Fig. 1

a Network structure. b Reaching task overview. c Arm trajectory of the reaching task. The target was shown at 12 o’clock in the first 500 trials, and the target was moved to 9 o’clock in the next 500 trials. d Raster plot of the last trial in c (left). Cortico-striatal synaptic weights changing in 1000 trials (right). e Maze task overview. f Trajectories of the agent for each trial of maze t