Csaba Molnar ethologist, science journalist

Summary of doctoral theses

In my research I studied various aspects of dog-dog and dog-human acoustic communication. According to basic concept of communication I studied whether dog barks emitted in different situations and by different individuals had different acoustic traits. By studying receivers' responses I investigated if dogs and humans were able to distinguish among barks recorded in different contexts or from different individuals.

In my first experiment I applied a machine learning approach. This software is able to recognize the context specific and individual specific features of barks input and, on the basis of this knowledge, categorize unfamiliar barks into classes. According to our results the software was able to classify new barks into the adequate categories with efficiency higher than expected by chance. In our opinion this suggests that barks have context specific and individual specific acoustic features.

In my next investigation I studied the communicative role of bark in dog-dog communication. Applying the 'habituation-recovery' paradigm I played barks for dogs in the laboratory. The played material consisted of barks recorded from the same individuals in different situations in one experiment and barks from different individuals in the same situation in the other experiment. The dogs' orientation responses were recorded and analyzed. The results showed that subjects distinguished between barks recorded in different contexts and from different individuals as well.

In my following experiments I studied what kind of information the humans could gain by listening to barks. I showed that humans were able to categorize barks into the adequate context categories with a higher accuracy than expected by chance and they described the possible motivational states of dogs similarly. The performances of people with different type of experiences with dogs were not different. I showed that the decisions of subjects about motivational states of barking dogs were affected by a few basic acoustic features of barks (e.g. mean frequency, tonality, and time interval between two individual barks). This role of acoustic features in the judging of inner states is similar to differences in vocal signals of individuals of several vertebrate species having opposite motivational states. This finding might suggest that this ability of people is homologous with the similar ability of animals.

I showed that people, independently form their previous experience with dogs, were not able to discriminate between two individual barks (or two bark sequences consisting 5-5 individual barks) and decide whether they were recorded from the same or different dogs. I found the most significant differences between humans' abilities and the abilities of dogs or computers. Furthermore I showed that blind people with no visual experiences with dogs were able to categorize the heard barks into context categories and describe the motivational states of dogs similarly to sighted subjects. When studying the ontogenesis of human ability for evaluation of barks I showed that even in the six years old age group the performances of subjects owned a dog and those not owned one were not different. The performances of subjects got better with their age.

Summarizing the results of my study I showed dog bark is an effective communicative system in the dog-dog and dog-human communication.

Download the extended summary of theses: PDF