Skip to content

Rationality

Version
v3 · 2026-09-28 · History
Prime #
1544
Domain group
Humanities
Origin domain
Philosophy
Aliases
Reason-responsiveness, Rational agency
Related primes
Bounded Rationality, Normativity, Decision, Evidence

Core Idea

Rationality is the structural quality of forming beliefs, choosing actions or governing procedures in a way that is answerable to reasons. A rational commitment is not simply one that succeeds or feels compelling. It connects available evidence, relevant aims, logical and practical constraints, and a norm for evaluation so that the commitment can be explained and revised. Philosophical accounts disagree over the final norm, but they converge on reason-responsiveness as the field's organizing problem.

The Prime has seven load-bearing roles: an agent or decision process; representations of the situation; a set of accessible reasons or evidence; aims or truth-directed standards; feasible commitments; coherence and consequence tests; and a revision relation triggered by new information.

How would you explain it like I'm…

Good-Reasons Thinking

Being rational means having good reasons for what you think and do. If you think it's raining because you see wet windows, and then you learn the sprinkler did it, you change your mind. Good thinkers listen to the reasons and change when the reasons change.

Thinking With Good Reasons

Rationality means making up your mind using good reasons. You look at the clues you have, think about what you are trying to do, and choose the belief or plan that fits them best. A rational person can explain why they believe something. Being rational does not mean never being wrong: it means that when you find a new clue, you are willing to change your mind. Getting lucky is not the same as being rational, and a choice that feels strong is not rational just because it feels strong.

Reason-Responsive Thinking

Rationality is the quality of forming beliefs and choosing actions in a way that answers to reasons—evidence, goals, and logical consistency. There are two main kinds: epistemic rationality (what to believe, and how strongly, based on evidence) and practical rationality (what to do, given your goals and options). Being rational is not the same as being successful; a lucky guess that works out is not rational, and a careful choice can still fail. It is judged by what the person could actually know at the time, not by facts they had no access to. A key part is revision: a rational thinker updates when the balance of evidence changes. People disagree about which standard is best—logic, probability, getting the best expected result—but all agree it is about responding to reasons.

 

Rationality is the property of beliefs, actions, or procedures that are answerable to reasons: they connect accessible evidence, relevant aims, logical and practical constraints, and an explicit evaluative norm so that the commitment can be justified and revised. Its structure has several load-bearing roles—an agent or decision process, a representation of the situation, an information set, aims or truth-directed standards, feasible options, coherence and consequence tests, and a revision relation triggered by new information. Drop the norm and you have mere behavior prediction; drop the information set and you judge agents by facts they could not access; drop revision and reasons become post-hoc rationalization. Epistemic rationality governs belief and credence; practical rationality governs intention and choice; they interact without reducing to each other. The standards are plural—deductive validity, probabilistic coherence, evidential support, expected utility maximization, satisficing, and public justifiability can each ground a judgment—so the standard in use must be declared. Bounded rationality describes how limited agents use tractable heuristics, qualifying rather than defining the concept. Finally, rationality is corrigible: it is marked less by never erring than by calibrated confidence and willingness to change when reasons shift.

Broad Use

Epistemic Belief Revision. The carrier contains an agent, evidence, competing propositions and confidence. A rationality assessment asks whether beliefs answer proportionally to evidence, coherence and calibrated uncertainty. The domain residual is truth-directed warrant, not mere confidence or social agreement. The reusable structure is literal: a reason-responsive agent or process holds representations, evaluates possible commitments against evidence and aims, chooses or revises, and remains answerable to counterevidence. What changes is the source of reasons and the consequence of error. A well-typed application states whose perspective and information set matter, which norm is used, and whether the judgment concerns a belief, an action, a procedure, or an institution.

Clarity

A clear rationality claim names the object of evaluation. Beliefs are assessed for evidence and coherence; actions for means-end fit and admissible constraints; procedures for how they gather reasons and correct error. Moving silently among these objects makes familiar disputes look deeper than they are.

The claim also declares the perspective and time. Ex ante rationality uses information available before the outcome.

Manages Complexity

Rationality compresses an open field of facts into a decision-relevant representation. It filters evidence by relevance, groups consequences by aim and constraint, and narrows alternatives without pretending that excluded detail never matters. The compression is accountable because its selection rule and residual uncertainty remain visible.

It separates object-level reasons from meta-level allocation. An analyst reasons about which bridge design is safest while also deciding how much analysis, testing and consultation the decision warrants.

Abstract Reasoning

  1. Type the commitment as belief, confidence, intention, act, policy, inference or procedure; different norms attach to each.
  2. Fix the temporal information set and distinguish inaccessible facts from ignored available evidence.
  3. List feasible alternatives, including delay, information gathering and reversible experiments when they are genuine options.
  4. State the evaluative standard and identify which aims or values it treats as inputs rather than conclusions.

Knowledge Transfer

The transfer skeleton is stable: an agent or process represents a situation, receives reasons, evaluates commitments under a norm, acts or believes at calibrated strength, observes consequences and revises. Transfer succeeds when each target domain supplies literal occupants for those roles. It fails when 'rational' is used only as praise, when one domain's metric replaces the target's norm, or when a successful outcome is treated as proof of a sound process.

Relationships to Other Abstractions

Current abstraction Rationality Prime

Parents (1) — outgoing typed relations

  • Rationality is a kind of Normativity Prime

    The accepted reference-grade review places Rationality under Normativity because the child instantiates or depends on the parent's broader structure while retaining its own constitutive identity.

Children (7) — abstractions linked here by typed relations

  • Argument to moderation Domain-specific is a kind of Rationality

    The fallacy violates evidence-responsive rational evaluation by substituting positional symmetry.

  • Lottery paradox Domain-specific is a kind of Rationality

    The paradox tests consistency conditions for rational acceptance.

  • Paradox of analysis Domain-specific is a kind of Rationality

    The candidate literally instantiates prime:rationality.

  • Principle of double effect Domain-specific is a kind of Rationality

    What makes it its own entry: the domain-specific identity determined by the harmful effect is foreseen but not intended as either end or means, the underlying act passes the governing moral constraint, and the proportionality condition is separately defended.

  • Rationalism Domain-specific is a kind of Rationality

    The doctrine elevates rational cognition as an epistemic source.

Hierarchy path (1) — routes to 1 parentless root