The animal whose milk most people in the world drink can recognise individual members of its own herd, choose a preferred partner from among them, and respond physiologically to that partner’s presence in ways that look unmistakably like the behavioural signature of friendship. None of this is contested at the level of the data. It is contested only at the level of what the dairy industry is willing to say out loud about the cows it sells.
The popular framing of cattle goes like this: they are placid, interchangeable, mildly stupid, and emotionally flat. That framing is approximately right in its visual effect — a field of black-and-white animals does look uniform from a distance — and almost entirely wrong in its biology. The part of the story worth slowing down on is the work that the British animal welfare researcher Krista McLennan first presented in 2011, the heart-rate monitors, the cortisol samples, and the roughly fifteen years of subsequent research that has confirmed and extended the original finding.
What the 2011 research actually measured
The work that put cattle friendship onto the welfare-science map was conducted by McLennan at the University of Northampton and first presented publicly in 2011 at a Centre for Animal Welfare and Anthrozoology seminar at the University of Cambridge and at the Northampton postgraduate research conference. The consolidated findings appeared in her 2013 doctoral thesis on social bonds in dairy cattle. The methodology, by the standards of farm-animal welfare research, was unusually clean.
McLennan paired heifers with either a preferred social partner identified through earlier observational work, an unfamiliar animal, or no companion at all, isolated them for short periods, and measured heart rate and cortisol — the standard hormonal marker for stress response — across the conditions. The behavioural response was scored alongside the physiological data.
The headline result, the one that has carried into the popular framing of the work, was on heart rate. Heifers isolated alone, or isolated with an unfamiliar animal, showed elevated heart rates and behavioural signs of distress. The same animals isolated with their preferred social partner showed measurably lower heart rates and reduced behavioural stress. Cortisol, by contrast, did not differ in a statistically meaningful way between the preferred-partner condition and the familiar-stranger condition, a detail that gets dropped in most of the secondary coverage. The honest summary is that behavioural and cardiac measures responded to the preferred-partner effect; the hormonal measure did not. That is itself a useful finding, because heart rate variability and behavioural scoring are increasingly treated by welfare scientists as more sensitive indicators of acute social stress than point-in-time cortisol sampling.
The framing in the resulting press coverage was careful. McLennan did not claim the cows were experiencing friendship in the literary sense. She claimed that the data were consistent with the formation of preferred social partnerships, and that separation from those partners produced a physiological stress response that the partner’s presence measurably reduced. That is roughly the same evidentiary bar used to establish pair-bonding behaviour in primates and corvids.
What the dairy industry knew, and when
McLennan’s work was not buried in an obscure journal. It was widely covered in mainstream and agricultural trade press from 2011 onwards. It was discussed at welfare conferences attended by dairy industry representatives. It was incorporated, in subsequent years, into the welfare science curricula at agricultural universities that train the veterinarians and herd managers the industry hires.
The industry’s response, in the years since, has not been to dispute the finding. The response has been to absorb it as an operational consideration where convenient — some larger operations now try to keep stable social groups together during routine handling — and to leave it out of the consumer-facing description of how milk is produced. On most commercial dairy operations, calves are separated from their mothers within hours of birth, and this separation produces measurable stress responses in both animals.
The gap between what is known inside the industry and what is communicated outside it is not a secret. It is documented in the industry’s own welfare literature, which discusses social bonding, separation stress, and reunion behaviour in technical terms while the marketing departments of the same companies describe their cows as content.
What subsequent research has added
McLennan’s heart-rate work was not the last word. The intervening years have produced a substantial body of follow-up research on cattle sociality, much of it consistent with the original findings and some of it extending them in directions not initially tested.
Recent work on cattle used as therapy animals has documented selective social preferences that extend beyond herd familiarity to discriminations between individual humans. A 2024 study by Katherine Compitus of New York University and Sonya Bierbower of West Point, published in the journal Human-Animal Interactions, observed two therapy steers, Magnus and Callum, interacting with eleven human participants, and reported that the steers showed a strong preference for interactions with women compared to men. The sample is small, and the authors are appropriately cautious about whether the steers were actively seeking out women or whether the women were simply more likely to initiate contact, but the result is consistent with a broader pattern in the cattle behaviour literature: these are animals that make fine-grained social discriminations about the individuals around them, not just generic responses to familiar versus unfamiliar stimulus categories.
The behavioural ecologist Marc Bekoff, who has spent decades documenting emotional capacity in non-human animals, has written extensively about the gap between what the science shows and what cultural framing allows. In a recent piece, Bekoff argues that cows are highly intelligent and deeply sentient and emotional beings with distinct individual personalities. His broader work on animal emotions and animal sentience traces the long methodological history of how researchers came to accept, against
