Abhinav Saxena
Branch of philosophy

Epistemology

Also called theory of knowledge, philosophy of knowledge, gnoseology, pramāṇa-śāstra

In brief

Epistemology is the branch of philosophy that studies knowledge and justified belief: what knowledge is, where it comes from, how far it reaches, and what makes a belief reasonable rather than merely true.

By · 4 min read

The big questions

  1. What is knowledge, and how does it differ from a true belief that happens to be right?
  2. What makes a belief justified?
  3. Can we know anything at all, or could the sceptic be right?
  4. Does all knowledge come from experience, or can some of it be had by reason alone?
  5. When can we know something because someone told us so?
  6. Is knowing a matter of what goes on in the mind, or of how the mind is connected to the world?

You believe a great many things, and you take yourself to know a good number of them: that your train leaves at nine, that the earth goes round the sun, that seven and five make twelve, that your friend is upset with you. Epistemology asks what entitles you to the word “know” in each case, and whether, if you pressed hard enough, you would be entitled to it at all.

The questions look idle until something goes wrong. A witness is sure and mistaken; a forecast is right by accident; a teacher passes on a falsehood that every textbook printed. Each failure shows a condition that knowledge was supposed to meet and did not, and epistemology is the attempt to say what those conditions are.

Knowledge, belief and luck

The oldest answer is Plato’s. In the Meno, a man who guesses the road to Larisa correctly is as good a guide as one who knows it, for as long as his guess holds; knowledge is worth more because it is tied down by the reason why. In the Theaetetus Socrates tests the proposal that knowledge is true opinion with an account, and the dialogue ends without accepting it. The tradition that followed settled, for the most part, on a version of that proposal: knowledge is justified true belief.

In 1963 Edmund Gettier published three pages showing that a belief can be justified and true and still not be knowledge, because it is true by luck (the Gettier cases). The decades since have produced many repairs. Reliabilists say the belief must come from a process that tends to produce truths; others require that it be safe (it could not easily have been false) or sensitive (had it been false, you would not have believed it); Timothy Williamson proposed in 2000 that knowledge is not built out of belief at all, but is the more basic notion. None of these has won general agreement, which is itself a finding: the word “know” marks something we recognise easily and define with great difficulty.

Where knowledge comes from

The early modern debate between rationalists and empiricists was a dispute about sources. Descartes held that the clearest knowledge comes from reason, and that the senses must be vindicated by it; Locke and Hume held that every idea comes from experience, and Hume drew the conclusion that our confidence in cause and effect, and in the future’s resembling the past, rests on custom rather than on reasoning. Kant replied that experience itself has a structure the mind contributes, which can be known in advance of any particular experience.

Indian philosophy made the question of sources its organising question. A pramāṇa is a means of knowing, and the schools are partly defined by the lists they accept: the Cārvākas perception alone; the Vaiśeṣikas and the Buddhist logicians perception and inference; the Sāṃkhya school those two and testimony; the Nyāya school four, adding comparison; the Bhāṭṭa Mīmāṃsakas and the Advaitins six. The debates between them about whether testimony (śabda) is a source of its own or a kind of inference anticipate the modern argument between Hume and Reid in close detail.

Scepticism

The sceptic does not need to prove that we know nothing; it is enough to ask how we know that we are not dreaming, or deceived by a demon, or, in the modern version, brains in a vat. Pyrrhonian sceptics in antiquity cultivated suspension of judgement as a way of life. Descartes raised the doubts in order to answer them, and most readers have thought his answer weaker than his doubts. Responses since then have ranged from G. E. Moore’s insistence that he knew he had hands better than he knew any premise of a sceptical argument, to theories on which the standard for “knowing” shifts with the context in which the word is used.

Knowing from others

Almost everything any of us believes about history, geography, medicine or the size of the sun we believe because someone told us. Hume held that our trust in testimony is justified only by our experience of its reliability; Reid held that trust in what others say is a basic disposition of the mind, needing no more defence than trust in our own eyes. The disagreement is still live, and it has acquired new stakes. Social epistemology studies trust, expertise and disagreement, and Miranda Fricker’s work on epistemic injustice shows how prejudice can deny people their standing as knowers.

My own research is on this part of the field: whether a source must stand in a particular relation to the truth of what it says for its hearers to come to know from it, and why on current evidence language models do not (can AI testify?).

How to begin

Read the Meno passage first; it is short and it states the problem the rest of the subject keeps returning to. Then the Theaetetus, then Descartes’ wax, then Hume and Reid on testimony side by side. The reading path below sets them in order, and each text in the library comes with the passages that matter and a note on why. For a survey of the whole field, the Stanford Encyclopedia’s entry is the place to go next.

Start here

A first reading path, in order. Tick what you have read; your progress stays in this browser.

7 readings

  1. Plato, Meno: the road to LarisaIn the library · Plato

    Why knowledge is worth more than a true opinion that happens to be right. A few pages, and the problem in its first form.

  2. Plato, Theaetetus: what is knowledge?In the library · Plato

    The first sustained attempt to define knowledge, and the reasons each definition fails.

  3. Descartes, the piece of waxIn the library · René Descartes

    What the mind grasps when everything the senses report has changed.

  4. Hume, Of MiraclesIn the library · David Hume

    How far to trust what others tell us, and why a wise man proportions his belief to the evidence.

  5. Reid, the principle of credulityIn the library · Thomas Reid

    Hume's great critic: trust in testimony is natural and original, as basic as trust in the senses.

  6. Clifford, The Ethics of BeliefIn the library · William Kingdon Clifford

    Is it wrong, always and everywhere, to believe on insufficient evidence?

  7. James, The Will to BelieveIn the library · William James

    The reply to Clifford: some questions must be answered before the evidence is in.

Texts in the library

Plato · 385 BCEPlato's Meno: the road to Larisa and the statues of DaedalusWhy a true opinion is as good a guide as knowledge, and yet worth less: it runs away unless it is tied down by the reason why.Plato · 369 BCEPlato's Theaetetus: what is knowledge?The dialogue that first asks what knowledge is, and ends by rejecting every answer it tries, including true opinion with an account.René Descartes · 1641Descartes on the piece of wax (Meditations II)A piece of wax changes every quality it has near the fire and is still the same wax; men in the street are seen only as hats and cloaks. What we know, Descartes argues, we know by judgement.David Hume · 1748Hume on testimony: 'Of Miracles' (Enquiry X)Hume's case that our trust in testimony is founded on experience of its conformity to the facts, and that a wise man proportions his belief to the evidence.Thomas Reid · 1764Reid on the principles of veracity and credulityReid's reply to Hume: we are made both to speak the truth and to believe what we are told, two principles that tally with each other.John Locke · 1690Locke on testimony and probability (Essay IV.xv–xvi)Locke on the two grounds of probability, our own experience and the testimony of others, and on why each further remove from the first witness weakens the proof.William Kingdon Clifford · 1877Clifford's 'The Ethics of Belief': the shipownerA shipowner sends an unsound ship to sea, sincerely believing it safe. Clifford argues he is guilty, because he had no right to the belief.William James · 1896William James, 'The Will to Believe': two laws, not oneJames answers Clifford: seeking truth and avoiding error are different duties, and a rule that only avoids error can cost us truths we needed.Plato · 375 BCEPlato's allegory of the cave (Republic VII)Prisoners take shadows for things and echoes for speech. The image this site opens with, in full.Plato · 399 BCEPlato's Apology: the oracle and the examined lifeSocrates tests the oracle by questioning those reputed wise, and finds that knowing the limits of one's knowledge is the only wisdom he has.Chuang Tzŭ (Zhuangzi) · 300 BCEZhuangzi: the happy fish (the bridge over the Hao)Two friends on a bridge argue about whether anyone can know what a fish enjoys, and the argument turns on where knowledge comes from.David Hume · 1748Hume on induction and custom (Enquiry IV–V)Why we expect the future to resemble the past, and why, Hume argues, no reasoning can show that it will: the expectation is the work of custom.John Stuart Mill · 1859Mill on knowing only your own side (On Liberty, ch. II)Mill's argument that a true opinion held without knowing its grounds is a dead dogma, not a living truth.John Godfrey Saxe · 1872Saxe, 'The Blind Men and the Elephant' (full poem)Six men touch six parts of an elephant and each is sure he knows the whole: Saxe's poem after the ancient Indian parable, in full.

Philosophers

1711–1776David HumeScottish philosopher, historian and essayist of the Enlightenment, the most thoroughgoing of the British empiricists, whose arguments about causation, induction, the self, testimony and morality still set the terms of debate.Uncertain. The Nyāya Sūtra is usually dated to about the 2nd century CE (c. 150–200), though estimates range from the 6th century BCE to the 3rd century CE.Akṣapāda GautamaTraditional author of the Nyāya Sūtra, the founding text of the Nyāya school of Indian logic and epistemology, which lists four means of knowledge and sixteen categories of inquiry and treats debate as a discipline.18 May 1872 – 2 February 1970Bertrand RussellBritish philosopher, logician and public intellectual, a founder of analytic philosophy, who found the paradox that bears his name, co-wrote Principia Mathematica, gave the theory of descriptions, and campaigned for pacifism and nuclear disarmament.1859–1938Edmund HusserlPhilosopher and mathematician, born in Moravia and working in Germany, who founded phenomenology, the study of conscious experience, and shaped Heidegger, Sartre, Merleau-Ponty and much twentieth-century thought.1873–1958G. E. MooreEnglish philosopher at Cambridge who, with Russell, broke from British idealism and helped found analytic philosophy; known for the open question argument against defining good, his defence of common sense, and Moore's paradox.12 March 1685 – 14 January 1753George BerkeleyAnglo-Irish philosopher and Anglican bishop who argued that there is no material substance, only minds and the ideas they perceive, so that for ordinary things to be is to be perceived.22 April 1724 – 12 February 1804Immanuel KantGerman philosopher of the Enlightenment who argued that the mind structures all experience, so that we know appearances and not things in themselves, and that morality is a law of reason binding every rational agent.Uncertain: estimates for the Mīmāṃsā Sūtra range from the 4th century BCE to the 1st century CEJaiminiAncient Indian sage, traditionally the author of the Mīmāṃsā Sūtra and founder of Mīmāṃsā, the school that holds dharma to be known only through Vedic injunction and builds a method for interpreting the Vedic texts.1632–1704John LockeEnglish philosopher and physician, a founder of British empiricism and of liberalism, who argued that all ideas come from experience, that government rests on consent and natural rights, and that the state may not coerce belief.Uncertain. The Vaiśeṣika Sūtra is dated between the 6th century BCE and the 1st or 2nd century CE; the Stanford Encyclopedia gives c. 1st century CE.KaṇādaTraditional author of the Vaiśeṣika Sūtra, the founding text of the Vaiśeṣika school, which sorts reality into six categories and explains the material world by eternal atoms, in one of the earliest atomist theories.26 April 1889 – 29 April 1951Ludwig WittgensteinAustrian-born philosopher of logic and language whose Tractatus held that propositions picture facts, and whose Philosophical Investigations, published after his death, ties meaning to use in language-games; one of the most influential philosophers of the twentieth century.1238–1317 (the dating most often given); some scholars place him c. 1199–1278MadhvaIndian monk and philosopher of the thirteenth century, founder of Dvaita Vedānta, who taught that God (Viṣṇu), souls and matter are eternally and really different, and that liberation comes only through God's grace.Traditionally 599–527 BCE (Śvetāmbara dating; the Digambara tradition puts his death in 510 BCE). Many historians place him later, with his death in the 5th century BCE, perhaps c. 477–425 BCE.MahāvīraIndian ascetic of the sixth or fifth century BCE, the twenty-fourth and last Tīrthaṅkara of the Jain tradition, whose teaching of non-violence, the soul bound by karma and the many-sidedness of reality shaped Jain philosophy.c. 150 – c. 250 CENāgārjunaIndian Buddhist monk and philosopher of about the second or third century CE, founder of the Madhyamaka school, who argued in the Mūlamadhyamakakārikā that nothing exists by its own nature, so that all things are empty and dependently arisen.c. 428–347 BCEPlatoAthenian philosopher, pupil of Socrates and teacher of Aristotle, who founded the Academy and wrote dialogues on knowledge, the soul, justice and the ideal city, built around his theory of Forms.Traditionally 1017–1137; recent scholarship c. 1077–1157RāmānujaSouth Indian teacher of Śrī Vaiṣṇavism and founder of Viśiṣṭādvaita Vedānta, who held that Brahman is a personal God, Viṣṇu, with souls and the world as his body, and that devotion, not undifferentiated knowledge, frees the self.1596–1650René DescartesFrench philosopher, mathematician and scientist who made methodical doubt the starting point of modern philosophy, argued that the thinking self is the first certainty and that mind and body are distinct, and founded analytic geometry.8th century CE (c. 788–820 in older scholarship; c. 700–750 in some recent work)ŚaṅkaraIndian monk and teacher, probably of the eighth century CE, whose commentaries on the Upaniṣads, Brahma Sūtras and Gītā gave Advaita Vedānta its classic form: the self is Brahman, and ignorance of this binds us.c. 470–399 BCESocratesAthenian philosopher who wrote nothing and is known through Plato and Xenophon; his questioning, his admission of ignorance and his care for the soul began Western ethics, and he was executed in 399 BCE.

Key terms

Knowledge
True belief held in a way that makes its truth no accident; what exactly secures that has been disputed since Gettier.
Justification
What makes a belief reasonable to hold: on internalist views, reasons available to the believer; on externalist views, facts such as the reliability of the process that formed it.
Justified true belief
The classical analysis of knowledge: S knows that p just in case p is true, S believes that p, and S is justified in believing that p.
The Gettier problem
The problem, set by Edmund Gettier’s 1963 counterexamples, that a belief can be justified and true without being knowledge, so the classical analysis fails.
Epistemic luck
The way a belief can be true by accident, in a manner incompatible with its being knowledge, even when it is well formed or justified.
Reliabilism
The view that a belief is justified, or counts as knowledge, when it is produced by a reliable process: one that tends to yield true beliefs.
Safety and sensitivity
Two modal anti-luck conditions on knowledge: a belief is sensitive if one would not hold it were it false, and safe if it could not easily have been false.
Testimony
Coming to know something because someone told you; the source of most of what any of us knows.
Trust
Relying on someone while accepting vulnerability to their failure, in a way that, unlike mere reliance, can be betrayed.
Understanding
Grasping why something is so, or what words mean: an achievement beyond true belief that shows in the ability to connect, apply and extend what one grasps.
Epistemic Injustice
A wrong done to someone specifically in their capacity as a knower, as when a speaker’s word is given less credibility than it deserves because of who she is.
Śabda (testimony in Indian philosophy)
In classical Indian epistemology, verbal testimony, counted by the Nyāya school as an independent pramāṇa, or means of knowledge, when it is the statement of a reliable speaker (āpta).
The Tracking Minimum
A source can transmit testimonial knowledge that p only if the truth of p, or the evidence for it, figures in why the source asserts p, through its own operation.
Explanation
An account of why something is the case or why something was done; in AI, an account of why a system produced a given output.
The endogeneity requirement
The clause of the Tracking Minimum requiring that a source’s connection to the truth hold through its own production process, not through a proxy on which that process depends.
Tracking reliability
Reliability as counterfactual sensitivity: a source is tracking-reliable when its outputs would follow the truth, or the evidence, were these to vary within the reach of its competence.
Frequency reliability
Reliability as a high rate of true outputs over a fixed distribution of cases, whatever explains that rate.
Instrument reading
Coming to know by competently reading a reliable device, where warrant is generated by the reader’s use rather than transmitted from a speaker.
Epistemic opacity
The condition of a process whose epistemically relevant elements an agent cannot all know, as with a complex simulation or a trained neural network.
Computational reliabilism
The view that reliance on opaque computational processes can be warranted by verification, validation, robustness analysis and a track record, without insight into the process.
AI testimony
The contested thesis that outputs of conversational AI systems can be testimony, transmitting knowledge to users as a competent human informant does.
Testimonial chain
A sequence of tellings, A to B to C, along which warrant is preserved so that the last hearer’s knowledge traces back to the first source.
The assurance view of testimony
The view that in telling someone that p a speaker assumes responsibility for its truth, giving the hearer a distinctive, non-evidential reason to believe it.
The statement view of testimony
Jennifer Lackey’s view that testimonial knowledge is acquired from what speakers state rather than what they believe, so one can learn from a statement its speaker disbelieves.
The knowledge norm of assertion
The thesis, defended by Timothy Williamson, that one must assert that p only if one knows that p, so that assertion is constitutively governed by knowledge.
Selfless assertion
Asserting what one does not believe because one recognises that the evidence supports it, as in Jennifer Lackey’s case of a creationist teacher who teaches evolution.
Automation bias
The tendency to accept what an automated system says or recommends in place of one’s own checking, so that its errors pass unnoticed and its silences are taken for all-clear.
Stochastic parrot
A critical label for large language models, coined in 2021, on which such a system recombines sequences of words according to their probabilities in its training text without reference to their meaning.
ELIZA effect
The tendency to attribute understanding, feeling or intention to a computer program because its output resembles human conversation.
Symbol grounding problem
The problem of how the symbols a system manipulates can mean something to the system itself, rather than only to the people who interpret them.
Epistemic dependence
The condition of holding a belief on the strength of someone else’s knowledge or evidence, which one does not have and often could not obtain oneself.
Epistemic vigilance
The set of capacities by which hearers guard against being misinformed, assessing both whether a source is competent and honest and whether what it says is believable.
Explainable AI (XAI)
The effort to make the outputs of machine-learning systems understandable to the people who use or are affected by them, either by building models whose workings can be followed or by attaching explanations to models whose workings cannot.
Echo chamber
A social structure in which outside voices are not merely absent but actively discredited, so that members come to distrust every source that does not belong to it.
Deepfake
An audio or video recording, made with machine-learning methods, that realistically depicts a person saying or doing something they did not say or do.
Anthropomorphism
The attribution of human characteristics, such as beliefs, intentions, feelings or understanding, to something that is not human, including animals, natural forces and machines.
Intentional stance
Daniel Dennett’s name for the strategy of predicting and explaining a system’s behaviour by treating it as a rational agent with beliefs and desires, whatever it is made of.
Cognitive offloading
The use of physical action or external tools, such as writing a note, setting a reminder or asking a device, to reduce the mental work a task would otherwise require.

Thought experiments

Edmund GettierThe Gettier casesIs knowledge the same as justified true belief, or does knowing need something more?PlatoThe allegory of the caveWhat is the difference between living among appearances and knowing what is real, and what can education do about it?Hilary PutnamThe brain in a vatHow can you know you are not a brain in a vat, and if you cannot, can you know anything about the world outside your mind?Nick BostromThe simulation argumentIf advanced civilisations would run many simulations of their ancestors, should you believe you are probably living in one?

Read free elsewhere

Full texts, free online

Encyclopedias and bibliographies

On this site

Research by Abhinav SaxenaEpistemologyQuestionDo large language models know what they say?QuestionCan AI testify, or is a language model's answer something else?QuestionShould you trust what ChatGPT tells you?QuestionWhat is the Tracking Minimum?EssayYour Chatbot Is Not a WitnessEssayWhat Nyāya Can Teach a Language ModelTry it in the labDoes the Machine Know?Try it in the labThe Epistemic Chain

Questions people ask

What is the difference between epistemology and metaphysics?

Metaphysics asks what there is and what it is like: whether minds are physical, whether we have free will, what causation is. Epistemology asks what we can know about any of it, and how. The two meet constantly: a theory of what the world is like has to leave room for our knowing it.

What are the sources of knowledge?

The usual list is perception, memory, reasoning, introspection and testimony. Indian philosophy made the list its central question, as the theory of pramāṇas: the Cārvākas accepted perception alone, the Nyāya school four sources (perception, inference, comparison and testimony), and the Bhāṭṭa Mīmāṃsakas and Advaitins six.

Is knowledge the same as justified true belief?

For a long time many philosophers assumed so. In 1963 Edmund Gettier described cases in which a person has a justified true belief that is true only by luck, and so seems not to know. No repair of the definition has won general agreement since, and some philosophers now treat knowledge as basic rather than as belief plus extra conditions.

Why does epistemology matter today?

Most of what anyone believes comes from sources they cannot check for themselves: experts, the news, search engines and now AI systems. Whether those sources can give us knowledge, and what we owe to the people we tell things, are epistemological questions with practical consequences.