Odor Pleasantness Modulates Functional Connectivity in the Olfactory Hedonic Processing Network

Abstract

Olfactory hedonic evaluation is the primary dimension of olfactory perception and thus central to our sense of smell. It involves complex interactions between brain regions associated with sensory, affective and reward processing. Despite a recent increase in interest, several aspects of olfactory hedonic evaluation remain ambiguous: uncertainty surrounds the communication between, and interaction among, brain areas during hedonic evaluation of olfactory stimuli with different levels of pleasantness, as well as the corresponding supporting oscillatory mechanisms. In our study we investigated changes in functional interactions among brain areas in response to odor stimuli using electroencephalography (EEG). To this goal, functional connectivity networks were estimated based on phase synchronization between EEG signals using the weighted phase lag index (wPLI). Graph theoretic metrics were subsequently used to quantify the resulting changes in functional connectivity of relevant brain regions involved in olfactory hedonic evaluation. Our results indicate that odor stimuli of different hedonic values evoke significantly different interaction patterns among brain regions within the olfactory cortex, as well as in the anterior cingulate and orbitofrontal cortices. Furthermore, significant hemispheric laterality effects have been observed in the prefrontal and anterior cingulate cortices, specifically in the beta ((13-30) Hz) and gamma ((30-40) Hz) frequency bands.

Keywords: brain connectivity; electroencephalography; hedonic evaluation; lateralization; olfaction.

Figure 1

Overview of the study’s methodology. EEG signals recorded in response to pleasant olfactory stimuli exposure (A) were preprocessed and source localization was performed to estimate the cortical sources generating scalp level signals (B). Then, functional connectivity between source-level EEG signals was estimated using the weighted phase lag index (wPLI), yielding $80\times 80$ adjacency matrices that store pairwise connectivity values between sources (ROIs). The wPLI matrices were subject to further thresholding based on orthogonal minimum spanning trees (OMST) (C), resulting in functional connectivity networks that were analyzed (D) to estimate graph theoretic metrics (nodal degree, clustering coefficient, betweenness centrality). Using said metrics, statistical analyses were carried out to quantify cortical dynamics induced by olfactory stimuli of different perceived pleasantness. This included direct comparisons of experimental factors via paired samples t-tests, as well as utilizing graph metrics to assess hemispheric laterality effects.

Figure 2

Location of selected AAL ROIs underlying major regions involved in olfactory, affective and reward processing. Regions comprise a total of 14 nodes: The middle frontal gyrus (${\mathrm{MFG}}_{\mathrm{L}/\mathrm{R}}$), the superior frontal gyrus, medial part (${\mathrm{SFGmed}}_{\mathrm{L}/\mathrm{R}}$), the anterior cingulate gyrus (${\mathrm{ACG}}_{\mathrm{L}/\mathrm{R}}$), the middle frontal gyrus, orbital part (${\mathrm{ORBmid}}_{\mathrm{L}/\mathrm{R}}$), the inferior frontal gyrus, orbital part (${\mathrm{ORBinf}}_{\mathrm{L}/\mathrm{R}}$), the olfactory cortex (${\mathrm{OLF}}_{\mathrm{L}/\mathrm{R}}$) and the amygdala (${\mathrm{AMYG}}_{\mathrm{L}/\mathrm{R}}$).

Figure 3

Paired-samples t-tests on behavioral response data reveal significant differences between high and low pleasantness stimuli; no meaningful discrepancies in fragrance intensity were observed. FDR corrected p-values are depicted; error bars show the standard error.

Figure 4

Statistically significant differences in graph theoretic metrics between high- and low pleasantness stimuli (individual regions, paired samples t-tests), including results of graph null-hypothesis testing (*: p < 0.01). Plots show results relating to the weighted nodal degree (A) and clustering coefficient (B). All events were observed in the beta ($\beta$) frequency band. FDR-corrected p-values are depicted; error bars show the standard error.

Figure 5

Chord diagrams outlining normalized connectivity preferences of the olfactory cortex (OLF) to regions included in the proposed hedonic evaluation model (consisting of OFC, mPFC, OLF, ACC, AMYG), for high- (panel A) and low pleasantness (panel B) stimuli. Only the beta band ([13–30] Hz) was considered.

Figure 6

Top 10 strongest connections (from the wPLI adjacency matrices) within the hedonic olfactory model, averaged across subjects and trials. (Panel A) shows the high-pleasantness condition, (panel B) the low-pleasantness counterpart. Beta band ([13–30] Hz) results are displayed.

Figure 7

Statistically significant differences in laterality coefficients between high- and low pleasantness stimuli (individual regions, paired samples t-tests), including results of graph null-hypothesis testing (*: p < 0.05, $**$: p < 0.01). Coefficients were computed based on graph metrics; positive coefficients indicate a left-dominant response, right-dominance is indicated by negative coefficients. FDR corrected p-values are depicted; error bars show the standard error.