Watching a vertebrate brain learn and behave in a virtual environment

点击翻译按钮获取中文摘要

DESCRIPTION Abstract: In order to study the neural basis of navigation, learning, and memory, it is important to simultaneously record the activity of large populations of neurons involved in execution of behavior. This has been difficult or impossible in awake, freely moving preparations and the anesthetics and paralytics used to facilitate recordings often alter or abolish the relevant behaviors. I propose to establish a fictive swimming assay that allows paralyzed larval zebrafish to navigate through- and interact with a virtual environment. The sensory feedback provided to the animals will consist of visual cues through video projection screens and water flow provided by a computer controlled valve system. The fish are allowed to control these stimuli via activity in their motor-neurons recorded by a set of external suction electrodes. This arrangement makes it possible for immobilized fish to navigate a virtual environment entirely through power of activity in their motorneurons. As a first step we will combine this assay with 2-photon laser-scanning microscopy and allow transgenic fish lines that express genetically encoded calcium indicators in all neurons to navigate through virtual environments while neuronal activity is monitored. The small size and perfect translucence of the larval zebrafish provides in principle no barrier or restriction to these recordings and thus we have access to the whole brain at single cell resolution during these behavioral tasks. We can thus independently monitor the activities of hundreds or thousands of neurons as animals navigate their virtual worlds. As a subsequent step we will implement different learning assays that the animals have to execute in this virtual environment. In principle this will allow us to follow the flow of neural information in a single animal, before, during and after specific training sessions. The large volumes of resulting data will be analyzed and distilled with automated algorithms and the results used to sketch out potential models of how animals store information in neural networks in order to generate adapted behaviors. Subsequently, we can apply specific manipulations using genetically encoded optic tools for activating and silencing targeted subsets of neurons to test and develop these emerging theories. This platform will offer an unprecedented ability to track and manipulate the large- and smallscale activity patterns that underlie innate as well as learned behaviors.