The Retina as a Dynamical System

Bruno Cessac
DOI: https://doi.org/10.48550/arXiv.2012.05501
2020-12-10
Abstract:Considering the retina as a high dimensional, non autonomous, dynamical system, layered and structured, with non stationary and spatially inhomogeneous entries (visual scenes), we present several examples where dynamical systems-, bifurcations-, and ergodic-theory provide useful insights on retinal behaviour and dynamics.
Neurons and Cognition
What problem does this paper attempt to address?
### What problems does this paper attempt to solve? The paper "The Retina as a Dynamical System" attempts to explore the dynamic behavior of the retina and its ability to process visual information from the perspective of dynamical system theory. Specifically, the author attempts to answer the following key questions: 1. **How does the retina pre - process visual information?** - The retina is not just a simple "camera" that converts photon streams into action potentials, but pre - processes visual information through its specific layered structure and complex neural circuits to reduce redundancy and improve the speed, efficiency, and reliability of visual responses. - Especially in motion processing, the retina can detect objects with different motions on the background, predict the trajectories of moving objects, and respond to changes in their trajectories. 2. **What are the dynamic characteristics of the retina?** - The retina is described as a high - dimensional, non - autonomous dynamical system with a layered structure and spatially non - uniform inputs (visual scenes). The author uses dynamical systems, bifurcation theory, and ergodic theory to provide useful insights into retinal behavior and dynamics. - For example, retinal waves are spontaneously generated in the early developmental stage and are crucial for the organization of the visual system. The multi - scale spatio - temporal dynamic characteristics of these waves and their sensitivity to changes in physiological parameters are the focus of research. 3. **How do local circuits in the retina process visual information?** - There are specific local circuits in the retina, composed of specific types of neurons and non - linear synaptic connections, which can efficiently process local visual information. These circuits form regular mosaics across the entire retina, providing parallel information - processing channels. - For example, the roles of horizontal cells and amacrine cells and their interactions are crucial for the processing of visual information. 4. **How do changes in retinal dynamics affect its function?** - The behavior of the retina may be affected by its structural or physiological changes. For example, in the early developmental stage, retinal waves occur in the retina, which is very important for the structuring of the visual system. In later life, the retina may be affected by specific pathologies, leading to impaired vision. - In addition, drugs can affect the retina's response to visual stimuli in a controllable manner. These changes and their effects on retinal dynamics can also be studied through dynamical systems and bifurcation theory. 5. **What are the dynamic mechanisms of retinal waves?** - In stage II, retinal waves are generated by Starburst Amacrine Cells (SACs) through acetylcholine (Ach) interactions. SACs emit spontaneous calcium bursts, triggering the activity of surrounding SACs, thereby initiating the waves. - The propagation of the waves is limited by the slow After - Hyperpolarization Current (sAHP), which prevents the waves from propagating again in the same area for a certain period of time. ### Summary In general, this paper aims to gain an in - depth understanding of how the retina, as a complex organ, processes visual information through the perspective of dynamical system theory, especially the specific mechanisms in motion processing, the generation and propagation of retinal waves, etc. In this way, the author hopes to reveal the uniqueness of the retina and provide a new perspective for understanding how the brain processes information.