Evolving choice hysteresis in reinforcement learning: comparing the adaptive value of positivity bias and gradual perseveration | Synapse