The present paper presents the results of an experiment aimed at assessing the impact of different levels of immersion on performance in a Spatial Perspective Taking (SPT) task. Since SPT is an embodied skill, the hypothesis was that the more immersive a tool is, the better the performance should be. Ninety-eight students from a local primary school have played with three different versions of a game: (i) completely immersive with a Head Mounted Display, (ii) semi immersive on a computer screen and (iii) non-immersive where no movements were possible for the player. Results showed that in the immersive versions of the game, players obtained higher scores than in the non-immersive version, suggesting that an immersive tool can better support performance in a SPT task.