Research
Our laboratory aims to elucidate the mechanisms of subjective experience through mathematical theories. We broadly categorize the problems of consciousness into three pillars: 1) the quality of consciousness, 2) the levels of consciousness, and 3) the location of consciousness. We address fundamental questions such as: What determines the qualitative difference between visual and auditory experiences? Why does consciousness fade during deep sleep? Our primary working hypothesis is Integrated Information Theory (IIT), which posits that the essence of consciousness lies in "intrinsic information". While we utilize the framework of IIT as a starting point, our primary focus is to overcome the theoretical and computational limitations that currently hinder the scientific testing of such mathematical theories.
To establish a rigorous basis for experimental verification, we tackle two critical hurdles. The first is the empirical challenge: How can we quantify subjective experiences as mathematical objects in psychophysical experiments? To address this, we propose a structural approach that characterizes the relational geometry of subjective experience, which we term "qualia structures." The second is the mathematical and computational challenge: How should we practically estimate theoretical quantities (e.g., integrated information) in large-scale neural systems? To handle this complexity, we employ tools from information geometry, dynamical systems theory, network control theory, and stochastic thermodynamics. These methods allow us to characterize causal relationships in the brain and develop practical algorithms for computing information-theoretic measures that were previously infeasible.
By combining these mathematical methodologies with experimental collaborations, we aim to bridge the gap between theory and data. For example, we analyze neural activity during sleep and anesthesia to verify if theoretical measures correctly track changes in consciousness levels, and we investigate whether the mathematical structure of subjective experiences correlates with psychophysical data. Ultimately, our goal is to build an empirical foundation for testing not only IIT but other consciousness theories in general. By facilitating fair comparisons among competing hypotheses, we aim to advance beyond current theoretical limitations and bridge the gap between subjective experience and neural activity.
Below, we introduce specific research projects addressing these topics.
Qualia structure paradigm
We quantify the quality of consciousness, or qualia, based on the relational structure of subjective experiences. By obtaining qualia structures from psychological experiments, we can compare them across individuals or relate them to the relational structures of neural activity.
Key Publications
Levels of consciousness
In addition to the standard network theory approach, we have recently applied thermodynamics to characterize irreversibility of dynamics, or the thermodynamic "cost", relating this quantity to the difference between conscious and unconscious states.
Key Publications
Location of consciousness
Experimental evidence suggests that both feed-forward and feedback processing are necessary for the brain to generate a conscious experience. To identify the location of consciousness in the brain, we are trying to locate a bidirectionally connected network core.
Key Publications
Control theory for revealing causal relations
A control theory framework can be useful for revealing causal relations in brain networks. Recently, we have proposed practical methods for quantifying controllability, controllable directions, and control costs in neural systems.
Key Publications
Integrated Information Theory (IIT)
We are working on the theoretical development and experimental verification of Integrated Information Theory (IIT). Specifically, we are developing practical algorithms to quantify IIT-based quantities, such as integrated information and the "complex" (informational core) in real neural data.
