Rationale The widespread deficits in cognitive flexibility observed across psychiatric disorders call for improved rodent tests to understand the biology of cognitive flexibility and development of better psychotherapeutics. Current reversal learning paradigms have a forced-choice setup that challenges the interpretation of results. Objectives We aimed at developing a free-choice reversal learning test, where images are presented sequentially and animals are free to move, to enable investigation of the cognitive sub-processes that occur during reversal. Methods Behavior in female C57BL/6JOlaHsd mice was characterized using chronic fluoxetine as a reference compound. Additional tests were included to support the interpretation of results and exclude confounding pharmacological effects. Behaviors in vehicle-treated mice were furthermore analyzed for relatedness to deepen the understanding of parameters measured. Results We found that exploitation of the previously rewarded image was independent of exploration and acquisition of the new reward contingency and could be differentially modulated by fluoxetine, supporting recent theories that these processes are not mutually exclusive. Specifically, fluoxetine reduced mistake rate, premature and perseverative responses, and promoted conservative strategies during reversal without affecting hit rate. These effects appeared to be most prominent during the late stage of reversal learning, where accuracy was above chance level. Analysis of behaviors in vehicle-treated mice suggested that exploitation was related to an impulsive-like deficit in response inhibition, while exploration was more related to motivation. Conclusions This new schedule was feasible, easy to implement, and can provide a deeper understanding of the cognitive sub-processes during reversal.