How Psychedelic-Assisted Treatment Works in the Bayesian Brain

This paper (2022) provides the first account of the interaction between pharmacological and psychological effects in psychedelic-assisted treatment. The paper builds on the previously proposed REBUS hypothesis and the contextual model of Wampold. It is assumed that psychedelics attenuate the precision of high-level predictions, making them more revisable by bottom-up input and psychotherapy is an important source of such input.


“Psychedelics are experiencing a renaissance in clinical research. In recent years, an increasing number of studies on psychedelic-assisted treatment have been conducted. So far, the results are promising, suggesting that this new (or rather, rediscovered) form of therapy has great potential. One particular reason for that appears to be the synergistic combination of the pharmacological and psychotherapeutic interventions in psychedelic-assisted treatment. But how exactly do these two interventions complement each other? This paper provides the first account of the interaction between pharmacological and psychological effects in psychedelic-assisted treatment. Building on the relaxed beliefs under psychedelics (REBUS) hypothesis of Carhart-Harris and Friston and the contextual model of Wampold, it argues that psychedelics amplify the common factors and thereby the remedial effects of psychotherapy. More precisely, psychedelics are assumed to attenuate the precision of high-level predictions, making them more revisable by bottom-up input. Psychotherapy constitutes an important source of such input. At best, it signalizes a safe and supportive environment (cf. setting) and induces remedial expectations (cf. set). During treatment, these signals should become incorporated when high-level predictions are revised: a process that is hypothesized to occur as a matter of course in psychotherapy but to get reinforced and accelerated under psychedelics. Ultimately, these revisions should lead to a relief of symptoms.”

Authors: Daniel Villiger



Psychedelics are on the rise again, with 17 clinical trials in 2020 (1). Michael Pollan’s 2018 book How to Change Your Mind became a No. 1 New York Times bestseller.

The 1950’s and 1960’s constitute the first flowering of psychedelic research, but the United Nations declared psychedelics a potential health and security risk and demanded strict regulation. The United States Drug Enforcement Agency then classified psychedelics as Schedule I substances, which eventually drained the entire research area.

Psychedelics were first examined in healthy participants at the end of the 20th century, but have since been used in clinical contexts to treat depression, OCD, anxiety, and alcohol abuse.

The relaxed beliefs under psychedelics (REBUS) hypothesis (11), based on the free-energy principle (12), and the common factors theory of psychotherapy (15), is a comprehensive account of psychedelic-assisted treatment (16), considering both its pharmacological and its psychological effects as well as their interaction.

The paper is structured as follows: pharmacological view, psychological view, and integrative view.


LSD, psilocybin and DMT are all psychedelics that cause profound changes in perception and mood. Taking a 5-HT2AR antagonist before taking a psychedelic substantially attenuates its typical phenomenological effects.

To understand the neurological role of 5-HT2ARs, we have to look at how the brain is organized and discuss the free-energy principle, which is a theory that explains how organisms resist entropy by avoiding surprising or uncertain states.

The Bayesian brain hypothesis, which is based on the free-energy principle, claims that the brain is an active probabilistic prediction machine that optimizes both internal probabilistic representations of an organism’s environments and how these environments are sampled.

Predictive processing is closely related to the free-energy principle and has received support from a vast range of theoretical and experimental studies, including functional magnetic resonance imaging and transcranial magnetic stimulation.

Predictive processing assumes that the brain is hierarchically organized, and that predictions at lower levels should accommodate the bottom-up sensory input as efficiently as possible. When predictions do not match the sensory input, the brain can either adjust the predictions or act upon the environment. Perceptual inference depends on the precision assigned to top-down predictions and bottom-up sensory input, and on the weighting assigned to specific predictions at all levels of processing.

Pyramidal cells are thought to play a key role in precision-weighting, where superficial pyramidal cells pass predictions forward and deep pyramidal cells pass predictions downward. Psychedelics are thought to affect deep pyramidal cells by stimulating 5-HT2ARs.

The densest expression of 5-HT2ARs can be found in the visual cortex and high-level association regions, and psychedelics affect the precision-weighting of top-down predictions in these areas. This is important because high-level predictions comprise our most fundamental assumptions. Psychedelics may affect lower- and intermediate-level predictions, but higher-level predictions might still smooth out their effects. For example, if higher-level predictions have little precision compared to that of lower-level predictions, we might actually see the wall breathing.

The REBUS hypothesis presumes that the brain is a hierarchically organized prediction machine that minimizes prediction error by accommodating bottom-up sensory input. Psychedelics weaken the precision of higher-up predictions, leading to psychedelic experiences.

The REBUS hypothesis is grounded in two theories: the entropic brain hypothesis and the criticality theory. The criticality theory argues that the brain can be ordered but still somewhat flexible, while the order-disorder theory argues that the brain can be highly ordered but also disordered.

Psychedelics increase entropy of spontaneous brain activity and decrease prominent low-frequency rhythms such as , , and . This leads to a decrease in top-down constrained brain activity and an increase in entropy, paving the way for psychedelic effects.

The entropic brain hypothesis can be naturally integrated into the predictive processing account of psychedelics, whereby stimulating 5-HT2ARs on deep pyramidal cells produces such entropic effects on spontaneous brain activity.

Psychedelics are believed to have remedial effects on mental disorders by decreasing the influence of high-level predictions and increasing the entropy of spontaneous brain activity. This allows the brain to access different brain states that were previously blocked by top-down override.

The anarchic brain is characterized by an attenuation of high-level predictions’ precision-weightings, a greater influence of bottom-up sensory input and a disinhibition of the limbic system, which can explain the intense and uncontrollable emotional experiences that usually come along with taking psychedelics.


Psychedelic-assisted treatment is embedded in a larger treatment, which typically is some form of psychotherapy. The prescription of psychedelics is thus also a psychotherapeutic intervention, and thus we have to consider how they work from a psychotherapeutic and thus a psychological perspective.

The present paper will focus on Wampold’s contextual model, which consists of one prerequisite and three pathways. The key element of such a bond is trust, which is of particular relevance in psychotherapy. The first encounter with the therapist is decisive for trust. The patient assesses the therapist’s trustworthiness based on the therapist’s appearance and the environment, as well as on the patient’s prior attitude toward psychotherapy.

When psychedelics are prescribed, the patient has already participated in multiple therapy sessions, so the prerequisite of an initial therapeutic bond has been fulfilled. The proposal of psychedelic treatment by the therapist does not jeopardize the initial therapeutic bond, as most psychedelic-assisted treatments are part of scientific trials and are executed within university psychiatric departments. Moreover, the patient is free to choose whether they want to take part in one or several psychedelic sessions.

The real relationship between the patient and the therapist is characterized by genuineness, realistic perceptions and confidence, safety, support and positivity. Three key characteristics of the therapeutic setting contribute to this, including confidentiality, non-disclosure of difficult material and empathic, appreciative and caring interactions.

Psychedelics seem to strengthen the real relationship between therapist and patient by elevating the patient’s feelings of connectedness to the therapist and by enabling the patient to experience uncontrollable experiences. This leads to new insights that are often of high personal relevance and meaning.

Expectations constitute the second pathway for therapeutic effects. Subliminal cues from the therapeutic setting can also elicit unconscious remedial expectations.

Psychotherapy creates new expectations for the patient by providing an alternative explanation for the patient’s mental problems. This explanation reinforces the patient’s feelings of self-efficacy and mastery, which in turn facilitates the initiation of change.

A recent meta-analysis showed that patients’ pre- or early treatment outcome expectations were positively correlated with their posttreatment outcomes. Moreover, there is indirect evidence from research on the alliance suggesting that expectations influence treatment outcomes.

The way in which expectations influence remedial effects has been extensively discussed in another research area: placebo studies. In open-label studies, giving participants a rationale for why placebos work appears to be crucial.

Psychedelic-assisted treatment provides a powerful basis to build up expectations. It provides a convincing pathological rationale and treatment rationale, and is embedded in preparation and integration sessions, indicating its therapeutic significance.

It has been argued that the effect sizes of psychedelics in randomized control trials are likely over-estimated because of high levels of response expectancy and deblinding of patients. Additionally, a placebo study has demonstrated that strong expectations can affect experience.

The third pathway involves specific ingredients. These ingredients promote new behavior, insights and perception that are favorable to health, and thereby reinforce recovery.

In psychedelic-assisted treatment, the psychedelic session is a specific ingredient. It is a well-defined specific ingredient because it is a conscious act, it has clear therapeutic effects, and it is well-attuned to the overall treatment.

Psychedelics fit well into Wampold’s contextual model of psychotherapy, strengthening the real relationship, increasing remedial expectations and being a well-defined therapeutic intervention.


Psychedelics induce an anarchic state in the brain, which is ideal for revisions of high-level predictions. However, whether these revisions ultimately lead to more functional high-level predictions depends on the bottom-up signaling and thus the “environment” generating them.

Humans are eusocial species with a basic need for social connection. This social connection helps us to maintain homeostasis by reducing free-energy and making use of positive interdependencies.

The therapist-patient relationship provides a highly controllable and supportive environment that helps the patient minimize prediction error. As time goes on, the patient gradually adapts to the therapeutic environment and updates their high-level predictions that were formed in less controllable environments.

Psychedelics appear to elevate feelings of connectedness to the therapist(s) when taken.

The precision-weightings of high-level predictions lighten and we become more sensitive to bottom-up sensory input when we are in a real relationship with a therapist. This process might not directly end after the acute psychedelic hot state has worn off.

Al Hubbard brought music, flowers and pictures into the treatment room to prime patients for a revelatory experience or to divert a trip when it took a bad turn. Timothy Leary emphasized the importance of the set in shaping psychedelic experiences.

In the last section, we discussed how expectations can have therapeutic effects. Predictive processing provides a prominent framework for how placebo effects work, and it argues that the more precise the prediction of remedial effects, the more likely prediction error gets minimized via active inference.

By inducing top-down predictions of amelioration and recovery, psychotherapy initiates inferential processes, which lead to a consequent experience of relief. Additionally, highly precise predictions of recovery launch a self-fulfilling prophecy, thereby acting on the body and its environment in such a way that they come true.

Psychedelics may interact with the pathway of expectations by making the brain more suggestible and elevating synaptic plasticity. This may allow induced expectations that have been suppressed by contrasting higher-level predictions to flow up the hierarchy and alter the generative model. After the acute psychedelic hot state, the patient is back in control and has to integrate the psychedelic experience into their life. Their expectations are essential for how they interpret their experiences.

Psychedelics are thought to boost the pathway of expectation because they induce a state where high-level predictions become revisable, and because they produce an unreal and ineffable experience that provides immense scope for interpretation.

Psychedelics are similar to placebos in that they give us what we expect to get from them. Psychotherapy is also a kind of placebo, but on rocket boosters.

Psychotherapy defines a patient’s set and setting, and psychedelics induce an anarchic state that boosts the pathway of the real relationship. Similarly, psychedelics boost the pathway of expectations, and the patient can find remedial cues when interpreting the psychedelic experience.

Three final notes: psychedelics are a specific ingredient of psychedelic-assisted treatment, so here we do not have an interaction between pharmacological and psychological effects.

Although indirect evidence for the REBUS hypothesis has been presented, there is also evidence that psychedelics increase top-down influence in some regions, namely from the parahippocampal gyrus to visual cortex. (2) The free-energy principle might not be an empirically testable theory, and (3) cognitive biases cast doubt on the assumption of a Bayesian brain. The authors of the paper evaluate the neurophysiological evidence for predictive processing and the entropic brain hypothesis, and conclude that the theories are promising but vulnerable to critique. The authors also propose new hypotheses that are speculative but testable by future research.

From the present paper: expectations about a psychedelic session should correlate with its therapeutic success; a familiar therapeutic environment; a real relationship between therapist and patient; and an alliance are all predictive of therapeutic success.


Psychedelic-assisted treatment works by stimulating 5-HT2ARs on deep pyramidal cells, making high-level predictions more prone to revision by the upwards flowing sensory input. This in turn makes rigid, dysfunctional high-level predictions more revisable, and the psychotherapeutic intervention becomes an essential source from which the bottom-up sensory input proceeds.

Sixty years ago, it was understood that set and setting had a significant influence on psychedelic experiences. This is what makes psychedelics and psychotherapy a promising combination.

Study details

Topics studied

Study characteristics
Theory Building

0 Humans

PDF of How Psychedelic-Assisted Treatment Works in the Bayesian Brain