MORABLES: A Benchmark for Assessing Abstract Moral Reasoning in LLMs with Fables

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Marcuzzo, Matteo, Zangari, Alessandro, Albarelli, Andrea, Camacho-Collados, Jose, Pilehvar, Mohammad Taher
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918141560356864
author Marcuzzo, Matteo
Zangari, Alessandro
Albarelli, Andrea
Camacho-Collados, Jose
Pilehvar, Mohammad Taher
author_facet Marcuzzo, Matteo
Zangari, Alessandro
Albarelli, Andrea
Camacho-Collados, Jose
Pilehvar, Mohammad Taher
contents As LLMs excel on standard reading comprehension benchmarks, attention is shifting toward evaluating their capacity for complex abstract reasoning and inference. Literature-based benchmarks, with their rich narrative and moral depth, provide a compelling framework for evaluating such deeper comprehension skills. Here, we present MORABLES, a human-verified benchmark built from fables and short stories drawn from historical literature. The main task is structured as multiple-choice questions targeting moral inference, with carefully crafted distractors that challenge models to go beyond shallow, extractive question answering. To further stress-test model robustness, we introduce adversarial variants designed to surface LLM vulnerabilities and shortcuts due to issues such as data contamination. Our findings show that, while larger models outperform smaller ones, they remain susceptible to adversarial manipulation and often rely on superficial patterns rather than true moral reasoning. This brittleness results in significant self-contradiction, with the best models refuting their own answers in roughly 20% of cases depending on the framing of the moral choice. Interestingly, reasoning-enhanced models fail to bridge this gap, suggesting that scale - not reasoning ability - is the primary driver of performance.
format Preprint
id arxiv_https___arxiv_org_abs_2509_12371
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MORABLES: A Benchmark for Assessing Abstract Moral Reasoning in LLMs with Fables
Marcuzzo, Matteo
Zangari, Alessandro
Albarelli, Andrea
Camacho-Collados, Jose
Pilehvar, Mohammad Taher
Computation and Language
Artificial Intelligence
68T50
I.2.7
As LLMs excel on standard reading comprehension benchmarks, attention is shifting toward evaluating their capacity for complex abstract reasoning and inference. Literature-based benchmarks, with their rich narrative and moral depth, provide a compelling framework for evaluating such deeper comprehension skills. Here, we present MORABLES, a human-verified benchmark built from fables and short stories drawn from historical literature. The main task is structured as multiple-choice questions targeting moral inference, with carefully crafted distractors that challenge models to go beyond shallow, extractive question answering. To further stress-test model robustness, we introduce adversarial variants designed to surface LLM vulnerabilities and shortcuts due to issues such as data contamination. Our findings show that, while larger models outperform smaller ones, they remain susceptible to adversarial manipulation and often rely on superficial patterns rather than true moral reasoning. This brittleness results in significant self-contradiction, with the best models refuting their own answers in roughly 20% of cases depending on the framing of the moral choice. Interestingly, reasoning-enhanced models fail to bridge this gap, suggesting that scale - not reasoning ability - is the primary driver of performance.
title MORABLES: A Benchmark for Assessing Abstract Moral Reasoning in LLMs with Fables
topic Computation and Language
Artificial Intelligence
68T50
I.2.7
url https://arxiv.org/abs/2509.12371