This week we will consider the implications of interpreting multilayer perceptrons, and in particular examining distributed representations in such networks, and will use this consideration to place connectionist modeling within a productive cognitive science.