Development of New Food-Sharing Relationships in Vampire Bats.

Some nonhuman animals form adaptive long-term cooperative relationships with nonkin that seem analogous in form and function to human friendship [1-4]. However, it remains unclear how these bonds initially form, especially when they entail investments of time and energy. Theory suggests individuals can reduce the risk of exploitation by initially spreading out smaller cooperative investments across time [e.g., 5] or partners [6], then gradually escalating investments in more cooperative partnerships [7]. Despite its intuitive appeal, this raising-the-stakes model [7] has gained surprisingly scarce empirical support. Although human strangers do "raise the stakes" when making bids in cooperation games [8], there has been no clear evidence for raising the stakes during formation of social bonds in nature. Existing studies are limited to cooperative interactions with severe power asymmetries (e.g., the cleaner-client fish mutualism [9]) or snapshots of a single behavior within established relationships (grooming in primates [10-13]). Raising the stakes during relationship formation might involve escalating to more costly behaviors. For example, individuals could "test the waters" by first clustering for warmth (no cost), then conditionally grooming (low cost), and eventually providing coalitionary support (high cost). Detecting such a pattern requires introducing random strangers and measuring the emergence of natural helping behaviors that vary in costs. We performed this test by tracking the emergence of social grooming and regurgitated food donations among previously unfamiliar captive vampire bats (Desmodus rotundus) over 15 months. We found compelling evidence that vampire bats selectively escalate low-cost grooming before developing higher-cost food-sharing relationships.

A method for rapid testing of social learning in vampire bats.

Designing experiments on social learning using an untested behaviour or species requires baseline knowledge of how the animals will perform. We conducted a pilot study of a procedure for rapidly testing social learning in the highly social common vampire bat (Desmodus rotundus) using a simple maze. To create demonstrators, we allowed captive bats to learn to exit a three-dimensional maze, which reunited them with their colony as a reward. We then filmed naive bats in the same maze, comparing their ability to exit the maze before, during and after the addition of a trained demonstrator. The presence of a demonstrator increased the exit rates of naive bats, presumably by attracting the attention of the naive bats to the maze exit. Four of the five naive bats that exited in the presence of a demonstrator retained the ability to exit without the demonstrator. No naive bat exited during trials without a potential demonstrator present. This experimental procedure appears to be a promising approach for efficient tests of social learning in vampire bats because maze difficulty can be manipulated to adjust learning rates and each trial requires only 15 min.