Over the past several decades, the geoscience community has made great strides toward discovering the kinematics and dynamical processes associated with plate tectonics. Convection of the Earth’s silicate mantle is qualitatively regarded as the primary driving force for tectonic processes; however, the manner in which this integrated system operates remains fundamentally unclear. A critical obstacle is our lack of understanding the first-order, dynamical nature of mantle convection. Seismological observations are increasingly finding evidence for compositional and thermal heterogeneity over multiple length scales within Earth’s mantle. However, the manner in which this heterogeneity affects mantle dynamics remains poorly understood, and several competing hypotheses of large-scale, thermochemical mantle convection currently exist. It is important to distinguish which, if any, of these conceptual models are representative of the actual Earth because each has significantly different consequences toward our understanding of heat and mass transport, thermal and chemical evolution, and the driving forces that generate plate tectonics. Numerical modeling of mantle convection, combined with seismic, geochemical, and geologic observations, provides a powerful tool to explore the dynamical feasibility of particular hypotheses and to provide observational predictions that can be tested by geophysical methods. Here, I will review recent progress toward discovering the nature of large-scale mantle convection within Earth’s interior by combining geodynamical modeling with geochemical and seismological observations of compositional heterogeneity within the mantle.