Why a social partner can teach more than a recording can
A recording gives an animal sound; a group mate gives it a sound plus a context, a relationship and a reason to answer. A review of primate vocal learning argues that standard learning theory, built on isolated lab animals trained with artificial rewards, fits real primate vocal development poorly. Instead, the authors propose that young primates often learn by acquiring "scripts" — mental representations of how typical social events unfold — with group mates acting as the agents that cause events, even when nothing directly rewards the learner [4].
That framing helps explain why vocal convergence shows up where animals form stable social groups. In greater spear-nosed bats, call structure changed with the composition of the social group, and the authors argue that where forming stable groups matters, social learning of vocalizations should be favored [3]. The change was not the appearance of brand-new call types but subtle shifts within calls the bats already had [3] — a reminder that "learning a repertoire" can mean fine-tuning existing sounds rather than adding new ones.
What social vocal learning looks like in real animals
Killer whales offer one of the clearest natural cases. Juvenile males changed their call repertoires after a shift in social association, in what the researchers describe as an adventitious cross-socializing experiment — an unplanned grouping change that functioned like a test [5]. The finding matters because social dialects are rare among mammals, so the whales are an unusually visible case of group membership leaving a mark on vocal output [5].
Marmosets show a related pattern from the other direction: their vocal behavior is shaped by social feedback and by convergence toward, or divergence from, the calls of those around them [1]. In both species the signal is a change in vocal output tied to social company, not simply exposure to a soundscape.
How do you tell social learning from just hearing sounds?
The core experimental move is to break the link between hearing a sound and interacting with the animal that makes it. One approach isolates individuals and uses an automated setup so that vocalizations can be conditioned without a social partner or a live vocal model present [2]. In bats, this demonstrated vocal flexibility and vocal usage learning — the animals could learn when to use calls — under conditions stripped of social cues [2]. That kind of result is the control case: it shows what vocal change looks like when social influence is removed.
The complementary approach keeps the social partner but manipulates who is present. Work on marmosets explicitly aims to distinguish social learning from auditory exposure, using social feedback as the variable of interest [1]. If vocal output tracks the identity or response of a partner rather than merely the acoustic input, social learning is the better explanation. The bat isolation experiments and the marmoset social-feedback work are most informative read together: one establishes that vocal change can happen without social cues, the other tests whether social cues change vocal output beyond what hearing alone would predict [1][2].
What is still being pinned down
The evidence supports social influence on vocal output, but the strength of that claim varies by species and study. The primate review's "scripts" account is offered as an alternative to reinforcement-based learning theory rather than as a demonstrated mechanism [4], and the marmoset work is framed around examining the role of social feedback rather than settling it [1]. The killer whale case rests on an unplanned grouping change, which is suggestive but not a designed manipulation [5].
A practical implication follows for anyone designing such experiments: a study that only plays sounds cannot rule out social learning, and a study that only observes groups cannot rule out simple exposure. The informative designs are the ones that vary social access while holding the acoustic environment as steady as possible — the logic behind both the isolated automated setups and the social-feedback manipulations [1][2].
Sources used in this answer
Convergence and Divergence: Forces Shaping Marmoset Vocal and Social Interactions
Marmoset vocal behavior is shaped by social feedback and by vocal convergence and divergence with group mates, and the work uses methods intended to distinguish social learning from auditory exposure.
Volitional control of social vocalisations and vocal usage learning in bats
An automated setup that isolates individual bats shows they can be conditioned to control social vocalizations, demonstrating vocal flexibility and vocal usage learning without social or auditory cues from a partner.
Vocal learning by greater spear-nosed bats
In greater spear-nosed bats, call structure changed with social group composition, involving subtle shifts within existing call types rather than new call types, which the author argues favors social learning where stable groups form.
Socially scripted vocal learning in primates
A review of primate vocal learning argues that reinforcement-based learning theory fits real primate vocal development poorly and proposes that primates often learn by acquiring scripts of how social events typically unfold, with conspecifics as event-causing agents.
Evidence for vocal learning in juvenile male killer whales, Orcinus orca, from an adventitious cross-socializing experiment
Juvenile male killer whales changed their call repertoires after an unplanned shift in social association, providing evidence of vocal learning tied to social grouping in a species with social dialects.