Serotonergic psychedelics have revitalized psychiatric research by producing rapid and potentially sustained clinical benefits in selected neuropsychiatric conditions. However, the mechanisms linking receptor activation to neural plasticity, altered conscious experience, and clinical improvement remain incompletely understood, while the evolutionary origins of psychedelic signaling and the translational value of animal models continue to be actively debated. In this review, we integrate evolutionary biology, molecular pharmacology, comparative neuroscience, and computational psychiatry into a unified framework for understanding psychedelic action across biological scales. Building upon predictive processing and the Relaxed Beliefs Under Psychedelics (REBUS) framework, we discuss how serotonergic psychedelics transiently reshape hierarchical inference, providing computational principles that can be experimentally investigated across species despite profound differences in neural architecture and subjective experience. Comparative analyses from invertebrates to non-human primates further suggest that conserved receptor pharmacology coexists with species-specific circuit organization, highlighting that animal models should be interpreted as complementary platforms for investigating molecular, cellular, and computational mechanisms rather than direct proxies for human hallucinatory experiences. We further discuss preclinical and emerging translational evidence suggesting that biased receptor signaling and non-hallucinogenic psychoplastogens may partially dissociate plasticity-related therapeutic actions from hallucinogenic effects, providing new opportunities for mechanism-based drug development. By integrating evolutionary biology with molecular psychiatry, this review provides a mechanistic framework linking receptor pharmacology, neural plasticity, circuit computation, and therapeutic outcomes, thereby offering new conceptual and experimental strategies for developing next-generation psychedelic therapeutics.
