University and Critical Thinking in the Digital Age: A Systematic Review of Pedagogical Practices for Student Intellectual Autonomy
TL;DR
This systematic review synthesises empirical and conceptual evidence on university pedagogical practices that cultivate critical thinking and, through it, student intellectual autonomy in digitally mediated learning environments and identifies gaps concerning long-term transfer, equity, and the assessment of autonomy in AI-saturated contexts.
Abstract
Background. The digital age has reconfigured the epistemic environment of higher education. Information abundance, algorithmic curation, misinformation, and the rapid diffusion of generative artificial intelligence (AI) have made critical thinking a defining graduate attribute, yet they simultaneously threaten the intellectual autonomy that critical thinking is meant to secure. Objectives. This systematic review synthesises empirical and conceptual evidence on university pedagogical practices that cultivate critical thinking and, through it, student intellectual autonomy in digitally mediated learning environments. Methods. Following the PRISMA 2020 guidelines, a structured search of five databases (Scopus, Web of Science, ERIC, Education Source, and Scientific Electronic Library Online) was conducted for records published between January 2015 and February 2026. After de-duplication, title–abstract screening, and full-text eligibility assessment against pre-registered criteria, 68 studies were retained and analysed through a thematic synthesis. Results. Five families of pedagogical practice were associated with gains in critical thinking and autonomy: (i) active inquiry-based instruction (problem-, project-, and case-based learning), (ii) argumentation, dialogue, and Socratic questioning, (iii) blended and flipped designs that reallocate class time to higher-order tasks, (iv) metacognitive and self-regulatory scaffolding, and (v) critical digital and AI literacy. Reported effects on critical thinking were moderate to large (standardised mean differences typically between 0.55 and 1.08), but the strength and durability of gains depended on the quality of dialogic facilitation, the presence of explicit reflection, and the deliberate design of student agency. A recurrent autonomy paradox emerged: the same digital and AI tools that can scaffold reasoning can, when used uncritically, foster cognitive offloading and dependence that erode autonomous thought. Conclusions. Intellectual autonomy is not a by-product of technology but of intentional pedagogy. The review proposes a set of design principles and identifies gaps concerning long-term transfer, equity, and the assessment of autonomy in AI-saturated contexts.