Human listeners attend to size information in domestic dog growls

Abstract

Vocal expressions of emotions follow simple rules to encode the inner state of the caller into acoustic parameters, not just within species, but also in cross-species communication. Humans use these structural rules to attribute emotions to dog vocalizations, especially to barks, which match with their contexts. In contrast, humans were found to be unable to differentiate between playful and threatening growls, probably because single growls' aggression level was assessed based on acoustic size cues. To resolve this contradiction, we played back natural growl bouts from three social contexts (food guarding, threatening and playing) to humans, who had to rate the emotional load and guess the context of the playbacks. Listeners attributed emotions to growls according to their social contexts. Within threatening and playful contexts, bouts with shorter, slower pulsing growls and showing smaller apparent body size were rated to be less aggressive and fearful, but more playful and happy. Participants associated the correct contexts with the growls above chance. Moreover, women and participants experienced with dogs scored higher in this task. Our results indicate that dogs may communicate honestly their size and inner state in a serious contest situation, while manipulatively in more uncertain defensive and playful contexts.

Auditory Communication in Domestic Dogs

“…As formant dispersion is controlled by VTL, this means that unlike F0, formants have the potential to provide accurate or 'honest' information about the caller (Fitch, 1997;Reby & McComb, 2003). A direct negative correlation between formant dispersion and body size has been confirmed in the signals of many species, including dog growls (Riede & Fitch, 1999;Taylor et al, 2008). When we analyse formant dispersion within any growl, it is thus possible to predict the body size of the caller.…”

“…Within dog breeds, those of comparable size to wolves show similar levels of body mass dimorphism, which was found to relate to a difference of 5%-7% in vocal tract length (VTL) between adult males and females for Portuguese water dogs (Plotsky et al, 2013). However, the size difference between sexes decreases in smaller breeds (Frynta et al, 2012), and no sex-related differences in vocal anatomy or in F0 have been identified across breeds when controlling for body weight (Riede & Fitch, 1999;Taylor et al, 2008). However, artificial selection by humans has led to an exceptionally large level of size variation between breeds, giving dogs the highest level of morphological variation of any mammal (Wayne, 1993).…”

“…However, artificial selection by humans has led to an exceptionally large level of size variation between breeds, giving dogs the highest level of morphological variation of any mammal (Wayne, 1993). Consequently, across breeds, F0 can provide a broad indication of body size, with smaller breeds producing growls with a higher F0 than larger breeds (Taylor et al, 2008). On an individual level, F0 is influenced by changes to respiration and muscular control of the tension of the vocal folds.…”

“…On an individual level, F0 is influenced by changes to respiration and muscular control of the tension of the vocal folds. Therefore, F0 is highly variable both within and between calls, across individuals, and across breeds (Feddersen-Petersen, 2000;Yin, 2002;Taylor et al, 2008).…”

“…In a psychoacoustic investigation, Taylor et al (2008) demonstrated that human listeners are able to accurately attribute associated body size to growls that had been resynthesised to vary only in formant dispersion, with all other acoustic parameters standardised across stimuli. Interspecific size assessment is also interesting to consider in the context of the predatory potential of domestic dogs.…”