Methods for Fitting Olfactory Psychometric Functions: A Case Study Comparing Psychometric Functions for Individuals...

In psychophysics, the detection of sensory signals can be depicted by a psychometric function (PF)-a sigmoid function determined by the intercept (i.e., threshold) and the slope (i.e., the rate of increase of detection probability). Fitting psychometric functions is, however, unpopular in chemosensory research, particularly in olfaction. Most olfactory studies adopt ad hoc methods involving a fixed-performance criterion, which result only in a threshold estimate rather than a complete detection profile. This study illustrates the method selection process for fitting olfactory PFs, using a unique odorant-β-ionone-as an exemplar. This particular odorant has a bimodal threshold distribution across the population, which is known to associate with individuals' genotypes for rs6591536 . The characteristics of β-ionone enabled us to use a receiver operating characteristic analysis to assess the different odor threshold estimation methods. The results showed that the available methods based on PFs discriminated between individuals of different genotypes with greater accuracy than the ad hoc method. By fitting PFs to the separate genotypic groups, we also made the first observation of the relationship between genotypic variation and the slope parameter of olfactory PFs. The genetically identified β-ionone "sensitive" group ( s = 1.102-1.883) was shown to have significantly shallower PFs than the β-ionone "insensitive" group ( s = 0.542-0.724). The slope difference may indicate potential discrepancies in the combinatorial coding scheme of odors between these 2 genotype groups, inviting future research to identify the physiological basis.