Computational Maps in the Visual Cortex
     Figure 12.8
MiikkulainenBednarChoeSirosh
Home    
About the Authors
Back Cover    
Table of Contents 
Sample Chapter 
Figures    
References    
Errata    
Demos     
Talks/Courses 
Software    
Credits    
Purchase online at:

springeronline.com
amazon.com

Click on the image to see a PDF version (for zooming in)

Fig. 12.8. Overcoming noise with a long refractory period. A network of 30 neurons with global excitatory lateral connections was simulated for 500 iterations. A significant level of noise (7%, 70 times the noise in Section 12.3.2) was added to the membrane potential at each iteration. Such high noise cannot be tolerated by just increasing the lateral excitatory contribution &gammaE. However, increasing the absolute refractory period can make the model robust even in this case. (a) When there is no absolute refractory period (tr = 0), the activities are random. (b) When the absolute refractory period lasts for three iterations, the activities start to synchronize loosely. (c) When it lasts for five iterations, the activities are strongly synchronized. With longer periods between firing, the noise is effectively washed out.