Simulations

The network described above was implemented in the PLANET neural network simulation environment (Miyata, 1991) on a Sun SPARC 1 computer. PLANET is an environment in which simple neural network simulations may be constructed using a language developed specifically for that purpose. Users can interactively examine activations within the network and its contents, interactively ``train'' the network to associate inputs with outputs, and observe the resulting error rates. The PlaNet code representing the network and the code for presenting stimuli in the context of either the affective valence identification or lexical decision task is included in Appendix F.

The network was used to simulate reaction times and error rates on the lexical decision and valence identification tasks, using varying training sequences and simulated SD's. The network's aggregate performance over a series of 50 trials in each condition was averaged and compared to the average performance of people in the analogous condition. Additionally, between trial variations were analyzed to see how well they approximated individual differences observed on the tasks.