Katsushi Kagaya

Research Interests

animal physiology, animal behavior, neurophysiology, soft robotics, data science

Overview

I am interested in understanding how animals behave as a combination of diverse environments, bodies, and brains, with a particular focus on the mechanisms that generate "flexibility" and "spontaneity". My field mainly targets the functions of individual animals and is known as animal physiology. However, this issue arises in several interdisciplinary fields, including soft robotics. I have focused on studying crustaceans such as crayfish and mantis shrimp. My approach includes three elements:

1. Behavioral physiological methods based on measurements of nerve cells, muscle activity, and body movements
2. Statistical inference methods that construct and evaluate models based on these measurements
3. Soft robotics methods that build and evaluate actual machines

I also look forward to the prospect of converting animal mechanisms into human technology (biomimetics).

Mechanisms of Flexible Elastic Deformation

The interaction between the body of an individual animal and its external environment, acquired through evolution, generates remarkable diversity. The brain, body, and environment are not separated but are coupled flexibly (from the viewpoint of embodiment). How can we systematize and understand this diversity? We have been working on this problem, targeting arthropods (crayfish, mantis shrimp, and trap-jaw ants) and mollusks (octopuses), which can realize deformational movements using their exoskeletons. For example, we have shown that individual crayfish consistently produce larger or smaller claws when constructing their exoskeletons. Our recent research has focused on soft robotics, aiming for biomimetics (Inoue+2021, Ito+2021, Kagaya+2022).

Neural Basis of Spontaneity

Animals can initiate voluntary and spontaneous movements without explicit external sensory stimuli. What are the neural mechanisms underlying the initiation of spontaneous behavior? To answer this question, we have been conducting research using the crayfish as our subject. We have discovered that, as with primates, there is a "readiness potential" that precedes the onset of spontaneous movements in crayfish (Kagaya & Takahata, 2010). We have proposed that the generation of the readiness potential is not due to the spontaneous activity of specific, single neurons, but rather the operation of whole recurrent neural circuits (Kagaya & Takahata, 2011). Furthermore, we has released a preprint that explains how this circuitry can be viewed as an integrate and fire type, high dimensional dynamical system showing avalanche-like behavior explained as a self-organized critical phenomenon (Kagaya, et al., 2022).

Interdisciplinary Data-Driven Research

- Behavioral Biology (Crabs) [Harada+2020]

We proposed a method to predict the individuality of behavioral characteristics of each animal based on the viewpoint of prediction (generalization error) using hierarchical and non-hierarchical stat models. The question was whether there was individuality in the behavior of the decorator crab, which creates a cap by processing a sponge. The analysis showed that each crab had a consistent, individual tendency to create larger or smaller caps. In evaluating singular models with structured hierarchical models, statistically valid conditions do not hold, and the AIC based on the maximum likelihood method cannot be used. Therefore, we used WAIC based on the statistical learning theory by Dr. Watanabe (cf. WAIC and WBIC).

- Epidemiology (Humans) [Mizumoto+2020a, b, c]

Scientific exploration is conducted based on data. The outbreak of the new coronavirus since the spring of 2020 has not only been uncertain, but has also widely recognized the importance of obtaining information from data in unknown situations. We conducted collaborative research on data analysis in mathematical epidemiology with Dr. Kenji Mizumoto. In particular, we reported a statistical inference of the proportion of asymptomatic individuals based on data from the Diamond Princess cruise ship (Mizumoto+2020).

- Chromosome Biology (Human Cells) [Kagaya+2020]  

We statistically inferred the influence on the phenotype of the cell cycle when sister X chromosomes were fused using a cell line that emits fluorescence when chromosomes are fused, developed by Dr. Makoto T. Hayashi. We concluded that the generation of a tiny nucleus structure resulted in a longer interphase and instability when fusion was induced. Applying hierarchical Bayesian modeling and its evaluation method, we considered whether groups of cells derived from the same cell should be considered as one group from the predictive point of view.

〒113-0033 Tokyo, Bunkyo-ku, Hongo 5-24-5, Nagase Hongo Building 8F

※ Transferred to Kohei Nakajima's lab from April 2020.