Saved in:
Bibliographic Details
Main Authors: Ma, Qingchuan, Wu, Yuhang, Zheng, Xiawu, Ji, Rongrong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.23833
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918038820880384
author Ma, Qingchuan
Wu, Yuhang
Zheng, Xiawu
Ji, Rongrong
author_facet Ma, Qingchuan
Wu, Yuhang
Zheng, Xiawu
Ji, Rongrong
contents In this paper, we aim to establish a simple, effective, and theoretically grounded benchmark for rigorously probing abstract reasoning in Large Language Models (LLMs). To achieve this, we first develop a mathematic framework that defines abstract reasoning as the ability to: (i) extract essential patterns independent of surface representations, and (ii) apply consistent rules to these abstract patterns. Based on this framework, we introduce two novel complementary metrics: \(\scoreGamma\) measures basic reasoning accuracy, while \(\scoreDelta\) quantifies a model's reliance on specific symbols rather than underlying patterns - a key indicator of true abstraction versus mere memorization. To implement this measurement, we design a benchmark: systematic symbol remapping in rule-based tasks, which forces models to demonstrate genuine pattern recognition beyond superficial token matching. Extensive LLM evaluations using this benchmark (commercial API models, 7B-70B, multi-agent) reveal:1) critical limitations in non-decimal arithmetic and symbolic reasoning; 2) persistent abstraction gaps despite chain-of-thought prompting; and 3) \(\scoreDelta\)'s effectiveness in robustly measuring memory dependence by quantifying performance degradation under symbol remapping, particularly highlighting operand-specific memorization. These findings underscore that current LLMs, despite domain-specific strengths, still lack robust abstract reasoning, highlighting key areas for future improvement.
format Preprint
id arxiv_https___arxiv_org_abs_2505_23833
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Benchmarking Abstract and Reasoning Abilities Through A Theoretical Perspective
Ma, Qingchuan
Wu, Yuhang
Zheng, Xiawu
Ji, Rongrong
Computation and Language
In this paper, we aim to establish a simple, effective, and theoretically grounded benchmark for rigorously probing abstract reasoning in Large Language Models (LLMs). To achieve this, we first develop a mathematic framework that defines abstract reasoning as the ability to: (i) extract essential patterns independent of surface representations, and (ii) apply consistent rules to these abstract patterns. Based on this framework, we introduce two novel complementary metrics: \(\scoreGamma\) measures basic reasoning accuracy, while \(\scoreDelta\) quantifies a model's reliance on specific symbols rather than underlying patterns - a key indicator of true abstraction versus mere memorization. To implement this measurement, we design a benchmark: systematic symbol remapping in rule-based tasks, which forces models to demonstrate genuine pattern recognition beyond superficial token matching. Extensive LLM evaluations using this benchmark (commercial API models, 7B-70B, multi-agent) reveal:1) critical limitations in non-decimal arithmetic and symbolic reasoning; 2) persistent abstraction gaps despite chain-of-thought prompting; and 3) \(\scoreDelta\)'s effectiveness in robustly measuring memory dependence by quantifying performance degradation under symbol remapping, particularly highlighting operand-specific memorization. These findings underscore that current LLMs, despite domain-specific strengths, still lack robust abstract reasoning, highlighting key areas for future improvement.
title Benchmarking Abstract and Reasoning Abilities Through A Theoretical Perspective
topic Computation and Language
url https://arxiv.org/abs/2505.23833