To feedback or not to feedback? Some open questions for brain-computer interfaces use

Brain-Computer Interfaces (BCI), allowing to operate the enviroment without muscular activation, can be used as assistive technology as well as neurorehabilitation tool. A BCI exploits brain signal changes induced by event-related potentials (P300) or by voluntary control of the user (Slow Cortical Potentials or Motor Imagery) and this can be achieved using a classifier. A key feature for BCI systems is user-classifier mutual learning, which can be modulated by feedback. There is experimental evidence that feedback can have a positive as well as a negative effect and this is relevant in using BCI both as assistive technology and as neurorehabilitation tool. The aim of this study was to investigate a possible sequence effect on overall accuracy in administering a first training session without feedback and then a second one with feedback. Moreover, we evaluated also a gender effect. We considered a sample of sixteen subjects (ten males), using an EEG-based BCI and executing a Motor Imagery task. Our results, combined with data from literature, indicate the presence of a "cognitive carry-over" effect for continous feedback in motor imagery, with gender differences not statistically significant. Such a cognitive effect means that the presence or absence of feedback changes the cognitive status of the BCI user and thus the information embedded in the features the classifier uses. Therefore, there should be a sort of cognitive coherence in BCI-training and the assessment of user attitude toward feedback could improve BCI usefulness and effectiveness.