Towards relating physiological signals to usability metrics: a case study with a web avatar

Inferring the user’s approval of a graphical interface with non-invasive devices can be effective in improving its design and in implementing adaptive pleasant interactions. This paper investigates how 3 common physiological signals, i.e. skin conductance, heart rate and respiration, can be exploited to infer users’ approval of an online avatar embedded in a health care Website. A between group experiment is performed with participants who have the avatar support and participants who do not. During the experiment, skin conductance, hearth rate and respiration were monitored, together with traditional usability metrics (visited pages, completion times, errors, etc). At the end of each experiment, a feedback questionnaire is proposed to infer information related to the user experience, ease of use and approval. Results indicate that the respiration overshoot rate is closely related to the users’ appreciation of the avatar based interaction. Further steps of our research will consider improvements in the results by investigating and exploiting mutual effects induced by the multiple collected signals.