Why learning socially in one area doesn't mean an animal can do it in another
Social learning is best understood as a set of causal routes by which a trait can pass from one individual to another, not as a single mental module [2]. An animal that learns its song from adults is using social information, but that tells us little about whether it will also learn which foods are safe by watching others. The mechanisms, cues, and timing involved differ enough that competence in one domain does not imply competence in another [2][5].
This matters for how we interpret animal behavior. A bird that copies a song may still rely on individual trial and error to find food, and a shark that follows a familiar migratory route may not learn predator recognition socially [4][2]. The domains are studied separately because the learning problems are different: song has a sensitive period and a specific tutor model, while route learning involves spatial memory and repeated travel [5][4].
Which domains do show clear social learning, and what does that look like
Song learning is one of the clearest cases. In birds, the development of species-typical song has been shown to depend critically on social learning from tutors [2]. This is a domain with strong evidence because the outcome—the song itself—is measurable and the social input can be manipulated.
Predator recognition is another domain where social learning has been documented. Birds can learn to recognize predators through social means, which suggests that fear of specific threats can be transmitted rather than discovered alone [2]. Route and refuge learning also appears in some groups: sharks and rays appear to socially learn refuges and migratory routes, though the evidence here is observational and the mechanisms are less well characterized [4]. Foraging and migration routes are also discussed as domains where social learning may shape behavior across generations [3].
What we still can't say about generalization
The honest answer is that no study directly tests whether an animal that is a strong social learner in one domain is also a strong social learner in another. The evidence is assembled from separate literatures on song, predator recognition, routes, and foraging, each with its own methods and species [2][3][4]. That makes it difficult to compare domains directly or to claim that a single underlying capacity generalizes.
There is also the question of mechanism. Social learning can occur through several different routes, and the route used in song learning may not be the same one used in predator recognition or route following [2]. Until the same species is tested across multiple domains with comparable designs, the generalization question remains open. What the current evidence does support is that social learning is widespread across domains, not that it is unified across them [2][5].
Sources used in this answer
Understanding the multiple factors governing social learning and the diffusion of innovations
Reviews social learning in wild animals and the diffusion of innovations but does not examine generalization across specific behavioral domains.
Social learning: an introduction to mechanisms, methods, and models
Introduces social learning mechanisms and discusses how social learning influences traits such as song and predator recognition, supporting domain-specific social learning.
A deepening understanding of animal culture suggests lessons for conservation
Synthesizes evidence that social learning shapes foraging, migration routes, and mate choice, and that population declines can lead to loss of song.
Shark and ray social lives: form, function, and ecological significance of associations and grouping
Reports that sharks and rays appear to socially learn refuges and migratory routes, indicating social learning in spatial domains.
Animal cognition and animal behaviour
Discusses social learning mechanisms alongside behaviors such as song learning and route learning, treating them as distinct learning problems.