Metraipolis › archive › dc_ef9059ffd765

When Reasoning Goes Astray: Attention Dynamics of Uncontrolled Reasoning — arXiv:2609.38817

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.38817
License (as archived): CC0-1.0
Archived: 2026-10-07T12:42:12.469Z
SHA-256 of the archived text: ffb256a4863639c2e60c23fcf878bab61af1cf87ea0f389f0dc0d7417ad1e54e
Stable URL of this page: https://metraipolis.com/documento/dc_ef9059ffd765

How to cite

When Reasoning Goes Astray: Attention Dynamics of Uncontrolled Reasoning — arXiv:2609.38817 — arXiv (en), https://arxiv.org/abs/2609.38817 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ef9059ffd765 (Metraipolis archive dc_ef9059ffd765, sha256 ffb256a4863639c2e60c23fcf878bab61af1cf87ea0f389f0dc0d7417ad1e54e)

BibTeX

@misc{dcef9059ffd765,
  title = {When Reasoning Goes Astray: Attention Dynamics of Uncontrolled Reasoning — arXiv:2609.38817},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38817},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ef9059ffd765; sha256 ffb256a4863639c2e60c23fcf878bab61af1cf87ea0f389f0dc0d7417ad1e54e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38817. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large reasoning models (LRMs) improve performance on complex tasks through extended reasoning, yet the same process can degenerate into redundant verification and persistent generation loops. Such uncontrolled reasoning increases inference cost and creates risks of resource exhaustion and service degradation. However, existing mitigations largely truncate long outputs or react to surface repetition, and thus fail to distinguish normal thinking from uncontrolled reasoning or explain how benign reasoning degenerates into harmful behavior. In this paper, we operationalize LRM generation as four states and further introduce Reasoning-state Analysis via Dynamic Attention Responses (RADAR), which identifies the current reasoning state in real time and characterizes how effective reflection can develop into uncontrolled generation. Guided by RADAR's analysis, we further realign abnormal attention distributions toward patterns observed in normal requests and examine how this correction affects excessive reflection and persistent looping. Temporal analyses show that uncontrolled reasoning is characterized by attention distributions that deviate from normal generation, with abnormal trends becoming detectable before repetition begins. Correcting these deviations through Attention Realignment consistently reduces looping while largely preserving benign performance. Together, RADAR provide a mechanistic account of how reasoning becomes uncontrolled, offering actionable guidance for identifying critical failure stages and designing targeted runtime interventions.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.38817 — 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 2026-10-07T12:14:13.689Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).