Understanding understanding source code with functional magnetic resonance imaging

Abstract

Program comprehension is an important cognitive process that inherently eludes direct measurement. Thus, researchers are struggling with providing suitable programming languages, tools, or coding conventions to support developers in their everyday work. In this paper, we explore whether functional magnetic resonance imaging (fMRI), which is well established in cognitive neuroscience, is feasible to more directly measure program comprehension. In a controlled experiment, we observed 17 participants inside an fMRI scanner while they were comprehending short source-code snippets, which we contrasted with locating syntax errors. We found a clear, distinct activation pattern of five brain regions, which are related to working memory, attention, and language processing-all processes that fit well to our understanding of program comprehension. Our results encourage us and, hopefully, other researchers to use fMRI in future studies to measure program comprehension and, in the long run, answer questions, such as: Can we predict whether someone will be an excellent programmer? How effective are new languages and tools for program understanding? How should we train developers?

Leveraging Biometric Data to Boost Software Developer Productivity

“…One of the early approaches in the software development domain mentioning biometric sensor technology is the Ginger2 environment by Torii and colleagues that included an eye-tracker and a skin resistance level sensor to empirically study developers [60]. In the context of software development, biometric sensors have been and are being used predominantly to gain a better understanding of developers' program comprehension either using eye-tracking [61]- [64] or by measuring brain activity [65], [66]. Recently, especially with the advances in sensor technology and the availability of more affordable biometric sensors, a few software engineering researchers have also looked at other aspects, for instance, using measures of cerebral blood flow, measures of sub-vocal utterances captured with electromyography (EMG), or EDA, eye-tracking and brain activity measures to asses task difficulty of small code snippets or programming tasks [67]- [69].…”

The last line effect explained

“…Given Example 2, any of the two statements could be counted as the one containing the duplication. However, reading and writing source code typically happens from top to bottom and from left to right (Siegmund et al 2014). Therefore, the only natural assessment is to flag lines and statements as problematic according to this strict left-right and top-down visual reading order: In Example 2, only when we have read the second statement do we know it is a duplicate of the first.…”

“…Our a-posteriori repository mining approach assumes a top-to-bottom reading order of blocks and a left-toright reading order for individual lines. We know that developers "jump" in the code when reading a file, only focusing on what seems important to solve the task at hand (Busjahn et al 2015;Siegmund et al 2014). However, in order to understand small coherent logical units, such as micro-clones, developers must necessarily read in the control-flow-direction of the software -which is top-to-bottom, left-to-right.…”

The Last Line Effect

“…For replication purposes, a free trial of PVS-Studio is publicly available. 6 2) Inspect the corpus of warnings from PVS-Studio and remove false-positives and warnings not related to microcloning. 3) For each faulty micro-clone, count the total number of lines (RQ1) or statements (RQ2) and denote which lines or statements are faulty.…”

“…Given Example 2, any of the two statements could be counted as the one containing the duplication. However, reading and writing source code typically happens from top to bottom and from left to right [6]. Therefore, the only natural assessment is to count the line as problematic according to this strict left-right/top-down-reading order: In Example 2, only when we read the second statement do we know it is a duplicate of the first.…”