The contribution of Artificial Neural Networks in psychometrics can help integrate the explanatory approach that often requires strong assumptions on data and the predictive one that, on the contrary, only needs mild assumptions on input data. Predictive techniques are able to identify data patterns and generate accurate predictions of output values starting from new sets of data, which is relevant in different psychometrics domains. Here we provide three examples of applications related to psychometric data analysis, language analysis and spatial cognition. These studies show that integrating machine learning techniques into traditional psychometric data analysis allows to work on different input data, including not standard ones, to complement traditional procedures with new ones and to identify unexpected patterns.