In human electrophysiological experiments, stimuli typically take the form of simple, isolated, discrete events, such as flashes and beeps. Such stimulation is not always ideal for examining cognitive processes. We are interested in developing novel methods of continuous stimulation to allow us to address several important and outstanding questions involving selective attention in humans using electrophysiology. In particular we are interested in attention to natural speech - the so called 'cocktail party problem'; that is, our ability to easily attend to one speaker in a multi-speaker environment.