Research Interests

I am broadly interested in how neural representation and computation govern cognition and affect. My goal is to uncover the computational mechanisms shared across biological and artificial systems. Previously, I was an fMRI guy who developed fMRI-based predictive models of pain with human brain representations. Currently, I am investigating the sense of agency in an artificial agent by using the PV-RNN algorithm, which combines predictive coding with variational recurrent neural networks. My focus is on how the sense of agency emerges from hierarchical generative models linking action generation and sensory prediction.
You can find my CV here.

keywords: computational neuroscience; bayesian inference; generative models; predictive coding; recurrent neural networks, reinforcement learning

Publications

Main papers are highlighted. My publications are also listed on Google Scholar.

6

6 Total citations

6 Recent citations

5.82 Field Citation Ratio

n/a Relative Citation Ratio

This study shows a comprehensive overview about modeling targets and options of brain imaging-based pain predictive models through a literature review and benchmark analysis.

5

5 Total citations

5 Recent citations

4.85 Field Citation Ratio

n/a Relative Citation Ratio

As a commentary paper for Hoeppli et al. (2022), this paper examined that a multiple regression model including brain imaging, sociodemographic, and psychological measures could predict individual differences in pain self-report. We also identified brain regions associated with these differences using fMRI data.

23

23 Total citations

19 Recent citations

6.34 Field Citation Ratio

1.48 Relative Citation Ratio

This study explored dynamic changes in functional brain networks during pain in the mouth, which provides new insights into how dynamic interactions between brain systems organize and modulate the experience of pain.

66

66 Total citations

52 Recent citations

21 Field Citation Ratio

4.18 Relative Citation Ratio

This study identified different brain representations of pain processing across brain regions. Some brain regions such as vmPFC show high variability across individuals, whereas other regions such as SMC show stable patterns.

Projects

Research works are highlighted.

Can artificial neural networks replicate human-like shifts in sense of agency in response to sensorimotor conflicts? If so, are the shifts in sense of agency asymmetric between gaining and losing controllability? We address this question through PV-RNN and robot arm data.

Skills: Python, Numpy, PyTorch, Hydra, Matplotlib, MySQL, Docker

code

RL-Codebook is a structured reinforcement learning repository that connects mathematical theory to runnable code, offering clean implementations of classic and modern RL algorithms (e.g., Q-learning and PPO) with reproducible experiments and systematic evaluation.

Skills: Python, Numpy, PyTorch, PyYAML, Gymnasium

Thanks to Jon Barron for providing the template of this website.