This is the current undocumented, messy source code of the processing neural network I wrote, along with DiRusso's data. You may notice that the neural network has only one layer of weights, making it really just a matrix. ;-) However as it is trained by the same algorithm used to train backpropagation networks, I feel it is important to call it a neural network. In any case, if nonlinear associations need to be made in the future, we will need to add more layers.
Also included is the OpenEEG VEP data acquisition program I wrote with the help of Jim Peters.