Projected-Entailment, Non-Mapping, and Causal-Overreach Audit¶
Analytic audit — instantiates Metaphor Mapping and Interpretive Control
Derives every inference a mapping invites, grades each as supported, uncertain, generative, misleading, or prohibited against target evidence, and fences the source features and causal claims that must not transfer.
The Projected-Entailment, Non-Mapping, and Causal-Overreach Audit takes a correspondence map and asks, of each row, "what does this invite a reader to conclude?" — then judges every such inference against target-domain evidence rather than against how natural it feels in the source. Each projected entailment is labelled supported, uncertain, generative-hypothesis, misleading, harmful, or prohibited; the source features, values, and remedies that must not travel are written down as explicit non-mappings; and one rule governs the whole exercise: source causality never establishes target causality. Its defining move is that it operates on inferences, not structure — it neither builds nor improves the map, it grades what the map licenses and fences what it must not.
Example¶
A science writer drafting an explainer works from the "the brain is a computer" correspondence map — memory ↔ storage, perception ↔ input, thinking ↔ processing — and runs the audit over it. "The brain stores memories like files" → uncertain, since human memory is reconstructive rather than file-retrieval, so it is relabelled a loose analogy. "Thinking is computation" → generative-hypothesis: useful for building models, not an established fact, so it is flagged as speculation, not banned. "A mind could be uploaded to new hardware" → prohibited: there is no target evidence for substrate independence, so the inference is fenced. "More memory means smarter" → misleading. "A damaged region is a broken chip you swap out" → harmful, because it erases plasticity.
The causal firewall bites hardest on the move "computers work by X, therefore brains must work by X" — the resemblance is not evidence about the brain. The output is a graded entailment ledger plus a travelling non-mapping list ("mind uploading," "swap the part," "RAM equals intelligence") that must accompany the metaphor wherever it is reused. Crucially, "prohibited" here means unsupported, not disproven: the audit fences mind-uploading for lack of evidence, not because it has been shown false.
How it works¶
- Enumerate entailments per correspondence. For each row of the map, state the inference a reader is invited to draw and the action it implies — not only the intended reading.
- Grade against the target, not the source. Assign supported / uncertain / generative / misleading / harmful / prohibited by the target model's evidence, never by how plausible the claim sounds in the source domain.
- Apply the causal firewall. Flag every place the mapping is being used as causal proof about the target, and demote it to hypothesis or prohibition.
- Write the non-mappings. List the source features, values, roles, and remedies that must not transfer, each paired with a counterexample and the point where literal language must take over.
Tuning parameters¶
- Entailment net width — whether to derive only the intended inferences or every plausible over-reading. A wider net catches more harmful transfers but costs effort and can invent strawmen nobody would infer.
- Grading resolution — a coarse safe/unsafe split versus the full six labels. Fine resolution preserves the generative-hypothesis category instead of collapsing useful speculation into "false."
- Causal strictness — how aggressively source-causal reasoning is demoted to hypothesis. Stricter prevents overreach but can suppress a genuinely productive leap.
- Non-mapping prominence — whether fenced features are logged privately or published beside the claim. Publishing protects downstream reuse; logging is lighter but travels less.
- Evidence bar for "supported" — how much target evidence promotes an entailment to supported rather than uncertain.
When it helps, and when it misleads¶
Its strength is that it is the firewall between vivid and true: it catches the inference a clean-looking mapping still smuggles in, and it preserves generativity by labelling speculation as speculation rather than either banning or believing it. Its failure mode is that it can only grade the entailments it manages to enumerate, and the dangerous one is often the inference nobody wrote down; worse, a confident grade plus repetition can quietly harden a "generative hypothesis" into a "supported" fact — the fallacy of misplaced concreteness, mistaking a useful abstraction for the concrete thing it models.[n1] The classic misuse is running the audit to bless a metaphor's causal story ("the ledger grades it supported, so the brain really is a computer"). The guarding discipline is to judge only against target evidence, keep the causal firewall absolute, and re-run the audit whenever the mapping is stretched to a new claim.
How it implements the components¶
correspondence_projection_entailment_and_evidence_map— it supplies the projection-and-entailment half of this component: deriving and evidence-grading the inferences each correspondence invites (it reasons over a map, and does not itself build the correspondence structure).non_mapping_literalization_overextension_and_failure_boundary— the fenced source features, prohibited remedies, and the "source causality is not target causality" rule are the non-mapping and failure-boundary list.
It grades inferences but does not build the two independent domain models or align their structure into the map it reasons over (independent_source_and_target_domain_model) — that is the Source–Target Structure and Correspondence Matrix — and it reasons on paper, so it does not measure what real audiences infer, feel, or do (audience_comprehension_inference_emotion_memory_and_action_test), which is the Audience Inference, Emotion, Memory, and Action Experiment.
Related¶
- Instantiates: Metaphor Mapping and Interpretive Control — the inference-grading firewall of the pattern.
- Consumes: Source–Target Structure and Correspondence Matrix supplies the correspondence rows whose entailments this audit derives and grades.
- Sibling mechanisms: Source–Target Structure and Correspondence Matrix · Literalization, Boundary-Case, and Overextension Test · Alternative-Metaphor and Countermetaphor Comparison · Audience Inference, Emotion, Memory, and Action Experiment · Metaphor Drift, Harm, and Downstream-Decision Audit
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Projected-Entailment, Non-Mapping, and Causal-Overreach Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it derives every inference a mapping invites, grades each as supported, uncertain, generative, misleading, or prohibited against target evidence, and fences the source features and causal claims that must not transfer.
Independent corroboration: The frozen evidence defines Projected-Entailment, Non-Mapping, and Causal-Overreach Audit as 'Derives every inference a mapping invites, grades each as supported, uncertain, generative, misleading, or prohibited against target evidence, and fences the source features and causal claims that must not transfer', so its operative form is Assessment, Review & Assurance.
Nearest alternative: Analysis, Modeling & Optimization — Projected-Entailment, Non-Mapping, and Causal-Overreach Audit includes features of an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution, but its defining operation is a bounded evaluation of existing evidence or work that produces a finding or disposition.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Linguistics & Semiotics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Projected-Entailment, Non-Mapping, and Causal-Overreach Audit is most plausibly rooted in the linguistics_semiotics tradition because its characteristic form depends on language, signs, presupposition, and the structure of meaning. The assignment tracks that formative lineage, not the many settings in which the mechanism can now be applied.
Related originating lineages:
- Literature & Literary Theory — The literature_literary_theory tradition materially shaped Projected-Entailment, Non-Mapping, and Causal-Overreach Audit through its own practice of narrative form, textual interpretation, and the disciplined control of figurative meaning.
- Philosophy — The philosophy tradition materially shaped Projected-Entailment, Non-Mapping, and Causal-Overreach Audit through its own practice of logic, argument, normative analysis, and conceptual distinction.
- Rhetoric — The rhetoric tradition materially shaped Projected-Entailment, Non-Mapping, and Causal-Overreach Audit through its own practice of argument quality, persuasive form, audience handling, and ordered presentation.
Review resolution: Both blind reviewers agree that linguistics semiotics is the primary origin. Explicit reconciliation resolves reported ambiguity. Formative alternate lineages are retained as literature_literary_theory, philosophy, rhetoric; later breadth of use is recorded separately as domain_reach=multi_domain, while origin_mode=cross_disciplinary_synthesis describes the relationship among origin lineages.
Attribution caveat: The exact encyclopedia label appears to synthesize established practices; the primary domain identifies the strongest formative lineage, while the alternates record material ingredients rather than downstream uses. The comprehensive audit is encyclopedia-synthesized from several discourse and reasoning traditions.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
The audit's most abused label is "prohibited," which is easy to hear as "false." It is not: a prohibited entailment is one the target evidence does not license, whether because it is wrong or simply unknown. Fencing on absence of evidence, not on disproof, is what keeps the audit honest and keeps a genuinely open question from being smuggled in as settled science under cover of a familiar image.
[n1] Alfred North Whitehead's fallacy of misplaced concreteness — treating an abstract model or category as if it were the concrete reality it stands for. A computational model of cognition is a productive abstraction; sliding from "the brain can be modeled as a computer" to "the brain is a computer" commits exactly this fallacy, which is why the audit keeps a standing causal firewall between a mapping's usefulness and its literal truth. ↩