Naturalistic quantification of canine olfactory sensitivity

Despite the large and growing dependence on dog-handler teams for solving "real world" problems of odor detection, recognition and localization, no comprehensive methodology for quantifying the capabilities of such teams has been reported. We developed an approach in which each dog is housed with its owner-handler, deprived of neither food nor water and indicates to its handler which of five Teflon boxes contains the target odorant of n-amyl acetate (nAA). In two dogs (Rottweiler, Standard Schnauzer) trained in this way, precisely controlled concentration ranges of nAA were systematically lowered over the course of several weeks, in blocks of three 9-trial sessions, until chance performance was seen. Data for each concentration were expressed in terms of a logistic regression equation relating concentration to the binomial probability that the observed performance (or better) would be seen by chance alone. That concentration corresponding to a probability of 0.05 was defined as threshold. In this admittedly small sample, the values we obtained (1.9 and 1.14 ppt) are roughly 30- to 20,000-fold lower than the range of thresholds reported by Krestel et al. [Krestel, D., Passe, D., Smith, J.C., Jonsson, L., 1984. Behavioral determination of olfactory thresholds to amyl acetate in dogs. Neurosci. Biobehav. Rev. 8, 169-174] in their conditioned suppression study of beagle sensitivity to nAA. Thus, it appears that there are significant advantages to our approach, though the reasons for differences in results are unclear. The "find the target" aspect of this new method makes it readily applicable to odor processing tasks much more complex than detection of single compounds. (c) 2005 Elsevier B.V. All rights reserved.