Computational Maps in the Visual Cortex
     Figure 5.7
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. 5.7. Self-organized V1 afferent and lateral weights. The weights of the neuron in Figure 5.5 are shown after self-organization. In (a), the OFF weights were subtracted from the ON weights, as in Figure 4.6c. This neuron prefers a line oriented at 60o, i.e. diagonal from bottom left to top right, and responds most strongly to a white line overlapping the light portion of its RF, surrounded by black areas overlapping the dark portions. Other neurons developed similar RFs with different preferred orientations (Figure 5.8). This type of RF structure is commonly seen in biological V1 neurons (Figure 2.2d; Hubel and Wiesel 1962, 1968). In the lateral weight figures (b, c, and other later such figures), the following convention is used: The hue (i.e. color) represents the orientation preference of the source neuron, according to the key along the top. The saturation of the color (i.e. its fullness, or intensity) represents how selective the source neuron is for this orientation; unselective neurons are shown in gray. The value of the color (i.e. its brightness) indicates the strength of the connection, with nonexistent or zero-weight connections shown as white. The jagged black outline traces the original lateral connections, and a small white square (in c) identifies the neuron itself. Using such a scale, plot (b) displays the lateral excitatory weights of this neuron. All connected neurons are strongly colored blue or purple, i.e. orientations similar to the orientation preference of this neuron. The lateral inhibitory weights are plotted in (c). After self-organization and connection pruning, only connections from neurons with similar orientations remain, and they are extended along the preferred orientation of this neuron. The connection pattern is patchy, because connections from neurons with opposite preferences are weak or have been pruned away entirely. Such patchy, orientation-specific connection patterns are also seen in biological V1 neurons (Figure 2.7; Bosking et al. 1997; Sincich and Blasdel 2001).