B03 Large-scale neural basis for generative model acquisition in active inference in non-human primates
We aim to investigate how the updating generative models in the primate cerebral cortex for predictive information processing, in order to develop and verify a unified theory regarding the information processing involved in the prediction and behavior of animals.
The animals constantly makes predictions and adapts their behavior by generalizing to and predicting stimuli that are inputted every moment. It is believed that during this process, self-organizing circuit reorganization occurs in the brain, and frameworks such as “predictive coding” and “active inference” have been proposed as universal computational models. However, their implementation in the brain are still unclear, particularly with regard to how circuit updates before and after learning occur at the whole-brain level. In this study, we will simultaneously measure the neural activity in the primate cerebral cortex during sensory prediction and learning tasks, and widely share this data within the project in order to achieve a unified understanding of brain information processing through theoretical verification and cross-species comparison.
Principal Investigator: Misako Komatsu
Specially Appointed Associate Professor, Institute of Innovative Research, Tokyo Institute of Technology
Collaborator: Masafumi Takaji
Researcher, RIKEN Center for Brain Science