Whereas a user visually processes geographic information faster and more accurate than a computational system, a system processes a large number of geographic data much faster than a user. To minimise this large capability discrepancy, we propose a cognitively-adequate design of geovisualisation as an external aid to visual information processing. Our main objective is to release the limited cognitive workload of users by adapting the design of geovisualisations to their visual scanning capabilities. For a proof of concept, we design 2D thematic maps and 3D non-photorealistic visualisations, and evaluate the efficiency of visual scanning strategies with a computational attention-model and the eye-movement recording method.