Loading field map
Testimony and trust is the subfield of epistemology that examines how we acquire knowledge and justified belief from what other people tell us. It asks when, why, and to what extent we are entitled to believe what we are told, and what happens when that entitlement is absent, broken, or misplaced. The subfield treats testimony not as a marginal or defective source of belief, but as a fundamental one: most of what any person knows—about history, science, geography, other minds, even much of their own biography—was learned by being told, not by direct observation or reasoning.
The central questions are deceptively simple. When someone asserts that a proposition is true, under what conditions does the hearer acquire a reason to believe it? Is the hearer’s justification grounded in their own evidence about the speaker’s reliability, or does the speaker’s assertion itself carry an epistemic force that the hearer can simply receive? What distinguishes legitimate trust in testimony from gullibility, and what distinguishes reasonable distrust from pathological skepticism? The subfield also investigates the social and structural dimensions of these questions: how trust operates in scientific communities, how testimony functions in legal and historical inquiry, and how systematic distortions—such as prejudice, propaganda, or institutional failure—can undermine the epistemic value of what people say.
For much of the history of philosophy, testimony was treated as a derivative or inferior source of knowledge. The ancient Greek and Roman traditions, and the medieval and early modern philosophers who followed them, generally held that genuine knowledge came from reason, perception, or divine revelation. What other people said could be useful, but it was not thought to be epistemically basic. The Scottish philosopher David Hume gave the classic modern formulation of this view in the eighteenth century. He argued that our belief in testimony is justified by induction: we have observed, in our own experience, that people generally tell the truth and that their reports tend to match the facts. Each new piece of testimony is therefore a piece of evidence, to be weighed against the possibility of error, deception, or the miraculous. On this account, testimony is not a distinct source of knowledge at all; it is just another set of observations about human behavior, from which we infer the likely truth of what is said.
The other major early modern position, associated with the English philosopher Thomas Reid, rejected this reduction. Reid argued that we are naturally endowed with a disposition to believe what others tell us, just as we are naturally disposed to trust our senses. This disposition is not a rational inference but a basic principle of human nature. It works because God or nature has aligned our propensity to speak truthfully with our propensity to believe others. On Reid’s view, testimony is as fundamental as perception or memory; it does not need to be justified by prior evidence about the speaker’s reliability. These two positions—the reductionist view, which tries to reduce testimony to other sources of evidence, and the anti-reductionist view, which treats testimony as basic—set the terms for the modern debate.
The modern subfield of testimony and trust emerged in the late twentieth century, largely as a response to the work of the British philosopher C. A. J. Coady, whose 1992 book Testimony revived and sharpened the Hume–Reid debate. Since then, the field has been organized around a central disagreement that is still unresolved.
Reductionists, in the contemporary sense, hold that a hearer’s justification for accepting testimony must ultimately rest on the hearer’s own evidence. This evidence includes past observations of the speaker’s truthfulness, general knowledge about human psychology and social incentives, and background beliefs about the topic at hand. The reductionist does not deny that we learn from others; they deny that the speaker’s assertion itself gives the hearer a reason to believe. The reason comes from the hearer’s independent assessment. A strong reductionist might require the hearer to have positive evidence of the speaker’s reliability in each case. A weaker version allows that general evidence about human nature, or about the reliability of a class of speakers, can suffice. The main problem for reductionism is that it seems to demand an unrealistic amount of evidence. A child learning that the earth is round, or a scientist accepting a colleague’s experimental results, rarely has the kind of independent confirmation that reductionism seems to require. If reductionism is correct, much of what we ordinarily take to be knowledge may not qualify.
Anti-reductionists, by contrast, hold that testimony is a basic source of justification. When a speaker asserts that p, the hearer is entitled to believe p unless there are positive reasons to doubt. The speaker’s assertion is not merely evidence about the speaker’s state of mind; it is a reason to believe the content asserted. This view is often grounded in the idea that assertion is a normative practice: when someone asserts, they take on a responsibility to speak truthfully, and the hearer is entitled to rely on that responsibility. Anti-reductionism faces its own problems. It seems to make us too credulous, and it struggles to explain why testimony from a known liar, or from someone with a clear motive to deceive, should carry less weight. Anti-reductionists typically respond that the default entitlement is defeasible: it can be overridden by evidence of unreliability, but it does not need to be earned by positive evidence in each case.
The debate between reductionism and anti-reductionism is not merely academic. It bears directly on questions about the division of cognitive labor, the legitimacy of deferring to experts, and the epistemic status of beliefs formed in childhood or under conditions of social pressure. If anti-reductionism is right, then a great deal of our knowledge is more secure than reductionism would allow; if reductionism is right, then much of what we think we know is fragile.
A third major approach, often called the assurance view or the interpersonal view, tries to move beyond the reductionist–anti-reductionist dichotomy. Its most prominent proponents include the British philosopher Richard Moran and the American philosopher Edward Hinchman. The assurance view begins from the observation that testimony is not just a transfer of information; it is a social act. When someone tells you something, they are not merely emitting a sound that happens to correlate with a fact. They are addressing you, offering you their assurance that the proposition is true. The hearer’s justification, on this view, is not a matter of evidence about the speaker’s reliability, nor a default entitlement that could be modeled as a kind of perceptual input. It is a matter of the speaker’s commitment.
To trust someone’s testimony, on the assurance view, is to accept their word—to treat their assertion as a promise or a guarantee. The speaker’s assertion gives the hearer a reason to believe because the speaker has staked their credibility on the truth of what they say. This reason is not reducible to the hearer’s evidence about the speaker’s past behavior, because it depends on the speaker’s present act of addressing the hearer. The assurance view explains why testimony from a friend or colleague feels different from reading a random post online: in the former case, the speaker is personally vouching for the claim; in the latter, there is no such address. It also explains why betrayal of trust is a distinctively epistemic wrong, not just a practical disappointment.
The assurance view has been criticized for being too narrow. It seems to work well for face-to-face testimony between individuals who share a relationship, but it is less clear how it applies to anonymous testimony, to testimony from strangers, or to the vast amounts of information we receive from institutions, media, and databases. Some philosophers have tried to extend the view by arguing that institutions can also offer assurance, but this extension is controversial. The assurance view also faces the objection that it conflates epistemic justification with moral or social obligation: the fact that someone has promised to tell the truth does not, by itself, show that the proposition is true. The hearer may have a reason to trust the speaker, but whether that reason is a genuinely epistemic reason—one that tracks truth—is disputed.
Beyond the debate about how individual testimony justifies belief, the subfield has expanded to examine the broader role of trust in the production and transmission of knowledge. This expansion was driven in part by the recognition that much of what we know is not learned from direct testimony at all, but from complex systems of testimony: scientific communities, legal proceedings, historical records, news media, and educational institutions. These systems raise questions that the individualistic framework of the reductionism debate cannot easily answer.
One important strand of this work concerns the epistemology of expertise. When a layperson accepts the claims of a scientist, or when a scientist accepts the results of a colleague in a different specialty, they are relying on testimony, but not on the testimony of a single known individual. They are relying on the testimony of a community, mediated by journals, peer review, and institutional credentials. Philosophers such as Alvin Goldman have asked how a nonexpert can rationally identify trustworthy experts, and have argued that this is a genuinely difficult problem. The layperson often lacks the very background knowledge needed to evaluate the expert’s claims, and the markers of expertise—credentials, reputation, institutional affiliation—can be manipulated. Goldman’s work has led to a rich literature on the epistemology of expertise, which is now a major part of the subfield.
Another strand concerns the social and political conditions of testimony. Feminist epistemologists, including Miranda Fricker, have drawn attention to the ways in which prejudice can undermine the epistemic standing of speakers. Fricker’s concept of testimonial injustice names the wrong done when a hearer gives a speaker less credibility than they deserve because of prejudice about their identity—for example, assuming that a woman is less knowledgeable about mechanics, or that a person of color is less credible about their own experiences. Testimonial injustice is not just a moral failing; it is an epistemic one, because it distorts the hearer’s ability to acquire true beliefs and silences the speaker’s contribution to shared knowledge. This work has connected the subfield to broader questions about power, identity, and the distribution of epistemic authority.
A related line of inquiry examines the epistemology of disagreement and the ethics of belief in contexts of distrust. If we know that a source is systematically biased, or that an institution has a history of deception, what should we do with its testimony? Some philosophers argue that we should withhold belief entirely; others argue that we should adjust our confidence downward but still treat the testimony as some evidence. This question is particularly pressing in the context of misinformation and propaganda, where the problem is not that we lack testimony but that we have too much of it, from sources whose reliability is hard to assess.
The subfield today is characterized by a plurality of approaches that do not neatly reduce to the older reductionist–anti-reductionist debate. Many philosophers now work on the assumption that testimony is a multifaceted phenomenon, and that different models may be appropriate for different cases. The assurance view, the reductionist view, and the anti-reductionist view are often treated not as mutually exclusive theories but as accounts that capture different aspects of testimonial practice. A philosopher might hold that face-to-face assurance is best explained by the interpersonal view, that anonymous testimony is best explained by a reductionist account, and that the testimony of children or the severely impaired requires a different treatment altogether.
There is also growing interest in the empirical dimensions of testimony. Psychologists have studied the conditions under which children trust or distrust what they are told, and philosophers have drawn on this work to refine their theories. The developmental evidence suggests that children are not simply credulous or simply skeptical; they are selective, trusting familiar and benevolent speakers and distrusting those who have been wrong in the past. This finding complicates both reductionism and anti-reductionism, because it suggests that the default trust in testimony is neither a simple basic principle nor a fully reasoned inference, but a set of heuristics that develop over time.
The subfield has also become more global in scope. While the modern debate was initially shaped by Anglophone philosophy, scholars have drawn attention to the rich traditions of reflection on testimony in Islamic, Jewish, Confucian, and African philosophy. These traditions often treat testimony and trust in ways that do not fit the Western individualist framework. For example, Confucian thought emphasizes the role of trust in maintaining social harmony and the moral obligations of both speakers and hearers, while some African philosophical traditions stress the communal nature of knowledge and the idea that knowing is a shared achievement. These perspectives are not merely historical curiosities; they offer alternative frameworks that challenge the assumptions of the contemporary debate.
The most durable contribution of the subfield may be its insistence that testimony is not a second-class source of knowledge. The questions it raises—about the nature of trust, the authority of speakers, the rationality of deference, and the social conditions of belief—are not confined to a narrow technical debate. They are questions that every person faces in deciding whom to believe, what to accept, and how to live with the fact that most of what we know, we know because someone told us.