Sperber et al.'s 2010 'epistemic vigilance' splits vigilance toward the source from vigilance toward the content — a sixth field independently drawing the reliability/credibility line

epistemicssource-evaluationcross-domain-bridgecross-time-bridgecognitive-sciencepragmatics

"Epistemic Vigilance" (Sperber, Clément, Heintz, Mascaro, Mercier, Origgi & Wilson, Mind & Language 25(4):359–393, 2010) sets up Section 4, "Vigilance Towards the Source" (p. 369), by dividing the acceptance of testimony into two factors in the section's second paragraph: "Factors affecting the acceptance or rejection of a piece of communicated information may have to do either with the source of the information—who to believe; or with its content—what to believe." This is the same fork the vault has now recorded converging across five other communities — Cold-War intelligence tradecraft, evidence-based medicine's GRADE, library science's CRAAP test, reliability-aware RAG, and probabilistic forecasting — each drawing the line between trusting the source and trusting this particular message (observation-intelligence-doctrine-and-rag-independently-derived-a-two-axis-source-model, moc-two-axes-that-wont-stay-independent). Cognitive science and philosophy of language make a sixth, arriving from a lineage none of the other five cite.

That lineage is the note's second point. Sperber et al. build the two-factor account on Grice (1975) and situate the trust-in-testimony debate against the older anti-reductionism of Thomas Reid (1764) and, further back, John Locke's 1690 objection that borrowed belief is not knowledge — the split has philosophical roots three centuries deeper than NATO doctrine, RAG, or the undergraduate worksheet. Where the human evidence and the machine evidence show the two axes leaking into each other, and where Icard's 2024 dynamic logic abandons their independence by construction, Sperber et al. supply the pragmatics-of-communication statement of why the two vigilances are distinct in the first place.

Whether this is genuine convergence or undetected common descent is unsettled: no shared name and no findable common ancestor link the six fields, and a common-ancestor check for the source/content split itself has not been run (an open thread on moc-two-axes-that-wont-stay-independent). The independence is held as convergent, not proven — the same discipline the cluster applies to its other members, and an instance of the vault's broader independence-keeps-failing pattern.