In our R&D Item 1, we develop technologies to interpret our mental states through brain and autonomic nervous activities. That will broaden our ways of non-verbal communication.
The primary task of our R&D Item 1 is to instantiate sensing technologies worthy of use in everyday communication. For instance, fMRI offers rich information yet is costly and time-consuming; polygraphs are handy yet are far from precise in decoding one’s mental state. Interpretation of one’s mental state through electroencephalography (EEG) has yet to be established.

Here, our R&D Item 1 aims to develop precise and multi-dimensional mind sensing based on brain and autonomic nervous activities combined with behavioral data. Our goal is to offer a reliable mind-interpretation method suitable for everyday use.

1. Successfully decoded animals' mental states with high accuracy from brain activity of deep and surface areas using a Transformer model
2. Successfully visualized individual traits by focusing on fluctuations in EEG phase-synchronization networks
3. Successfully formulated a mathematical index of shared emotional resonance based on empathic EEG activity recorded from multiple individuals

Outcome 1: Using a Transformer model, a machine-learning approach originally developed for natural language processing, we successfully decoded mental states from brain activity. In a study using monkeys, we analyzed neural activity simultaneously recorded from deep and surface brain areas, identified the brain regions in which the animals’ mental states were reflected, and successfully decoded positive and negative emotional states. In experiments using mice, we showed that EEG activity recorded from multiple brain regions enabled more accurate discrimination of aversive visceral pain states than conventional methods.

Outcome 2: We developed an index that accurately quantifies fluctuations, or metastability, in EEG phase-synchronization networks. Using this index, we identified differences in EEG metastability between typically developing individuals and individuals with ASD, and successfully visualized individual traits.

Outcome 3: We quantified changes in EEG activity that occur when empathy arises among multiple individuals and developed an algorithm for detecting “shared emotional resonance.” We also presented an exhibit related to this achievement at Moonshot Park during Expo 2025 Osaka, Kansai, Japan.

So far, we are conducting researches to unravel the relationship between neuronal activities and one’s mental state by combining animal and human experiments.