Reversal learning tasks are commonly used to test cognitive flexibility by allowing humans and experimental animals to learn one set of rules to gain rewards and then respond to changes in those rules. In this experiment, male and female Long-Evans rats were trained to discriminate distinct visual stimuli associated with different reward probabilities (a 2-alternative choice), requiring them to nosepoke the stimuli on a touchscreen. A range of reward likelihoods from probabilistic (i.e., 90:30, 90:10, 70:30) to deterministic (100:0) were administered. We explored how sex may affect performance on the following measures: probability correct (learning), latencies to nosepoke correct/incorrect stimuli (decision processing speed), trial initiation (attention), and reward collection (motivation). Across separate cohorts of rats, we found differences during the reversal phase mainly centered on decision processing speed, with females performing more slowly than males with longer initiation latencies. Attention measures (trial initiation) were also attenuated in females compared to males in both discrimination and reversal learning. Males failed to initiate more trials in 70:30 discrimination learning than in 90:30. [KD1] Importantly, overall learning (probability of correct choice and rewards received per session) did not differ by sex, suggesting all animals learned equivalently despite performance drawbacks. Our results provide some of the first fine-grained evidence for sex differences in stimulus-based reversal learning. Ongoing experiments are aimed at testing animals on more certain 90:10 and 100:0 reward probabilities. We predict there will be less pronounced sex differences on attention and processing speed as reward probability increases.