Military organizations worldwide are aiming to mature artificially-intelligent agents (e.g., computer-based intelligent tutoring systems (ITS) and virtual humans) to lead, support, and tailor training to the needs of individuals and small units. Goals for ITS are: to match or exceed the learning effect of expert human tutors; reduce the cost of authoring, delivering, and managing training; lower entry skills needed to author ITS; and develop quality standards, accessibility, and flexibility for the learner. This paper focuses on improving learning effect and explores how learning gains (e.g., knowledge and skill acquisition, and enhanced performance) might be realized in ITS for tutoring both individual and small unit tactical tasks. To this end, an adaptive tutoring learning effect chain (ATLEC) for both individual and team learning is put forth. Originally developed by Sottilare (2012), ATLEC for individual tutoring models the relationships of concepts for learner data (behavioral, physiological, historical, and trait), learner states (cognitive and affective), instructional strategy selection, and learning gains. This model is a key methodology incorporated within the Generalized Intelligent Framework for Tutoring (GIFT). This paper expands the ATLEC model to include small unit tutoring and an expanded array of learning gains (e.g., accelerated learning and enhanced retention). A key to learning gains in human tutoring is the ability of the tutor to detect and interpret behavioral cues from the learner to aid them in assessing the learner's cognitive (e.g., engagement) and affective (anxiety, frustration, boredom and confusion) states in order to optimally select their next instructional strategy. ITS must use other means (e.g., behavioral and physiological sensors) to detect and interpret learner states which is an advantage over human tutors. The product of this paper will be a model of learning effect that can be used to drive standards and the development of ITS for training and education.