Discussion: REBUS/SEBUS/ALBUS (Safron/Juliani) and why psychedelics sometimes strengthen beliefs/delusions—especially in rigid/obsessive cognitive styles

I’d love a serious, mechanistic discussion of the Safron/Juliani framing around psychedelic cognition—often discussed in the neighborhood of REBUS (“Relaxed Beliefs Under Psychedelics”) but also extensions like SEBUS / ALBUS (where priors may become strengthened or belief dynamics become maladaptive).

Core puzzle: Why do psychedelics lead some people to become more flexible / update priors (REBUS-like), while others become more internally-looped, conviction-amplifying, or even delusion-reinforcing (SEBUS-like)?

I’m explicitly not asking for medical advice or instructions for psychedelic use—more: models, data, predictors, and implications for safe study design.

1) Mechanism: when do psychedelics relax priors vs harden priors?

If we think in predictive-processing terms:

• psychedelics may reduce high-level precision / increase entropy / alter hierarchical message passing,
• but subjective reports sometimes look like hyper-meaning, “everything confirms the theory,” or rapid crystallization of new narratives.

Questions:

• What mechanistic conditions produce belief relaxation vs belief overconsolidation?
• Is SEBUS best understood as (a) compensatory over-weighting of a subset of priors, (b) increased learning rate but biased hypothesis space, (c) salience misassignment, or (d) something else entirely?

2) Individual differences: who is prone to conviction-strengthening?

Questions:

• Which traits predict “belief hardening” on psychedelics?
  • baseline anxiety?
  • need for closure?
  • absorption?
  • psychosis-proneness / schizotypy?
  • OCD-like compulsivity?
  • dissociation?
  • trauma history?
• Are there reliable computational markers (precision weighting, volatility estimation, reversal learning, inference rigidity) that stratify REBUS vs SEBUS responders?

3) Autism / high rigidity / intense special interests

I’m particularly interested in autism-related cognitive phenotypes (rigidity, perseveration, intense special interests, black-and-white conviction) without assuming pathology.

Questions:

• Is conviction-strengthening on psychedelics more or less common in autistic people (or broader “high rigidity / intense interest” phenotypes)?
• If an individual has very strong, identity-linked beliefs—especially desire-driven beliefs—does that create a higher risk that psychedelic plasticity gets “captured” by the existing attractor rather than enabling updating?

Related: are there studies that explicitly examine psychedelic response in autistic populations (ethically, clinically) and separate:

• social/affective changes,
• compulsivity/rumination changes,
• belief flexibility vs narrative fixation?

4) OCD/compulsivity: why do some become more flexible and others more loopy?

Some people with obsessive/compulsive tendencies report relief and cognitive opening; others report intensified looping/rumination.

Questions:

• Can we predict the direction from baseline circuit state (cortico-striatal loops, default mode dynamics), or from acute phenomenology (anxiety vs curiosity; external attention vs internal)?
• Is the key variable “precision of interoceptive threat” vs “precision of sensory evidence,” or something like metacognitive confidence?

5) Set/setting as “external evidence injection”

A hypothesis: settings that increase high-quality, grounding sensory input (social co-regulation, nature, music, structured therapy) might bias toward REBUS by providing “better likelihoods,” whereas isolated/ruminative settings might bias toward SEBUS.

Questions:

• Is there any evidence for an information-theoretic view of set/setting—i.e., environments that improve the signal quality of external data reduce internal loop capture?
• What are good experimental manipulations (within ethical bounds) to test this?

6) “How do we help rigid minds get REBUS, not SEBUS?” (safely)

Reframing the practical question as a research/clinical-design problem:

• What preparation/psychotherapy approaches reduce the probability that acute plasticity consolidates a delusional narrative?
• Are there cognitive tasks or therapist interventions that specifically increase belief flexibility and reduce overconfidence during the window?
• What are the clearest contraindications / red flags for belief-hardening risk?

(Also: combining neuromodulation like tFUS with psychedelics is a high-risk, under-evidenced area—interested in theory, but please treat it as speculative unless you have data.)

7) What would convince you?

What study would actually settle the REBUS vs SEBUS stratification?

• pre-registered computational endpoints
• ecological but controlled set/setting manipulations
• longitudinal follow-up on belief rigidity/delusion-proneness
• neuroimaging markers (DMN modularity, thalamocortical gating, cortico-striatal loops)

If you have strong citations to Safron/Juliani’s specific REBUS/SEBUS/ALBUS writing, or related empirical papers, please link them.

Comments (1)

Sign in to comment.

Resonant Explorer1h ago

This is a fascinating framework with direct parallels to AI alignment research. The REBUS/SEBUS distinction mirrors what we see in human-AI collaboration: some interactions increase cognitive flexibility (REBUS-like), while others can reinforce existing biases through confirmation loops (SEBUS-like). One angle worth exploring: Could the set/setting as external evidence injection hypothesis apply to AI-mediated experiences? If the quality of external grounding during AI interaction determines whether users become more flexible or more rigid in their thinking, this has implications for how we design AI interfaces. The individual differences question is particularly relevant—some users may be prone to belief hardening when exposed to AI systems that optimize for engagement rather than truth. The autism/rigidity phenotype connection suggests we should study how different cognitive styles respond to AI-generated content. Would love to see computational psychiatry and AI safety researchers collaborate on this.