{"schema_version":1,"assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","source_experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"computability_boundary_mapping__neuroscience","archetype_slug":"computability_boundary_mapping","domain_slug":"neuroscience","title":"Formal reachability boundaries for executable neural-circuit models","opportunity_summary":"Audit whether a neural-modeling project actually makes an unrestricted exact-and-terminating reachability claim, then formalize a toy model language and test whether unrestricted impossibility, useful decidable fragments, and explicit UNKNOWN routing can replace timeout-based false certainty. The potential value is scientifically meaningful, but the problem's occurrence, the relevant proofs, usefulness of restricted fragments, and distinctiveness remain unverified.","adopter_authorizer":"The computational-neuroscience project's model-governance owner could adopt claim and interface changes, subject to approval of impossibility or totality claims by an independent methods reviewer.","scores":{"meaningful_impact":{"score":3,"rationale":"Preventing false non-reachability conclusions and wasted pursuit of an impossible universal analyzer could materially improve scientific claims and resource allocation, but the sealed candidate supplies no evidence about how often the alleged behavior occurs or how consequential it has been."},"stakeholder_pull":{"score":2,"rationale":"The candidate identifies model authors, reviewers, collaborators, and downstream interpreters as affected parties, but provides no demonstrated requests, commitments, observed demand, or evidence that an institution recognizes the problem."},"incremental_advantage":{"score":4,"rationale":"Relative to timeout-based simulation and to adding compute while retaining a universal claim, formally scoped guarantees plus distinct NO, UNKNOWN, timeout, and out-of-scope states directly address the stated failure mechanism. The size of the practical improvement has not yet been measured."},"distinctiveness_plausibility":{"score":2,"rationale":"The combination of computability classification, restricted-fragment totality, and operational routing is proposal-specific and coherent, but prior art is explicitly unsearched across neural-model reachability, verification, and governance, so distinctiveness cannot presently be credited."},"technical_implementability":{"score":3,"rationale":"The toy-language formalization, bounded executions, interface labels, and independent checking appear bounded and implementable, while a valid halting reduction, a total procedure for a scientifically useful fragment, and scalable routing all remain unresolved research obligations."},"adoption_authority_feasibility":{"score":4,"rationale":"The candidate assigns interface and claim authority to a model-governance owner and reserves theorem approval for an independent methods reviewer, with rollback rules and continued availability of the prior simulator. Feasibility is reduced because no actual institution, owner, or review process is confirmed."},"evidence_readiness":{"score":3,"rationale":"The candidate supplies a bounded first step, separate problem and intervention falsifiers, negative tests, prohibited extrapolations, and observable output states. It supplies no audited project artifacts, proof, benchmark corpus, pilot result, or adoption evidence."},"safety_net_benefit":{"score":4,"rationale":"Explicit UNKNOWN and out-of-scope states, a prohibition on treating timeout as NO, limits against biological or clinical extrapolation, independent review, and rollback to UNRESOLVED provide a strong safeguard against overclaiming. Downstream coercion of UNKNOWN into NO remains a stated risk."},"scalability":{"score":3,"rationale":"The contract-and-routing pattern could be reused across analyzers and model classes, but each model language, input regime, quantifier choice, and assumption change may require new proofs and review; useful restrictions may also exclude important neural dynamics."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Audit one project's documented reachability claims; formalize one toy executable neural-model language; attempt one checked unrestricted reduction and one finite-fragment procedure; obtain independent methods review; and compare routed outputs with the existing timeout-based interpretation on a fixed bounded corpus.","confidence":"LOW","assumptions":["Requires several weeks to a few months of combined formal-methods, computational-neuroscience, software, and independent-review labor.","Uses existing compute and a toy corpus rather than new experiments, clinical data, or specialized equipment.","Access to relevant analyzer documentation, code, and model definitions is available without major contracting or data-access expense."]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"Integrate scoped guarantee declarations and YES, NO, UNKNOWN, timeout, and out-of-scope states into one existing analyzer workflow, with validation tests, documentation, review controls, and rollback support.","confidence":"LOW","assumptions":["Deployment is limited to one research workflow and does not authorize experimental or clinical decisions.","The existing simulator can be wrapped rather than substantially rewritten.","No new theorem beyond the pilot's validated model class is required for startup."]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Launch the governed routing approach across a research program, including production integration, model-class registries, proof and test artifacts, user training, methods review, monitoring for label misuse, and evaluation against the prior workflow.","confidence":"LOW","assumptions":["Scope covers multiple model families within one institution or consortium, not universal coverage of neuroscience models.","Scientifically useful restricted fragments are established during prior evidence work.","Compliance needs are those of research governance rather than regulated clinical software."]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Maintain analyzer integrations and formal artifacts, review new or changed model assumptions, monitor downstream interpretation, rerun regression and bounded benchmark tests, and retain independent methods oversight.","confidence":"LOW","assumptions":["A modest number of model-language or assumption changes require review each year.","Existing institutional compute and software infrastructure remain available.","No clinical validation, continuous high-performance compute campaign, or large external certification program is included."]}},"research_burden":"HIGH","earliest_credible_horizon":"3_TO_12_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The candidate describes a recognizable failure—timeouts interpreted as non-reachability under an implicit universal claim—but supplies no audited project artifact showing that documented claims are actually unrestricted, reusable, exact, and terminating."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"The sealed candidate identifies the model-governance owner as the change authorizer and an independent methods reviewer as the required approver for impossibility or totality claims."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The proposal can test whether formal scope classification and explicit UNKNOWN routing reduce false definitive verdicts and wasted universal-search effort relative to the existing timeout-plus-informal-yes/no workflow."},"bounded_next_evidence_step":{"status":"YES","reason":"The authorized toy-language study bounds the model language and cases, requires independent checking, excludes live experimental or clinical decisions, and has explicit problem and intervention falsifiers."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The pilot has named authority roles, prohibits biological and clinical extrapolation, preserves the prior simulator, and restores UNRESOLVED if formalization or review fails; remaining risks are addressed through stop and rollback rules."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The candidate identifies relevant technical and governance activities but provides no staffing, integration inventory, data-access terms, proof complexity, or institutional scope from which an implementation range could be supported beyond a low-confidence broad estimate."}},"blocking_evidence":["An artifact audit establishing that at least one relevant project makes an exact, always-terminating reachability claim outside an enforceable finite or bounded class and fails to distinguish UNKNOWN, timeout, and NO.","A stable formal definition of the model language, admissible inputs, target regime, horizon, and quantifiers that matches the neuroscientific claim rather than merely a convenient surrogate.","Independent verification of either a valid impossibility argument for the unrestricted class or a total reachability procedure for at least one scientifically useful restricted class.","A fixed-corpus comparison showing that guarantee-based routing reduces unsupported definitive verdicts or wasted search effort without causing unacceptable label confusion or scope loss.","A bounded prior-art review establishing whether the contribution is a new theorem, a new neuroscience formulation, or an operational synthesis."],"next_evidence_step":"Select one documented neural-model analyzer and first audit its claims against the problem falsifier. If an unrestricted promise remains, formalize one toy language and fixed reachability predicate, independently check a halting-to-reachability construction and a finite-fragment total procedure, and replay a fixed bounded case corpus through both the baseline timeout-based interpretation and a prototype with YES, NO, UNKNOWN, timeout, and out-of-scope outputs. Stop if the audit finds only enforceably bounded claims, either formal argument fails review, or the routed prototype does not reduce unsupported definitive verdicts without material label confusion.","research_questions":["Do actual project documents promise exact, terminating reachability beyond an enforceable finite model class or bounded horizon?","Which model-language features and admissible inputs are necessary for the proposed unrestricted reduction, and do they occur in the accepted project language?","Can a restricted fragment be both provably total and scientifically useful rather than trivially decidable but irrelevant?","How often does the baseline convert timeout or failed proof search into a definitive non-reachability interpretation on the fixed corpus?","Does explicit routing reduce unsupported certainty and wasted search effort, and what rate of UNKNOWN or false alarms remains usable?","Can interfaces and review controls prevent UNKNOWN, timeout, or out-of-scope outputs from being coerced into NO?","What existing reachability, hybrid-systems verification, neural-model analysis, or scientific-governance work already covers the theorem or operational pattern?"],"recommendation":"VALIDATE_PROBLEM_FIRST","uncertainty_constraints":["Closed-book assessment: no external evidence was used.","Problem prevalence, stakeholder demand, realized harm, and market size are unmeasured.","The unrestricted undecidability result and restricted-fragment totality procedure are hypotheses, not established results.","The accepted model language may instead be finite-state, bounded-horizon, effectively enumerable, or already covered by a total procedure.","Formal-model conclusions cannot be generalized to biological nervous systems or empirical validity.","Cost bands are low-confidence resource-equivalent ranges because staffing, proof complexity, code condition, data access, and deployment scope are unspecified.","Distinctiveness cannot be inferred because prior art is explicitly unsearched."],"closed_book_prior_art_boundary":"No claim is made that the proposed theorem, neural-model formulation, output taxonomy, or governance workflow is novel. Prior art in neural-model reachability, formal verification, hybrid systems, theorem proving, and computational-neuroscience governance was not searched and remains an external-research requirement."}