Limitations of emotion recognition in software user experience evaluation context

This paper concerns how an affective-behavioural-cognitive approach applies to the evaluation of the software user experience. Although it may seem that affect recognition solutions are accurate in determining the user experience, there are several challenges in practice. This paper aims to explore the limitations of the automatic affect recognition applied in the usability context as well as to propose a set of criteria to select input channels for affect recognition. The results are revealed via a semi-experiment based on the case study of an educational game. As a result, a number of concerns were identified, providing a list of pros and cons for affective computing methods applied in the usability testing context. The lessons learned might be interesting for both researchers that develop emotion recognition algorithms and for practitioners, who apply them to diverse areas.