The development of the cat lateral geniculate nucleus depends on a combination of activity dependent and activity independent processes. A Hebbian model of the development of topography and ocular segregation in the LGN was first presented by Keesing, Stork and Shatz (1992). This paper investigates two key features of the model: the normalisation schemes used and the nature of the retinal inputs. It is shown that the normalisation of weights into an LGN cell is not necessary, and that small positive between eye correlations do not prevent normal development.
Download compressed postscript file