Engagement in Code Review: Emotional, Behavioral, and Cognitive Dimensions in Peer vs. LLM Interactions

Code review is a socio-technical practice, yet how software engineers engage in Large Language Model (LLM)-assisted code reviews compared to human peer-led reviews is less understood, especially as artificial intelligence (AI) tools are increasingly integrated into software engineering (SE) workflows. We report a two-phase qualitative study with 20 software engineers to understand such dynamics. In Phase I, participants exchanged peer reviews and were interviewed about their affective responses and engagement decisions. We also prompted them to discuss their submitted code review generated by ChatGPT 4o. In Phase II, we investigated and introduced a new prompt to match engineers’ stated preferences for the review and probed how content characteristics shaped their reactions. We develop an integrative account linking emotional self-regulation to behavioral engagement and likelihood of a resolution. We identify a repertoire of participant-reported self-regulation strategies that engineers use to regulate their emotions in response to negative feedback: reframing, dialogic regulation, avoidance, and defensiveness. These strategies are inspired by the engineers’ commitment to code quality and values: accountability and growth mindset. Engagement proceeds through social calibration; engineers seem to align their responses and behaviors to the relational climate and team norms. Trajectories to likelihood of a resolution, in the case of peer-led review, vary by locus (solo/dyad/team) and an internal sense-making process. With the LLM-assisted review, emotional costs and the need for self-regulation seem lower, as reported by participants in our sample. When LLM feedback appear to better align with engineers’ cognitive expectations (e.g., clear structure, concise scope, neutral tone, actionable), participants reported reduced processing effort and a potentially higher tendency to adopt. As reported by our participants, our findings demonstrate that LLM-assisted review redirects engagement from managing emotions and social affect to managing cognitive load. We contribute with an integrative model of engagement, linking emotional self-regulation \(\leftrightarrow\) behavioral engagement \(\rightarrow\) likelihood to a resolution, showing how affective and cognitive processes may influence feedback adoption in peer-led and LLM-assisted code reviews. We conclude that AI is best positioned as a supportive partner to potentially reduce cognitive and emotional load while preserving human accountability and the social meaning of peer review and similar socio-technical activities.