COMPUTATIONAL MODELS OF RETINAL AND BRAIN FUNCTION

点击翻译按钮获取中文摘要

DESCRIPTION (Adapted from applicant's abstract): The long-range goal of the research program is to understand the neural basis of visual behavior. The principal investigator proposes to work towards this goal by studying the visual system of the horseshoe crab (Limulus polyphemus). Its retina contains the one of the largest neural networks for which a quantitative model exists. In addition, a circadian clock modulates the function of the retina, and it is relatively well known what the animal can see underwater day and night. The computational models have yielded new insights about visual coding during the day. Building on this work, the principal investigator plans to combine theoretical and experimental techniques to investigate what information the eyes send to the brain for the animal to see at night. Moreover, he plans to study how the brain decodes the information it receives from the eye. Specific aims are to investigate: 1. Retinal coding of visual information underlying behavior at night. How does the animal see so well at night? During the day, across its ensemble of optic nerve fibers, the eye functions as a global feature detector generating "neural images" of mate-like objects. Does the eye use such a distributed code at night? 2. Brain networks for target detection. How do brain networks decipher the neural codes they receive from the eye day and night? From multiple recordings of brain cells, the principal investigator will develop simplified models of brain network and then drive the models with neural images generated by his cell-based models of the retina. 3. Visual information in the animal's natural environment. The principal investigator will determine whether natural fluctuations in environmental lighting once thought to be noise increase retinal signal-to-noise performance and enhance the visibility of objects important to the animal. Preliminary experiments indicate they do. The principal investigator plans a systematic study involving computation, electrophysiology, neuroanatomy and behavior. He has developed a realistic cell-based model of the daytime state of the 1000-neuron retina. Furthermore, he verifies model predictions by recording responses from single optic nerve fibers of a behaving animal, while recording what the animals sees with a miniature underwater video camera, "CrabCam." And he has developed parallel techniques for studying brain networks. His new results and new techniques establish Limulus as a good model for analyzing the link between neural coding and visual behavior. The proposed studies promise to reveal basic information-processing mechanisms in sizable neural networks. These studies may be applicable to even more complex networks--a major goal of neuroscience.