B04 Hidehiko Takahashi

B04　Understanding the Predictive Mechanisms of the Human Brain and its aberration

We will examine how predictive information processing suggested by theory and animal experiments is carried out in the human brain, and what kind of alterations are observed in psychiatric disorders. In addition to standard 3T fMRI, we will use high spatio-temporal resolution measurement of brain activity using cortical EEG. fMRI is difficult to separate bottom-up and top-down predictive information processing in the brain. For this reason, we use a deep learning model without a prediction mechanism, contrast the brain signals with the internal signals of the deep learning model, separate the prediction signals, and identify the neural basis of information processing associated with the prediction mechanism. The same method will be applied to patients with schizophrenia and developmental disorders to identify the neural basis of abnormalities in the predictive mechanism of psychiatric symptoms. Brain information data from patients at rest or during the prediction task will be shared with theoretical and animal experimental researchers to validate the pathological model as an alteration of predictive information processing proposed by the theory. Brain information data before and after symptom recovery will also be examined to clarify the mechanism of treatment effect and recovery from the viewpoint of predictive information processing. With the cooperation of the Department of Neurosurgery at Tokyo Medical and Dental University, we will utilize the higher spatial and temporal resolution than normal EEG obtained with stereotactic intracranial EEG to measure brain information during prediction tasks. This data, along with data from other animal species, will be shared with theoretical researchers to validate and extend prediction theory.