Action decision congruence between human and deep reinforcement learning agents during a coordinated action task.