Maintaining long-term odor memory and detection performance in dogs

Abstract

Detection dogs represent an essential resource for security and defense sectors, yet logistical and security constraints limit the frequency in which relevant maintenance training can occur. Further, the extents and limits of canine odor memory are unknown. The aims of this study were to examine the duration of long-term odor memory in detection dogs, and to evaluate effects of single-target maintenance training on the recall of previously trained targets, as well as generalization to a set of related but untrained targets. Detection dogs (n = 18) were trained on a set of 10 target odors followed by an assessment of baseline performance in odor discrimination and operational search tests. During the subsequent 12 months, half of the dogs received minimal intermittent odor detection maintenance training (approximately 30 min total per month) with a single target odor from the trained set, while the other half received commensurate training not involving odor detection (i.e., obedience training). Detection of the targets not experienced over the 12-mo period was then re-assessed. Generalization to untrained variants of the trained targets, a critical aspect of optimal detection that may be affected by extensive training with specific odors, was also assessed at baseline and after the 12-mo period. Accuracy in the odor discrimination test significantly decreased from 99 % (SE = .25) at baseline to 72 % (SE = 4.36) after 12 months for dogs that received no odor detection maintenance training, and significantly increased from 94 % (SE = 1.57) to 99 % (SE = .53) for dogs that received single-odor maintenance training over the 12-mo period. However, the maintenance training did not appear to maintain operational search performance. Further, results from the generalization test were less straightforward, suggesting that single-target maintenance training may not be sufficient for optimal operational detection of untrained variants. Our results indicate that long-term odor memory is largely robust in dogs, and that odor discrimination accuracy can be effectively maintained and even improved through minimal maintenance training using a single target odor, but that the task of searching in operational environments may require more regular maintenance.