# What Does Activation Steering Control? Attribution Across Answer Encodings and Output-Sensitive Subspaces

arXiv: https://arxiv.org/abs/2608.22985 · HTML: https://mlsanity.com/paper/2608.22985/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability.
- Builds on 28 of its 28 references that are in the map.
- First posted on arXiv 24 Aug 2026.

## Placed in

- Mechanistic interpretability / [Activation steering methods](https://mlsanity.com/mechinterp/activation-steering-methods/)

## Builds on (most cited first, 28 of 28)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [Representation Engineering: A Top-Down Approach to AI Transparency](https://mlsanity.com/paper/2310.01405/) (2023)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [The Linear Representation Hypothesis and the Geometry of Large Language Models](https://mlsanity.com/paper/2311.03658/) (2023)
- [Eliciting Latent Predictions from Transformers with the Tuned Lens](https://mlsanity.com/paper/2303.08112/) (2023)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [Social Chemistry 101: Learning to Reason about Social and Moral Norms](https://mlsanity.com/paper/2011.00620/) (2020)
- [Analyzing the Generalization and Reliability of Steering Vectors](https://mlsanity.com/paper/2407.12404/) (2024)
- [AtP*: An efficient and scalable method for localizing LLM behaviour to components](https://mlsanity.com/paper/2403.00745/) (2024)
- [NormBank: A Knowledge Bank of Situational Social Norms](https://mlsanity.com/paper/2305.17008/) (2023)
- [Towards Reliable Evaluation of Behavior Steering Interventions in LLMs](https://mlsanity.com/paper/2410.17245/) (2024)
- [Instruction-following Evaluation through Verbalizer Manipulation](https://mlsanity.com/paper/2307.10558/) (2023)
- [Exploring and steering the moral compass of Large Language Models](https://mlsanity.com/paper/2405.17345/) (2024)
- [Internal Value Alignment in Large Language Models through Controlled Value Vector Activation](https://mlsanity.com/paper/2507.11316/) (2025)
- [On the Non-Identifiability of Steering Vectors in Large Language Models](https://mlsanity.com/paper/2602.06801/) (2026)
- [Where Do Apparent LLM Clinical Triage Failures Arise? Localizing the Multiple-Choice Format Effect](https://mlsanity.com/paper/2605.29889/) (2026)
- [Controllable Value Alignment in Large Language Models through Neuron-Level Editing](https://mlsanity.com/paper/2602.07356/) (2026)
- [How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities](https://mlsanity.com/paper/2603.02578/) (2026)
- [On the Context Sensitivity of LLM Moral Judgment](https://mlsanity.com/paper/2603.23114/) (2026)
- [Cultural Value Alignment Via Latent Activation Steering in Large Language Models](https://mlsanity.com/paper/2605.26365/) (2026)
- [Where Steering Signals Come From: Activation Source Selection in Activation Steering](https://mlsanity.com/paper/2607.25270/) (2026)
- [Scenario-based Probing and Steering Cultural Values in Large Language Models--Extended Version](https://mlsanity.com/paper/2606.11399/) (2026)
- [DecodeShare: Tracing the Shared Subspace of LLM Decode-Time Decisions](https://mlsanity.com/paper/2607.20469/) (2026)

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

- [Replay-Gated Neural Execution: Decoupling Persistent Behavioral Specifications from Neural Realizations in Frozen Language Models](https://mlsanity.com/paper/2609.22243/) (2026)

## Abstract (arXiv)

Activation steering is often evaluated under the answer encoding used to construct the direction. A reported gain may reflect the intended judgment or compatibility with answer identifiers seen during construction. We introduce Cross-Encoding Steering Evaluation, which freezes an intervention while re-encoding answers to the same held-out items. On NormBank, after A/B/C identifiers are reassigned, contrastive activation addition (CAA) induces larger target-versus-source score changes for the extraction indices than for the semantic labels under the new mapping. We call this extraction-index following. Varying identifier vocabulary (A/B/C, X/Y/Z, or 1/2/3) and row order shows that the effect tracks extraction index rather than row position. After matching direction norms across layers, extraction-index following emerges mainly at later depths. A low-rank output-sensitive component containing 15.4% of the direction's squared norm retains 96.3% of this effect. An Inference-Time Intervention (ITI)-style method also favors extraction-index over semantic-label following on NormBank in three models. In aggregate, MNLI favors extraction-index following, whereas Social Chemistry 101 (SC101) favors semantic-label following. Multiple-choice and open-ended evaluations can yield different behavioral conclusions. Thus, a steering gain under one answer encoding does not by itself identify what the intervention controls.