Sniff Olfactometer (SO) Protocols

Most olfactometers used to study human olfaction have stimulus durations of more than 1 second and often lasting minutes(Dravnieks 1975; Leland et al. 2001; Schmidt and Cain 2010). During long stimulations, olfactory receptor responses and their resulting behaviors are modulated by adaptation and habituation to the stimulus(Pellegrino et al. 2017; Rankin 2009; Wilson and Linster 2008). For example, EOG results from the first deorphanized olfactory receptor tissue reached a maximum in [~]1 s, dropping to 1/2 maximum in the next second, and showing little signal reduction until the stimulation stopped after 6 seconds(Zhao et al. 1998). Longer stimulations can result in complete habituation; receptors still respond even though the behavior shows complete habituation (Barwich 2014). To minimize the effects of adaption and habituation on stimulus responses, the sniff olfactometer (SO) combined the precision of a blast olfactometer with the gentleness of a stream olfactometer by blasting a brief odorant puff (70ms duration) into a subjects self-imposed inhalation air stream(Rochelle 2017; Rochelle et al. 2017b; Wyckoff and Acree 2017). Here we describe SO protocols for threshold determinations of odorants in aqueous headspaces using odorant recognition probabilities associated with Log(odorant-concentrations(Rochelle et al. 2017a)). During a single trial a subject, preconditioned to associate a veridical name with a given odor (e.g., a pyrazine with "nuts" when the odor was detected and "not nuts" if it wasnt) was cued to "inhale" and 750ms later, a 15ml-70ms puff of odorant headspace was delivered into their inhalation airstream. A session consisted of 12 randomized double-blind trials of 3 different odorant concentrations. Additional sessions with different concentrations were conducted until the response probability to the samples ranged from below 0.2 to above 0.8. The robustness of the fitted function and the size of their confidence intervals depended on the difference between the concentrations of the odorants during a single session: small differences in sample concentration resulted in the data failing to fit a logistic function; larger concentration differences resulted in a better fit to the model. However, if one of the stimuli had no odorant at all i.e., a blank, the response to the blank was random.