What counts as knowledge—how do we know that we know?

We use “know” constantly—I know the way, she knows the answer, we know smoking causes cancer. But what turns a confident belief into knowledge? Philosophers call this the epistemic question. Below is a compact tour of the main ideas and a practical take-away on how to tell when you really know.

The classic starting point: Justified True Belief (JTB)

A natural thought is that to know a claim p you must:

1. Believe p,
2. p is true, and
3. You are justified in believing p.

This “JTB” picture fits much of ordinary life. If you believe the train leaves at 8:05, it does leave at 8:05, and you checked the timetable, you seem to know.

The Gettier shock

In 1963, Edmund Gettier described cases where someone has a justified true belief by luck. Example: you look at a reliable clock; unknown to you it stopped yesterday at 8:05, but right now it is 8:05. Your belief is true and justified, yet it feels wrong to call it knowledge. Moral: knowledge is not just J+T+B—it must also be non-accidental.

What to add beyond JTB? (Major proposals)

• No-luck requirements
  • Sensitivity (Nozick): If p were false, you wouldn’t believe p. (Good against some luck, bad for necessary truths and induction.)
  • Safety (Sosa): In nearby situations where you form beliefs the same way, you wouldn’t easily be wrong. (Explains why “fake-barn county” fails: your method is too risk-prone.)
• Reliabilism (Goldman)
  Beliefs count as knowledge when produced by a reliable process (vision in good light, calibrated instruments, good inference). Internal reflection isn’t required; what matters is objective reliability.
• Defeasibility
  You know only if there are no undefeated counter-evidence facts lurking. If there’s a police bulletin saying all the town clocks are broken, that would defeat your justification.
• Virtue epistemology (Sosa, Greco, Zagzebski)
  Knowledge is the apt success of an intellectual virtue—true belief because of cognitive skill (careful observation, fair-minded reasoning), not because of luck.
• Knowledge-first (Williamson)
  Take knowledge as primitive: it’s a mental state that can’t be reduced to belief + extras. Other notions (evidence, justification) are explained in terms of knowledge, not vice versa.

Where knowledge comes from (sources)

• Perception: sight, touch, hearing—when conditions are normal.
• Introspection: awareness of your own thoughts, pains, intentions.
• Memory: carries earlier knowledge forward, barring defeaters.
• Testimony: most of what we “know” (history, science news) comes from others; trust depends on their reliability and our checks.
• Inference: deductive, inductive, abductive (inference to the best explanation).

How justification is structured

• Foundationalism: some beliefs are basic (e.g., simple perceptual reports) and support the rest.
• Coherentism: justification from mutual support within a web of beliefs.
• Infinitism: reasons can, in principle, extend without end; what matters is the ability to keep giving further reasons.

Internalism vs. Externalism:

• Internalists: justification must be reflectively accessible to you.
• Externalists: what matters is whether your belief hooks up to the world in the right way, even if you can’t show that from the armchair.

The skeptic’s challenge

What about dreaming, brain-in-a-vat, or perfect deepfakes? Skeptics argue our evidence underdetermines the truth. Replies include:

• Moorean shift: I know I have hands; therefore radical skeptical scenarios are false.
• Contextualism: standards for “know” shift with context—everyday claims can be known though they’d fail under hyper-demanding standards.
• Pragmatic encroachment: when the stakes rise (Should I wire all my savings?), the bar for “know” rises too.

Knowledge vs. understanding vs. know-how

• Knowing-that (propositional) differs from knowing-how (practical skill). You might “know-that” how to swim works but not know-how to swim.
• Understanding emphasizes grasping patterns and explanations (how pieces fit), not just isolated truths. Science typically aims at both knowledge and understanding.

The social dimension

Knowledge is often collaborative:

• Peer disagreement: equally informed peers who disagree give each other defeaters that call for suspension or further inquiry.
• Epistemic injustice (Fricker): unfairly deflating someone’s credibility can block knowledge flow.
• Institutions (journals, courts, standards bodies) act as error-correcting systems, raising reliability.

Science and probability

Science rarely yields certainty; it offers models with measured error. A Bayesian gloss: rational agents update credences with evidence; at some threshold (context-dependent) high credence + reliability + absence of defeaters = action-worthy knowledge.

So… how do you know that you know?

Use this five-part check:

1. Truth-tracking
   • Would your method still get it right in nearby cases? (Safety)
   • If the claim were false, would you likely not believe it? (Sensitivity, where appropriate)
2. Method reliability
   • Is your belief formed by a process that’s generally reliable here (good conditions, calibrated tools, competent testimony)?
3. Defeater sweep
   • Have you checked for live counter-evidence or reasons to doubt (conflicts of interest, known illusions, data anomalies)?
4. Cross-method convergence
   • Do independent lines (perception + testimony + calculation) point the same way?
5. Stakes and context
   • Are the stakes low enough for your current standard? If the stakes are high, raise your bar: seek replication, direct evidence, or expert consensus.

If a belief passes these, you have more than a lucky hunch; you have something that—by many leading theories—counts as knowledge. If it fails, you may still have a reasonable belief, but intellectual humility (and sometimes further checking) is the wise move.

Bottom line

Knowledge is not mere true belief; it’s true belief through the right kind of connection to the world—robust against luck, formed by reliable methods, and standing firm against defeaters. We “know that we know” when our beliefs survive this pressure-test across methods, contexts, and counterfactuals. That’s the difference between being right this time and being the kind of knower who tends to be right for good reasons.