In a coordinated five‑laboratory study of ~200 mice, psilocybin (2 mg/kg IP) produced robust, replicable acute behavioural effects — increased anxiety- and avoidance-related behaviour and reduced fear expression — but did not reproducibly produce persistent (24 h) anxiolytic, antidepressant, fear‑extinction or pro‑social effects. The multi‑institutional design therefore both delineates the reliable acute actions of psilocybin and provides a model for improving reproducibility in psychedelic research.
Studies reporting novel therapeutic effects of psychedelic drugs are rapidly emerging. However, the reproducibility and reliability of these findings could remain uncertain for years. Here, we implemented a multi-institutional collaborative approach to define the robust and replicable effects of the psychedelic drug psilocybin on mouse behavior. Five laboratories performed the same experiments to test the acute and persistent effects of psilocybin (2 mg/kg, IP) on various behaviors that psychedelics have been proposed to affect, including anxiety-related approach-avoidance, exploration, sociability, depression-related behaviors, fear extinction, and social reward learning. Through this coordinated approach, we found that psilocybin had several robust and replicable acute effects on mouse behavior, including increased anxiety- and avoidance-related behaviors and decreased fear expression. Surprisingly, however, we found that psilocybin did not have replicable effects 24 hours post psilocybin administration on reducing anxiety- and depression-like behaviors or facilitating fear extinction learning. Additionally, we were unable to observe psilocybin-induced alterations in social preference or social reward learning. Overall, our comprehensive characterization of psilocybin’s acute and persistent behavioral effects using ∼200 total male and female mice per experiment spread across five independent labs demonstrates with unique certainty several acute drug effects and suggests that psilocybin’s persistent effects in mice may be more modest and inconsistent than previously suggested. We believe this unusual multi-laboratory, highly coordinated research effort serves as a model for facilitating the generation of replicable results and consequently will reduce efforts based on unreliable and spurious results.
Papers cited by this study that are also in Blossom
Carhart-Harris, R. L., Bolstridge, M., Rucker, J. et al. · Lancet Psychiatry (2016)
Raison, C. L., Sanacora, G., Woolley, J. D. et al. · JAMA (2023)
Goodwin, G. M., Aaronson, S. T., Alvarez, O. et al. · New England Journal of Medicine (2022)
Ross, S., Bossis, A. P., Guss, J. et al. · Journal of Psychopharmacology (2016)
After decades of limited investigation, interest in psychedelic drugs has resurged because single or few doses of compounds such as psilocybin, MDMA, LSD and ibogaine appear to produce rapid and sometimes enduring therapeutic effects across several neuropsychiatric conditions. Earlier preclinical and clinical reports have been inconsistent: some studies report antidepressant- or anxiolytic-like effects in rodents and humans, while others find null or contradictory results. These inconsistencies raise concerns about robustness and replicability of reported psychedelic effects and about how well results from single laboratories generalise. To address that gap, Lu and colleagues implemented a coordinated, multi-institutional approach to identify behavioral effects of psilocybin in mice that are robust and replicable across laboratories. The consortium aligned on strain (C57BL/6J), age range, sex inclusion, dose (2 mg/kg IP), core experimental protocols and timepoints, and tested a battery of assays probing acute (minutes after dosing) and persistent (around 24 hours after dosing) effects on head twitch response, anxiety-related approach–avoidance and exploration, sociability, depression-related passive coping, cued fear expression and extinction, and social reward learning. The study aimed to determine which psilocybin effects are reliably observed across independent labs and which are small, variable, or absent under these standardised conditions.
The study used a multi-site coordinated design in which five independent laboratories performed the same set of behavioural experiments in male and female C57BL/6J mice. The extracted text does not clearly report the total sample size per experiment in this Methods section. Psilocybin powder obtained from a national source was prepared at 0.2 mg/mL in 0.9% saline and administered intraperitoneally at 10 mL/kg to achieve a dose of 2 mg/kg for all experiments. Stability and concentration of the drug preparation were checked using liquid chromatography–high resolution mass spectrometry (LC-HRMS) and a series of stability conditions. Behavioural assays included: head twitch response (HTR) monitored visually for 20 minutes post-injection as an index of hallucinogenic-like activity; open field test (OFT) for locomotion and anxiety-related centre avoidance; elevated plus maze (EPM) for avoidance of open arms; novel object exploration (NOE); three-chamber social interaction test (SIT) to assess social preference; forced swim test (FST) and tail suspension test (TST) in repeated designs to probe antidepressant-like effects; cued fear conditioning with drug administered immediately before or after retrieval to assay fear expression and extinction; and a social conditioned place preference (sCPP) protocol to test social reward learning and whether psilocybin could reopen a putative juvenile critical period in adults. Details such as apparatus dimensions, centre zone definitions for OFT, trial timing for fear conditioning and sCPP protocols were harmonised but varied slightly across sites and are reported in supplementary materials. Procedures to reduce bias included blinding of experimenters at several sites (Stanford, Berkeley 1, Berkeley 2, UCSF 2), automated video-based scoring where possible (Ethovision, Biobserve, DeepLabCut and custom scripts), and manual scoring by trained blinded observers when indicated. Pre-experiment handling, acclimation, and other methodological variables were recorded for each lab. For statistics, the team used two-way ANOVAs (drug and sex), repeated-measures three-way ANOVAs when appropriate, and linear mixed-effects models to estimate effect sizes and partition variance attributable to treatment, methodological fixed effects, and random effects such as lab or cage. Simulations were run by z-scoring each lab's data to their saline controls, resampling 10,000 times, and calculating the distribution of p-values to estimate replicability probabilities for observed effect sizes.
Head twitch response: Psilocybin produced a robust and highly replicable increase in head twitch responses across all laboratories. A time-course analysis indicated the effect was strongest in the first ten minutes post-injection. Female mice tended to exhibit more head twitches than males. Acute anxiety-related behaviour (OFT, EPM): In the open field test, four of five labs observed that psilocybin acutely reduced time spent in the centre; the fifth lab showed a strong trend in the same direction. Pooling all data revealed a statistically strong anxiogenic effect despite substantial variability across measurements. In the elevated plus maze, three of five labs reported reduced time in the open arms after psilocybin, with the remaining labs showing effects in the same direction; pooled analysis again showed a significant acute anxiogenic effect. Overall locomotor activity in the open field was not altered acutely or persistently. Novel object exploration (NOE): Results varied between labs. Three labs reported an acute decrease in exploration of novel objects after psilocybin, while two labs reported inconsistent effects. When data were pooled, a significant decrease in novel object exploration was present in females and a nonsignificant trend in the same direction in males. Persistent (24-hour) effects on NOE were not observed in pooled analyses; four of five labs reported no persistent effect and the remaining lab reported an increase. Social preference (SIT): Across all laboratories, mice preferred the chamber with a conspecific, but psilocybin produced no acute or persistent alteration in social preference in any individual lab or in pooled analyses. Antidepressant-like assays (FST, TST): Using a repeated-design FST (pre-test, drug, post-test), the typical increase in immobility between test days was not consistently observed after saline across labs, and four of five labs found no significant antidepressant-like effect of psilocybin in the post-drug FST; pooled data were null. Repeated TST protocols showed a consistent across-day increase in immobility in all five labs (stress effect), but psilocybin did not reduce immobility in any lab or in pooled data. Naïve TSTs also revealed no main effect of psilocybin. Fear conditioning and extinction: When psilocybin was administered immediately before retrieval, every lab observed an acute reduction in freezing during retrieval, indicating reduced fear expression. Persistent effects on extinction (measured 24–48 hours later) were largely absent: four of five labs detected no persistent facilitation of extinction. Pooled analysis showed a small but significant persistent effect driven principally by a single lab (UCSF 1) that reported a large effect size. Administration of psilocybin after retrieval produced no persistent effects on extinction in four of five labs or in the pooled data. Social reward learning (sCPP): Using a simplified social conditioned place preference assay, three of five labs replicated a juvenile-only social preference (juveniles P34–41 showing a preference; adults P88–106 not). Two labs did not find robust juvenile social reward learning. In the three labs that did establish juvenile social reward learning, adult mice pre-treated with psilocybin or saline four days before conditioning showed no social reward learning, indicating psilocybin did not reopen the putative critical period. Effect size, replicability and methodological influences: Linear mixed-effects modelling across experiments indicated a large acute effect of psilocybin on HTR and modest acute effects on OFT, EPM, NOE and fear retrieval; persistent effects were small across assays. Simulations showed that persistent effects of the observed magnitude had low probabilities of yielding statistically significant results across independent replications, whereas acute robust effects had much higher replicability probabilities. Methodological variables (handling, injection acclimation, apparatus dimensions, mouse weight) significantly influenced some behavioural outcomes, but with the exception of handling on psilocybin's effect in novel object exploration, these variables did not systematically alter psilocybin's effect size. Among approximately 80 independent behavioural experiments, four isolated, non-replicable positive findings were observed, consistent with the expected rate of false positives at α=0.05. The authors therefore attribute the lack of consistent persistent effects primarily to small or nonexistent true persistent effect sizes at the tested dose rather than broad methodological differences.
Lu and colleagues interpret their findings to mean that psilocybin produces several robust, replicable acute effects in mice—most notably a consistent increase in head twitch responses, acute anxiogenic-like effects in approach–avoidance assays (OFT, EPM), and a reproducible reduction in fear expression during retrieval. By contrast, they found little evidence for replicable persistent (24–48 hour) effects on anxiety-, depression- or fear extinction-related behaviours, and no evidence that psilocybin reopens a social reward learning critical period in adult mice under the tested conditions. The authors position these results against prior literature that has reported both positive and null findings, arguing that some previously reported persistent effects may reflect small true effect sizes, spurious positives, or dependence on particular experimental conditions. They note that the acute anxiogenic effects align with higher-dose rodent studies and with human reports of transient anxiety during psilocybin sessions, while the consistent reduction in fear expression is concordant with most prior studies and may warrant mechanistic follow-up given its potential translational relevance for exposure-based therapies. Several limitations and uncertainties are acknowledged. There was substantial between-lab and within-subject variability in behavioural measures; sample sizes and effect magnitudes influence replicability; certain outcomes may be dose- or assay-dependent (other doses or behavioural paradigms could yield different persistent effects); and standard laboratory mice may not capture aspects of the human psychedelic experience that are necessary for enduring therapeutic benefit. The extracted text also notes that persistent effects might manifest in pathological or stress-model animals rather than in unstressed laboratory mice. In terms of implications, the investigators advocate for multi-site, highly coordinated experimental designs as a model for improving robustness and replicability in psychedelic research. They recommend caution in interpreting single-laboratory findings and suggest that collaborative approaches can more efficiently identify which drug effects are reliable and worthy of mechanistic or translational investment. Supplementary statistical reports are referenced for detailed analyses.
After decades of limited research, there is now renewed interest in studying psychedelic drugs. A single or few doses of various psychedelic drugs including psilocybin, MDMA, LSD, and ibogaine, appear to elicit rapid and enduring therapeutic effects for a wide range of neuropsychiatric disorders, including anxiety, depression, substance use disorder, and post-traumatic stress disorder. These preliminary clinical findings have generated great excitement in both the lay and scientific press because they suggest that psychedelic drugs may have unique therapeutic properties that could dramatically transform minutes after drug administration) and persistent (24 hours after drug administration) effects of the drug and focused on testing behaviors that classic psychedelics, including psilocybin, have been reported to affect.
To identify the replicable effects of psilocybin on behavior, it was critical to select an appropriate dose. We chose 2 mg/kg because this dose: (1) triggers head twitch responses in mice which are commonly considered a correlate of hallucinogenic activity, () is low enough to avoid gross motor changes, and () is similar to doses used in other preclinical studies. After confirming the concentration and stability of our drug preparation (Extended Data Fig.), each lab tested drug efficacy by monitoring head twitch responses for 20 minutes after administering psilocybin intraperitoneally (Fig.). Psilocybin consistently increased head twitch responses, both in each individual lab (Fig.) and when summarizing data across all labs (Fig.). Modest sex differences were evident, with female mice generally exhibiting more psilocybin-induced head twitches than male mice. Consistent with previous reports, a detailed time course analysis suggested that psilocybin-induced head twitches were most prominent in the first ten minutes after drug administration (Fig.).
Having confirmed the behavioral activity of psilocybin within the setting of each individual lab, we assessed the acute and persistent effects of the drug on behaviors that psychedelics are thought to alter, including behaviors related to anxiety, exploration, sociability, depression, fear extinction learning, and social reward learning. To assess psilocybin's acute effects on anxiety-related approach-avoidance behaviors, we first performed the open field test(Fig.). Four of five labs found psilocybin significantly reduced time in the center, with the Stanford lab observing a strong trend in the same direction (Fig.). Pooling data from all labs revealed a strong effect of the drug despite the great variability in individual measurements (Fig.). We next tested the acute effects of psilocybin in the elevated plus maze; a second anxiety-related behavioral test(Fig.). Three of five labs found that psilocybin-treated mice spent significantly less time in the open arms, with the other two labs generally observing effects in the same direction (Fig.). Pooling results from all labs again revealed a highly significant anxiogenic effect of psilocybin despite great variability in individual measurements (Fig.). Having found that psilocybin acutely increases anxiety-related avoidance behaviors in a replicable manner across labs, we performed the same assays 24 hours after psilocybin administration. In the open field test (Fig.), no individual labs detected a persistent effect of psilocybin on center time (Fig.), and no effect was revealed when results from all labs were pooled (Fig.). Similarly, in the elevated plus maze (Fig.), four of five labs found no persistent effects of psilocybin on the time spent in the open arms (Fig.), and again no effect was revealed when all results were pooled (Fig.). We also examined overall locomotion in the open field and observed no acute or persistent effects of psilocybin (Extended Data Fig.).
Avoidant coping strategies are frequently observed in anxiety. Therefore, we next tested whether psilocybin influenced avoidance of novel objects (Fig.). Although there was variability in results between labs, with three labs finding that psilocybin acutely decreased novel object exploration and the other two labs reporting inconsistent effects (Fig.), pooling results from all labs revealed a significant decrease in females and a nonsignificant effect in the same direction in males (Fig.). We also tested the persistent effects of psilocybin on novel object exploration by monitoring behavior in this test 24 hours after drug administration (Fig.). Four of five labs found no effects of psilocybin on object exploration at this timepoint with the remaining lab reporting an increase in exploration (Fig.). As expected from the individual labs' results, pooled results revealed no persistent effects of psilocybin (Fig.).
Psychedelics have been reported to influence social interactions. Therefore, we next examined psilocybin's acute and persistent effects on social approach/avoidance using the three-chamber social interaction test(Fig.). All labs found that subjects spent more time exploring the cup containing a conspecific, with individual lab social preference indexes ranging on average from as low as ~0.2 to as high as ~0.6. However, no lab observed either an acute (Fig.) or persistent (Fig.) effect of psilocybin on social preference.
Psilocybin has shown promise as a potential treatment for depression, motivating us to examine psilocybin's effects in the commonly used forced swim test (FST). Specifically, we replicated a protocol in which the FST is performed 1 day pre-and 1 day post-drug administration (Fig.) because several psychedelic drugs and analogs have been reported to elicit antidepressant activity as demonstrated by reduced immobility in the post-drug assay. While this repeated FST design has been reported to induce an increase in immobility on the second day (often interpreted as a "depression-like phenotype"), this effect was not replicable, as only two labs observed an increase in immobility during the second FST following saline administration (Fig.). Furthermore, psilocybin's antidepressant-like effects in this assay were not replicable, as four out of five labs found no significant effects of psilocybin on immobility in the second FST following psilocybin administration (Fig.), and pooled results from all labs did not yield significant effects (Fig.). Additional experiments replacing the FST with the tail suspension test (TST)(Fig.) yielded a consistent increase in immobility across days in all five labs (Fig.), but again, no antidepressant-like effects of psilocybin in this assay were observed by any lab (Fig.) or in the pooled data (Fig.). TST experiments performed in naïve mice also revealed no significant main effects of psilocybin (Extended Data Fig.). Taken together, these results demonstrate that psilocybin does not cause reliable changes in two commonly used assays of antidepressant activity in mice.
Psilocybin has been reported to enhance fear memory extinction potentially through its plasticitypromoting effects. To examine the replicability of these results, we performed a standard cued fear conditioning experiment during which psilocybin was administered either immediately before or after retrieval. Administration of psilocybin before fear memory retrieval (Fig.) caused a robust decrease in freezing during retrieval in all labs (Fig.). However, no persistent effects of psilocybin on fear extinction were observed 24 and 48 hours after drug administration in four out of five labs (Fig.). While psilocybin had a small but significant persistent effect on facilitating fear extinction in the pooled results (Fig.), this effect was driven by a single lab (UCSF 1) finding a large effect size. Administration of psilocybin after fear memory retrieval (Fig.) produced no persistent effects on fear extinction in four out of five labs (Fig.) and in the pooled data (Fig.).
Recently, it has been suggested that a broad array of psychedelic drugs can reopen a social reward learning critical period, which was defined by the ability of young (~21-80 days old) but not adult (>90 days old) mice to exhibit modest preference for a context in which they previously had social interactions with cagemates for 24 hours over a context in which they were socially isolated for 24 hours. In that study, psilocybin facilitated social reward learning in mice ~100 days old even when administered 2 weeks prior to the conditioning procedure. To test the replicability of this phenomenon, we used a simplified social reward learning protocol (Fig.), designed to mimic a conditioned place preference assay, which is commonly used to test the rewarding properties of drugs of abuse. Using this assay, three of five labs found that juvenile (34-41 days old) but not adult (88-106 days old) mice exhibited a place preference for the context in which they had previously experienced social interactions (Fig.), thereby suggesting the presence of a critical period for social reward learning. Although we used a slightly different assay for social reward learning, it is noteworthy that the magnitude of the social preference in the juvenile and adult mice that we observed was virtually identical to that previously reported (Juveniles = ~1.2, Adults = ~1.0). The remaining two labs did not observe robust social reward learning in juvenile mice (Extended Data Fig.). To test whether psilocybin reopens this social reward learning critical period, the three labs that observed social reward learning in juvenile mice administered saline or psilocybin to adult mice (88-106 days old) four days before the start of the social conditioned place preference protocol. No lab observed social reward learning following either saline or psilocybin treatment (Fig.), indicating that psilocybin did not reopen a social reward learning critical period.
The surprising lack of replicable persistent effects observed in this study could arise because: (1) the persistent effects of psilocybin are small and thus not able to be readily replicated, (2) methodological differences between labs preclude replicability, and/or (3) the positive results are spurious. To assess the role of small effect size in the replicability of our experiments, we first estimated the effect size of psilocybin across all experiments with linear mixed effects modeling (Methods). Consistent with our pooled datasets, psilocybin had a substantial acute effect on the head-twitch response and modest but significant acute effects on behaviors in the open field test, elevated plus maze, novel object exploration, and fear conditioning. Persistent effects were small across all experiments (Extended Data Fig.). To assess the influence of psilocybin's effect size on replicability, we estimated the probability of observing a significant main effect of psilocybin with simulation experiments (Methods). Persistent effects with small effect sizes had low replicability probabilities, which paled in comparison to the probabilities for the more robust, acute effects observed (Extended Data Fig.). Next, to explore whether mouse-, environmental-, and experimental-specific conditions systematically altered mouse behaviors, which could limit replicability, we added these conditions as fixed effects to our linear mixed effects models. In agreement with the well-established influence of environmental conditions on mouse phenotyping, we found that methodological variables significantly influenced mouse behavior in several experiments (Extended Data Fig.). For experiments with nonreplicable persistent effects (namely elevated plus maze, novel object exploration, forced swim test, and fear extinction), several methodological variables significantly influenced behavior, including handling, injection acclimation, apparatus dimensions, and mouse weight. However, these variables did not significantly influence psilocybin's effect size, with the exception of handling on psilocybin's effect in novel object exploration (Extended Data Fig.), suggesting that overall, nonreplicable persistent effects were unlikely to result from the influence of methodological differences. A final possibility which could explain the absence of replicable persistent effects is that isolated positive findings were spurious, false positives. When observing α=0.05, 5% of independent experiments are expected to be false positives, and indeed in our study of close to 80 total independent behavioral experiments, there were 4 occasions of singular, non-replicable positive effects. Together, these analyses suggest that nonreplicable persistent effects observed in this study are likely a result of psilocybin having a small or nonexistent persistent effect at the dose tested, rather than methodological differences.
We have presented results from a multi-institutional collaborative effort designed to elucidate the replicable acute and persistent effects of psilocybin on a range of mouse behaviors. Despite large quantitative variance in many of our behavioral measurements due to both lab-to-lab variability and individual subject variability, overall, many effects of psilocybin were replicable between labs (Extended Data Fig.). Specifically, we found that psilocybin acutely increased anxiety-related behaviors and decreased fear expression. Surprisingly, however, we found no evidence for replicable persistent effects of psilocybin 1-2 days after drug administration on anxiety-, depression-, and fear extinction-related behaviors. Furthermore, psilocybin did not reopen a critical period for social reward learning in adult subjects. These findings call into question the widespread notion that psilocybin promotes profound, long-lasting behavioral changes in commonly used behavioral assays. They also emphasize the advantages of this collaborative research effort and that, as many scientists appreciate but rarely explicitly state, results from any individual laboratory must be interpreted with caution. The reliable acute anxiogenic effects of psilocybin in the OFT and EPM reported here are consistent with previous work on the effects of psilocybin in the open field using higher doses (3-5 mg/kg). They also are consistent with human subjects' reports of psilocybin generating acute anxiety. However, these results are inconsistent with those for the psychedelic 2,5dimethoxy-4-iodoamphetamine (DOI), which has been reported to decrease avoidance behaviors in mice that were assessed using the elevated plus mazeor marble burying test. Additional work will be necessary to determine whether the inconsistent reported effects of psychedelics on anxiety-related behaviors are due to differences in the drugs being studied, methodological differences in the assays being performed, and/or generation of unreliable or spurious results. A second acute effect of psilocybin that was remarkably replicable across labs was its effect on reducing fear expression. Every single lab observed this effect, and to our knowledge, every study in the literature has reported similar results. Considering the replicability of this effect, future mechanistic studies of this effect are warranted, and if translatable to humans, this finding suggests the potential utility of psilocybin for suppressing fear responses during post-traumatic stress disorder exposure therapy. It was surprising that replicable persistent effects of psilocybin on anxiety, depression, and fear extinction-related behaviors were not detected. It should be noted, however, that in several of these assays, one lab found statistically significant effects. We propose three possible interpretations for these isolated observations: (1) psychedelics may have some long-lasting effects on these behaviors, but that the true effect size is much smaller than previously anticipated and thus not able to be consistently replicated (Extended Data Fig.), (2) the results are spurious and reflect chance variation, or (3) the results are extremely dependent on idiosyncratic testing conditions. We do not think that differences in results between labs were due to differences in our experimental methods because overall, psilocybin's effects varied minimally between labs (Extended Data Fig.), and methodological variables did not significantly influence psilocybin's effect size when non-replicable effects were observed -apart from handling on novel object exploration (Extended Data Fig.). It is possible that experiments conducted using different behavioral assays or doses might yield different effects. For example, psilocybin (1.5 or 3 mg/kg IP) was reported to have anxiolytic effects in the novelty-suppressed feeding testbut not in the open field test when assayed four hours after drug administration. Furthermore, psilocybin's effect of facilitating fear extinction was reported to have an unusual dose-dependence with maximal effects at 1 mg/kg and null effects at 2 mg/kg, albeit another study reported significant effects using 2.5 mg/kg. The relevance of results using mice to the effects of psychedelics in human subjects remains to be determined. Although we observed that psilocybin's persistent behavioral effects in standard laboratory behavioral paradigms are minimal in mice, it is possible that factors unique to the human psychedelic experience cannot be mimicked in mice and may be critical for the potential enduring therapeutic effects of psilocybin. Another possibility is that psilocybin has effects that specifically manifest in pathological states but are not observed in standard laboratory mice, which have not been subjected to interventions such as stress. Nevertheless, it seems very likely that there will continue to be an increase in efforts devoted to elucidating the mechanisms of psychedelics in experimentally tractable species such as miceand that the results from these studies will be used to interpret the therapeutic actions of psychedelics in human subjects. Given the typical trajectory of scientific discovery from individual investigators, the replicability of new findings will only slowly be realized. Furthermore, basing conclusions and future studies on unreliable findings will impede, not accelerate, scientific understanding and progress. Thus, we believe that the advantages and importance of collaborative research efforts of the type described here are self-evident. We hope that our efforts to define the replicable effects of psilocybin on a variety of mouse behaviors using a collaborative approach inspires others to adopt similar approaches. In this way, we believe we can make the most rapid and important advances to psychedelic science. Statistical reports for all analyses are provided in Supplementary Table..
Male and female C57BL/6J mice (Jackson Laboratory, stock number: 000664) were used for all experiments. Records of experimental mouse ages, weights, and housing numbers are detailed in Supplementary Table. Mice were maintained on the following light:dark cycles in each lab:
Psilocybin powder (obtained from the National Institute of Drug Abuse) was diluted in 0.9% sterile saline to a concentration of 0.2 mg/mL. Psilocybin solution was then administered intraperitoneally (IP) at a volume of 10 mL/kg to achieve a dose of 2 mg/kg for all experiments.
Psilocybin certified reference material was purchased from Cerilliant Ⓡ P-097 1mL; 1.0 mg/mL in acetonitrile:water (1:1); lot#FE07232102. For use as an internal standard (IS), 5-Hydroxy-L-Tryptophan (5-HTP) was purchased from TCI America™ (≥98.0% purity determined by HPLC).
A psilocybin standard reference solution was serially diluted with saline to obtain a series of standard working solutions, and a 10 mM stock of 5-HTP in DMSO was diluted to give a 500 µg/mL IS working solution. The calibration curve was prepared by spiking 20 µL of each standard working solution into 200 µL water followed by addition of 20 µL IS. A calibration curve was prepared fresh with each set of samples and comprised a range of 0.5-50 µg/mL psilocybin in 200 µL aliquots.
A sample of psilocybin used in the behavioral assays reported herein (0.2 mg/mL) was prepared in normal saline (0.9% NaCl, pH = 7.4, rt), aliquoted (200 uL) into triplicate 1.5 mL plastic tubes, and subjected to a variety of conditions: 1) freeze-thaw cycles (1x or 3x), 2) incubation at 4°C (3, 7, and 14 days), 3) incubation at 21°C under ambient 700 lux light (24 hours), 4) incubation at 21°C in the dark (8 or 24 hours), 5) incubation at pH = 10.05 and 21°C in the dark (100 mM sodium bicarbonate, 24 hours), or 6) incubation at pH = 2.05 and 21°C in the dark (0.1% formic acid, 24 hours). Samples for each condition were frozen at -80°C prior to analysis, whereupon they were thawed, vortexed, and diluted 20-fold in water followed by addition of 20 µL IS (500 µg/mL 5-HTP). Samples were then vortexed and centrifuged for 5 min at 10,000 rpm at 4°C, and 5 µl of supernatant was subjected to liquid-chromatography high resolution mass spectrometry (LC-HRMS).
A unified LC-UV-HRMS method was used to quantify psilocybin concentrations across sample conditions with a Waters Acquity H Class UPLC Plus Bio Quaternary system equipped with Orbitrap Exploris 240 (Thermo Fisher Scientific) HRMS. Liquid Chromatography was performed on an Atlantis® T3 (3 µm, 2.1 mm x 100 mm) column held at 40 °C using a linear gradient of mobile phases: A) 20 mM ammonium formate + 0.1% formic acid in water and B) 20 mM ammonium formate + 0.1% formic acid in acetonitrile/methanol/water 45:45:10. Elution was performed using an initial hold at 5% B for 3 minutes, followed by a gradient of 5-95% in 2 minutes, then held at 95% for 2 minutes; total run time was 10 minutes at a flow rate of 250 returned directly to their home cages for 24 hours before behavioral testing. For all experiments, behavioral equipment was cleaned with 70% EtOH before experiments, after experiments, and between individual animals. Experimenters at Stanford, Berkeley 1, Berkeley 2, and UCSF 2 were always blinded to the treatment group. For a more comprehensive report of experimental conditions for each mouse that could influence behavioral results (i.e., sexes, ages, housing numbers, dimensions of materials and apparatuses, handling counts, acclimation times, experiment times of day, number of mice tested concurrently, prior experimentation, and treatment details), see Supplementary Table.
The head twitch response is a rapid, side-to-side, rotational head movement that is considered a proxy of hallucinogenic activity in mice. To quantify psilocybin-induced head twitch responses, mice were injected with psilocybin or saline IP and monitored for 20 minutes. Head twitches were visually quantified by a blinded, trained observer.
To assess psilocybin's effects on locomotion and anxiety-like behaviors, mice were placed into an open field box and allowed to behave freely for 30 minutes (acute experiments) or 15-30 minutes (persistent experiments). Experiments were video recorded, and the average velocity (cm/s) and percent time spent in the center were automatically quantified using Ethovision, Biobserve, or DeepLabCut & custom MATLAB scripts. The center zone size varied from 39-57% of the total area of the open field.
To assess psilocybin's effects on anxiety/avoidance-like behaviors, mice were placed into the elevated plus maze, facing the open arm opposite the experimenter, and allowed to freely behave for 10 minutes for both acute and persistent experiments. Experiments were video recorded, and the percentage of time the animal's center body was in the open arms was automatically computed using Ethovision, Biobserve, or DeepLabCut & custom MATLAB scripts.
To assess psilocybin's effects on social preference, mice underwent the social interaction test. In this test, two wire cups (to allow contact between mice) were placed into the left and right chambers of a three-chambered apparatus. Next, an ovariectomized female mouse or same-sex juvenile mouse was positioned inside one of the cups. Afterwards, the experimental mouse was placed into the center chamber of the three-chambered apparatus and allowed to explore freely for 10 minutes for both acute and persistent experiments. Experiments were video recorded, and the time exploring both the empty and social cups was quantified using automated methods (Ethovision, Biobserve, or DeepLabCut & custom MATLAB scripts). For all automated methods, an animal was considered exploring if its nose was within a zone defined around the cups. The social interaction index was defined as (time spent exploring the social cup -time spent exploring the empty cup) / (total time exploring both cups). Animals placed inside the cups were habituated to the cups prior to experiments, and experimental mice were habituated to the 3-chambered apparatus (with cups but no animals) prior to the experiment.
To assess psilocybin's effects on exploration of a novel object, two identical objects were placed into an open field box. Then, mice were placed into the open field and allowed to explore freely for 10 minutes. Experiments were video recorded, and the total time spent exploring the objects was quantified using DeepLabCut & custom MATLAB scripts. The animal was considered exploring if its nose was within an object's zone.
To assess psilocybin's persistent effects on passive coping behaviors in naïve mice, mice underwent the tail suspension test as previously described. Briefly, mice were hung by their tails, at least 25 cm from the ground, for 6 minutes. Experiments were video recorded, and the time the animal spent immobile was quantified with Ethovision.
To assess psilocybin's persistent effects on passive coping behaviors in stressed mice, we used a protocol similar to one that has yielded antidepressant-like effects of psychedelics. Specifically, for the repeated tail suspension test, mice were first handled for one to three consecutive days (1 minute per mouse) prior to experimentation (exception: UCSF 2). On day 1 of the experiment, mice underwent a 6-minute TST pre-test, following the protocol described above, to induce a depression-like phenotype. On day 2, mice were administered saline or psilocybin IP. On day 3, mice underwent a 6-minute TST post-test to assess whether psilocybin could reduce stress-induced increases in passive coping. All experiments were video recorded, and immobility was quantified using Ethovision. The repeated forced swim test was conducted exactly as previously described. Briefly, mice were first handled for one to three consecutive days (1 minute per mouse) prior to experimentation (exception: UCSF 1). On day 1 of the experiment, mice underwent a 6-minute FST pre-test, in which mice were placed in a clear Plexiglas cylinder filled with 24 ± 1°C water to induce a depression-like phenotype. On day 2, mice were administered saline or psilocybin IP. On day 3, mice underwent a 6-minute FST post-test to assess whether psilocybin could reduce stressinduced increases in passive coping. All experiments were video recorded, and immobility scores were determined via manual scoring by a trained, blinded observer.
To assess psilocybin's persistent effects on fear extinction learning, we performed fear conditioning experiments. On day 1 of the experiment (encoding session), mice were placed into context 1, which included a lemon scent, metal bar flooring and rectangular walls. The encoding session consisted of a 300 second habituation period, followed by five tone-shock pairings (30 seconds, 2.9 or 5 kHz, 65-75 dB pip tone, concluding with a 2 second, 0.75 mA footshock). The intertrial interval ranged from 55-65 seconds. On day 2 of the experiment (retrieval session), mice were placed into context 2, which included an anise scent, hard plastic flooring, and round walls. The retrieval session consisted of a 180 second habituation period, followed by 2 tones (same tones as during encoding), separated by a 50 second intertrial interval. On day 3 and 4 of the experiment (extinction sessions), mice were again placed into context 2. The extinction sessions consisted of a 180 second habituation period, followed by 10 tones (same tones as during encoding), with intertrial intervals ranging from 55 to 65 seconds. At the conclusion of each session, mice were removed from fear boxes ~60 seconds after the final tone and returned to their home cage or holding cage. UCSF labs used holding cages to transfer mice between the holding room and experimental room. For injection before retrieval experiments, mice were injected 5-10 minutes before the retrieval session and placed into individual holding cages. For injection after retrieval experiments, mice were injected 5-10 minutes after the retrieval session and returned to their home cages directly after. All experiments were video recorded and analyzed by a blinded, trained observer or through automated methods (FreezeFrame or custom MATLAB scripts).
To assess whether psilocybin could reopen the social reward learning critical period, mice underwent a social conditioned place preference task. In this task, mice were first single-housed for three days prior to any experimentation to prime a desire for social interaction. On day 1 of the social conditioned place preference experiment, mice underwent a 30-or 60-minute pre-test, during which mice were placed into skinner boxes or custom-made boxes that had two distinct contexts on the left and right sides. Textured / smooth flooring and striped / polka-dot walls were used to create distinct contexts. Then, on days 2-4, mice experienced 1-hour periods of social conditioning, in which they were re-united with 1-4 cage mates in one context, and 1-hour periods of isolation conditioning, in which they were alone in the other context. Finally, on day 5, mice underwent a 30-or 60-minute post-test, in which they had free access to explore both contexts in the skinner boxes. To quantify social reward learning, we identified the mouse's body center using automated methods (Biobserve, Ethovision, or DeepLabCut) and compared time spent in the social context during the pre-test versus post-test. The social preference score was calculated as the time spent in the social context during the post-test divided by the time spent in the social context during the pre-test. Mice that exhibited strong baseline preferences for one chamber, defined as spending more than 75 percent of time in one chamber during the pre-test, were excluded. Using this task, each lab first tested whether juvenile mice (P34-41) experienced social reward learning, defined as having a social preference score that was significantly different from 1.0 (a value indicative of no social reward learning). In labs that established social reward learning in juvenile mice, adult mice (P88-106) were then tested to determine if, as previously reportedsocial reward learning was absent at this age. To test whether psilocybin could 'reopen' the social reward learning critical period in adult mice, mice were pre-treated with saline or psilocybin four days prior to the start of experiments (while mice were still group housed, 1 day prior to singlehousing).
To determine the distribution of p-values expected if 10,000 labs were to repeat our experiments, we first z-scored each lab's data to each lab's saline data, for males and females separately to determine the normalized size of the effect in each experiment. Next, we resampled a kernel density estimation of our z-scored data 10,000 times and used 2-way ANOVAs (factors: drug and sex) or 3-way ANOVAs (for fear conditioning; factors drug, sex, and cue number) to analyze each sample. The resulting distribution of p-values for the main drug effect was then plotted. The replicability probability was calculated for experiments in which at least one lab showed a positive result and was defined as the percentage of experiments expected to yield a p-value less than 0.05.
To determine whether psilocybin had a significant effect on mouse behavior, and whether there were any sex-specific effects, we performed two-way ANOVAs using Prism 10 (Graphpad), with drug and sex as factors. Significance for main effects of drug and/or sex are visualized with asterisk(s) next to either the drug legend or next to sex conditions, respectively. When there was a significant interaction between drug treatment and sex, Sidak's multiple comparisons were used to test for any sex-specific effects of the drug, and significance from post-hoc comparisons is denoted on the main body of the graphs. Statistical significance was *p<0.05, **p<0.01, ***p<0.001. Data are presented as mean ± SEM.
Repeated measures three-way ANOVAs were performed when there was an additional third variable of interest (i.e., timepoint, cue number, object type, cup type). Factors included drug, sex, the third variable of interest, and all interactions. Statistical significance was *p<0.05, **p<0.01, ***p<0.001. Data are presented as mean ± SEM.
The percentage of variance explained by psilocybin treatment (Extended Data Fig.and persistent (pNOE) experiments, social preference index from acute (aSIT) and persistent (pSIT) social interaction test experiments, day 3 immobility times (s) from the repeated forced swim test (rFST) and repeated tail suspension test (rTST), immobility times (s) from the persistent tail suspension test (pTST) in stress-naïve mice, average % time freezing to cues during retrieval for fear conditioning injection before retrieval experiments (FC -pre -retr), average % time freezing to cues during the first extinction session for fear conditioning injection before retrieval (FC -pre -ext1) and injection after retrieval (FC -post -ext1) experiments, average % time freezing to cues during the second extinction session for fear conditioning injection before retrieval (FC -pre -ext2) and injection after retrieval (FC -post -ext2) experiments, and social preference scores for adult saline versus psilocybin social conditioned place preference (sCPP) experiments. The abbreviated names in parentheses are what is presented in the extended data figures. To determine psilocybin's effect size in z-score (Extended Data Fig.), we first z-score normalized all data points in each lab to each lab's saline groups for each sex separately before fitting linear mixed effects models. Outcome variables used for each model are as described above. Model statistics were evaluated using R packages MuMIn 70 and lmerTest. To determine fixed effects due to treatment or methodology and random effects due to lab or cage, we used additional linear mixed-effects models. Fixed effects, with abbreviated names in brackets, included: i. Features that had the same value for all mice tested across the five labs were excluded from the model. The fixed effect size of each feature in each experiment was estimated with 95% confidence intervals. Outcome variables used for each model are as described above. For the methodological variables that had a significant influence on mouse behavior in non-replicable persistent experiments, simple linear regression was performed to quantify the effect of that variable on psilocybin's effect size, where x = selected methodological variables, and y = z-scored psilocybin data for the selected experiment. Slope values from linear regression were estimated with 95% confidence intervals. To assess how mouse behavioral outcomes depended on the laboratory in which they were conducted, we evaluated the degree of clustering by comparing the intraclass correlation (ICC) across all behavioral paradigms. ICC was calculated for both the raw data to identify the inter-lab variability in the raw measurements (Extended Data Fig.), as well as for the z-score normalized data to identify the inter-lab variability in psilocybin's effect (Extended Data Fig.).
Create a free account to open full-text PDFs.
Schindowski, E. M., Jungwirth, J., Schuldt, A. et al. · EClinicalMedicine (2023)
Mitchell, J., Ot’alora G, M., van der Kolk, B. et al. · Nature Medicine (2023)
Gasser, P., Holstein, D., Michel, Y. et al. · Journal of Nervous and Mental Disease (2014)
Cherian, K. N., Keynan, J. N., Anker, L. et al. · Nature Medicine (2024)
Frampton, C. M., Yazar-Klosinski, B., Nollar, G. E. · The American Journal of Drug and Alcohol Abuse (2017)
Carhart-Harris, R. L., Giribaldi, B., Watts, R. et al. · New England Journal of Medicine (2021)
Hesselgrave, N., Troppoli, T. A., Wulff, A. B. et al. · PNAS (2021)
Jefsen, O., Højgaard, K., Christiansen, S. L. et al. · Acta Neuropsychiatrica (2019)
Cameron, L. P., Patel, S. D., Vargas, M. V. et al. · ACS Chemical Neuroscience (2023)
Halberstadt, A. L., Chatha, M., Klein, A. K. et al. · Neuropharmacology (2020)
Passie, T., Seifert, J., Schneider, U. et al. · Addiction Biology (2002)
Fadahunsi, N., Lund, J., Breum, A. W. et al. · Translational Psychiatry (2022)
Shao, L-X,, Liao, C., Gregg, I. et al. · Neuron (2021)
Nardou, R., Sawyer, E., Song, Y. J. et al. · Nature (2023)
Nardou, R., Lewis, E. M., Rothhaas, R. et al. · Nature (2019)
Cameron, L. P., Tombari, R. J., Lu, J. et al. · Nature (2020)
Vargas, M. V., Dunlap, L. E., Dong, C. et al. · Science (2023)
Dong, C., Ly, C., Dunlap, L. E. et al. · Cell (2021)
Moliner, R., Girych, M., Brunello, C. A. et al. · Nature Neuroscience (2023)
Papers in Blossom that reference this study
Jacobs, E., Zahid, Z., Hinkle, J. et al. · BMJ (2026)