Epistemology of disagreement

The epistemology of disagreement is the subfield of epistemology concerned with the rational significance of discovering that others — particularly those with comparable competence and access to the same evidence — hold beliefs that conflict with one’s own.¹⁶ The question is deceptively simple: when you and someone you regard as equally reliable reach opposite conclusions about the same matter, what should you do? Should you revise your belief, hold firm, or suspend judgment? The debate, which has developed rapidly since the early 2000s, bears directly on the rationality of conviction in politics, science, philosophy, and religion. Its implications are particularly acute for religious epistemology, where the fact that intelligent and informed people hold incompatible theological commitments has been cited as a challenge to religious belief since at least the Enlightenment.^{1, 8}

The concept of epistemic peerhood

The debate presupposes a working notion of an epistemic peer: someone who is roughly equal to oneself in the relevant cognitive virtues — intelligence, thoughtfulness, familiarity with the evidence, argumentative skill, and freedom from obvious bias — with respect to the question at hand.⁷ The concept is an idealization; in practice, two individuals are never perfectly matched across all relevant dimensions. But the idealization serves to isolate the philosophically interesting question: if the only relevant difference between two agents is that they reached different conclusions, what follows?

Defining peerhood precisely has proven difficult. Jennifer Lackey has argued that shared evidence is a crucial component — two agents who have access to different bodies of evidence are not genuine peers even if they are equally competent, and the existence of unshared evidence can change the rational response to disagreement significantly.¹² Thomas Kelly has noted that peerhood assessments are themselves fallible: you might mistakenly regard someone as a peer who is in fact much less (or much more) reliable than you, and the consequences of this mistake differ depending on which direction the error runs.⁵ These complications notwithstanding, the core of the debate concerns cases where the symmetry between the two agents is as close to perfect as possible, so that the disagreement itself is the primary datum requiring a rational response.

Conciliationism and the equal weight view

Conciliationism holds that the discovery of peer disagreement provides a reason to revise one’s beliefs significantly in the direction of the peer’s position. The strongest form of conciliationism, the equal weight view, maintains that one should give a peer’s assessment the same weight as one’s own — that, upon discovering disagreement, one should update one’s credence to something like the average of the two agents’ initial credences.⁷ Adam Elga, who developed the equal weight view in its most precise form, argued that prior to discovering the disagreement, each agent should regard the other’s assessment as equally likely to be correct. Once the disagreement is revealed, treating one’s own prior assessment as privileged would be rationally unjustifiable — it would amount to giving oneself special epistemic status without any basis for doing so.⁷

Richard Feldman has offered a related but distinct conciliationist position. Feldman argues that in cases of recognized peer disagreement, the rational response is typically to suspend judgment rather than to split the difference.¹ If two equally competent agents have evaluated the same evidence and reached incompatible conclusions, neither conclusion is better supported than the other from a neutral standpoint, and the appropriate attitude is withholding belief. Feldman has applied this analysis explicitly to religious disagreement, arguing that the existence of equally competent non-believers who have considered the same evidence should lead religious believers, and vice versa, to reduce their confidence or suspend judgment.^{1, 2}

David Christensen has developed a version of conciliationism that emphasizes the role of higher-order evidence. Disagreement with a peer provides evidence not about the first-order question directly but about the reliability of one’s own reasoning: it is evidence that one may have made a cognitive error somewhere in the process of evaluating the first-order evidence. Christensen argues that this higher-order evidence should lead to a reduction in confidence even if one cannot identify any specific error in one’s own reasoning, because the peer’s disagreement is itself evidence that such an error exists.^{3, 4}

The steadfast view

The steadfast view holds that, at least in some cases, it is rationally permissible to maintain one’s original belief in the face of peer disagreement. Thomas Kelly has been the most prominent defender of this position. Kelly argues that the conciliationist’s demand to give equal weight to a peer’s assessment ignores the asymmetry in each agent’s epistemic position: each agent has first-person access to the reasoning that produced their conclusion, and this access provides a legitimate basis for maintaining confidence in one’s own view even after learning that a peer disagrees.⁵

Kelly draws a distinction between the “total evidence” available to an agent and the “extra-evidential” fact of disagreement. On his view, the first-order evidence is what ultimately justifies or fails to justify a belief, and the mere fact that someone else has evaluated the evidence differently does not alter the evidential relationship between the first-order evidence and the proposition in question. A person who has carefully reasoned to a conclusion may reasonably regard the peer’s disagreement as evidence that the peer has made an error, rather than as an undercutting defeater for one’s own position.⁶

Kelly also raises a bootstrapping concern about conciliationism: if every peer disagreement requires revising toward the average, then the process of revision can be manipulated. An agent could engineer disagreements with less reliable thinkers who happen to qualify as peers on superficial criteria, forcing revisions that move the agent further from the truth. Moreover, in philosophy itself, where virtually every substantive position is disputed by someone who qualifies as a peer, consistent application of the equal weight view would require near-total suspension of philosophical belief — a consequence that many philosophers find self-defeating.⁶

Religious disagreement

The epistemology of disagreement has particularly significant implications for religious belief, because the fact of deep and persistent religious disagreement is among the most conspicuous features of the human epistemic landscape. Billions of people hold mutually incompatible religious beliefs with great conviction: Christians, Muslims, Jews, Hindus, Buddhists, and adherents of other traditions each affirm doctrines that contradict the doctrines of the others, while secular naturalists reject the central claims of all religious traditions.^{8, 11}

Feldman has argued that this disagreement is epistemically significant precisely because many of the disagreeing parties are epistemic peers in any reasonable sense of the term. Devout Christians and devout Muslims include individuals who are highly intelligent, deeply thoughtful, well-read in theology and philosophy, and sincerely committed to following the evidence and arguments where they lead. If such individuals count as peers, then conciliationism implies that the discovery of their disagreement should lead to significant reductions in religious confidence, potentially to suspension of judgment on contested theological claims.¹ This conclusion connects directly to the argument from religious diversity, which holds that the existence of incompatible religious traditions, each supported by apparently sincere and competent adherents, is evidence against the truth of any particular tradition.⁹

Defenders of reformed epistemology have challenged this analysis on several fronts. Alvin Plantinga has argued that if Christian belief is produced by a properly functioning cognitive faculty (the sensus divinitatis), then the believer has a source of evidence — a quasi-perceptual awareness of God — that the non-believer lacks, and the two are therefore not genuine peers.¹⁴ On this view, what appears to be peer disagreement is actually a case of unshared evidence: the believer has access to a source of warrant (direct divine self-disclosure) that the non-believer, whose cognitive faculties are impaired by sin, does not possess. Critics respond that this move is available to every religious tradition — the Muslim can claim a corresponding faculty, as can the Hindu — and therefore cannot provide a basis for preferring one tradition over another without begging the question.^{9, 15}

David Basinger has explored a middle position, arguing that the appropriate response to religious disagreement depends on the specific epistemic situation. In some cases, a believer may have access to evidence (personal religious experience, the coherence of a particular theological framework with lived experience) that is not fully shareable, and in such cases some degree of steadfastness may be warranted. But Basinger also argues that intellectual honesty requires religious believers to engage seriously with the perspectives of those who disagree, to consider the possibility that their own convictions are mistaken, and to resist the temptation to dismiss disagreement by impugning the intelligence or sincerity of those who see things differently.¹¹

The self-defeat problem

One of the most significant objections to conciliationism is that it appears to be self-undermining. The epistemology of disagreement is itself a subject of peer disagreement: conciliationists and steadfasters disagree about what disagreement rationally requires, and both camps contain highly competent philosophers. If conciliationism is correct, then the conciliationist who discovers that Thomas Kelly disagrees with her should reduce her confidence in conciliationism — a revision that, pushed far enough, would undermine the very position that demands the revision.^{3, 16}

Elga has responded by restricting the scope of the equal weight view: it applies to first-order disputes but not to the second-order dispute about how to respond to disagreement itself.⁷ Christensen, by contrast, has bitten the bullet, accepting that conciliationism requires the conciliationist to be less than fully confident in conciliationism, and arguing that this is not a fatal defect but rather an honest acknowledgment of the difficulty of the metaepistemological question.³ The self-defeat problem remains one of the most actively debated aspects of the field.

Broader implications

The implications of the debate extend well beyond religious epistemology. In political philosophy, the existence of deep and persistent disagreement about justice, rights, and the good life raises the question of whether political conviction can be rationally maintained in the face of intelligent dissent. In religious pluralism, the epistemology of disagreement intersects with questions about whether multiple religious traditions can be simultaneously valid or whether the diversity of religious belief itself constitutes evidence against exclusivist claims.¹⁵

The debate also raises questions about the relationship between evidence and belief-formation. Conciliationism presupposes that beliefs should track the balance of evidence and that the beliefs of peers constitute a form of evidence about the reliability of one’s own cognitive processes. The steadfast view challenges this by insisting that first-person access to one’s own reasoning provides a legitimate asymmetry that external evidence (including the fact of disagreement) cannot override.^{5, 6} The tension between these positions reflects a deeper tension in epistemology between internalist and externalist accounts of justification — a tension that the epistemology of disagreement has brought into unusually sharp focus.¹³

What makes the epistemology of disagreement particularly challenging is the difficulty of applying its conclusions without either collapsing into a thoroughgoing skepticism (if conciliationism is applied consistently across every domain of peer disagreement) or retreating into a dogmatism that insulates one’s beliefs from relevant counter-evidence (if steadfastness is applied too liberally). Navigating between these extremes requires a nuanced account of when and how the beliefs of others should influence one’s own — an account that the field is still in the process of developing.¹⁶