Increased Intra-Participant Variability in Olfactory Sensitivity Correlates with Autistic Trait Levels

Background:   It is now well established that both a diagnosis of Autism Spectrum Disorder (ASD) and/or an increased level of autistic traits results in increased reporting of sensory difficulties (Robertson & Simmons, 2013). Typical olfactory symptoms reported by individuals with ASD include difficulties in tolerating “perfumed” environments like pharmacies and restaurants, problems being around people who have recently been smoking, or, alternatively, a preference for strong-tasting foods like garlic and chilli. Despite these data, laboratory-based findings on olfactory performance in ASD have provided mixed results, with a number of studies finding no significant differences between ASD and control populations (e.g. Tavassoli & Baron-Cohen, 2012).

Objectives:   To obtain data on olfactory performance in a sample taken from the general population, whilst simultaneously measuring autistic trait level and reported sensory symptoms.

Methods:  106 adults participated in the experiment (mean age, 24 years; 74 females). Olfactory performance was measured using “Sniffin’ sticks” (Burghart, Messtechnik), a standardized method for measuring olfactory detection, discrimination and identification. Autistic traits were measured using the Autistic Spectrum Quotient (AQ; Baron-Cohen et al, 2001). A subset of participants also filled in a questionnaire about their own sensory issues (Robertson & Simmons, 2013). For some analyses, we separated the top 17 highest scorers in the AQ (27.8 ± 5.6) from the 18 lowest scorers (6.3 ±1.6), representing approximately the top and bottom 15% of the sample.

Results:  There was no significant relationship between AQ score and olfactory detection thresholds, neither did the high AQ scorers differ significantly from the low AQ scorers. However, we also obtained a measure of intra-participant variability: the variance in individual mean detection threshold. We found a small, but significant, correlation between this variability measure and AQ score (r=.266, p<.01, R²=.071) and a significant difference between high- and low AQ scorers [t(21.71) = 2.23, p<.05, r=.277]. There were no significant correlations with AQ score, or significant differences between high and low scorers, in measurements of olfactory discrimination or identification thresholds, even after controlling for gender and native language. We did, however, find a significant correlation between AQ score and reported sensory difficulties [r=.433, p<.001, R²=.187], confirming previous results.

Conclusions:   These data suggest that whilst there are no differences in mean olfactory detection, discrimination or identification performance with varying autistic traits, detection threshold variability, within a given individual, is higher in those scoring higher on the AQ. Therefore an apparent lack of difference in mean scores on sensory threshold tests could mask an underlying variability in performance. One potential cause of intra-participant variability in threshold performance is the existence of significant levels of internal neural noise, which can alter perception on a moment-to-moment basis (Dinstein et al, 2012; Milne, 2011; Simmons et al, 2009). Higher internal noise levels may thus explain the apparent conundrum of increased sensory difficulties with increasing autistic trait levels which are unaccompanied by a change in mean thresholds.