3.6.1 Odor perception and chemistry
At first, the functioning of olfaction can appear as a counterintuitive. For instance, contrary to popular belief, human beings have a sharp sense of olfaction. We are extremely good at detecting and discriminating odors at faint concentrations such as 0.77 parts per million in the case of isoamyl, or at spotting molecular alterations (Laska, Ayabe-Kanamura, Hübener, & Saito, 2000; Laska &
Teubner, 1999a). Yet, we are very bad at identifying and naming smells on the basis of free recall, although our performances improve drastically when verbal labels are available (Cain, 1979; Lawless
& Engen, 1977). Additionally, olfactory sensitivity is subject to strong interindividual variability (de Araujo et al., 2005; Herz & Clef, 2001; Keller, Zhuang, Chi, Vosshall, & Matsunami, 2007), as shown by the prevalence of selective anosmias (Bremner, Mainland, Khan, & Sobel, 2003; Wysocki, Dorries, &
Beauchamp, 1989), and the differences existing in the qualitative discrimination of enantiomers (Laska & Teubner, 1999b). These differences mostly arise from population variability in the expression of olfactory receptors (Keller et al., 2007). Nevertheless, the ability to sense to certain molecules, such as androstenone, can be acquired by anosmic people (Wysocki et al., 1989) through systematic exposure, while it is also possible to learn to discriminate indistinguishable odor enantiomers, by conferring them different affective value through associative learning (Li et al., 2008). All these apparent contradictions governing odors detection and perception reflect the way odors valence is determined, since an odorant’s pleasantness depends on its chemical structure, but also, for a large proportion, on personal factors and values.
As previously described, odors have a distinctive and salient hedonic tone, preferentially transduced thanks to the particular organization of the neural system. Yet, still little is known about how odor borne pleasantness subjective experience is elicited, in comparison with other sensory modalities. In this regard, part of the mechanism could depend on the intrinsic properties of a fragrance, as shown by the experiment of Khan and colleagues (Khan et al., 2007) (Figure 19c). By comparing the relations between the primary axes of a verbally-based, olfactory perceptual space and a chemically-based, analogous physicochemical space, they were able to predict, to a certain extent, olfactory pleasantness on the grounds of the odorant chemical structure. Similarly, Kermen and colleagues (Kermen et al., 2011) (Figure 19 a, b) found a relation between the molecular complexity of an odorant, its pleasantness and its perceived complexity assessed by the quantitative use of olfactory notes descriptors, with low complexity molecules rated as more aversive, in line with studies suggesting that some aspects of olfactory pleasantness may be innate (Soussignan, Schaal, Marlier, &
Jiang, 1997).
62 Nevertheless structure and odor quality are not always linked, as similar chemical compounds can evoke very different odors. This is the case with eugenol and vanillin, whose smells are very distinguishable although the two molecules are almost identical. On the contrary, very dissimilar compounds can evoke the same odor, such as benzaldehyde and hydrogen cyanide which both smell of almonds (Figure 19d). For instance, it should be noted that the primary axis of physicochemical properties used for odor pleasantness prediction in Khan’s work only accounted for 32% of the variance. Additionally, the simplified assumption of reducing odors to single molecules is not ecologically valid, since most naturally occurring smells are usually constituted by several tens or hundreds of compounds (Stevenson & Wilson, 2007).
Figure 19. Chemistry and subjective olfactory experience.
a) The complexity of monomolecular odorants has an effect on the perceived olfactory notes, which in turn influences the perceived pleasantness of odors (b). Taken from Kermen et al., 2011. c) A metric based on the first principal component (PC1) of physicochemical space predicts odorant pleasantness in humans Taken from, Arzi and Sobel, 2011, adapted from Khan et al., 2007. d) Odor quality and structure mismatch: benzaldehyde and hydrogen cyanide both smell of almonds, while clove and vanilla smell derive from very similar molecules.
3.6.2 The plasticity of odor-borne emotions: differential influences
“Three-quarters of the universe may find the rose's scent delicious without that serving either as evidence upon which to condemn the remaining quarter which might find the smell offensive, or as proof that this odor is truly agreeable.”
Marquis de Sade
3.6.2.1 Overview
In addition to the interindividual variability brought by genetics (Keller et al., 2007), odor-elicited emotions also seem to rely on learned association and values, as evidenced by the fact that hedonic
63 perception is extremely plastic, and can be influenced by many factors deriving from context and personal expectations. Likewise, neural representations of odor quality seem to undergo some plasticity as new sensory experiences modify odor perception and quality coding in the brain. Many studies have demonstrated a link between the familiarity and the pleasantness of odors (see section 3.2), which is particularly true in the case of pleasant smells: the more familiar the odors, the more pleasant they are perceived (Delplanque et al., 2008). This effect illustrates how experience can influence liking when it comes to odor, given how it affects olfactory processing at the behavioral (Kermen et al., 2011) and at the neural level (Delon-Martin, Plailly, Fonlupt, Veyrac, & Royet, 2013;
Gottfried & Dolan, 2003; Gottfried, 2008; Li et al., 2006; Plailly, Delon-Martin, & Royet, 2012; Royet, Plailly, Saive, Veyrac, & Delon-Martin, 2013). Experience and familiarity also probably account for the consistently reported interindividual variability across cultures, since different populations are exposed and used to different odors, and thus perceive unfamiliar and foreign odors as less pleasant (Ayabe-Kanamura et al., 1998; Distel et al., 1999; Ferdenzi, Roberts, et al., 2013), see also Ferdenzi et al., 2011 for a review of these effects. These cultural differences are also reflected in more refined odor-borne feelings. As previously described in section 3.4.4 (GEOS), when refining the affective experience beyond the traditional “I like” vs. “I do not like” dichotomy, regional variations emerge under the form of culture specific descriptive dimensions. For instance, odors can evoke feelings such as hunger and thirst in western countries (USA, UK, Brazil) vs. intellectual stimulation or spirituality in Asia (Singapore) (Ferdenzi et al., 2011; Ferdenzi, Delplanque, et al., 2013).
3.6.2.2 Inner state, motivations and expectations
Specific motivational states (e.g. hunger) can also influence the subjective experience elicited by external stimuli through appraisal. Alliesthesia and sensory-specific satiety (Rolls 1999) both illustrate this mechanism. Alliesthesia designates the change in pleasantness of sensory inputs modulated by internal body signals (e.g. glycaemia) (Cabanac, 1971), while in the case of sensory-specific satiety, the change in reward value is induced by an external sensory stimulation (e.g. a specific taste, flavor, odor), and it is specific to that external stimulus (Rolls, 2001). As a consequence, a given food eaten to satiety will undergo a decrease in reward value (O’Doherty et al., 2000; Small, Zatorre, Dagher, Evans, & Jones-Gotman, 2001), since it is not regarded as necessary anymore to meet the organism’s goals (need for calories). Food-odors are sensitive to both phenomena, as it is possible to modulate the reward value of a given odor by having participants eating the corresponding food (O’Doherty et al., 2000), or by manipulating the participants’ general state of hunger (Plailly et al., 2011). In particular, the latter alliesthesia effect depends on the qualitative significance of the odor, as it is amplified by smells related to relevant, high energy, fat-containing foods (Plailly et al., 2011). The induced reward-value changes in odors are noticeable at the neural level, since reward-sensitive OFC
64 areas have been shown to respond accordingly (Critchley & Rolls, 1996; O’Doherty et al., 2000) (Figure 20).
Figure 20. Sensory specific satiety effect in the orbitofrontal cortex.
Mean percent change in the BOLD signal across subjects of significantly activated clusters in the orbitofrontal cortex in response to sensory-specific satiety. Activation in the orbitofrontal cortex decreases to the odor a food eaten to satiety (banana), but not to the odor of a food that has not been eaten (vanilla). Taken from O’Doherty et al., 2000.
3.6.2.3 Contextual influences on olfactory processing
The affective experience of odor is also influenced by contextual elements (Gottfried, 2008). For example, the hedonic value of an odor can be impacted if presented in the context of a more pleasant or unpleasant odor (see Mohanty & Gottfried, 2013 for a review), while this relative change in pleasantness is reflected neurally in the anterolateral OFC (Grabenhorst & Rolls, 2009). Other types of contextual cues (i.e. sensory, semantic, verbal) have also been reported to affect olfactory processing: different color hues can cross-modally alter the perception of a smell’s intensity (Gilbert, Martin, & Kemp, 1996; Zellner & Kautz, 1990) or quality (Morrot, Brochet, & Dubourdieu, 2001).
Additionally, the perception of olfactory stimuli is facilitated by the presence of congruent visual semantic, which is translated at the brain level by enhanced neural activity in the anterior hippocampus and the rostromedial OFC (Gottfried & Dolan, 2003). Thus, as mentioned earlier, identification but also of perception of odors improves when provided with relevant cues, such as verbal labels (Cain, 1979; Lawless & Engen, 1977). However, verbal labeling can also affect the hedonic perception of odors: the same odorant, described as “cheese”, will be perceived as more pleasant than when described as “vomit” (Herz & Clef, 2001) or “body odor” (de Araujo et al., 2005), Figure 21a). This semantic influence occurs at the neural level too, since the activity in odor-sensitive areas in response to a given odorant (amygdala and rostral anterior cingulate cortex / OFC) correlates with the perceived pleasantness according to a given label (de Araujo et al., 2005) (Figure 21b).
65
Figure 21. Influence of verbal labeling on olfactory perception and processing.
a) Subjective pleasantness rating to odors are modulated by different hedonic verbal labels. b-c) Activations in the rostral anterior cingulate cortex adjoining to the medial OFC. d-e) Bilateral activations in the amygdala, extending to the primary olfactory cortex.
Threshold: p<0.0001 uncorrected. e-f) Parametric plots showing the percentage of BOLD change correlating with the pleasantness ratings in the b and d regions respectively. Adapted from de Araujo et al. 2005.
3.6.2.4 Preference formation: influence of choice and learning
Interestingly, olfactory subjective experience can also be influenced by choice, as shown by the experiment of Coppin and colleagues (Coppin, Delplanque, Cayeux, Porcherot, & Sander, 2010).
Participants were required to choose the odor they preferred within a pair of odors they previously rated as similarly pleasant. The chosen odors were subsequently perceived as more pleasant than the rejected ones (Figure 22), unveiling the existence of post choice, decision-making related preference changes in smell. Experience and learning also account for the acquisition of preferences, through the emotional associations they generate. A striking example of this is the differential hedonic evaluation of eugenol (clove) by participants according to whether they are afraid or not of dental procedures (Robin, Alaoui-Ismaïli, Dittmar, & Vernet-Maury, 1998). Likewise hedonic responses to odors can be acquired or altered through associative learning with emotional experiences (Barkat, Poncelet, Landis, Rouby, & Bensafi, 2008; Herz, Schankler, & Beland, 2004; Herz, 2005): a novel, pleasant odor presented paired to a negative emotional experience (frustration) is subsequently evaluated as more unpleasant, while an unpleasant odor paired to a positive emotional experience (positive mood induction) is evaluated as less unpleasant (Herz, Schankler, et al., 2004).
Likewise, the pleasantness of neutral odors can be enhanced by appetitive conditioning with pleasant taste (Barkat et al., 2008). Additionally, the robustness of this learning directly depends on the
66 number of odor-taste associations, also advocating for a key role of experience in shaping olfactory preferences.
Figure 22. Choice-induced change of preference for odors.
Mean z scores of pleasantness ratings of chosen vs. rejected odors before and after (prechoice and postchoice phase) being chosen or rejected. Odors were initially rated equally before the choice. Higher pleasantness ratings are obtained for chosen vs. rejected odors after the choice. Taken from Coppin et al., 2010.