Lune

ACL Findings2026

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

Leonardo Bertolazzi, Sandro Pezzelle, Raffaella Bernardi

Abstract

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.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

  • LLMs Encode Harmfulness and Refusal Separately cites Refusal in Language Models Is Mediated by a Single Direction
  • Refusal in Language Models Is Mediated by a Single Direction cites The Linear Representation Hypothesis and the Geometry of Large Language Models
Builds onCited byLeft to right by year; the more cited, the higher.

Builds on9

  1. Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
  2. Refusal in Language Models Is Mediated by a Single DirectionAndy Arditi, Oscar Obeso, Aaquib Syed, Daniel Paleka et al.NeurIPS 2024 · 1,166 citations
  3. The Linear Representation Hypothesis and the Geometry of Large Language ModelsKiho Park, Yo Joong Choe, Victor VeitchICML 2024 · 461 citations
  4. In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space SteeringSheng Liu, Haotian Ye, Lei Xing, James Y. ZouICML 2024 · 244 citations
  5. LLMs Encode Harmfulness and Refusal SeparatelyJiachen Zhao, Jing Huang, Zhengxuan Wu, David Bau et al.NeurIPS 2025 · 93 citations

Related papers

  1. Mitigating Content Effects on Reasoning in Language Models Through Fine-Grained Activation SteeringMarco Valentino, Geonhee Kim, Dhairya Dalal, Zhixue Zhao et al.AAAI 2026 · 15 citations
  2. The Geometry of Reasoning: Flowing Logics in Representation SpaceYufa Zhou, Yixiao Wang, Xunjian Yin, Shuyan Zhou et al.ICLR 2026 · 29 citations
  3. Thinking in Schemas: Robust Syllogistic Reasoning in LLMsFederico Ranaldi, Leonardo Ranaldi, Fabio Massimo Zanzotto, Shay B. CohenACL 2026
  4. Content-free Logical Modification of Large Language Model by Disentangling and Modifying Logic RepresentationXin Wu, Yuqi Bu, Yifei Chen, Yi CaiAAAI 2025 · 1 citation
  5. Calibrating Reasoning in Language Models with Internal ConsistencyZhihui Xie, Jizhou Guo, Tong Yu, Shuai LiNeurIPS 2024 · 37 citations
Dusk over the sea between two cliffs drawn in fine vertical lines