What olfactory psychophysics and ethology can contribute to work with sniffer dogs – Presented by Simon Gadbois

In this theoretical and methodological talk, I will examine two traditional methodologies often used with scent dogs: 1) In field work, the use of transects (especially with conservation canines). Here, ethology and sensory ecology can contribute some insight. We will discuss the problems with transects (for the dogs), and the “natural” alternative. 2) In lab work, especially in biomedical detection, for mostly historical reasons, scent training with canines has used the mAFC model, i.e., forced choice tasks (i.e., line-ups, wheels, carousels). Robust data from psychophysics would suggest an alternative, if not a replacement, depending on the goal of the study: “true” detection. I say true detection (which includes yes/no and go/no-go procedures) because mAFC tasks are actually discrimination tasks. In some applications, such as for alert, or even for a clear idea on performance in diagnostic work, there is a strong argument to at least use discrimination (mAFC) and detection in tandem, if not detection alone. I will present arguments from Signal Detection Theory that will demonstrate that such a methodology provides significantly more information about the dog’s performance (accuracy and precision), especially if combined, by addressing both sensitivity (detection task) and specificity (discrimination task).

Coming soon.