Recently, it has been proven that the emotional aspect directly influences the learning process, so that, based on data mining techniques, this behavior has been sought to be characterized. This has made clustering techniques become one of the most used techniques for this purpose. However, studies where emotional data obtained from a person’s brain activity are used, are rare. For this reason, the present study aims to implement and compare advanced clustering techniques based on emotional metrics obtained through Brain-Computer Interfaces, captured in an AR-Sandbox, which fulfills the role of a learning environment. The evaluation of these techniques is carried out using internal criteria such as silhouette coefficient, Composed Density Between and within, Calinski-Harabasz and other statistical measures. When carrying out this study, it was obtained as a result that, the Density-Based Spatial Clustering of Application with Noise and Density-Based Hierarchical Spatial Clustering of Noisy Applications algorithms as the Density-based clustering methods, presented a better level of well-separation, cohesion and compaction, in comparison to the rest of the techniques implemented.