A01Mutual reference approach bewteen artificail inteligence and neural correlates for investigation of value emergence
Our behavior is guided by the subjective values of objects. Physical and chemical information from the external environment is captured by sensory organs such as the eyes and mouth, and then is transformed to abstract value information. Previous functional MRI (magnetic resonance imaging) studies revealed involvement of the orbitofrontal cortex, cingulate cortex and striatum in value processing, however, what processes are required for transformation from object to value are unknown. In contrast, it is well known that object recognition is supported by the hierarchical visual stream from V1 to inferior temporal cortex. I hypothesize that value processing stream would also have hierarchical structure, and thus plan to use artificial neural network to model human value processing. In this project, I will compare value representations across brain regions, and furthermore, reveal unknown neural correlates by comparison with the informational structure represented by the model. During the fitting procedure, the artificial neural network will be modulated so that it mimics human value processing. I aim to develop the next generation of artificial intelligence by combining deep learning and functional MRI.