People with profound Intellectual Disability (ID) have difficulties expressing their needs due to their limited communication skills. These skills are usually limited to gestures and vocalizations many of which so idiosyncratic that can only be interpreted correctly by their caregivers/parents often with a large gap between their interpretations. In this paper we explore the application of Brain Computer Interface (BCI) in designing a methodology to interpret the mental state of these individuals during interaction within their environment, focusing on the phenomenon of surprise in interaction. We report on our progress towards predicting the participants' reactions to expected and unexpected events by classifying their brain activities monitored by electroencephalography (EEG signals). This study takes a major step forward towards improving the communication skills with individuals with ID in particular with those outside their family circle.