Multidimensional Projections (MPs) are key tools used to support the analysis of multidimensional data. MPs can project data to a low dimensional representation, typically visualized as a 2D scatterplot where similar elements are conveniently positioned in close neighborhoods. However such visualizations tells us which points are similar, but not why. Our aim is, thus, to enrich 2D MP scatterplots with explanatory information telling users which key dimensions make closely-projected points similar