Being able to display social behavior is key to the welfare of domesticated animals, who are all social species. Oxytocin (OT) has received increased attention for its involvement in the proximate mechanisms of social behavior, offering opportunities to elucidate the perception or processing of social stimuli.

This review aims to critically evaluate the validity and robustness of OT as an indicator of animal welfare related to the social environment. We focus on the primary literature on OT in domesticated animal species, because although OT has been well researched in human, non-human primates and rodents (Winslow et al., 2003; Neumann, 2009; Cavanaugh et al., 2016; Freeman and Young, 2016), an emerging theory is that domestication may have influenced the oxytocinergic system (Nagasawa et al., 2015), similarly to the effects of domestication on quantitative behavioral changes (Price, 2002). We discuss findings on the impact of experimental manipulations on endogenous OT concentration, differences in OT receptor gene polymorphisms and OT administration on behavior and human-animal interaction. We excluded the special case of maternal behavior, covered by previous reviews (Neumann, 2009; Kim and Strathearn, 2016), and studies of human-animal interaction focused on the human experience. This review does not intend to exhaustively cover the literature on OT in domesticated species but rather focuses on aspects relevant to behavior and welfare, highlighting findings and gaps in research. Approaches to study animal welfare and animal welfare assessment are covered elsewhere (Fraser, 2008; EFSA, 2012).

We found 32 relevant studies on OT in domesticated species, with 78% of them published in the last 5 years (Table 1). Studies used different methods: measures of central OT (in cerebrospinal fluid 3%, in brain sections 3%) or peripheral OT (in blood plasma or serum 31%, in urine 13%), administration of exogenous OT to study the animal's response (47%), study on brain OT gene expression (3%), and more recently OT receptor gene polymorphisms (13%), with 9% of studies using more than 1 approach. For studies that measured endogenous OT, 47% took a pre- and post-stimulus sample, 40% took repeated samples during the presentation of the stimulus, and 13% only took a sample at 1 time point (comparing between individuals). Only 22% of studies had a control group, whereas the rest used within-subject designs.

Average sample size was 48 subjects, ranging from 5 to 207 subjects. Most studies used dogs (53%: 44% mixed breeds and 9% single breed, and 6% with wolf as a comparison), then pig (22%), cattle (13%), sheep (9%), and cat (3%). Studies on dogs involved a variety of adult age dogs (over 12 months) whereas studies on farm animals involved mostly young, pre-pubertal weaned subjects. As mentioned earlier, we excluded studies of maternal behavior. Studies involved mixed sexes (56%, intact or castrated), females only (34%), males only (6%), or unreported (3%). The more pronounced effects of OT administration in females than males is well-recognized (Rault et al., 2013a; Nagasawa et al., 2015; Kovács et al., 2016; Oliva et al., 2016a), but whether sexes differ in endogenous OT remains unclear as the few studies that included both sexes did not report testing for a sex effect or individual OT profiles or had insufficient sample size.

The majority of OT studies were on human-animal interaction (51%), demonstrating that OT's function cross species boundaries. The rest was composed of intra-species interaction (22%) and social isolation (27%), with 24% of studies using more than 1 paradigm. Furthermore, a variety of experimental testing conditions have been used (Table 1).

The following sections explore the premise of OT as an indicator of social, positive and psychological well-being in domesticated animals based on the research to date and the main factors identified for the scope of this review. We discuss research from different animal species when available, but do not assume results should be similar across species given different ethological or evolutionary importance, which we refer to as species-specific social behavior.

A large number of studies investigated the impact of social stimuli on endogenous OT concentration. A range of social settings trigger an OT response; the presence and magnitude of the response depending on a range of experimental factors such as familiarity of setting and partner, voluntary contact, and form of contact.

In the quest for indicators of positive welfare states, OT is often proposed to reflect situations of positive valence. However, few studies have compared positive to negative or neutral situations. For instance, that urinary OT increases in three positive situations does not prove OT as a “biomarker of positive emotions” (Mitsui et al., 2011) unless a non-positive situation would have been included, although cortisol was included as a measurement of arousal.

Environmental context can modulate OT's actions. For instance, OT administration promoted positive social behaviors of dogs toward both their owners and familiar dogs (Romero et al., 2014), but reduced friendliness toward the owner in the presence of an approaching stranger, as discussed earlier (Hernádi et al., 2015). Opposite findings were found in pigs, in which OT administration in familiar groups reduced social contact in neutral or positive situations but increased it in negative situations (Camerlink et al., 2016). Conversely, CSF (endogenous) OT increased in pigs following positive human interaction, but not negative human interaction (Rault, 2016), although valence and familiarity of the partner were confounded.

The social context (e.g., partner familiarity) may also modulate OT's actions. Calves with high basal plasma OT postnatally showed higher social engagement, both affiliative and agonistic behaviors, in later life (Yayou et al., 2015), and exogenous studies showed that OT administration can increase aggression in pigs (Rault et al., 2013a, 2015). However, these studies involved animals mixed with unfamiliar conspecifics and in unfamiliar environments, i.e., stressful situations. Altogether, these findings are consistent with the in-group vs. out of group OT theory in humans (De Dreu, 2012), with OT's positive actions toward existing social partners and negative actions toward unfamiliar partners.

In summary, OT does not necessarily correlate with positive situations or outcomes. The OT literature appears biased toward investigating positive valence, with a lack of controlled paradigms for valence and arousal. There is evidence that negative situations also mobilize OT. We propose that OT may be evolutionarily linked to social coping strategies (Buisman-Pijlman et al., 2014; Cavanaugh et al., 2016), as the social arm of homeostatic processes, and as such neither positive nor negative but simply adaptive. The valence of OT's actions are modulated by the environmental and social contexts, and OT's theoretical function of preserving existing social bonds (Tops et al., 2014). Environmental and social factors are therefore important to consider in study design and interpretation (Olff et al., 2013).

Oxytocin is often referred to as the “feel-good” hormone, or as an indicator of positive emotions (Mitsui et al., 2011). Rodent and human data highlight the effect of exogenous OT in increasing trust and reading of social cues, reducing anxiety and other psychological processes (Lee et al., 2009). There is no direct neurobiological evidence yet in domesticated species to support the role of OT in psychological, and particularly emotional, processes. Studies extrapolate their findings to psychological implications based on analogy with human studies (Mitsui et al., 2011). However, OT's role in human psychological processes is still debated (Nave et al., 2015). Particularly, whether the affective “feel good” effect is a direct or indirect effect of OT is unclear, given that OT antagonists do not block these effects (Uvnas-Moberg, 1998) and that the oxytocinergic system interact with other reward systems, notably opioidergic and dopaminergic systems that also increase in response to social interactions (Odendaal and Meintjes, 2003; Buisman-Pijlman et al., 2014; Tops et al., 2014) and impact on the HPA axis (Buisman-Pijlman et al., 2014; Tops et al., 2014).

Most of the knowledge in psychology is about the effect of intranasal OT administration, rather than correlative studies between endogenous OT and psychological states. Interestingly, OT administration induces a positive cognitive bias in dogs to ambivalent food cues (Kis et al., 2015).

The stage at which OT affects socio-cognitive processes currently debated in humans (perception vs. processing of social cues) has been followed up in dogs, with OT administration posited to reduce the attentional bias to social cues (Kovács et al., 2016), whereas others argue that OT does not alter perceptual salience of social cues or social anxiety but rather motivates social engagement (Romero et al., 2014).

The social motivation vs. social reward hypothetical functions, which appears in the human literature, is also relevant to domesticated animals. The hypothesis that OT increases social motivation is supported by exogenous OT studies, with dogs administered OT initiating more contact toward a familiar dog and owner (Romero et al., 2015), even when owners were instructed to ignore or only briefly reciprocate (Romero et al., 2014; Nagasawa et al., 2015). The hypothesis that OT conditions the rewarding value of social cues is supported by endogenous OT studies, where the failure from humans to reciprocate contact results in lower plasma OT concentration over time compared to the initial reunion (Rehn et al., 2014), but no change in urine OT (Nagasawa et al., 2015). More frequent measurements of OT over time could allow discerning appetitive from consummatory motivations.

In summary, there is currently limited evidence that OT's actions are linked to psychological states. Nevertheless, it remains to be demonstrated that it is a direct effect of OT per se. This is a worthwhile area of research given the increasing interest in affective states (feelings, emotion, and cognition) in psychology and animal welfare science.

While findings are coming at a quick pace, the few and discrepant findings make it premature to conclusively decide on OT's potential as an animal welfare indicator.

Focusing on the biological significance of OT in the regulation of psychological and behavioral states may help reconcile findings. A greater understanding of the effects of genetic, epigenetic and ontogeny on the oxytocinergic system is highly relevant to domesticated animals. Accumulating evidence in other species also shows that OT's actions are moderated by context and inter-individual differences. This is determinant to the use of OT as an animal welfare indicator sensitive to the state of interest and robust to extraneous factors. Furthermore, classic parameters for animal welfare measures such as sensitivity, specificity, and repeatability remain to be tested. Indeed, research on OT in domesticated species brings the advantage of potentially well-controlled experiments. It also has direct implications for animal welfare given the importance of social factors and the ability for human management practices to include situations conducive to OT system's development and stimulation.

Mv screened the existing literature and drafted Table 1. JR and FB analyzed and interpreted the literature database and wrote the draft of the manuscript. JR, Mv, and FB reviewed and approved the final manuscript.