TABLE OF CONTENTS Neurons in visual cortex of cat and monkey can usually be driven by both eyes, however one of the eyes usually dominates. Many models have tried to account for the development of this ocular dominance segregation.
The simplest model in which we can describe ocular dominance, is composed of just two inputs, one for the left eye and one for the right eye. For the type of Hebbian rules examined here the receptive fields are determined by the correlation function. If it is a PCA (Equation 5.5) rule that does not have the Oja normalization term, the dynamics are dominated by the first PC (see Equation 5.4) and the receptive field is the first PC.
In this simple 2D case we have two weights ( m l, m r) and two inputs ( d l, d r). The correlation function Q is a two dimensional matrix:
where <> denote an average over the input distribution. Due to the symmetry of the two channels < d ld l >=< d rd r >= a and < d ld r >=< d rd l >= b where a and b are two constants and a ? b.
Thus Q has the form
This matrix has two eigenvectors:
with corresponding eigenvalues
In such models ocular dominance emerges if ? 2 > ? 1
