Behavioral flexibility enables us to tailor actions to the ever-changing demands and constraints of our environment. Different regions of the prefrontal cortex, such as the orbitofrontal cortex (OFC) and the medial prefrontal cortex (mPFC), are known to be differentially involved; yet, their complementary roles in behavioral flexibility remain unclear. In order to dissect their relative contributions, we developed a reversal learning task using Freibox, a custom-built, open-source platform for behavioral experiments. Reversal learning performance in this task depends on the OFC, which we demonstrated by optogenetically inhibiting the OFC around reward delivery. In order to differentiate the OFC’s involvement from that of other prefrontal areas, we expanded the task into a serial reversal learning paradigm, allowing for three intrasession reversals. We first characterized baseline performance during the task and found that mice consistently made more errors during the first reversal, compared to subsequent ones, indicating a switch between different cognitive processes across reversals. Next, we performed one-photon calcium imaging, revealing correlates of task aspects such as trial initiation, licking response and post-reward processing across all four blocks in both OFC and mPFC. Stochastic rewards in particular revealed prominent reward anticipation and feedback processing in both prefrontal structures, suggesting that both OFC and mPFC contribute to the flexible decision-making necessary in the task. Based on our optogenetic inhibition experiment and the literature, we hypothesize that the OFC is particularly involved in early reversals, whereas the mPFC might be more involved in late reversals, which seem to require different cognitive strategies.