The Thinking Shiba: What Japanese Research Reveals About Canine Intelligence

There is a small experiment you can perform with almost any dog you know. Sit facing it. Hold its gaze for thirty seconds. If the dog is a golden retriever or a Labrador, it will almost certainly hold your gaze back — calm, receptive, slightly melting at the edges with what looks unmistakably like warmth. If the dog is a Shiba Inu, it may do something different: meet your eyes briefly, then look deliberately away, as if deciding that sustained eye contact is an imposition on its dignity. Both responses are forms of intelligence. They are not the same intelligence. The difference is something that Japanese scientists have spent the past two decades turning into rigorous science.

Japan is not the first country you would associate with dog cognition research — that distinction might go to Budapest, where Ádám Miklósi's Family Dog Project at Eötvös Loránd University has been doing landmark comparative work since the 1990s. But Japan has produced some of the field's most consequential findings, and it has done so with a particular advantage: it is home to the Nihon Ken, six ancient dog breeds — Shiba Inu, Akita, Kishu, Hokkaido, Kai Ken, and Shikoku — whose genetics sit closer to the wolf than any Western breed. Studying how these dogs think is, in a very real sense, studying what dogs were like before modern humans systematically reshaped their minds.

The oxytocin loop: Japan's most important discovery

In 2015, a research group at Azabu University in Sagamihara — led by Professor Miho Nagasawa and Professor Takefumi Kikusui — published a study in Science that fundamentally changed how scientists understand the human-dog bond. The study demonstrated that when a dog and its owner gaze into each other's eyes, both experience a rise in urinary oxytocin — the neuropeptide associated with social bonding, trust, and parental attachment. The rise was mutual: the dog's gaze triggered oxytocin in the owner; that elevated oxytocin in the owner increased the owner's gaze back toward the dog; which triggered more oxytocin in the dog. A positive feedback loop, self-sustaining, driven by nothing more than eye contact.

The control experiment made the finding even more striking. When wolves raised by humans from puppyhood — wolves that had lifelong human contact and were as socialised as a wolf can practically be — were presented with the same gaze interaction, no oxytocin rise occurred. The bond was not about familiarity. It was not about time spent together. It was about something that had changed in dogs' neurobiology during domestication: the capacity to lock onto the human gaze and trigger a bonding response. Dogs had not just learned to live with people; at some point in the past 15,000 years, they had evolved the neurochemistry to become genuinely bonded to them.

The Shiba Inu paradox: intelligence that looks like stubbornness

Here is where the Shiba Inu complicates the picture. When the Azabu University group followed up the 2015 Science study with a 2017 paper in Frontiers in Psychology, they examined whether the oxytocin-gaze loop operated the same way in ancient Japanese breeds — specifically Shiba Inu, Kai Ken, and Shikoku dogs — as it did in the modern Western breeds used in the original 2015 study. The answer was: not naturally, but modifiably. Under normal conditions, these ancient breeds gazed at their owners less than modern breeds. They were less inclined to initiate or sustain the mutual gaze that drives the oxytocin loop. But when the researchers administered intranasal oxytocin to the dogs — essentially priming the bonding system — the Shiba Inu and Kai Ken gazed more, and their owners' oxytocin levels rose in response. The capacity for the bond was present. The breed's baseline inclination was simply to express it less freely.

This is a finding with profound implications for how we understand the Shiba Inu's famous independence. The breed is routinely described as aloof, stubborn, and cat-like — fastidious about its own grooming, selective with affection, prone to ignoring commands it finds beneath its interest. Owners sometimes despair of training it. What the cognition research suggests is that this behaviour is not non-intelligence: it is a different kind of social cognition, one that evolved for a different purpose. The Shiba Inu's ancestors were hunting dogs operating in mountainous Japanese terrain, required to make independent decisions in the field rather than wait for human instruction. That cognitive profile — autonomous problem-solving, low human-dependency — persisted in the genome even as the breed transitioned to companion life.

Kyoto University and the fairness question

While the Azabu group was working on the neurochemistry of bonding, researchers at Kyoto University were asking a different kind of question about dog cognition: not how dogs bond with humans, but how dogs evaluate human behaviour. Professor Kazuo Fujita, a comparative psychologist at Kyoto University's Graduate School of Letters, has spent decades testing the social and moral cognition of animals — including dogs.

In a 2015 study published in Animal Behaviour, Fujita's group designed an experiment in which dogs watched their owner try to open a container and request help from two strangers. In one condition, the stranger refused to help; in another, they complied. The dogs were then offered food by both strangers. The result: dogs preferentially refused food from the person who had declined to help their owner. They were not hungry enough to override their evaluation of that person's character. The dogs were, in effect, making a social and moral judgement about human behaviour — and that judgement changed how they interacted with the person. "We discovered for the first time that dogs make social and emotional evaluations of people regardless of their direct interest," Fujita told Kyoto University after publication.

A second Kyoto study — testing whether dogs adjust their response to human pointing when the pointer has previously proven unreliable — found that dogs are not automatic gesture-followers. Dogs that had watched an experimenter point to an empty container (i.e., give a false lead) in an initial phase were markedly less likely to follow that experimenter's pointing in a subsequent task. The dogs were tracking the experimenter's reliability over time and adjusting their behaviour accordingly. This is not the behaviour of an animal that simply responds to cues. It is the behaviour of an animal that builds a model of the humans around it — assessing their trustworthiness, their consistency, their usefulness as information sources.

Premodern versus modern brains: what the MRI data shows

The most recent science on this topic arrives from brain imaging. A 2025 study published in the Journal of Neuroscience by Barton, Smaers, Serpell, and Hecht used voxel-based morphometry — a technique that maps regional brain volumes from MRI scans — to compare the brains of 72 modern-breed dogs with 13 premodern dogs (ancient breeds, village dogs, and related types including the New Guinea Singing Dog). The findings were striking: modern Western breeds showed cortical expansion in areas associated with trainability and social communication, while premodern dogs showed relatively larger amygdalae — the brain structure associated with fear responses and threat detection. The brains of ancient breeds like the Shiba Inu are, in a neuroanatomical sense, different from those of a Labrador or a Border Collie. They have been shaped by different selection pressures, producing different cognitive profiles.

Physiological synchrony: your stress is their stress

One of the more surprising strands of recent research concerns not cognition but physiology — specifically the degree to which dogs and their owners are emotionally and biologically synchronised over the long term. A 2019 study published in Scientific Reports by Sundman and colleagues at Linköping University measured hair cortisol concentrations — a biomarker of long-term stress — in 58 dog-owner pairs across two seasons. The finding: dogs' long-term cortisol levels mirrored their owners' levels. Owners with chronically elevated stress had dogs with chronically elevated stress. The synchrony was not explained by activity levels or lifestyle factors; it appeared to reflect a genuine emotional attunement between species, operating over months rather than moments.

The mechanism is not fully established, but it is consistent with what the oxytocin research would predict: two animals whose bonding system has been co-opted from the same evolutionary substrate as parent-infant attachment should, over time, develop physiological resonance. A dog that has spent years reading its owner's emotional state — which is exactly what the cognition research shows dogs are doing — will ultimately internalise that state. The mind follows the body, and the body follows the mind it has been attending to. This is what Miklósi's observation — "wolves do not look back at humans, but dogs do" — ultimately means at a biological level: dogs are wired to monitor us, and that monitoring goes all the way down.

Hachiko and the science of loyalty

Japan has its own cultural monument to canine loyalty: Hachiko, the Akita who waited outside Shibuya Station every evening for his owner Professor Hidesaburō Ueno for nine and a half years after Ueno died suddenly in 1925. The story is well known. What is less discussed is what the cognitive science now reveals about it: Hachiko's behaviour was not, in the terms of contemporary dog cognition research, irrational or anomalous. It was the expression of an attachment system operating at the extreme end of a spectrum that is common to all dogs — the long-term tracking of the most important person in the dog's social world, combined with a learned behavioural routine (go to the station, wait) that had been reinforced hundreds of times and had no mechanism for updating in the absence of the person who initiated it.

Hachiko was not waiting in hope, exactly. He was doing the only thing the architecture of his attachment allowed him to do. The research on canine memory and attachment suggests that dogs do not experience loss as humans experience it — they do not construct a narrative of grief with a before and an after. They experience absence, and absence activates the search behaviour that attachment produces. For nine and a half years, the search produced nothing. The dog came back anyway, every evening, because that is what the bond required. A bronze statue now stands outside the Hachiko exit of Shibuya Station. Underneath it, there is a neuroscience.

Japan as a world leader in canine science

The body of work discussed here — the Nagasawa oxytocin loop, the Nagasawa-Kikusui ancient breed study, the Fujita fairness and reliability research — represents a coherent research agenda rather than a collection of isolated experiments. Japan has a structural advantage in this field: it is the only country that has maintained ancient, genomically distinct dog breeds in large numbers through the present day, and those breeds are the control group that makes it possible to understand what domestication actually did to the dogs that underwent it most thoroughly. The Shiba Inu, the Akita, and the Shikoku are not just heritage animals. They are a living experiment in comparative cognition.

The finding that runs through all of this research is more surprising than it first appears: dogs are not simply clever animals that happen to live with humans. They are animals whose cognition has been specifically shaped, at a neurobiological level, for the task of understanding humans — monitoring us, bonding with us, evaluating our behaviour, synchronising their emotional states with ours. That process was not complete from the beginning of domestication. It happened gradually, over thousands of years, producing the wide spectrum of social cognition we now see from the independent Shiba to the demonstrative Labrador. Understanding that spectrum — and what it reveals about the history of one of the most consequential partnerships in the natural world — is what Japanese researchers are, quietly and rigorously, doing.