A small 1980 study that helped launch modern research into the human-animal bond
Much of the research interest in the human-animal bond traces back to a 1980 study by Erika Friedmann and colleagues, which followed 92 patients discharged from a coronary care unit after a heart attack. Of the 53 pet owners in the study, 3 had died within a year, compared with 11 of the 39 non-owners. It was a striking finding, and a widely reported one. It was also observational, meaning it could show an association but could not establish that owning a pet caused the difference in survival.
The sample was small, and later research complicated the picture. In a larger observational follow-up of 369 patients recovering from a heart attack, dog ownership was associated with one-year survival after adjusting for disease severity and other psychosocial factors, while overall pet ownership and social support showed weaker statistical associations with survival in the same analysis. This does not mean the original finding was wrong. It means a single study, however striking, is a starting point for research rather than a conclusion. That is a fair description of how the whole field has developed - promising early results, followed by more careful and more mixed replication.
What actually happens in the body during human-dog interaction
A widely cited 2012 review in Frontiers in Psychology, led by Andrea Beetz, examined 69 original studies on human-animal interaction that met predefined criteria concerning peer review, sample size and study design. The review found reasonably well-documented effects on social attention and interpersonal interaction, on stress-related measures such as cortisol, heart rate and blood pressure, on self-reported fear and anxiety, and some cardiovascular measures. It found more limited evidence, meaning fewer qualifying studies rather than a lack of effect, for outcomes like trust, empathy and pain management.
Several individual studies included in that review are worth naming. Odendaal and colleagues reported that interaction with a dog, whether the person's own dog or an unfamiliar one, was associated with changes in cortisol and several other physiological measures in both the person and the dog. Interaction with one's own dog produced a stronger effect than interaction with an unfamiliar dog, suggesting that the existing relationship between person and dog may influence the physiological response. These findings came from specific, relatively small experimental studies and should not be generalised to every human-dog interaction. A separate randomised study by Cole and colleagues found that a 12-minute therapy-dog visit to hospitalised adults with advanced heart failure was associated with lower anxiety and short-term changes in several cardiovascular and stress-related measures, including systolic pulmonary artery pressure, compared with a visit from a person alone or usual care.
The review's authors proposed that activation of the oxytocin system may contribute to some of these effects, given oxytocin's established role in social bonding and affiliative behaviour more generally.
Frontiers in Psychology, 2012 · HABRI - Human Animal Bond Research Institute
The gaze loop that made headlines - and the scrutiny that followed
In 2015, a study published in Science by Nagasawa and colleagues proposed something specific: that longer mutual gaze between dogs and owners was associated with increased urinary oxytocin in both, and that this pattern was not observed in a comparison group of hand-reared wolves and their handlers. The researchers framed this as a loop that may have developed during dog domestication.
Dog-to-owner gaze predicted oxytocin change in both owners and dogs, but this pattern was not observed between wolves and their handlers.
The study was widely reported, and it is worth reporting fairly - including its limits. A 2016 commentary on the same data argued that the dog-wolf comparison was difficult to interpret, pointing to the small wolf sample of just 11 animals, differences in owner sex distribution between the two groups, differences in the dogs' and wolves' rearing and socialisation history, differences in baseline oxytocin levels, and questions about whether the statistical analysis directly established a dog-specific effect. The commentary does not disprove the original findings. It shows why the study is better treated as an influential early result than as settled evidence of a uniquely human-dog oxytocin loop.
This is a useful example of how the human-animal bond is actually studied - not through one dramatic result, but through a slow back and forth of findings, replications, critiques and refinements, the same process that applies to any area of behavioural science.
Science, 2015 - Nagasawa et al. · Frontiers in Psychology, 2016 - Commentary
The bond is not just a side effect of assistance and therapy work - it may be part of the mechanism
For a pet dog, the human-animal bond is a source of comfort and companionship. For an assistance or therapy dog, the relationship with its handler is an important part of the working partnership. Trust, familiarity and effective communication may support cooperation, engagement and the handler's sense of safety. For a working dog, the quality of that relationship is likely to matter: trust and responsiveness are part of what allows a dog to work reliably, whether that means responding to a trained cue, providing a physical task or remaining settled in an unfamiliar environment.
Some of the research suggests that the quality and familiarity of a relationship may influence physiological responses, rather than the presence of an animal alone. But a strong bond is not a substitute for temperament, task training, generalisation, health, welfare or appropriate matching, and the studies reviewed here do not test assistance-dog task performance directly. They point toward a plausible and important contributing factor, not a proven mechanism on their own.
Why the bond sits at the centre of our model
Every dog in our Cohort 1 programme is raised and trained by a Hero, one of our foster-trainers, in her own home for many months before any question of rehoming with purpose comes up. That is a deliberate design choice, not a scheduling constraint. A dog that spends months being trained by the same person in a household environment has a different training history and relationship context from a dog trained mainly in short sessions or kennel-based settings.
We treat this relationship as part of the training outcome, not just the training method. When a dog is eventually matched with a person it will support, whether in a school, a hospital or a home, the goal is to build a comparable relationship there too, one grounded in trust rather than compliance. The science on the human-animal bond is still developing, and we try to represent it honestly rather than reach for the most dramatic version of a finding. But the consistent thread across several decades of research is that interactions and relationships between people and animals can be associated with measurable physiological and psychological effects. That is worth taking seriously, and it is a large part of why our programme is built around long-term bonds between dogs and the women who raise them, rather than around training alone.
