Lesson 3 · Evidence, Knowledge and Intellectual Responsibility

How do you know what you think you know?

Estimated study time: 45–55 minutes · Stage I: Foundations of Reasoning

Before we begin

You can be utterly sincere, profoundly convinced and mistaken. This observation applies to believers and atheists alike. If an argument is intended to persuade someone who doesn't share your starting assumptions, conviction cannot substitute for publicly assessable reasons.

Epistemology is the study of knowledge, justification and rational belief. We will not settle every debate about the nature of knowledge in an introductory lesson. Instead, we'll learn questions that make disagreements more productive: What supports this claim? How trustworthy is the evidence? What alternatives are possible? What would change our minds?

1. Belief, truth and justification are different

A belief is an attitude of accepting a proposition. Truth concerns whether the proposition is actually the case. Justification concerns the grounds someone has for accepting it.

Someone may believe a true proposition for a dreadful reason. For example, correctly guessing the result of a coin toss doesn't ordinarily constitute knowledge. Conversely, a person may have strong evidence for a claim later overturned by new information. Epistemic responsibility isn't omniscience; it's proportioning confidence to available reasons and being prepared to revise.

The famous "justified true belief" analysis of knowledge is historically influential but faces important objections, including Gettier cases where a belief is true and apparently justified partly by luck. This is a reminder that philosophy has complications beyond slogans. Our practical focus is whether reasons genuinely support conclusions.

2. Different kinds of evidence answer different questions

Empirical evidence arises from observation, measurement and experiment. Historical evidence includes documents, artefacts, inscriptions and testimony interpreted in context. Logical and mathematical arguments assess relations among propositions; they can establish what follows from assumptions, but cannot alone establish every empirical premise. Personal experience may be powerful evidence for the person having it, though its interpretation can be disputed.

Ask whether the evidence is suitable for the claim. A laboratory result may help establish the age of an object. It doesn't by itself tell us the moral significance of the object. A philosophical argument may evaluate whether a concept is coherent. It doesn't automatically demonstrate that something matching the concept exists.

"Archaeology confirms a town mentioned in the Bible, so every event narrated in the Bible must be true."

The existence of the town is relevant to some historical claims. It doesn't follow that every adjacent narrative claim has been independently verified, especially a supernatural claim. Evidence has a scope; don't stretch it beyond that scope.

3. Evidence should be assessed comparatively

Suppose a person reports an extraordinary experience during a medical emergency. One explanation is an encounter with a supernatural reality. Another involves altered cognition under unusual physiological and psychological conditions. Others may involve memory reconstruction or a combination of causes.

The experience itself shouldn't be dismissed simply because it seems remarkable. But establishing that something was experienced is different from establishing which explanation caused it.

Good reasoning asks what each hypothesis predicts, how well it fits the evidence and whether it requires unsupported additional assumptions. It also asks whether the set of proposed alternatives is sufficiently broad. A hypothesis doesn't win merely because its advocate has dismissed one competitor.

4. Testimony: neither worthless nor infallible

Nearly everything we know about distant events and specialised subjects depends to some extent on testimony. Credible testimony is evidence. The challenge is evaluating its reliability.

Consider the witness's opportunity to observe, possible errors of perception or memory, independence from other witnesses, consistency over time, interests and incentives, documentary provenance, and independent corroboration.

Ten people repeating the same online account may provide only one underlying source. By contrast, several independently obtained lines of evidence can be more powerful than a thousand copies of one anecdote.

The most useful question isn't simply "How many people say it?" but "How many genuinely independent reasons are there to accept it?"

5. Probability, prior expectations and new evidence

Probability offers a way of expressing uncertainty rather than pretending knowledge comes in only two states: absolute certainty and complete ignorance.

The general Bayesian idea is that new evidence updates a prior assessment:

Posterior odds = Prior odds × Likelihood ratio.

The likelihood ratio compares how probable the evidence would be under one hypothesis against how probable it would be under a competing hypothesis.

Imagine a medical test. A positive result is more likely in someone with a disease than in someone without it. Yet if the disease is exceptionally rare, many positive results may still be false positives. To interpret the result, we must consider both the test's accuracy and the original prevalence.

The same general discipline is helpful when discussing miraculous claims. How probable would the testimony be if the event occurred? How probable might similar testimony be without the event? How should background knowledge bear on the comparison? Philosophers disagree about answers, but questions of this form are clearer than "I find it convincing, so it's proved".

Caution: writing a Bayesian formula doesn't supply numerical values. If an apologist claims an exact probability for God, or an atheist claims an exact probability against God, ask where those estimates came from. Made-up numbers don't become evidence by appearing in an equation.

6. Burden of proof and degrees of commitment

The speaker who asserts a proposition should be prepared to defend it. But complex debates often involve multiple claims, and both participants can carry burdens for the positive claims they make.

Compare:

• "I am not convinced God exists." This reports a position of withheld assent.
• "No gods exist." This asserts a broader proposition for which reasons may reasonably be requested.
• "The Christian God exists." This is another substantive claim requiring reasons.

None of those statements automatically settles the truth of the others. Criticise the actual claim being made, not a stronger position imposed upon the speaker.

There are contexts where a presumption or default rule is appropriate, but such rules are methodological choices and must be justified. "The burden is always yours" is not a universal shortcut around substantive dialogue.

7. Falsifiability, testing and the limits of science

In empirical inquiry, a hypothesis is valuable partly when it yields predictions that expose it to possible failure. Falsifiability is an important principle associated with Karl Popper, but it isn't a complete, universally accepted definition of all scientific reasoning, nor a general test of whether any proposition in philosophy is meaningful.

Not every historical claim can be rerun as a laboratory experiment, and not every metaphysical proposition is an experimental hypothesis. This doesn't make all standards of assessment disappear. Historical and philosophical disciplines have their own methods of evaluating evidence and reasons.

"Science can't disprove my interpretation of a miracle, therefore it happened."

That doesn't work. Limited scientific testability doesn't supply positive historical evidence. Conversely, the limits of science don't entail that philosophical or theological claims are automatically false.

8. Independent corroboration and source criticism

When evaluating a source, ask:

1. Who produced it, and when?
2. What did the author personally observe, and what was reported second-hand?
3. How has the text or testimony been transmitted?
4. What independent sources support or contradict it?
5. What explanation would best account for both confirming and disconfirming evidence?

For biblical studies, these questions apply to the Gospels and Paul's letters, just as analogous questions apply to ancient secular historians. Religious texts don't automatically become worthless because they're religious, nor do they become independently verified because they're sacred to a community.

9. Cognitive biases and motivated reasoning

Confirmation bias encourages us to notice evidence that supports existing views and discount contrary material. Motivated reasoning can shape how rigorously we scrutinise competing claims. Selection bias appears when we examine only the cases that made it into a preferred collection.

Ask yourself whether you'd accept the same form of argument from a religious tradition you reject.

If a Christian testimony is evidence for Christianity, how would you assess a comparably detailed Muslim or Hindu testimony? You needn't conclude that all traditions are equally likely. You do need criteria that don't silently change depending on which tradition is under discussion.

An atheist has an analogous obligation: if a nonreligious witness provides evidence you find credible, you must explain why otherwise comparable religious testimony is weaker, rather than dismissing it solely because of its supernatural interpretation.

10. When to say "I don't know"

An intellectually responsible conclusion is often narrower than the one we would prefer.

"The evidence doesn't currently establish this interpretation" may be defensible even when "this interpretation is impossible" is not. "I have a philosophical reason to favour theism" differs from "I have conclusively demonstrated the Trinity". "A naturalistic explanation is plausible" differs from "we have proved there was no supernatural event".

A willingness to suspend judgement isn't weakness. It's an essential part of inquiry. If the evidence doesn't yet settle the question, the responsible response is to identify the uncertainty and explain what evidence might reduce it.

Worked argument clinic

Original post: "Thousands of people have reported near-death experiences. Those reports prove consciousness survives death, which proves Christianity."

First question: what is documented? Reports of remarkable experiences around life-threatening events exist. They deserve careful investigation.

Second question: what follows? The existence of reports doesn't, on its own, settle when the experiences were formed, what physiological states accompanied them or whether consciousness continued independently of brain function. Those are empirical and philosophical questions requiring additional evidence.

Third question: which theology? Even if consciousness were shown to survive bodily death, that wouldn't by itself identify Christian doctrine as the correct account. Many traditions posit survival in various forms.

Improved argument: Select well-documented cases, scrutinise timing and corroboration, compare competing hypotheses, consider cross-cultural differences, and then explicitly separate evidence of unusual experience from the larger theological inference.

Practice exercises

1. A hundred websites repeat the same unsupported anecdote. How many independent evidential sources may that amount to?
2. What two quantities, in addition to the result itself, matter when interpreting a diagnostic test for a rare condition?
3. Does an argument for the existence of a supernatural cause automatically establish that the cause is the God of Christianity? What additional steps would be needed?
4. Distinguish the sentences "I don't believe that P" and "I believe that not-P".
5. Name one piece of evidence that would make you revise a currently held belief. Explain why it would matter.

Lesson summary

• Sincerity, truth and justified belief aren't interchangeable.
• Evidence must be suited to the claim and not stretched beyond its actual scope.
• Independent corroboration matters more than sheer repetition.
• Probability requires both background assumptions and the evidential fit of competing hypotheses.
• Scientific, historical and philosophical questions require appropriate methods.
• A clear admission of uncertainty is better than false certainty.

Further reading: Stanford Encyclopedia of Philosophy, Epistemology, Stanford Encyclopedia of Philosophy, Bayesian Epistemology and Stanford Encyclopedia of Philosophy, Testimony.