Simulation environments (SE) are becoming important tools that can be leveraged to implement training protocols and educational resources. Electroencephalography (EEG) is used to compare the effects of different types of feedback in SE, but it can be challenging to know which aspects of EEG represent the impact of different types of feedback on neural processing. For this study, the Minimum-Redundancy-Maximum-Relevance (MRMR) algorithm, and the MRMR in conjunction with the Mann-Whitney U statistical test were applied to select the most relevant EEG features associated with haptic and non-haptic feedback in a simulated drilling task. EEG was analyzed based on the extraction and selection of different machine learning features.