656329
research-article2016
CDPXXX10.1177/0963721416656329HuberHow Dogs Perceive and Understand Us
How Dogs Perceive and Understand Us Ludwig Huber
Current Directions in Psychological Science 2016, Vol. 25(5) 339–344 © The Author(s) 2016 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0963721416656329 cdps.sagepub.com
Messerli Research Institute, Vienna, Austria
Abstract In recent years, researchers have become increasingly interested in how dogs understand us humans, given that they show impressive abilities for interacting and communicating with us. Such understanding of heterospecifics is especially interesting because decoding of social signals across the species boundary is challenging. Latent learning during a pet dog’s life in the human environment seems to be a major promotor of this ability. This article reviews recent research on one aspect of this ability: the reading of the human face for the acquisition of important social information, such as identity and emotional expression. Our knowledge of how dogs perceive the human environment and use this information to solve their everyday problems is important for understanding why they fit so well into the human environment. Keywords dogs, cross-modal representation, faces, emotion Among the many research topics that have been addressed using domesticated dogs as study subjects has been how dogs are able to understand us humans. To understand this ability, the phylogenetic and ontogenetic origins of dogs’ interspecies communication abilities need to be considered, from the perspective of domestication as well as individual cognitive development. The success of dogs among humans, including their adoption of the numerous roles humans give to them, likely depends on an interaction of both kinds of developmental factors. On the one hand, one may assume that dogs have acquired a special sensitivity toward human gestures, speech, and behavior as a result of human selection over thousands of years. On the other hand, dogs collect an enormous amount of experience during their life with humans, which is often characterized by close, intimate relationships. An understanding of how dogs perceive elements of their environment and use this knowledge to make decisions is important for disentangling the two kinds of developmental factors (Bensky, Gosling, & Sinn, 2013).
The Visual Abilities of Dogs This article concerns the perceptual basis of dog-human understanding restricted to the visual modality. Of course, from a purely physiological point of view, this would appear to be paradoxical, given that dogs are well known
for their other sensory capacities, such as olfaction and hearing. In comparison, the dog’s visual system may seem poorly developed. However, this does not mean that it is deficient. It has evolved to exploit a particular ecological niche through enhanced visual performance under low-light conditions while still retaining good function under a wide array of lighting conditions, including daylight. The retina of the dog contains about 150,000 ganglion cells. Because the dominant photoreceptors in dogs are rods (only 3% of their photoreceptors are cones), which are particularly well suited to detecting motion and shape, it is probable that dogs are much more sensitive to moving objects than they are to stationary ones (Miller & Murphy, 1995). The adaptation to lowlight conditions may have resulted in reduced color perception in the ancestors of dogs, the latter of which have dichromatic color vision. The two classes of cone pigments have spectral peaks at 429 and 555 nm, respectively (Neitz, Geist, & Jacobs, 1989). We therefore should not expect severe physiological limitations on dogs’ ability to discriminate or even classify colored stimuli, provided the discriminatory cues are not restricted to shades of red or tiny fragments of pictures. Corresponding Author: Ludwig Huber, Messerli Research Institute, 1210 Vienna, Austria E-mail:
[email protected]
340
Multimodal Representations An interesting question is whether dogs can integrate information across modalities. The first study addressing this question investigated whether dogs can form visual-auditory representations of their owners (Adachi, Kuwahata, & Fujita, 2007). The authors used the auditory- visual expectancy- violation procedure (Aronson & Rosenbloom, 1971) to examine whether dogs could recall their owner’s face upon hearing the owner’s voice. A photograph of either the owner’s face or an unfamiliar person’s face was presented to the dog after a vocalization was played. If hearing the owner’s voice activated the visual representation of the owner’s face, but the face of an unfamiliar person appeared (incongruent condition), dogs should be “surprised” (i.e., their expectation should be violated) and they should exhibit extended looking. If, however, the vocalization and face matched (i.e., came from the same person; congruent condition), the dog’s expectation should be confirmed and the duration of its gaze should be comparably briefer. This is what was found. If dogs are able to form a bimodal (auditory plus visual) representation of humans, they may also be able to do that with conspecifics. Indeed, this has been shown in two studies using a preferential-looking paradigm (Faragó et al., 2010; Taylor, Reby, & McComb, 2011). Both studies examined whether dogs are able to spontaneously relate information about body size from the acoustic domain to the appropriate visual category. In agonistic contexts, such as when a dog is defending its bone, its growl will contain information about its body size. Faragó and colleagues (2010) projected two pictures of a growling dog. One picture matched the size of the dog; the other was either 30% larger or 30% smaller. The dogs looked sooner and longer at the matching pictures than the nonmatching pictures, whereas control stimuli (pictures of cats, geometric figures) failed to generate such looking differences, which suggests that dogs utilize a mental representation of the signaler after hearing its vocalization. A later study using stuffed models of differently sized dog breeds confirmed these findings, which is important given the potential methodological limitations of using video displays with nonhuman subjects (Taylor et al., 2011).
The Perception and Discrimination of Human Faces Faces are an important visual category for many taxa because they differ in subtle ways and possess many idiosyncratic features, thus providing a rich source of perceptual cues. For instance, dogs can rely on human facial expressions when making decisions about approaching other objects (Merola, Prato-Previde, & M arshall-Pescini,
Huber 2012), and their gazing behavior has analogies with that of primates. They spontaneously focus attention on informative objects, such as eyes, and prefer looking at upright over inverted faces and familiar over strange faces (Somppi, Törnqvist, Hänninen, Krause, & Vainio, 2012, 2014). In a recent study, Huber, Racca, Scaf, Virányi, and Range (2013) examined dogs’ ability to discriminate the faces of people by active choice (approaching and touching) rather than by (passive) looking. The humans in this study were familiar to the dogs, one being the owner and the other being a highly familiar person; thus, the discrimination couldn’t have simply been based on familiarity. In successive stages of the experiment, the humans were shown in increasing abstractness: first their whole body as they sat on the floor; then only their head, looking through a hole; then only a picture of their face, projected onto a big screen; and finally face pictures in which the outer parts of their faces were occluded with a balaclava hood (Fig. 1). The dogs showed large interindividual and interstage variance in performance, indicating differences across dogs in their learning ability as well as their selection of discriminative cues. The performance of the group decreased when dogs were presented with pictures of human heads after having learned to discriminate the real heads, and when—after relearning—they were confronted with pictures of the same people that showed only the central parts of their faces (eyes, nose, mouth). Only two dogs mastered the final stage. Therefore, it was reasonable to conclude that dogs are in principle able to discriminate people on the basis of visual information from their faces and by making active choices. A subsequent study that used a touch screen as the presentation device (Range, Aust, Steurer, & Huber, 2008; Steurer, Aust, & Huber, 2012) corroborated this conclusion (Pitteri, Mongillo, Carnier, Marinelli, & Huber, 2014). In the first experiment, dogs were required to discriminate between two simultaneously presented humans, their owner and a stranger, of which only the eyes, the nose, or the mouth was shown. All 7 dogs acquired this task, but discrimination of the eye region required fewer sessions compared to both the mouth and the nose region. When asked to discriminate between the two persons on the basis of the whole face, learning was significantly faster when the face had no covered parts than when the eye region was covered, but not when the nose region or the mouth region was covered. This finding suggests that dogs rely less on the nose or the mouth than on the eyes for human face discrimination. The fact that the eye region was easier to discriminate than either the nose or the mouth region confirms the importance of human eyes for dogs. Dogs follow the human gaze (Wallis et al., 2015), and they judge the attentional state of their owners by using eye contact and eye orientation as cues (Schwab & Huber, 2006).
How Dogs Perceive and Understand Us
a
341
b
Fig. 1. The stimuli used by Huber, Racca, Scaf, Virányi, and Range (2013). Stimulus pairs consisted of the subject’s owner and another highly familiar person. In the first and second training phases, dogs discriminated these people while they were seated on the floor and looking through a hole, respectively. In the third training phase, pictures of the two individuals’ faces (a) were projected onto a big screen; in the fourth training phase, only the inner face area (“balaclava condition”) was shown (b).
A second experiment aimed at examining the relative importance of configural and part-based processing. For this purpose, the faces of the owner and the stranger from Experiment 1 were shown in a grayscale and a blurred condition, both of which provided configural information, as well as in an inverted and a scrambled condition. The 7 experienced dogs from Experiment 1 generalized well to the novel faces in all four conditions. However, a group of 7 experimentally naive dogs, when tested with the same picture manipulations after being trained to discriminate the intact faces, showed a significant drop in performance for the scrambled and inverted faces but less of a drop for grayscale and blurred faces. The authors suggested that the performance of these dogs was affected by the loss of configural information through inversion and pixilation (Pitteri, Mongillo, Carnier, Marinelli, & Huber, 2014). Such a decrement of
face-discrimination ability in inverted and scrambled conditions has been known in humans for some time, but it has also been demonstrated in primates, sheep, and pigeons (reviewed in Parr, 2011). In addition, in dogs, the configural face bias may be grounded in an overall preference for global over local information in the processing of visual stimuli (Pitteri, Mongillo, Carnier, & Marinelli, 2014).
The Perception of Human Emotions Dogs appear to get information about more than a person’s identity from looking at his or her face. It has been suggested that the increased readiness of dogs to look at the human face provides the basis for complex forms of dog-human communication (Miklósi et al., 2003). By monitoring human faces, dogs seem to obtain important
342 social information, ranging from communicative gestures to attentive states (Kaminski & Nitzschner, 2013; Schwab & Huber, 2006). Although these competences are well supported by empirical evidence, the evidence for the ability of dogs to discriminate between human emotional expressions is ambiguous. On the one hand, differential gaze bias was observed when dogs were presented with dog faces with threatening or friendly expressions, but not when they were shown human faces with angry or happy expressions (Racca, Guo, Meints, & Mills, 2012). On the other hand, dogs showed empathy-like behavior upon encountering a crying human, but not a talking or humming human (Custance & Mayer, 2012). Still, it is not known whether the dogs discriminated the facial mimics of the experimenters rather than responding differentially to their vocalizations or body expressions. Also, a study in which stimuli were photographs showing human faces with two different emotional expressions did not yield conclusive results (Nagasawa, Murai, Mogi, & Kikusui, 2011). Although dogs learned to discriminate between happy (smiling) faces and neutral faces of their owner and subsequently generalized the contingency to novel faces of unfamiliar people, it is not clear whether the dogs simply used a salient discriminatory cue, such as the visibility of teeth in the happy faces, to solve both the discrimination and the generalization task. In order to clarify the previous findings, Müller, Schmitt, Barber, and Huber (2015) presented pet dogs with a discrimination task that they could solve only by attending to the emotional expression (happy or angry) in the presented human face. This was assured by showing only “semifaces”—either the lower or the upper half of the faces—in training and then testing dogs with the other semiface in the transfer test (Fig. 2). Given that the simple discriminatory cues in one half of the faces—such as teeth in the lower half—were absent in the other half, the authors could test the dogs’ ability to spontaneously categorize novel pictures on the sole basis of the emotional expression, not just by using local cues. Although not all dogs reached the high learning criterion in training, those that did maintained their level of accuracy on training trials in the test phase and— importantly—performed above chance level when confronted with the test stimuli (Fig. 3). This successful transfer suggests that dogs are able to discriminate between emotional expressions in a different species. Compared to emotion recognition in conspecifics (cf. Parr, Waller, & Heintz, 2008), discriminating emotional expressions in heterospecifics is particularly challenging. For instance, humans open their mouth when showing their teeth while laughing, whereas dogs express aggression when showing their teeth. Therefore, the ability to recognize emotional expressions in members of a different species must depend on individual experience. It is
Huber
Fig. 2. A dog in front of the touch-screen apparatus (shown here in Müller, Schmitt, Barber, & Huber, 2015). The dog owner (not shown) sits on a chair at the opposite end of the room, and the experimenter (not shown) on the side of the learning chamber. Neither can see the stimuli, and thus they cannot interfere.
possible that when they were exposed to the pictures of the human faces, dogs could remember something from their daily experiences with their owner or other familiar people and then used this information in the artificial laboratory environment. At least, this study would suggest picture-object equivalence in the dogs. Especially when living in an intimate relationship with a human partner, dogs may acquire representations of human emotional expressions. These may consist either of global memories (“template matching”) or of associations between different parts of the face showing the same expression. With reference to such memories, dogs in the study by Müller and colleagues (2015) could have
How Dogs Perceive and Understand Us 1.0
343 **
**
**
**
**
Standard Trial
Novel Face, Trained Hemiface
Trained Face, Untrained Hemiface
Novel Face, Untrained Hemiface
Trained Face, Left Half
Proportion Correct
.8 .6 .4 .2 .0 Training
Condition Fig. 3. The proportion of correct choices for 11 dogs in the last five sessions of training and in the five conditions of the test phase of Müller, Schmitt, Barber, and Huber (2015). In the test phase, probe trials were interspersed semi-randomly within sessions of standard trials (in which stimuli were identical to those presented in the training phase); stimuli across the probe trials consisted of novel faces with the same hemiface as in training, training faces with the other semiface, novel faces with the other semiface, and training faces with the left half shown. Face stimuli were drawn from the Amsterdam Dynamic Facial Expression Set (Van der Schalk, Hawk, Fischer, & Doosje, 2011) and the Radboud Faces Database (Langner et al., 2010).
learned that a happy expression in the lower half of the face has the same meaning as a happy expression in the upper half of the face and generalized this association to similar but novel faces that expressed the same emotions in the same way. Crucially, this study could not demonstrate that the dogs recognized or understood the emotional content of the human faces because it did not include a planned assessment of a specific recognition response. However, recent eye-tracking studies have supported the hypothesis that dogs look at emotional human faces differently, not by being forced through discrimination learning but by a spontaneous evaluation of the emotional content (Barber, Randi, Müller, & Huber, 2016; Somppi et al., 2016).
Conclusion Many studies have provided evidence that domestic dogs possess sufficient perceptual and cognitive resources to solve difficult visual tasks, such as subtle discriminations and polymorphous, open-ended categorizations. They can extract and combine task-relevant features even when tested in an artificial laboratory environment and with small computer images. When confronted with either conspecifics or humans, dogs retrieve from their memory complex representations consisting of bimodal or multimodal sensory information. Beyond learning subtle idiosyncratic features, dogs appear to obtain
important social information from living as companion animals in a human environment. The human face is a rich source of information allowing dogs to interact and communicate with humans. From this visual resource dogs obtain information signaled through communicative gestures and attentive and emotional states. It seems that their sensitivity and/or competences in the visual domain are beneficial for living and working with humans. Recommended Reading Bensky, M. K., Gosling, S. D., & Sinn, D. L. (2013). (See References). Provides a comprehensive review of research on dog cognition in both the social and nonsocial domains. Hare, B., Brown, M., Williamson, C., & Tomasello, M. (2002). The domestication of social cognition in dogs. Science, 298, 1634–1636. Proposes the domestication hypothesis of dog’s social cognition, which assumes that cooperation and communication with humans were selected for in dogs during their domestication. Kaminski, J., & Marshall-Pescini, S. (Eds.). (2014). The social dog: Behaviour and cognition. San Diego, CA: Academic Press. Provides a comprehensive review of the growing body of research on dogs’ social behavior and cognition, and highlights some of the controversial issues and open questions. Range, F., & Virányi, Z. (2015). Tracking the evolutionary origins of dog-human cooperation: The “Canine Cooperation Hypothesis.” Frontiers in Psychology, 5, Article 1582. doi:10.3389/fpsyg.2014.01582. Proposes the Canine Cooperation Hypothesis, suggesting that wolves are
344 characterized by high social attentiveness, tolerance, and cooperativeness, which likely provided a good basis for the evolution of dog-human cooperation.
Declaration of Conflicting Interests The author declared no conflicts of interest with respect to the authorship or the publication of this article.
Funding The research described in this article was supported by the Messerli Foundation, the Austrian Science Fund (Fonds zur Förderung der wissenschaftlichen Forschung Grant P21418) and the Vienna Science and Technology Fund (Wiener Wissenschafts-, Forschungs- und Technologiefonds Grant CS11-005).
References Adachi, I., Kuwahata, H., & Fujita, K. (2007). Dogs recall their owner’s face upon hearing the owner’s voice. Animal Cognition, 10, 17–21. doi:10.1007/s10071-006-0025-8 Aronson, E., & Rosenbloom, S. (1971). Space perception in early infancy: Perception within a common auditory-visual space. Science, 172, 1161–1163. Barber, A. L. A., Randi, D., Müller, C.A., & Huber, L. (2016). The processing of human emotional faces by pet and lab dogs: Evidence for lateralization and experience effects. PLoS ONE, 11(4), e0152393. doi:10.1371/journal.pone.0152393 Bensky, M. K., Gosling, S. D., & Sinn, D. L. (2013). The world from a dog’s point of view: A review and synthesis of dog cognition research. Advances in the Study of Behavior, 45, 209–406. Custance, D., & Mayer, J. (2012). Empathic-like responding by domestic dogs (Canis familiaris) to distress in humans: An exploratory study. Animal Cognition, 15, 851–859. doi:10.1007/s10071-012-0510-1 Faragó, T., Pongracz, P., Miklósi, A., Huber, L., Virányi, Z., & Range, F. (2010). Dogs’ expectation about signalers’ body size by virtue of their growls. PLoS ONE, 5(12), e15175. doi:10.1371/journal.pone.0015175 Huber, L., Racca, A., Scaf, B., Virányi, Z., & Range, F. (2013). Discrimination of familiar human faces in dogs (Canis familiaris). Learning and Motivation, 44, 258–269. doi:10.1016/j .lmot.2013.04.005 Kaminski, J., & Nitzschner, M. (2013). Do dogs get the point? A review of dog–human communication ability. Learning and Motivation, 44, 294–302. doi:10.1016/j.lmot.2013.05.001 Langner, O., Dotsch, R., Bijlstra, G., Wigboldus, D.H.J., Hawk, S.T., & van Knippenberg, A. (2010). Presentation and validation of the Radboud Faces Database. Cognition & Emotion, 24, 1377–1388. doi:10.1080/02699930903485076 Merola, I., Prato-Previde, E., & Marshall-Pescini, S. (2012). Dogs’ social referencing towards owners and strangers. PLoS ONE, 7(10), e47653. doi:10.1371/journal.pone.0047653 Miklósi, A., Kubinyi, E., Topál, J., Gácsi, M., Virányi, Z., & Csányi, V. (2003). A simple reason for a big difference: Wolves do not look back at humans, but dogs do. Current Biology, 13, 763–766. doi:10.1016/s0960-9822(03)00263-x Miller, P. E., & Murphy, C. J. (1995). Vision in dogs. Leading edge of medicine—A review. Journal of the American Veterinary Medical Association, 207, 1623–1634.
Huber Müller, C. A., Schmitt, K., Barber, A. L. A., & Huber, L. (2015). Dogs can discriminate emotional expressions of human faces. Current Biology, 25, 1–5. doi:10.1016/j .cub.2014.12.055 Nagasawa, M., Murai, K., Mogi, K., & Kikusui, T. (2011). Dogs can discriminate human smiling faces from blank expressions. Animal Cognition, 14, 525–533. doi:10.1007/s10071011-0386-5 Neitz, J., Geist, T., & Jacobs, G. (1989). Color vision in the dog. Visual Neuroscience, 3, 119–125. Parr, L. A. (2011). The evolution of face processing in primates. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 1764–1777. doi:10.1098/rstb.2010.0358 Parr, L. A., Waller, B. M., & Heintz, M. (2008). Facial expression categorization by chimpanzees using standardized stimuli. Emotion, 8, 216–231. doi:10.1037/1528-3542.8.2.216 Pitteri, E., Mongillo, P., Carnier, P., & Marinelli, L. (2014). Hierarchical stimulus processing by dogs (Canis familiaris). Animal Cognition, 17, 869–877. doi:10.1007/s10071-013-0720-1 Pitteri, E., Mongillo, P., Carnier, P., Marinelli, L., & Huber, L. (2014). Part-based and configural processing of owner’s face in dogs. PLoS ONE, 9(9), e108176. doi:10.1371/journal .pone.0108176 Racca, A., Guo, K., Meints, K., & Mills, D. S. (2012). Reading faces: Differential lateral gaze bias in processing canine and human facial expressions in dogs and 4-year-old children. PLoS ONE, 7, e36076. doi:10.1371/journal.pone.0036076 Range, F., Aust, U., Steurer, M., & Huber, L. (2008). Visual categorization of natural stimuli by domestic dogs (Canis familiaris). Animal Cognition, 11, 339–347. doi:10.1007/ s10071-007-0123-2 Schwab, C., & Huber, L. (2006). Obey or not obey? Dogs (Canis familiaris) behave differently in response to attentional states of their owners. Journal of Comparative Psychology, 120, 169–175. Somppi, S., Törnqvist, H., Hänninen, L., Krause, C., & Vainio, O. (2012). Dogs do look at images: Eye tracking in canine cognition research. Animal Cognition, 15, 163–174. Somppi, S., Törnqvist, H., Hänninen, L., Krause, C. M., & Vainio, O. (2014). How dogs scan familiar and inverted faces: An eye movement study. Animal Cognition, 17, 793–803. Somppi, S., Törnqvist, H., Kujala, M. V., Hänninen, L., Krause, C. M., & Vainio, O. (2016). Dogs evaluate threatening facial expressions by their biological validity—Evidence from gazing patterns. PLoS ONE, 11(1), e0143047. doi:10.1371/ journal.pone.0143047 Steurer, M., Aust, U., & Huber, L. (2012). The Vienna Comparative Cognition Technology (VCCT): An innovative operant conditioning system for various species and experimental procedures. Behavior Research Methods, 44, 909–918. doi:10.3758/s13428-012-0198-9 Taylor, A. M., Reby, D., & McComb, K. (2011). Cross modal perception of body size in domestic dogs (Canis familiaris). PLoS ONE, 6(2), e17069. doi:10.1371/journal.pone.0017069 Van der Schalk, J., Hawk, S. T., Fischer, A. H., & Doosje, B. (2011). Moving faces, looking places: validation of the Amsterdam Dynamic Facial Expression Set (ADFES). Emotion, 11, 907–920. Wallis, L. J., Range, F., Müller, C. A., Serisier, S., Huber, L., & Virányi, Z. (2015). Training for eye contact modulates gaze following in dogs. Animal Behaviour, 106, 27–35.
