot be ruled out despite matching and statistical controls; self-selection bias may have skewed the Users sample toward those with favourable experiences; exclusion of individuals on psychiatric medication limits generalisability to clinical populations; and design choices may have affected power to detect subtle effects. Overall, the authors conclude that their results point to a nuanced, region- and emotion-specific reorganisation of emotion-processing networks associated with naturalistic psychedelic use, and they position these findings as complementary to controlled clinical work while calling for longitudinal and mechanistically focused studies to clarify causality and temporal dynamics.

The study reports that habitual naturalistic users of classic psychedelics show selective behavioural advantages in recognising angry faces and regionally specific neural differences during emotional face perception that overlap partly with patterns reported in controlled clinical studies. Because of its cross-sectional design and sample selection constraints, the study does not establish causality; nevertheless, the authors conclude that the findings contribute to an emerging picture of persistent, nuanced neurofunctional alterations in emotional processing associated with long-term psychedelic use and underscore the need for longitudinal and experimentally controlled follow-up work.