Active learning research emerged from the undergraduate STEM education communities of practice, some of whom identify as discipline-based education researchers (DBER). Consequently, current frameworks of active learning are largely inductive and based on emergent patterns observed in undergraduate teaching and learning. Alternatively, classic learning theories historically originate from the educational psychology community, which often takes a theory-driven, or deductive research approach. The broader transdisciplinary education research community is now struggling to reconcile the two. That is, how is a theory of active learning distinct from other theories of knowledge construction? We discuss the underpinnings of active learning in the geosciences, drawing upon extant literature from the educational psychology community on engagement. Based on Sinatra et al. engagement framework, we propose a model for active learning in the geosciences with four dimensions: behavioral, emotional, cognitive, and agentic. We then connect existing literature from the geoscience education community to the model to demonstrate the current gaps in our literature base and opportunities to move the active learning geoscience education research (GER) forward. We propose the following recommendations for future investigation of active learning in the geosciences: (1) connect future GER to our model of active learning in the geosciences; (2) measure more than content learning; (3) document research methods and outcomes with effect sizes to accumulate evidence; and (4) prioritize research on dimensions of active learning essential to the geosciences.