Psilocybin-assisted neurofeedback for the improvement of executive functions: a randomized semi-naturalistic-lab feasibility study

Executive function (EF) deficits are a transdiagnostic problem across psychiatric disorders, impairing everyday functioning and limiting recovery and treatment efficacy. The fronto-cingulo-parietal network underpins EFs, with frontal-midline (fm) theta oscillations (a 4–8 Hz EEG rhythm) implicated as a marker and mediator of top‑down control; higher fm-theta power typically relates to better EF performance. Neurofeedback (NF) trains participants to self‑regulate specific neural signals such as fm-theta and has shown medium-sized effects on components of EF, notably working memory updating, but substantial inter-individual variability and a notable proportion of 'non-responders' limit clinical translation. Impaired neuroplasticity in psychiatric populations may underlie reduced NF responsiveness, and recent preclinical and clinical evidence suggests that serotonergic psychedelics, including psilocybin, act as potent psychoplastogens via 5‑HT2A and TrkB/BDNF-related signalling.

Enriquez-Geppert and colleagues conducted a randomised semi-naturalistic-lab feasibility study. Thirty-seven participants were recruited from an online microdosing workshop and randomised (Matlab rand()) to an experimental group or a passive control group; the passive control group was blind to the NF component to reduce motivational differences. Exclusion criteria included personal or family history of psychotic disorders, normal/corrected vision, and three weeks' drug abstinence prior to baseline. Pre- and post-assessments were conducted online in small Zoom groups, and NF took place in a sound-attenuated EEG lab at the University of Groningen. The experimental protocol comprised a one-week microdosing adjustment phase (participants self-selected an optimal dose within 0.5–2.0 g range of fresh Psilocybe mexicana truffles across three exploratory days), followed by a second week containing three microdose sessions paired with three psilocybin-assisted fm-theta NF sessions (Monday, Wednesday, Friday). Participants were advised to take the chosen microdose about two hours before each NF session. The passive control group had a two-week break during this period; both groups completed identical pre/post online testing schedules. Primary outcomes included two sets: objective experimental EF tasks delivered online (working memory updating via the reference-back task, set-shifting via a switching task, conflict monitoring via a picture-word interference task, and motor inhibition via a stop-signal task) and self-reported everyday EFs using four BRIEF-A subscales (Working Memory, Shift, Task Monitor, Inhibit). Participants also listed personal training priorities and rated pre/post functioning and satisfaction. Placebo and expectation proxies were assessed with standardised questionnaires (optimism, suggestibility, microdose expectancy, and NF/psilocybin‑NF attitudes). EEG NF used a 15-electrode cap; online processing drew fm-theta power from five electrodes (Fz, FC1, FC2, FCz, Cz), computed in 2 s segments shifted every 200 ms. Each laboratory NF session lasted ~70 minutes including six 5-minute NF blocks separated by breaks, with 5-minute resting baselines at start and end; feedback was a colour-saturation square updated every 200 ms indicating relative fm-theta power. Statistical analyses included repeated-measures ANOVAs for NF learning indices (session-to-session and block-wise changes), two-way RM ANOVAs for task RTs and accuracy, and one-sided t-tests for BRIEF-A subscales given directional hypotheses. Data preparation excluded participants with overall/subtask accuracy <60% and outlier SSRTs, resulting in analytic Ns of 31 (reference-back), 35 (switching), and 33 (stop-signal) for the respective tasks. A significance threshold of p ≤ .05 was used; where assumptions were marginally violated, nonparametric checks were performed and parametric results reported when consistent.

Thirty-seven participants (mean age 24 years, SD = 4.3; 20 female, 16 male, 1 other) completed the study. Most were students; 33 were not on regular medication. Twenty participants reported previous psychedelic use. The experimental group took an average fresh-truffle microdose of 1.14 g (SD = 0.52), consuming it on average 2.2 hours (SD = 0.35) before NF sessions. Participants reported high motivation and commitment and rated NF strategy effectiveness and session difficulty at moderate levels. Neurofeedback outcomes used two learning indices. The session-to-session learning index (mean relative theta amplitude across the six blocks per session, analysed across SESSION 1–3) approached significance: RM ANOVA F(2,34) = 3.185, p = .054, partial eta squared η2p = .158, indicating a moderate effect size after only three sessions. Within-session block-wise changes (start baseline, NF blocks 1–6, end baseline) were not significant after Greenhouse–Geisser correction: F(2.457, 41.765) = 1.303, p = .285, η2p = .071. Placebo and expectancy measures (optimism, suggestibility, microdose expectancy, and NF/psilocybin‑NF attitudes) showed no significant differences between experimental and passive control groups, with medium expectations for microdosing and moderately high expectations for NF. Objective online EF tasks produced no notable group-by-condition effects. For example, the reference-back RT and accuracy, task-switching RT and accuracy, and stop-signal RT analyses all returned non-significant GROUP, CONDITION, and interaction effects (selected statistics: reference-back RT GROUP F(1,29) = 1.473, p = .235; reference-back ACC CONDITION F(1,29) = 5.551, p = .025 driven by non-switch > switch improvement but not interacting with GROUP). The picture-word conflict monitoring task was excluded due to low performance (12/37 below the 60% accuracy threshold). In contrast, self-reported everyday EFs on BRIEF-A showed significant, sizeable improvements in the experimental group versus the passive control group across all four targeted subscales: Working Memory t(35) = 2.69, p = 0.0054, Cohen's d = 0.885; Shifting t(35) = 2.08, p = 0.0226, d = 0.683; Monitoring t(35) = 2.63, p = 0.0063, d = 0.865; Inhibition t(35) = 2.63, p = 0.0067, d = 0.857. The experimental group shifted Working Memory scores from a pre-intervention deficit range into a range interpreted as normal functioning. Analyses of participant-defined priority areas found significant pre–post improvements in Cognition (t(16) = 3.895, p = .001, d = .876), Presence (t(17) = 3.566, p = .002, d = 1.052), and Mood (t(12) = 2.520, p = .027, d = .966). Connection did not show significant changes. Participants reported medium-to-good satisfaction with achieved outcomes in these areas.

Enriquez-Geppert and colleagues interpret their findings as a feasibility demonstration for combining repeated psilocybin microdosing with fm-theta neurofeedback to potentially enhance neuroplasticity and NF learning. They highlight a modest, near-significant session-to-session increase in fm-theta self-regulation after three psilocybin-assisted NF sessions, but no reliable within-session changes across NF blocks. The investigators suggest that psilocybin's acute effects—even at low doses—could influence within-session self-regulation and that future work should determine optimal timing to preserve neuroplasticity benefits while minimising disruptive acute effects. While no improvements were detected on objective laboratory EF tasks after this shortened three-session protocol, the experimental group reported large, significant improvements in everyday EF functioning across Working Memory, Shifting, Monitoring and Inhibition domains. The authors propose that subjective, ecologically valid gains may emerge earlier or be more sensitive to this combined intervention than standard computerised measures, and that a greater number of NF sessions (the typical protocols extend to seven or eight sessions) might be needed to transfer objective task performance. The discussion notes distinctive features of the microdosing procedure used here: participants self-titrated an individual ‘‘optimal’’ microdose during an adjustment week rather than applying a fixed dose. The authors argue this approach mirrors real-world practice and may reduce inter-subject variability arising from metabolic and receptor differences. They also emphasise the semi-naturalistic design as a strength for ecological validity. Key limitations acknowledged by the investigators are the reduced number of NF sessions and the absence of an active control group, which constrain causal attribution. Although explicit expectancy and placebo proxy measures did not differ between groups, the authors concede that implicit placebo effects cannot be fully excluded. They recommend future studies include an active control condition, larger samples, and systematic exploration of dosing, timing of drug administration relative to NF, and longer NF schedules. Overall, the authors conclude that psilocybin-assisted NF is practicable and shows promise as a transdisciplinary, potentially transdiagnostic approach to augment NF by leveraging transient increases in neuroplasticity. They call for replication and optimisation studies to establish efficacy, dosing/timing parameters, and the durability of effects.