The eye is the mirror of technology acceptance: segmented capturing of trust in self-driving vehicles with eye-tracking

Abstract

THE AIMS OF THE PAPER

Based on the mainstream technology acceptance models (TAM, UTAUT), trust is a prominent factor influencing technology acceptance. Despite the fact that literature has identified trust as a key driver of technology acceptance in self-driving vehicles, few related empirical studies have been conducted. The available studies tend to use traditional survey methods to examine the population in general as a large homogeneous group, but there is little information on how trust is reflected in physiological responses, or whether any differences can be detected in segments with different characteristics. Hence, the main objective of our research is to investigate, using neuroscience, trust as a key determinant of technology adoption in relation to self-driving vehicles for different consumer segments.

METHODOLOGY

The study used real-time eye-tracking measurements on a 102-item sample, supplemented by a self-report survey. In a laboratory setting, we conducted measurements in which our participants watched videos recorded by cameras inside self-driving vehicles. We tracked the individuals' eye movements to measure the number and composite length of fixations and pupil movements. 

MOST IMPORTANT RESULTS

Based on the number of fixations, composite length, and pupil dilation, we empirically verified that the confidence level of the "traditionalist rejectionist" segment is the lowest, and based on gaze direction and number of fixations, that the confidence level of all five segments is relatively low. An increase in the level of trust can be detected when subjects are provided with additional information about the trip (tablet, repeated scenes). Furthermore, an important finding is that the level of trust based on self-report is not always reflected in the eye movement results.

RECOMMENDATIONS

Methodologies based on interviewing and measuring unconscious reactions complement each other well and their combined use allows us to get a more accurate picture of the level of trust in each segment.