# Interpreting Multi-Attribute Confounding through Numerical Attributes in Large Language Models

arXiv: https://arxiv.org/abs/2511.04053 · HTML: https://mlsanity.com/paper/2511.04053/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 22 of its 46 references that are in the map.
- First posted on arXiv 6 Nov 2025.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)

## Builds on (most cited first, 22 of 22)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Calibrate Before Use: Improving Few-Shot Performance of Language Models](https://mlsanity.com/paper/2102.09690/) (2021)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories](https://mlsanity.com/paper/2212.10511/) (2022)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Large Language Models Can Be Easily Distracted by Irrelevant Context](https://mlsanity.com/paper/2302.00093/) (2023)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Mechanistic Interpretability for AI Safety -- A Review](https://mlsanity.com/paper/2404.14082/) (2024)
- [Adaptive Chameleon or Stubborn Sloth: Revealing the Behavior of Large Language Models in Knowledge Conflicts](https://mlsanity.com/paper/2305.13300/) (2023)
- [In-Context Learning Creates Task Vectors](https://mlsanity.com/paper/2310.15916/) (2023)
- [Not All Language Model Features Are One-Dimensionally Linear](https://mlsanity.com/paper/2405.14860/) (2024)
- [Language Models Use Trigonometry to Do Addition](https://mlsanity.com/paper/2502.00873/) (2025)
- [The Semantic Hub Hypothesis: Language Models Share Semantic Representations Across Languages and Modalities](https://mlsanity.com/paper/2411.04986/) (2024)
- [Beyond Single Concept Vector: Modeling Concept Subspace in LLMs with Gaussian Distribution](https://mlsanity.com/paper/2410.00153/) (2024)
- [Evaluating LLMs' Mathematical Reasoning in Financial Document Question Answering](https://mlsanity.com/paper/2402.11194/) (2024)
- [Language Models Do Hard Arithmetic Tasks Easily and Hardly Do Easy Arithmetic Tasks](https://mlsanity.com/paper/2406.02356/) (2024)
- [Do Language Models Know the Way to Rome?](https://mlsanity.com/paper/2109.07971/) (2021)
- [More Samples or More Prompts? Exploring Effective In-Context Sampling for LLM Few-Shot Prompt Engineering](https://mlsanity.com/paper/2311.09782/) (2023)
- [CoME: An Unlearning-based Approach to Conflict-free Model Editing](https://mlsanity.com/paper/2502.15826/) (2025)
- [The Geometry of Numerical Reasoning: Language Models Compare Numeric Properties in Linear Subspaces](https://mlsanity.com/paper/2410.13194/) (2024)
- [Related Knowledge Perturbation Matters: Rethinking Multiple Pieces of Knowledge Editing in Same-Subject](https://mlsanity.com/paper/2502.06868/) (2025)
- [ChatGPT Doesn't Trust Chargers Fans: Guardrail Sensitivity in Context](https://mlsanity.com/paper/2407.06866/) (2024)

## Abstract (arXiv)

Although behavioral studies have documented numerical reasoning errors in large language models (LLMs), the underlying representational mechanisms remain unclear. We hypothesize that numerical attributes occupy shared latent subspaces and investigate two questions:(1) How do LLMs internally integrate multiple numerical attributes of a single entity? (2)How does irrelevant numerical context perturb these representations and their downstream outputs? To address these questions, we combine linear probing with partial correlation analysis and prompt-based vulnerability tests across models of varying sizes. Our results show that LLMs encode real-world numerical correlations but tend to systematically amplify them. Moreover, irrelevant context induces consistent shifts in magnitude representations, with downstream effects that vary by model size. These findings reveal a vulnerability in LLM decision-making and lay the groundwork for fairer, representation-aware control under multi-attribute entanglement.