Psychophysiological Modeling of Trust In Technology

Abstract

Trust as a precursor for users' acceptance of artificial intelligence (AI) technologies that operate as a conceptual extension of humans (e.g., autonomous vehicles (AVs)) is highly influenced by users' risk perception amongst other factors. Prior studies that investigated the interplay between risk and trust perception recommended the development of real-time tools for monitoring cognitive states (e.g., trust). The primary objective of this study was to investigate a feature selection method that yields feature sets that can help develop a highly optimized and stable ensemble trust classifier model. The secondary objective of this study was to investigate how varying levels of risk perception influence users' trust and overall reliance on technology. A within-subject four-condition experiment was implemented with an AV driving game. This experiment involved 25 participants, and their electroencephalogram, electrodermal activity, and facial electromyogram psychophysiological signals were acquired. We applied wrapper, filter, and hybrid feature selection methods on the 82 features extracted from the psychophysiological signals. We trained and tested five voting-based ensemble trust classifier models using training and testing datasets containing only the features identified by the feature selection methods. The results indicate the superiority of the hybrid feature selection method over other methods in terms of model performance. In addition, the self-reported trust measurement and overall reliance of participants on the technology (AV) measured with joystick movements throughout the game reveals that a reduction in risk results in an increase in trust and overall reliance on technology.

Exploring Central-Peripheral Nervous System Interaction Through Multimodal Biosignals: A Systematic Review

“…Some papers have evaluated their subjects by Annett Handedness Scale [43], mini-mental state evaluation, state-trait anxiety, and Hamilton rating scale [47], Epworth sleepiness scale [49]. In a few of the cases, right-handed people are preferred [18], [21], [22], [25], [39]- [41], [44], [45], [50], [51], [53]- [57]. Subject's consent is also taken in some literature [21], [22], [24], [26]- [35], [37]- [43], [45]- [47], [52], [54], [56], [58]- [62].…”

“…Some of the most used stimuli are musical videos [18], [48], [63]- [65], visual stimuli [23], [40], [41], [45], [53], [54], [67], images [31], [33], [36], [46], [50], [51], [68], [69], Audio [29], [34], [37], [52], task based stimuli [24], [25], [27], [38], [39], [42]- [44], [48], [55]- [57], [58], [62], [70]- [75], film clips [76], [77], normal video clips [26], [62], [78]. In task-based stimuli either the subjects are instructed to do the mental task (mathematics-related problems, memorizing, computer-based gaming, and reading) or physical task (coldpressor test, rope skipping, surgical task, and fatigue exercise).…”

“…Both single and multichannel bio-signals are extracted at different sampling rates at a range of 30-2048 Hz. In some cases those signals are downsampled to reduce processing time [25], [26], [30], [39], [43], [53], [54], [57], [64], [73], [78]. Among database-based emotion-related articles, five papers have used a dimensional…”

“…Physical stress is induced by cold pressor task [70], [71], rope skipping [25], fatiguing exercise [22], [56], handgrip task [44], [75]. Cognition is evaluated using go/no go task [45], surgical task [74], audio stimuli [29], images [67], [69], visual stimuli [41], film clips [77], reading comic strips [73], audio visual stimuli [43], gamming [57], and deceptive task [28]. Sleep based task is also done in some cases namely [21], [27], [30], [34], [35], [49], [59], [60], [66].…”

“…For EEG, the following time domain features are extracted namely N550 latency and amplitude [34], mean [18], [50], [57], [69], maximum [18], minimum [18], standard deviation [50], [57], [76], skewness [50], [76], kurtosis [50], [76], variance [57], [69], peak-to-peak amplitude [57], [69], mean of the absolute values of the first difference of raw signals [18], [50], and mean of the absolute values of the first difference of normalized EEG signal [50]. Some of them also have used power spectra of one second EEG epoch [55], approximate entropy [46], corrected conditional entropy (CCE) [38], duration of overall NREM [21], number of Slow oscillation events [21], global power [36], the correlation coefficient between two channels [69], root mean square value and energy of EEG signal [69], median frequency (MF) and mean power frequency (MPF) of resting state EEG.…”