MIRAGE: Evaluating and Explaining Inductive Reasoning Process in Language Models

TL;DR

MIRAGE introduces a synthetic dataset to evaluate LLMs' inductive reasoning, revealing their reliance on neighbor-based reasoning rather than rule-based generalization, and analyzing factors affecting their reasoning process.

Contribution

The paper presents MIRAGE, a comprehensive dataset for evaluating inductive reasoning in LLMs, and provides insights into their reasoning strategies and limitations.

Findings

LLMs often do not rely on correct rules for inductive reasoning.

Models tend to focus on similar observed facts near test examples.

Neighbor-based reasoning significantly improves deductive performance.

Abstract

Inductive reasoning is an essential capability for large language models (LLMs) to achieve higher intelligence, which requires the model to generalize rules from observed facts and then apply them to unseen examples. We present MIRAGE, a synthetic dataset that addresses the limitations of previous work, specifically the lack of comprehensive evaluation and flexible test data. In it, we evaluate LLMs' capabilities in both the inductive and deductive stages, allowing for flexible variation in input distribution, task scenario, and task difficulty to analyze the factors influencing LLMs' inductive reasoning. Based on these multi-faceted evaluations, we demonstrate that the LLM is a poor rule-based reasoner. In many cases, when conducting inductive reasoning, they do not rely on a correct rule to answer the unseen case. From the perspectives of different prompting methods, observation numbers, and task forms, models tend to consistently conduct correct deduction without correct inductive rules. Besides, we find that LLMs are good neighbor-based reasoners. In the inductive reasoning process, the model tends to focus on observed facts that are close to the current test example in feature space. By leveraging these similar examples, the model maintains strong inductive capabilities within a localized region, significantly improving its deductive performance.