Rational Inattention¶
Economic choice with optimally selected partial information under an attention cost or capacity limit.
Core Idea¶
Rational inattention treats information as a scarce input to economic choice. Instead of assuming the agent sees every relevant state perfectly, a model lets the agent choose a signal or attention allocation under a processing-capacity constraint or information cost and then choose an action. The resulting action may appear inferior under a full-information benchmark yet be optimal once acquiring better information is costly.
Sims's foundational formalization uses a finite Shannon information channel to restrict how observations can influence actions. A later empirical application by Joo estimates an information-cost discrete-choice model in a laundry-detergent market and evaluates the introduction of Tide Pods. Its finding that evaluation friction can offset benefits is a model-dependent result, not a universal verdict on new products or a direct measurement of individual attention budgets.
Scope of Application¶
The central friction is chosen information processing, not any unexplained mistake.
- Macroeconomics. Model slow or noisy adjustment to economic signals.
- Consumer choice. Estimate consideration and evaluation costs.
- Pricing. Study what shocks firms attend to when setting prices.
- Policy analysis. Compare model-dependent outcomes under changed information environments.
Clarity¶
An optimizing agent selects what information to process under a cost or capacity limit, then acts on the partial signal. Sims formalizes finite Shannon capacity; Joo fits a detergent-choice model and reports a conditional Tide Pods welfare result. A missed price tag without modeled information choice is not evidence for rational inattention.
Manages Complexity¶
The agent must solve a joint signal-and-action problem. A richer signal can improve the decision but consumes capacity; an enlarged option set can improve matches yet raise evaluation burden. Empirical work must separate attention frictions from changing tastes, search exposure and ordinary noise, so the model's latent process is not directly observed.
Abstract Reasoning¶
Specify uncertainty, actions and payoffs; model signal capacity or cost; optimize information and response jointly; compare with full information and qualify empirical identification.
Knowledge Transfer¶
The scarce-information logic can inform psychology or machine learning analogies, but this node is the economic model of optimizing an information structure and action under explicit costs. A person casually overlooking a detail is only analogous until that mechanism is specified.
Neighborhood in Abstraction Space¶
Rational Inattention sits in a crowded region of the domain-specific corpus (32nd percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.
Family — Decisions Under Constraint & Commitment (9 abstractions)
Nearest neighbors
- Buridan's ass — 0.92
- Illusion of control — 0.90
- Non-Consequential Reasoning — 0.89
- Strategic Planning — 0.88
- Robust Optimization — 0.87
Computed from structural-signature embeddings · 2026-10-08