Lähetä tietue sähköpostilla: Global remapping emerges as the mechanism for renewal of context-dependent behavior in a reinforcement learning model