1 Title and one-sentence summary

Effects of Attentional Refocusing and Self-Acceptance on Perceptual Priors and Sensitivity

A four-arm randomized trial involving 360 adults will test whether four weeks of training reduces specific perceptual prior weights, while separating changes in sensory sensitivity, response criterion, and task learning.

Applicant: Future Mind Institute, a US nonprofit; priority A, consciousness modeling and multiple levels of reality. This expanded brief follows the requested length; conventional typesetting will exceed one page.

2 Scientific question and background

Can training change the perceptual computations measured in a controlled task? Raz et al. (2005) showed that suggestion reduced Stroop interference and associated brain activity in highly hypnotizable participants. This supports modulation of some automatic processing, without establishing wholesale removal of perceptual assumptions. Carhart-Harris and Friston (2019) proposed reduced precision of high-level priors as a theoretical mechanism; this account cannot be directly extrapolated to nonpharmacological training.

Cascio et al. (2016) associated self-affirmation with self-related and valuation systems. In a randomized study of 48 participants, Gu et al. (2019) found changes in early feedback-related ERP components but no P3 effect. Neither study demonstrated improved sensory thresholds. The opportunity in 2026 is to use computational psychophysics to distinguish prior influence, sensory evidence, and decision bias, giving otherwise broad claims explicit experimental boundaries.

3 Origin of the hypotheses

The Seth Material serves strictly as a hypothesis generator. Claim alternate-focus-2, in seth-speaks §19, proposes that alternating focus can loosen assumptions that constrain perception. Claim affirmation-11, in personal-reality §675, proposes that self-affirmation can permit broader perceptual engagement. We operationalize these suggestions as attentional refocusing and self-acceptance interventions. Acceptance is not identical to the affirmation manipulations in the cited studies; that translation is itself an assumption under test.

Provenance is recorded in the supplied dataset, DOI 10.5281/zenodo.22484474, licensed CC BY 4.0. These source statements provide neither empirical evidence nor a distinctive quantitative computational model. Their contribution is a pair of candidate intervention pathways.

4 Primary hypotheses and falsifiable predictions

Each intervention is predicted to reduce prior weighting relative to expectancy-matched attention training. Prior weight and sensory sensitivity, d′, are co-primary outcomes. Increased d′ constitutes a separate sensory-improvement prediction: reducing reliance on a valid prior does not necessarily improve performance.

The statistical null is no between-group difference in change. A competing mechanistic model permits changes in response criterion, lapses, and learning while holding prior and sensory parameters unchanged. Planning assumptions are a −0.50 baseline-standard-deviation effect on prior weighting and a +0.50 effect on d′. The smallest meaningful effect is an absolute standardized difference of 0.30. These values are design assumptions, not estimates supplied by the source material or existing studies.

Mechanistic support requires recoverable parameters and validation on held-out tasks. A criterion change alone does not support altered perceptual computation. This distinction prevents more frequent reports of seeing a signal from being mistaken for better detection.

5 Research design

Randomize 360 adults equally to attentional refocusing, self-acceptance, expectancy-matched attention training, or neutral activities. Conceal allocation and stratify by site and baseline performance. Training lasts four weeks, with 20-minute sessions on five days each week and matched contact time. Refocusing alternates attention between local and global features; acceptance practices nonjudgmental awareness; the active control practices sustained attention; the neutral condition presents general knowledge.

At baseline, week four, and week eight, participants complete visual signal-detection tasks independently crossing signal probabilities of 25%, 50%, and 75% with three sensory-noise levels. Include signal-absent trials, counterbalanced response keys, and fixed incentives. Calibrate individual stimulus intensity at baseline and retain that intensity thereafter. A two-interval task without probability cues and a held-out stimulus set assess transfer. Equalize practice and exclude assessment stimuli from training.

Allowing 15% attrition leaves approximately 76 participants per arm. For four primary tests, a conservative two-sided alpha of .0125 and baseline–posttest correlation of .50 give approximately 80% power for a standardized effect of .47 under a baseline-adjusted approximation. We therefore plan around .50. Smaller effects and equivalence conclusions may remain underpowered.

Before confirmatory recruitment, generative simulations must establish power, false-positive rates, and parameter recovery for the actual task and analysis. Failure triggers design revision and renewed preregistration. An independent feasibility sample will not enter the confirmatory analysis.

Participants will not be told which intervention is favored; complete masking of training format is impossible. Assessors, EEG staff, and analysts remain blind to allocation. Preregistration covers intervention scripts, models, exclusions, contrasts, multiplicity, and stopping rules. Expectancy and adherence are measured to assess the credibility and delivery of the interventions.

6 Data and analysis plan

Week four is the primary endpoint. Baseline-adjusted comparisons test each intervention against the active control on prior weight and d′, with Holm correction across the four primary tests. A hierarchical generative model jointly estimates sensory gain, prior weighting, response criterion, lapses, and learning. Parameter uncertainty is propagated into group comparisons, with signal-detection estimates providing a cross-check.

Simulation must establish whether the model can distinguish prior weighting from a criterion that changes with probability cues. If these explanations remain observationally indistinguishable, the study will not assign a perceptual mechanism to the result. This identification requirement is central to the project's adjudicative value.

Secondary outcomes include persistence at follow-up, transfer across tasks, criterion, response time, and comparisons with the neutral group. A prospectively stratified random subsample of 120 participants, 30 per arm, receives EEG assessment. The secondary EEG endpoint is posterior P1 amplitude 80–130 milliseconds after stimulus onset, using electrodes and artifact rules fixed in advance. EEG is supporting evidence and cannot independently adjudicate the mechanism.

Use intention-to-treat analysis with sensitivity analyses for missing-data assumptions; adherence-based analyses are exploratory. Release de-identified behavioral data, shareable EEG, dictionaries, code, and synthetic data. Raw files that cannot safely be made public will have controlled access. The resulting package should allow independent investigators to reproduce parameter recovery, primary analyses, and model comparisons.

7 Milestones and duration

The project lasts 18 months. Months 1–3 establish ethics approval, intervention manuals, and the adversarial agreement. Months 4–5 complete the independent pilot, simulations, and preregistration. Months 6–12 cover recruitment and training; months 13–14 complete follow-up. Months 15–16 deliver blinded analyses. Months 17–18 cover unmasking, public data release, and manuscript submission, including publication of null findings.

Progression to the main trial depends on the simulation and feasibility gate, rather than the appearance of promising pilot effects.

8 Budget scale

Request US$450,000, within the proposed US$300,000–550,000 range:

• Personnel: $200,000.
• Participant compensation and recruitment: $70,000.
• EEG facilities and technical support: $50,000.
• Task development, pilot work, and computing: $40,000.
• Independent methodological review and open outputs: $20,000.
• Indirect costs and contingency: $70,000.

This estimate assumes access to existing equipment. Final rates and institutional overhead arrangements require confirmation by the administering institution.

9 PI profile and candidate teams or institutions

The PI should have demonstrated expertise in computational psychophysics, hierarchical modeling, and randomized intervention research, supported by an EEG specialist and an independent statistician. Candidate collaborators are Philip R. Corlett and Yale's Belief, Learning, and Memory Lab for prior modeling; Emily B. Falk and Penn's Communication Neuroscience Lab for intervention specification and expectancy controls; and Devin B. Terhune at King's College London for suggestion and subjective-report methodology.

These are proposed roles, with no commitment implied. Proponents of intervention effects and proponents of the criterion-only explanation should jointly specify competing models and sign off on success and failure criteria before data collection. This creates an adversarial collaboration around a precise, testable disagreement.

10 Risks and failure criteria

The principal risks are parameter confounding, practice effects masquerading as intervention effects, and failure to generalize beyond one task. If simulations cannot identify the relevant mechanisms, confirmatory recruitment pauses pending redesign.

If multiplicity-adjusted equivalence tests place both interventions' effects on prior weight and d′ within ±0.30 standardized units, abandon the claim of a practically meaningful effect at this dose. Wide intervals instead imply uncertainty, not demonstrated absence. Criterion-only changes, effects confined to the trained task, or isolated exploratory EEG findings do not support a generalized perceptual mechanism.

Even positive results would challenge only a specified model in which these interventions change reporting without changing measured perceptual computation. They would not refute predictive processing as a whole. Threshold changes also cannot establish an additional information channel.

11 Ethics and compliance

Obtain IRB approval and informed consent before recruitment. Provide clear withdrawal rights, compensation arrangements, and procedures for responding to discomfort. Disclose funding and the intellectual provenance of the hypotheses without promoting beliefs or claiming therapeutic benefit.

Future Mind Institute will act as a neutral consciousness-and-life-sciences funder. Its Seth Material Research Program is one public program within the institution. Agreements must protect independent analysis and complete publication, including unfavorable findings; the sponsor cannot suppress null results. Data sharing must remain consistent with consent and privacy obligations.

12 Fit with existing funders

Bial's scientific research program provides a plausible fit through psychophysiology and the combined behavioral–EEG design. Fetzer's stated mission connects science with human flourishing, making the acceptance intervention potentially relevant without turning this trial into a test of that institution's worldview.

Templeton World Charity Foundation's Cogitate program supplies a methodological precedent for preregistration, open science, and adversarial collaboration. BICS, UVA DOPS, and IONS are additional reference points for future dialogue, rather than interchangeable funding channels. These are assessments of thematic or methodological fit, not statements that a call is open, that eligibility has been established, or that support has been secured.