CRII: SHF: Building Visibility into the Cognitive Processes of Software Engineers via Biosensors

点击翻译按钮获取中文摘要

The goal of this research project is to study cognitive effort associated with tasks that computer programmers perform. This is important because the limits of cognitive load present a barrier to improving a programmer's productivity and quality. Using newly available biometric technologies such as fMRI, EMG and eye-tracking, it is possible to conduct studies of cognitive processes in programmers. Such studies make it possible to provide scientific answers to questions such as which programming language constructs are easier to use, what kinds of training produce better programmers, or how different programming environments and situational variables are more conducive to writing the best code. The main thrust of this project is to overcome the significant challenges that currently prevent researchers in the software engineering community from widely adopting these methods. This project performs some of the groundbreaking work that is needed to provide robust measurements of cognitive processes in programmers using the biometric technologies. The resulting tailored methodologies and measurement devices will support a community of researchers working in this space. The project will develop techniques to use fMRI to measure difficulty in understanding certain programming constructs. This will be done by measuring cognitive effort, as measured by neural efficiency, and level of concentration via neural deactivation. Similarly, measurements of cognitive processes activated during other code comprehension tasks will be obtained. The objective is to obtain a validated set of measures that predict the cognitive load associated with understanding code, which is necessary to create the scientific basis for the studies. The work will lead to combining different measurements, such as eye-tracking and EMG, to establish more sensitive measurement tools. The project will address technical concerns such as maintaining clean signal data from sensors and learning how to calibrate devices to maintain validity of results, as well as practical concerns such as how to perform measurements on human subjects without interfering with the measurements. Ultimately, such studies should lead to a better understanding what constitutes best practices in software development and training of software engineers.