This chapter starts off with an overview of the networks used in this simulation. It will then go on to describe the training regime and will finally present the results of the simulations.
Two networks are used in the current simulation. The first one is a control network, similar to Shillcock and Monaghan (2003) where the words have the same frequency at each fixation position. The second one, the ’fixation’ network, has different fre- quencies for the different fixation positions as in table 1.1 from previous chapter. The control network is essential in order to be able to compare the performance of the fixa- tion network and to see if there is any difference in the behaviour of the two networks.
The networks both have two input layers, one corresponding to the right visual field and the other one to the left visual field. Both of these are connected to separate