Prevalent human brain analyses have succeeded in measuring the brain activities with perception, thought and motion. However a lot of researches have never estimated what one percepts or thinks from one's brain activities. Decoding techniques are key methods in order to "read one's mind" and control computers or robots through the brain activities.

We develop the decoding techniques reading the states of mind by non-invasive brain imaging, fMRI or MEG. Especially, we aim to reveal subjective perceptual and cognitive contents.

<Link>

To understand the dynamic mechanisms of human brain activities, it is necessary to develop brain imaging techniques with high spatial and temporal resolution. We research the hierarchical Bayesian estimation which integrats high spatial measurement fMRI or NIRS and high temporal measurement MEG or EEG.
Moreover, we develop Brain Network Interfaces in order to control computers or robots by human brain activities.

<Link>

To understand the brain function from the viewpoint of information processing, one must try to reconstruct the same function artificially. That is the meaning of creating the brain. It may not be sufficient to reconstruct solely the brain but also the body and environment. That is the meaning of creating the human being. That is where humanoid robots become a tool for brain research. For all research in humanoid robotics to progress, hardware and software development, learning tools etc (?) are needed.

<Link>

Using non-invasive neuroimaging methods (e.g., fMRI and MEG), we investigate learning mechanisms in the human brain underlying flexible adaptation to environments. Various types of internal representations are known to exit in the brain. These representations are thought to dynamically change as learning proceeds. We aim to reveal the change using appropriate experimental paradigms, functional brain imaging and decoding of neural activity.

<Link1>�@<Link2>