<p>Bidimensional regression was developed by Waldo Tobler since 1960s to evaluate the resemblance between two plans. This is the most used way in geography to highlight differences in shape between maps. However, bidimensional regression can only compare the shape of two background maps simultaneously. This article presents two ways of generalizing bidimensional regression in order to be able to compare more than two geographic maps using a geometric morphometric approach. For this purpose, we took inspiration from the generalization of the ordinary Procrustes algorithm. An example shows that the generalized bidimensional regression makes it possible to compare the shape, and therefore the structure of several cognitive maps, thanks to the creation of a morphospace. More precisely, it makes it possible to produce a typology of maps based on their shape on the one hand, and to identify the main variations in shape of all the maps analyzed on the other.</p>