Over recent years, scientific recognition of the complex social skills possessed by dogs has increased, with canine social cognition often being considered analogous to that of humans [1
]. A prime example lies in the type of relationship between humans and dogs where, similar to a human infant, dogs often display behaviors indicative of attachment bonds, such as proximity-seeking during stress [2
]. Although the biological mechanisms underlying dogs’ human-like social competence remain unclear, oxytocin is regularly proposed as a primary mediator [3
Oxytocin is a hormone and neuropeptide produced centrally, in the hypothalamus, and peripherally, in organs such as the heart [4
]. Evidence suggests that the release of central oxytocin inhibits the activation of neural stress pathways, such as the hypothalamic-pituitary adrenal (HPA) axis, and threat-processing centers, such as the amygdala [5
]. Peripheral oxytocin release is also believed to inhibit stress reactivity, observed in the cardiovascular system via reduced heart rate and increased heart rate variability [6
]. Oxytocin also plays an essential role in mediating the neural reward of social information [7
]. These mechanisms are likely to explain, at least in part, the increase in canine pro-social behaviors observed with elevated oxytocin concentrations, including increased gaze towards the owner [8
] and increased affiliative behaviors such as sniffing, licking, body contact, and play [9
Much of the research investigating oxytocin responses and canine social behavior has used either exogenous oxytocin administration, to increase oxytocin concentrations, or a genetic approach, examining associations between genetic polymorphisms in the oxytocin receptor gene and canine behavior [10
]. Moreover, several correlational studies have investigated the influence of human interaction on canine endogenous oxytocin, documenting an almost universal increase in oxytocin concentrations [11
]. The literature is consistent in that only one study, to our knowledge, has reported no change in oxytocin following human interaction [9
]. The interaction protocols of each study vary significantly, examining interaction types such as gaze [14
], stroking, petting, talking [11
], and free-form interaction in which the dog leads the interaction type [16
]. One study also examined the influence of exercise on endogenous oxytocin concentrations in nine dogs and reported a positive effect after 15 min of “trotting” [13
]. However, the combined effect of exercise and human–dog interaction through activities such as owner-led dog-walking remains undocumented.
Dogs’ behavior toward humans varies significantly with familiarity [17
] and closeness of the relationship. For example, dogs may direct most of their attention toward a familiar person who is primarily responsible for their care (e.g., in the form of exercising and feeding) compared to a familiar person who is responsible for less than half their care [18
]. Given these behavioral differences, it is plausible the strength of attachment between owner and dog may also influence a dog’s oxytocin response to human interaction. Pilot data suggest a positive correlation exists between owner-reported strength of the human–dog bond and canine oxytocin concentrations. Alternatively, the owner’s perception of the difficulty in caring for their dog or the likelihood of their dog making a mess was negatively correlated with canine oxytocin concentrations [19
Only a limited number of empirical investigations have examined endogenous oxytocin concentrations following human–dog interactions, and the sample size used in each study are typically small [9
]. The primary aim of the current study was to investigate the canine endogenous oxytocin response to owner-led dog-walking and affiliative human–dog interaction. A secondary aim was to investigate the moderating effect of the owner-reported strength of the human–dog bond on canine oxytocin responses.
The primary aim of this study was to examine responses in the concentration of canine endogenous oxytocin to dog-walking and affiliative human–dog interactions. Contrary to most of the current literature [11
], we did not observe a significant effect of owner-led walking or affiliative human–dog interactions on urinary oxytocin concentrations. Our results align with one previous study that reported no effect of human–dog interactions on urinary oxytocin concentrations when examining affiliation between owner and dog, or the proximity of dogs to their owners [9
A secondary aim was to investigate the moderating effect of the owner-reported strength of the human–dog bond on canine oxytocin responses to dog-walking or affiliative human–dog interactions. Again, there was no significant change in oxytocin concentration following interactions in dogs whose owners reported greater or lower levels of attachment to their dogs. These findings are discordant with those of Handlin and Nilsson [19
], who reported a positive association between canine oxytocin concentrations and the total MDORS score. A positive correlation has also been documented with individual indicators of the human–dog bond, such as the frequency of kissing the dog [19
]. However, kissing dogs may be a feature of attachment that only certain owners have with their dogs and may not be representative of all owner–dog bonds. Furthermore, some dogs do not enjoy having humans blow or breathe on their faces [28
], so they may self-select out of such face-to-face interactions.
The effects of oxytocin differ based on sex in both human and non-human animals [29
]. Recent reports highlight a positive effect of oxytocin administration on urinary oxytocin concentrations in female dogs only, with males displaying no change in concentration following intranasal administration and a 30-min human–dog interaction [8
]. Several studies have also documented the positive effects of oxytocin administration on human-directed gazing behavior in female dogs only [14
]. Therefore, it is plausible that the present use of an entirely male sample may have influenced results, possibly explaining the null findings. Differences in the samples analyzed, namely, the use of urinary samples compared to the use of plasma samples in most previous studies [12
] may also contribute to the discrepancies between our findings and those of previous studies. Alternatively, it has been suggested that replicating findings from human–animal interaction research is difficult due to small effect sizes, small sample sizes, and the variety of research designs seen throughout the literature [32
]. It is also possible that selective reporting of findings based on their direction and statistical significance may have skewed the research on human–animal interactions [32
]. Therefore, the relatively consistent positive influence of human interaction on canine endogenous oxytocin concentrations seen throughout the literature may, to some extent, be a reflection of publication bias, in which significant results are more likely to be published than null results [33
To our knowledge, this is the first study to examine the influence of human–dog interactions on canine endogenous oxytocin concentrations within the dog’s home environment. It is plausible that human–dog interactions performed in clinical settings for research purposes are not indicative of authentic interactions between an owner and their companion dog. Therefore, testing privately owned pet dogs in the home environment, as in the current design, is likely to better simulate “real-world” conditions. The use of a within-subject, cross-over study design may also be considered a strength of the current study.
However, a number of possible limitations must be noted. Firstly, differences in the mechanisms underlying central and peripheral oxytocin release remain unclear [5
]. Currently, it is unknown how central and peripheral release patterns differ and whether peripheral oxytocin concentrations are indicative of central oxytocin concentrations [34
]. Therefore, findings based on peripheral oxytocin concentrations should be interpreted with caution [34
]. Secondly, it is possible that canine oxytocin responses were influenced by the researcher’s presence during experimental conditions, despite the study protocol having been designed to reduce any arousing effect of the researcher’s arrival at the home and despite the concerted efforts of researchers not to interact with the dog. For example, the researcher’s arrival at the participant’s home, approximately 15 min prior to baseline sample collection, could have elevated baseline concentrations, thereby reducing the observed effect of stimuli on oxytocin concentrations. The use of a convenience sampling method may have produced a sample of dog owners with greater levels of attachment to their dog compared to the general population. Therefore, it is not known whether our results based on a binary divide of participants into “low” and “high” levels of attachment can be extrapolated to the general population. A lack of detailed information regarding the dogs’ breeds, particularly in the case of cross-breeds, also limits the sub-division of dogs into breed groups because this relied on attempts to classify the dogs according to putative breed or cross. Similarly, there are limitations in using the group age median as a cut-off given the differences in life expectancy for various breeds. However, our study did not include any breeds with drastically shorter life expectancies, and we can assume that dogs aged below the median of 7 years were in the early or mid-stages of their life, while dogs aged over 7 years were likely to be in the mid to late stages of life. Finally, due to pressure on resources, urine was collected at the minimum time point of the suggested range (approximately 30 min). Additionally, interactions were not coded, meaning we were not able to examine possible associations between the oxytocin response and the type of interaction or time spent engaged in each interaction behavior.
Future research is warranted to replicate and extend our findings. To begin with, it may be helpful to apply the current methodology to female dogs. Studies investigating the moderating effect of individual differences on the canine oxytocin response would also be of interest, including the influence of genetics, early life experience, and the attachment style of the dog [36
]. Accordingly, previous studies have highlighted an association between genetic variation in the oxytocin receptor gene and human-directed social behavior following intranasal oxytocin administration [37
]. The need for such studies is further supported by recent reports that oxytocin does not elicit purely pro-social or anxiolytic effects on canine behavior. For example, Pekkin and Hänninen [38
] documented a positive association between baseline concentrations of urinary oxytocin and general fearfulness, noise fear frequency, and reactivity.
To conclude, dog-walking and affiliative human–dog interactions did not have an effect on urinary oxytocin concentrations in this sample of male, privately owned, companion dogs. Further research is needed to improve and optimize oxytocin measurement standards and to establish the effect of different types of human–dog interactions on endogenous canine oxytocin concentrations in larger samples including both male and female dogs.