20.01.02 · philosophy / epistemology

Theories of justification: foundationalism, coherentism, reliabilism

stub3 tiersLean: nonepending prereqs

Anchor (Master): Quine, W. V. O. — Epistemology Naturalized (1969)

How do you know that what you believe is actually true? Having asked what knowledge is in the previous unit, we now ask a second question that is in some ways harder: what makes a belief justified? The JTB analysis treated justification as a single component, but the internal structure of justification — what grounds a reason, where the chain of warrant stops, whether the subject must be able to access the grounds — is itself one of the most contested terrains in epistemology. Three structural theories have dominated the modern debate. Foundationalism organises justification as a pyramid resting on basic beliefs. Coherentism organises it as a web of mutual support. Reliabilism looks past structure entirely and locates justification in the causal track record of the process that formed the belief. A fourth programme, Quine's naturalised epistemology, dissolves the normative question into empirical psychology. This unit reconstructs the regress argument that forces the choice, examines each theory and its signature objections, and extends to the contemporary positions — virtue epistemology, proper functionalism, evidentialism, contextualism, Bayesian and formal epistemology, and the social, feminist, and decolonial critiques that have reshaped the field since the 1990s.

Intuition Beginner

How do you know that what you believe is actually true? Philosophers call this the problem of justification. A belief counts as justified when you have a good reason to hold it. But what makes a reason good, and where does the chain of reasons end? Three main theories compete to answer these questions, and the choice between them shapes everything else in epistemology.

Foundationalism says some beliefs are basic. They need no support from other beliefs because they are self-evident or directly experienced. If you are seeing red right now, or grasping that two and three make five, you do not need further reasons — the belief carries its own warrant. Every other belief rests on this foundation, supported directly or indirectly by the basic ones. Descartes built his system on the cogito; Russell pointed to immediate acquaintance with sense-data as the base.

Coherentism denies that any belief is more basic than another. A belief is justified when it fits into a coherent system — a web where each strand supports the others. No single strand holds up the whole structure; mutual support does the work. Quine's image of the web of belief captures the idea: beliefs near the centre shift rarely, while those at the edge revise under pressure from experience.

Reliabilism, developed by Alvin Goldman, shifts the focus entirely. What matters is not the structure of your beliefs but how they were formed. If the process that produced a belief — vision in good light, intact memory, careful reasoning — is generally reliable, the belief is justified. Wishful thinking and hasty generalisation are not. Quine later proposed naturalising epistemology: turning it into a chapter of psychology, studying how people actually form beliefs rather than asking how they ought to.

Visual Beginner

Theory	Structure	Where justification lives	Main weakness
Foundationalism	Pyramid: basic beliefs support derived ones	In the basic beliefs at the base	What justifies the basic beliefs themselves?
Coherentism	Web: beliefs mutually support each other	In the coherence of the whole system	Could a coherent fiction be justified?
Reliabilism	Process: reliable mechanisms produce beliefs	In the track record of the process	How do you specify the relevant process?

Worked example Beginner

Justifying "there is a cup on the desk"

Suppose you are sitting at your desk and you believe there is a cup in front of you. The belief is true — there really is a cup. But why is it justified? Each of the three theories gives a different answer.

Foundationalism. Your perceptual experience — the direct visual awareness of a white cylindrical object on a brown surface — is a basic belief. It needs no further support. Your belief that there is a cup is derived from this basic perceptual belief plus background knowledge about cups. The chain of reasons stops at experience.

Coherentism. The belief fits coherently with hundreds of others: you remember putting the cup there, you can reach for it and feel it, others in the room confirm it, the laws of physics permit a cup to rest on a desk. No single belief is foundational. The cup-belief is justified because removing it would tear a hole in the web.

Reliabilism. The belief was produced by vision under normal lighting at close range — a process that has reliably delivered true beliefs about nearby objects throughout your life. The justification does not depend on whether you can articulate the structure of your reasons. It depends on the process being the kind that tracks truth.

The three theories agree that this belief is justified. They disagree about why. That disagreement matters when cases get harder — illusions, expert testimony, scientific instruments, perceptual evidence in courtrooms — and the simple answers start to diverge.

Check your understanding Beginner

Formal definition Intermediate+

A theory of justification specifies the structural conditions under which a subject $S$ is justified in holding a belief that $p$ . The three classical positions differ on where the justificatory work is done.

Foundationalism. Justification has a two-tier structure. There are basic beliefs — beliefs justified independently of any other belief — and derived beliefs, justified by their inferential relation to basic beliefs. Formally, $S$ 's belief that $p$ is foundationally justified if either (i) $p$ is basic for $S$ , or (ii) $p$ is inferred by $S$ from a set of basic beliefs (or from beliefs ultimately supported by basic beliefs) via a truth-preserving or warrant-transmitting inferential chain. Classical foundationalism (Descartes) demands that basic beliefs be infallible — immune to error. Moderate foundationalism (Pollock, Alston, Plantinga) relaxes this to prima facie justification: a basic belief is justified unless the subject possesses specific defeaters.

Coherentism. A belief is justified by its coherence with the subject's total belief system. There are no basic beliefs; every belief's justification depends on its relations to others. Formally, $S$ 's belief that $p$ is justified iff $p$ belongs to a system of beliefs $Σ_{S}$ that is maximally coherent — where coherence is a function of explanatory integration, consistency, and comprehensiveness. The unit of justification is $Σ_{S}$ , not $p$ in isolation ^{[BonJour 2010 Ch. 4]}.

Reliabilism. A belief is justified iff it was produced by a reliable cognitive process — one that yields a sufficiently high ratio of true to false beliefs across an appropriate range of possible situations. Formally, $S$ 's belief that $p$ is reliably justified iff there exists a process type $π$ such that $p$ was produced by an instance of $π$ , and the actual-world reliability of $π$ exceeds a threshold $r$ . Reliabilism is externalist: $S$ need not know, or be able to identify, that $π$ produced the belief ^{[Goldman 1976]}.

The three theories respond differently to Agrippa's trilemma — the ancient argument that every attempt to justify a belief encounters one of three terminations: infinite regress (each justifying belief needs another), circularity (the chain loops back on itself), or dogmatic stopping (the chain halts at an unjustified belief). Foundationalism accepts the third horn but denies that the stopping beliefs are dogmatic — they are justified without being justified by other beliefs. Coherentism accepts a restricted form of the second horn: mutual support is not vicious circularity but holistic warrant. Reliabilism rejects the trilemma's framing: justification is not a chain of beliefs at all but a property of the process, so the regress never arises.

The regress problem

The regress argument, reconstructed precisely:

P1. For any belief $p$ that $S$ holds, either $p$ is justified or $p$ is not justified.

P2. If $p$ is justified, it is justified either (a) by some further belief $q$ that $S$ holds, or (b) by something that is not a belief (perception, intuition, the reliability of a process), or (c) by nothing at all.

P3. If every justified belief is justified by further beliefs (option (a) exclusively), then either the chain of justifying beliefs is infinite, or it terminates, or it loops.

P4. Finite minds cannot entertain infinitely many justifying beliefs. So infinite chains are unavailable to us.

P5. A chain that terminates in an unjustified belief transmits no warrant. So unadorned termination fails.

P6. Circular justification is epistemically valueless: a belief cannot receive warrant from itself, even via a long detour.

C. Therefore, if justification runs only through beliefs, it is impossible. To secure justification, we must accept either non-doxastic justifiers (foundationalism's basic beliefs or reliabilism's reliable processes) or holistic mutual support (coherentism's web).

This argument does not decide among the three theories. It establishes that purely inferential, linear justification is untenable. Each theory fills the gap differently.

Key argument: the structure of justification and Agrippa's trilemma Intermediate+

Foundationalism and the myth of the given

Classical foundationalism, in its strongest Cartesian form, requires basic beliefs to be infallible. Descartes' cogito — "I think, therefore I am" — was the prototype: a belief so secure that not even an omnipotent deceiver could make it false. Sense perception, memory, and testimony do not meet this standard. They can deceive. If only infallible beliefs count as basic, the foundation is narrow — too narrow to support the edifice of ordinary empirical knowledge. This is the first pressure on classical foundationalism.

Moderate foundationalism widens the base. A perceptual belief like "there is a table before me" is prima facie justified — justified in the absence of defeaters — even though it is fallible. The subject need not be infallible; they need only lack reason to doubt. This generosity comes at a cost. If fallible perceptual beliefs can be basic, what excludes any other fallible belief from the foundation? The criterion for basicness becomes harder to state without arbitrariness.

Wilfrid Sellars's "Empiricism and the Philosophy of Mind" (1956) delivered the deepest critique. Sellars attacked the "Myth of the Given": the idea that experience provides non-inferential, pre-conceptual justifiers that ground empirical knowledge. For Sellars, every awareness is already conceptual. To have a perceptual belief is to apply concepts, and conceptual application presupposes other beliefs. There is no raw "given" prior to belief that could serve as a foundation. If Sellars is right, the foundationalist's basic beliefs are either inferential (and the regress returns) or non-justificatory (and the foundation is inert) ^{[BonJour 2010 Ch. 3]}.

Coherentism: BonJour's web and its problems

Laurence BonJour's The Structure of Empirical Knowledge (1985) offered the most rigorous coherentist defence. On BonJour's account, a belief is justified when it coheres with the subject's overall system — where coherence encompasses explanatory integration (the system explains why the belief is true), consistency (no contradictions), and comprehensiveness (the system covers a wide range of phenomena). The "observation" that the system must accommodate is not a basic belief but a constraint: the system must explain why the subject has the experiences they have.

Two objections dominate the literature. The isolation problem: a perfectly coherent fairy tale is internally consistent and explanatorily integrated, yet its propositions are not justified. Coherence alone cannot distinguish a true account of the world from a well-constructed fiction. The input problem: how does the web connect to reality? If experience enters only as a belief within the system, then the system's contact with the external world is unexplained — the web floats free. BonJour acknowledged these problems and eventually judged them fatal. In his 2003 paper "A Version of Internalist Foundationalism," he abandoned coherentism and returned to foundationalism, arguing that "cognitive contact" with reality through non-inferential awareness is indispensable ^{[BonJour 2010 Ch. 4-5]}.

Reliabilism: Goldman's process theory

Alvin Goldman's process reliabilism relocates justification outside the subject's reflective access. A belief is justified iff it was produced by a reliable cognitive process — vision in normal conditions, intact memory, careful logical reasoning. Unreliable processes — wishful thinking, hasty generalisation, reasoning under emotional duress — fail to justify. The theory handles Gettier-style cases and ordinary perceptual knowledge without requiring the subject to identify or justify the process.

The generality problem asks how finely to individuate the process. "Looking at a clock" might be reliable; "looking at this particular stopped clock at 2:00 PM" is not. Since any instance can be described at many levels of generality, and reliability varies across descriptions, the theory needs a principled criterion for selecting the right level. No consensus exists.

The new evil demon problem is more damaging. An envatted brain — perfectly deceived by a demon — has the same internal cognitive processes as a normally embodied subject. The processes are unreliable in the demon world, yet the envatted subject seems as justified as anyone. If reliabilism says the subject is unjustified, it conflicts with the strong intuition that the subject is doing everything an epistemically responsible agent can do. Reliabilists have distinguished justification (which the demon-world subject has) from knowledge (which they lack), but the boundary remains contested ^{[Goldman 1976]}.

Counterexamples to common slips

"Foundationalism requires certainty." Only classical foundationalism demands infallible basic beliefs. Moderate foundationalism allows fallible, defeasible basic beliefs — perceptual beliefs that are prima facie justified but can be overturned. Conflating the two versions is a standard misreading.
"Coherentism says any consistent system is justified." Coherence is stronger than consistency. A coherent system must also be explanatorily integrated and comprehensive. A fairy tale may be consistent, but it does not explain the subject's actual experiences; it fails the comprehensiveness and explanatory constraints that coherentists impose.
"Reliabilism is just about truth ratio." The reliability of a process involves not only the proportion of true beliefs it produces but also the range of situations over which it operates. A process that is reliable in normal environments but unreliable in abnormal ones raises subtle questions about which environments are relevant — questions that push the theory toward "normal world" or "relevant alternatives" amendments.

Exercises Intermediate+

Exercise 1 (easy, multiple choice).

Agrippa's trilemma says that any attempt to justify a belief by appeal to further beliefs terminates in one of three ways. Which of the following is not one of the three?

A. Infinite regress

B. Circular justification

C. A self-evident axiom known by intuition

D. An arbitrary stopping point with no justification

Hint

The trilemma is meant to show that linear justificatory chains are problematic. Which option would be a solution rather than a problem?

Answer

Option C. The three horns of Agrippa's trilemma are infinite regress, circularity, and arbitrary termination. A self-evident axiom known by intuition is the foundationalist's response to the trilemma — a non-inferential justifier that halts the regress without arbitrariness. It is not itself one of the three problematic terminations.

Exercise 3 (medium, argument reconstruction).

Reconstruct the isolation objection to coherentism in premise-conclusion form. Then explain how a coherentist might respond.

Hint

The objection turns on whether internal coherence alone can distinguish a true account of the world from a well-constructed fiction.

Answer

P1. A coherentist system is justified by its internal coherence — consistency, explanatory integration, comprehensiveness.

P2. It is possible to construct a fictional system (a detailed, self-consistent fantasy world) that is as coherent as the subject's actual belief system.

P3. If two systems are equally coherent, the coherentist has no basis for declaring one justified and the other not.

C. Therefore, coherence alone is not sufficient for justification; the theory cannot distinguish genuine knowledge from sophisticated fiction.

A coherentist response appeals to the input constraint: the actual subject's system must explain their actual experiences, while the fictional system need not. The coherence relation is not purely internal — it connects to experience via the requirement that the system explain the subject's observational beliefs. Whether this response succeeds depends on whether the observational constraint can be formulated without reintroducing foundational basic beliefs.

Exercise 4 (medium, short answer).

Explain the generality problem for reliabilism. Why does it threaten the theory's ability to deliver verdicts about specific beliefs?

Hint

The same belief-forming event can be described at many levels of generality, and the reliability verdict changes with the description.

Answer

The generality problem arises because any given belief-forming process can be type-identified at many levels of specificity. Consider someone who forms the belief "it is 2:00" by looking at a clock. Is the relevant process "visual perception" (highly reliable), "clock-reading" (generally reliable), or "reading this particular stopped clock" (unreliable)? The reliability verdict flips depending on which description we choose. Since the theory says a belief is justified iff produced by a reliable process, it must specify which process type is the relevant one — but no principled, non-ad hoc criterion for selecting the level of generality has won consensus. Without such a criterion, reliabilism cannot deliver determinate verdicts: for any belief, there is a description of the process under which it is reliable and another under which it is not.

Exercise 5 (medium, defence).

Quine's naturalised epistemology replaces normative questions ("how ought we to form beliefs?") with descriptive ones ("how do we in fact form beliefs?"). In 4-6 sentences, explain one objection to this replacement and one reason a naturalised epistemologist might give in reply.

Hint

The objection concerns the loss of the normative dimension — the whole point of epistemology, traditionally, was to tell us what we should believe. The reply might appeal to the continuity of science and philosophy.

Answer

The standard objection is that naturalised epistemology changes the subject. Traditional epistemology is normative: it asks what justifies a belief, what we ought to believe given our evidence, what distinguishes knowledge from mere opinion. A purely descriptive psychology of belief-formation cannot answer these questions. Describing how people actually reason does not tell us whether their reasoning is any good — the is-ought gap blocks the inference.

A Quinean reply is that the normative and descriptive projects are continuous. Our epistemic norms are themselves products of the same scientific enterprise that studies perception and cognition. We evaluate reasoning by the same empirical standards we use elsewhere: a method is "good" if it tends to produce accurate predictions and reliable theories. The normative does not vanish; it is naturalised — grounded in the empirical track record of cognitive methods rather than in a priori stipulation.

Exercise 6 (hard, critique).

BonJour defended coherentism in 1985 and abandoned it in 2003, returning to a form of internalist foundationalism. In 5-7 sentences, reconstruct the reasoning that led to his recantation. What problem did he conclude coherentism cannot solve?

Hint

BonJour came to believe that coherentism lacks something indispensable: a connection between the belief system and the external reality it purports to describe. What is this "cognitive contact," and why can coherence alone not secure it?

Answer

BonJour's central insight was that coherentism, however refined, cannot secure the cognitive contact between a belief system and the reality it represents. The isolation and input problems are two faces of this deficiency. A purely coherent system is self-supporting, but self-support is not truth-conducive: the system could be a elaborate fantasy. For the system to count as knowledge of an external world, there must be some non-coherent link — a constraint imposed by reality itself on the beliefs the subject forms. But any such link, if it is to be justificatory, must take the form of a non-inferential awareness — exactly the kind of foundational basic belief that coherentism was designed to eliminate. BonJour concluded that the regress argument, properly understood, demands not mutual support but a stopping point in experience. Coherence is necessary for justification but not sufficient; it must be anchored by something the subject encounters directly.

Advanced results Master

The internalism-externalism divide

The deepest fault line in contemporary epistemology runs between internalism and externalism about justification. Internalism — defended by Conee and Feldman, BonJour (in his later work), and much of the traditional rationalist and empiricist tradition — holds that the justificatory status of a belief is determined by factors accessible to the subject's reflection. If two subjects are internally alike (same evidence, same reasoning, same conscious perspective), they are equally justified, regardless of external differences. Externalism — paradigmatically Goldman's reliabilism — denies this. What justifies a belief may be a fact about the world (the reliability of the process, the proper functioning of the cognitive faculty) that the subject cannot access by reflection alone.

The dispute is not merely technical. It determines whether epistemology is a first-personal discipline (concerned with what the subject can see from the inside) or a third-personal one (concerned with objective facts about cognitive processes). The new evil demon problem sharpens the tension: an internally identical duplicate in a demon world has the same reflective access, but if reliabilism is right, only the non-demon subject is justified. Internalists take this as a reductio of externalism; externalists accept the consequence and argue that the demon-world subject has justification (an internal matter) but not knowledge (which requires external reliability).

Virtue epistemology and proper functionalism

Ernest Sosa's virtue epistemology reframes justification and knowledge in terms of intellectual competence. A belief is apt — Sosa's term for knowledge-grade justification — when it is true because produced by a competent faculty. The analysis has three tiers: a belief is accurate if true, adroit if produced by a competence (vision, memory, logical reasoning), and apt if accurate because adroit. The "because" matters: in a Gettier case, the belief may be accurate and adroit but not apt, because the truth is not creditable to the competence ^{[BonJour 2010]}.

Sosa models belief on archery. An archer's shot can be accurate (hit the target), adroit (skilful form), or apt (on target because of the skill). Gettier cases are lucky shots that happen to hit; the skill is present but did not cause the success. Virtue epistemology handles the generality problem by relocating it: instead of specifying the "process," the theorist specifies the "competence" — a stable disposition of a cognitive faculty to deliver true beliefs in appropriate conditions. The specification is still nontrivial, but the vocabulary of competence is richer than the vocabulary of process.

Linda Zagzebski's virtue responsibilism shifts the focus from faculties to character. Knowledge requires the exercise of intellectual virtues — open-mindedness, thoroughness, intellectual courage — understood as traits of the whole person, not just isolated cognitive mechanisms. The responsibilist programme connects epistemology to ethics: intellectual virtues are a species of moral virtue, and the virtuous knower is someone whose character is reliably oriented toward truth.

Alvin Plantinga's proper functionalism analyses warrant (his term for the property that turns true belief into knowledge) in terms of proper function. A belief has warrant iff it is produced by cognitive faculties functioning properly, in the cognitive environment for which they were designed, according to a design plan successfully aimed at truth. Plantinga's reformed epistemology extends this to religious belief: belief in God can be properly basic — warranted without inferential support — if produced by a functioning sensus divinitatis in the appropriate environment. The framework is controversial, but it illustrates how externalist resources can be deployed beyond the reliabilist paradigm.

Evidentialism, contextualism, and pragmatic encroachment

Evidentialism, developed by Feldman and Conee, holds that epistemic justification supervenes on the subject's evidence. What you ought to believe is entirely determined by the evidence available to you. In their "fake country" case, a subject has overwhelming evidence that a country called "Terra" exists (encyclopedia entries, maps, testimony), but Terra is in fact a fictional invention that has fooled everyone. The evidentialist verdict: the subject is justified in believing Terra exists, because the evidence supports it, even though the belief is false. This makes justification strongly factive-independent: evidence, not truth, is the arbiter.

Contextualism, developed by DeRose and Lewis, holds that the truth-conditions of knowledge attributions (" $S$ knows that $p$ ") vary with the conversational context. In DeRose's "bank cases," a depositor who needs to cash a cheque on Saturday asks whether the bank will be open. In a low-stakes version (no urgency), we attribute knowledge: "She knows the bank is open." In a high-stakes version (a large cheque that must clear Monday), the same evidence no longer suffices: "She doesn't know." The contextualist says the standards for "knows" shift with the practical context of the attributor, not the subject.

Pragmatic encroachment, defended by Stanley and by Fantl and McGrath, goes further: it is not the attributor's context but the subject's practical situation that matters. Whether $S$ knows that $p$ depends in part on what is at stake for $S$ . High stakes raise the threshold for knowledge; if being wrong would be costly, more evidence is required. This is a form of subject-sensitive invariantism: the standards for knowledge are fixed (not context-variable) but sensitive to the subject's practical interests. The debate between contextualism and pragmatic encroachment turns on whether the shift occurs at the level of language (what "knows" means in a context) or at the level of fact (what knowledge actually requires given the stakes).

BonJour's coherentism and its recantation

BonJour's 1985 The Structure of Empirical Knowledge remains the most rigorous coherentist treatise. The argument proceeds in two stages. First, the regress argument establishes that linear inferential justification is untenable — every option (infinite regress, circularity, dogmatic stopping) fails. Second, coherentism is presented as the only remaining alternative: a belief is justified by membership in a system that meets stringent coherence constraints (explanatory integration, consistency, comprehensiveness). BonJour insisted that the "cognitive contact" requirement — the system must be constrained by experience — could be met without reintroducing basic beliefs.

By 2003, BonJour had concluded that this was wrong. The experience-constraint either reduces to a basic belief (in which case coherentism collapses into foundationalism) or is merely causal (in which case it has no justificatory force). In "A Version of Internalist Foundationalism," BonJour argued that non-inferential seeming — the conscious awareness that things are thus-and-so — provides genuine justificatory support without being a belief. This is internalist (the seeming is accessible to reflection) and foundationalist (it halts the regress without inference). The recantation was significant: the most prominent coherentist of the late twentieth century concluded that coherentism is untenable.

Gettier descendants: tracking, safety, and causal theories

The Gettier problem (covered in 20.01.01) generated a family of conditions intended to replace or supplement JTB. Goldman's causal theory (1967) holds that $S$ knows that $p$ iff the fact that $p$ is causally connected to $S$ 's belief that $p$ . This handles perceptual knowledge (the table causes the visual belief) but fails for mathematical and historical knowledge, where the causal link is obscure.

Nozick's tracking theory holds that $S$ knows that $p$ iff (i) $p$ is true, (ii) $S$ believes $p$ , (iii) if $p$ were false, $S$ would not believe $p$ (sensitivity), and (iv) if $p$ were true, $S$ would believe $p$ (adherence). The counterfactual analysis handles Gettier cases: in the stopped-clock case, if it were not 2:00, the subject would still believe it is 2:00 (the clock reads 2:00 regardless), so the sensitivity condition fails. But sensitivity is not closed under known entailment, which generates paradoxical results for deductive inference.

Sosa's safety condition — $S$ 's belief could not easily have been false — avoids some of sensitivity's problems and connects naturally to virtue epistemology. A safe belief is one produced by a competence that would not easily go wrong. The fake-barns case (Goldman 1976) is the canonical stress test: the subject's belief that there is a barn is true, but unsafe — in nearby worlds, they would have looked at a facade and formed a false belief.

Bayesian and formal epistemology

Bayesian epistemology models degrees of belief as probabilities conforming to the probability calculus. A rational agent's credence in $p$ , conditional on evidence $E$ , is updated by Bayes' theorem: $P (p ∣ E) = P (p) \cdot P (E ∣ p) / P (E)$ . Conditionalisation — updating by setting the posterior equal to the prior conditional on the new evidence — is the normative rule for belief revision. Dutch book arguments (Ramsey, de Finetti) show that agents whose credences violate the probability axioms are vulnerable to sure-loss betting arrangements, providing a pragmatic vindication of probabilistic coherence.

The programme divides between subjective Bayesianism (any prior distribution satisfying the probability axioms is permissible; the evidence does the constraining) and objective Bayesianism (additional principles — maximum entropy, the principle of indifference — constrain the prior). The debate connects to the problem of induction: Bayesian learning can show how evidence raises the probability of a hypothesis, but it cannot, on its own, justify the assumption that the future will resemble the past.

Jaakko Hintikka's Knowledge and Belief (1962) inaugurated formal epistemic logic. The logic introduces operators $K_{a} p$ ("agent $a$ knows that $p$ ") and $B_{a} p$ ("agent $a$ believes that $p$ "), interpreted over possible worlds: $K_{a} p$ holds at world $w$ iff $p$ holds at every world accessible from $w$ . The KK thesis — if $S$ knows $p$ , then $S$ knows that $S$ knows $p$ — is valid on some accessibility relations and invalid on others. Moore's paradox (" $p$ but I do not believe that $p$ ") is a constraint on the interaction of belief and truth. Epistemic logic provides the formal substrate for contemporary work on distributed knowledge, common knowledge, and the logic of information flow.

Social, feminist, and decolonial epistemology

Traditional epistemology treats the knower as a solitary reasoner. Social epistemology — inaugurated by Goldman's Knowledge in a Social World (1999) — asks how knowledge is produced, transmitted, and assessed in communities. Testimony is the primary mechanism: most of what anyone knows, they learned from others. The reductionist holds that testimony is justified only insofar as it can be backed by non-testimonial evidence (the testifier's track record). The anti-reductionist holds that testimony is a basic source of knowledge — we are entitled to trust it unless there are specific reasons not to.

Miranda Fricker's Epistemic Injustice (2007) showed that social power distorts epistemic practice. Testimonial injustice occurs when a speaker receives less credibility than they deserve because of identity prejudice — the gaslit woman whose testimony is dismissed as "hysterical," the Black defendant whose account is disbelieved because of racial stereotype. Hermeneutical injustice occurs when a group lacks the conceptual resources to make sense of their experience because the dominant interpretive frameworks exclude them — the concept of "sexual harassment" did not exist before the 1970s, leaving women without the vocabulary to name and contest a pervasive harm. Kristie Dotson extends the analysis to epistemic violence, where the testimony of marginalised groups is rendered unintelligible by the frameworks imposed on them.

Feminist epistemology challenges the ideal of the detached, value-neutral observer. Sandra Harding's standpoint epistemology argues that knowledge from marginalised positions has epistemic privilege: those at the bottom of a social hierarchy must understand both their own perspective and the dominant perspective that governs them, while the dominant group need only understand their own. Helen Longino's Science as Social Knowledge argues that objectivity is not achieved by individual neutrality but through the critical interaction of diverse perspectives — a community with heterogeneous assumptions catches biases that a homogeneous community cannot see. Lorraine Code's What Can She Know? examines how gender shapes epistemological practice, and Lynn Hankinson Nelson develops a feminist naturalised epistemology that integrates these insights with Quine's programme.

Decolonial epistemology presses the critique further. Walter Mignolo's epistemic disobedience calls for the delinking of knowledge production from Eurocentric frameworks. Raewyn Connell's Southern Theory argues that metropolitan sociology has systematically ignored knowledge produced in the Global South. Boaventura de Sousa Santos's epistemologies of the South argues that the global North has monopolised the production of "valid" knowledge, and that the cognitive traditions of the South — oral, embodied, spiritual, communal — are epistemic resources, not objects of study. Linda Tuhiwai Smith's Decolonizing Methodologies (1999) shows that research itself has been a colonial practice, and argues for Indigenous control over the research conducted about Indigenous communities — from question formation through publication.

Connections Master

Connection to 20.01.01 — Epistemology: knowledge, justification, and truth

This unit extends the analysis of justification introduced in 20.01.01. Where 20.01.01 asked what knowledge is (the JTB analysis, the Gettier problem), this unit asks what justification is — its internal structure, its regressive character, and the three classical theories that answer the regress. The Gettier descendants covered in the Master tier (tracking, safety, causal theory) presuppose the material on JTB from 20.01.01.

Connection to 20.01.03 — Skepticism

The theories of justification examined here frame the skeptical challenge that the next unit takes up. Foundationalism faces the skeptic's demand for indubitable foundations; coherentism faces the isolation objection that the skeptic exploits; reliabilism faces the new evil demon problem, which is structurally identical to the skeptical scenario. The hook hooks_out: 20.01.03 records this dependency: one cannot evaluate skeptical arguments without first fixing a theory of justification.

Connection to 20.08.NN — Philosophy of science

The structure-of-justification debate applies directly to scientific knowledge. A theory of scientific evidence — what counts as support for a hypothesis — presupposes a position on whether justification is foundational (individual observations ground theoretical claims), coherentist (a theory is justified by its fit with the total system), or reliabilist (the scientific method is a reliable process). The Duhem-Quine problem (a hypothesis cannot be tested in isolation) is a coherentist challenge to naive foundationalism about scientific evidence.

Connection to 25-logic and 42-mathematical-logic — Epistemic and modal logic

Hintikka's epistemic logic (covered in the Master tier) provides the formal substrate for modelling knowledge and belief. The possible-worlds semantics for $K_{a} p$ connects directly to the modal logic curriculum in §25 and §42. A formal treatment of the KK thesis, Moore's paradox, and the closure principle for knowledge requires the tools of §25.

Connection to 29-psychology — Naturalised epistemology

Quine's naturalised epistemology dissolves the normative question into empirical psychology. The programme connects to cognitive psychology (how do people actually form beliefs?), developmental psychology (how does knowledge arise in the child?), and the cognitive sciences of perception and memory. The reliability of belief-forming processes — the central reliabilist notion — is an empirical matter that psychology is equipped to investigate.

Historical and philosophical context Master

The problem of justification is ancient. Aristotle's Posterior Analytics assumed a foundational structure: scientific knowledge (episteme) rests on first principles known by intuition (nous), which are indemonstrable but more certain than what is derived from them. This is the earliest statement of the foundationalist programme. The Stoics developed a similar picture: cognitive impressions (kataleptike phantasia) — impressions so clear and vivid that they could not be false — provided the foundation for all knowledge. The Stoic-Skeptical debates of the Hellenistic period turned on whether such indubitable impressions exist.

Descartes's Meditations (1641) gave the regress argument its modern form. By systematically doubting every belief that could possibly be false — sensory beliefs (dreams), mathematical beliefs (evil demon) — Descartes searched for a belief immune to doubt. The cogito ("I think, therefore I am") was the result: a belief whose truth is guaranteed by the act of thinking it. The Cartesian programme attempted to rebuild the entire structure of knowledge on this indubitable foundation, using only inference chains as strong as the deduction itself. The programme's ambition was total; its success was partial. The cogito secured one belief, but the bridge from the isolated thinking self to the external world proved hard to construct without circularity — a problem the British empiricists inherited and worsened ^{[Russell 1912]}.

Locke's Essay Concerning Human Understanding (1689) grounded knowledge in ideas derived from sensation and reflection. Simple ideas — the direct products of experience — are the foundation; complex ideas are built from them. Hume's Enquiry (1748) pushed empiricism to its skeptical conclusion: even the causal connection between experiences cannot be directly perceived, and induction has no rational foundation. Kant's Critique of Pure Reason (1781) attempted to rescue knowledge from Humean skepticism by arguing that the mind contributes a priori structures (the categories) that organise experience. Knowledge requires both sensory input and conceptual form. The Kantian synthesis defined the terms of epistemological debate for the next century and a half.

The twentieth century opened with logical positivism's verification principle: a proposition is meaningful only if it is analytically true or empirically verifiable. The programme grounded knowledge in protocol sentences — reports of immediate experience — which were to serve as the foundational base. The programme collapsed under internal critique (the verification principle is itself neither analytic nor verifiable) and under Sellars's attack on the Myth of the Given (1956), which argued that even protocol sentences are concept-laden and inferential ^{[BonJour 2010 Ch. 3]}.

W. V. O. Quine's "Epistemology Naturalized" (1969) marked a decisive rupture ^{[Quine 1969]}. Quine argued that the Cartesian project — grounding science in sense-data by rational reconstruction — had failed. The Humean and Carnapian programmes had not delivered the promised foundation. Quine's proposal was to reconceive epistemology as part of psychology: the study of how sensory stimulation produces theory. The normative question ("what ought we to believe?") gives way to the descriptive one ("how do we in fact form beliefs?"). The web-of-belief metaphor, introduced in "Two Dogmas of Empiricism" (1951) and developed in Word and Object (1960), pictured the belief system as a field of sentences whose interior beliefs shift rarely and whose peripheral beliefs revise under pressure from experience — a holistic, coherentist image that stood against the foundational pyramid.

Alvin Goldman's reliabilism, introduced in "A Causal Theory of Knowing" (1967) and developed in "Discrimination and Perceptual Knowledge" (1976) ^{[Goldman 1976]} and Epistemology and Cognition (1986), opened the externalist programme. By locating justification in the objective reliability of the belief-forming process rather than in the subject's reflective access, Goldman broke with the internalist consensus that had dominated since Descartes. The subsequent decades saw the internalism-externalism debate become the organising question of the field, with virtue epistemology (Sosa 1991, Zagzebski 1996), proper functionalism (Plantinga 1993), and evidentialism (Feldman and Conee 1985) taking positions on the map.

The 1990s and 2000s brought two further shifts. Social epistemology (Goldman 1999, Fricker 2007) expanded the focus from the solitary knower to communities and institutions. Feminist and decolonial epistemology (Harding 1986, Longino 1990, Smith 1999, Mignolo 2000) challenged the universality of the knowing subject, arguing that the standard picture encodes a particular — male, European, colonial — standpoint as neutral. These critiques did not replace the classical theories; they reframed the questions the theories were trying to answer, insisting that who knows and under what conditions are not incidental to whether the belief is justified.

Bibliography Master

Descartes, R. (1641). Meditationes de Prima Philosophia. English trans. Meditations on First Philosophy, Cambridge University Press (1996).
Locke, J. (1689). An Essay Concerning Human Understanding.
Hume, D. (1748). An Enquiry Concerning Human Understanding.
Kant, I. (1781). Critik der reinen Vernunft. English trans. Critique of Pure Reason, Cambridge University Press (1998).
Russell, B. (1912). The Problems of Philosophy. Oxford University Press.
Sellars, W. (1956). "Empiricism and the Philosophy of Mind." In Minnesota Studies in the Philosophy of Science, vol. 1, pp. 253-329. University of Minnesota Press.
Quine, W. V. O. (1951). "Two Dogmas of Empiricism." The Philosophical Review 60, 20-43.
Quine, W. V. O. (1969). "Epistemology Naturalized." In Ontological Relativity and Other Essays, pp. 69-90. Columbia University Press.
Goldman, A. I. (1967). "A Causal Theory of Knowing." The Journal of Philosophy 64, 357-372.
Goldman, A. I. (1976). "Discrimination and Perceptual Knowledge." The Journal of Philosophy 73, 771-791.
Goldman, A. I. (1979). "What Is Justified Belief?" In Justification and Knowledge, ed. G. S. Pappas, pp. 1-23. Reidel.
Goldman, A. I. (1986). Epistemology and Cognition. Harvard University Press.
Goldman, A. I. (1999). Knowledge in a Social World. Oxford University Press.
BonJour, L. (1985). The Structure of Empirical Knowledge. Harvard University Press.
BonJour, L. (2003). "A Version of Internalist Foundationalism." In Epistemic Justification, ed. L. BonJour and E. Sosa, pp. 3-96. Blackwell.
BonJour, L. (2010). Epistemology: Classic Problems and Contemporary Responses, 2nd ed. Rowman and Littlefield.
Plantinga, A. (1993). Warrant and Proper Function. Oxford University Press.
Sosa, E. (1991). Knowledge in Perspective: Selected Essays in Epistemology. Cambridge University Press.
Sosa, E. (2007). A Virtue Epistemology: Apt Belief and Reflective Knowledge, Vol. I. Oxford University Press.
Zagzebski, L. (1996). Virtues of the Mind: An Inquiry into the Nature of Virtue and the Ethical Foundations of Knowledge. Cambridge University Press.
Feldman, R. and Conee, E. (1985). "Evidentialism." Philosophical Studies 48, 15-34.
DeRose, K. (1992). "Contextualism and Knowledge Attributions." Philosophy and Phenomenological Research 52, 913-929.
Lewis, D. (1996). "Elusive Knowledge." Australasian Journal of Philosophy 74, 549-567.
Stanley, J. (2005). Knowledge and Practical Interests. Oxford University Press.
Nozick, R. (1981). Philosophical Explanations. Harvard University Press.
Hintikka, J. (1962). Knowledge and Belief: An Introduction to the Logic of the Two Notions. Cornell University Press.
Ramsey, F. P. (1926). "Truth and Probability." In Foundations of Mathematics and Other Logical Essays, pp. 156-198. Routledge and Kegan Paul (1931).
Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford University Press.
Harding, S. (ed.) (1986). Feminism and Methodology: Social Science Issues. Indiana University Press.
Longino, H. (1990). Science as Social Knowledge: Values and Objectivity in Scientific Inquiry. Princeton University Press.
Code, L. (1991). What Can She Know? Feminist Theory and the Construction of Knowledge. Cornell University Press.
Mignolo, W. (2000). Local Histories/Global Designs: Coloniality, Subaltern Knowledges, and Border Thinking. Princeton University Press.
Connell, R. (2007). Southern Theory: The Global Dynamics of Knowledge in Social Science. Polity.
Santos, B. de Sousa (2014). Epistemologies of the South: Justice Against Epistemicide. Paradigm.
Smith, L. T. (1999). Decolonizing Methodologies: Research and Indigenous Peoples. Zed Books.
Dotson, K. (2011). "Tracking Epistemic Violence, Tracking Practices of Silencing." Hypatia 26, 236-257.
Fantl, J. and McGrath, M. (2009). Knowledge in an Uncertain World. Oxford University Press.

Prerequisites

20.01.01 pending

Used in

20.01.03

Tier anchors

beginner: Russell, B. — The Problems of Philosophy (1912)
intermediate: BonJour, L. — Epistemology: Classic Problems and Contemporary Responses (2010)
master: Quine, W. V. O. — Epistemology Naturalized (1969)

References

Russell, B. — The Problems of Philosophy (Oxford, 1912) · Ch. 1-2 Appearance and reality, knowledge by acquaintance
BonJour, L. — Epistemology: Classic Problems and Contemporary Responses, 2nd ed. (Rowman & Littlefield, 2010) · Ch. 3-5 Foundationalism, coherentism, reliabilism
Quine, W. V. O. — Epistemology Naturalized · In Ontological Relativity and Other Essays (Columbia, 1969)
Goldman, A. I. — Discrimination and Perceptual Knowledge · J. Philos. 73 (1976) 771-791; reliabilism

Estimated time

beginner: 15m
intermediate: 30m
master: 45m