Beyond facilitation and inhibition: a configurational mechanism study of cognitive transitions in human–AI collaboration

Ongoing debates in higher education regarding whether artificial intelligence should be further integrated or deliberately constrained call for empirical research that offers a more explanatory analytical framework. However, existing studies on the human-AI collaboration (HAC) paradox are largely grounded in a binary logic of facilitation versus inhibition, leaving the dynamic mechanisms underlying complex cognitive processes insufficiently explored. To address this issue, this study adopts a dialectical perspective on explicit and tacit knowledge in knowledge conversion, and systematically examines the mechanisms of cognitive conflict, regulation, and equilibrium underlying cognitive transitions in HAC. Drawing on data collected from 316 participants in an authentic instructional context, this study constructs a configurational model incorporating social interaction (SI), tacit knowledge acquisition (TKA), internalization (I), self-motivation (SM), and trust in AI (TiAI), and employs fuzzy-set qualitative comparative analysis (fsQCA) to examine multiple equifinal pathways leading to higher-level cognition. The findings identify two distinct types of driving mechanisms underlying cognitive transitions: a high human-centered engagement pathway in the absence of AI, and a compensatory pathway in which AI offsets deficiencies in human-centered conditions. These results suggest that cognitive transitions emerge not from the linear effect of isolated factors but from the dynamic counterbalancing and configuration of psychological characteristics and technological conditions. In this specific educational context, AI functions as a mediating and compensatory agent that mitigates cognitive imbalance. Methodologically, this study demonstrates the logical compatibility between fsQCA and knowledge spiral; theoretically, it extends the explanatory boundaries of the HAC paradox; and practically, it provides evidence-based guidance for the structural deployment of AI support in higher education.