mlsanityData to 14 Oct 2026

How Language Models Conflate Logical Validity with Plausibility: A Representational Analysis of Content Effects

arXiv 2510.06700 ↗First posted 8 Oct 2025cs.CLcited by 3

Not on the map. Why?

Abstract (arXiv)

Both humans and large language models (LLMs) exhibit content effects: biases in which the plausibility of the semantic content of a reasoning problem influences judgments regarding its logical validity. While this phenomenon in humans is best explained by the dual-process theory of reasoning, the mechanisms behind content effects in LLMs remain unclear. In this work, we address this issue by investigating how LLMs encode the concepts of validity and plausibility within their internal representations. We show that both concepts are linearly represented and strongly aligned in representational geometry, leading models to conflate plausibility with validity. Using steering vectors, we demonstrate that plausibility vectors can causally bias validity judgements, and vice versa, and that the degree of alignment between these two concepts predicts the magnitude of behavioral content effects across models. Finally, we construct debiasing vectors that disentangle these concepts, reducing content effects and improving reasoning accuracy. Our findings advance understanding of how abstract logical concepts are represented in LLMs and highlight representational interventions as a path toward more logical systems.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 8 of its 30 references that are in the map. First posted on arXiv 8 Oct 2025.

Builds on

8 of 30 references.

References · 8

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  2. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,085
  3. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  4. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 712
  5. 2023In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steeringcited by 245
  6. 2024Not All Language Model Features Are One-Dimensionally Linearcited by 212
  7. 2025LLMs Encode Harmfulness and Refusal Separatelycited by 88
  8. 2025Mitigating Content Effects on Reasoning in Language Models through Fine-Grained Activation Steeringcited by 18

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026When Decodability Is Not Enough: Logical Validity Representations, Behavioral Dissociation, and Causal Tests in Language Modelscited by 1
  2. 2026Reproducing and Stress-Testing Two Approaches to LLM Reasoning Reliability: Test-Time Probability Aggregation and Logic-Representation Editingcited by 0
  3. 2026When Scientific Contradictions Are Lost in Translationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2022Language models show human-like content effects on reasoning tasks
  2. 2026Convergence Without Understanding: When Language Models Agree on Representations but Disagree on Reasoning
  3. 2023A Systematic Comparison of Syllogistic Reasoning in Humans and Language Models
  4. 2026Do LLMs Share Human-Like Biases? Causal Reasoning Under Prior Knowledge, Irrelevant Context, and Varying Compute Budgets
  5. 2024Comparing Inferential Strategies of Humans and Large Language Models in Deductive Reasoning
  6. 2026The Reasoning Error About Reasoning: Why Different Types of Reasoning Require Different Representational Structures
  7. 2025Towards Evaluting Fake Reasoning Bias in Language Models
  8. 2024Dissociation of Faithful and Unfaithful Reasoning in LLMs