Metraipolis › archive › dc_2f1f74f300bf

Counting on Thinking: Tracing Evidence Integration in Language Models — arXiv:2609.32932

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.32932
License (as archived)
CC0-1.0
Archived
2026-10-07T22:36:59.140Z
SHA-256 of the archived text
fabaecaa8f6abd6325714f4d557ebcac9a2da180484ec529623bc19080563044
Stable URL of this page
https://metraipolis.com/documento/dc_2f1f74f300bf

How to cite

Counting on Thinking: Tracing Evidence Integration in Language Models — arXiv:2609.32932 — arXiv (en), https://arxiv.org/abs/2609.32932 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2f1f74f300bf (Metraipolis archive dc_2f1f74f300bf, sha256 fabaecaa8f6abd6325714f4d557ebcac9a2da180484ec529623bc19080563044)
BibTeX
@misc{dc2f1f74f300bf,
  title = {Counting on Thinking: Tracing Evidence Integration in Language Models — arXiv:2609.32932},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32932},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2f1f74f300bf; sha256 fabaecaa8f6abd6325714f4d557ebcac9a2da180484ec529623bc19080563044},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32932. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Finite computational resources force a tradeoff between automatic System 1 processes and costly System 2 thinking. Large language models (LLMs) can spend extra computation on hard problems, yet direct answers struggle even with counting, an elementary operation humans and animals perform automatically. We ask why this requires thinking in LLMs. Evidence integration has long been used in psychology and neuroscience to probe decision-making. Our evidence-integration task presents one letter per conversational turn and asks which of two target letters appeared more often. A running count difference solves the task optimally by weighting every letter equally; tokens at each turn could represent and update this difference. Direct responses instead weighted evidence unevenly, with strong recency effects, and assigned less probability to the correct answer as difficulty increased. Thinking improved performance and made integration weights nearly uniform, yet final-query attention remained concentrated on the sequence ends in both modes. Reasoning trajectories showed models revisiting input, recounting letters, and checking intermediate counts that informed the answer, suggesting that thinking constructs the accumulated count that direct responses lack rather than reading out one already formed. Reasoning-token costs grew with the number of letters far more than with coherence. Outcome feedback did not bring this computation into direct responses: under in-context reinforcement learning (ICRL), performance deteriorated over repeated games and recency effects strengthened, yet models grew more confident. Humans and animals amortize such computations into automatic processes, whereas current LLMs still pay for them with thinking on every trial. Which operations learning can make directly available remains central to how future models allocate computation.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.32932 — 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 2026-10-07T22:14:27.810Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).