# Frame-Conditioned Moral Computation in LLaMA 3.1-8B-Instruct: A Mechanistic Interpretability Audit of Ethical Reasoning

arXiv: https://arxiv.org/abs/2606.15507 · HTML: https://mlsanity.com/paper/2606.15507/

## 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 10 of its 47 references that are in the map.
- First posted on arXiv 13 Jun 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, 10 of 10)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Emergent Introspective Awareness in Large Language Models](https://mlsanity.com/paper/2601.01828/) (2026)
- [Tracing Moral Foundations in Large Language Models](https://mlsanity.com/paper/2601.05437/) (2026)
- [Auditing the Ethical Logic of Generative AI Models](https://mlsanity.com/paper/2504.17544/) (2025)
- [Investigating Political and Demographic Associations in Large Language Models Through Moral Foundations Theory](https://mlsanity.com/paper/2510.13902/) (2025)
- [Direction-Flipped Influence Audits Reveal Hidden Structure in Moral Choices of LLMs](https://mlsanity.com/paper/2602.22831/) (2026)
- [From Stability to Inconsistency: A Study of Moral Preferences in LLMs](https://mlsanity.com/paper/2504.06324/) (2025)
- [Reasoning or Rhetoric? An Empirical Analysis of Moral Reasoning Explanations in Large Language Models](https://mlsanity.com/paper/2603.21854/) (2026)

## Abstract (arXiv)

Behavioral audits of Large Language Models on moral prompts measure what the model says, not the internal computation producing it. We use Transluce, an AI-driven mechanistic-interpretability platform, to examine LLaMA 3.1-8B-Instruct on 54 moral prompts in four batteries: 17 dilemmas, policy, and meta-ethical questions (B1); 6 role-playing scenarios (B3); and a controlled trolley contrast varying the switching mechanism with people fixed (B4, 15 prompts) or identity attributes with mechanism fixed (B5, 16 prompts). Two complementary metric families, five cluster-level metrics and a six-metric neuron-level panel, converge on a Situational Anchor Effect: domain-specific representations dominate the top of the activation list across every battery. The model's ethics-labeled capacity stays essentially constant; its salience (rank, priority, top-of-list presence) is highly sensitive to the interpretive frame the prompt selects. The B4-vs-B5 contrast confirms the model attends to whichever surface feature varies: aggregate ethics metrics are indistinguishable, but the dominant non-ethics distractor mirrors the design. A multi-temperature audit identifies a candidate ethics neuron (L16/N3837) stable across temperatures; a cross-model behavioral proxy on two frontier models yields preliminary evidence of divergence in self-reported moral focus, consistent with an Alignment Wrapper in which RLHF re-orders surface text without removing underlying domain-first frames. We unify these as Frame-Conditioned Moral Computation: the prompt's surface vocabulary selects a feature manifold, and the moral conclusion is downstream of that selection. Behavioral alignment must be supplemented by Mechanistic Alignment: a research program asking whether ethics-related features can be shown causally privileged under controlled frame variation, not merely loud in the explanation.