⁂ Concepts · Knowledge · No. 39
Belief

Disregard the evidence: an instruction to perform an action that is not available.

Belief

Taking something to be the case - the basic attitude that can be true or false, and the one you cannot adopt at will.

Theme
Knowledge, truth and reason
Tradition
Cross-tradition
Term
δόξα · doxa
Term fixed
c. 380 BC

In plain terms

Try to believe, right now, that it is snowing indoors. You can say the words, picture it, and pretend. You cannot do it.

That is the strangest thing about belief and the most revealing. Actions are up to you and belief is not, which means it is not something you do. It is more like something that happens to you when evidence arrives - and it is why offering someone money to believe something never works, while offering them evidence sometimes does.

The complications follow immediately. People act against what they say they believe. People hold beliefs they would deny having. And people go on reacting to what they know is false, which suggests that belief is not one thing but several, imperfectly connected.

As a philosopher would put it

On the standard account belief is a propositional attitude with a mind-to-world direction of fit: it aims at truth and is answerable to evidence, unlike desire, which aims to change the world. Dispositional accounts identify believing with a pattern of dispositions to assert, infer and act, which handles the fact that most beliefs are never entertained.

Doxastic involuntarism is the standard position, and it constrains the ethics of belief. If believing is not an action, then telling someone they ought to believe something can only be an indirect instruction about evidence and attention. Clifford's principle - that it is wrong always to believe on insufficient evidence - and James's reply about cases where belief must run ahead of proof are the two poles of that argument.

The alternatives are live. Bayesian accounts replace belief with degrees of credence, which fits reasoning under uncertainty and fits ordinary talk badly. And eliminativists argue that belief belongs to a folk theory that will not survive neuroscience - a claim Churchland presses and almost everyone else resists, on the grounds that no vocabulary predicts human behaviour half as well.

Where you meet it

Judges routinely instruct juries to disregard evidence - an inadmissible confession, a prior conviction, testimony struck from the record. The instruction is a cornerstone of trial procedure, and appellate courts treat it as an adequate remedy.

It does not work. Experimental studies of mock juries have found repeatedly that instructed jurors are influenced by the excluded material, and in some designs more influenced than jurors who never heard it - an effect attributed to the instruction drawing attention to it and to reactance against being told what to think. The finding has held across decades and jurisdictions.

The legal system is asking people to perform an action that is not available: to stop believing something on command. Judges are not naive about this - the instruction persists because no better remedy exists short of discharging the jury. What the practice demonstrates, at scale and with consequences, is that belief does not answer to instruction, only to evidence.

What is actually disputed

Whether belief comes in kinds or degrees.

The traditional picture has a single attitude, held or not held. The Bayesian picture has credences from zero to one, which handles evidence and uncertainty far better and makes flat statements of belief an approximation. Attempts to reconcile them founder on the lottery paradox: a rational person can assign high credence to each ticket losing while believing that some ticket will win, so belief cannot simply be credence above a threshold.

Separately, the fragmentation problem is unresolved. People sincerely assert one thing and act on another - and whether that shows two beliefs, a belief plus something that is not a belief, or that the category is too coarse, is exactly what the implicit-attitude literature has spent twenty years failing to settle.

Commonly confused with

  • Faith c. 1270 Faith on most religious accounts involves commitment and trust beyond what evidence settles. Belief is the bare attitude, and one can have it with no commitment at all.
  • Knowledge c. 369 BC Knowledge requires truth and grounds. Belief requires neither, which is why the vast majority of beliefs anyone has held have been false.

Defined against

  • Doubt Peirce's picture makes the pair a cycle rather than an opposition: doubt is the irritation that enquiry works to settle back into belief

Philosophers

  • Plato Belief as opinion, short of knowledge
  • David Hume Belief as a manner of conceiving, not an additional idea
  • Charles Sanders Peirce Belief as a settled habit of action, and doubt as its irritation
  • William James The cases in which belief may legitimately run ahead of evidence
  • Ludwig Wittgenstein Believing as shown in what one does, not in an inner act

Works

Schools

Related concepts

Knowledge c. 369 BC · Truth c. 330 BC · Justification 1641 · Doubt 1641 · Testimony 1764

Plato's doxa is opinion as distinct from knowledge; the modern treatment as a propositional attitude with a direction of fit belongs to the twentieth century.