Covariance-based decoding reveals content-specific feature integration and top-down processing for imagined faces versus places

Visual imagery consists in endogenous image generation in the absence of exogenous visual stimulation. Previous neuroimaging studies have shown that visual imagery activates occipital, parietal and frontal areas. Sparse activity has been observed throughout this extended brain network when the imagery targets are high-level visual objects. The aim of the present study was to investigate whether we can decode face vs. place imagery based on functional connectivity measures. In particular, we tested whether short-range connections in occipital and inferior-temporal areas-involving visual processing-are rich enough to discriminate face vs. place imagery or if long-range connections between occipital, parietal and frontal areas-involving memory-guided attentional modulations-are also important. In order to address this question we designed an MEG experiment (N= 11). Previous neuroimaging studies used a …