Reinforcement learning deficits in people with schizophrenia persist after extended trials