Variations in mean PISA scores have not been adequately explained to date, suggesting the limits of our current understanding of the relationship between educational practices and students' performance. In contrast to previous research that applies existing theories to explain observed variations, this study attempts to extend our existing theoretical horizon using PISA-derived data. We first introduce findings of PISA-Science data that run counter to the fundamental assumptions of both student-centered and teacher-centered learning theories; namely, countries having lower levels of students' initiative to design and carry out their own projects had higher scores. We then propose an alternative theory of learning (Type II learning) to explain this counterexample by rethinking the learning process at its philosophical and ontological depths. We conclude by noting a surprising paradox: the Type II learning made visible through PISA data appears to undermine the core premise of the OECD's whole approach to PISA itself.