Behavior Automation

Princeton University, Princeton NJ

Investigators

Linked publications, trials & patents

Abstract

Project Summary: Core 3, Behavior Automation    Working memory, the ability to temporarily hold multiple pieces of information in mind for  manipulation, is central to virtually all cognitive abilities. This multi-component research project  aims to comprehensively dissect the neural circuit mechanisms of this ability across multiple  brain areas. The individual parts of the project cohere conceptually, in part, because they all  involve rodents trained to perform a type of decision-making task that is based on the gradual  accumulation of sensory evidence and thus relies on working memory. To produce enough  subjects for these experiments, this Core will scale up an existing high-throughput rat training  facility run by technicians and adapt it for mice. This expansion will quintuple the project?s  capacity for rodent training. To do so, we will take advantage of the expertise of the project  leader in developing and managing such a training facility for sophisticated cognitive tasks in  an existing virtual reality infrastructure, software, and hardware. Once this facility is operational,  the Core will manage it and troubleshoot problems as needed. It will develop new hardware  and software components for training rigs to make technician interventions as reproducible and  error-free as possible. Because the most crucial and time-consuming aspect of mouse virtual  training is ensuring that the head-fixed animal is properly aligned to the ball and the projection  system, the Core will develop an automated alignment system based on image registration and  actuators to replace the current manual alignment. It will develop software tools and  standardized technician procedures to ensure consistency in rodent training, prevent training  errors, detect hardware failures, and monitor the health of the animals. This centralized facility  will promote rigor and reproducibility by reducing variability in animal training across labs,  increase the rate of data acquisition, and free personnel to focus on designing and carrying out  creative experiments. In the long run, the entire neuroscience community will benefit from this  effort, as the software and hardware tools and management protocols produced will be made  freely available, along with their documentation. These tools will enable other researchers to  introduce automated training for well-controlled cognitive tasks in their own laboratories,  leading to improved efficiency, rigor, and reproducibility in behavioral research across the field  of neuroscience.