Seemingly Plausible Distractors in Multi-Hop Reasoning: Are Large Language Models Attentive Readers?

TL;DR

This paper investigates whether large language models genuinely perform multi-hop reasoning or exploit superficial cues, revealing their vulnerabilities to plausible yet incorrect reasoning chains and proposing a challenging benchmark.

Contribution

The study uncovers subtle ways LLMs bypass multi-hop reasoning and introduces a new benchmark with plausible distractors to evaluate their reasoning capabilities.

Findings

LLMs' performance drops up to 45% with plausible distractors

Models tend to ignore lexical cues but struggle with misleading reasoning paths

Proposed benchmark reveals vulnerabilities in current LLM reasoning abilities

Abstract

State-of-the-art Large Language Models (LLMs) are accredited with an increasing number of different capabilities, ranging from reading comprehension, over advanced mathematical and reasoning skills to possessing scientific knowledge. In this paper we focus on their multi-hop reasoning capability: the ability to identify and integrate information from multiple textual sources. Given the concerns with the presence of simplifying cues in existing multi-hop reasoning benchmarks, which allow models to circumvent the reasoning requirement, we set out to investigate, whether LLMs are prone to exploiting such simplifying cues. We find evidence that they indeed circumvent the requirement to perform multi-hop reasoning, but they do so in more subtle ways than what was reported about their fine-tuned pre-trained language model (PLM) predecessors. Motivated by this finding, we propose a challenging multi-hop reasoning benchmark, by generating seemingly plausible multi-hop reasoning chains, which ultimately lead to incorrect answers. We evaluate multiple open and proprietary state-of-the-art LLMs, and find that their performance to perform multi-hop reasoning is affected, as indicated by up to 45% relative decrease in F1 score when presented with such seemingly plausible alternatives. We conduct a deeper analysis and find evidence that while LLMs tend to ignore misleading lexical cues, misleading reasoning paths indeed present a significant challenge.