# When Does the Concept of "Dog" Emerge in an Audio LLM?

arXiv: https://arxiv.org/abs/2609.33458 · HTML: https://mlsanity.com/paper/2609.33458/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 5 of its 19 references that are in the map.
- First posted on arXiv 27 Sep 2026.

## Builds on (most cited first, 5 of 5)

- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Qwen2.5-Omni Technical Report](https://mlsanity.com/paper/2503.20215/) (2025)
- [Designing and Interpreting Probes with Control Tasks](https://mlsanity.com/paper/1909.03368/) (2019)
- [Verbalizable Representations Form a Global Workspace in Language Models](https://mlsanity.com/paper/2607.15495/) (2026)

## Abstract (arXiv)

Multimodal large language models answer audio questions, but how they represent auditory semantics and use them in decisions remains unclear, limiting our understanding of response formation. We study dog barking in Qwen2.5-Omni-7B using Jacobian lens (J-lens) readout and directional interventions. We define the dog direction as a J-lens-derived hidden-state vector associated with dog; adding or removing its component modulates dog-related information. We find this information decodable without dog/bark prompt cues or animal-identification requirements. Directional interventions change response tendencies and some final answers, with effects concentrated in late-layer states immediately before generation across species classification, vocalization classification, and sound description. The dog direction shows no comparable advantage over controls in animal/other classification. These results provide causal-intervention evidence that the dog direction affects output scores in a task-dependent manner, most consistently at L22 and L24 immediately before generation.