Metraipolis › archive › dc_b2006986446e

Emotion Is Not External Knowledge: Internal Attention Signals for Affective State Reasoning — arXiv:2610.21347

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.21347
License (as archived)
CC0-1.0
Archived
2026-10-16T09:57:41.486Z
SHA-256 of the archived text
7d3b80841617dca73827e75e3c3bc6dfc3839c6db44c17e94da03a599e227416
Stable URL of this page
https://metraipolis.com/documento/dc_b2006986446e

How to cite

Emotion Is Not External Knowledge: Internal Attention Signals for Affective State Reasoning — arXiv:2610.21347 — arXiv (en), https://arxiv.org/abs/2610.21347 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b2006986446e (Metraipolis archive dc_b2006986446e, sha256 7d3b80841617dca73827e75e3c3bc6dfc3839c6db44c17e94da03a599e227416)
BibTeX
@misc{dcb2006986446e,
  title = {Emotion Is Not External Knowledge: Internal Attention Signals for Affective State Reasoning — arXiv:2610.21347},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.21347},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b2006986446e; sha256 7d3b80841617dca73827e75e3c3bc6dfc3839c6db44c17e94da03a599e227416},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21347. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large Language Models (LLMs) increasingly operate in multi-turn interactions that require maintaining latent conversational states, not only explicit facts. We study affective state as an implicit state inferred from sparse contextual cues that remain relevant after topic shifts, and ask whether LLMs expose internal attention signals for routing such evidence at inference time. Prior attention interventions rely on labeled in-domain head profiles and externally specified target spans, limiting transfer across affective-evidence distributions. We propose Self-targeted Internal Attention Boosting (SIAB), which localizes sparse affective heads via causal localization (in detail activation patching) on synthetic minimal pairs and boosts, at the answer position, the content tokens these heads already attend to; both edited heads and target tokens are derived from the model's own behavior, enabling the same head set to be reused across this http URL evaluate on ECC and GoEmotions across three LLMs and multiple prompting settings. Across settings, SIAB improves prompt-matched baselines by up to +10.3 macro-F1 on ECC and +5.5 on GoEmotions. Combined with PASTA-style external supervision, it achieves up to +13.84 accuracy points and +12.51 macro-F1 points, while mechanistic analyses trace the gains to sparse, emotion-enriched heads and show that layer proximity affects additivity.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.21347 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T08:44:47.642Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).