Serotonergic psychedelics have been identified as promising next-generation therapeutic agents in the treatment of mood and anxiety disorders. While their efficacy has been increasingly validated, the mechanism by which they exert a therapeutic effect is still debated. A popular theoretical account is that excessive 5-HT2a agonism disrupts cortical dynamics, relaxing the precision of maladaptive high-level beliefs and making them more malleable and open to revision. We extend this perspective by developing a simple energy-based model of cortical dynamics based on predictive processing which incorporates effects of neuromodulation. Using this model, we propose and simulate hypothetical computational mechanisms for both 5-HT2a and 5-HT1a agonism. Results from our model are able to account for a number of existing empirical observations concerning serotonergic psychedelics effects on cognition and affect. Using the findings of our model, we provide a theoretically-grounded hypothesis for the clinical success of LSD, psilocybin, and DMT, as well as identify the design space of biased 5-HT1a agonist psychedelics such as 5-MeO-DMT as potentially fruitful in the development of more effective and tolerable psychotherapeutic agents in the future.
Classic psychedelics all have significant affinity for both the 5-HT2a and 5-HT1a receptors. Although 5-HT2a is responsible for the main psychedelic effects, 5-HT1a also plays a significant modulating role. We set out to computationally characterize both of these roles.
We adopt the predictive processing framework and an energy-based model in which neural responses are the result of an optimization process on an energy landscape. During inference 'energy' is minimized, and during learning the 'predictive error' is minimized. Within this framework, many mental disorders (depression, OCD, etc) are understood as pathologies of optimization. Overly-precise and maladaptive priors manifest as local minima with steep gradients within the energy landscape, a phenomenon sometimes called canalization.
We model 5-HT2a as injecting noise into the energy landscape, and 5-HT1a as smoothing it. The former results in acute overfitting during inference, while the latter in acute underfitting. Since many psychedelic (PSI, LSD, DMT) are mixed agonists, both happen simultaneously.
The results of our simulations suggest that 5-HT2a is responsible for long-term therapeutic effects, but at the cost of short-term acute tolerability. In contrast, 5-HT1a is acutely therapeutic and tolerable, but provides little long-term efficacy. Things get interesting when you mix both.
In our model mixed agonists have greater long-term efficacy than 5-HT2a alone, while also being significantly more acutely tolerable. We find that if you want to optimize for both long-term and acute therapeutic effects an optimal agonism bias is towards 5-HT1a over 5-HT2a.
5-MeO-DMT, a highly-biased 5-HT1a agonist, has received clinical attention for its potential to treat depression. Likewise for the co-administering of MDMA and LSD. There is a whole space of biased 5-HT1a agonists such as 5-MeO-MIPT which may also be worth exploring. Our work points to the importance of non-5HT2a receptor targets in the efficacy and tolerability of psychedelic therapy. Perhaps not surprisingly, the tryptamines have this profile, and the clinical success of psilocybin may be attributable to its unique mixed profile.
@article{juliani2024dual,
title={A dual-receptor model of serotonergic psychedelics},
author={Juliani, Arthur and Chelu, Veronica and Graesser, Laura and Safron, Adam},
journal={bioRxiv},
pages={2024--04},
year={2024},
publisher={Cold Spring Harbor Laboratory}
}