Appetite for change: How psilocybin reshapes food reward learning through striatal dopamine function

Abstract

Psilocybin has emerged as a promising therapeutic agent for psychiatric disorders characterised by cognitive rigidity and disrupted reward processing, including anorexia nervosa. While its pro-cognitive effects have been mechanistically probed almost exclusively through serotonin receptor subtype antagonism, the downstream contributions of dopaminergic systems to these outcomes remain poorly understood. Here, we examined how psilocybin (1.5 mg/kg) alters cognitive flexibility across multiple operant paradigms in female rats and whether ventral striatal dopamine, nutritional state or prior activity-based anorexia (ABA) exposure moderate these effects. In food-restricted rats, psilocybin improved reversal learning with an apparent temporal delay relative to our previously published ad libitum fed cohort. In vivo fiber photometry revealed that psilocybin broadly amplified nucleus accumbens core (NAc) dopamine transients time-locked to both expected and unexpected outcomes during probabilistic reversal learning across 7 days. Computational modelling identified psilocybin-specific increases in learning rate and reductions in prior value weighting, consistent with strengthened feedback-driven updating. In touchscreen paradigms, psilocybin enhanced discrimination accuracy and accelerated reversal learning acquisition when administered prior to initial discrimination, but impaired serial reversal accuracy when administered after initial discrimination. ABA exposure constrained psilocybin's pro-cognitive effects, abolishing discrimination accuracy benefits, suggesting that metabolic disruption and stress alter the neural substrates required for psilocybin-enhanced neuroplasticity and reward learning. These findings provide the first direct evidence that psilocybin modulates striatal dopamine signalling in a behaving animal and demonstrate that current nutritional state and prior experience with pathological weight loss and reward learning moderate its cognitive effects.