Neuroanatomy is one of the most challenging sections of anatomy to learn, partially related to the intricate relation of multiple 3D structures. As part of the medical student curriculum, it is usually taught in 2D using illustrations and plastinated brain section, since the number of hours devoted to anatomy have dropped in the curriculum, making the dissection of brain too time-consuming to be done. In this study we are analyzing the role of innate spatial ability of novices in learning some basic structures and placing them back in a 3D volumetric brain. Two tasks are performed after a short training session: the first one is to localize the ventricular tip as would be required during a temporal lobectomy, and the second task requires that the subject 'reconstruct' 3D anatomical structures within the context of our 3D brain model. We report our findings on the performance scores obtained from a population of subjects of differing backgrounds and spatial abilities.