When a dog gazes into its owner’s eyes, oxytocin can rise in both of them, creating a bonding loop involving the same hormone released between mothers and babies — Japanese researchers found wolves did not show the same effect

Some of the strongest evidence that a dog loves you arrived, unromantically, in a specimen cup.

A team at Azabu University in Japan put dogs and their owners together in a quiet room for thirty minutes and told them to behave normally. Some owners talked to their dogs, others patted them, and a few just held eye contact and let that do the work. Urine samples were taken before and after, because the chemical being measured leaves the body that way. The findings, published in Science by Miho Nagasawa and colleagues, described something closer to a circuit than a cuddle.

A hormone with a day job

Oxytocin gets sold as a love potion, which undersells how busy it is. It drives contractions during labour and milk release during breastfeeding, and in humans it tracks closely with the behaviour of new parents. A review co-authored by the same Azabu University team, with Kazutaka Mogi and Takefumi Kikusui, points to research finding that mothers with higher circulating levels gaze at their babies more, touch them more affectionately and check on them more often. Parent and infant levels also rise and fall together during play, and the connection tightens when the pair are well synchronised.

So the hormone is already implicated in the oldest bond humans have. The question the Japanese team asked was whether a completely different species had found a way in.

How the loop was measured

Dogs in the first experiment were sorted by how long they looked at their owners. Among the pairs who exchanged long gazes, owner levels climbed steeply and the dogs’ levels rose too, though more modestly. Where the eye contact was brief, nothing much moved.

Then came the part that made the finding hard to wave away. In a second experiment the researchers gave dogs a nasal dose of oxytocin, which prompted the female dogs to spend more time watching their owners, whose own levels then rose in response. Push the hormone in at the dog end and it comes out at the human end, carried by nothing but a look. Writing in the same issue, Evan MacLean and Brian Hare said dogs had hijacked the human bonding pathway.

Eleven wolves in a room

The comparison group is what gave the study its teeth. Eleven hand-raised wolves went through the identical procedure with their owners, who fed them and saw them every day. These were not wild animals dragged in from a forest. They simply did not hold their owners’ gaze, and no hormonal response followed.

To a wolf, sustained staring reads as confrontation, not affection. Nagasawa’s team read the split as evidence that the gaze loop emerged during domestication, as dogs were selected, deliberately or otherwise, for looking at people.

The arguments the paper started

One paper is one paper, and this one attracted careful objections within a year.

A commentary by Zoltán Kekecs and co-authors at Baylor University pointed out that the headline comparison rested on 28 dogs and 11 wolves. At that size, they calculated, a real effect in the wolves would have slipped through undetected in most samples, so the absence of a correlation proves little. Rebuilding the statistical models with the original data, they found no significant difference between the two groups. Reliably detecting a real difference, per their simulations, would need around 44 wolves, matched by as many dogs.

Separately, Sylvain Fiset and Vickie Plourde questioned the analogy at the heart of the interpretation. Human bonding of this kind is usually studied between a child and a caregiver. The dogs in the experiment were bonded to their owners, which is a different relationship wearing similar clothes.

Wolves get another hearing

Seven years later, a Stockholm University ethologist went at the underlying question from a different angle. Christina Hansen Wheat and colleagues hand-raised wolf and dog puppies from ten days old under identical conditions, then ran them through the Strange Situation Test. That protocol was built to measure attachment in human infants, and it works by having a familiar person and a stranger take turns entering a room.

The wolves behaved much like the dogs: both told the familiar person apart from the stranger without trouble. The wolves, unlike the dogs, also showed clear signs of stress and paced the room, and the pacing stopped once their hand-raiser returned. As Hansen Wheat put it, the familiar human was working as a stress buffer. If wolves can already do some of this, attachment to people may have been raw material for domestication rather than a product of it.

Not everyone was convinced. Márta Gácsi, Ádám Miklósi and József Topál argued that the test as run departed too far from the original paradigm for the results to be compared with earlier work. A handful of behavioural preferences, they wrote, falls short of demonstrating an attachment system.

None of which touches the dog on the floor, still looking. Whatever the hormone turns out to be doing, the behaviour itself is unmistakable and unique among the animals that share a house with people. Settling the evolutionary argument, though, apparently requires 44 hand-raised wolves and their devoted owners, all willing to sit quietly in a room and hand over a urine sample afterwards.