@inproceedings{rabern-etal-2026-logicskills,
title = "{L}ogic{S}kills: A Structured Benchmark for Formal Reasoning in Large Language Models",
author = "Rabern, Brian and
Mondorf, Philipp and
Plank, Barbara",
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.98/",
pages = "2035--2047",
ISBN = "979-8-89176-462-0",
abstract = "Large language models perform well on many logical reasoning benchmarks, but it remains unclear which core logical skills they truly master. To address this, we introduce LogicSkills, a benchmark that isolates three fundamental logical skills: (i) formal symbolization$\unicode{x2013}$translating premises into first-order logic; (ii) countermodel construction$\unicode{x2013}$showing that an argument is logically invalid by constructing a finite countermodel; and (iii) validity assessment$\unicode{x2013}$determining whether a conclusion follows from a set of premises. Items are drawn from the two-variable fragment of first-order logic without identity and are presented in both English and a Carrollian nonce-word language. All instances are solver-verified with Z3 for correctness and non-triviality. Across conventional instruction-tuned LLMs, performance is high on validity assessment but substantially lower on formal symbolization and countermodel construction, highlighting that high task-level accuracy can mask weaknesses in core logical skills. In contrast, recent reasoning-tuned models perform strongly across all three tasks, suggesting a more systematic logical skill profile."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="rabern-etal-2026-logicskills">
<titleInfo>
<title>LogicSkills: A Structured Benchmark for Formal Reasoning in Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Brian</namePart>
<namePart type="family">Rabern</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Philipp</namePart>
<namePart type="family">Mondorf</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Barbara</namePart>
<namePart type="family">Plank</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>Large language models perform well on many logical reasoning benchmarks, but it remains unclear which core logical skills they truly master. To address this, we introduce LogicSkills, a benchmark that isolates three fundamental logical skills: (i) formal symbolization\unicodex2013translating premises into first-order logic; (ii) countermodel construction\unicodex2013showing that an argument is logically invalid by constructing a finite countermodel; and (iii) validity assessment\unicodex2013determining whether a conclusion follows from a set of premises. Items are drawn from the two-variable fragment of first-order logic without identity and are presented in both English and a Carrollian nonce-word language. All instances are solver-verified with Z3 for correctness and non-triviality. Across conventional instruction-tuned LLMs, performance is high on validity assessment but substantially lower on formal symbolization and countermodel construction, highlighting that high task-level accuracy can mask weaknesses in core logical skills. In contrast, recent reasoning-tuned models perform strongly across all three tasks, suggesting a more systematic logical skill profile.</abstract>
<identifier type="citekey">rabern-etal-2026-logicskills</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.98/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>2035</start>
<end>2047</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T LogicSkills: A Structured Benchmark for Formal Reasoning in Large Language Models
%A Rabern, Brian
%A Mondorf, Philipp
%A Plank, Barbara
%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 rabern-etal-2026-logicskills
%X Large language models perform well on many logical reasoning benchmarks, but it remains unclear which core logical skills they truly master. To address this, we introduce LogicSkills, a benchmark that isolates three fundamental logical skills: (i) formal symbolization\unicodex2013translating premises into first-order logic; (ii) countermodel construction\unicodex2013showing that an argument is logically invalid by constructing a finite countermodel; and (iii) validity assessment\unicodex2013determining whether a conclusion follows from a set of premises. Items are drawn from the two-variable fragment of first-order logic without identity and are presented in both English and a Carrollian nonce-word language. All instances are solver-verified with Z3 for correctness and non-triviality. Across conventional instruction-tuned LLMs, performance is high on validity assessment but substantially lower on formal symbolization and countermodel construction, highlighting that high task-level accuracy can mask weaknesses in core logical skills. In contrast, recent reasoning-tuned models perform strongly across all three tasks, suggesting a more systematic logical skill profile.
%U https://aclanthology.org/2026.findings-emnlp.98/
%P 2035-2047
Markdown (Informal)
[LogicSkills: A Structured Benchmark for Formal Reasoning in Large Language Models](https://aclanthology.org/2026.findings-emnlp.98/) (Rabern et al., Findings 2026)
ACL