Revealed Rationality: Label-Free Evaluation and Regularization from Representation Theorems

• URL: http://arxiv.org/abs/2608.05015v1
• Date: Wed, 05 Aug 2026 16:21:18 GMT
• Title: Revealed Rationality: Label-Free Evaluation and Regularization from Representation Theorems
• Abstract summary: theorems in decision theory that behavior satisfies certain axioms if and only if it can be rationalized by a well-defined objective.<n>I argue that this if and only if'' structure provides a potentially useful foundation for label-free evaluation and regularization of LLMs and other AI systems.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Representation theorems in decision theory establish that behavior satisfies certain axioms if and only if it can be rationalized by a well-defined objective. I argue that this ``if and only if'' structure provides a potentially useful foundation for label-free evaluation and regularization of LLMs and other AI systems. Axiom compliance can be checked from the model's own responses to synthetic choice problems, with no external labels or human feedback, and the penalties are readily computable. Because the axioms are necessary and sufficient, the resulting checks exhaust the implications of the relevant rationality standard for the elicited data: a model that passes cannot be rejected on rationality grounds by any further test of the same data. I discuss three instantiations: probabilistic coherence via a theorem of de Finetti, preference rationality via Afriat's theorem, and subjective expected utility via a theorem of Echenique and Saito (2015), each yielding a continuous penalty that is zero whenever behavior can be rationalized. Since coherence does not restrict which objective rationalizes behavior, these penalties complement rather than replace other evaluation and training signals.

Related papers

• LiveMathematicianBench: A Live Benchmark for Mathematician-Level Reasoning with Proof Sketches [61.30693283718321]
  We present LiveMathematicianBench, a dynamic multiple-choice benchmark for research-level mathematical reasoning.<n>By grounding evaluation in newly published theorems, it provides a realistic testbed beyond memorized patterns.<n>It employs a proof-sketchguided distractor pipeline that uses high-level proof strategies to construct plausible but invalid answer choices.
  arXiv  Detail & Related papers  (2026-04-02T08:22:17Z)
• Reconstruction of finite Quasi-Probability and Probability from Principles: The Role of Syntactic Locality [0.0]
  Quasi-probabilities appear across diverse areas of physics, but their conceptual foundations remain unclear.<n>We develop a principled framework that derives quasi-probabilities and their conditional calculus from structural consistency requirements.
  arXiv  Detail & Related papers  (2026-02-12T19:00:08Z)
• Probabilistically stable revision and comparative probability: a representation theorem and applications [0.0]
  We show that a selection function can be representable as a strong-stable-set operator on a finite probability space.<n>The resulting logic of probabilistically stable belief revision exhibits strong monotonicity properties.<n>We point out two applications of our main result to the theory of simple voting games and to revealed preference theory.
  arXiv  Detail & Related papers  (2025-09-02T16:42:47Z)
• Formal Power Series Representations in Probability and Expected Utility Theory [0.24578723416255752]
  We advance a theory of coherent preference that surrenders restrictions embodied in orthodox doctrine.<n>Unlike de Finetti's theory, the one we set forth requires neither transitivity nor Archimedeanness nor boundedness nor continuity of preference.<n>Representability by utility is a corollary of this paper's central result, which at once extends H"older's Theorem and strengthens Hahn's Embedding Theorem.
  arXiv  Detail & Related papers  (2025-08-01T03:34:39Z)
• DeepTheorem: Advancing LLM Reasoning for Theorem Proving Through Natural Language and Reinforcement Learning [67.93945726549289]
  DeepTheorem is a comprehensive informal theorem-proving framework exploiting natural language to enhance mathematical reasoning.<n>DeepTheorem includes a large-scale benchmark dataset consisting of 121K high-quality IMO-level informal theorems and proofs.<n>We devise a novel reinforcement learning strategy (RL-Zero) explicitly tailored to informal theorem proving, leveraging the verified theorem variants to incentivize robust mathematical inference.
  arXiv  Detail & Related papers  (2025-05-29T17:59:39Z)
• Rational Inference in Formal Concept Analysis [0.4499833362998487]
  Defeasible conditionals are a form of non-monotonic inference.<n>KLM framework defines a semantics for the propositional case of defeasible conditionals.
  arXiv  Detail & Related papers  (2025-04-07T20:15:20Z)
• An Embedding-based Approach to Inconsistency-tolerant Reasoning with Inconsistent Ontologies [12.760301272393898]
  We propose a novel approach to reasoning with inconsistent semantics based on the embeddings of axioms. We show that our embedding-based method can outperform existing inconsistency-tolerant reasoning methods based on maximal consistent subsets.
  arXiv  Detail & Related papers  (2023-04-04T09:38:02Z)
• Nested Counterfactual Identification from Arbitrary Surrogate Experiments [95.48089725859298]
  We study the identification of nested counterfactuals from an arbitrary combination of observations and experiments. Specifically, we prove the counterfactual unnesting theorem (CUT), which allows one to map arbitrary nested counterfactuals to unnested ones.
  arXiv  Detail & Related papers  (2021-07-07T12:51:04Z)
• Measuring Association Between Labels and Free-Text Rationales [60.58672852655487]
  In interpretable NLP, we require faithful rationales that reflect the model's decision-making process for an explained instance. We demonstrate that pipelines, existing models for faithful extractive rationalization on information-extraction style tasks, do not extend as reliably to "reasoning" tasks requiring free-text rationales. We turn to models that jointly predict and rationalize, a class of widely used high-performance models for free-text rationalization whose faithfulness is not yet established.
  arXiv  Detail & Related papers  (2020-10-24T03:40:56Z)
• Logical Neural Networks [51.46602187496816]
  We propose a novel framework seamlessly providing key properties of both neural nets (learning) and symbolic logic (knowledge and reasoning) Every neuron has a meaning as a component of a formula in a weighted real-valued logic, yielding a highly intepretable disentangled representation. Inference is omni rather than focused on predefined target variables, and corresponds to logical reasoning.
  arXiv  Detail & Related papers  (2020-06-23T16:55:45Z)
• Invariant Rationalization [84.1861516092232]
  A typical rationalization criterion, i.e. maximum mutual information (MMI), finds the rationale that maximizes the prediction performance based only on the rationale. We introduce a game-theoretic invariant rationalization criterion where the rationales are constrained to enable the same predictor to be optimal across different environments. We show both theoretically and empirically that the proposed rationales can rule out spurious correlations, generalize better to different test scenarios, and align better with human judgments.
  arXiv  Detail & Related papers  (2020-03-22T00:50:27Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.