Brain-Computer Interfaces (BCI) can provide innovative solutions beyond the medical domain. In human research, visuospatial attention is often assessed from shifts in head or gaze orientation. However in some critical situations, these behavioral features can be dissociated from covert attention processes and brain features may indicate more reliably the spatial focus of attention. In this context, we investigate whether EEG signals could be used to enhance the behavioral performance of human subjects in a visuospatial attention task. Our results demonstrate that a BCI protocol based on adaptive or warning displays can be developed to shorten the reaction time and improve the accuracy of responses to complex visual targets.