You can't surpress it because that's what reinforcement learning is.