Willmore et al. — Neural Representation of Natural Images in Visual Area V2 (2010)

Ben D. B. Willmore, Ryan J. Prenger, and Jack L. Gallant — Neural Representation of Natural Images in Visual Area V2 doi:10.1523/JNEUROSCI.4099-09.2010

The authors explore response properties in area V2 using natural images. The spatial but not the temporal statistics of natural images are present in their stimuli. They use the Berkeley Wavelet Transform as a non-linear preprocessing step to generate components to relate to V2 responses by reverse correlation. Phases are quadrature encoded. The authors find that the BWT fits V1 and some V2 cells well with only a few excitatory components. Other cells in V2 are a mixture of several wavelet components as well as suppression from other components. Qualitatively, it appears that components are summed to create robust, broad-band edge detectors in V2. The authors note that inhibition plays a large role in the receptive field properties and that natural input statistics are important for recruiting realistic inhibition.

No comments:

Post a Comment