In this commentary, we respond to Jaakkola, (2025), who raised several concerns regarding our findings on vocal labeling in marmosets (Oren et al. Science, 385, 996–1003, 2024). We address each point in turn, clarifying that marmosets use socially learned, arbitrary vocal labels for specific conspecifics – labels that are neither imitations nor acoustically derived from the receiver’s own calls. We show that classifier models trained on individual callers reliably identify the intended receiver, while cross-caller models reveal family-specific label conventions. We also provide evidence that vocal accommodation does not account for the observed behavior, and we argue that the consistent, receiver-specific use of labels reflects a stable mapping between individuals and calls – indicative of an internal representation of identity. Taken together, our findings support cognitively controlled social calling in marmosets and suggest a primate precursor to name-like referential communication.
