FIGURE 2.9 Decay rates. This figure demonstrates the effects of changing the decay constant for the hidden and output layers. A decay constant of 0.8 means each node loses 20% of its activation each time step; a decay constant of 1 means no loss. Each line represents the average of five independent peak procedure simulations. The heavy black line represents the standard value for this parameter in all other simulations in this work.

Finally, there is a learning parameter that governs how much change can occur in the model's weights each time step. This constant is used in Equations 1 and 3 of Figure 2.5. On nonfood trials the learning constant was 0.01, and on food trials it was 0.1. How various values of this constant affect the interval timing functions is shown in Figures 2.10 and 2.11.

0 0

Post a comment