Individual Differences in Mental Imagery Modulates Effective Connectivity of Scene Regions at Rest

Successful navigation relies on the human ability to identify, perceive and correctly process the spatial structure of a scene. It is well known that mental imagery plays a crucial role in navigation. Cortical regions encoding navigationally-relevant information are active during both perception and mental imagery of navigational scenes. The spatial navigational system relies on the activity of at least three visual cortical areas selectively responding to scenes: the Occipital Place Area (OPA) in the dorsal occipital lobe, the Retrosplenial Complex (RSC) within the parietal-occipital sulcus, and the Parahippocampal Place Area (PPA) at the boundary between the posterior parahippocampal cortex and the anterior lingual gyrus. Although functional connectivity between these regions has been studied, it remains unknown which is the causal influence of these regions on each other and whether their connectivity reflects the individuals’ ability to imagine a scene. Forty-two healthy volunteers (mean age = 32.21, SD = 4.17, 24 female) underwent two fMRI sessions including two localizer scans for scene-selective regions and two resting-state scans. fMRI data were acquired on 3T Philips Allegra and Siemens Achieva scanner: for Allegra 242 (localizer scans) and 128 (rest scans) volumes, TR=2s, TE=30ms; for Achieva 249(localizer) and 160(resting-state) volumes, TR=1.9s, TE=25ms. The localizer fMRI experiment consisted of passive viewing of eight alternating blocks of face and place pictures presented for 300 ms every 500 ms, interleaved with a fixation period of 15 s on average. In the resting-state fMRI scans subjects lied at rest with eyes closed and no experimental task was imposed. The Vividness of Visual Imagery Questionnaire (VVIQ) (Marks, 1973 ) was administered to measure the vividness of visual imagery. Scene-responsive regions (PPA, RSC and OPA) were defined in each subject as the regions responding more strongly to places than to faces. Six head motion regressors, CSF, WM signal and discrete cosine transform were specified as regressors on resting-state General Linear Model. Spectral Dynamic Causal models (DCM) for resting state fMRI (Friston et al., 2014) estimating intrinsic effective connectivity from observed BOLD responses were applied to regional time series computed as the principal eigenvariates of resting-state time series after standard preprocessing and removal of most typical confounds and artefact sources (Behzadi Y., 2007). Full-model DCMs were specified and inverted for each session separately. To estimate the average connectivity at the subject level we specified Parametric Empirical Bayes (PEB; Friston et al., 2016) modelling average connectivity (within subject) over sessions. Moreover, to define whether individual differences in mental imagery are predicted by the intrinsic connectivity between scene-selective regions, we performed a further PEB analysis with the VVIQ score as a covariate.