BabySim-Agent: A Natural Language-Based Multi-Agent Framework for Simulating Parent-Infant Interaction Narratives and Evaluating care-giving Strategies

DOI

10.2991/978-2-38476-611-6_38How to use a DOI?

Keywords

Large Language Model; Multi-Agent Simulation; Developmental Psychology; Parent-Infant Interaction; Computational Framework; Attachment Theory

Abstract

Traditional infant psychological research requires recruiting families, longitudinal tracking over months, and substantial funding. We propose BabySim-Agent, the first Large Language Model (LLM) multi-agent framework for generating parent-infant interaction narratives and computationally evaluating care-giving strategies. The framework consists of two LLM-powered agents and one rule-based assessment module. We conducted 243 simulated interaction episodes spanning three age points (4, 9, 15 months), three temperamental profiles (easy, difficult, slow-to-warm), three care-giving scenarios, and three replicates. One-way ANOVA results demonstrated that the developmentally-informed care-giving strategy significantly outperformed both random and raw-LLM baselines across all four dimensions (p < 0.001 for all comparisons). Bootstrap 95% confidence intervals for eta-squared ranged from [0.078, 0.231] to [0.700, 0.794]. These findings demonstrate the viability of LLM-based multi-agent systems as low-cost, rapid prototyping tools for exploratory hypothesis testing in developmental psychology.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 3.0 International License (http://creativecommons.org/licenses/by-nc/3.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Hangyu Wu
PY  - 2026
DA  - 2026/09/07
TI  - BabySim-Agent: A Natural Language-Based Multi-Agent Framework for Simulating Parent-Infant Interaction Narratives and Evaluating care-giving Strategies
BT  - Proceedings of the 2026 12th International Conference on Digital Humanities and Frontiers in Social Sciences (DHFSS 2026)
PB  - Atlantis Press
SP  - 337
EP  - 347
SN  - 2352-5398
UR  - https://doi.org/10.2991/978-2-38476-611-6_38
DO  - 10.2991/978-2-38476-611-6_38
ID  - Wu2026
ER  -