Beyond Surface Forms: Symbolic Edits as a Test for Logical Reasoning with LLMs
2026-08-31 • Computation and Language
Computation and LanguageArtificial Intelligence
AI summaryⓘ
The authors studied how well large language models (LLMs) can do logical reasoning by changing parts of logic problems in a controlled way. They created a tool that edits logical operators in problems while keeping the meaning the same, then tested how models handle these changes. They found that LLMs often give inconsistent answers when logical structures are changed, showing they don't always fully understand the logic. This method helps evaluate and measure how reliable these models are at reasoning.
large language modelslogical reasoningfirst-order logicconstraint satisfaction problemslogical operatorsnatural language processingmodel evaluationdeductive reasoningrobustness testing
Authors
Ramya Keerthy Thatikonda, Wray Buntine, Ehsan Shareghi
Abstract
Logical reasoning with large language models (LLMs) is a critical capability, as it reflects a system's ability to correctly deduce hypotheses from a given context using faithful deductive processes. However, LLM reasoning has often been shown to be sensitive to small surface-level variations in problem formulation, raising questions about whether models truly follow the underlying logical structure. Studying this behavior is challenging because the symbolic components of logical problems, such as operators and predicates, are difficult to systematically manipulate in natural language. We introduce a tool-driven framework for generating controlled, label-preserving edits to logical reasoning problems. Our method operates on symbolic representations of first-order logic and constraint satisfaction problem tasks, enabling targeted modifications to logical operators and other structural components before translating them back into natural language. Using this framework, we evaluate various LLMs under cumulative and individual operator edits and analyze their behavior in response to these changes. Our quantitative and qualitative analyses show that LLM reasoning behavior under controlled operator edits is inconsistent, regardless of model size or family: models sometimes adapt correctly to structural changes but often fail to track their logical consequences. The results from this automated stress test enable an evaluation of language models across different dimensions and help measure the reliability of their reasoning.