Faith and evidence
A Socratic walk-through of faith and evidence — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #How can a believer and a sceptic reach opposite conclusions from the same world?
Two careful people look at the same world. One finds in the fine-tuning of physical constants, the fact of consciousness, and a friend's unexpected recovery a reason to believe the universe has an author. The other looks at the same three items and finds a selection effect, an unsolved problem, and statistics. Neither is lying, and often neither is less intelligent or less honest than the other.
The tempting explanation is that one of them is not really looking at the evidence. That is too easy, and it is also wrong about how evidence works. The more interesting question is how the same datum can move two rational people in opposite directions — and this turns out to have nothing specifically to do with religion.
Reasoning it through
REASONING #Start with something uncontroversial. A doctor sees a patient with a fever and a cough. During a flu surge she reads it as flu; in a traveller returning from a malaria region she orders a blood film. The symptom is the same. What changed is what she thought likely before she saw it.
That is the whole mechanism. Evidence does not deliver conclusions; it multiplies. What a piece of evidence does to your confidence is a matter of how much more expected it is under one hypothesis than another — and where you end up depends on that ratio applied to where you started. Two people with the same likelihood assessments but different starting confidences must end at different places. That is not a failure of reasoning. It is what updating on evidence is.
Now push one step further, because the mechanism is stronger than the arithmetic suggests. The two observers often do not even agree on what the evidence is. A recovery no one predicted — is that datum "an answered prayer" or "an incompletely understood remission"? Describing it already commits you. And no single observation confronts a claim about God on its own: it meets a whole web of background commitments about what such a being would be like and what a world containing it should look like. That web absorbs the shock. The believer who learns of a child's suffering can adjust auxiliary beliefs about divine purpose rather than the core one; the sceptic who learns of a striking coincidence can adjust estimates of how many coincidences a large world produces.
So the disagreement rarely turns on a fact either side denies. It turns on priors, on the description of the datum, and on which part of the web takes the strain.
Given that, three positions are worth naming, because they are the honest ones and they disagree cleanly.
Evidentialism holds that a belief is justified only in proportion to the evidence for it, and that this standard applies to religious belief exactly as to any other. W. K. Clifford put it at its sharpest in 1877 — it is wrong, always and everywhere, to believe anything on insufficient evidence.
Reformed epistemology, associated with Alvin Plantinga, Nicholas Wolterstorff and William Alston, denies that the bar is the right one. It argues that belief in God can be properly basic: not inferred from other beliefs but formed directly in certain circumstances — much as your belief that other people have minds is not the conclusion of an argument and is none the worse for it. If belief arises in a person on encountering beauty or guilt or the night sky, the demand for an argument may misdescribe the case. The obvious objection — that this would license any belief whatever — Plantinga answers by tying warrant to a faculty functioning properly and aimed at truth, which is where critics say the dispute really lies.
The ambiguity thesis takes a third line: the evidence is genuinely balanced. John Hick argued that the world is religiously ambiguous, capable of being consistently interpreted either way, and that this is not a defect but a condition of free response. J. L. Schellenberg turns the same observation against theism — if a loving God existed, non-resistant non-belief should not be possible, and it plainly is.
Notice that none of these is settled by looking harder at the world, because each is a claim about what looking at the world can establish.
The analogy
THE ANALOGY #Two experienced editors read the same manuscript. One reads it as a flawed work by a serious writer and finds the rough passages purposeful; the other reads it as competent imitation and finds the same passages derivative. Neither has misread a sentence. Each brought a hypothesis about what kind of book this is, and the sentences were evidence given that hypothesis.
the editors can eventually settle their disagreement by learning who wrote it and what else the writer produced, whereas the disputants here have no agreed independent check available — which is precisely what makes the case hard rather than merely stubborn.
Clarifying the model
THE MODEL #Three clarifications matter here, and the third most of all.
First, "faith" is not one thing. Some traditions treat it as trust in something already taken to be well grounded — Aquinas held reason could establish a great deal before faith took over. Others treat it as commitment made precisely where evidence runs out, closer to Kierkegaard. A criticism aimed at one often misses the other entirely.
Second, priors are not arbitrary just because they differ. They encode a lifetime of experience and argument, and the honest question is not whether someone has them — everyone does — but whether they are responsive to anything.
Third, and to be explicit: nothing above is a critique of religious belief. The mechanism described is general epistemology, the same structure that produces durable disagreement between economists reading one recession or physicists reading one anomaly. Anyone hoping the analysis vindicates their side should notice that it applies to their own reasoning identically — and this walk-through deliberately does not adjudicate between the three positions, because that adjudication is exactly what is unresolved.
A picture of it
THE PICTURE #How to readread top to bottom as time. The first two solid arrows are the identical datum arriving at both parties — everything below is downstream of that one shared input. The notes beneath them are the background commitments each brings. The self-arrows are each party describing the datum, and the description has already diverged before any inference happens. The dashed arrows are the question each puts back to the world, and the spanning note records the crucial point: even with the same answer to that question, different starting confidences give different destinations.
What became clearer
WHAT CLEARED #Evidence is never processed neat. It is read against a background that decides what the datum even is and how surprising it would be, so identical input can rationally move two people apart rather than together. That is a general fact about inference under uncertainty, not a diagnosis of anyone's faith or doubt — and it explains why this disagreement has proved so durable, without telling you who is right.
Where to go next
ONWARD #- The epistemology of peer disagreement: what you should do, if anything, on learning that an equally informed person concluded the opposite.
- Whether pragmatic arguments — Pascal's wager, William James's case for believing ahead of the evidence — can justify a belief on grounds other than its truth.
Key terms
TERMS #| Term | What it means |
|---|---|
| Prior | the confidence you assign a hypothesis before taking a new piece of evidence into account. |
| Evidentialism | the view that belief is justified only in proportion to the evidence supporting it. |
| Properly basic belief | a belief justifiably held without being inferred from other beliefs, such as belief in other minds. |
| Religious ambiguity | Hick's thesis that the world admits of a consistent interpretation both with and without God. |
Every term the collection defines is gathered in the glossary.