A Poll Oriented Classifier for Affective Brain Computer Interfaces

Affective Computing and Brain Computer Interface (BCI) are two innovative and rapidly growing fields of research. Affective Computing aims at equipping machines with the human capabilities of observe, understand and express affecting features; BCI aims at discovering novel communication channels and protocols, through the monitoring of the brain activity. Emotion recognition plays a central role in both these research fields. In this work we present an EEG poll based classification algorithm for self-induced emotional states used for BCI. We tested the approach using three emotions: the disgust produced by remembering an unpleasant odor (a stink), the pleasantness induced by the memory of a fragrance and a relaxing state. Preliminary experimental results are also reported.