Classification of Tactile Perception and Attention on Natural Textures from EEG Signals

Abstract

Brain-computer interface (BCI) allows people who have lost their motor skills to control robot limbs based on electroencephalography (EEG). Most BCIs are guided only by visual feedback and do not have somatosensory feedback, which is an important component of normal motor behavior. The sense of touch is a very crucial sensory modality, especially in object recognition and manipulation. When manipulating an object, the brain uses empirical information about the tactile properties of the object. In addition, the primary somatosensory cortex is not only involved in processing the sense of touch in our body but also responds to visible contact with other people or inanimate objects. Based on these findings, we conducted a preliminary experiment to confirm the possibility of a novel paradigm called touch imagery. A haptic imagery experiment was conducted on four objects, and through neurophysiological analysis, a comparison analysis was performed with the brain waves of the actual tactile sense. Also, we obtained the average classification performance of 68.07(±4.0)% and 57.55(±1.83)% with basic machine learning algorithms.