Skip to content

Preference

Prime #
None
Origin domain
Decision Theory Economics
Also from
Psychology & Behavioral Sciences, Philosophy, Political Science
Aliases
Ordering, Ordered Valuation

Core Idea

Preference is an ordering or comparative evaluation over possible outcomes, actions, states, or bundles, from the standpoint of an evaluator (agent, system, criterion, or model). It specifies which alternatives are favored, disfavored, indifferent, or incomparable. The abstraction is substrate-neutral: utility functions, rankings, revealed choices, policy priorities, qualitative value orderings, and ML reward signals are all implementations. The core commitment is the ordering relation — the evaluator carries some disposition over alternatives that, when consulted, picks out which are better.

How would you explain it like I'm…

Liking One More

If someone offers you chocolate or vanilla ice cream, you probably want one more than the other — that wanting-more is your preference. It just means you have a way of ranking the choices: this one first, that one second, maybe these two are tied.

Ranking Your Choices

A preference is how someone (or something) ranks their choices. If you'd rather have pizza than salad, and salad than cereal, that's a preference ordering — pizza is first, salad second, cereal last. Preferences don't have to come with numbers; they just say what beats what. Almost every decision rule, from picking lunch to a computer choosing a move in a game, starts from some kind of preference over the options.

Ordering over alternatives

A preference is an agent's ordering over a choice set on some evaluative dimension — a disposition that, when consulted, says which options are favored, disfavored, equally good, or unable to be compared. Four things travel with the prime: a choice set (what is being ordered), an evaluator (whose preference it is — a person, a criterion, an algorithm), an ordering relation over the set (which can be complete or partial, consistent or messy), and a context where the ordering does work (guiding selection, predicting behavior, supplying an objective for optimization). Preferences can be expressed as utility numbers, ranked lists, observed choices, or learned reward signals — these are all implementations of the same abstract ordering.

 

Preference is an agent's ordering over a choice set on some evaluative dimension — a disposition that, when consulted, says which alternatives are favored, disfavored, indifferent, or incomparable. Four roles travel with the prime: a choice set (the alternatives being ordered), an evaluator (the agent, criterion, system, or model whose preference this is), an ordering relation over the set (which may be complete or partial, transitive or inconsistent, strict or weak), and a context in which the ordering does work — guiding selection, predicting behavior, or supplying the objective that optimization will then maximize. The abstraction is substrate-neutral: utility functions, rankings, revealed choices (preferences inferred from observed decisions), policy priorities, qualitative value orderings, and learned reward signals are all implementations of the same ordering relation. Preference is not the act of selecting (that is decision), not the assumption that comparisons share a common metric (that is value commensuration), not the search for a best feasible option (that is optimization, which presupposes preference), and not the case of multiple evaluators disagreeing (that is preference heterogeneity and conflict, a child case).

Broad Use

  • Economics / decision theory: utility functions, preference orderings, indifference curves, revealed preference, expected utility.
  • Behavioral / cognitive psychology: choice under risk and uncertainty (loss aversion, prospect theory), time preference and discounting.
  • Political theory / public choice: voter preferences, Arrow's theorem on aggregating heterogeneous preferences, social welfare orderings.
  • Machine learning: reward models, preference learning from pairwise comparisons (RLHF and friends).
  • Biology / behavioral ecology: revealed preference in animal foraging and mate choice.
  • Aesthetics / criticism: ranked judgments over artworks, performances, or styles from a critic's or audience's standpoint.

Clarity

Preference names the ordering relation itself, distinct from the cluster of concepts that get confused with it. It is not the act of selecting (that is Decision); it is not the condition that comparisons share a common metric (that is Value Commensuration); it is not the search for a best feasible option (that is Optimization); and it is not the case of multiple evaluators disagreeing (that is Preference Heterogeneity and Conflict). What Preference adds is the bare commitment of comparability under an ordering from an evaluator's standpoint — a disposition that, when consulted, picks out which alternative is favored. That disposition can be revealed or stated, complete or partial, transitive or inconsistent, but it is always somebody's or something's ordering over a choice set. Naming this prime separates "we ranked it" from "we chose it," "we measured it on a scale," and "we found the optimum."

Manages Complexity

Preference decomposes a valuation situation into four named roles that turn an opaque "what's better here?" into a structured problem. There is a choice set (the alternatives being ordered); an evaluator (the agent, criterion, system, or model whose preference this is); an ordering relation over the choice set (which may be complete or partial, transitive or inconsistent, strict or weak); and a context in which the ordering guides selection, prediction, or explanation. Once those four roles are visible, downstream phenomena fall out by topology. Multiple evaluators with different orderings becomes preference heterogeneity; an ordering that flips with framing or time becomes temporal inconsistency or preference reversal; an ordering inferred from behavior rather than reports becomes revealed preference; an ordering aggregated across many evaluators raises the social-choice problem (Arrow). The same role-vocabulary that handles a consumer choosing between bundles handles a voter ranking candidates, an RLHF labeler comparing model outputs, and a foraging animal choosing patches. The analyst can locate the live question — is the choice set ill-defined? is the evaluator ambiguous? is the ordering partial or intransitive? is the context shifting? — rather than re-deriving the structure for each domain.

Abstract Reasoning

Preference supports a family of substrate-neutral operations on the ordering once it is named. The analyst can take transitive closure (if X ≻ Y and Y ≻ Z, infer X ≻ Z under suitable conditions) to extend a partial ordering; they can check consistency (transitivity, completeness, independence of irrelevant alternatives) and use violations as diagnostic evidence that the evaluator's underlying disposition is structured differently than assumed (e.g., framing effects, reference-dependence, intransitive cycles). They can perform aggregation across multiple evaluators (voting rules, social welfare functions) and confront the impossibility results that come with it. They can run counterfactuals: "if the choice set were enlarged, would the favored option remain favored?" "if the framing were inverted, would the ordering reverse?" These operations let preference supply the objective that downstream procedures like optimization, decision, and choice consume. Subordinate primes inherit the operations: time_preference, risk_aversion, and loss_aversion are all specifications of which ordering an evaluator has over particular structured choice sets (temporal, probabilistic, gain-vs-loss-framed) — they presuppose the bare preference relation and add curvature or asymmetry on top.

Knowledge Transfer

The vocabulary travels intact across substrates. An economist studying utility functions, a behavioral psychologist studying prospect theory, a political theorist working on Arrow's impossibility, an ML researcher training a reward model from pairwise comparisons, and an ecologist documenting revealed foraging preference in pollinators are all working with the same four-role structure: choice set, evaluator, ordering, context. The transfer is structural, not metaphorical. The animal case is especially load-bearing for substrate independence — a bumblebee's revealed preference between flower types has no utility function written down anywhere, no stated valuation, no language at all, and yet the same ordering-from-an-evaluator's-standpoint pattern fits cleanly. That rules out the suspicion that preference is a specialty of economics or human cognition; it is a structural relation that any system capable of differential selection can instantiate. Reward modeling in ML pulls the same trick from the other direction — the "evaluator" is a learned scoring function rather than an agent, but the ordering over candidate outputs still does the work.

Example

Consider an RLHF training run that consumes a labeler's pairwise comparisons over model completions. The choice set is the pool of generated completions on a given prompt; the evaluator is the human labeler (or a population of labelers, aggregated); the ordering relation is the labeler's revealed pairwise judgments ("completion A is better than completion B"); and the context is the training pipeline that will use that ordering to fit a reward model. The four roles are named, and the structure is immediately legible: the ordering is partial (the labeler did not compare every pair), possibly inconsistent across labelers (heterogeneity), possibly intransitive within a single labeler under fatigue or framing, and the trained reward model is an attempt to extrapolate the labeler's preference disposition to unseen pairs. The same four-role pattern fits a consumer choosing between grocery-store bundles, a voter ranking candidates, and a foraging bee picking flowers — what changes is which slot fills with what, not the structure of the relation itself.

Relationships to Other Primes

One-hop neighborhood: parents above, mutual partners to the right, children below.Preferencesubsumption: Approach-Avoidance ConflictApproach-Avoida…composition: Expected UtilityExpected Utilitycomposition: Indifference CurvesIndifferenceCurvessubsumption: Loss AversionLoss Aversioncomposition: Marginal UtilityMarginal Utilitycomposition: Preference Heterogeneity and ConflictPreference Hete…subsumption: PrioritizationPrioritizationcomposition: Risk AversionRisk Aversionsubsumption: Time Preference (Discounting Future)Time Preference…composition: Two-Sided MatchingTwo-SidedMatching

Foundational — no parent edges in the catalog.

Children (10) — more specific cases that build on this

  • Approach-Avoidance Conflict is a kind of Preference — Approach-avoidance conflict is a specific kind of preference where a single goal carries both positive and negative valence simultaneously.
  • Loss Aversion is a kind of Preference — Loss aversion is a specialization of preference in which the ordering is reference-dependent and weighs losses more than equivalent gains.
  • Prioritization is a kind of Preference — Prioritization is a kind of preference: it expresses an evaluator's ordering over competing claims and acts on that ordering.
  • Time Preference (Discounting Future) is a kind of Preference — Time preference is a specialization of preference in which the ordering systematically discounts delayed outcomes relative to present ones.
  • Expected Utility presupposes Preference — Expected utility presupposes preference because probability-weighted aggregation requires a prior utility function ranking outcomes.

Not to Be Confused With

  • Not Decision: a preference can exist without commitment to action. I can prefer X to Y without ever choosing between them. Decision is the act of selecting; preference is the disposition that may (or may not) inform it.
  • Not Value Commensuration: preferences need not be expressed in a common metric. I can prefer Beethoven to Bach without committing to a numerical value on each. Commensurability is the condition that comparison along a single metric is possible; preference is the ordering, which may be partial or qualitative.
  • Not Optimization: optimization searches for a best feasible option under a fixed objective. Preference is the ordering that supplies that objective. Optimization presupposes preference (or some target function); preference is the more primitive notion.
  • Not Preference Heterogeneity and Conflict (existing prime): that node is specifically about the conflict variant — what happens when multiple evaluators have different preferences. It would become a child of preference (the conflict case of preferences in tension).

Notes

ChatGPT Pro's R16 pass surfaced this as a clear gap, observing that the catalog already has a CONFLICT-flavored node (preference_heterogeneity_and_conflict) but not the underlying ordering prime. The case is strong: the risk/utility/discounting/loss-aversion cluster of behavioral economics and decision theory is genuinely orphaned, and "preference" is the canonical vocabulary across all those fields. If accepted, the candidate children re-home cleanly; preference_heterogeneity_and_conflict becomes a sub-case (preferences differing across agents) rather than the umbrella. Open question worth flagging: relationship to value (if a value prime exists or is added) — preference might decompose to ordering-over-value, but the simpler near-root positioning is probably right for v1.