Visual cognitive ability of dog for social stimuli
Dogs have undoubtedly established a niche for themselves in human society, as companion animals that are closest to and the most loved by humans. Recently, it was found that the visual cognitive ability of dogs is superior to that of wolves and apes, in terms of understanding human pointing gestures (Hare, Brown, Williamson, & Tomasello, 2002; Miklósi, Kubinyi, Topál, Gácsi, Virányi, & Csányi, 2003). Dogs' visual cognitive ability is considered to have developed during their domestication (Hare & Tomasello, 2005). Examinations were conducted on different species' understanding of human pointing gestures (domestic animals like goats, horses, and cats included), and some of these yielded positive results depending on the experimental methods used (Kaminski, Riedel, Call, & Tomasello, 2005; McKinley & Sambrook, 2000; Miklósi, Pongracz, Lakatos, Topál, & Csányi, 2005; Miklósi & Soproni, 2006; Udell, Dorey, & Wynne, 2008; Viranyi, Gácsi, Kubinyi, Topál, Belényi, Ujfalussy, & Miklósi, 2008). Perhaps the animal that is organizing a society should maintain the ability for directional detection among conspecifics at some level. It is difficult to discuss the significance of the social cognitive ability of each species simply on the basis of these examination results. However, it was found that when dogs faced difficulties, they immediately looked to their owners for help, whereas wolves attempted to solve problems on their own (Miklósi et al., 2003). Cats, who have also carved a niche for themselves in human society, were found to have similar pointing gesture understanding abilities; however, they did not display certain attention-seeking behaviors that dogs did (e.g. alternating gaze between the direction of the hidden food and the human: Miklósi et al., 2005). Given the more stable results of the examinations on dogs, these studies suggest the distinctiveness of their social cognitive ability with respect to humans.
In investigating the human-dog relationship, the possibility that dogs possess a human-like visual cognitive ability is very important. Each animal has species-specific cues for social bonding. Sharing social cues and distinguishing a particular individual are indispensable to the survival of animals, bringing them physiological and ethological benefits, such as social buffering (see below). In humans, one of the most important cues is the visual cue (e.g. in the human mother-infant relationship (Dickstein, Thompson, Estes, Malkin, & Lamb, 1984). Humans gaze into other human's eyes with particular interest and respond to the direction of other's gaze (Baron-Cohen, Wheelwright, & Jolliffe, 1997; Driver, Davis, Ricciardelli, Kidd, Maxwell, & Baron-Cohen, 1999; Macrae, Hood, Milne, Rowe, & Mason, 2002; Morton & Johnson, 1991). Moreover, human infants are more comfortable maintaining eye contact with adults (Batki, Baron-Cohen, Wheelwright, Connellan, & Ahluwalia, 2000; Farroni, Csibra, Simion, & Johnson, 2002). With respect to dogs, some studies have shown that dogs can understand where human attention is directed (Call, Bräuer, Kaminski, & Tomasello, 2003; Miklósi, Polgárdi, Topál, & Csányi, 2000). Other studies have demonstrated that dogs tended to accept food from people whose faces were visible rather than people who were blindfolded or whose heads were covered with a bag (Gácsi, Miklósi, Varga, Topál, & Csányi, 2004; Virányi, Topál, Gácsi, Miklósi, & Csányi, 2004). Virányi, Topál, Miklósi, and Csányi (2006) suggested the possibility that dogs think that humans do not know what they cannot see. These studies indicate the psychological and behavioral harmony between dogs and human beings. The fact that dogs use gaze to communicate proves their high social cognitive ability, and is very important in investigating the specific relationship between human and dogs.
Autonomic responses to specific stimuli in dogs
The capacity for individual discrimination is essential in the development of long-lasting bonds between mating partners and biological relatives such as mothers and infants, and for maintaining individual life and species life. Intraspecies individual discrimination employs species-specific sensory organs. It is widely known that ewes reject other lambs without bonding with them on the basis of their odor (Keverne & Kendrick, 1992). Primates can discriminate between conspecific individuals using visual cues (Dasser, 1987).
Adachi, Kuwahata, and Fujita (2007) also showed the ability of dogs to discriminate between human beings. They tested whether dogs form a crossmodal representation of human beings. Dogs were shown a monitor with a picture of either their owner or a stranger after playing back a voice recording of one of those persons. They tended to look longer at the visual stimulus when the voice and face did not match than when they did; this suggests that dogs form a multimodal concept of their owners using visual and auditory cues.
These results indicate that dogs can discriminate between their owners and unfamiliar persons. Could it be that dogs are able to not only discern humans individually, but also distinguish a specific person from others as the attachment object? Topál, Miklósi, Csányi, and Dóka (1998) investigated the ability of dogs to discriminate between individuals using the Strange Situation Test (SST; Ainsworth & Wittig, 1969). The SST was originally developed to examine the balance between attachment and exploratory behaviors on the basis of the different reactions of infants, for example, reactions to their mothers compared with unfamiliar persons in a moderately stressful environment. The results showed that, like human infants, dogs behave differently with their owners and with strangers. Topál, Gácsi, Miklósi, Virányi, Kubinyi, and Csányi (2005) also carried out the SST on hand-reared wolves, and found that the wolves did not behave differently with strangers than they did with their owners. Using physiological approaches, we tried to examine whether dogs can discriminate between a particular object, that is, the owner, and others, and to verify whether they can exhibit specific responses. Even though it has been proven that dogs can discriminate between human beings, there is no significant evidence indicating the biological functions of the discrimination. The results of this study provide significant insight into the attachment shown by dogs toward humans from the biological perspective. We investigated whether dogs show specific emotional reactions to their owners, by assaying the heart rate variability (HRV) as an autonomic parameter (Nagasawa, Mitsui, Mogi, & Kikusui, unpublished data, 2008).
The heart rate (HR) is controlled by the sympathetic and parasympathetic autonomic nervous systems. Under resting conditions, the ECG of healthy individuals exhibits periodic variation in beat-to-beat intervals. The HRV refers to the beat-to-beat alterations in the HR. A common frequency domain method is the application of the discrete Fourier transformation to the beat-to-beat interval time series. This provides an estimation of the amount of variation at specific frequencies. A low-frequency band (LF) between 0.04 and 0.15 Hz derives from both parasympathetic and sympathetic activities, while a high-frequency band (HF) between 0.15 and 0.40 Hz is driven by respiration, and appears to derive mainly from the parasympathetic nervous system. The general ratio of LF to HF is considered to be the activity of the sympathetic nervous system (Bootsma, Swenne, Van Bolhuis, Chang, Cats, & Bruschke, 1994; Furlan, Barbic, Piazza, Tinelli, Seghizzi, & Malliani, 2000; Montano, Ruscone, Porta, Lombardi, Pagani, & Malliani, 1994), and HF to reflect the activity of the parasympathetic nervous system (Hayano, Sakakibara, Yamada, Mukai, Fujinami, Yokoyama, Watanabe, & Takata, 1991; Pomeranz, Macaulay, Caudill, Kutz, Adam, Gordon, Kilborn, Barger, Shannon, & Cohen, 1985). Maros, Dóka, and Miklósi (2008) tested the HR and HRV responses in dogs during different physical activities and arousal situations, and found that the HR, but not the HRV, increased during periods of increased physical activity and body position. In contrast, the HRV significantly increased when dogs oriented toward their favorite toy. This study indicated that the body position significantly affected the HR in dogs, and that the HRV may be indicative of their attentive state.
In order to examine the discriminatory ability of dogs, we conducted the habituation-dishabituation test (Halpin, 1974) on dogs, using their owners as the test stimulus and unfamiliar persons as the neutral stimulus. Seven adult dogs and their owners (OW) participated in the study. During the experiments, a dog was kept in an opaque enclosure with a metal grid opening so that the stimuli were visible. Three unfamiliar persons, who had never interacted with the dog before, were shown to each dog as strangers (ST). These unfamiliar persons were the same sex as and of similar appearance to the dogs' owners.
The dogs were introduced into the testing area one at a time, and presented with four stimuli: the first three were denoted as ST1, ST2, and ST3, and the last as OW. After a 5-min resting period, the first stimulus (ST1) was presented for 4 min, followed by a 3-min interval. Then, ST2, ST3, and OW were presented in the same manner. We compared the HR and LF/HF values and the duration of the exhibited behavior between the presentation of ST1 and ST3, and ST3 and OW. The results showed that during the habituation phase (ST1–ST3), the HR and HRV (LF/HF) decreased significantly (ST1 vs. ST3). In the test trial for OW, the HR and HRV significantly increased compared with those for ST3 (Figure 1). These results indicated that presenting the owners as stimuli activated the autonomic system in dogs more than when strangers were presented; this suggests that dogs can not only discriminate between owners and strangers, but also exhibit emotional responses when reunited with their owners. Therefore, dogs have the ability to discriminate between and respond differently to different humans, and this ability fulfills some of the criteria for the formation of a social bond between humans and dogs.
Figure 1. (a) Heart rate variability (HRV) and (b) heart rate (HR) of dogs in the habituation-dishabituation test of attachment between humans and dogs. The dog's owner was assigned as the test stimulus (OW) and unfamiliar subjects were the neutral stimuli (ST1, 2, 3). Both HRV and HR gradually decreased during the habituation phase (ST1, 2, 3), and increased significantly in the test phase (OW). *P < 0.05.
Download figure to PowerPoint
Neuroendocrine mechanisms for attachment
Attachment, which has significant biological implications in each species, may possibly activate the internal mechanism as a vital reaction and homeostatic function (Hofer, 1995; Kraemer, 1992). Furthermore, the attachment system involves the interactions initiated by the infant's attachment behavior, followed by the mother's care-giving behavior and subsequent interactions. Social signals (social cues) from members in the dyad, and the subsequent endocrinological changes, can activate the attachment system, which not only increases the survival probability of infants, but also contributes to the enhancement of their parents' biological fitness. However, because the biological social attachment of both human and dogs has not been manifested yet, we have to apply the results from studies in other species.
In animal experiments, the meanings of attachment and bond based on attachment are more critical and clearly described. When a social bond is broken, animals exhibit separation anxiety and the stress-endocrine response is activated. In contrast, in polygamous mating pairs, for example, separation does not elicit the stress-endocrine response. This kind of relationship (called social affiliation) has to be distinguished from attachment and bond (DeVries, 2002). It is suggested that attachment is not only related to the behavioral control system for avoiding the danger of predation, but is also closely related to the feedback system of the neuroendocrinological system.
It is necessary to form and maintain a social bond with a particular object in order to maintain a stable social life, both mentally and physically (Kikusui, Winslow, & Mori, 2006). Social bonding refers to various relationships, such as between a mother and infant, mating pairs, and conspecific companions. Attachment between bonding partners has stress-buffering effects and is critical for the survival of many social species. As mentioned earlier, attachment, as defined by Bowlby on the basis of the behavioral control system, is the safety regulation system. The social buffering effects observed in dyadic relationships comply with Bowlby's definition. Studies investigating the social buffering effects of the stress-endocrine response have focused on the hypothalamo-pituitary-adrenal (HPA) axis. The first evidence of social buffering in nonhuman primates was demonstrated in mother-infant bonding in squirrel monkeys (Coe, Mendoza, Smotherman, & Levine, 1978; Mendoza, Coe, Lowe, & Levine, 1978; Wiener, Johnson, & Levine, 1987), and details of these findings were later reviewed by Levine and colleagues (Levine, 2000, 2001; Levine, Lyons, & Schatzberg, 1997). In the mother-infant separation paradigm, the cortisol response was observed in infants, but was completely attenuated when the infants were reunited with their mothers, in parallel with the decrease of distress-like behavior. Being in a dyadic relationship has social-buffering effects, not only in the case of social separation stress, but also when faced with natural threats. For example, it has been found that systematic aversive stimulation increased cortisol levels in squirrel monkeys and that the presence of a familiar partner decreased the neuroendocrine response to the stimuli, as demonstrated by the classical fear-conditioning paradigm (Stanton, Patterson, & Levine, 1985). In this paradigm, the monkeys were first conditioned to the cue stimuli (lighting) and aversive stimuli (shock). Thereafter, the monkeys were exposed to the cues, and an increase in the cortisol levels was observed. However, if a monkey was with a familiar partner, there was a lesser increase in the cortisol levels and fear responses. These data indicate that the presence of social partners ameliorates the neuroendocrine response to various types of stressors.
The neuropeptide oxytocin is the key mediator for bonding between infant-mother dyads and mating pairs (Carter, Williams, Witt, & Insel, 1992). Moreover, an increase in oxytocin levels can have social buffering effects in the dyad (Kikusui et al., 2006). For example, the oxytocinergic system in ewes regulates the formation of memories for specific maternal behavior toward their own lamb (Keverne & Kendrick, 1992). Similarly, in rats, vaginocervical stimulation during parturition can activate oxytocin release and facilitate the onset of maternal behavior (Pedersen, Ascher, Monroe, & Prange, 1982). Oxytocin is also involved in infants' memory formation of the caregiver (Insel & Young, 2001; Nelson & Panksepp, 1996; Wilson & Sullivan, 1994), and in the process of pair-bond formation after the mating of monogamous species such as prairie voles (Young & Wang, 2004; Young, Wang, & Insel, 1998). In contrast, these behaviors are inhibited by oxytocin antagonists administered to prairie voles just before mating. In sum, oxytocin is the most important neurotransmitter that is responsible for social bonding.
Oxytocin has been suggested as a candidate molecule for the transmission of social-buffering effects. During physiological and psychological stress responses, oxytocinergic neurons are activated, particularly in the paraventricular nucleus of the hypothalamus, and secreted into the blood (Ludwig, 1998; Neumann, 2002). Increases in oxytocin release from the hypothalamus inhibit stress neuroendocrine activity (DeVries, Glasper, & Detillion, 2003; Neumann, 2002; Neumann, Kromer, Toschi, & Ebner, 2000). Oxytocin also causes behavioral modifications to the stress response. In experimental rodents, it has been demonstrated that oxytocin has anxiolytic properties and plays a significant role in the stress-coping style (Ebner, Bosch, Kromer, Singewald, & Neumann, 2005; Haller, Makara, Barna, Kovacs, Nagy, & Vecsernyes, 1996; Neumann, 2002). It also induces excessive grooming behavior when injected into the brain (Pedersen, Caldwell, Drago, Noonan, Peterson, Hood, & Prange, 1988). Moreover, it has been suggested that central oxytocin is released in response to physical contact (Uvnas-Moberg, 1997).
A study on humans showed that breastfeeding after physical exercise attenuated the release of cortisol as well as adrenocorticotrophic hormones (ACTH) into the bloodstream (Altemus, Deuster, Galliven, Carter, & Gold, 1995). Recent studies have shown that the intranasal administration of oxytocin, which passes directly into the brain (Born, Lange, Kern, McGregor, Bickel, & Fehm, 2002), suppressed cortisol response to psychological stress as well as attenuated emotional functions after stress episodes (Heinrichs, Baumgartner, Kirschbaum, & Ehlert, 2003). Therefore, in humans, the inhibitory effect of oxytocin on HPA activity is also mediated in the central nervous system. In sum, these results indicate that oxytocin attenuates not only the HPA axis response to stress, but also reduces negative emotions. Although there is no direct evidence that oxytocin mediates social bonding in humans, some studies have suggested oxytocin related infant-mother bonding and social relationships (Fries, Ziegler, Kurian, Jacoris, & Pollak, 2005; Heinrichs et al., 2003; Kosfeld, Heinrichs, Zak, Fischbacher, & Fehr, 2005) Thus, oxytocin plays an essential role in the neural mechanisms of social bonding and stress buffering.
Oxytocin in human-dog bonding
The visual cognitive ability, particularly gazing, is the most fundamental and important social cue in humans. Visual communication abilities are considered to have developed in dogs as a by-product of their domestication, and as a behavior required for a symbiotic relationship with humans (Hare & Tomasello, 2005). It is possible that dogs, like humans, use gazing as a social cue. Focusing on the gaze of dogs, we constructed a hypothesis that the concentration of oxytocin in the urine of dog owners is affected by their dog's gaze, which functions as an attachment behavior. To test this hypothesis, the urinary oxytocin concentration of dog owners was measured before and after 30-min interactions with their dogs (Nagasawa, Kikusui, Onaka, & Ohta, 2009).
The participants in this study were 55 volunteers and their dogs. Prior to the experiment, the participants' degree of relationship with their dogs was measured on the basis of two questions: “How much are you satisfied with your dog?” and “How much do you feel you can communicate with your dog?” The answers to these questions were rated on a 5-point scale.
Two types of experiments, interaction experiment and control experiment, were carried out per participant on different days. The interaction experiment consisted of three settings: a 20-min rest period, a 30-min testing period involving normal interaction with the dogs, and a 20-min rest period after a 30-min testing period. During the testing period, the participants were allowed to freely interact with their dogs. Before and after the testing period, urine samples were collected from the participants. In the control experiment, the participants were forbidden to look at their dogs directly. They were made to sit facing a wall, which inhibited the gaze stimuli from the dog to the owner.
Using a cluster analysis, the owners were divided into two groups: one group received a longer gaze duration from their dogs and reported a higher degree of relationship with their dogs (LG), while the other group received a shorter gaze duration and reported a lower degree of relationship (SG). The urinary oxytocin levels were higher in the LG group than in the SG group after a normal interaction with their dogs (interaction experiment); however, this was not the case in the control experiment. In the interaction experiment, a high correlation was found in the LG group between the frequency of behavioral exchanges initiated by the dogs' gaze and the increase in urinary oxytocin levels, suggesting that the dogs' gaze stimulated the owner's oxytocinergic system.
These results indicated the strong possibility that oxytocin plays an important role in human-dog bonding. It is still unclear whether a dog's gaze at its owner is innate or learned, and whether the dog's gaze has different functional meanings for humans and conspecifics. However, the human response to attachment-related signals from other species can possibly be explained by human nature. Humans tend to anthropomorphize other animals and objects (Albert & Bulcroft, 1987; Salmon & Salmon, 1983). Human infants react to the motion of a robot by exhibiting communicative gaze behavior (Itakura, Ishida, Kanda, Shimada, Ishiguro, & Lee, 2008). One can assume that such behavior is very similar to that observed in human infants; hence, an owner who is gazed upon by his or her dog speculates about the dog's emotional status and regards the gaze as attachment behavior. Therefore, a dog's gaze is considered to be a significant cue for social contact in humans.
The results of this study suggest that, of all the interactions observed between dogs and their owners, a dog's gaze, as a factor that contributes to social bonding, has a particularly strong effect on its owner's neuroendocrine system. In humans, only indirect findings of the relation between oxytocin secretion and attachment have been reported (Fries et al., 2005). It is not yet clear whether oxytocin concentrations in the urine reflect the activity of the central nervous system; however, it has been shown that there is a strong positive relationship between the oxytocin concentration in the urine and that in plasma (r = .89: Amico, Ulbrecht, & Robinson, 1987), and that plasma oxytocin is related to hypothalamus activity (Ludwig, Callahan, Neumann, Landgraf, & Morris, 1994). Based on these data, it is likely that urinary oxytocin, at least to some extent, reflects the hypothalamic oxytocinergic activity.
This study provides a clue to the neural mechanisms through which interacting with dogs affects human physical and mental health. It is said that animals have species-specific attachment styles. This study suggests that humans and dogs may have a common attachment style, and this may partially explain why dogs can adapt to human society.