Perspectives

The experimental findings discussed in thesis give rise to several research questions that could be tested in the future:

(1) The neural correlates obtained for the sensory pleasure category in the lateral OFC and the ventral insula suggests that the category could account for ingestion related behaviors (e.g.

salivation, appetite) beyond valence only. Yet these neural loci also neighbor regions involved in the processing of unpleasant odors. Do these regions account for the global relevance of an odor stimulus (food value vs. toxicity), or do they have separate representations?

(2) Relaxation scores anti-correlated with activity in the basolateral amygdala. Do these relate to enhanced arousal? How are they related to odor perception induced activation in more medial-dorsal aspects of the amygdala?

(3) The OFC is essential to the processing of odor valence, but also it is sensitive to contextual influences and higher order representations since it is a highly integrative region. These characteristics suggest that the OFC could be an ideal candidate for encoding complex odor

194 borne feeling beyond valences only, for example through differential patterns of activity, and calls for a thorough investigation.

(4) Although accessing less complex representations of odor than the OFC, the piriform cortex plays a key role in determining odor valence in its anterior aspect, and odor category in its posterior aspect. Given that different feelings can arise from odors with a similar global pleasantness score, could the PPC bear a representation of these complex feelings, whilst constant valence induce no changes in the APC?

Questions (1), (2) (3) and (4) could be assessed using MVPA techniques, coupled to high resolution MRI, and manual segmentation of the ROIS

(5) No specific neural correlates for the “refreshing” category were identified in study 2. Perhaps this feeling category reflects specifically the trigeminal experience, which can induce a cooling sensation. Although some our odors were trigeminal (e.g. eucalyptus), we did not formally test whether increasingly trigeminal odors could induce increasingly “refreshing”

subjective experience, and whether this experience would be reflected at the neural level.

(6) Given the affective importance of odors, smells disorders alter the quality of life in varied ways, spanning from flavor impoverishment to social isolation (see section 3.7 of the theoretical part). The EOS scale could constitute a complementary tool to achieve a better delineation of the affective consequences of partial smell loss or smell distortion, or to help assessing early olfactory symptoms signaling neurological conditions (e.g.: Alzheimer or Parkinson diseases).

(7) fMRI allows great insight in spatial representation of neural activity, yet offers limited temporal resolution for the study of the dynamics of olfactory perception and its associated emotions (for a discussion, see Gottfried & Zald, 2005). Although the olfactory neural structures are known, only few studies have assessed the temporal sequence of activation involved in olfactory processing (Bathellier et al., 2008; Lascano et al., 2010). Important dynamic differences between the contralateral vs. ipsilateral sides, and temporal vs. frontal have been revealed. They underline the importance both spatial and temporal features of odor perception, and call for a better dynamic characterization of complex odor-borne feelings elicitation at the neural level, using intracranial EEG techniques.

(8) The results from study 2 led us to wonder whether there could be consciousness / top-down attentional dissociation in the anterior insula for the olfactory system. If so, does it involve other structures such as the ventral pallidum or the mammillary bodies? Do the ACC and / or the olfactory bulb play a role in olfactory attention?

(9) The attentional manipulation performed in study 2 did not appear to induce a corresponding up- or downregulation of activity was observed in the odor-related areas. Would

195 manipulating the affective value of the competing attention modality (i.e. sounds) facilitate the diverting attention away from odors and thus downregulate the activity in olfactory cortices?

(10)The results obtained in study 3 suggest that a pleasant odor context generally enhanced the awareness of aversive stimuli, following a general broadening of attention (Broaden and Build effect, see Fredrickson, 2004, and section 5.4 of the theoretical part), see, without interfering with aversive associative learning. Yet, this effect could also be caused by a contrast with the aversive US. In turn, the acquisition of new emotional values appeared to be facilitated by a congruent aversive context, although this was only visible during extinction since strong malodors may have obscured the CR during acquisition. We thus propose to perform a reverse paradigm, involving appetitive conditioning in a pleasant or unpleasant context, using low intensity odors to instate a more subtle affective setting (see section 4.3.4 of the experimental part), followed by an fMRI version of the 2 experiment. If indeed a Broaden and Build mechanism is at stake under a positive context, then differences in the recruitment of global attentional networks could be expected in both aversive and appetitive conditioning. Conversely, if affective congruency guides associative learning, then preference (or aversion) acquisition (appetitive or aversive conditioning) should be facilitated in a positive (or negative) context. In the case of facilitated aversive learning, enhanced amygdala activation should be expected (Büchel, Morris, Dolan, & Friston, 1998; Knight, Nguyen, & Bandettini, 2005; Labar, Gatenby, Gore, Ledoux, & Phelps, 1998; E. A. Phelps, Delgado, Nearing, & Ledoux, 2004),

General conclusions 4

Study 1 provided new evidence for a fine-grained distinction among olfactory feelings at the brain level. The range and nature of olfactory-elicited feelings appeared to be more complex than simple valence, as suggested by persisting neural activations obtained when directly contrasting GEOS categories to hedonicity alone. Moreover, each of the feeling categories elicited unique and specific neural profiles, despite their apparent intercorrelations. These feeling profiles could potentially be related to the putative functions of smell, such as food-related behaviors, including appetite or disgust, but also memory, emotional regulation, and social interactions. For instance, relaxation feelings are positively related to activation pattern of the default mode network, whereas sensory pleasure appears to be linked to the appetite regulation functions and approach behaviors. In turn, unpleasant feeling appears to be more linked to disgust and avoidance. Intimate links with motor control were also evidenced by activations associated with olfactory disgust, underscoring the vital

196 importance of smell and affective responses in order to signal noxious compounds and ensure an efficient avoidance reaction.

Taken together, the findings obtained in Study 2 reveal an attentional asymmetry for odor perception, with a complete lack of attentional modulation of neural activity in the olfactory cortical and subcortical areas when exposed to bimodal stimuli (odor + sound), while attention to sounds increased activity in auditory areas. Importantly, the amygdala was responsive to odor presence, but not affected by attention nor modulated by subjective hedonic value. Our results also show differential effects of odor valence both with and without attention. While activity in medial OFC and right anterior insula showed a pure effect of valence regardless of attention (positive and negative, respectively), the left anterior insula was significantly modulated by valence depending on whether attention was allocated to odors or not, supporting distinct roles for the right and left insula in processing saliency and disambiguating task-relevant stimuli, respectively. Unpleasant and pleasant odors were still discriminated in the right insula irrespective of attentional focus, but generated similar responses in the left insula when attention was directed away from the odors. These data suggest that the right AI could mediate a general salience detection/alarm-like system for responding to potentially harmful stimuli even without attention; whereas the left anterior insula might be more sensitive to top-down attentional control and task relevance by integrating odor cues with other contextual information in order to enact an appropriate behavioral response.

Study 3 found that the hedonicity of contextual odors can affect low level affective processes, by influencing aversive conditioning outcomes. As pointed out by behavioral and physiological indicators, aversive conditioning appeared to be more effective by generating resistance to extinction when performed with unpleasant odors, in an affective congruent fashion. However, the important attentional drive induced by aversive smells appears to mask the effects of associative learning during the acquisition phase. In comparison, pleasant odors did not facilitate aversive conditioning, but appeared to generally increase the awareness of aversive stimuli (angry face, unpleasant white noise), in line with the broadening of attention enabled by pleasant emotions according to the Broaden and Build theory.

As stated in the introduction, valence and trigeminality are the most salient characteristics of odors, and thus raise the ultimate question of whether odors could possess an innate preparedness inherent to these features for the detection of chemical threat, as it is the case in visual or auditory domains. In these sensory modalities, fear relevant stimuli such as angry or fearful facial expressions (Forgas & Eich, 2006) spiders or snake evoking shapes (Soares & Ohman 1993) and some vocal sounds such as cries have been proposed to to evoke emotional reactions partly based on an innate,

197 inherited biological predisposition to fear threats. Nevertheless, our data suggests that if odors indeed had a preparedness dimension, it would not be supported by valence, but rather by trigeminality as shown by the skin conductance increases. Alternatively, since evolution has primed us to rely more and more on vision and audition to detect danger, one could assume that olfaction only plays a residual role in threat signaling. In fact, given its notorious plasticity, the olfactory system would need confirmatory congruent intermodal cues to disambiguate aversiveness from a real hazard.