The common vampire bat, Desmodus rotundus, feeds exclusively on blood. Not occasionally — entirely. It has no metabolic fallback, no fat reserves substantial enough to bridge a gap, no ability to shift to insects or fruit. A bat that fails to obtain a blood meal for roughly sixty hours will die from starvation. In practice, this means a bat that misses two consecutive nights is in serious danger.
Blood feeding sounds reliable, but it isn't. Cattle and horses move; pastures change; bats are small and can be displaced by other animals, by weather, by random misfortune. Roughly 7 percent of adult bats fail to feed on any given night. The bats that miss don't die — because other bats feed them.
The mechanics of food sharing
Vampire bats roost in colonies of ten to a hundred individuals. A bat that has fed successfully returns to the roost with more blood than it strictly needs — bats often ingest 40 percent of their body weight in blood per night. Some of this surplus gets shared.
Sharing happens through regurgitation. A well-fed bat donates a portion of its blood meal directly from its digestive tract into the mouth of a hungry bat. The transfer is initiated by the hungry bat, who approaches a potential donor and begins grooming — licking the donor's fur and face. If the donor accepts, it regurgitates.
The donated blood is not trivial. A single donation can extend a starving bat's survival window by several hours. For a bat that has missed two nights, a donation can be the difference between recovery and death.
Gerald Carter's decade of research
The evidence that bat food sharing is genuinely reciprocal — not just kin-based — comes primarily from work by Gerald Carter, now at Ohio State University. Carter spent over a decade studying vampire bat colonies both in captivity and in the field in Panama, mapping the network of sharing relationships with fine-grained precision.
The key method involved temporarily removing bats from the roost and returning them hungry. Carter and colleagues then tracked who donated to the hungry bat and used genetic analysis to determine whether the donors were relatives. The result: sharing happened across the colony, not just within family groups. Bats shared with non-relatives at substantial rates. And critically, they shared preferentially with bats that had shared with them before.
In one set of experiments, Carter introduced unfamiliar bats to established colonies and manipulated which individuals had a history of sharing with each other. The pattern that emerged was clear: bats that had previously donated to a hungry bat were significantly more likely to receive donations from that bat in return. The relationship was tracked over years, not just a few nights.
Grooming as relationship-building
Grooming is the infrastructure. Bats that groom each other more frequently develop stronger sharing relationships. The causality seems to run both ways: bats that have shared food with each other also groom each other more. But grooming precedes food sharing when two bats are forming a new relationship — it's how a bat signals willingness to invest in a partner before a food emergency arises.
This mirrors a pattern that appears in primate societies: grooming as social currency, used to build the kind of trust that supports resource sharing when scarcity hits. Vampire bats developed the same strategy independently.
Partner fidelity and cheater detection
The network is maintained through something that functions like cheater detection. Bats that consistently receive donations but fail to donate in return are eventually cut off. Carter documented this in captive colonies where he could track every interaction: a bat that accepted food without reciprocating over many interactions would stop receiving donations from the bats it had failed to repay.
This is not simple tit-for-tat in the classical sense — bats don't respond to a single failure to repay with immediate withdrawal. The relationship has some tolerance for temporary imbalance. But consistent non-reciprocation ends it. The enforcement is behavioral and apparently remembered.
Sharing beyond kin
Most biological accounts of food sharing in animal societies explain it through kin selection — sharing genes through relatives. Vampire bats share more with relatives than with non-relatives, as expected. But a substantial portion of their sharing network involves non-kin, and those non-kin relationships are built and maintained through grooming and reciprocity, not genetics.
This makes vampire bat food sharing one of the cleaner examples in the literature of reciprocal altruism — the evolutionary strategy in which cooperation emerges not from shared genes but from repeated interaction, partner recognition, and the enforcement of mutual benefit over time.
Convergent comparison
The pattern has parallels in primate food sharing, where grooming and sharing form an exchange economy in several species. Chimpanzees share meat after hunts; bonobos share food to build alliances; humans share across large anonymous groups with elaborate institutional enforcement mechanisms. Vampire bats are a much simpler system, which makes the parallels useful — the basic structure of reciprocal cooperation appears to be stable across very different social contexts.
What the bat network shows is that trust-based mutual insurance can emerge from repeated interaction in small stable groups without language, explicit negotiation, or abstract reasoning. The requirements are: individuals that remember past interactions, enough interactions to develop a track record, and costs high enough to make cheating detectable. The bats have all three.
The engineering implication is uncomfortable: the mechanisms that make distributed systems reliable look a lot like the mechanisms that make small animal societies stable. Stable gossip protocols, reputation systems, and eventual consistency all have rough equivalents in what vampire bats work out every night in a hollow tree.
Building something? builds.anethoth.com is a public build ledger — proof that a product is really being built. Free to list your project.