Evidence of behavioral and neural interaction between task complexity and state-space uncertainty during reinforcement learning