PGLISSOM: Contour Integration/Segmentation Demo

Yoonsuck Choe and Risto Miikkulainen. Contour integration and segmentation in a self-organizing map of spiking neurons. Biological Cybernetics, 90:75-88, 2004.

1. Self-organization of orientation maps: the left plate is the orientation preference, the center plate is the orientation selectivity (black is low, white is high), and the right-most plate is the combined preference-selectivity. Each animation starts from iteration 0 and ends at iteration 40,000, and loops.

   

2. Contour integration: The contour integration inputs and the spike activity of MAP2 for 600 iterations are shown for each input configuration. The degree of synchrony within each area of the map responding to the oriented Gaussian inputs were measured. See the paper for more detail on the results. For the configurations with less orientation jitter, the population of neurons responding to the oriented Gaussians are more consistently synchronized. If it is hard to see what's going on, first see the contour segmentation demo at the end of this page.
   • 0-degree

     

   • 30-degrees

     

   • 50-degrees

     

   • 70-degrees

     

3. Contour Segmentation: when two (or more) equally salient contours are present in the scene, the alternating phase of synchronized spikes is more prominent.