Human Adults and LLMs as Scientists: Who Benefits from Active Exploration?
Why humans excel at learning rules when they get to ask the questions
Adults are notoriously bad at figuring out how multiple causes work together—but only when they're passively watching. When researchers let adults actively test their own hypotheses in a causal learning task, their ability to understand conjunctive rules (where multiple things must happen together) improved dramatically. Large language models, by contrast, showed similar struggles to conjunctive reasoning even with active exploration, and explored less efficiently than humans.
Understanding how humans learn from experimentation has direct applications for designing educational tools, scientific training, and human-AI collaboration. The finding that active control reshapes how people reason about causality suggests that giving learners agency—rather than just showing them data—unlocks cognitive abilities they appear to lack in passive settings. It also identifies a significant gap between human and AI reasoning that matters for tasks where language models are used to model or assist with scientific discovery.