Integrating task objectives for improved performance of preference-based reinforcement learning in different tasks | Synapse