Psilocybin therapy increases cognitive and neural flexibility in patients with major depressive disorder

The study investigates whether psilocybin therapy produces enduring improvements in cognitive flexibility and related neural dynamics in patients with major depressive disorder (MDD). Cognitive flexibility—the ability to adaptively switch between cognitive operations—is often impaired in MDD and may represent a transdiagnostic mechanism relevant to treatment. While psilocybin-assisted therapy has demonstrated rapid and durable antidepressant effects, its mechanistic underpinnings remain unclear. Prior work implicates the anterior cingulate cortex (ACC) and its connectivity, including with the posterior cingulate cortex (PCC), in cognitive flexibility and in the acute effects of psychedelics. The authors hypothesized that psilocybin therapy would improve cognitive flexibility, modulate ACC-related functional connectivity dynamics (dFC), and alter neurochemical markers (glutamate and N-acetylaspartate, NAA) in regions supporting flexibility (ACC and hippocampus). They further tested whether baseline neural flexibility predicts post-treatment changes in cognition and mood.

The introduction reviews mixed findings on acute serotonergic modulation of cognitive flexibility: 5-HT2A receptor blockade can both impair and enhance flexibility in animal models, whereas 5-HT2A agonists (psychedelics) have been reported to acutely impair flexibility in humans and variably affect it in animals. The 5-HT2C receptor’s role is complex and may mitigate impairments from 5-HT2A activation. The ACC is central to cognitive flexibility, with human and animal data linking ACC activity and connectivity to adaptive rule switching. Psychedelics acutely modulate ACC activity and network connectivity, especially within salience and default mode networks, and may induce neural plasticity. Neural flexibility (signal variability/entropy) is reduced in MDD, particularly in ACC/PCC, and psychedelics acutely increase measures of brain signal variance/entropy, although durability in patients is unclear. Prior small studies in healthy adults did not observe enduring ACC-related changes one to four weeks after psilocybin. Antidepressants including ketamine, SSRIs, and rTMS can improve cognitive flexibility, potentially via hippocampal plasticity, but many patients remain treatment-refractory, motivating investigation of psychedelic therapy’s effects on flexibility.

Design: Open-label trial with immediate and delayed treatment groups in adults with MDD. N=24 (8 males), age 24–59 (M=39.83, SD=12.23). The delayed group served as a within-subject control for non-treatment-related changes during an 8-week delay. Participants: Recruited via ads/referrals; screened by surveys, phone, in-person medical/psychiatric evals. Key exclusions: current antidepressants, substantial lifetime or recent use of ketamine/classic psychedelics, significant medical conditions, personal/family psychotic disorder history, substance use disorder, or other major psychiatric disorders. Intervention: Two psilocybin sessions ~1 week apart after preparatory psychotherapy. Doses administered orally in opaque capsules: first session moderately high dose (reported as 200 mg/70 kg in text) and second session high dose (30 mg/70 kg). Sessions in a supportive setting with eyeshades and music; integration meetings followed each session. Assessments and timeline: Immediate group completed baseline clinical/neuroimaging, then psilocybin. Delayed group completed clinical/cognitive assessments at screening and at 5 and 8 weeks during delay, then baseline neuroimaging just prior to therapy, followed by the same intervention. Clinical and cognitive measures: Depression measured with the 17-item GRID-HAMD by blinded clinicians at screening, 1 week, and 4 weeks post-treatment; also during delay (weeks 5 and 8). Cognitive flexibility measured with the Penn Conditional Exclusion Test (PCET), a set-shifting task assessing perseverative errors (lower errors indicate better flexibility). Other cognitive tasks (response inhibition, selective attention, abstract reasoning) are reported in Supplementary Information. Some participants had missing/refused data; final PCET N≈22 with one additional missing 4-week post-treatment data point. Imaging: 7T MRI (Philips Achieva, 32-channel head coil) at ~4 weeks before first psilocybin session (baseline) and 1 week after second session. Acquisitions: T1 MPRAGE; resting-state BOLD fMRI (eyes open); 1H-MRS (STEAM, short TE) voxels in ACC (30×20×20 mm3), left and right hippocampus (35×15×15 mm3). Some scans excluded for incomplete sessions, missing resting state, abnormal findings, or poor SNR, yielding approximately N=20 for rs-fMRI and ACC MRS, N=18–19 for hippocampal MRS. Preprocessing: fMRI preprocessing in SPM12: realignment, co-registration, normalization to EPI template; nuisance regression including 24 motion parameters; temporal denoising. Due to 7T susceptibility-induced artifacts, only atlas nodes with acceptable SNR (approximately 89 nodes) were included for connectivity analyses. Connectivity metrics: Static functional connectivity (sFC): Fisher z-transformed Pearson correlations between node time series (3916 edges). Dynamic functional connectivity (dFC): computed using dynamic conditional correlations (DCC) for each edge, with dFC defined as the variance of the correlation time series. Analyses focused on ACC and PCC nodes due to their roles in cognitive flexibility and psychedelic effects. MRS quantification: Spectra processed with in-house software (FID-A based) and quantified with LCModel. Focus on glutamate (Glu) and N-acetylaspartate (NAA) due to reliability and relevance to depression and psychedelics. Quality control included visual inspection and exclusion based on SNR and fitting criteria. Statistical analyses: Repeated-measures ANOVAs and paired t-tests comparing baseline to post-psilocybin (1 and 4 weeks) for clinical, cognitive, MRS, and connectivity outcomes. The delayed group’s pre-delay versus baseline comparisons and Bayes factors assessed practice/expectancy effects. Exploratory Pearson correlations examined associations between changes in imaging/metabolites and changes in GRID-HAMD or PCET errors. Predictive modeling: Connectome-based predictive modeling (CPM) using baseline sFC and dFC features to predict changes in depression and cognitive flexibility post-treatment, and to predict baseline symptom/cognitive levels. Feature selection thresholds varied (p<0.01–0.05); models evaluated with cross-validation (e.g., leave-one-out), reporting correlations between observed and predicted behavior.

- Depression: Robust decreases in GRID-HAMD scores from baseline to 1 and 4 weeks post-psilocybin (time effect F(2,46)=69.695, p<0.001, η²=0.715). In the delayed group, no change during the pre-treatment delay (F(2,20)=0.25, p>0.25; Bayes factor supporting null). - Cognitive flexibility: PCET perseverative errors decreased at 1 week post-psilocybin and remained improved at 4 weeks (time effect F(2,40)=10.509, p<0.001, η²=0.335). Delayed group showed no practice-related change prior to treatment (Bayes factor supported null). Changes in depression were not correlated with changes in PCET errors (all |r|<0.25, p>0.25). - MRS neurochemistry: At 1 week post-therapy, ACC glutamate decreased (t(19)=2.54, p=0.020, d=0.57; 95% CI [0.11, 1.10]) and ACC NAA decreased (t(19)=3.95, p=0.007, d=0.68). No significant changes in Glu or NAA in left/right hippocampus (ps>0.25). Changes in ACC Glu/NAA did not correlate with changes in GRID-HAMD or PCET errors (all |r|<0.36, ps>0.13). - Functional connectivity: Across the brain, sFC tended to decrease and dFC tended to increase post-therapy. For ACC-PCC, sFC change was not significant (t(19)=0.67, p>0.25; 95% CI [-0.12, 0.23]). dFC between ACC and PCC increased at 1 week (t(19)=2.87, p=0.010, d=0.424; 95% CI [0.02, 0.21]). Exploratory associations suggested that greater increases in ACC-PCC dFC were linked to less improvement in PCET errors (consistent with the abstract’s summary), while ACC-PCC connectivity changes were not associated with GRID-HAMD changes (all |r|<0.10, p>0.25). - Predictive modeling: Baseline ACC-centered dFC features predicted post-treatment improvements in cognitive flexibility. In models, greater baseline dFC from ACC predicted better baseline cognitive flexibility but smaller improvements after therapy, indicating a ceiling effect or nonlinearity. dFC features were more frequently selected than sFC features in predictive models.

The findings demonstrate that psilocybin therapy produces enduring improvements in cognitive flexibility in MDD, alongside robust antidepressant effects, yet the two changes were not correlated, suggesting partially distinct mechanisms. Neurochemically, reductions in ACC glutamate and NAA one week post-treatment indicate regionally selective alterations consistent with modulation of excitatory neurotransmission and neuronal integrity/energy metabolism. Functionally, increased neural flexibility (dFC) between ACC and PCC at one week reflects enhanced dynamics within salience-default mode circuits implicated in cognitive control and self-referential processing. However, larger increases in ACC-PCC dFC were associated with less improvement in cognitive flexibility, and higher baseline dFC predicted less post-therapy improvement despite better baseline performance. This pattern suggests a nonlinear relationship wherein insufficient baseline neural flexibility may limit cognitive flexibility, while excessive increases in neural dynamics may not yield proportional cognitive benefits. Together, results support the view that psychedelic therapy may open a window of neural and psychological plasticity during which targeted psychotherapeutic processes could be optimized to translate neural dynamics into cognitive gains.

This study provides convergent behavioral, neurochemical, and functional connectivity evidence that psilocybin therapy enhances cognitive and neural flexibility in MDD. Improvements in cognitive flexibility persisted for at least four weeks and were accompanied by decreased ACC glutamate and NAA and increased ACC-PCC dFC at one week. Connectome-based models showed that baseline neural flexibility predicts subsequent cognitive change, highlighting individual variability and potential patient stratification markers. Future research should employ randomized, placebo- or active-controlled designs, larger samples, and longer follow-up to determine durability, causality, and clinical relevance; integrate task-based fMRI and ecological cognitive measures; and test whether tailoring psychotherapy to the period of enhanced neural dynamics maximizes cognitive and clinical outcomes. Identifying optimal ranges of neural flexibility that confer psychological benefit will be key for precision psychedelic-assisted interventions.

- Open-label, non–placebo-controlled design limits causal inference and leaves room for expectancy effects; though delayed-group analyses reduce concerns about practice effects on PCET. - Small sample size with some missing data and imaging exclusions due to 7T artifacts/SNR constraints reduces generalizability and power. - Dose reporting in text appears inconsistent; exact dosing details may affect interpretation and replication. - Neuroimaging focused on nodes with acceptable SNR and on ACC/PCC, potentially overlooking broader network effects. - Correlational analyses were exploratory; some reported associations were modest and require replication. - Lack of direct mechanistic measures of synaptic plasticity in humans; MRS markers are indirect proxies.