The part family/machine group identification (or formation) is the crux of implementing Group Technology, and a well-studied subject. However, most of the approaches have neglected the original family concept proposed by Burbidge, that there are already naturally occurring families existing. A desirable approach should be that of identifying these families rather than forcing to form the families. This paper describes a neurocomputing model which is inspired by the way the biological neuronal systems reach intelligent decisions. A comprehensive survey of previous approaches is presented. The simulation results from an example are provided to show how the model is used to identify the part families and the machine groups. The advantages of the neurocomputing model and future directions are discussed.