Humans Probably Didn’t Feed Wild Animals to Domesticate Them

1. A distinction that matters

“Feed the chickens, and later you will get more eggs.” That sentence treats an outcome as a purpose. Chickens may lay more after feeding. Early humans scattering grain around a camp did not necessarily have eggs in mind. A behavior can help domestication without having domestication as its function.

Cross-species feeding is common, and the mechanisms behind it vary. Riedman’s 1982 review found alloparental care and adoption of non-offspring in more than 120 mammal species and 150 bird species. Calling all 150 bird cases “altruism” would hide the differences. Some may involve kin selection. Some may be recognition errors. Some may carry no detectable fitness effect. Human feeding of other animals belongs in that same spread of mechanisms.

This article separates four behaviors often lumped together as “providing food.” It then tests four explanations: altruism and reciprocity, display and signaling, live storage, and non-adaptive byproducts. It ends with domestication as a late population-level aftereffect.

2. Four different behaviors

“Providing food” covers at least four behaviors. They differ in cognition, selection pressure, and time scale.

Active provisioning: a person hands or throws food to an animal of another species.

Tolerated feeding: food sits somewhere, and humans do not chase away animals that eat it.

Leftover discard: scraps go around a camp, and animals take them. The human action aims at waste, not at feeding.

Live storage: when food is plentiful, a person lets a potential prey animal eat, planning to kill it later.

Zeder’s commensal pathway describes animals drawn to human settlements and food-processing waste. They become tolerated scavengers and eventually live permanently around people. The key move in that pathway is human non-refusal, not active feeding. If we call the whole process “humans providing food,” we erase the object of the human action: the waste, not the animal.

That distinction limits which explanations apply. For leftover discard, altruism and reciprocity lack a target. The behavior is not aimed at a recipient. The narrower question is why humans did not drive away animals that came to feed. That question is easier to test.

3. Low-cost tolerance

Food-sharing studies among hunter-gatherers offer a useful comparison. Among the Aché, hunters who brought in large, unpredictable resources did not receive more later because they gave more. The pattern of generous giving without matching return is hard to explain by delayed reciprocity. Hawkes proposed tolerated theft: when consumption yields diminishing returns and people are similar in competitive ability, defending food costs more than letting others take some. Widespread sharing can follow from that economics without reciprocity as a goal.

The cross-species version is limited but useful. People may let animals eat because chasing them off costs more than tolerating them. Around an early camp, driving away a scavenger takes energy and risks a bite. Letting it eat adds little cost. That is low-cost tolerance. It is neither altruism nor reciprocity.

The hypothesis has an advantage: it needs no complex psychology. It has a weakness: it is too easy to satisfy. To make it substantive, we need conditions. When is tolerance cheap enough that driving an animal away is not worth it? Food concentration, animal size and aggression, length of camp occupation, and abundance of alternative food all matter. Archaeology and ethnography can read some of these conditions, but the readings disagree. Critics of the garbage-dump hypothesis for wolf-dog domestication note that Paleolithic hunter-gatherers may not have produced enough sustained, concentrated refuse to support a scavenging wolf population. Wolves also rarely scavenge. Low-cost tolerance is logically possible. It is not empirically guaranteed.

4. Display and signaling: three constraints

Costly signaling theory gives “sharing shows you can get food” a formal shape. If sharing is public, observable, and hard to fake, it can signal a sharer’s qualities. That can bring status, reputation, or mating opportunities. Smith and Bliege Bird’s Meriam turtle-hunting study is a standard case. Hunters who caught large sea turtles gave meat away widely and kept little. Their reputation and social opportunities rose.

Extending the logic to cross-species feeding needs three conditions. Cross-species cases often fail all three.

First, an audience must be present and able to read the signal. Feeding at an empty camp edge gives display theory nothing to work with.

Second, the signal must be costly or hard to fake. Dropping rotting meat and carefully tending an animal differ sharply as signals. The first says little about quality.

Third, the jump from showing off to people to showing goodwill to animals is unverified. Mate provisioning in other animals has a clear recipient and a return structure. A human handing food to an animal does not get mating returns in the same currency. The analogy can suggest that sharing tendencies can be triggered by non-reproductive targets. It cannot show that cross-species feeding is an adaptation shaped by sexual selection.

Display may be a side benefit in some cases. It is a weak candidate for the main selection explanation, especially without an audience, repetition, or measurable fitness return. Analogies to bridewealth or payment for sex produce rhetoric, not an explanation.

5. Live storage: an analogy that needs evidence

“Fatten the animal when food is plentiful; kill it when food is scarce.” This treats feeding as delayed capture. It is logically coherent. It needs no altruism or reciprocity. The sharer moves surplus energy into another organism for later.

Surplus killing by predators is often used as a comparison. Kruuk defined surplus killing as killing prey without eating it when prey is freely available. Foxes in henhouses and wolves killing ungulates in deep snow are recorded cases. Some explanations say carcasses can be saved for later, as a larder. The same literature gives another reading. Surplus killing may be a hunting reflex running out of control when prey defenses fail. It may not be adaptive storage. In some cases it can deplete local prey and hurt the predator.

The human “fatten prey” hypothesis has larger gaps. Storage by surplus killing depends on the predator being able to control or return to the carcass. A human who feeds a free-ranging animal cannot guarantee recapture. Live storage requires conditions: the animal must be sedentary or trackable, humans must have practical control over its range, and killing it later must beat preserving food now. These conditions are possible without preservation technology. They are not automatic. They need archaeological and ethnographic testing. Treating “fatten, then kill” as the default explanation for early cross-species feeding has no stronger empirical base than “discard leftovers.”

6. Byproducts: mistaken triggering and borrowed mechanisms

Before looking for adaptation, consider a simpler option. Cross-species feeding may not be maintained by selection at all. It may be a byproduct of parental care or social reward systems triggered by unusual stimuli.

Interspecific feeding among birds gives the clearest model. Shy listed 65 bird species engaged in interspecific feeding. Riedman’s records cover 150 bird species that adopt non-offspring young. In most reported cases, reproductive error is the most reasonable explanation. Two species breed in the same area. Their young are similar in size and appearance before a certain stage. Begging calls are strong acoustic stimuli. A parent with high prolactin may care for any nearby young making those sounds, even if its own nest has failed. Begging can act as a supernormal stimulus. A parasitic chick exploits the host’s sensory bias. The host gains no fitness benefit.

The general lesson: a behavior can occur regularly without an adaptive function if the trigger falls inside an existing response range. Humans are sensitive to small, begging, dependent stimuli. That sensitivity was selected in parental care and intraspecific food sharing. It does not necessarily police species boundaries. Handing food to cats and dogs, or tolerating begging, may show how permissive the proximate system is. It does not prove an adaptation for cross-species reciprocity or domestication.

Primates offer supporting cases. Reviews of interspecific social behavior in primates report carrying, embracing, food sharing, and adoption of other species in several taxa. These behaviors look continuous with intraspecific affiliation. There is no evidence that each evolved a separate cross-species function. Langurs and chital deer in central India provide another case. Chital snatch leaves dropped by langurs and may use langur anti-predator alarms. Langurs sometimes rest on chital backs. Newton’s analysis treats the association as commensal rather than strictly mutualistic, with uneven benefits. Cross-species food transfer can produce stable ecological associations. Those associations do not require the feeder to benefit.

7. Domestication: a late, unintended population aftereffect

Zeder’s commensal pathway describes animals entering the human niche while humans tolerate rather than manage them. Dogs, cats, and pigs may have started there. The model has critics. The garbage-dump hypothesis for wolf-dog domestication faces the objection that wolves rarely scavenge and Paleolithic camps may not have produced enough refuse to support a scavenging wolf population. Domestication may have rested more on early cooperative social relationships between humans and wolves than on passive scavenging.

The debate matters for method. The causal distance between “providing food to animals” and “domestication” is much longer than the proximity of outcomes suggests. Even if the commensal model holds for a species, the human behavior was “not preventing animals from eating garbage.” Even if that process ended in domestication, the people in it did not need domestication intentions. Domestication is a population-level, multigenerational, unintended coevolutionary aftereffect. Treating it as the function of early feeding mixes temporal scales and causal levels.

Two aftereffects need separation. One is the fitness consequence of a behavior. Feeding may bring animals around a camp more often, which raises later hunting opportunities. Individual selection can capture that. The other is genetic and behavioral change accumulated over many generations: domestication itself. Individual-level selection does not directly cover that. It involves group structure, isolation, reproductive control, and mechanisms beyond a single feeding event. Confusing the two is the usual route by which teleology slips in.

8. Three constraints

Cross-species feeding is a case of multiple realizability. Giving food to another species can be maintained by mistaken parental care, low-cost tolerance, display, delayed capture, or a particular way of handling leftovers under local conditions. These mechanisms can coexist. No single label such as altruism or domestication covers them all.

A serious explanation should follow three constraints.

First, separate the object of the behavior from its consequences. Throwing food around a camp aims at waste disposal. Animals eating it is a consequence. This is not semantic fussiness. It decides which hypotheses have a target.

Second, separate proximate from ultimate causes. Oxytocin, dopamine, or parental hormones explain why the act feels natural or pleasant. They do not explain why the behavior persisted in evolutionary history. The two can vary independently. A behavior can be driven by reward systems and remain fitness-neutral.

Third, separate individual-level function from population-level aftereffect. Domestication accumulates over generations. It cannot serve as the motive for an individual behavior. Outcomes can be useful. Usefulness is not the cause.

Early humans feeding other animals most likely did not feed in order to domesticate, and did not share in order to be altruistic. A positive feedback system built for intraspecific parental care and social affiliation kept meeting begging signals and feeding opportunities from other species. Some of those encounters later opened paths to symbiosis and domestication. The explanation is less flattering. It is also better placed: not “what is this for?” but “under what conditions, and at which layer of mechanism, is it maintained?”