Dogs' scent detection performance with rapidly changing targets

Abstract

Scent detection dogs trained to detect one target scent are sometimes needed to detect a different target. Recent studies have demonstrated that scent detection animals can be trained to detect multiple scents simultaneously. However, in some cases, it may be problematic if the dogs indicate on previous targets. Therefore, it would be useful to know how quickly they can learn to detect new target scents, and not to indicate on previously trained target scents. The aim of this study was to evaluate dogs’ ability to learn to detect new target scents while simultaneously rejecting all previously trained target scents. Firstly, five pet dogs were trained to detect a training scent (amyl acetate) using an automated apparatus. The dogs were then required to detect a four new target scents within four phases. Target scents were treated as non-target scents in all subsequent phases; indications on previously trained target scents were not reinforced. Non-target scents in each phase consisted of three randomly selected chemicals that were presented in a randomised rotation. The dogs were required to complete a phase with one target and meet the criteria of more than 80% for correct indication and correct rejection responses, without performing an indication response more than once on the previous target scent for four out of five sessions. The results of the experiment demonstrated that (1) the dogs were able to successfully discriminate between the specified target scent and all non-target scents within each phase; (2) the persistence of indication behaviour on previously trained target scents decreased more rapidly with each phase; and (3) there was a re-emergence of indication responses when each previously trained target scent was presented after extinction conditions. In summary, the present study demonstrates that dogs can be retrained to indicate on new targets and reject previous targets. The learning processes associated with the dogs learning to indicate on new targets, and learning not to indicate on previous target scents must both be considered. These results have significant practical implications and it is hoped that the they will improve our ability to employ individual scent-detection animals to find multiple targets.