The goal of this research is to develop tools and techniques for studying voluntary control of cortically recorded event related potentials (ERP) in humans. The application of this research is focused on the development of a Direct Brain Interface (DBI) to help persons with severe communication-impairing disabilities. Signal processing techniques for estimating ERP quality are developed and evaluated. A method of estimating the Signal to Noise Ratio (SNR) of ERPs is implemented as the primary tool for measuring quality. The approach developed is shown to be very effective in identifying ERPs useful for the DBI.
The quality measurement techniques are the basis of a new on-line, real-time instrument, an ``ERP Analyzer.'' The instrument, which was developed as an integral part of this dissertation research, is used to identify high-quality ERPs for further experimentation, such as feedback training. It provides a test bed upon which ERP experiments and different protocols can be conducted. The instrument was tested extensively with pre-recorded human electrocortical signals as well as during experiments with human subjects. The ERP Analyzer also includes the first real-time prototype of the University of Michigan Direct Brain Interface. A protocol was designed and tested to investigate voluntary control of cortically recorded ERPs in human subjects. Visual feedback of ERP quality was used in a first attempt to ``train'' subject's to shape or modify their ERPs. The signal processing techniques and ERP Analyzer together were key elements in the design of the protocol. The experimental protocol was used in a study of two subjects and the results are presented as case studies. While the experimental results did not answer the question of whether cortical ERPs can be voluntarily controlled, they demonstrated that such experiments are feasible with the tools and techniques developed. Suggestions are made for improvements to the protocol and it is hoped that these will lead to exciting results in the near future.