SOCIAL REFERENCING AND UNDERSTANDING OF HUMAN EMOTIONAL EXPRESSIONS IN DOGS

Summary Social referencing and understanding of human emotional expressions in dogs. Introduction The literature on dog cognition and in particular social cognition has grown incredibly in recent years, however no studies have focused on the process of Social referencing in dog-human communication and interaction, and only few preliminary studies have investigated dogs’ understanding of human emotions. Social referencing is a process characterized by the use of another person’s perceptions and interpretation of a situation to form one’s own understanding and guide action. Human infants use this process to go beyond the information given by an informant to construct a more general interpretation of the meaning of a stimulus and they can successfully link the emotions expressed to a referent object. Social referencing has been investigated not only in humans, but also in a few primate species, namely chimpanzees and barbary macaques, although with mixed results. In fact, results from primate studies are overall ambiguous, with the strongest evidence of social referencing coming from human-raised chimpanzees, where it is exhibited with their human caretaker. There is, however, more consistent evidence that non-human primates can refer conspecific and human emotional expressions to a referent object. The dog-human relationship is a very special one and the more recent literature shows that the dog-human bond is similar in many respect to an infantile attachment. Furthermore, studies suggest that dogs are sensitive to a number of behaviours potentially revealing the person’s attentional states; can follow a number of human referential cues, and, preliminary evidence, seems to suggest that they can also discriminate between some expressions of human emotions. Finally there is also some evidence that dogs can communicate intentionally and referentially with humans (although this is still a mater of some controversy). Considering the above, dogs are particularly good candidates for the comparative investigation of social referencing and of the ability to understand human’s emotional expressions as referring to specific objects. Four studies are presented in this thesis which aimed to investigate Social referencing in dog-human dyads and dogs’ capacity to understand the referential nature of a person’s emotional message. The aims of the present research were to add to the literature on dogs’ socio-cognitive abilities and of human-dog communication by studying (1) the presence of Social referencing, both referential looking and behavioural regulation, in dogs towards humans, (2) the potential selectivity of this process, based on the relationship with the informant (owner vs. stranger), (3) the effect of a particular kind of training experiences (i.e. water rescue training) on this process, and (4) the ability of dogs to refer different emotional expressions toward two objects. To answer these questions we carried out four different experiments: Study 1 Social referencing in dog-human dyads was investigated using the “new object paradigm”. As no study has been carried out on dogs on this topic so far, we set up a new procedure suitable to this species, that was similar to the one used in the infant. In particular in our “new object paradigm” we presented dogs with a new and potentially scary object (a fan with plastic ribbons attached to it) in presence of their owner as the informant. The aim was to evaluate whether, in a social referencing paradigm, dogs would show referential looking and behavioural regulation toward the owner acting as the informant and hence approach the object more having witnessed a positive vs. a negative message. We tested 75 dog-owner dyads at the Canis Sapiens Lab of the University of Milan. There were two different groups of dogs: in one group the owner was delivering a positive emotional message towards the object, whereas in the second group the owner delivered a negative emotional message. The results of this first study showed that dogs, like human infants, use referential looking towards a familiar person (their owner) in a situation of ambiguity. However, differently from infants, dogs showed no clear evidence of behavioural regulation after receiving an emotional message from the owner. Study 2 Since the results obtained in the first study showed a clear presence of referential looking in dogs, but not clear evidence of behavioural regulation toward the owner, we modified the testing procedure to further evaluate behavioural regulation. We investigated the presence of Social referencing in dogs with the same potentially scary object (a fan with plastic ribbons attached on it) in presence of their owner vs. a stranger. The study had two main aims. First, since results obtained in the first study provided only unclear evidence of behavioural regulation toward the object, we wanted to see whether using a procedure more closely resembling the one used with infants we would be able to find evidence of behavioural regulation. The second aim was to assess the influence of the informant’s identity on social referencing. Selectivity is an important aspect of children’s Social referencing: they seem to use referential looking toward both a familiar and an unfamiliar person, but they regulate their behaviour only when a familiar person is the informant or when she/he is present in the experimental set-up together with the unfamiliar informant. We tested 90 dog-owner dyads and there were four groups of dogs: two groups were tested with the owner as the informant (either expressing a positive or a negative emotion), and two were tested with a stranger as informant (either expressing a positive or a negative emotion). Results provided clear evidence that dogs use referential looking not only towards their owner but also towards a stranger, with no difference between the two persons. Furthermore, dogs regulated their behaviour towards the object after receiving a positive or a negative emotional expression by the owner, but not when the stranger was acting as the informant. Study 3 Since in Study 2, besides referential looking, we found a clear evidence of behavioural regulation with the owner, whereas no clear results emerged with the stranger acting as the informant, we tested dogs exposed to a particular type of training: water rescue dogs. In fact, during their training these dogs become used to focusing their attention on unfamiliar person. Thus we investigated the presence of Social referencing in dog-human dyads using the same procedure and scary object as in study 2 but with a stranger acting as the informant. For this study we tested 22 dog-owner dyads: the group of trained water rescue dogs were tested with their owners in Naples (Università degli Studi di Napoli “Federico II”), while the breed, age and sex matched control group was tested in Milano at the Canis Sapiens Lab of the University. Dogs of both groups showed referential looking toward the stranger and this result supports those obtained in study 2. However, highly trained water rescue dogs did not change their behaviour towards the object (behavioural regulation) when the stranger was exhibiting a positive emotional expression towards it. Surprisingly, the control (untrained) group dogs did change their behaviour toward the ambiguous object, approaching it more than the trained dogs when the positive emotional message was given by the stranger. These findings are discussed in relation to our previous results (in Study 2) where no such effect was found with the stranger acting as the informant (we believe the likely cause is the breed of dogs tested in Study 3) and in relation to the potential inhibitory effect of training on dogs’ behavioural regulation in this context. Study 4 In the previous three studies dogs were tested always in the presence of only one object (the potentially scary fan), and, although the informant’s message always referred to that object, the goal of these studies was not strictly to assess whether dogs were capable of appropriately referring the human’s emotional expression to the object itself. So far only one study has tested dogs’ ability to attribute a human’s emotional reaction (facial expression and short vocalization) to a specific object (i.e. to grasp that emotions can be referred to something in the outside world). Thus the aim of this study was to assess dog’s understanding of human emotional expression as referential. In particular, we evaluated if dogs can discriminate between three different emotional expressions (fear, happiness and neutral) and whether they have a perception that a human’s (owner vs. stranger) emotional expression can refer to specific objects in the environment. We tested 95 dog-owner dyads at the Canis Sapiens Lab of the University of Milan. We adapted a procedure used with infants, and more recently also with chimpanzees and dogs, in which the informant expresses two different emotions towards two identical (hidden) objects. In a control group, the same procedure is adopted but in the absence of the objects (“no-object” condition). After observing the emotional expressions being conveyed by the informant (owner vs. stranger) dogs were free to approach the objects. Dogs showed a clear preference for the hidden object eliciting the positive emotion, compared to one eliciting the negative one when the owner was acting; on the contrary no preference emerged when the stranger was acting as informant. Furthermore, dogs didn’t show a choice behaviour when the owner was expressing the emotion in the “no-object condition”. A follow-up study, contrasting the positive and negative emotion in turn with a neutral one, showed that dogs tended to approach the object eliciting a positive emotion rather than avoid the one eliciting a negative reaction. Taken together these results, show that dogs do in fact appreciate that an emotional message can relate to a specific object if the owner is the informant but they do not do so if the stranger is the informant. Furthermore, their performance seems to be based on approaching the positive stimulus rather than avoiding the negative one. Together these results suggest that prior experience with their owner using positive emotional expressions, has allowed dogs to associate these emotions to particular objects. Finally, considering the non-specific behaviours exhibited by dogs in the no-object control group, dogs may even have come to expect that an emotional message refers to a specific object. Conclusions The experiments presented in the current thesis reveal some new and interesting aspects of dog social cognition and communication with humans. On the one hand they provide the first evidence of social referencing in dogs: they show that dogs, like infants, can show referential looking and behavioural regulation in ambiguous situations when the emotional message is delivered by the owner, or by a stranger (in presence of the owner). Finally, these findings suggest that dogs have some understanding that emotional expressions are referential, in the sense of being directed to specific stimuli in the environment as has been reported for infants and non human primates.