Fast habituation effect following prolonged presentation of odorants is observed in adults. However, this habituation phenomenon of the olfactory system is not fully understood. The olfactory system is one of the first sensory systems to be functional during fetal life and has a high behavioral importance following birth¹. Olfactory fMRI is considered to be challenging due to (i) strong orbito-frontal signal loss, (ii) fast habituation of the olfactory system to prolonged activation^2,3 and (iii) complex spatio-temporal dynamics of the olfactory system⁴. The aim of this study is to better characterize habituation effect of sustained odorant stimulation and to investigate if this effect is already present in newborns.

Acquisition. fMRI was acquired at 3T (Siemens Trio, Erlangen, Germany) during olfactory stimulation in 11 healthy volunteers (6 women; median age: 29.6 years, range: 25-43 years) and in 28 full-term newborns (mean gestational age=39.5 weeks) using respectively a 12-channel head-receive coil and an 8-channel neohead coil (LMT medical systems, Lübeck, Germany). Neonates were tested in their first week of life, during natural sleep or while resting quietly, without any sedation. Functional images were obtained using a single-shot T₂^*-weighted GE-EPI sequence (TR=1800ms, TE=25ms, 30 slices, voxel size=2.2x2.2x3.5mm³). In adults, a B₀ field-map was acquired to correct geometric distortions induced by local magnetic field inhomogeneities. A high-resolution 3D-T₁ (MPRAGE, voxel size=1x1x1mm³) or a T₂-weighted (113 coronal slices, voxel size=0.78x0.78x1.2mm³) images were also acquired respectively in adults and infants for anatomical reference.

Stimulations. Three different odorants (banana, cabbage and eucalyptol) were delivered using a home-made four-way odorant delivery system. Each odorant was presented in a pseudo-randomized order during 20 seconds and separated with a neutral odor (water). Two runs of 10’30 minutes including five repetitions of each odorant were acquired for each participant.

fMRI preprocessing. Functional images were pre-processed with SPM8 and included: realignment and unwarping, slice-timing correction, coregistration on the structural image, normalization to MNI space or to a T₂ neonatal template (voxel size=2x2x2 mm³) and spatial smoothing using an isotropic Gaussian kernel of 8mm (adults) or 6mm (infants).

Habituation modelling. Habituation h(t) was modelled as a decreasing exponential $$$h(t)=e^{-\frac{t}{\tau}}$$$ where the time constant $$$\tau$$$ varies from 0.05s to 1000s with 21 different values sampled exponentially. We also modelled the cases of no habituation ($$$\tau\rightarrow+\infty, h(t)=1$$$) or immediate habituation ($$$\tau\rightarrow0, h(t)=\delta$$$ where $$$\delta$$$ stands for the Dirac function). The 20-second bloc stimulation is then weighted by these habituation functions and convolved by the canonical hemodynamic response function (Figure 1).

First-level analyses. For each subject and each $$$\tau$$$, a General Linear Model (GLM) was built including a regressor for each odorant. To accommodate the high level of motion in infants, images with framewise displacement superior to 1mm as well as the previous image and the two following images were excluded⁵ and sessions including at least one stimulation block of each odorant were built with the remaining images. Motion parameters were included into the model as covariates and low-frequencies were removed using a discrete cosine transform basis set with a filter cut-off period of 256s.

Second-level analyses. For each different habituation model and for each odorant, a second level analysis was performed using a random-effect GLM analyses in which the inputs are the contrast maps obtained during the first-level analysis. For each odorant, combined activation maps were built using the different habituation models.

Habituation maps. For each odorant, a habituation map was built by extracting optimal habituation parameter ($$$\tau$$$) for each voxel considered significantly activated (p<0.005).

Whereas no significant activation were detected in adults without modelling habituation (p<0.001, uncorrected), we obtained activation of bilateral piriform cortex, amygdala and parahippocampal gyrus (primary olfactory cortex), bilateral fronto-orbital cortex (secondary olfactory cortex) and bilateral insula after modelling habituation (Figure 2). The corresponding habituation maps for the different odorants show a very fast habituation in adults (Figure 3). In newborns, we observed activations in piriform cortex, orbitofrontal cortex and anterior cingulate cortex. However, contrary to adults, the habituation effect during the 20-second sustained stimulation was not observed (Figure 4).Activations in the primary and secondary olfactory cortices were observed both in adults and newborns. Habituation effect of sustained odorant stimulation was strong in adults, but was not present in newborns in the tested time-period even if it was previously demonstrated after repetitive stimulations⁶. This absence of habituation to sustained stimulation may be explained by the immaturity of inhibitory system at this age⁷. This study shows that the olfactory cortex of newborns is highly functional soon after birth and that the habituation effect of sustained stimulation is negligible compared to adults.