Individual Contributions as Intrinsic Exploration Scaffolds for Multi-agent Reinforcement Learning | Synapse