Applied Psychology I

To enhance satisfaction and improve seating comfort, objective and accurate monitoring systems are necessary to evaluate and optimize seating solutions in various contexts such as aviation, automotive and office environments. Existing systems, like questionnaires, pressure maps or physiological measurements each lack in some of these requirements. We present a novel and objective measurement approach aiming for an easily accessible and contactless method to automatically detect and assess episodes of discomfort. To this end, we explored the feasibility of an automated video-based discomfort detection using state of the art computer vision methods. For acquiring learning-data, N = 30 participants attended for two 150-minutes session in a simulated flight cabin mockup (economy class seats) in one ‘high-‘ and one ‘low-discomfort’ condition. Data streams consisted of 4 synchronized camera streams, ECG sensor-data and acceleration-data from sensors placed on wrist and ankle. Additionally, discomfort was rated on Corlett and Bishop's body part discomfort scales (BPD) by the participants. For all datasets behavioral discomfort events (e.g. painful facial expression, head leaning, back rotation, hip and leg movements) were manually annotated by 10 raters. Convolutional neural networks were then trained on 2D–coordinates of estimated body keypoints to classify 20 distinct movement classes. Performance on the test-set revealed significant classification rates for identifying various discomfort related movements. For future applications, the development of 3D computer-vision models may help to further improve the usability in daily life conditions.

The limited research on follower preferences has shown that followers often prefer relationship-oriented (considerate) leaders to task-oriented (initiating structure) leaders. However, little is known about potential moderators of followers’ preferences. In this study, we investigated whether situational pressure moderates the preference of followers for a considerate over an initiating structure leadership style. In a laboratory experiment, we presented students (N = 319) with the situation of a fictive company, varying how high the pressure (high vs. low) in this situation was. Afterwards participants were shown the profiles of two department heads, including information about their leadership style. Based on these profiles, participants then chose in which of these (otherwise identical) departments they would rather want to work in the future. In addition, they also rated the qualification and their liking of the two department heads. Our results show that the department of the considerate leader was chosen significantly more often than the department of the initiating structure leader. More importantly, this effect was moderated by situational pressure, as only in the low pressure condition a significant preference for the considerate leader was found, while in the high-pressure situation the initiating structure leader was chosen as often as the consideration leader. In a currently running second study, we test bank employees instead of students, and the results, so far, confirm the findings from the student sample.

As digitalization increasingly generates challenges and opportunities for future leaders, organizations attempt to navigate these novel environments by leveraging promising technologies. One emerging technology is social robotics, in which robots with the capability of human-like interactions are used to augment day-to-day experiences. Prior studies have shown that humans i) enjoy being led by robots as long as it increases their efficiency and ii) are willing to be motivated by a robot if they perceive it as displaying authority. While first evidence suggests that robot leadership can be successful, specific leadership styles – fundament of leadership success in human leaders – have not yet been studied in robot leaders. To fill this empirical gap, we implemented three leadership styles in robot (Pepper by SoftBank Robotics) behavior (i.e., transformational leadership, transactional leadership, and minimal leadership as control) and tested how Executive MBA students (N = 67) reacted to those leaders in a between-subjects design. Preliminary results show that participants who engaged with the transformational robot perceive the robot to be significantly more competent (F(2,64) = 3.19, p < 0.05) and trustworthy (F(2,64) = 3.97, p < 0.05) than the other two groups. These results indicate that the perceptions evoked by transformational robots are similar to those of transformational human leaders. Upon completion of data collection, we will investigate how different leadership styles influence followers’ task engagement and performance. Our study is a first step in establishing whether evidence based on human leaders applies to robot leaders as well.