Deep learning refers to an educational approach that prioritizes conceptual comprehension and the development of higher order thinking skills, as opposed to surface learning which emphasizes memorization of facts. In this context, mental acts as described by Harel serve as crucial internal cognitive processes, including comparing, relating, generalizing, and abstracting, which support the meaningful construction of knowledge. This study applies a systematic review of literature, drawing on sources from international journals, nationally accredited journals, academic books, and conference proceedings. The data were examined through thematic analysis to identify the forms of mental acts, the instructional strategies that facilitate their emergence, and their influence on deep learning. The review reveals that while mental acts play an important role in fostering critical, creative, and metacognitive thinking, much of the existing research still concentrates on instructional strategies, with limited attention to affective and metacognitive dimensions. This article contributes a conceptual synthesis that situates mental acts as the central element of deep learning and proposes an integrated framework to strengthen educational practices for the twenty first century.