Takahashi & Shiramatsu Lab · 2024–2026
Bayesian inference in neuronal cultures
Master's thesis research on collective Bayesian inference and predictive coding in networks of cultured neurons.
Talk at IEEJ 2025
Wataru Kawakami, M.D.
PhD Student, The University of Tokyo
Department of Mechano-Informatics, Graduate School of Information Science and Technology
Takahashi & Shiramatsu Lab (Bio-Intelligence Lab)
I am a PhD student at the University of Tokyo, advised by Prof. Hirokazu Takahashi. I trained in medicine at Kyoto University (M.D., 2024) and hold a Japanese medical license.
I am interested in consciousness, and in the information processing that perception is rooted in. I study it through temporal prediction, sensorimotor inference, and learning and memory in the cerebral cortex. My work spans in vivo experiments (microelectrode-array recordings and calcium imaging), in vitro experiments with neuronal cultures, and computational models of predictive coding, Bayesian inference, and associative memory.
Previously, I worked on reconstructing 3D representations from human fMRI with AI models in the Kamitani Lab at Kyoto University, and on Japanese medical language models at Preferred Networks.
News
-
Poster at Neuroscience 2026 (Society for Neuroscience), Washington, DC: “Acoustic rearing scales auditory cortical information processing while preserving a music-aligned temporal profile.”
-
Oral presentation at Models of Consciousness 7 (MoC7), University of Copenhagen: “Guarded Reflection: How Retention Dissolves the Regress Problem of Self-Consciousness.”
-
Started my PhD at the University of Tokyo, and was selected for SPRING GX, a JST-supported doctoral research fellowship.
-
Completed my Master of Information Science and Technology at the University of Tokyo.
-
First-author paper at the GenAI4Health Workshop, NeurIPS 2025: “Stabilizing Reasoning in Medical LLMs with Continued Pretraining and Reasoning Preference Optimization.” arXiv
Research
Current projects
Synaptic perturbations and memory selection
Earlier projects
Takahashi & Shiramatsu Lab · 2024–2026
Auditory cortex and the structure of music
Auditory information processing in the rat auditory cortex, and how it relates to the structure of music.
Poster at SfN 2026
Preferred Networks · 2024–2025
Reasoning in medical language models
Led research on continued pretraining and reasoning preference optimization for Japanese medical language models.
Kamitani Lab, Kyoto University · 2021–2024
3D representations from human fMRI
Worked on reconstructing three-dimensional representations from human fMRI data using AI models and analyses of neural representations.
Methods & expertise
- Experimental neuroscience
- In vivo electrophysiology (microelectrode-array recordings), mouse calcium imaging (two-photon and Inscopix nVista miniscope), human fMRI, and neuronal culture.
- Computational neuroscience
- Mathematical modeling of neural computation, predictive coding, associative memory, and synaptic plasticity.
- Machine learning & data analysis
- Automated calcium-imaging pipelines (Suite2p), neural decoding, deep learning, and reproducible research software development.
Publications
-
NeurIPS 2025Workshop paper & poster
Stabilizing Reasoning in Medical LLMs with Continued Pretraining and Reasoning Preference Optimization
GenAI4Health Workshop at NeurIPS 2025
-
SIGGRAPH Asia 2024Conference demonstration
Online Chat with Living Neuronal Cultures
SIGGRAPH Asia 2024 Real-Time Live!
Selected Talks & Posters
-
SfN 2026Poster
Acoustic rearing scales auditory cortical information processing while preserving a music-aligned temporal profile
Neuroscience 2026, Society for Neuroscience (SfN) · Washington, DC, USA
-
MoC7 2026Oral presentation
Guarded Reflection: How Retention Dissolves the Regress Problem of Self-Consciousness
Models of Consciousness 7 (MoC7), University of Copenhagen · Copenhagen, Denmark · 14 October 2026
-
IEEJ 2025Oral presentation
Measurement of variational free energy across electrically connected distributed neuronal cultures
IEEJ Electronics, Information and Systems Conference · Ishikawa, Japan · Presented in Japanese (title translated)
Awards & Funding
-
SPRING GX
2026 – 2029JST-supported doctoral research fellowship, The University of Tokyo. Selected student.
-
MVP Runner-up Award
2026CPC Spring Camp.
-
IRCN Retreat Brainstorming Award
2025International Research Center for Neurointelligence (IRCN), The University of Tokyo. Research proposal support; project funding JPY 800,000.
-
WINGS-IIW
2024 – 2029International Graduate Program of Innovation for Intelligent World, The University of Tokyo. Selected student.
-
Independent Research and Activities Support
FY2024WINGS-IIW. Project funding JPY 500,000.
Education & Experience
Education
-
PhD Student, The University of Tokyo
Apr 2026 – presentDepartment of Mechano-Informatics, Graduate School of Information Science and Technology. Advisor: Prof. Hirokazu Takahashi.
-
Master of Information Science and Technology, The University of Tokyo
Mar 2026Department of Mechano-Informatics, Graduate School of Information Science and Technology. Advisor: Prof. Hirokazu Takahashi.
Thesis research: Computational modeling of collective Bayesian inference and predictive coding. -
M.D., Kyoto University
Mar 2024Faculty of Medicine. Medical License, Japan (2024).
Experience
-
Research Assistant, The University of Tokyo
Apr 2024 – presentTakahashi & Shiramatsu Lab (Bio-Intelligence Lab).
-
Part-time Engineer, Preferred Networks, Inc.
Mar 2024 – Mar 2025Continued pretraining and reasoning preference optimization for Japanese medical language models.
-
Student Researcher, Kyoto University
Jul 2021 – Mar 2024Kamitani Laboratory.
Additional AI experience: Head of Artificial Intelligence, artics (Nov 2024 – Dec 2025); AI Researcher, Starley Inc. (Feb 2024 – May 2024).