Thesis title: Neurophysiological assessment of user’s experience for an advanced and transversal user-centered approach in daily life applications.
RATIONALE. Over the past few decades, the concept of User-Centered Design (UCD) has gained increasing significance in daily life applications. The field of industrial neurosciences has emerged to support this trend, focusing on understanding the human mental state during interactions with the external environment and tools. Numerous studies in literature have demonstrated the benefits of incorporating neuroscience into various industrial contexts, employing different neuroscientific metrics to analyze various aspects of the human mental state. However, the methods for characterizing users' mental state are poorly standardized, highly varied, and strictly related to the application domains.
GENERAL OBJECTIVES. The activity of the present Ph.D. thesis aimed to address this gap by investigating and establishing standards for four key neuroscientific metrics derived from electrophysiological signals (e.g., EEG) and signals related to the autonomic nervous system (ANS): heart rate (HR) and electrodermal activity (EDA). These metrics consist of mental effort, attention, pleasantness, and emotion (EI).
EXPERIMENTAL DESIGN AND METHODS. The first part of the research activity focused on methodological research. In this phase, weaknesses in the EEG processing chain were investigated, introducing methods to make the signal more reliable, such as the envelope and the standardization of the global field power (GFP) and a threshold method for cleaning index outliers. In addition, the modification of indices upon reduction of the number of electrodes (system invasiveness is a crucial topic when bringing these tools in daily applications) and upon reduction of sample size, was explored. Finally, the research focused on solving some limitations related to the emotion assessment, by proposing a new method for calculating the EI, validated on a controlled dataset. In a second part of the Ph.D. activity, three case studies were conducted to test the indices in three different contexts of the industrial neurosciences: 1) neuromarketing, evaluating the response to different odors exposition at short and long term, 2) automotive, investigating the mental state of users while performing simulated driving tasks in urban and highway, and 3) neuroaesthetics, estimating the human experience while observing an Etruscan artifact exposed in a museum or reproduced in Virtual Reality.
RESULTS. Results on methodological research showed that the proposed methods improved the features of the EEG signals while maintaining the information conveyed by the indicators as demonstrated by a significant correlation of the indices (p < 0.05). Both the workload and the attention remained consistent at the reduction of the number of electrodes with a significant correlation between indices computed with the different configuration, while the pleasantness presented the higher distortion. The index reliability showed to be strictly dependent from the considered sample size, pointing out a threshold of subjects for which almost all the comparisons were significant, the outcomes remained comparable (p < 8×10^−5), the data dispersion and the committed error were acceptable (mean squared error < 0.1). The ANOVA performed over the new EI values computed for videos with different valence, showed a significant effect of valence (p < 0.001) with EI significantly higher for positive emotions and significantly lower for negative emotions. Three case studies demonstrated the sensitivity of the indices in three different contexts of application. The application in neuromarketing allowed to discriminate the odour conditions at the different time scale. Results in the field of automotive characterized the response of the indices in different road contexts and secondary tasks. The study in neuroaesthetics pointed out an employment of different mental resources during the observation task performed in the museum compared to the VR.
CONCLUSIONS. The methods for the signal processing proposed in this thesis allowed to improve the reliability of the four indices for monitoring the user’s experience in daily life applications, as well as to identify some guidelines for standardizing methodological procedures such as the sample sizing, the choice of electrodes and the signal processing steps. The indices demonstrated to be sensitive in characterizing the human neurophysiological response in all the contexts in which they have been applied: neuromarketing, automotive and neuroaesthetics. These promising results are the first demonstration of how reliable and transversal the proposed indices can be, encouraging their use in the UCD approach, independently from the context of application.