Metacognition has been a subject of considerable interest in school settings, particularly its implications on learning and performance in mathematics. While metacognition has been widely studied as a multi-faceted construct comprising of metacognitive knowledge, regulation and experiences in various combinations, few have examined these three facets within a single study. Based on a mixed-method design, the validity and empirical relationships among the three dominant components of metacognition were investigated using a person- and variable-centred approach. Convergent and discriminant validity were supported in which robust relationships were found among the three components, but some aspects of metacognition differed in their associations with mathematics achievement. Expanding on the quantitative results, student interviews and classroom data were collected to deepen the understanding of metacognitive experiences, and students' learning of mathematics. Collectively, the triangulated results showed that low achieving students in mathematics tended to rely more on affective learning approaches when engaging with mathematics tasks. By contrast, students who were high achieving tended to employ cognitive information processing strategies. Nonetheless, metacognitive experiences, with respect to feelings and emotions, are important and should not be overlooked regardless of whether students are high or low achieving in mathematics. The results suggest that the path towards more inclusive, active participation of low achieving students in the learning of mathematics could be better mitigated through the heart (i.e., the affect) rather than wholly via the brain (i.e., cognition).