Analysing facial expressions of basic emotions with FaceReader: An experimental study [Review]

The aim of the study was to examine, using FaceReader, participants' facial expressions in response to the presentation of images extracted from the IAPS database, by manipulating the levels of valence and arousal in order to study seven basic emotions. The results showed that FaceReader detected a higher proportion of neutral facial expressions than basic emotions, and that the basic emotions detected showed an indistinguishable pattern in the four conditions resulting from manipulating the values of valence (positive and negative) and arousal (high and low). This leads to the conclusion that FaceReader may not be the most suitable tool for recording and distinguishing among emotional experiences. The fact that the results of the work are null is not a problem, as it may be useful for future studies. The main objection lies in the assertion made at several points in the manuscript that "no previous studies have been published to date that analyse the facial expressions detected by FaceReader as a result of the changes in subjective experience evoked by watching IAPS". And then, in the discussion "it is surprising that previous research have not yet combined them in an experimental study design". Indeed, the combination of IAPS and FaceReader deserves to be studied regardless of the results, but unfortunately this is an issue for which there is already previous literature. Specifically, at https://doi.org/10.3389/fpsyg.2020.01388, in 2020, the authors carry out a very similar experimental manipulation. FaceReader is tested with IAPS images by manipulating their valence and arousal values, and also comparing FaceReader output with Electromyography and Skin Conductance data. The results are favorable to FaceReader in some respects and null in others. And, although in a conference paper from 2009, we can also cite here another work (somewhat more limited than the previous one): “Benţa, Kuderna-Iulian & Kuilenburg, Hans & Xolocotzin Eligio, Ulises & Uyl, M. & Cremene, Marcel & Hoszu, Amalia & Creţ, Octavian. (2009). Evaluation of a System for RealTime Valence Assessment of Spontaneous Facial Expressions”, with very similar experimental manipulations and results also limited in scope. Additionally, it should be noted that in both previous studies the presentation of the images was randomised in terms of the emotional category to which they belonged in order to avoid possible habituation effects, so it would be necessary to explore in detail the impact that the block presentation of the present study may have had on the low proportion of emotions detected.