A Simple, Spectral-Change Based, Electrocorticographic Brain–Computer Interface

Abstract

A brain–computer interface (BCI) requires a strong, reliable signal for effective implementation. A wide range of real-time electrical signals have been used for BCI, ranging from scalp recorded electroencephalography (EEG) (see, for example, [1, 2]) to single neuron recordings (see, for example, [3, 4]. Electrocorticography (ECoG) is an intermediate measure, and refers to the recordings obtained directly from the surface of the brain [5]. Like EEG, ECoG represents a population measure, the electrical potential that results from the sum of the local field potentials resulting from 100,000 s of neurons under a given electrode. However, ECoG is a stronger signal and is not susceptible to the artifacts from skin and muscle activity that can plague EEG recordings. ECoG and EEG also differ in that the phenomena they measure encompass fundamentally different scales. Because ECoG electrodes lie on the cortical surface, and because the dipole fields [7] that produce the cortical potentials fall off rapidly (forumala presented)., the ECoG fundamentally reflects more local processes.