Sequential sampling without comparison to boundary through model-free reinforcement learning | Synapse