Handbook
General Reasoning Intelligence
Transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where…
Updated
Definition
Transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate.
Coverage role
Provides the shared reasoning kernel used by domain intelligences and cross-intelligence plans. It coordinates decomposition, claim relations, evidence, constraints, causality, trade-offs, robustness, and decision status without pretending that every question is reducible to one scoring method.
Scope
Included
- problem framing and decision context
- decomposition typing and coverage audit
- claim normalization and logical relation classification
- evidence and argument graphs
- hard-constraint extraction and satisfiability reasoning
- option generation and feasibility filtering
- causal temporal and sequential dependency modeling
- trade-off and sensitivity analysis
- failure risk and robustness analysis
- reasoning trace audit
- decision status and reopening rules
Excluded and non-goals
- claiming certainty where evidence is conflicted or absent
- treating a preference score as a hard fact
- using a universal proposition as evidence that its subject exists
- calling every overlapping taxonomy MECE
- inferring causation from association without a declared causal model
- inventing stakeholder values or criterion weights
- authorizing irreversible or regulated actions without accountable approval
Boundary rules
- Route to general.reasoning when the requested outcome requires one or more declared capabilities and the output can be represented by its canonical artifacts.
- Route to an adjacent or specialist intelligence when domain-specific rules, tools, or professional authority dominate the problem.
- Use general.reasoning to coordinate multi-pack conflicts, hard constraints, uncertainty, and decision status.
- Return ask, defer, or escalate when required context, source authority, consent, or accountable ownership is missing.
Capability semantics
| Capability ID | Name | Semantic intent |
|---|---|---|
capability.general.reasoning.frame_problem |
Frame Problem | Define the decision, objective, affected parties, scope, time horizon, stakes, constraints, output, and stop conditions. |
capability.general.reasoning.classify_reasoning_structure |
Classify Reasoning Structure | Classify the problem as partition, cover, facets, sequence, causal system, taxonomy, constraint problem, decision table, game, or hybrid. |
capability.general.reasoning.decompose_and_audit_coverage |
Decompose And Audit Coverage | Build a decomposition appropriate to the declared structure and test formal or qualitative gaps, overlaps, mixed dimensions, and boundary ambiguity. |
capability.general.reasoning.normalize_claims_and_relations |
Normalize Claims And Relations | Normalize propositions with quantifier, subject, predicate, polarity, modality, scope, time, and assumptions; classify entailment, equivalence, contradiction, contrariety, compatibility, and unknown relations. |
capability.general.reasoning.build_evidence_and_argument_graph |
Build Evidence And Argument Graph | Represent support, refutation, provenance, uncertainty, attacks, counterarguments, and four-valued evidence states without collapsing conflicts. |
capability.general.reasoning.extract_and_solve_constraints |
Extract And Solve Constraints | Separate hard constraints from preferences, formalize compatible rules, and seek a satisfying witness, uncovered case, overlap witness, or unsatisfied core. |
capability.general.reasoning.generate_and_filter_options |
Generate And Filter Options | Generate a structurally diverse option portfolio, remove infeasible and dominated options, and record rejected alternatives with reasons. |
capability.general.reasoning.model_causal_and_temporal_dependencies |
Model Causal And Temporal Dependencies | Represent causal assumptions, confounders, interventions, timing, feedback, information dependencies, and sequential decisions. |
capability.general.reasoning.evaluate_tradeoffs |
Evaluate Tradeoffs | Compare feasible options through declared criteria, Pareto analysis, decision tables, utility or MCDA methods, and sensitivity analysis. |
capability.general.reasoning.analyze_failure_and_risk |
Analyze Failure And Risk | Identify failure modes, initiating events, consequences, controls, detectability, residual risk, and escalation conditions. |
capability.general.reasoning.stress_test_decision |
Stress Test Decision | Test a provisional decision across plausible scenarios, uncertain parameters, adversarial cases, model error, and minimax-regret or threshold criteria. |
capability.general.reasoning.audit_reasoning_trace |
Audit Reasoning Trace | Check that material outputs trace to evidence, assumptions, policy rules, deterministic calculations, or accountable human judgments. |
capability.general.reasoning.decide_next_step |
Decide Next Step | Return decide, ask, investigate, repair, defer, escalate, monitor, or reopen with explicit rationale and trigger conditions. |
Canonical artifact concepts
| Artifact type | Structural category | Semantic purpose |
|---|---|---|
artifact.general.reasoning.decision_frame |
decision | Decision Frame for General Reasoning Intelligence: A bounded decision or routing record that exposes alternatives, governing constraints, evidence, rationale, ownership, status, and reopening conditions. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.reasoning_structure_classification |
structured_output | Reasoning Structure Classification for General Reasoning Intelligence: A typed domain output that preserves purpose, context, evidence, assumptions, status, and next action. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.decomposition_and_coverage_audit |
analysis | Decomposition And Coverage Audit for General Reasoning Intelligence: An evidence-linked assessment that states criteria, findings, uncertainty, limitations, status, and action. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.claim_relation_graph |
model | Claim Relation Graph for General Reasoning Intelligence: A purpose-bounded representation of entities, relationships, assumptions, boundaries, and evidence of adequacy. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.evidence_and_argument_ledger |
ledger | Evidence And Argument Ledger for General Reasoning Intelligence: A controlled collection of uniquely identified entries with ownership, provenance, status, and review state. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.constraint_model_and_solver_result |
model | Constraint Model And Solver Result for General Reasoning Intelligence: A purpose-bounded representation of entities, relationships, assumptions, boundaries, and evidence of adequacy. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.option_portfolio_and_feasibility_record |
ledger | Option Portfolio And Feasibility Record for General Reasoning Intelligence: A controlled collection of uniquely identified entries with ownership, provenance, status, and review state. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.causal_and_temporal_dependency_model |
model | Causal And Temporal Dependency Model for General Reasoning Intelligence: A purpose-bounded representation of entities, relationships, assumptions, boundaries, and evidence of adequacy. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.trade_off_and_sensitivity_report |
comparison | Trade Off And Sensitivity Report for General Reasoning Intelligence: A normalized comparison that applies hard constraints before preferences and preserves evidence, uncertainty, dominance, and sensitivity. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.risk_and_control_model |
model | Risk And Control Model for General Reasoning Intelligence: A purpose-bounded representation of entities, relationships, assumptions, boundaries, and evidence of adequacy. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.robustness_report |
analysis | Robustness Report for General Reasoning Intelligence: An evidence-linked assessment that states criteria, findings, uncertainty, limitations, status, and action. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.reasoning_trace |
structured_output | Reasoning Trace for General Reasoning Intelligence: A typed domain output that preserves purpose, context, evidence, assumptions, status, and next action. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
artifact.general.reasoning.decision_status_and_reopening_rule |
decision | Decision Status And Reopening Rule for General Reasoning Intelligence: A bounded decision or routing record that exposes alternatives, governing constraints, evidence, rationale, ownership, status, and reopening conditions. It supports the pack purpose of transform an ambiguous problem into a traceable reasoning graph that separates facts, assumptions, values, constraints, options, uncertainty, risks, decisions, and reopening conditions; apply deterministic checks where the structure is formalizable and preserve an explicit route to ask, repair, defer, or escalate. |
Method families and limits
| Method ID | Method | Use when | Avoid when |
|---|---|---|---|
method.general.reasoning.decision_framing_and_scope_contract |
decision framing and scope contract | Use when an ambiguous request must be converted into a decision-relevant, answerable, bounded question or requirements brief. | Avoid endless clarification, leading questions, hidden presuppositions, unnecessary personal-data collection, and solutioning before the decision context is known. |
method.general.reasoning.provenance_graph_and_reasoning_trace_audit |
provenance graph and reasoning trace audit | Use when the reliability of a conclusion depends on locating original evidence, comparing independent sources, preserving scope, and identifying conflicts. | Avoid equating repetition with independence, citing summaries for claims available in primary sources, or merging conflicting evidence into a single unsupported score. |
method.general.reasoning.mece_partition_and_cover_audit |
MECE partition and cover audit | Use when propositions, partitions, evidence states, attacks, rules, or lifecycle transitions can be made explicit and checked for consistency or coverage. | Avoid importing classical implications without their assumptions, treating an overlapping cover as a partition, or allowing an ordering-dependent rule table to masquerade as unique. |
method.general.reasoning.generalized_square_and_graph_of_opposition |
generalized square and graph of opposition | Use when propositions, partitions, evidence states, attacks, rules, or lifecycle transitions can be made explicit and checked for consistency or coverage. | Avoid importing classical implications without their assumptions, treating an overlapping cover as a partition, or allowing an ordering-dependent rule table to masquerade as unique. |
method.general.reasoning.four_valued_evidence_lattice |
four-valued evidence lattice | Use when propositions, partitions, evidence states, attacks, rules, or lifecycle transitions can be made explicit and checked for consistency or coverage. | Avoid importing classical implications without their assumptions, treating an overlapping cover as a partition, or allowing an ordering-dependent rule table to masquerade as unique. |
method.general.reasoning.structured_argumentation_and_defeasible_reasoning |
structured argumentation and defeasible reasoning | Use when propositions, partitions, evidence states, attacks, rules, or lifecycle transitions can be made explicit and checked for consistency or coverage. | Avoid importing classical implications without their assumptions, treating an overlapping cover as a partition, or allowing an ordering-dependent rule table to masquerade as unique. |
method.general.reasoning.csp_sat_and_smt_constraint_checking |
CSP SAT and SMT constraint checking | Use when variables, domains, rules, invariants, or proof obligations can be represented formally and a witness, counterexample, or unsatisfied core is decision-relevant. | Avoid encoding ambiguous natural-language rules as hard constraints without review or treating solver satisfiability as evidence that the real-world model is complete. |
method.general.reasoning.decision_tables_with_declared_hit_policy |
decision tables with declared hit policy | Use when propositions, partitions, evidence states, attacks, rules, or lifecycle transitions can be made explicit and checked for consistency or coverage. | Avoid importing classical implications without their assumptions, treating an overlapping cover as a partition, or allowing an ordering-dependent rule table to masquerade as unique. |
method.general.reasoning.morphological_analysis_and_cross_consistency_assessment |
morphological analysis and cross-consistency assessment | Use when the task needs a deliberately diverse option space before convergence and novelty must remain relevant to a declared brief. | Avoid judging too early, generating cosmetic variants, copying an analogy without mapping its mismatch, or treating novelty as usefulness. |
method.general.reasoning.causal_dag_and_influence_diagram |
causal DAG and influence diagram | Use when a decision concerns intervention, mechanism, feedback, delay, confounding, or counterfactual change rather than description alone. | Do not treat association, temporal sequence, stakeholder belief, or repeated why-questions as proof of causation; use them to propose testable structures. |
method.general.reasoning.pareto_frontier_mcda_and_sensitivity_analysis |
Pareto frontier MCDA and sensitivity analysis | Use after infeasible alternatives are removed and several legitimate criteria or trade-offs remain. | Do not use before hard constraints are applied or without accountable ownership of criteria, weights, thresholds, and risk appetite. |
method.general.reasoning.fault_tree_event_tree_fmea_and_bow_tie_analysis |
fault tree event tree FMEA and bow-tie analysis | Use when potential harm must be detected, prioritized, contained, routed, and recorded without exceeding available authority. | Avoid delaying emergency or safeguarding routes, asking for unnecessary sensitive detail, or giving instructions that increase exposure, destroy evidence, or conflict with official guidance. |
method.general.reasoning.scenario_stress_testing_and_minimax_regret |
scenario stress testing and minimax regret | Use when material uncertainty cannot be represented credibly by one forecast and a decision must remain acceptable across plausible futures. | Avoid presenting scenarios as predictions, omitting adverse but plausible futures, or choosing a preferred scenario after seeing the result. |
method.general.reasoning.decision_lifecycle_state_machine |
decision lifecycle state machine | Use when propositions, partitions, evidence states, attacks, rules, or lifecycle transitions can be made explicit and checked for consistency or coverage. | Avoid importing classical implications without their assumptions, treating an overlapping cover as a partition, or allowing an ordering-dependent rule table to masquerade as unique. |
Pack-specific invariants
The pack also inherits the core invariants in core/core_invariants.yaml.
| Invariant ID | Statement | Failure action |
|---|---|---|
invariant.general.reasoning.01_scope_time_modality_quantifier_and_domain_assumptions_are |
Scope, time, modality, quantifier, and domain assumptions are normalized before two claims are classified as contradictory. | fail_or_repair |
invariant.general.reasoning.02_a_universal_proposition_does_not_imply_an_existential_prop |
A universal proposition does not imply an existential proposition unless non-emptiness or existential import is independently established. | fail_or_repair |
invariant.general.reasoning.03_every_decomposition_is_labelled_partition_cover_facets_seq |
Every decomposition is labelled partition, cover, facets, sequence, causal, taxonomy, or hybrid before a MECE claim is accepted. | fail_or_repair |
invariant.general.reasoning.04_hard_constraints_are_evaluated_before_preference_scoring_a |
Hard constraints are evaluated before preference scoring, and infeasible options cannot win by compensating scores. | fail_or_repair |
invariant.general.reasoning.05_criterion_weights_and_risk_appetite_come_from_an_accountab |
Criterion weights and risk appetite come from an accountable owner or declared policy, not silent model invention. | fail_or_repair |
invariant.general.reasoning.06_association_prediction_intervention_and_causation_are_repr |
Association, prediction, intervention, and causation are represented as different relations. | fail_or_repair |
invariant.general.reasoning.07_supporting_and_refuting_evidence_are_preserved_independent |
Supporting and refuting evidence are preserved independently so conflicted and undetermined states remain visible. | fail_or_repair |
invariant.general.reasoning.08_every_material_conclusion_traces_to_evidence_a_declared_as |
Every material conclusion traces to evidence, a declared assumption, policy rule, deterministic calculation, or accountable human judgment. | fail_or_repair |
invariant.general.reasoning.09_a_winner_that_changes_under_plausible_weights_or_assumptio |
A winner that changes under plausible weights or assumptions is labelled contingent and receives a reopening trigger. | fail_or_repair |
invariant.general.reasoning.10_the_reasoning_lifecycle_always_exposes_ask_repair_defer_es |
The reasoning lifecycle always exposes ask, repair, defer, escalate, monitor, and reopen paths where applicable. | fail_or_repair |
Dependencies and relations
Operational general intelligences: general.question-framing, general.validation-verification
Foundation concepts: general.semantic-representational, general.logical-constraint, general.epistemic-evidence, general.probabilistic-statistical, general.causal-counterfactual, general.decision, general.planning-execution, general.systems, general.temporal, general.metacognitive
Life intelligences: None
Source roles
| Source ID | Claim roles | Runtime resolution required |
|---|---|---|
nap.science_engineering_practices |
concept_definition, research_and_evidence_practice | True |
nist.engineering_statistics |
concept_definition, measurement_or_security_practice, risk_control | True |
nap.reproducibility_replicability |
concept_definition, research_and_evidence_practice | True |
nasa.systems_modeling |
concept_definition, engineering_method, verification_and_validation | True |
nasa.systems_engineering |
concept_definition, engineering_method, verification_and_validation | True |
nasa.models_simulations |
concept_definition, engineering_method, verification_and_validation | True |
w3c.prov |
concept_definition, interoperability_or_formal_contract | True |
jsonschema.2020-12 |
concept_definition, interoperability_or_formal_contract | True |
nist.ai_rmf |
concept_definition, measurement_or_security_practice, risk_control | True |
Risk and authority boundary
Risk class: high
- The decision may create immediate safety, medical, legal, financial, security, or rights consequences.
- Material evidence remains conflicted or absent and the action is difficult to reverse.
- Stakeholders disagree about values, risk appetite, or authority and no accountable owner is available.
- A formal solver reports inconsistency or no feasible option and relaxing constraints requires human authorization.
Accountable people retain consent, value, regulated, irreversible, and residual-risk decisions. This semantic page does not claim a deployed implementation.
Maturity
semantic: S3_reviewable_definition
contract: C3_testable_contract
implementation: I0_not_included
evidence: E2_source_roles_mapped
governance: G2_controls_and_review_defined
overlay: O2_paired_seed_complete
Paired overlay
Runtime maturity
- Capabilities: 13
- Technical floor→ceiling:
I1_runtime→I2_integrated - Autonomy floor→ceiling:
A1_supervised→A1_supervised - I1_runtime count: 11
- Evidence class:
contract: 11,semantic: 2 - Full hierarchy: MATURITY-HIERARCHY.md
Evaluation suite status
- CI status:
pass - Suite:
evaluation_suites/reasoning.yaml - Cases defined: 7
- Capabilities under test: 13