n.Newfeed

Philosophy · 6 min read

When Another Person's Word Counts as Evidence

A map of philosophical accounts of testimony, trust, competence, defeaters, and socially distributed knowledge.

When Another Person's Word Counts as Evidence

Most knowledge arrives from other people

Few people have personally measured Earth's age, sequenced a virus, inspected the archives behind a historical date, or verified the electrical grid before switching on a lamp. Much ordinary and scientific knowledge depends on testimony: one person presents something as true, and another person accepts it partly because it was said. The category includes conversation, textbooks, news reports, expert opinions, maps, databases, and instrument readings reported by researchers.

Calling testimony evidence does not settle when it justifies belief. A reliable friend says the train leaves at six. An anonymous account posts the same claim. A railway timetable lists it but may be outdated. The proposition is identical; the epistemic position is not. Philosophers of testimony ask what the hearer needs, what the speaker must possess, and how social practices distribute responsibility between them.

The literature offers no single test that turns another person's word into knowledge. It does identify recurring elements: the speaker's access and competence, sincerity, the hearer's background reasons, the presence of warning signs, and the chain through which the claim traveled. Those elements support calibrated reliance rather than a choice between trusting everyone and verifying everything oneself.

Reductionism asks for independent support

Reductionist accounts hold, roughly, that testimony is justified through more basic sources such as perception, memory, and inference. David Hume's discussion of reports of miracles is an influential ancestor of this approach. A hearer considers whether people in the relevant circumstances are generally truthful and accurate, then weighs the report against other evidence.

The attraction is clear. If a stranger makes an extraordinary medical claim, the absence of an immediate contradiction is not enough. Positive reasons might include the speaker's track record, institutional credentials, access to records, agreement with independent sources, and exposure to correction. Local reductionists focus on reasons concerning this speaker or occasion; global versions appeal to the general reliability of testimony as a practice.

Reductionism faces a regress and a scale problem. Much of the evidence used to certify a source—who holds a degree, which journal is reputable, whether a measurement was replicated—also comes from testimony. Children learn language and basic facts before they can assemble independent reliability records. If justification required each hearer to reconstruct the evidence behind each report, large portions of ordinary knowledge would become inaccessible.

That objection does not show that checking is useless. It shows that "independent" support is often partial and socially supplied. A reader may not rerun a clinical trial but can check whether it was registered, peer reviewed, replicated, and incorporated into a systematic review. The resulting reasons remain testimonial in part, yet they are not identical to accepting the original claim at face value.

Non-reductionism begins with a qualified entitlement

Non-reductionist or anti-reductionist views treat testimony as a basic source of justification. Thomas Reid argued in the eighteenth century that humans have natural dispositions toward truth-telling and credulity. Contemporary versions are more qualified: a hearer may be entitled to accept an intelligible assertion when there is no relevant reason for doubt, rather than needing positive evidence of reliability on every occasion.

This approach fits the speed and dependence of social learning. Conversation would stall if each mundane statement required an audit. It also explains how children can learn from caregivers before possessing a theory of speaker reliability. But permissiveness creates the central objection. A fabricated website and a careful reference work can both present fluent assertions. If the hearer's lack of known defeaters were sufficient, ignorance of warning signs could look like justification.

Hybrid theories try to preserve default trust while requiring monitoring appropriate to the stakes and context. The Stanford Encyclopedia of Philosophy's survey of testimony describes this as a live family of positions rather than a settled compromise. Some accounts emphasize assurance: in telling a hearer something, a speaker undertakes responsibility for its truth. Others emphasize reliable social processes rather than reasons consciously available to the hearer.

The disagreement is partly about what justification must be. Internalist theories ask what reasons the believer can access; externalist theories may count a reliably formed belief as knowledge even if the believer cannot articulate the process. Testimony makes that fault line visible because the evidence can reside partly in another person's competence.

Competence and sincerity answer different questions

A sincere speaker can be mistaken, and a knowledgeable speaker can deceive. Responsible reliance therefore separates at least two assessments. Competence concerns whether the source is positioned to know: relevant training, direct access, sound methods, and a domain-matched record. Sincerity concerns whether the source aims to communicate what they take to be true. Neither can be inferred safely from confidence or fluency.

Elizabeth Fricker's work stresses the hearer's duty to monitor speakers for trustworthiness rather than accept assertions indiscriminately. Jennifer Lackey's cases complicate the idea that testimony merely transmits knowledge already possessed by the speaker. A creationist teacher may accurately present the scientific account of evolution from a reliable curriculum while privately rejecting it. Students can appear to gain knowledge even though the speaker does not believe the proposition in the ordinary way. Such cases motivate the view that testimonial exchanges can generate knowledge through a reliable statement and social practice, not simply pass a speaker's belief intact.

Other cases point the opposite way. A speaker may know a fact but offer it through an unreliable process—guessing which true item from memory applies, for example. The hearer should not automatically inherit the speaker's knowledge. The epistemic quality of the act depends on the connection between the source, the assertion, and the truth.

Defeaters change what a hearer may accept

Philosophers use "defeater" for information that undermines justification. A rebutting defeater supports the claim's falsity: the railway issues a cancellation notice. An undercutting defeater attacks the route to belief without establishing the opposite: the timetable is dated last year. The distinction helps explain why uncertainty should sometimes rise even when no alternative answer is available.

Defeaters are sensitive to stakes. A friend's recollection may justify choosing a café but not administering a drug. Higher stakes do not necessarily change truth, but they can raise the amount of checking required for responsible action. Expertise is also granular. A physicist's standing does not supply authority in epidemiology, and an epidemiologist may accurately describe population evidence without knowing what treatment is right for one patient.

Disagreement adds another layer. When comparably informed people reach different conclusions, the disagreement is evidence about one's own fallibility, but philosophers dispute how much each side should revise. Real disputes rarely involve perfectly equal "epistemic peers." Differences in data access, models, priors, and incentives matter, and identifying them may itself rely on testimony.

A practical conclusion with philosophical limits

The literature supports several habits without producing an algorithm. Ask how the speaker could know, whether the claim lies within that access, what incentives and correction mechanisms operate, and whether independent lines converge. Look for defeaters proportionate to the stakes. Preserve the chain of attribution so later readers can inspect more than the final assertion.

These habits do not reduce all testimony to private verification. They improve the social routes through which knowledge travels. Peer review, audit trails, professional norms, and adversarial checks are ways of making testimony more reliable; none makes each report infallible. Trust can be rational because it is selective and revisable, not because error has been removed.

What is established is our dependence on testimony and the inadequacy of both blanket credulity and total self-verification. What remains contested is whether testimonial justification is basic, inherited from a speaker, generated by a reliable practice, or reducible to other evidence. Philosophy maps those conditions. It cannot certify a particular witness without the factual work of examining that witness and claim.

Sources and review note

4 sources, none of them with a link, so treat them as reading, not proof.

  1. Stanford Encyclopedia of Philosophy, "Epistemological Problems of Testimony" (2021, substantive revisions tracked by the encyclopedia).
  2. Elizabeth Fricker, "Telling and Trusting: Reductionism and Anti-Reductionism in the Epistemology of Testimony," Mind 104 (1995).
  3. Jennifer Lackey, Learning from Words: Testimony as a Source of Knowledge (Oxford University Press, 2008).
  4. C. A. J. Coady, Testimony: A Philosophical Study (Oxford University Press, 1992).

The schools described contain internal variants. The draft uses them as an evidence briefing on a debate, not as a claim that philosophers have agreed on a decision rule.

Want the reader to remember this spot? Sign in — it is free.

Next Moral Luck and the Limits of Responsibility

The app

Keep going, or take it with you.

Free accounts read twenty pieces a day. Pro is US$1.99 a month or US$11.99 a year for unlimited reading.