# Knowledge-First Epistemology ## Intro For most of the history of philosophy, people tried to explain knowledge by breaking it into parts. Knowledge, the old recipe said, is a belief that is true and that you have good reason for. In 1963 Edmund Gettier showed that the recipe leaks: you can have a true belief with good reasons and still only be right by luck. Knowledge-first epistemology turns the project around. Instead of building knowledge out of belief, truth and reasons, it treats knowledge as the basic thing and uses it to explain the others. The view comes from the Oxford philosopher Timothy Williamson. On his account, knowing is a state of mind that links you to the world, the way seeing or remembering does. You cannot be in it unless what you know is true. This matters in apologetics because the view shows up in real debates, often used by skilled atheists. A Christian who misreads it will lose the exchange. A Christian who understands it can ask the question it leaves open: what kind of world must this be, for minds like ours to be in states that reach the truth at all? ![Head-and-shoulders photograph of philosopher Timothy Williamson, smiling, in glasses and a white shirt against a stone wall](/codex/assets/timothy-williamson-portrait.jpg) *Timothy Williamson, 2014. Photo: A Williamson, [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/), via [Wikimedia Commons](https://commons.wikimedia.org/wiki/File:Professor_Timothy_Williamson.JPG), re-encoded.* ## In full **Knowledge-first epistemology** is the program, associated above all with [Timothy Williamson](/codex/timothy-williamson/)'s *Knowledge and Its Limits* (Oxford University Press, 2000), that reverses the traditional order of explanation in the theory of knowledge. Instead of analysing knowledge as justified true belief plus some anti-Gettier condition, it takes knowledge as explanatorily basic and uses it to illuminate belief, evidence, justification and assertion. Its central theses are: 1. **Knowledge is a mental state**, not a hybrid of an inner state (belief) and an outer condition (truth). 2. **Knowledge is the most general factive mental state.** A factive state is one you can be in only if its content is true: seeing that, remembering that, realising that. Each of these entails knowing. 3. **Knowledge is unanalysable** into more basic components. The failure of the post-Gettier industry to find a counterexample-proof analysis is, on this view, the expected result rather than an embarrassment. 4. **Evidence is knowledge (E=K).** Your evidence consists of exactly the propositions you know. 5. **No non-trivial condition is luminous** (the anti-luminosity argument). There is no interesting state, knowledge included, such that whenever you are in it you are always in a position to know that you are. 6. **Knowledge requires safety.** If you know, you could not easily have been wrong in a relevantly similar case. The program does not say knowledge is mysterious or beyond explanation. It says knowledge is explained by its role and connections, not by a recipe of ingredients. ## The theses in detail ### 1. Knowledge as a factive mental state The traditional picture treats knowledge as a belief (in the head) that happens to meet outside conditions (truth, a good connection to the facts). Williamson argues that this gets the joints of the mind wrong. Seeing that the door is open is a way of being related to the open door, and it is still a state of the seer. Knowledge is the general form of such states. Two consequences follow: - **Knowledge entails truth.** If you know that p, then p. This is factivity. - **Knowledge entails belief**, on Williamson's view, but belief is explained by reference to knowledge, not the other way round. Roughly, to believe p is to treat p as if you knew it. ### 2. Unanalysability Williamson argues by analogy with colour. Being red is not analysed as "being coloured plus some further condition", yet nobody concludes that redness is unreal. Likewise, knowledge entails belief and truth without being *composed* of belief and truth plus an extra ingredient. The Gettier cases are then not puzzles to be solved by a fourth condition. They are evidence that the JTB recipe was the wrong kind of project. See [Justified True Belief](/codex/justified-true-belief/) and [Edmund Gettier](/codex/edmund-gettier/). ### 3. E=K If your evidence is what you know, several things follow: - **Evidence is propositional.** It is not raw sensation but known facts, such as *that the road is wet*. - **Evidence is true.** Since knowledge is factive, nothing false can be part of your evidence. - **The deceived subject has less evidence.** A person in a Cartesian scenario who seems to see a hand does not have the same evidence as a person who sees one. The two cases feel alike from the inside, but feeling alike is not having the same evidence. This is Williamson's answer to the skeptic's assumption that the good case and the bad case share their evidence. See [Cartesian Skepticism](/codex/cartesian-skepticism/). ### 4. Anti-luminosity A condition is *luminous* if whenever you are in it, you are in a position to know you are in it. Williamson's argument runs through a slow change, such as feeling cold at dawn and gradually warming. At each moment you are almost exactly as you were a moment before. If knowledge requires a margin for error, there will be moments when you feel cold but are not in a position to know it, because you could easily have been wrong. So feeling cold is not luminous, and by the same reasoning nothing interesting is. The practical upshot: **you can know without knowing that you know.** Williamson rejects the "KK principle", the idea that knowing p entails being in a position to know that you know p. ### 5. Safety Knowledge requires that your belief could not easily have been false in similar cases. This explains why the lucky Gettier believer fails to know: in a nearby case, the same method would have produced a false belief. Safety is a modal condition, which is why knowledge-first epistemology is closely tied to Williamson's work on [Modal Logic](/codex/modal-logic/) and possibility. ### 6. The knowledge norm of assertion Williamson also argues that knowledge is the norm of assertion: one should assert p only if one knows p. This explains why "p, but I don't know that p" sounds absurd. ## What "conceptually fundamental" does and does not mean This is the point most often garbled in live debate. Saying that knowledge is **conceptually fundamental** means: - the *concept* of knowledge is not defined in terms of more basic epistemic concepts; - explanation runs from knowledge to belief and evidence, not the reverse. It does **not** mean: 1. **That the facts known are mind-dependent.** The opposite: factivity ties knowledge to truths that hold whether or not anyone knows them. Williamson is a realist. 2. **That knowledge exists apart from knowers.** Knowledge is a mental state, so every instance of it is someone's knowing. But nothing in the view says the knower must be human. 3. **That God is invoked.** Conceptual priority is a claim about the order of explanation among concepts. It is not a claim about what grounds the existence of minds, truths, or the fit between them. Three senses of "fundamental" need to be kept apart: | Sense | Question it answers | What knowledge-first claims | |---|---|---| | Conceptual | Can the concept be defined by simpler concepts? | No, knowledge is primitive | | Explanatory | What explains what within epistemology? | Knowledge explains evidence, belief, justification | | Ontological | What does the existence of knowers and truths depend on? | Nothing; the view is silent here | An argument that slides from the first row to the third, for example *"knowledge is conceptually fundamental, so it is mind-dependent, so it depends on God"*, is a non sequitur. A transcendental argument has to supply the third-row premise by separate argument. It cannot extract it from Williamson's vocabulary. ## Critics Knowledge-first epistemology is influential and contested. The main lines of criticism: - **Internalist evidentialists** such as Earl Conee and Richard Feldman (*Evidentialism*, 2004) hold that evidence is fixed by what is available to the subject's mind, so the deceived subject and the normal subject share their evidence. E=K, on this view, makes evidence inaccessible in exactly the cases where we need it. See [Evidentialism](/codex/evidentialism/). - **The new evil demon intuition.** Stewart Cohen (1984) pressed the intuition that a victim of a deceiving demon is just as *justified* as we are. Knowledge-first theorists answer that the victim is blameless but lacks knowledge and so lacks justification in the full sense. Critics find this too costly. - **Is evidence really knowledge?** Juan Comesaña and Holly Kantin ("Is Evidence Knowledge?", *Philosophy and Phenomenological Research*, 2010) argue that E=K cannot handle cases where a person reasons from a falsehood and still gains knowledge. - **The knowledge norm of assertion.** Jennifer Lackey ("Norms of Assertion", *Noûs*, 2007) argues that one can assert properly without knowing, for instance when passing on a well-supported belief one is required to share. - **Anti-luminosity.** Selim Berker ("Luminosity Regained", *Philosophers' Imprint*, 2008) and others challenge the margin-for-error premises of the argument. A full survey of the debate is Patrick Greenough and Duncan Pritchard (eds.), *Williamson on Knowledge* (Oxford University Press, 2009). ## What the view means for theistic and presuppositional arguments ### What it takes away 1. **The "define knowledge as JTB" trap no longer works.** A presuppositionalist who demands that an opponent state the necessary and sufficient conditions of knowledge, and then pounces on Gettier, will meet an opponent who denies that knowledge has such conditions. The demand has to be rephrased. 2. **"How do you know that you know?" is not a defeater.** Anti-luminosity rejects KK. The opponent can grant that he cannot always tell whether he knows and still hold that he knows. Pressing the question as if it ended the debate concedes that you have not understood the view. 3. **E=K blocks the easy skeptical move.** The Christian cannot simply say "your evidence is the same as a deceived person's, so you know nothing." Williamson denies the premise. ### What it leaves open 1. **The ontological question.** Knowledge-first epistemology tells you how knowledge, evidence and belief relate. It does not tell you why there are minds capable of factive states, or why human cognition is the kind of thing that can be safely connected to truths about electrons, numbers and other minds. That is the question pressed by the [Argument from Reason](/codex/argument-from-reason/), the [Argument from the Reliability of Reason](/codex/argument-from-the-reliability-of-reason/) and the [Transcendental Argument for God](/codex/transcendental-argument-for-god/). A Christian can ask it without misrepresenting Williamson. 2. **Safety needs a world with stable structure.** Safety is defined over nearby possible cases. A world in which our belief-forming methods are safe across a wide range of cases is a world with a deep fit between mind and nature. Theism explains that fit by design ([Genesis 1:27](/codex/genesis-1-27/); [Imago Dei](/codex/imago-dei/)). Naturalism must explain it by selection for survival, which is where Plantinga's evolutionary argument against naturalism applies. 3. **Parity cuts both ways.** If an atheist may know without being able to show that he knows, so may a believer. Knowledge-first externalism sits comfortably with [Reformed Epistemology](/codex/reformed-epistemology/): a person can know God through properly functioning faculties, or through the [Sensus Divinitatis](/codex/sensus-divinitatis/), without first producing an argument that he does. [Romans 1:19-20](/codex/romans-1-19-20/) describes exactly this kind of knowledge: "that which is known of God is manifest in them" ([Romans 1:19](/codex/romans-1-19/), ASV). ### A useful tension to probe In debate, Williamson's view is often paired with an announced commitment to fallibilism. The combination is coherent but needs explaining. On E=K, whatever you know is part of your evidence, so **your evidence entails what you know**. Several epistemologists therefore describe the view as a kind of *infallibilism about evidence*, while it remains fallibilist in the ordinary sense that a knower cannot always tell a case of knowledge from a case of error. A fair cross-examination question is: *"In which sense are you a fallibilist, given that on your own view your evidence entails what you know?"* See [Epistemic Fallibilism](/codex/epistemic-fallibilism/). ## Better lines for live debate - **Instead of** "So knowledge is mind-dependent, so you need God": *"I agree knowledge is primitive on Williamson's view. My question is not what knowledge is, but what explains the existence of minds that can be in factive states about a mind-independent world. Does your view answer that, or leave it open?"* - **Instead of** "How do you know that you know?": *"On your view you need not know that you know. Fine. Then you should grant that a Christian may know God without first proving that he does. Are you applying the same standard to both of us?"* - **Instead of** "Gettier refutes your epistemology": *"Gettier refutes the JTB analysis, which you don't hold. So let us talk about safety. What makes your methods safe, and what makes the world one in which they could be?"* ## Scholarship Published academic works this page draws on, each matched to a publisher, index or library catalog record. - Juan ComesañA and Holly Kantin, "Is Evidence Knowledge?" *Philosophy and Phenomenological Research* 80 (2010). [doi:10.1111/j.1933-1592.2010.00323.x](https://doi.org/10.1111/j.1933-1592.2010.00323.x) - Earl Conee and Richard Feldman, *Evidentialism* (Oxford University Press, 2004). [doi:10.1093/0199253722.001.0001](https://doi.org/10.1093/0199253722.001.0001) - Patrick Greenough and Duncan Pritchard, *Williamson on knowledge* (Oxford University Press, 2009). [Open Library](https://openlibrary.org/works/OL16920160W) - Jennifer Lackey, "Norms of Assertion," *Noûs* 41 (2007). [doi:10.1111/j.1468-0068.2007.00664.x](https://doi.org/10.1111/j.1468-0068.2007.00664.x) - Timothy Williamson, *Knowledge and Its Limits* (2000). [Open Library](https://openlibrary.org/works/OL2981213W) ## See also - [Timothy Williamson](/codex/timothy-williamson/), the view's architect - [Justified True Belief](/codex/justified-true-belief/), the analysis knowledge-first epistemology rejects - [Edmund Gettier](/codex/edmund-gettier/), whose counterexamples set the stage - [Epistemic Fallibilism](/codex/epistemic-fallibilism/), knowledge without infallible justification, and the tension with E=K - [Closure Principle](/codex/closure-principle/), which Williamson defends - [Cartesian Skeptical Argument and Christian Responses](/codex/cartesian-skeptical-argument-and-christian-responses/), where E=K offers a reply to the skeptic - [Foundationalism](/codex/foundationalism/), the structural view knowledge-first theorists often combine with it - [Evidentialism](/codex/evidentialism/), the internalist rival account of evidence - [Reformed Epistemology](/codex/reformed-epistemology/), the Christian externalism it sits closest to - [Presuppositionalism](/codex/presuppositionalism/), whose live-debate use of the view needs repair - [Argument from the Reliability of Reason](/codex/argument-from-the-reliability-of-reason/), the question knowledge-first epistemology leaves open - [Epistemology](/codex/epistemology/), the master hub
## Common questions this page answers **Q: What is knowledge-first epistemology?** It is the view, developed by Timothy Williamson in *Knowledge and Its Limits* (2000), that knowledge cannot be broken down into belief, truth and justification. Knowledge is treated as a basic mental state, the most general state that requires its content to be true, and it is used to explain belief, evidence and justification instead of being explained by them. **Q: What does E=K mean?** E=K is Williamson's thesis that a person's evidence is exactly what that person knows. It implies that evidence is always true and that a person deceived by a skeptical scenario has less evidence than a person seeing the world correctly, even if the two cases feel the same from the inside. **Q: Does knowledge-first epistemology solve the Gettier problem?** It changes the question rather than adding a fourth condition. Williamson argues that the search for a counterexample-proof analysis of knowledge failed because knowledge has no such analysis. Gettier cases still fail to be knowledge, and the safety condition helps explain why: the lucky believer could easily have been wrong. **Q: If knowledge is "conceptually fundamental", does that mean it depends on the human mind or on God?** No. Conceptual fundamentality means the concept of knowledge is not defined by simpler concepts. It says nothing about whether the facts known are mind-dependent (on Williamson's realist view they are not) and nothing about what grounds the existence of knowers. An argument for God has to make that further case on its own. **Q: Can a Christian accept knowledge-first epistemology?** Yes. Nothing in the view conflicts with Christian theism, and its externalism fits well with Reformed epistemology's claim that a person can know God without first proving that he knows. The Christian's apologetic question is the one the view leaves open: why minds like ours can be in states that reach the truth at all. **Q: What is the anti-luminosity argument?** It is Williamson's argument that no interesting mental condition is such that you can always know when you are in it. Using gradual changes, such as slowly warming from cold, he argues that a margin for error means there will be moments when you are in a condition without being in a position to know it. One result is that you can know something without knowing that you know it.