A new study from the researchers of the Department of Ethology at Eötvös Loránd University (ELTE), Hungary reveals that dogs can recognize their owner by voice alone, and they make use of some of the same voice properties as humans do.
The study is published in Animal Cognition.
Sure, our dogs recognize us. But are they also capable of it, when neither vision nor smell is there to guide them, so by using voice alone? If so, what is it about voice, that helps them?
To find out, researchers at the Department of Ethology, ELTE, Hungary, invited 28 owner-dog pairs to play hide-and-seek in the lab. Dogs had to find their owner behind one of two hiding places while a stranger hid behind the other one.
They played the owner’s voice from the owner’s hiding place, and a stranger’s voice from the other hiding place, both reading out recipes in a neutral tone. The dogs’ task was to choose from a distance and find their owners.
The game had multiple rounds and the owner’s voice was paired with 14 different strangers’ voices, some more similar to the owner’s voice, some more different.
Dogs found their owner in 82% of the cases. To make sure that smells did not help dogs here, in the last two rounds the researchers played the owner’s voice from where the stranger hid, and the dogs still went for the voice showing that they did not use their nose in this task.
The researchers also explored what exactly in the voices helped dogs to choose.
“People mostly make use of three properties: pitch (higher or lower), noisiness (cleaner or harsher), and timbre (brighter or darker) to differentiate others. Dogs may make use of the same voice properties or different ones. If two voices differ in a property that matters for dogs, decisions should be easier,” explains Anna Gábor, lead author of the study.
The time dogs looked in the direction of the owner’s voice while waiting for the sign to go showed how sure they were in their decisions.
It turned out that if the owner’s and the stranger’s voice differed more in pitch and noisiness it helped dogs to recognize their owners voice, but timbre and other sound properties did not.
“This is the first demonstration that dogs can tell apart their owner’s voice from many others. The study also shows that dogs make use of some, but only some of the same voice properties as humans do to recognize who is talking” – concludes Andics Attila, leader of the Neuroethology of Communication Lab, where this study was conducted.
Understanding Human Emotions: How Dogs Read Our Faces and Listen to Our Voices
Interspecies emotional communication is in part facilitated by chemosignals (D’Aniello et al., 2018), but, faces are in addition an important visual category for many species because they provide a rich source of perceptual cues, including many idiosyncratic features, and thus facilitate important discriminations.
In the specific case of dogs, it has been suggested that their increased readiness to look at the human face provides a basis for complex forms of dog-human communication (Miklósi et al., 2003). By monitoring human faces, dogs seem to obtain important social information, ranging from communicative gestures to attentive states (Schwab and Huber, 2006; Kaminski and Nitzschner, 2013).
Dogs can quickly find out what features are relevant or informative for making important decisions. They also spontaneously focus on the eyes to infer where humans attend, what they are interested in, and even what they intend to do next (see eye movement studies like for example Somppi et al., 2014).
Gaze following is present in many species, but dogs outperform even nonhuman primates in following human gaze in object choice tasks (Hare et al., 2002; Cooper et al., 2003). Like in the case of human infants, their gaze following is modulated by ostensive cueing, such as direct gaze and addressing by the person, which is evidence that it is more than simply a product of reflexive and learnt mechanisms (Téglás et al., 2012).
Dogs also follow human’s gaze into distant space (Wallis et al., 2015), and they use the eyes of humans to judge their attentional state. In one study, dogs were tempted with sausages but told by the caregiver not to take them. The dogs obeyed more or less depending on the caregiver’s attention to them (Schwab and Huber, 2006). When being watched by the caregiver, dogs stayed lying down most often or for the longest time, but when the caregiver read a book, watched TV, turned her back on them, or left the room, their patience ceased. Obviously, they were using eye contact and eye orientation as cues.
Human faces provide much more information than simply looking patterns. A great number of idiosyncratic features allow humans to identify and recognize others. Would dogs also profit from this rich source of information? Could they also identify and recognize their caregiver and other familiar humans?
In one study we put these questions to test and asked dogs to discriminate between their caregiver and another highly familiar person by active choice (approaching and touching; Huber et al., 2013). The task could not simply be solved on the basis of familiarity (approaching the familiar person), which is considered an easier task (Wilkinson et al., 2010), but required a fine-grained distinction of familiar people.
Dogs could do so, even when they saw only the (real) face of the humans, but had difficulties when the face was only projected as a picture to a big screen. Only a minority of dogs could finally identify the caregiver on face pictures in which the outer parts of their faces were occluded with a balaclava hood.
A further study confirmed the importance of human eyes for dogs, because they rely less on the nose or the mouth than on the eyes for human face discrimination (Pitteri et al., 2014). They also prefer looking at upright over inverted faces, exactly as we ourselves do (Somppi et al., 2012, 2014).
On the basis of our findings that dogs are competent enough to extract subtle, idiosyncratic features of a face in order to identify a human person, despite changes of color, hair style, make-up, jewelry, hats, etc., we went one step further and asked whether dogs may also learn from our facial expressions.
It has been already shown that dogs can rely on human facial expressions when making decisions about approaching other objects (Merola et al., 2012a). However, a study in which the stimuli were photographs showing human faces with two different emotional expressions did not yield conclusive results (Nagasawa et al., 2011).
Although dogs learned to discriminate between happy (smiling) faces and neutral faces of their caregiver and subsequently transferred 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 the Clever Dog Lab in Vienna, we asked dogs to discriminate “hemifaces” – either the lower or the upper half of the faces – of women showing different (happy and angry) emotions. With this trick we could investigate whether dogs solve the task solely by attending to the emotional expression rather than any inadvertent cues in the presented human face (Müller et al., 2015).
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, provided globally and not just by local cues. Indeed, the dogs did not only manage to learn the training task, but they were also able to transfer the extracted rule to novel faces, even if they had been presented a hemiface not shown in training.
These findings provide strong evidence that dogs are able to discriminate between emotional expressions in a different species, which, compared to emotion recognition in conspecifics, is particularly challenging (cf. Parr et al., 2008). For instance, humans open their mouth and show their teeth while laughing, whereas dogs express the underlying emotions of aggression by showing their teeth. Therefore, dogs cannot rely on genetic predispositions, but need to individually learn the emotional expressions of humans.
The fact that dogs could spontaneously generalize from one face half to the other without the possibility to use cues learned during training strongly supports the idea that they remembered something from their daily experiences with their caregiver or other familiar people and then used this information in the artificial laboratory environment. As they had not been explicitly trained, it seems that they had acquired the competence by latent learning.
Humans express their emotions not only visually but also their voices convey information about affects. Dogs may exploit these contingencies by extracting and integrating bimodal sensory emotional information from humans. From the combination of visual and auditory cues they may form multimodal representations.
Using a cross-modal preferential looking paradigm, researchers at the University of Lincoln (United Kingdom) managed to show that dogs spontaneously combine human or dog faces with different emotional valences (happy/playful versus angry/aggressive) with a single vocalization from the same individual of the same positive or negative valence (Albuquerque et al., 2016). This result points to the possibility that dogs recognized or understood the emotional content of the human faces, not just discriminated them perceptually. Recent eye-tracking studies have supported this hypothesis (Barber et al., 2016; Somppi et al., 2016).
The ability of dogs to integrate information of humans across modalities has also been investigated by using the expectancy-violation procedure (Adachi et al., 2007). A photograph of either the caregiver’s face or an unfamiliar person’s face was presented to the dog after a vocalization was played.
The vocalization used was from the same person or another person, thus matched or mismatched the image. According to the expectancy-violation logic, dogs should be surprised if the visual and auditory cues mismatch and thus look longer than when the two cues match.
This is what happened. After hearing the caregiver’s voice when the face of an unfamiliar person appeared (incongruent condition), dogs exhibited extended looking, while in the case when the vocalization and face matched (i.e., came from the same person; congruent condition), the duration of their gaze was comparably briefer. These findings lend support to the hypothesis that dogs recall their caregiver’s face upon hearing the caregiver’s voice.
Taken together, there is cumulating evidence that dogs obtain social information from their experiences with humans, specifically from their facial expressions. They can recognize and remember individual humans. They understand to a significant degree what these humans attend to, what they are interested in, and what they intend to do next.
They can discriminate, individually learn from, and categorize emotional expressions, and they integrate information coming from vocalizations into their understanding of humans and their emotions. Thus, they form multi-modal representations of humans and their emotions, integrating emotions, facial expressions, and vocalizations.
reference link : https://www.frontiersin.org/articles/10.3389/fpsyg.2020.584037/full
reference link : Original Research: Open access.
“The acoustic bases of human voice identity processing in dogs” by Anna Gábor et al. Animal Cognition