Human Adults and LLMs as Scientists: Who Benefits from Active Exploration?

2026-06-04

AI Summary

This study investigates whether human adults experience difficulty in reasoning about conjunctive causal rules—where multiple causes must co-occur—and whether active exploration mitigates this challenge. It further provides the first systematic comparison of the strategies employed by humans and large language models (LLMs) in active causal learning. Using a modified “blicket detector” task, both human participants and LLMs freely intervened to infer causal relationships under conjunctive and disjunctive causal structures. The results demonstrate that active intervention significantly improves human accuracy in reasoning about conjunctive rules. While some LLMs achieve human-level inference accuracy, they exhibit lower exploration efficiency and similarly display a performance gap between conjunctive and disjunctive conditions.

Abstract

A long-standing finding in the causal learning literature is that adults struggle to identify conjunctive causal rules, where an effect requires the simultaneous presence of multiple causes, while performing better in disjunctive settings. However, most demonstrations of this ``conjunctive handicap'' rely on passive observation paradigms with limited evidence, where learners have no control over evidence generation. This paper asks whether this bias persists when adults are granted agency through active exploration. Using a modified ``blicket detector'' task, adult participants freely intervened to identify causal objects under conjunctive or disjunctive rule structures. We show that active exploration substantially improves adults' conjunctive causal reasoning, although conjunctive rules still require more tests to infer than disjunctive rules. We further compare human performance to a range of large language models in the same setting. While some state-of-the-art models approach human-level performance on hypothesis inference accuracy, they often exhibit less efficient exploration strategies and similar conjunctive-disjunctive performance gaps.

Problem

• conjunctive causal reasoning
• active exploration
• causal learning
• large language models
• disjunctive rules

Innovation

• active exploration
• conjunctive causal reasoning
• causal learning
• large language models
• blicket detector