|Scent Detection in Dogs: Effects of Training, Handling, and Selection on Performance
|Saturday, May 23, 2020
|12:00 PM–12:50 PM
|Marriott Marquis, Level M2, Marquis Ballroom 5
|Area: AAB/EAB; Domain: Applied Research
|Chair: Erica N. Feuerbacher (Virginia Tech)
|CE Instructor: Erica N. Feuerbacher, Ph.D.
Domestic dogs are routinely used for scent detection tasks in increasingly wide-ranging domains, from explosives, to disease, pest, and rare species. How training, handling, and selection of dogs and handlers influences performance are important questions for this expanding field. This symposium reports research into the effects of these dimensions on scent dog performance. First, detection threshold of dogs and dogs' ability to generalize to new odor concentrations will be explored as it relates to the odor concentration they were initially trained on. Such research provides insight into applied questions of what concentrations to train on, as well as basic questions of stimulus generalization. A second paper investigates the influence of handler knowledge of target locations, an essential question given that teams often test in situations where the handler knows target locations, but do not when they are in the field. Finally, a third paper investigates the feasibility of training owned dogs with their owners to serve as citizen-science teams detecting agricultural diseases and pests, and whether selection tests based on reinforcer efficacy and persistence can predict success. Given the ubiquitous nature of conservation and agricultural issues, training and deploying pet dogs could greatly increase communities' ability to protect their environment.
|Instruction Level: Intermediate
|Keyword(s): detection dog, dog behavior, dog training, scent detection
Practitioners interested in applied nonhuman research and how results might apply to human training issues Practitioners interested in basic research and how results might apply to human training issues Researchers interested in applications of basic and applied research to nonhuman domains
|Learning Objectives: Be able to identify differences in stimulus generalization based on the trained stimulus Be able to identify the effects of handler knowledge on detection dog behavior Be able to identify the predictive value of reinforcer efficacy tests on detection dog performance
|Odorant Concentration as a Dimension for Stimulus Generalization
|NATHANIEL HALL (Texas Tech University), Mallory Tatum DeChant (Texas Tech University)
|Abstract: Detection dogs are required to detect trace quantities of substances, many times in the parts per billion or parts per trillion range. Frequently, detection of trace quantities is not explicitly trained but rather assumed when dogs show proficiency at higher concentrations. The aim of this study was to evaluate the effect of the odor concentration of the training sample on the minimum concentration dogs will subsequently detect. We expected that dogs may not spontaneously generalize to trace odor concentrations when trained with higher concentrations, but when trained to a lower range of concentrations, dogs will show superior trace odor detection. Two groups of dogs (n=5 per group) were trained to alert to amyl acetate at 0.01% odor dilution (v/v with mineral oil) using a 3-alternative forced choice test. Once reaching proficiency, odor detection threshold was assessed using a descending staircase procedure. Next, experimental dogs received training to systematically lower concentrations of amyl acetate and threshold re-assessed. Control dogs were yoked to experimental dogs, but only received training to the 0.01% dilution between threshold assessments. Experimental dogs showed improved detection thresholds, outperforming control dogs that failed to alert at the lowest concentrations. Results suggest that explicitly training for low concentrations is critical for trace odor detection.
|Effect of Handler Knowledge on Detection Dog Performance
|MALLORY TATUM DECHANT (Texas Tech University), Nathaniel Hall (Texas Tech University)
|Abstract: Detection dogs are commonly trained and tested under conditions in which the handler or the evaluator know the true presence or absence of a target odor. Previous research has demonstrated that when handlers are deceived in regards to the presence of a target odor, more false alerts occur. However, many detection teams operate under unknown conditions, and it remains unclear how handler knowledge (or lack thereof) of odor presence/absence influences the dog’s behavior. The aim of this study was to evaluate if knowing the number of hides placed influenced detection dog performance in an applied search environment. Professional (n=12) and sport (n=39) detection handler-dog teams were asked to search three separate areas (area 1 had one hide, area 2 had one hide, area 3 was blank). Handlers in the Unknown Group were not told the number of hides whereas the Known Group was told there were a total of two hides in the three areas. The Unknown Group spent a longer duration (62.98sec) searching in area 3 compared to the Known Group (P = 0.008) (Figure 1). Further, dogs in the Unknown group looked back to the handler more frequently than dogs in the Known group (~6 more lookbacks on average). Critically, however, there was no difference in false alerts between the Known and Unknown groups. Overall, when handlers knew the number of hides, it led to significant changes in search behavior of the detection team, but did not influence the overall false alert rates.
|Using Community Dogs and Owners to Create a Citizen-Science Agricultural Scent Dog Detection Program
|HANNAH DECKER (Virginia Tech), Erica N. Feuerbacher (Virginia Tech)
|Abstract: Currently, dogs are being used to detect agricultural diseases and pests. Traditionally, scent detection dogs are specifically selected, trained, and handled by professionals. However, given that conservation and agricultural issues are ubiquitous, having local detection dogs could protect the environment and produce sustainable agriculture. Owners have successfully trained pet dogs to compete in nosework competitions, suggesting they could be trained and deployed as local detection dogs. We recruited local dogs and owners to participate in a citizen-science program aimed at determining whether pet dogs could be trained to detect agricultural diseases and pests. Recruited dogs were tested for sensitivity to toy and food reinforcement. Dogs were randomly assigned to one of two scent targets (powdery mildew or spotted lanternfly) and completed weekly training with a scent trainer. After completing training, dogs were tested in the field. We assessed whether the reinforcer efficacy tests predicted dogs’ success in training and performance as measured by sensitivity and specificity on real-world detection tests and latency to indicate target odor. Dogs had varied performance on the reinforcer efficacy tests and performance during training and in field. We will discuss results in terms of likelihood that citizen-science projects can be help address local conservation issues.