Implications for psychedelic-assisted psychotherapy: functional magnetic resonance imaging study with psilocybin

Psilocybin is a classic psychedelic historically used in psychotherapy on the premise that it lowers psychological defences and facilitates emotional insight. Earlier literature and case reports describe spontaneous autobiographical recollections or "relivings" under psychedelics that resemble dream-like sequences or flashbacks, and prior work by the group found decreased resting activity in medial prefrontal cortex after psilocybin—a region thought to exert top-down inhibitory control over limbic structures. These observations motivated a hypothesis that psilocybin might facilitate access to salient personal memories via neural disinhibition of limbic and sensory systems. This study set out to test whether psilocybin augments subjective and neural responses to positive autobiographical memories. Using a placebo-controlled, within-subject cross-over design with functional magnetic resonance imaging (fMRI), the investigators compared responses to personal positive memory cues under intravenous psilocybin (2 mg) versus saline. They predicted increased activations in medial temporal and visual association regions and greater subjective vividness and emotionality of recollections under psilocybin. Positive cues were used deliberately to reduce the risk of adverse reactions during the drug state.

A balanced, within-participant cross-over design was used: each participant underwent two fMRI sessions separated by at least 7 days, receiving 2 mg intravenous psilocybin on one day and saline placebo on the other in counterbalanced order. Solutions (10 ml) were infused manually over 60 s during scanning; the behavioural memory task began about 7.5 min after the start of infusion, a period during which subjective effects of the 2 mg dose are reported to peak and persist for roughly 25–30 min. Scanning took place on a 3 T GE system and included a high-resolution structural scan and BOLD-weighted echo-planar imaging (TR/TE 3000/35 ms, 3 mm isotropic voxels); an initial resting-state scan was acquired but reported elsewhere. Participants were healthy volunteers recruited by word of mouth. The extracted text reports 15 enrolled individuals, with attrition and data loss leaving 10 participants with usable functional data for the memory task (nine male, mean age approximately 31 years). All included participants had prior psilocybin experience. Screening included psychiatric interview, blood tests, ECG and neurological examination. Autobiographical memory cues were derived from participants: at least 30 specific, positive, emotionally salient personal events were submitted, rated for potency and recency, then split into two matched sets of 15 cues (one set per drug condition). During the fMRI behavioural paradigm participants viewed each cue for 6 s with eyes open then closed their eyes for a 16 s recollection period; recollection periods were split into an early phase (first 8 s) and a late phase (last 8 s) for analysis. Memory blocks were interleaved with rest and an auditory attention task in a fixed sequence repeated 15 times to reduce carry-over. After scanning each memory was rated on 0–10 scales for vividness, emotional intensity, valence and visual imagery. Two-week follow-up ratings of changes in subjective well-being were collected on a −3 to +3 scale. Imaging analysis used FSL: first-level general linear models included regressors for early and late recollection, attention, instruction periods and temporal derivatives; data were high-pass filtered and spatially smoothed. Primary contrasts were early recollection v. rest and late recollection v. rest. Higher-level within-subject mixed-effects analyses compared psilocybin with placebo; cluster-thresholding was applied with thresholds reported in the text as P50.001 and P50.05 (the extraction appears to have lost the less-than signs). Regions of interest for bilateral parahippocampal gyrus and amygdala were defined from the Harvard–Oxford atlas. Pearson correlations were used to relate regional BOLD changes to subjective ratings.

Subjective ratings after the scan indicated that memories were experienced as more vivid, visual, emotional and positive under psilocybin than under placebo. Individually, vividness (t = not reported, P = 0.049) and visual imagery (P = 0.041) differed significantly between conditions, and a combined measure of the four correlated parameters was higher under psilocybin (P = 0.0003). Cued autobiographical recollection produced robust activations versus rest in both conditions. Early-phase activations (first 8 s) were predominantly subcortical and included limbic and medial temporal regions (amygdala, hippocampus), striatum (putamen, nucleus accumbens), mid-cingulate, pre-sensorimotor area and precuneus. Late-phase activations (last 8 s) included limbic and paralimbic regions (amygdala, subgenual cingulate, temporal pole) and prominently the medial prefrontal cortex and frontal pole, consistent with memory elaboration. Comparing conditions, there were no significant psilocybin–placebo differences in early-phase activations. In the late phase, psilocybin produced significantly greater activations in three large clusters: a left occipital pole and visual association region cluster, and bilateral clusters encompassing mid-insula, primary and secondary auditory cortex, temporal pole, primary and secondary somatosensory cortex and superior parietal regions (including superior parietal lobule and supramarginal gyrus). First-level percentage BOLD change analyses showed these sensory and parietal regions tended to be deactivated during recollection under placebo but were activated under psilocybin, especially the visual cluster. Correlational analyses across combined scans found a significant positive relationship between late-phase bilateral parahippocampal activation and emotion ratings (r = 0.4, P = 0.04, one-tailed) and a trend-level correlation between vividness and late anterior parahippocampal activation (r = 0.35, P = 0.07, one-tailed). Two-week follow-up ratings of well-being in the complete sample (n = 15) were higher after psilocybin than placebo (P = 0.03). In the subset of 13 participants who completed both memory tasks, vividness under psilocybin correlated with increased well-being at follow-up (r = 0.72, P = 0.004, one-tailed) after removal of one outlier. Notably, none of the participants reported dramatic "relivings" akin to those described in historical reports, and contrary to the authors' prediction there was no significant augmentation of medial temporal lobe activation under psilocybin relative to placebo.

Carhart-Harris and colleagues interpret these results as evidence that psilocybin augments the sensory components of autobiographical recollection: memories were rated as more vivid and visual under psilocybin and late-phase activations in visual and other sensory cortices shifted from deactivation under placebo to activation under drug. The authors suggest this sensory augmentation may account for the increased sense of "realness" or vivid reliving reported anecdotally with psychedelics. The observed pattern—early limbic activation followed by later medial prefrontal engagement—is consistent with prior autobiographical memory studies, and the correlation between parahippocampal activity and emotional intensity aligns with existing findings. However, the investigators did not observe the predicted enhancement of medial temporal lobe responses or dramatic relivings under psilocybin. They note that invasive stimulation studies link medial temporal lobe input to posterior sensory regions during relivings, and propose that future connectivity analyses (for example dynamic causal modelling) could test whether psilocybin increases medial temporal input to visual cortex or reduces medial prefrontal control over medial temporal structures during recollection. Several methodological considerations that may explain negative findings are acknowledged. Using only positive memory cues—an ethical choice to reduce risk—may have limited the likelihood of relivings, which are often associated with conflict-laden or traumatic material. Short recollection windows (16 s) may have constrained the depth of re-experiencing. The authors also emphasise limitations of the sample: the functional imaging analysis relied on a small subsample (n = 10), increasing the chance of false negatives, and participants were self-selecting with prior psychedelic experience, which could bias subjective reports and follow-up well-being ratings. Nevertheless, the authors argue that such biases are unlikely to explain the observed increases in sensory-region activation. In terms of clinical implications, the paper suggests potential applications of psilocybin-assisted psychotherapy. The enhancement of positive autobiographical recall could be harnessed to reverse negative cognitive biases in depression by facilitating access to positive life events; this idea is supported by prior reports of reduced depression scores and sustained personality changes after single psilocybin sessions. The authors recommend further work in clinical populations, designs that include conflict-laden memories within therapeutic frameworks, longer recollection periods, and connectivity-focused neuroimaging to clarify mechanisms such as medial prefrontal disinhibition of limbic and sensory networks.