The frontal cortex plays critical roles in learning and decision-making, particularly in representing reward-related variables such as reward magnitude and probability. Previous studies have identified frontal neural representations of these variables in expertly trained animals. However, it remains unknown how these representations emerge during learning from naive to expert. To address this gap, we use 2-photon microscopy to longitudinally record the activity of the same population of neurons in the dorsomedial frontal cortex. We record from head-fixed mice learning to perform Pavlovian or operant visually-guided behavioral tasks that involve parametric manipulation of reward magnitude or its probability. Our analyses comprise two levels: firstly, we examine the responses of the same neurons tracked over days and secondly, we examine the same neural populations throughout the entire learning process. This approach enables us to determine whether distinct neurons and neural populations represent different reward-related variables and how this representation evolves during learning. Our preliminary analysis suggests that neural population responses to reward-predicting stimuli emerge during learning and become relatively stable in expert performance, with only subtle day-to-day changes. In summary, this study will provide novel insights into frontal cortical representation of reward, and how plasticity in such representations might underlie reward learning.