Tutorial at ICDL 2026 (Kyoto, Japan)
Afternoon Session (13:45–17:05), September 15, 2026.
Digital Brain with Predictive-Coding-Inspired Variational RNN: Multiscale Neural Dynamics, Data Assimilation, and Virtual Interventions
Embedded Files
This tutorial introduces Predictive-Coding-Inspired Variational RNN (PV-RNN) for digital twin brain modeling, focusing on how a hierarchical generative model can support online assimilation, state estimation, prediction, and virtual intervention. We cover the core ideas needed to apply PV-RNN in practice and discuss key modeling choices—such as temporal hierarchy and precision control—that enable multiscale representations. Applications span neural recordings (EEG/ECoG/SEEG), brain–body physiology (e.g., ECG/respiration), and behavior. Attendees will gain practical perspectives on synchronizing models with real data, interpreting latent states, and using simulation to explore perturbations and intervention scenarios.
Target audience
This tutorial primarily targets graduate students and researchers who have a basic background in time-series generative modeling (e.g., VAE/RNN, state-space models, Bayesian inference) and are interested in computational frameworks for developmental learning, and adaptation. In particular, we envision participants working in cognitive development, computational neuroscience, predictive processing, and computational psychiatry, who wish to introduce generative-model-based integration, data assimilation, and virtual intervention for real-world datasets such as EEG/ECoG/SEEG, brain–body signals (ECG, respiration) and behavior.
Organizers
Tentative schedule（Half-day）
Afternoon Session (13:45–17:05), September 15, 2026.
̵ Part-A: PV-RNN Fundamentals (theory, implementation, and basics of data assimilation) — Tani
̵ Part-B: Hierarchical Principles (temporal hierarchy and precision control as design principles for inference and representation learning) — Idei
̵ Part-C: Digital Twin Brain with PV-RNN (data assimilation, state estimation, prediction, virtual intervention, plus multiscale/multimodal extensions and practical considerations) — Yamashita
̵ Q&A / Open discussion with participants
Lecturer
The tutorial lecture will be delivered by Prof. Jun Tani, Dr. Hayato Idei, and Dr. Yuichi Yamashita. Prof. Tani (OIST) proposed the original PV-RNN framework and has led research in cognitive neurorobotics grounded in predictive-coding and active-inference perspectives, including a comprehensive monograph on the topic. Dr. Idei (NCNP) has advanced PV-RNN–based hierarchical inference frameworks with precision control and multimodal integration, applying them to explain neuroscientific phenomena and to enable scalable, high-dimensional, multiscale robotic intelligence. Dr. Yamashita (NCNP), a clinical psychiatrist and computational psychiatry researcher, develops digital twin brain approaches with PV-RNN for modeling and interpreting neural and behavioral dynamics.
References
[1] Takahashi, Y., Idei, H., Komatsu, M. Tomita, H., & Yamashita, Y. Digital twin brain simulator for real-time consciousness monitoring and virtual intervention using primate electrocorticogram data. npj Digit. Med. 8, 80 (2025). https://doi.org/10.1038/s41746-025-01444-1
[2] De Domenico, M., Allegri, L., Caldarelli, G. et al. Challenges and opportunities for digital twins in precision medicine from a complex systems perspective. npj Digit. Med. 8, 37 (2025). https://doi.org/10.1038/s41746-024-01402-3
[3] Ahmadi, A & Tani, J. (2019). A Novel Predictive-Coding-Inspired Variational RNN Model for Online Prediction and Recognition. Neural Computation, 31(11), 2025–2074. doi: https://doi.org/10.1162/neco_a_01228
[4] Yamashita Y, Tani J (2008) Emergence of Functional Hierarchy in a Multiple Timescale Neural Network Model: A Humanoid Robot Experiment. PLoS Comput Biol 4(11): e1000220. https://doi.org/10.1371/journal.pcbi.1000220
[5] Idei, H., Ohata, W., Yamashita, Y. et al. Emergence of sensory attenuation based upon the free-energy principle. Sci Rep 12, 14542 (2022). https://doi.org/10.1038/s41598-022-18207-7
[6] J. Tani, Exploring Robotic Minds: Actions, Symbols, and Consciousness as Self-Organizing Dynamic Phenomena, New York, NY, USA: Oxford Univ. Press., 2016. https://doi.org/10.1093/acprof:oso/9780190281069.001.0001
[7] H. Idei, T. Miyake, T. Ogata, and Y. Yamashita. Scalable predictive processing framework for multitask caregiving robots, 2025, arXiv. doi: 10.48550/ARXIV.2510.25053 (2025).
[8] Soda T, Ahmadi A, Tani J, Honda M, Hanakawa T and Yamashita Y (2023) Simulating developmental diversity: Impact of neural stochasticity on atypical flexibility and hierarchy. Front. Psychiatry 14:1080668. https://doi.org/10.3389/fpsyt.2023.1080668
Acknowledgement
Jun Tani is partially supported by Grant-in-Aid for Transformative Research Area (A): unified theory of prediction and action (24H02175). Yuichi Yamashita is partially supported by JST CREST #JPMJCR21P4, JST Moonshot R&D #JPMJMS2031, AMED Multidisciplinary Frontier Brain and Neuroscience Discoveries (Brain/MINDS 2.0) #JP24wm0625407, JSPS KAKENHI #JP24H00076, #JP24K00499, #JP25H01173, and Intramural Research Grant (6-9, 7-9) for Neurological and Psychiatric Disorders of NCNP. Hayato Idei is partially supported by JST ACT-X No. JPMJAX24C2.
Page updated
Google Sites
Report abuse