Hypothesis Generation and Inductive Inference in Children and Language Models
The study explores hypothesis generation and inductive inference in children and language models. It compares how both groups infer latent causes in uncertain environments through a structured task. The findings indicate that while both adapt to environmental structures, their information-seeking behaviors differ significantly.
- ▪The research investigates the computational principles behind human inference and compares them to LLM-based agents.
- ▪Children's behavior is explained by subjective evidence reliability and online hypothesis generation.
- ▪LLM-based agents replicate children's responses but tend to over-observe and over-comply with instructions.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.24528 (cs) [Submitted on 23 May 2026] Title:Hypothesis Generation and Inductive Inference in Children and Language Models Authors:Jeffrey Qin, Wasu Top Piriyakulki, Zhuangfei Gao, Mia Radovanovic, Jessica Sommerville, Kevin Ellis, Marta Kryven View a PDF of the paper titled Hypothesis Generation and Inductive Inference in Children and Language Models, by Jeffrey Qin and 6 other authors View PDF HTML (experimental) Abstract:Real world decision-making requires constructing mental models under uncertainty over evidence, over the underlying causal rules, and over the state of the world itself.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
