# Rule vs.Consequence: Dissociable Internal Representations of Moral Reasoning in LLMs

arXiv: https://arxiv.org/abs/2610.16371 · HTML: https://mlsanity.com/paper/2610.16371/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 23 of its 31 references that are in the map.
- First posted on arXiv 12 Oct 2026.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)
- Mechanistic interpretability / [Probing deception, bias and beliefs](https://mlsanity.com/mechinterp/probing-deception-bias-and-beliefs/)

## Builds on (most cited first, 23 of 23)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Toy Models of Superposition](https://mlsanity.com/paper/2209.10652/) (2022)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Aligning AI With Shared Human Values](https://mlsanity.com/paper/2008.02275/) (2020)
- [Social Chemistry 101: Learning to Reason about Social and Moral Norms](https://mlsanity.com/paper/2011.00620/) (2020)
- [Moral Stories: Situated Reasoning about Norms, Intents, Actions, and their Consequences](https://mlsanity.com/paper/2012.15738/) (2020)
- [Moral Mimicry: Large Language Models Produce Moral Rationalizations Tailored to Political Identity](https://mlsanity.com/paper/2209.12106/) (2022)
- [Evaluating the Zero-shot Robustness of Instruction-tuned Language Models](https://mlsanity.com/paper/2306.11270/) (2023)
- [Instructed to Bias: Instruction-Tuned Language Models Exhibit Emergent Cognitive Bias](https://mlsanity.com/paper/2308.00225/) (2023)
- [Strategic Behavior of Large Language Models: Game Structure vs. Contextual Framing](https://mlsanity.com/paper/2309.05898/) (2023)
- [Open (Clinical) LLMs are Sensitive to Instruction Phrasings](https://mlsanity.com/paper/2407.09429/) (2024)
- [Disentangling the Drivers of LLM Social Conformity: An Uncertainty-Moderated Dual-Process Mechanism](https://mlsanity.com/paper/2508.14918/) (2025)
- [Normative Conflicts and Shallow AI Alignment](https://mlsanity.com/paper/2506.04679/) (2025)
- [Measuring Pragmatic Influence in Large Language Model Instructions](https://mlsanity.com/paper/2602.21223/) (2026)
- [Deontological Keyword Bias: The Impact of Modal Expressions on Normative Judgments of Language Models](https://mlsanity.com/paper/2506.11068/) (2025)
- [Normative Reasoning in Large Language Models: A Comparative Benchmark from Logical and Modal Perspectives](https://mlsanity.com/paper/2510.26606/) (2025)
- [Where Norms and References Collide: Evaluating LLMs on Normative Reasoning](https://mlsanity.com/paper/2602.02975/) (2026)
- [DeFrame: Debiasing Large Language Models Against Framing Effects](https://mlsanity.com/paper/2602.04306/) (2026)

## Built on by (most cited first, 1 of 1)

- [Steering Vectors for Competing Ethical Frameworks Are Interchangeable: A Competing-Target Test for Activation Steering](https://mlsanity.com/paper/2610.19359/) (2026)

## Abstract (arXiv)

Large language models are increasingly deployed in morally and legally sensitive contexts, yet little is known about how they internally represent distinctions between moral reasoning styles. We ask whether a model's differential response to rule-framed and consequence-framed moral questions reflects distinct internal structure or merely surface lexical sensitivity. We construct a 2×2 factorial dataset crossing reasoning style (rule-based vs. consequence-based framing) with judgment polarity (permissive vs. prohibitive), drawn from three established moral reasoning benchmarks and evaluated on a cleanroom test set of novel phrasings that includes conflict pairs where the two framings predict opposite labels. Using linear probing and activation steering, we identify two orthogonal directions in the residual stream: Dir-A tracks reasoning style and Dir-B tracks judgment polarity, each near-blind to the other factor (cross-correlation |r| < 0.04). This double dissociation replicates across three instruction-tuned models from different developers at comparable proportional depth, and it is not knife-edged in layer: the directions are recoverable and causally effective throughout an extended band of the network. Dir-A can causally change judgments, and in conflicting scenarios, it selectively corrects errors in rule-based judgments without changing consequence-based judgments, suggesting that it does not simply make the model more likely to accept everything. Generalization beyond the training templates is partial: the direction transfers across benchmark sources, but its advantage over lexical baselines on novel phrasings is modest. We therefore frame the result as moral reasoning style sensitivity, a geometrically dissociable and causally effective structure, rather than internalized normative reasoning.