Graphic showing Pathmind determining if an action should be taken again based on observations and the reward score