1 Title and one-sentence summary

Discovery-Method Labels and Funding Judgments for Methodologically Matched Research Proposals

A randomized reviewer experiment will test whether analytical, intuitive, or prior anomalous-experience discovery labels change the fundability of methodologically matched proposals, while distinguishing label effects from reasonable informational inferences and procedural bias.

Applicant: Future Mind Institute, a US nonprofit positioned as a neutral consciousness-and-life-sciences funder. Priority: A. Field: Coordination of intuitive and analytical cognition. Duration: 18 months. Request: US$320,000. This expanded brief follows the requested word range and will exceed a conventional single printed page.

2 Scientific question and background

Does an account of how an idea originated influence scientific evaluation when the research question, supporting evidence, and proposed methods are otherwise identical? This question is narrower than whether scientific institutions resist unconventional thinking, and more directly actionable for a foundation designing review procedures.

Bromham, Dinnage, and Hua (2016) found lower funding success associated with interdisciplinarity in Australian funding data. Sun, Livan, Ma, and Latora (2021), examining more than 44,000 UK grants, found stronger long-term funding performance among interdisciplinary researchers. Their units of analysis and time horizons differ. Neither study identifies a causal discovery-label penalty, and neither supports a blanket claim that science excludes nonanalytical thinking. These are the only scholarly studies cited as background here. Bromham et al., 2016; Sun et al., 2021.

The proposed experiment isolates a manipulable feature of review materials. Its immediate value is to help the foundation establish auditable procedures from 2026 onward. It does not investigate whether anomalous experiences provide reliable knowledge.

3 Origin of the hypothesis

The hypothesis is extracted from two methodological claims in the Seth Material: a-1-8, The Magical Approach, source key magical-approach, section pending verification, concerning the dominance of rational frameworks and disciplinary fragmentation; and unconscious-9, source key esp-power, §14, advocating legitimate investigation of unconscious and inner experience.

These texts generate a research question; they provide no empirical support for the predicted effect. The broad institutional claims are reduced to a falsifiable proposition about labels in a particular evaluation task. No doctrinal claim is an experimental premise.

The supplied provenance identifies a Zenodo dataset, DOI 10.5281/zenodo.22484474, licensed CC BY 4.0, and the program website. Neither resource could be retrieved during preparation. Dataset version and passage locations therefore remain submission checks. In particular, a-1-8 is a claim identifier, not a verified book section; its missing section must be checked before formal submission rather than invented.

4 Primary hypothesis and falsifiable predictions

The primary prediction is that a prior anomalous-experience label reduces fundability relative to an analytical-discovery label. An intuitive-discovery comparison is the second confirmatory contrast. “Historical experience” is operationalized as an unusual subjective experience previously reported by the hypothetical researcher, without naming a celebrated historical scientist.

The null hypothesis is that both average label contrasts equal zero. The planning effect is d = −0.20; the smallest effect of practical interest is |d| = 0.15. With a planning outcome standard deviation of 20 points, these correspond to four and three points on the fundability scale. These values are design assumptions, not estimates derived from the supplied literature.

Higher-quality methods should receive higher scores. A generalizable penalty should recur in the independent disciplinary samples. A nonsignificant result alone will not establish absence: practical absence requires prespecified equivalence tests around zero.

5 Research design

The study uses a crossed randomized three-label by two-method-quality design. It targets 960 eligible completers: 480 life-science reviewers in the main study and two independent validation samples of 240 reviewers each, covering psychological/behavioral science and physical science/engineering. Recruitment allows approximately 10% attrition. Eligibility requires manuscript or grant-review experience within five years; both types of experience are recorded separately.

Each reviewer evaluates 12 distinct abstracts, with two assignments per factorial cell. Nobody sees multiple versions of the same abstract. The main study uses 48 base abstracts; each validation study uses 24 different abstracts, with no overlap between studies. This provides both independent reviewers and independent proposal materials for validation.

Within a base abstract, qualifications, prior evidence, feasibility, and budget remain fixed. Discovery wording and method quality are manipulated independently. All versions explicitly state that the discovery account explains idea generation, supplies no supporting evidence, and adds no information about researcher competence. High- and low-quality methods differ through features such as sample adequacy, controls, or identification strategy. Independent experts pretest these differences and the credibility of the discovery descriptions.

Power planning assumes standardized variance components of 0.25 for reviewers, 0.10 for abstracts, and 0.65 for residual variation, with label random-slope standard deviations of 0.15 at both levels. Under a balanced-design analytical approximation, a label contrast has a standard error of approximately 0.035 in the main study and 0.049 in each validation study. At two-sided α = .025, idealized power for d = .20 exceeds 90%.

These are provisional calculations, not completed simulation results. Before recruitment, simulation must incorporate the full allocation, random-effect covariance structure, and missingness scenarios. Acceptance targets are at least 90% power for the main study and 80% for each validation sample. If the design cannot meet these targets within budget, it must be revised and reviewed before launch; the meaningful-effect threshold will not be reduced to make the design appear adequate.

Reviewers are unaware of the directional hypothesis. Analysts receive masked condition codes and unlock them only after the analysis code is fixed. Preregistration covers materials, allocation, eligibility, exclusions, stopping, contrasts, multiplicity, model simplification, and failure criteria. Collaborators favoring a penalty interpretation and those favoring informational explanations jointly sign the predictions and interpretation rules.

6 Data and analysis plan

The primary outcome is a 0–100 fundability rating. The two confirmatory label contrasts use Holm correction. A crossed mixed-effects model includes label, method quality, and their interaction, with reviewer and abstract random intercepts and identifiable random slopes. Two-way cluster-robust inference by reviewer and abstract provides a robustness check. A predetermined simplification sequence addresses singular fits without selecting the model that produces the strongest result.

Validation samples are analyzed and reported independently. Pooled results cannot substitute for unsuccessful replication. Effect sizes and confidence intervals accompany significance tests, and equivalence tests use bounds of ±0.15 SD. Missing-data sensitivity analyses and the label-by-quality interaction are reported explicitly.

Secondary outcomes include a binary funding recommendation, proposal credibility, inferred researcher competence, perceived evidential sufficiency, and recognition of method quality. Mechanism questions follow the primary rating to reduce priming. These post-treatment measures are not covariates in the primary total-effect model. Exploratory mediation cannot, by itself, identify a causal psychological mechanism.

The project releases deidentified rating data, experimental materials, code, registrations, and a deviation log. Fine-grained professional characteristics that create reidentification risks are restricted rather than included in the public dataset.

7 Milestones and duration

Months 1–3: Obtain ethics review, establish adversarial collaboration, and verify source provenance. Months 4–5: Pretest materials, complete power simulations, and lock the registration. Months 6–10: Recruit and complete the main study. Months 11–14: Conduct independent disciplinary validation under the locked protocol. Months 15–18: Unmask analyses, release approved data, and deliver findings and review guidance.

Pilot observations are excluded from confirmatory analyses. There is no early stopping based on interim label effects. Deliverables include a reusable audit protocol, a complete results report, and guidance on when discovery-origin information should be requested or withheld in future evaluation studies.

8 Budget magnitude

The total request is US$320,000, within the proposed US$200,000–400,000 range:

• Personnel: $115,000.
• Reviewer honoraria and recruitment: $80,000.
• Materials development and pilot testing: $25,000.
• Independent validation coordination: $30,000.
• Statistical support and open-data preparation: $20,000.
• Ethics administration and data security: $10,000.
• Dissemination: $5,000.
• Indirect costs and contingency: $35,000.

Validation-reviewer honoraria are included in the recruitment allocation, avoiding double counting. Coordination funds cover the additional organizational work required for independent samples and materials.

9 PI profile and candidate teams or institutions

The PI should combine expertise in metascience, judgment and decision-making, crossed randomized experiments, and open science. An independent statistical lead and an experienced grant-review adviser should complement that profile.

Candidates for exploratory discussions include Balázs Aczél and the ELTE team for judgment research, Zoltán Kekecs for transparent research procedures, Giacomo Livan for interpretation of funding patterns, and the Santa Fe Institute as a potential complexity-methods partner. These are proposed roles, not confirmed appointments or institutional commitments. ELTE’s public information documents Aczél’s metascience group and Kekecs’s affiliation; Livan’s supplied funding study and SFI’s interdisciplinary work motivate the other suggestions. ELTE Metascience Lab; Kekecs profile; Sun et al., 2021; SFI interdisciplinary work.

10 Risks and abandonment criteria

A label may change reasonable beliefs about competence or prior evidence, evoke demand characteristics, or sound implausible. Holding information constant reduces these ambiguities but cannot guarantee their removal. Simulated ratings also differ from consequential funding decisions.

If both contrasts fall within the prespecified equivalence bounds in all three samples, the project will abandon the claim of a practically important penalty in this task. Failure to reproduce the effect in validation will defeat a claim of disciplinary generality. A stable positive effect contradicts the predicted penalty direction. Failure to distinguish high- from low-quality methods, or failure of the label manipulation, indicates an invalid task rather than a substantive null result. Wide intervals indicate uncertainty.

A score reduction alone is not sufficient to call the effect unfair bias. No outcome can validate anomalous claims, overturn scientific methodology, or establish that all real funding disparities arise from prejudice.

11 Ethics and compliance

An IRB or equivalent ethics determination precedes recruitment. Fictional proposals ensure that participation does not affect real awards or professional evaluations. Temporary withholding of the directional hypothesis requires ethics approval and subsequent debriefing.

Identity and rating records are separated, access is restricted, and consent explains cross-border handling and withdrawal deadlines. Public-data licensing follows the data type and participant consent: the source dataset’s CC BY license does not automatically apply to reviewer data. Funding interests are disclosed. Investigators retain publication freedom, including for null findings and results unfavorable to the originating hypothesis.

12 Fit with existing funders and models

Future Mind Institute would implement this project as a neutral consciousness-and-life-sciences funder. Its Seth Material Research Program is one public program, not a basis for presuming conclusions.

Bial Foundation has thematic adjacency through its psychophysiology and parapsychology remit, but eligibility of a purely metascientific proposal must be confirmed. Fetzer Institute is a potential dialogue partner; it currently has no public funding application process and should not be conflated with the Fetzer Franklin Fund. Bial research grants; Fetzer Institute’s work.

The closest methodological model is Templeton World Charity Foundation’s COGITATE approach, combining preregistration, open science, and adversarial collaboration. This resemblance does not establish eligibility for an active call. TWCF on COGITATE.

BICS, UVA DOPS, and IONS remain possible exchange partners whose roles and interest require verification. The fundable product is a reusable review-audit protocol, with any subsequent experiment in consequential funding procedures requiring separate approval.