The impact of heterogeneous spatial autocorrelation on comparisons of brain maps

It is increasingly common to statistically compare macroscopic brain maps to assess how spatially similar they are. Due to the presence of spatial autocorrelation, statistical inference can be challenging; to address this, random permutation approaches based on null models are widely used. Here, we show how that the presence of heterogeneity in the spatial autocorrelation across brain maps may affect statistical inference in common approaches for spatially correlated maps. In response, we highlight the need to explicitly model spatial processes, including non-stationarity, for more accurate statistical inference. We illustrate how a Bayesian spatial regression approach can be applied to compare functional and structural cortical brain maps, even in the presence of heterogeneity. By explicitly modelling spatial processes underlying the data, a much wider and more sophisticated range of neurobiological questions can be answered about the relationship between brain maps than with current approaches.