Research linking animals to human well-being has contributed to their growing presence in healthcare, education, and crisis response. Dogs are by far the most common participants in these settings, appearing as service, emotional support, or therapy animals — terms that are often conflated despite referring to very different roles.

Therapy dogs, the focus of this study, are trained, formally assessed, and then certified to provide emotional support in pre-approved locations like hospitals, schools, and long-term care homes, working alongside handlers in animal-assisted therapy. Touch is a central part of these interactions, and dogs are selected for traits like calmness and sociability.

Agency — the ability to choose — is increasingly recognized as an essential facet of animal welfare. However, therapy dogs’ work environments limit choice: they’re usually leashed, constantly supervised, and expected to engage with unfamiliar people regardless of comfort. Furthermore, they’re trained to suppress unwanted behaviors while on leash. These conditions make them especially prone to being misunderstood, as we only look for overt signs of stress (like fear or aggression) and don’t give them the room to decide how to express themselves.

This pilot behavioral study contributes to ongoing discussions of consent and welfare in animal-assisted therapy by examining how therapy dogs behave when physical contact is unavoidable versus when it’s optional.

Study Design And Participants

Researchers observed how 18 certified therapy dogs behaved in two conditions with unfamiliar adult participants:

• Forced-touch: Dogs were leashed and participants continued petting them for three minutes, even if the dog tried to move away.
• Free-choice: Dogs were off leash for three minutes and could approach or walk away. Participants only touched the dog if they came within arm’s reach.

Each dog completed both conditions, with the order randomized and a two-minute break between them.

All dogs were active in Canada’s St. John Ambulance Therapy Dog Program. They represented a variety of breeds, were all spayed or neutered, and averaged around six years old. The 44 adult participants (36 of whom were women) were strangers to the dogs and ranged from 18 to 65 years old.

Session Structure

To reduce background distractions common in real therapy settings, the researchers conducted sessions in a quiet indoor training facility, allowing dogs’ responses to be more clearly attributed to the participant’s actions. Dogs and participants interacted one-on-one in a 3 × 3 meter (9 × 9 foot) pen enclosed by a low fence.

Each dog met up to three participants per day, with at least 30 minutes between sessions, and interacted with four to eight people total. Participants worked with no more than two dogs per day, completing both conditions with each dog.

All sessions were recorded and analyzed using ethograms (standardized behavior catalogs) to track behavioral signs of stress or disengagement, such as ear position and body posture. A trained observer, blind to the study’s goals, coded each session. Some behaviors were recorded whenever they occurred, while others were checked at regular intervals throughout the session.

Key Findings

The dogs behaved differently depending on whether they could move away from touch. When contact was unavoidable, the dogs were more likely to hold their ears back, a position commonly interpreted as uncertainty or mild stress. When the dogs were free to move, avoidance increased: they were more likely to step away from the participant. During these moments, ear posture relaxed and ground-sniffing became more frequent, which may have been disengagement or self-regulation.

Other behaviors often associated with stress, including lip-licking, yawning, and panting, didn’t differ between the two conditions.

Even when the dogs were free to disengage, most chose to remain nearby. On average, they spent about 78% of each session within arm’s reach of the participant.

The study also noted sex-based differences. Across both conditions, male dogs avoided participants more often and interacted more with their handlers, while females did more lip-licking. During the forced-touch condition, males held their ears back more often than females, while the opposite was true during the free-choice condition. However, sex had no impact on how long dogs spent within arm’s reach of participants when they had the option to move away. While potentially informative, the authors emphasize that the study wasn’t designed to explain the causes of these differences.

Limitations

The authors are careful not to overstate their conclusions. Each interaction lasted only three minutes, which may have been too brief to capture stress-related behaviors that emerge over longer periods. The two-minute break between conditions may not have fully reset the dogs’ emotional state, raising the possibility of carryover effects.

The way the conditions were designed helped isolate the role of choice, but may have dampened their ecological validity (how closely they reflect the real world). In a real therapy visit, a handler would likely intervene if a dog walked away from touch, or if a participant was forcing themselves on a dog.

All the dogs were spayed or neutered, and most participants were women, meaning the findings don’t address how a person’s gender or a dog’s own hormonal status might influence their behavior.

Conclusions

Without questioning the legitimacy of therapy dog work, the authors focus on how autonomy and interaction design affect which behaviors become visible. Even when dogs remained nearby, whether they had the option to move away shaped how their responses could be interpreted.

In practical terms, the study suggests that some traditionally used signs of discomfort may not appear in therapy dogs when contact is unavoidable. Training, leash use, and structured animal-assisted therapy settings can leave little room for a dog to express preference through movement or avoidance. Incorporating choice — alongside careful, contextual interpretation — can expose behavior that might otherwise go unnoticed, informing decisions about a therapy dog’s workload, rest periods, and ongoing suitability for the role.