Model and Large-scale Simulator of a Biological Retina, with Contrast Gain Control
Author | : Adrien Wohrer |
Publisher | : |
Total Pages | : 340 |
Release | : 2008 |
ISBN-10 | : OCLC:494148912 |
ISBN-13 | : |
Rating | : 4/5 (12 Downloads) |
Book excerpt: The retina is a complex neural structure. The characteristics of retinal processing are reviewed extensively in Part I of this work: It is a very ordered structure, which proceeds to band-pass spatio-temporal enhancements of the incoming light, along different parallel output pathways with distinct spatio-temporal properties. The spike trains emitted by the retina have a complex statistical structure, such that precise spike timings may play a role in the code conveyed by the retina. Several mechanisms of gain control provide a constant adaptation of the retina to luminosity and contrast. The retina model that we have defined and implemented in Part II can account for a good part of this complexity. It can model spatio-temporal band-pass behavior with adjustable filtering scales, with the inclusion of plausible mechanisms of contrast gain control and spike generation. The gain control mechanism proposed in the model provides a good fit to experimental data, and it can induce interesting effects of local renormalization in the output retinal image. Furthermore, a mathematical analysis confirms that the gain control behaves well under simple sinusoidal stimulation. Finally, the simulator /Virtual Retina/ implements the model on a large-scale, so that it can emulate up to around 100,000 cells with a processing speed of about 1/100 real time. It is ready for use in various applications, while including a number of advanced retinal functionalities which are too often overlooked.