@inproceedings{ding-etal-2026-scit,
title = "{SCIT}: Testing Causal Cache Carriers in Latent Chain-of-Thought Models",
author = "Ding, YI and
Huang, Lijun and
Yang, Menglin",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.846/",
pages = "17905--17946",
ISBN = "979-8-89176-462-0",
abstract = "Latent chain-of-thought models move intermediate reasoning into continuous states, obscuring which internal object causally carries a computation. We introduce SCIT, the Suffix Cache Interchange Test, which patches exact source{--}recipient counterfactuals and crosses cache-segment sufficiency with K/V splits, hidden-state and semantic-source controls, decoded validation, and matched corruption. In GPT-2 arithmetic, counterfactual transfer concentrates in value-cache suffix trajectories rather than hidden states, keys, answer slots, or single-token triggers. Full sufficiency-and-necessity closure holds only for the main CODI-GPT2 checkpoint; the Sim-CoT-style reproduction supports sufficiency and decoded transfer but not necessity. Across broader competent cells, arithmetic-like GPT-2/1B tasks retain latent-tail value/KV transfer, whereas 8B and repaired non-arithmetic tasks shift to prompt-prefix or full-cache K/V; ungated cells receive no call. SCIT thus provides a cache-level diagnostic, a checkpoint-specific arithmetic mechanism, and a competence-gated carrier map{---}not a universal latent-tail claim. Code: https://github.com/YIDING4869/scit-emnlp-2026"
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ding-etal-2026-scit">
<titleInfo>
<title>SCIT: Testing Causal Cache Carriers in Latent Chain-of-Thought Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Y</namePart>
<namePart type="given">I</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lijun</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Menglin</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Latent chain-of-thought models move intermediate reasoning into continuous states, obscuring which internal object causally carries a computation. We introduce SCIT, the Suffix Cache Interchange Test, which patches exact source–recipient counterfactuals and crosses cache-segment sufficiency with K/V splits, hidden-state and semantic-source controls, decoded validation, and matched corruption. In GPT-2 arithmetic, counterfactual transfer concentrates in value-cache suffix trajectories rather than hidden states, keys, answer slots, or single-token triggers. Full sufficiency-and-necessity closure holds only for the main CODI-GPT2 checkpoint; the Sim-CoT-style reproduction supports sufficiency and decoded transfer but not necessity. Across broader competent cells, arithmetic-like GPT-2/1B tasks retain latent-tail value/KV transfer, whereas 8B and repaired non-arithmetic tasks shift to prompt-prefix or full-cache K/V; ungated cells receive no call. SCIT thus provides a cache-level diagnostic, a checkpoint-specific arithmetic mechanism, and a competence-gated carrier map—not a universal latent-tail claim. Code: https://github.com/YIDING4869/scit-emnlp-2026</abstract>
<identifier type="citekey">ding-etal-2026-scit</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.846/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>17905</start>
<end>17946</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SCIT: Testing Causal Cache Carriers in Latent Chain-of-Thought Models
%A Ding, Y. I.
%A Huang, Lijun
%A Yang, Menglin
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F ding-etal-2026-scit
%X Latent chain-of-thought models move intermediate reasoning into continuous states, obscuring which internal object causally carries a computation. We introduce SCIT, the Suffix Cache Interchange Test, which patches exact source–recipient counterfactuals and crosses cache-segment sufficiency with K/V splits, hidden-state and semantic-source controls, decoded validation, and matched corruption. In GPT-2 arithmetic, counterfactual transfer concentrates in value-cache suffix trajectories rather than hidden states, keys, answer slots, or single-token triggers. Full sufficiency-and-necessity closure holds only for the main CODI-GPT2 checkpoint; the Sim-CoT-style reproduction supports sufficiency and decoded transfer but not necessity. Across broader competent cells, arithmetic-like GPT-2/1B tasks retain latent-tail value/KV transfer, whereas 8B and repaired non-arithmetic tasks shift to prompt-prefix or full-cache K/V; ungated cells receive no call. SCIT thus provides a cache-level diagnostic, a checkpoint-specific arithmetic mechanism, and a competence-gated carrier map—not a universal latent-tail claim. Code: https://github.com/YIDING4869/scit-emnlp-2026
%U https://aclanthology.org/2026.findings-emnlp.846/
%P 17905-17946
Markdown (Informal)
[SCIT: Testing Causal Cache Carriers in Latent Chain-of-Thought Models](https://aclanthology.org/2026.findings-emnlp.846/) (Ding et al., Findings 2026)
ACL