Afferent Effects on Brain Computer Interfaces: An Experimental Analysis

Abstract

A brain-computer-interface (BCI) or brain-machine-interface (BMI) uses brain signals to drive external devices without participation of the spinal and peripheral motor system. In most BCIs the users brain activity is acquired via amplifiers and filters and decoded using an on-line classification algorithm. In turn, this output is fed back to users, which allows them to modulate their brain activity. The feedback usually consists of sensory stimuli, such as visual and auditory or vibrotactile, varying proportionally to the classified brain activity, a discrete reward for a particular brain response, a verbal response (such as yes or no), the movements of a prosthesis or wheel chair, or direct electrical stimulation of muscles or brain. Thus, feedback of the consequences of the brain activity carried out to control the device is likely an essential part of a successful BCI. In my thesis I focus on the influences that afferent sensory information could elicit on the BCI performance. In EEG based BCIs the activity used is very disperse due to the volume conduction effect, making afferent information converging in areas used for classification. Indeed, in my thesis I study the effects of sensory activity on the classifier performance and on the feedback modality chosen to accomplish the goal of each BCI system. In my case I focus on the potential use of BCI technology coupled On-line with robotics to perform neurorehabilitation in stroke and amyotrophic lateral sclerosis (ALS) patients.

Date of Award	2012
Original language	English
Awarding Institution	Max Planck Graduate School of Neural and Behavioural Science/Hospital of the university of Tübingen