<![CDATA[Intelligent Tutoring Systems (ITS) have emerged as powerful AI-driven learning technologies capable of delivering personalized instruction at scale. This study provides a systematic review of literature published between 2020 and 2025 to examine the cognitive mechanisms through which ITS improve learning outcomes. A total of 20 empirical and high-impact studies were analyzed using predefined inclusion criteria focusing on adaptive feedback, knowledge tracing, scaffolding, metacognition, and motivational design. The findings reveal that ITS consistently produce small-to-moderate learning gains (d ≈ 0.30–0.70), with feedback and knowledge tracing identified as the most dominant and reliable mechanisms. Metacognitive supports demonstrate the strongest influence on deep learning and self-regulated learning, while scaffolding contributes to conceptual understanding and transfer. Motivation and engagement features, although less frequently studied, enhance persistence and learning intensity. The review also highlights emerging trends, including the integration of learning analytics, affective computing, and hybrid ITS architectures. However, challenges remain in model interpretability, standardization of outcomes, and alignment with pedagogical practices. This study contributes a synthesized framework linking cognitive mechanisms to learning effectiveness and provides directions for future research to develop more explainable, scalable, and pedagogically grounded ITS.]]>
Copyrights © 2026
