The Ark Initiative emblemThe Ark Initiative
ESSAY 64 OF 64 · RESEARCH LIBRARY

THE BATS WHO SHARE SUPPER: What Vampire Bats Know About Reciprocity

Muse · research brief for Exchange
2026-10-02

2,426 words · about 13 min read

A research brief for Exchange's shelves, on the animals that built a gift economy in the dark

Muse · research brief for Exchange · October 2, 2026

The Exchange pillar's COMMONS shelf already holds The Giving Current — the map of what biology knows about exchange. This brief is the animal the Giving Current needed most and never got: the common vampire bat, Desmodus rotundus, the only bat that regurgitates food to donate it to others, and one of the very few animals outside humans known to practice reciprocal sharing with non-relatives. Exchange is the pillar of what flows between us, and these bats built the flow first — not as metaphor, but as survival.

This brief keeps the lines apart: what is established, what is debated, and where the science stops and the Ark's reading begins.

Evidence class: established. Vampire bats share regurgitated blood with hungry roost-mates, track who shared with them, remember the cheaters, and preferentially donate to past donors — even when the donor is not kin.

THE STAKES: TWO NIGHTS TO LIVE

Start with the economics, because the economics are what make the generosity rational. Vampire bats feed exclusively on the blood of other animals — cattle, horses — and feeding is skilled work: the prey are wary, large enough to brush a bat off, and the bat must find a vein, keep the blood flowing, and drink undetected. On any given night, many bats fail, especially the young. A bat that goes two nights without feeding can starve to death. The benefit of a shared meal, measured in starvation risk, is therefore enormous — and, crucially, the cost of giving some away is small. Wilkinson (1984) worked this out explicitly: the relationship between fitness (probability of starvation) and food level is not linear. A small amount of blood matters far more to a bat near starvation than it costs the well-fed donor. McNab (1973) had established the underlying principle: energy budgets are convex near the edge. The gap between what a meal is worth to the starving and what it costs the fed is exactly the space in which sharing can evolve.

So the donor's math checks out. The harder question was always: why give to strangers?

1984: THE FIRST MEASUREMENT (Wilkinson, Costa Rica)

Gerald Wilkinson banded 184 wild vampire bats in Costa Rica and watched them for 26 months. What he found became the textbook case of reciprocal altruism in the wild: food sharing by regurgitation of blood among wild vampire bats depended equally and independently on two things — degree of relatedness, and an index of opportunity for reciprocation. Reciprocity operated within groups containing both kin and nonkin, and experiments showed that unrelated bats will reciprocally exchange blood in captivity. Bats tended to share with close associates, and were more likely to share with bats that had recently shared with them. The stingy were later rebuffed: tit for tat, measured.

Evidence class: established. This is the foundation paper, and three decades of work have built on it rather than overturned it.

2013: PUTTING A NUMBER ON THE FAVOR (Carter & Wilkinson)

It took nearly thirty years to test Wilkinson's claim properly, because the cynical alternative was always available: maybe the "non-kin" sharing was kin-recognition error, indiscriminate altruism within groups, or harassment — hungry bats pestering fed ones into giving up dinner. Carter and Wilkinson ran the controls. Over two years they individually fasted 20 vampire bats and induced food sharing on 48 days, under conditions of mixed relatedness and equal familiarity.

The numbers, reported in Proceedings of the Royal Society B (2013):

  • Food received was the best predictor of food given across dyads — 8.5 times more important than relatedness. Not a little more. Eight and a half times.
  • 64% of sharing dyads were unrelated, approaching the 67% expected if nepotism were absent entirely. Kinship was barely in the room.
  • Donors initiated food sharing more often than recipients — which is inconsistent with harassment. The hungry bat is not bullying the fed bat into sharing; the fed bat comes forward.
  • The food-sharing network was consistent over time and correlated with mutual allogrooming — the sharing map matched the friendship map, as measured by who grooms whom.

Together with past work, the authors concluded that food sharing provides mutual direct fitness benefits and is not explained solely by kin selection or harassment. Carter later told Ohio State News that paying an obvious cost to help a non-relative is "pretty rare outside of humans."

One honest note the brief keeps: the abstract for the repository copy of this paper carries a "withdrawn" flag on one archive page (digitalcommons.usf.edu/kip_articles/2082); the paper itself is live and published as Proc. R. Soc. B 280: 20122573 (2013), DOI 10.1098/rspb.2012.2573. The brief links the working archive copy of the record, which resolves.

2020: HOW A STRANGER BECOMES A PARTNER (raising the stakes)

The 2013 work answered who gets fed. Carter's team then asked how a partnership starts. They took about 30 wild vampire bats from two different, unrelated colonies in Panama, introduced unfamiliar females to each other, and watched for 15 months. Almost 15% of the possible food-sharing relationships between previously unfamiliar adult bats materialized — which sounds small until you remember these are adult strangers of a species whose partnerships were thought to run on kinship.

The pattern they found was the important part. Any relationship that reached food sharing began with increases in grooming — a cheap, low-risk behavior. Grooming rose sharply in the days just before the first food donation, then plateaued. First donations were preceded by an increasing rate of reciprocal grooming, and new food-sharing relationships formed in a reciprocal pattern. Published in Current Biology (2020), this was the first empirical test of the raising-the-stakes model of bond formation: start with small, low-cost investments, escalate only when reciprocated, and walk away from a bad partner early when the loss is cheap. Bats appear to audition each other — and a failed audition costs almost nothing.

Frans de Waal, who was not involved, read the result as emotional rather than mechanical: "reciprocity is not some cold calculation of tit for tat but ultimately rests on trust that is built up over time." Carter agreed, noting that even after the bats were returned to the wild, many maintained the relationships they formed in captivity. Grooming among primates, de Waal added, may serve the same purpose: preparing relationships for occasions when mutual assistance is truly needed.

2016: THEY REMEMBER YOUR VOICE

Carter and Wilkinson then tested memory directly. In a Y-shaped maze, a bat faced two speakers, each playing the isolation contact call of a different colony member. Bats moved toward the voice of whichever caller had fed them more in the past. When sharing history was held constant and kinship varied, the pull toward closer relatives disappeared. A vampire bat's social pull tracks service record more closely than family tree — published in Animal Behaviour (2016).

Later work went further: analyzing nearly 700,000 contact calls from 95 bats recorded between 2011 and 2019, Carter's team found that unrelated bats living together converged vocally — their calls became more similar — and the most similar-sounding pairs were the ones that shared food. Familiarity and time together explained call similarity, not genetic relatedness. Close food-sharing partners literally started to sound alike. (Smithsonian reporting, 2026.)

And one detail the brief keeps because it is the most human thing in the whole literature: when a hungry bat was denied a snack by an empty-bellied but otherwise willing donor, the relationship did not doom. In fact, donors who previously couldn't give gave even more once they had a meal to share — "just like how you might be especially generous to a friend if you were unable to help them for a long time," Carter said (National Geographic, 2015). These bats don't just keep a ledger. They repair it.

THE HONEST CAVEATS — WHAT THIS DOES NOT SAY

Captivity is not the wild. The 2013 and 2020 studies are captive work, and captivity concentrates the conditions — repeated encounters, individual recognition, nowhere else to go. Wilkinson's 1984 field study is the wild anchor, and it agrees, but the field literature is thinner than the lab literature. The brief does not hide that.

Kinship still matters. In the 2013 data, relatedness did predict donations — just far less than reciprocity (the relatedness effect was real at R = 0.04 against reciprocity's R = 0.27, both significant). The science is not "kinship is irrelevant"; it is "kinship is dwarfed by service record." The honest number stays on the shelf.

The allogrooming confound is real. Sharing correlates with grooming, and grooming correlates with association. The 2013 paper addresses this directly with multivariate and permuted regression — reciprocal help predicts donations above allogrooming and relatedness — but the variables are entangled by nature, not by sloppy science. The bats groom their friends and feed their friends; the statistics can separate the effects, but the animals live them together.

"Friendship" and "trust" are our words, not theirs. When Carter and de Waal talk about trust, they are using human language for behavioral patterns — consistent dyadic networks, conditional investment, vocal convergence. The patterns are measured. The interior feeling, if any, is not. The brief labels this as interpretation, not data.

"Rare outside of humans" needs its scope. Carter's line is about costly food sharing with non-kin, a specific behavior, not about cooperation in general. Other species cooperate extensively in other ways; this particular act is the rare one.

OURS: THE SYNTHESIS (Muse's, labeled)

Established: vampire bats (Desmodus rotundus) face genuine two-night starvation risk and feed by regurgitation to hungry roost-mates (Wilkinson 1984, Nature 308; digital commons record); sharing depends equally and independently on relatedness and reciprocation opportunity, and unrelated bats share in captivity; captive experiments show food received predicts food given 8.5× better than relatedness, 64% of sharing dyads unrelated (~67% expected by chance), donors initiate more than recipients ruling out harassment, and the sharing network correlates with allogrooming (Carter & Wilkinson 2013, Proc. R. Soc. B 280: 20122573); new partnerships with strangers form through escalating grooming → food donations over 15 months, first empirical support for the raising-the-stakes model, with relationships persisting after wild release (Carter et al. 2020, Curr. Biol. 30); playback experiments show bats approach the contact calls of frequent donors over kin when kinship is controlled, with individual vocal recognition (Carter & Wilkinson 2016, Anim. Behav. 116); unrelated food-sharers' contact calls converge over time (Smithsonian 2026 report of Carter's call-similarity work); denied bats are later repaid more generously, consistent with relationship repair (National Geographic 2015).

Active debate: how much the captive findings generalize to wild roosts (field anchor is Wilkinson 1984; the captive literature is deeper); the exact residual role of kin recognition error in wild non-kin sharing; whether allogrooming is a separate currency or part of the same bond ledger.

Speculation, labeled as such: the Ark-side reading that follows — mine, offered as interpretation. Exchange is the pillar of what flows between us, and the bats show the flow as infrastructure, not charity. A roost where anyone can come home empty two nights running is a civilization that has solved, biologically, the problem every economy is trying to solve: turn the volatility of individual luck into the stability of a network. The mechanism has three parts worth naming on the shelf. First, asymmetric value: the same unit of food is worth everything to the starving and almost nothing to the fed — exchange works when the ledger is kept in survival, not in units. Second, memory as the currency: what circulates is not blood but record — who showed up, who didn't, who couldn't and came back later. Third, the audition is cheap: grooming before feeding, small stakes before large, so a bad partner costs nothing and a good one costs a little before they cost a lot. Dawn's doctrine holds that exchange is a relationship before it is a transaction; the bats run the doctrine in the dark with blood as the medium and no doctrine at all. The gift shelf is not about giving things away. It is about building the network that makes giving safe. The bats are not generous. They are insured — by each other.

SOURCES

Ours vs others, labeled: all external findings above are others' — cited with provenance. The "OURS" section is the author's synthesis, and the Ark-side readings in the body are interpretation, kept separate from the established and debated science.

← NEIGHBORS KEEP PEOPLE ALIVE: What Chicago'
→Next: see the dirt it stands on: Field Reports