A process-based approach to hominin taxonomy provides new perspectives on hominin speciation.

A longstanding debate in hominin taxonomy is that between "lumpers" and "splitters." We argue that both approaches assume an unrealistically static model of speciation. Speciation is an extended process, of which fossils provide a record. Fossils should be interpreted in a more dynamic framework than is the norm. We introduce the process-based approach (PBA), in which we suggest that "splitters" recognize and name units at an earlier stage of speciation than "lumpers" do. The "determinants" of speciation can control the rate at which population isolates form, or the rate at which these complete the speciation process, or both. Embedded in the PBA, differences between existing lumped and split taxonomies are a heuristic tool to study these processes. We apply the PBA to show that not all hominin populations reached later stages of the speciation process and that populations have a disproportionate likelihood of doing so from ∼3.1 to ∼1.5 Ma. We outline and discuss resulting new research questions.