Emotions and facial expressions: theoretical, operational, methodological, and statistical considerations for modern open research

Emotion research has observed a resurgence in critical debates surrounding its core tenants and framework of operation. These debates have been fuelled by new methodological advances, such as the widespread adoption of mixed effects models, Bayesian statistics, and dynamic stimuli. Importantly, more recent findings call into question elements of the prevalent encoding-decoding approach to emotion research, finding more compatibility with the affective signalling framework. This requires us to carefully consider how we fundamentally reason about emotions, facial expressions/displays, and emotion recognition (human and automated). In this workshop, the elements of these debates will be discussed and exemplified. First, it will discuss how the emotion framework adopted by a researcher impacts the hypotheses that they can and cannot investigate. Second, the often-overlooked importance of operational definitions and operationalization in individual studies will be addressed, focusing on inference and generalizability. Third, considerations will be given to the emotion stimuli we employ and the open research movement. And finally, the workshop will make statistical analysis recommendations (frequentist and Bayesian), with practical examples, to aid researchers in getting the most out of their data.