Learning in the Cerebellar Circuit
The cerebellum is supposed to be involved in the acquisition of the dynamic models of the body and the environment. We propose biologically plausible models of network dynamics and plasticity in the cerebellum and investigate their performances in behavioral learning tasks, including arm and eye movements.

Compartmental Model of the Inferior Olive Neurons
The inferior olive is a small nucleus in the brain stem but it has been hypothesized to send the most critical signal for the learning in the cerebellum. Inferior olive neurons are characterized by their very low firing rates and strong electrotonic couplings. Based on the existing experimental data, we have constructed a compartmental model of the inferior olive neurons and are testing by simulations how the electrotonic coupling affect the information coding in the inferior olive.

Reinforcement Learning Models of the Basal Ganglia
The function of the basal ganglia has long remained enigmatic, but recently it has been suggested that they are specifically involved in evaluating the state of the body and the environment and accordingly selecting appropriate actions. We are exploring the roles of the basal ganglia in motion sequence and skill learning using reinforcement learning models that replicate experimental findings and by robotic experiments that simulate human sequential behaviors such as standing up.

Representation of Motion Sequences in the Frontal Cortex
As we learn to perform a particular sequence many times, we feel as if our body moves `on its own.' It seems that the same apparent movement can involve different representations and brain areas depending on the constraint in the task and the degree of experience. We are devising new behavioral experiments to distinguish what kind of representations are used in sequence learning, such as extrinsic, visuospatial coordinates and intrinsic, joint-angle coordinates, and network modeling to elucidate how such representations are acquired.

Dynamic Perception in the Visual Cortex
Neurons in the cerebral cortex are not just passively driven by the sensory input but actively select and interpret the sensory input based on the top-down or contextual information. We are exploring dynamic mechanisms of the cortical sensory processing using the lateral connections within each areas as well as reciprocal connections between different cortical areas. For example, we proposed a model of motion binding by diffusive lateral connection and confirmed the validity of the model by a psychophysical experiment.