Personalized learning is the idea of customizing an educational experience for each learner by accommodating their unique attributes like knowledge level, preference for learning, behavior, as well as emotional state. Supporting personalized learning are accurate and changing models of each learner. This paper presents a comparison of five approaches to learner modeling used in personalized e-learning. The comparison explains how each learner model and its components relate to cognitive, behavioral, affective, profile-situated and learning style attributes. The paper illustrates the methods used to develop and update the models while demonstrating the different degrees of complexity with respect to learner diversity. While recent approaches have made some strides toward understanding the multiple dimensions of learners, they have not yet fully and adequately demonstrated a model’s ability to be adaptive and emotionally sensitive to learners. The results imply that there is a need for more complete learner modeling frameworks that leverage new present day AI-mediated personalization techniques in order to advance engagement, support, and learning outcomes.