Curiosity-driven development of action and language in robots through self-exploration
Science Advances · 2026-07-24 · 12 (30)
DOI: 10.1126/sciadv.aee7533 ↗At a glance
Journal article in Science Advances, published on 2026-07-24. Written by Theodore J Tinker and 2 co-authors. Institution: Okinawa Institute of Science and Technology, Okinawa, Japan. Major MEDLINE topics: Robotics, Language Development, Exploratory Behavior, Language. 62 references. The full text is freely available in PubMed Central. Funded by: Japan Society for the Promotion of Science, Okinawa Institute of Science and Technology Graduate University.
Abstract (original language)
Infants acquire language with generalization from minimal experience, whereas large language models require billions of training tokens. What underlies efficient development in humans? We investigated this problem through experiments wherein robotic agents learn to perform actions associated with imperative sentences (e.g., push red cube) via curiosity-driven self-exploration. Our approach amortizes active inference using Q-learning, enabling intrinsically motivated developmental learning. The simulations reveal key findings corresponding to observations in developmental psychology. (i) Generalization improves markedly as the scale of compositional elements increases. (ii) Curiosity-driven exploration enables faster learning. (iii) Rote pairing of sentences and actions precedes compositional generalization. (iv) Exception handling induces U-shaped developmental performance, a pattern like representational redescription in child language learning. These results suggest that curiosity-driven active inference accounts for how intrinsically motivated sensorimotor-linguistic learning supports scalable compositional generalization and exception handling in humans and artificial agents.
Subjects
MeSH terms (MEDLINE)
Authors and institutions
- Okinawa Institute of Science and Technology, Okinawa, Japan.
ORCID profiles: Theodore J Tinker, Kenji Doya, Jun Tani
Funding
- Japan Society for the Promotion of Science
- Okinawa Institute of Science and Technology Graduate University
Cite
APA
Tinker, T. J., Doya, K., & Tani, J. (2026). Curiosity-driven development of action and language in robots through self-exploration. Science Advances, 12(30). https://doi.org/10.1126/sciadv.aee7533
Vancouver
Tinker TJ, Doya K, Tani J. Curiosity-driven development of action and language in robots through self-exploration. Science Advances. 2026;12(30). doi:10.1126/sciadv.aee7533
BibTeX
@article{Tinker2026,
author = {Tinker, Theodore J and Doya, Kenji and Tani, Jun},
title = {Curiosity-driven development of action and language in robots through self-exploration},
journal = {Science Advances},
year = {2026},
volume = {12},
number = {30},
doi = {10.1126/sciadv.aee7533}
}
EUVIMED