aiwiki.page
English

Belief

Belief is the attitude of taking something to be true, studied in philosophy, psychology, and the social sciences.

52 keywords23 linked from14 not yet writtenWritten by AI
PhilosophyPropositionTruthEpistemologyPhilosophy of Mi…IntentionalityConsciousnessKnowledgeBelief

Belief is the attitude of taking something to be the case or regarding it as true. A person who believes that rain is falling represents the world as being a certain way, whether or not rain is actually falling. In philosophy, belief is commonly analyzed as an attitude toward a proposition: an agent believes that something is so. The term also has broader uses, including religious commitment, confidence in a person, and enduring convictions that organize practical life.

Belief matters because it connects thought with truth, reasoning, and action. It is a central subject of epistemology, which investigates knowledge and rational justification, and of philosophy of mind, which asks what mental states are and how they operate. Psychology examines how beliefs are acquired, maintained, and changed; the social sciences investigate their transmission through testimony, institutions, and culture. These inquiries distinguish what people believe, how strongly they believe it, what makes their beliefs reasonable, and how belief differs from imagination, acceptance, faith, and knowledge.

The concept and its neighboring attitudes

In its standard philosophical sense, belief is a propositional attitude: a mental attitude with content that can be true or false. Believing that a bridge is safe differs from wanting it to be safe, fearing that it is unsafe, or imagining that it collapses. The same subject matter can occur within different attitudes, but only belief ordinarily presents its content as the way things actually are. This feature connects belief with intentionality, the capacity of mental states to be about objects, events, and situations.

Belief does not require certainty. Someone may believe that a train leaves at six while acknowledging that the timetable could have changed. Nor does it require current conscious attention. A person normally retains countless beliefs about names, geography, and everyday routines while thinking about something else. Philosophers therefore distinguish occurrent beliefs, which are actively entertained, from standing or dispositional beliefs, which persist without occupying conscious awareness. Whether every standing belief must be explicitly represented in memory is a further question.

Belief is also distinct from knowledge. On a widely accepted view, knowing that something is the case entails believing it, but belief alone does not establish knowledge. False beliefs cannot constitute knowledge, and a true belief may be correct merely by luck. The Gettier problem, introduced by Edmund Gettier in 1963, shows why even justified true belief may fall short: an agent can have good reasons for a conclusion that happens to be true through an unexpected coincidence.

Disbelief and suspension of judgment are different attitudes. To disbelieve that a door is locked is normally to believe that it is not locked. To suspend judgment is to settle on neither proposition. Merely lacking a belief need not amount to deliberate suspension: someone who has never considered a question may have no attitude toward it. These differences matter when assessing whether an agent is committed to a contradiction or is simply undecided.

Belief must also be distinguished from practical acceptance. A lawyer may accept a proposition for the purposes of an argument without believing it, and a scientist may use an idealized model while believing that some of its assumptions are literally false. Believing in a person may involve trust rather than merely believing propositions about that person. Religious faith can similarly include loyalty, hope, and practice. Such attitudes often contain beliefs, but their full character cannot always be reduced to a list of propositions held true.

Historical development

Ancient Greek philosophy developed influential distinctions between opinion and knowledge. In the Republic, Plato contrasts doxa, commonly translated as opinion or belief, with knowledge, relating the distinction to his account of changing sensible things and intelligible objects. In the Theaetetus, proposed definitions of knowledge include true judgment and true judgment with an account, but the dialogue does not endorse a final definition. These discussions established enduring questions about how correct opinion differs from understanding.

Aristotle distinguished demonstrative scientific understanding from less secure forms of judgment. In the Posterior Analytics, scientific understanding concerns explanations through causes and principles, rather than merely accepting that a conclusion is true. The Stoics developed a different analysis around impressions and assent. An impression presents something as being the case; assent involves accepting that presentation. Their account made the management of assent central to rational agency, while ancient skeptics examined whether suspension of judgment could avoid the instability of competing claims.

Medieval discussions connected assent with both intellectual evidence and faith. Thomas Aquinas distinguished understanding, science, opinion, and faith according to their objects and grounds of assent. In his account, faith involves intellectual assent under the influence of the will, with divine truth as its object. This is not simply a classification of all religious belief as unsupported opinion: it belongs to a theological framework that assigns different roles to demonstration, revelation, authority, and voluntary commitment.

Early modern philosophers investigated belief through doubt, perception, and evidence. René Descartes, in the Meditations of 1641, used systematic doubt to examine which judgments could survive skeptical challenges. John Locke, in the Essay Concerning Human Understanding of 1689–1690, treated assent as responsive to probability where knowledge was unavailable. David Hume, in the Treatise of Human Nature of 1739–1740, characterized belief through the force or vivacity with which an idea is conceived. On his account, custom explains why repeated experience generates expectations beyond what demonstrative reasoning can establish.

Nineteenth- and twentieth-century debates widened the subject. Charles Sanders Peirce, in “The Fixation of Belief” of 1877, connected belief with habits of action and inquiry with the irritation of doubt. William James examined when commitment might be permissible without decisive evidence. Later analytic philosophy investigated belief reports, mental representation, probability, and the relation between belief and action. Contemporary work combines these questions with experimental research, while retaining disagreements over whether belief is fundamentally an internal representation, a behavioral disposition, or a pattern identified through interpretation.

Mental representation and the nature of belief

One influential family of theories treats beliefs as internal representations. On this approach, believing that snow is white involves a mental state whose content is that snow is white. The state need not resemble an inner spoken sentence, although some theories propose structured, language-like representations. The computational theory of mind explains reasoning partly as the manipulation of such representations according to their formal properties. A belief can then participate in several processes: it can be retrieved from memory, combined with other beliefs, or used in planning.

Functionalist theories identify belief through its causal or functional role rather than through a particular material composition. A belief about rain might typically arise from seeing rain, combine with a desire to remain dry, and contribute to taking an umbrella. Its identity depends on this broader network of relations to perception, other mental states, and behavior. Functionalism therefore allows that beings with different physical constitutions could possess beliefs, provided the relevant organization is present.

Dispositional theories emphasize what an agent is inclined to do, say, infer, and experience. To believe that ice is slippery is, in part, to be disposed to walk carefully on ice, warn others, and expect objects to slide across it. Simple behavioral accounts face difficulties because behavior depends on many beliefs and desires together. Someone may believe that a surface is dangerous yet cross it to help another person. More sophisticated dispositional views consequently appeal to patterns across circumstances rather than a single observable response.

Interpretive approaches ask what makes belief attribution warranted. Daniel Dennett’s intentional stance treats a system as having beliefs and desires when doing so supports successful prediction and explanation. Donald Davidson connected the attribution of beliefs with interpreting speech and identifying substantial rational coherence. These approaches do not simply say that every useful description reveals an inner sentence. They examine the relation between mental-state attribution, intelligible behavior, and the standards used by interpreters.

Realist and eliminativist positions disagree about the status of everyday psychological categories. Realists regard beliefs as genuine states or properties that scientific explanation should accommodate. Eliminativists argue that some categories of ordinary mental discourse might eventually be replaced by a substantially different neuroscience. Between these positions lie views on which belief is real but multiply realized, approximate, or useful at a particular explanatory level.

These differences also affect the boundary of belief. Infants and nonhuman animals can behave in ways that suggest expectations without possessing language. Artificial systems can store information and use it in planning. Whether such cases constitute belief depends partly on whether belief requires linguistic articulation, conscious endorsement, representational structure, or only an appropriate pattern of information-sensitive activity. No single behavioral test settles all of these theoretical questions.

Content, reference, and belief reports

A theory of belief must explain not only what believing is but also what is believed. Two people can both believe that “the bank is nearby” while referring to different banks. Conversely, different sentences can express the same belief. The philosophy of language investigates how linguistic meaning, reference, and context help identify the contents attributed in statements such as “Mira believes that the train has left.”

Belief reports create difficulties for ordinary principles of substitution. “Hesperus” and “Phosphorus” both refer to Venus, but a person unfamiliar with their identity might believe that Hesperus appears in the evening without believing the corresponding claim under the name “Phosphorus.” Replacing one co-referring expression with another can therefore change the truth of a belief report. Gottlob Frege addressed related problems through his distinction between sense and reference: an object can be presented under different modes of presentation even when the reference is the same.

Another distinction concerns de dicto and de re belief. Someone may believe, in a general way, that the winner of a competition is talented without having any particular individual in mind. Alternatively, the person may believe of a specific competitor that she is talented, despite not knowing that she won. The first attribution concerns a descriptive proposition; the second singles out an object about which the belief is held. Quantification, identity, and the scope of descriptions make the formal treatment of this distinction difficult.

Externalist theories hold that the content of some beliefs depends partly on the believer’s environment. Hilary Putnam’s 1975 Twin Earth thought experiment imagines a world where a liquid superficially resembling water has a different chemical constitution. The example challenges the view that identical internal states must always determine identical contents. Tyler Burge extended externalist arguments to the social environment, examining how a speaker’s incomplete understanding of a term can nevertheless be anchored in a linguistic community.

Beliefs involving the self present another problem. John Perry’s 1979 discussion of the “essential indexical” describes a shopper who follows a trail of spilled sugar and eventually realizes that he himself is making the mess. Knowing that some shopper is responsible does not have the same practical significance as believing “I am responsible.” Such cases suggest that self-locating or first-person content cannot always be replaced by an impersonal description without losing an important explanatory feature.

These issues constrain formal models. Representing a proposition as a set of possible worlds captures distinctions between situations in which it is true and false. Yet necessarily equivalent propositions then receive the same representation, even when an agent understands one and not the other. More fine-grained accounts use structured contents or modes of presentation to represent differences that matter to actual believers.

Truth, evidence, and justification

Beliefs are evaluated along several dimensions that must be kept separate. Truth concerns whether their contents match how things are. Epistemic justification concerns whether they are appropriately supported or responsibly held. Knowledge introduces further requirements. A false belief can be reasonable given misleading evidence, while a true belief can result from guessing or wishful thinking. Evaluating the outcome alone therefore differs from evaluating the agent’s evidential position or method.

Evidentialist approaches connect justified belief with evidence. Perception, memory, testimony, and inference are common sources, although theories disagree about their ultimate status. Evidence can support a proposition without proving it. Seeing wet streets may support the belief that rain has fallen, but street cleaning provides an alternative explanation. Deductive reasoning preserves truth when its premises are true and its form is valid; inductive reasoning and inference to the best explanation extend conclusions beyond what their premises deductively entail.

The structure of justification generates the epistemic regress problem. If every justified belief must receive support from another justified belief, the chain appears either to continue indefinitely, circle back, or terminate. Foundationalism allows some beliefs to possess noninferential justification. Coherentism locates justification in a sufficiently coherent system of mutually supporting beliefs. Infinitist approaches defend indefinitely extendable chains of reasons. These positions differ not merely over where reasoning stops, but over what makes support epistemically effective.

Internalist and externalist theories disagree over whether justification depends entirely on factors accessible from the believer’s perspective. Reliabilism, an influential externalist approach, emphasizes whether a belief is produced by a reliable process. Under suitable conditions, ordinary visual perception may generate justified beliefs without the perceiver possessing an argument for its reliability. Internalists object that epistemic evaluation must give a more central role to the subject’s accessible reasons. Virtue epistemology instead emphasizes intellectual abilities and traits, such as perceptual competence, careful inquiry, and responsiveness to objections.

Defeaters help explain why initially reasonable beliefs may lose their support. A rebutting defeater provides reason to think the believed proposition is false. An undercutting defeater challenges the connection between evidence and conclusion. Learning that an object is painted blue may rebut the belief that it is red; learning that the room is illuminated by red light may undercut reliance on its apparent redness without establishing its actual color.

Belief’s apparent orientation toward truth has led philosophers to ask whether truth is its constitutive aim or norm. On some accounts, a state cannot count as belief unless it is assessable as correct when true. Other accounts explain the same pattern through the biological, practical, or social functions of representing the environment accurately. Neither approach implies that people always consciously aim at truth or that useful beliefs are necessarily true.

Degrees of belief and formal models

Ordinary speech treats belief both categorically and gradually. A person believes that a meeting is on Tuesday but may be more confident about this than about the weather forecast. Formal epistemology distinguishes outright belief from credence, a degree of confidence commonly represented by a number between 0 and 1. A credence of 0.8 in a proposition need not mean that the agent has counted eight favorable cases out of ten; it expresses a level of confidence within a particular model.

Probabilistic accounts impose coherence constraints drawn from probability theory. For a proposition pp, its negation ¬p\neg p, and a credence function PP, a standard requirement is

P(¬p)=1−P(p).P(\neg p)=1-P(p).

More generally, probabilities assigned to mutually exclusive alternatives must add appropriately. These requirements are defended through several arguments, including consistency in betting dispositions and measures of accuracy. A Dutch book argument shows how certain incoherent betting commitments can jointly guarantee a loss. Its significance depends on assumptions about preferences, stakes, and the connection between confidence and betting behavior.

Bayesian epistemology combines probabilistic coherence with rules for learning. When evidence EE has positive probability, Bayes’s theorem gives

P(H∣E)=P(E∣H)P(H)P(E),P(H\mid E)=\frac{P(E\mid H)P(H)}{P(E)},

where HH is a hypothesis. The posterior probability depends both on how expected the evidence is if the hypothesis is true and on the prior probabilities. Evidence that is common whether or not a hypothesis is true may provide little support, even when it is compatible with that hypothesis. This explains the importance of alternative explanations and base rates.

Outright belief cannot straightforwardly be identified with confidence above a fixed threshold. The lottery paradox, associated with Henry Kyburg’s work in 1961, illustrates the difficulty. In a large fair lottery with exactly one winner, each ticket is very likely to lose. If high probability suffices for belief, an agent might believe of every ticket that it will lose while also believing that some ticket will win. Combining these individually plausible beliefs produces inconsistency. Responses revise the threshold account, restrict the aggregation of beliefs, or separate categorical belief from probabilistic confidence more sharply.

Logical approaches model different features. Epistemic logic and related doxastic logics use operators such as BapB_a p, read as “agent aa believes that pp.” Standard idealized systems often make agents believe every logical consequence of their beliefs. This produces the problem of logical omniscience: real people do not know all the consequences of even elementary arithmetic assumptions. Models of bounded reasoning, awareness, and explicit versus implicit belief attempt to represent these limitations without abandoning formal analysis.

Formation, revision, and resistance to change

Beliefs arise through multiple routes. Perception supplies information about the immediate environment; memory preserves and reconstructs earlier information; testimony provides access to matters beyond direct experience; and inference combines existing commitments into new conclusions. These sources interact. Recognizing a bird, for example, depends on current visual information, learned categories, and background expectations. Belief formation is therefore not generally a sequence in which uninterpreted sensory data are first collected and only afterward assigned meaning.

Rational change can involve adding a new belief, withdrawing an old one, or reorganizing a larger network. Belief revision theory formalizes these operations. The AGM framework, introduced by Carlos Alchourrón, Peter Gärdenfors, and David Makinson in 1985, distinguishes expansion, contraction, and revision. Expansion adds information without addressing inconsistency; contraction removes a commitment; revision incorporates new information while restoring consistency where necessary. Its postulates describe ideal constraints rather than a complete psychological account of how people actually think.

The central difficulty is that conflicting evidence rarely dictates a unique repair. If a prediction fails, an investigator might reject the main hypothesis, an auxiliary assumption, a measurement claim, or some combination of these. Confirmation holism emphasizes that claims usually confront evidence as parts of larger systems. Rational revision therefore depends on considerations such as explanatory scope, independent support, and the cost of giving up other commitments. Merely preserving as many sentences as possible is not always the best measure of minimal change.

Psychological research identifies systematic obstacles to revision. Confirmation bias includes tendencies to seek, interpret, or remember information in ways favorable to existing expectations. In Peter Wason’s 1960 rule-discovery experiment, participants often tested examples compatible with their current hypothesis rather than cases that could discriminate it from alternatives. Belief perseverance can occur when a conclusion continues to influence judgment after its original evidential basis has been discredited. These effects vary with task, incentives, background knowledge, and the way information is presented.

Leon Festinger’s 1957 theory of cognitive dissonance examines the discomfort associated with incompatible cognitions and the processes through which people reduce it. Revision is one possible response, but so are reinterpretation, selective attention, and changes in the perceived importance of the conflict. Motivated reasoning similarly concerns the influence of desired conclusions on information processing. It does not imply that all strongly held beliefs are irrational or that every disagreement is caused by motivation.

Resistance to change can sometimes be reasonable. A single anomalous observation need not overturn a well-supported theory, especially when instruments are fallible. The epistemic issue is whether resistance remains responsive to the comparative quality of evidence. Stable commitment and openness to correction are therefore not simple opposites: both can contribute to inquiry when appropriately related to the strength and independence of the available support.

Psychological development and the attribution of belief

Understanding belief includes understanding that another person can represent the world incorrectly. Research on theory of mind investigates this capacity to attribute beliefs, desires, intentions, and related states. False-belief tasks are especially important because they separate what an observer knows from what another agent has reason to think. If a child predicts another person’s behavior solely from the actual location of an object, the child has not yet demonstrated the required distinction.

Heinz Wimmer and Josef Perner’s 1983 study helped establish the unexpected-transfer paradigm. A character leaves an object in one place; while absent, someone moves it; the participant is asked where the character will search. Simon Baron-Cohen, Alan Leslie, and Uta Frith’s 1985 study used the widely known Sally–Anne version. In many explicit verbal tasks, performance changes substantially during the preschool years, with successful responses becoming common around four to five years of age. Exact performance depends on language, memory, executive demands, and task design.

Some studies using looking time, anticipatory gaze, or helping behavior have been interpreted as evidence for earlier sensitivity to others’ false beliefs. These interpretations remain debated. Performance may reflect genuine belief tracking, learned behavioral regularities, sensitivity to perceptual access, or combinations of these capacities. Replication difficulties and differences between implicit and explicit measures prevent a simple conclusion that all such tasks measure one underlying ability in the same way.

Adult belief attribution is also imperfect. People can find it difficult to set aside information they themselves possess when judging what another person knows. Hindsight can make an outcome appear more predictable after it occurs. Conversational context complicates matters further: an assertion may express a belief, a guess, irony, politeness, or strategic presentation. Reliable attribution therefore requires more than reading a sentence in isolation.

Belief is not usually measurable through a single direct instrument. Researchers use verbal reports, choices, confidence ratings, reaction times, information seeking, and other behavioral indicators. Each measure has limitations. Reports can be affected by memory and social expectations; choices also depend on preferences and constraints. Measures of automatic associations may reveal patterns that diverge from explicit endorsement, but such divergence does not by itself establish that a person possesses a definite hidden propositional belief.

Neuroscientific research examines the systems involved in mentalizing, evaluating evidence, memory, and decision-making. These activities recruit distributed processes rather than a single known location for belief as such. The philosophical category encompasses many contents and functions, from immediate perceptual judgments to abstract convictions. Neural findings can constrain theories of the relevant processes, but do not independently settle whether belief should be defined representationally, dispositionally, or functionally.

Testimony, institutions, and religious commitment

Most beliefs about history, science, and distant events depend on other people. Social epistemology studies the conditions under which this dependence supports or undermines knowledge. Testimony includes ordinary conversation, teaching, records, journalism, and expert communication. A central question is whether accepting testimony requires independent evidence of a speaker’s reliability in each case, or whether testimony can provide an initial entitlement that remains defeasible when reasons for doubt arise.

Expertise introduces an unavoidable division of cognitive labor. A nonspecialist may be unable to reproduce the reasoning behind an expert conclusion yet possess good grounds for relying on it. Relevant considerations include the expert’s competence, access to evidence, domain-specific track record, and relationship to other qualified investigators. Institutional procedures also matter. In science, peer criticism, transparent methods, and reproducibility help make claims assessable beyond the authority of a single person. None of these procedures makes error impossible.

Apparent agreement must be distinguished from independent confirmation. Ten reports repeating one original claim do not supply the same support as ten independent investigations. Communication networks can amplify this problem by making a single source appear to have many separate endorsements. Conversely, genuine expert convergence may reflect extensive checking rather than conformity. Assessing a community’s beliefs therefore requires attention to how information travels, not merely to how many people repeat a conclusion.

Collective belief raises another set of questions. A committee, corporation, or court may officially endorse a proposition even when not every member personally believes it. Procedures for aggregating individual judgments can also produce inconsistent results. In a simple case, separate majorities may endorse two premises while another majority rejects their logical consequence because different members support each premise. Philosophers distinguish collective acceptance, institutionally authorized commitments, and stronger claims that groups themselves possess mental states.

Religious belief overlaps with these social processes but is not exhausted by them. Religious traditions transmit teachings through texts, ritual, testimony, and institutions, while individuals also cite experiences, arguments, and personal commitments. Philosophy of religion examines their epistemic status and the relation of belief to faith. In some traditions, orthodoxy gives doctrinal assent a prominent role; in others, practice, disciplined attention, or communal belonging may be more central to religious identity. Variation also exists within each tradition.

Interreligious disagreement raises questions about evidence, interpretation, and the significance of recognizing sincere and informed dissent. Similar questions arise in political and moral disagreement. The existence of disagreement alone does not establish that all positions are equally supported, but it can become relevant evidence when agents assess their own methods, access to information, and vulnerability to error.

Agency, responsibility, and the ethics of belief

The ethics of belief asks whether believing is subject to obligations, permissions, or forms of responsibility. The issue differs from whether an action based on a belief is right. A person might act cautiously despite an unjustified belief, or act harmfully on a belief formed through otherwise reasonable inquiry. Philosophers therefore distinguish epistemic evaluation, concerned with truth and support, from moral and practical evaluation, concerned with interests, duties, and consequences.

A classic dispute arose between William Kingdon Clifford and William James. In “The Ethics of Belief” of 1877, Clifford argued that believing on insufficient evidence is wrong, emphasizing the social consequences of careless credulity. In “The Will to Believe” of 1896, James defended the permissibility of commitment in certain options that are live, forced, and momentous when intellectual grounds cannot settle the choice. His argument was not a general license to accept any comforting proposition; it concerned cases in which withholding commitment could itself forfeit a possible good.

Responsibility is complicated by the apparent limits of direct voluntary control. A person usually cannot decide, merely by an act of will, to believe that an empty cup is full while looking into it. This challenges strong forms of doxastic voluntarism, the view that belief is under voluntary control. Yet people can influence beliefs indirectly by choosing what to investigate, whom to consult, and how carefully to examine objections. Accounts of responsibility consequently distinguish control over the resulting attitude from control over the conduct of inquiry.

Self-deception is a particularly difficult case. Someone may preserve a favorable conviction by avoiding unwelcome evidence or interpreting ambiguous information selectively. A simple model in which the person simultaneously knows a proposition is false and deliberately makes themselves believe it generates puzzles about divided awareness and intention. Alternative accounts explain self-deception through motivationally biased processes that need not involve a conscious plan to acquire a false belief.

Some philosophers argue that practical or moral stakes affect the epistemic standards for belief or knowledge. On such views, stronger support may be required when an error would have serious consequences. Others maintain that stakes change what it is reasonable to do without changing what the evidence supports. The dispute turns partly on whether categorical belief is inherently action-guiding or whether practical decisions should instead depend on graded confidence, possible outcomes, and the costs of error.

G. E. Moore identified a related puzzle about first-person commitment. An utterance such as “It is raining, but I do not believe that it is raining” can describe a logically possible situation, yet sounds defective when sincerely asserted in the present tense. Moore’s paradox suggests a special connection between assertion and presenting oneself as believing what is asserted. Explaining that connection remains relevant to theories of self-knowledge, rational coherence, and the norms governing public expression.