REBUS Model: How Psilocybin Relaxes the Brain's Beliefs

Psilocybin Entropic Brain Theory & REBUS Model

N=11 first direct REBUS test

The first direct experimental test of REBUS predictions: 25 mg psilocybin decreased confidence in negative self-beliefs both acutely and at four weeks post-dose, while 1 mg produced no effect.

Zeifman et al., Scientific Reports, 2025

2019

year REBUS was published by Carhart-Harris & Friston

5-HT2A

receptor through which psilocybin relaxes precision-weighted priors

REBAS

revised beliefs persist at 4 weeks, not just during the session

Zeifman et al., Scientific Reports, 2025

View as image ↗

Most mental health treatments target symptoms. The REBUS model targets something deeper: the confidence with which the brain holds its own beliefs. REBUS (Relaxed Beliefs Under pSychedelics) is a neuroscience theory published by Robin Carhart-Harris and Karl Friston in 2019 that proposes a specific mechanism for how psilocybin produces psychological change. The brain maintains beliefs not just as abstract propositions but as active predictions it uses to filter all incoming experience. REBUS proposes that psilocybin temporarily reduces how tightly the brain holds those predictions, creating a window in which beliefs that would otherwise be self-reinforcing can be revised.

The model synthesizes predictive coding and the entropic brain hypothesis into a mechanistic account of why a drug with a six-hour duration of action can produce psychological changes lasting weeks or months. For the broader context of where REBUS fits in the entropic brain framework, see The Entropic Brain: How Psilocybin Increases Neural Disorder to Heal. And for the empirical support connecting entropy measurement to clinical outcomes, see How Psilocybin Increases Brain Entropy (And Why That's Good).

What Predictive Coding Is (and Why It Matters)

To understand REBUS, you need one core concept from neuroscience: the brain is not a passive receiver of sensory input. It is an active predictor.

At every moment, higher brain regions are sending predictions downward to lower regions: "given what I know about the world, here is what I expect the next sensory signal to be." When incoming data from the senses matches those predictions, the mismatch is small and is suppressed; the brain treats it as non-news. When incoming data does not match predictions, a larger prediction error signal travels upward toward the higher regions, which then update their predictions accordingly.

This process is called predictive coding, and it explains a great deal about how experience works. Your ability to walk down a familiar street without consciously registering every detail is predictive coding at work: the brain's model of the environment is generating accurate predictions, so little prediction error needs to be processed. The moment something unexpected happens (a person behaves strangely, a car appears where none was expected), a large prediction error breaks through and captures conscious attention.

A key feature of this system is what researchers call precision weighting: how much confidence the brain assigns to any given prediction. High-precision predictions are held rigidly; incoming data that conflicts with them is discounted before it can trigger meaningful updating. Low-precision predictions are held tentatively; conflicting data gets through more easily. Precision weighting determines how open the brain is to revising its current model of the world in response to new evidence.

REBUS: Relaxed Beliefs Under pSychedelics

Carhart-Harris and Friston's 2019 paper in Pharmacological Reviews proposed a specific account of what psilocybin does to this system.^[2]

Psilocybin activates serotonin 2A (5-HT2A) receptors. These receptors are densely expressed in the high-level cortical areas that generate top-down predictions, specifically the regions of association cortex and prefrontal cortex that hold the brain's current model of itself and the world. When psilocybin activates these receptors, the precision weighting of top-down predictions decreases. The brain continues to generate predictions, but it holds them with less confidence. They loosen their grip.

The consequence is that bottom-up signals travel upward more freely. The limbic system (which processes emotional responses, interoception, and raw sensory input) sends signals upward that are normally suppressed or minimized by high-level prior beliefs. Under psilocybin, those signals break through the filtration that normal precision weighting would apply.

Carhart-Harris and Friston called the resulting brain state the "anarchic brain": when top-down priors lose their normal dominance, the brain's hierarchical command structure temporarily flattens. Perception and experience become more driven by incoming, bottom-up information than by accumulated expectations and beliefs.

The REBUS Mechanism

What Psilocybin Changes Is the Balance of Influence, Not the Machinery

Ordinary consciousness

Top-down control dominates

High-level priors are held with high precision. Bottom-up prediction errors that conflict with them are discounted before they can update the model — so entrenched beliefs, including negative self-beliefs, keep reinforcing themselves.

Under psilocybin — the "anarchic brain"

The hierarchy flattens

5-HT2A agonism in high-level cortex lowers the precision weighting of top-down priors. The same priors are still generated, but held loosely, so bottom-up limbic and sensory signals flow up more freely — opening a window in which rigid beliefs can be revised.

Empirically, when precision falls the effect is measurable: 25 mg psilocybin (but not 1 mg) reduced confidence in negative self-beliefs, and greater session entropy tracked greater belief revision (Zeifman et al., 2025).

Sources: Carhart-Harris & Friston, Pharmacological Reviews (2019); Zeifman et al., Scientific Reports (2025)

What the Anarchic Brain Actually Means

"Anarchic" sounds like chaos, but the REBUS sense is more specific. An anarchic brain is one in which the normal hierarchy of influence (high-level beliefs dominating lower-level sensory processing) is temporarily suspended. The high-level regions continue to function. The bottom-up signals continue to be generated. What changes is the relationship between them.

In ordinary consciousness, most of what you experience is shaped by what you expect. Your visual system does not hand raw photons to consciousness; it hands you a processed, expectation-filtered interpretation of the visual scene. Your emotional system does not inject raw fear signals into consciousness; those signals pass through multiple layers of appraisal and suppression before reaching awareness.

The default mode network sits at the top of the precision hierarchy. Its priors about the self generate the most persistent top-down filtering in waking life. For how psilocybin quiets the DMN specifically, see Psilocybin and the Default Mode Network: How It Quiets the Ego.

Under psilocybin, according to REBUS, the filtration becomes less thorough. Bottom-up signals that would ordinarily be suppressed as prediction errors reach higher-level processing. This is why the psilocybin experience includes visual phenomena, unexpected emotional surges, unusual associative connections between thoughts, and a loosening of the ordinary boundaries between categories of experience. These are not hallucinations in the clinical sense; the person typically knows they are under a drug. They are the result of a brain temporarily processing more of its own signal, less filtered by prior expectations.

Therapeutically, the anarchic state matters because it extends to self-referential beliefs. Beliefs about the self ("I am worthless," "I cannot change," "I am unlovable") are normally held with very high precision. They shape how self-relevant information is processed, filtering evidence that could challenge them. In an anarchic brain state, even these high-precision beliefs lose some of their hold, creating a window in which the evidence for self-revision can be received and processed.

The 2018 Bridge: How the Entropic Brain Theory Led to REBUS

The REBUS paper did not emerge from nowhere. Carhart-Harris had been developing the theoretical architecture for years, and the 2018 revision of the entropic brain theory is the most direct bridge.^[1]

The 2014 entropic brain theory described psychedelic states as high-entropy states and proposed that sub-critical, low-entropy brain states underlie depression and related conditions. It provided a framework for thinking about the therapeutic potential of entropy increase but stopped short of specifying the mechanism.

The 2018 revision addressed this gap. Carhart-Harris provided more specific entropy measures, proposed new clinical studies to test the framework, and began explicitly connecting entropy to the predictive coding literature that Friston had developed. The 2018 paper was the theoretical scaffolding on which REBUS was built: from "entropy increases" to "here is the mechanism by which entropy increase produces therapeutic belief revision."

This progression (2014 observation to 2018 mechanism-building to 2019 full model) shows that REBUS is not a standalone hypothesis but the current front of a decade-long theoretical development.

Clinical Evidence the Theory Predicted: TRD fMRI

The REBUS model made specific predictions about what psilocybin should do to brains in depression. Carhart-Harris et al.'s 2017 fMRI study in treatment-resistant depression provided data consistent with those predictions.^[3]

Nineteen patients with treatment-resistant depression received psilocybin therapy, with quality pre- and post-treatment fMRI obtained from 16. Post-treatment scans showed decreased cerebral blood flow in the amygdala (a limbic structure central to threat processing and negative emotional responses), which correlated with reduced depressive symptoms, alongside increased resting-state connectivity within the default mode network. Shifts in connectivity — increased ventromedial prefrontal-to-parietal coupling and decreased parahippocampal-to-prefrontal coupling — predicted treatment response at five weeks.

REBUS explains these changes. Psilocybin reduced the precision weighting of top-down beliefs, including the negative self-evaluations that characterize depression. With those beliefs loosened, bottom-up limbic signals (including the amygdala's threat-signaling) were processed differently; the amygdala's hyper-vigilant responses were no longer being sustained by top-down expectations of threat. The shifts in resting-state connectivity that tracked clinical response are consistent with a post-acute reorganization of the networks that sustain depressive thought patterns, which the authors framed as a network "reset." These changes align with what the model anticipates.

The limitation is that this was an open-label study without a placebo arm, so the brain changes cannot be attributed solely to psilocybin's pharmacological mechanism versus expectation effects. The design predates REBUS itself, so it is consistent with the theory rather than a prospective test of it.

The First Empirical Test: What REBAS Found

By 2025, REBUS had accumulated substantial circumstantial support but no direct empirical test of its core prediction: that psilocybin specifically reduces the precision of existing beliefs and enables their revision.

Zeifman et al. designed the first study aimed directly at this question.^[4]

Eleven healthy participants received both 1 mg and 25 mg psilocybin in a within-subjects design, the two doses given four weeks apart, with confidence in a set of personally held beliefs measured before, during, and four weeks after each dose. The 1 mg dose was below the threshold for any psychedelic effects, chosen to test whether the belief changes required a full psychedelic experience or could occur with minimal pharmacology.

The 25 mg dose significantly decreased confidence in the participants' own negative self-beliefs during the session. This belief confidence decrease persisted when participants were reassessed at four weeks post-dose. The 1 mg dose produced no such effect: no significant belief confidence change during the session, no change at four weeks.

Three correlations were found in the 25 mg condition: entropy increase during the session correlated with belief confidence decrease; the subjective experience intensity during the session correlated with belief confidence decrease; and the magnitude of belief confidence decrease strongly predicted four-week well-being gains.

Zeifman et al. extended the REBUS framework to name this process REBAS: Revised Beliefs After pSychedelics. REBUS describes the acute relaxation of beliefs during the session. REBAS describes the lasting revision of those beliefs in the weeks afterward.

The study's limitation is the small sample size (11 participants). Effect sizes in small studies are often inflated relative to replication in larger samples. But the dose-response pattern (1 mg nothing, 25 mg everything) is strong evidence against a purely expectation-driven interpretation.

What REBUS Predicts About Which Conditions Should Respond

The model's therapeutic logic applies to any condition in which pathologically over-precise beliefs maintain the disorder.

Depression fits the REBUS pattern directly: beliefs about the self and future are held with pathological precision, filtering incoming evidence that could challenge them. Addiction also fits: the belief that the rewarding stimulus is necessary overrides other value representations with abnormal precision. OCD has a related structure: beliefs about threat and contamination are held with such confidence that disconfirming evidence (nothing bad happened when I did not wash my hands) fails to update the belief.

REBUS does not predict equal efficacy across all these conditions. The precision weighting of different belief types involves different neural circuits, and psilocybin's 5-HT2A mechanism is concentrated in particular cortical regions rather than all of them equally. But the broad prediction is that conditions characterized by overly rigid, high-precision belief structures should be more amenable to REBUS-type interventions than conditions with different underlying mechanisms.

This is a testable prediction that the current literature has not yet fully examined. Clinical trial data on psilocybin for OCD and addiction is at earlier stages than the depression evidence.

The Academic Debates: Where REBUS Has Been Challenged

Rankaduwa and Owen (2023) published a systematic review of debates within the academic literature about the entropic brain theory and its extensions, including REBUS.^[5]

Several concerns appear in the peer literature. One is whether psychedelic states are accurately described as "higher" on a consciousness richness scale, or whether they are better understood as qualitatively different states with their own characteristics, neither higher nor lower than ordinary consciousness. The therapeutic benefit of psilocybin does not necessarily require that it "expands" consciousness in a value-laden sense; it may simply restructure which neural patterns are active.

A second debate concerns whether entropy is the right measure to capture what is therapeutically significant. Entropy measures unpredictability of neural signal patterns. The REBUS model uses entropy to operationalize the degree of precision-relaxation. But precision weighting operates at the level of belief content, which may not map cleanly onto signal diversity as a measure. Validating the link between MEG signal diversity and subjective belief confidence requires studies like Zeifman 2025, which are only beginning to appear.

A third debate is taxonomic: Rankaduwa and Owen argue that the classification of psychedelic states requires a more fine-grained taxonomy than EBT currently provides, distinguishing between different types of altered consciousness rather than treating all psychedelic states as points on a single entropy axis.

These are legitimate scientific debates and not simply skepticism from outside the field. Engaging with them is consistent with the evidence-based approach the REBUS framework itself claims. The model is the best current mechanistic account of psilocybin's therapeutic effects, but it is a theory under active development, not a settled framework. For the view from the clinical data, see Psilocybin for Depression: Results from the Largest Clinical Trials.

The Bottom Line

REBUS (Relaxed Beliefs Under pSychedelics) proposes that psilocybin reduces the precision-weighting of top-down beliefs via 5-HT2A agonism, liberating bottom-up limbic signals in what Carhart-Harris and Friston (2019) called the anarchic brain. The 2018 EBT revision bridged entropy data to predictive coding, setting up REBUS conceptually. The 2017 TRD fMRI study showed brain changes consistent with REBUS predictions. The 2025 REBAS study (Zeifman et al., N=11) provided the first direct empirical test: 25 mg psilocybin (not 1 mg) decreased negative self-belief confidence, session entropy correlated with belief change, and belief change predicted four-week well-being gains. Debates remain about whether entropy is the right measure and how to taxonomize psychedelic states, as documented in Rankaduwa and Owen (2023).

Frequently Asked Questions

What does REBUS stand for in psilocybin therapy?

REBUS stands for Relaxed Beliefs Under pSychedelics. It is a theory published by Carhart-Harris and Friston in 2019 proposing that psilocybin temporarily reduces the confidence with which the brain holds its top-down predictions and beliefs. By activating serotonin 2A receptors in high-level cortical areas, psilocybin loosens the precision weighting of existing beliefs, allowing bottom-up signals from the limbic system to influence experience more freely.

What is predictive coding in neuroscience?

Predictive coding is a framework describing the brain as a prediction machine. Higher brain regions constantly generate predictions about incoming sensory data and send those predictions downward. When incoming data matches predictions, the match is suppressed. When incoming data violates predictions, a prediction error signal travels upward, triggering belief updating. Precision weighting determines how confident the brain is in its predictions and how readily it updates them when evidence conflicts.

How does the REBUS model explain psilocybin's therapeutic effects?

REBUS proposes that depression, OCD, and addiction involve pathologically over-precise top-down beliefs that filter incoming evidence and maintain themselves. Psilocybin reduces the precision weighting of these beliefs via 5-HT2A receptor activation, creating a window in which they can be revised. The 2025 REBAS study (Zeifman et al.) directly tested this: 25 mg psilocybin decreased confidence in negative self-beliefs during the session and at four weeks, with entropy during the session correlating with the degree of belief revision.

What is the anarchic brain?

The "anarchic brain" is the term Carhart-Harris and Friston (2019) used to describe the psilocybin brain state in which top-down beliefs have temporarily lost their normal dominance over bottom-up sensory and emotional processing. In ordinary consciousness, high-level priors strongly filter incoming signals. In an anarchic brain, that filtration is reduced: bottom-up limbic signals, raw emotional responses, and sensory signals that would normally be suppressed reach higher-level processing more freely. This produces the unusual perceptual and emotional content of the psychedelic experience.

Is there evidence for the REBUS model?

The direct empirical evidence is limited but promising. Zeifman et al. (2025) conducted the first prospective test of REBUS predictions: 25 mg psilocybin (but not 1 mg) decreased confidence in negative self-beliefs, with this decrease correlated with session entropy and persisting at four weeks post-dose. Carhart-Harris et al. (2017) provided indirect evidence through TRD fMRI: post-psilocybin brain changes (decreased amygdala blood flow, increased within-default-mode-network connectivity) are consistent with what REBUS predicts but were not designed as REBUS tests. Rankaduwa and Owen (2023) review ongoing academic debates about the model.

Sources & References

1. 1RTHM-00943·Carhart-Harris, Robin L (2018). “The entropic brain - revisited..” Neuropharmacology.Study breakdown →PubMed →↩
2. 2RTHM-01038·Carhart-Harris, R L et al. (2019). “REBUS and the Anarchic Brain: Toward a Unified Model of the Brain Action of Psychedelics..” Pharmacological reviews.Study breakdown →PubMed →↩
3. 3RTHM-00463·Carhart-Harris, Robin L et al. (2017). “Psilocybin for treatment-resistant depression: fMRI-measured brain mechanisms..” Scientific reports.Study breakdown →PubMed →↩
4. 4RTHM-02216·Zeifman, Richard J et al. (2025). “From relaxed beliefs under psychedelics (REBUS) to revised beliefs after psychedelics (REBAS)..” Scientific reports.Study breakdown →PubMed →↩
5. 5RTHM-01773·Rankaduwa, Sidath et al. (2023). “Psychedelics, entropic brain theory, and the taxonomy of conscious states: a summary of debates and perspectives..” Neuroscience of consciousness.Study breakdown →PubMed →↩