Fig. 2.

The relation between lesion probability and overall lesion volume. Reanalysing the data presented in Mah et al. (2014), here we used Bayesian logistic regression (Makalic and Schmidt, 2016), performed independently at each voxel, to estimate the odds ratio for the relation between the volume of a lesion and the probability of damage at each location across the brain. The odds ratios are visualised as box glyphs whose colour and dimensions are proportional to the estimated value. Note the enormous variation in the odds across the brain, and the complex anatomical pattern—distinct from the pattern observed in Fig. 1—it follows. Using lesion volume as a regressor in a mass-univariate lesion-deficit model will thus unquestionably distort the inference, penalising voxels more commonly hit by large lesions such as those that reach the cortical periphery or fall on vascular territorial boundaries. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)