Abstract

Communication is an essential component of the translation of learning theory into the practical control of the behaviour of dogs. A handler sends a signal (e.g. a command), to which their dog responds. This response is dependent on the dog’s perception of the signal rather than the intention of the sender. Previous research has shown that a dog’s response can be influenced by specific changes in the verbal and non-verbal qualities of signals (i.e. the commands) used, but there has been little scientific evaluation of what happens in practice. Therefore in a first study, 56 dog handlers were videotaped giving their dogs a “sit” command and the significance of verbal and non-verbal factors on response was analyzed. Two factors were associated with a significant decrease in obedience: the dog’s attention to its handler and the handler giving additional verbal information preceding the actual verbal command. Based on these results, a second more controlled study was run with 12 dogs that were trained to a new (“uff”, i.e. jumping onto a raised surface) and a known (“sit”, “down” or “paw”) command. Once trained to predefined criteria, dogs were tested for their responsiveness with each of three additional types of verbal information preceding the command: the dog’s name, the dog’s name followed by a pause of 2 seconds and a “novel word”, i.e. a word with no established relationships in this context (“Banane”). The results suggest that the addition of the novel word significantly reduced response to both the known (p = 0.014) and the new (p = 0.014) commands. The name plus a pause preceding the command significantly reduced the response to the new command (p = 0.043), but not the known one. The use of the name before the command without a pause had no significant effect on performance. The dogs’ ability to generalize learned commands from the training context to a new context was tested by going through the same procedure in an unfamiliar environment. There was a significant reduction in correct responses only to the new command independent of the preceding verbal information (name (p = 0.028), name plus pause (p = 0.022) and novel word (p = 0.011)). This suggests that dogs may have more difficulties generalizing a less well-established command than an already known command.