Literature Reading and Discussion: The Impact of Gaze in Human-Computer Interaction on Human Decision-Making Processes and Game Strategies

Literature Review

In a paper published in 2021 in Scientific Robotics by Dr. Marwen Belkaid and other scholars, it was highlighted that gaze contact plays a crucial role not only in interpersonal interactions but also in interactions between humans and machines.

In their experiments, human participants interacted with humanoid robots through a simple game (Chicken Game). In different experimental conditions, the humanoid robot either engaged in direct gaze with humans or avoided eye contact. The experimental data indicated that the gaze contact or avoidance by the robot did not significantly affect the outcomes of the game for humans. However, it had a notable impact on human decision-making processes.

The influence of robot gaze on human decision-making processes manifested in two aspects: cognitive processes and game strategies. Compared to gaze avoidance, when the robot engaged in direct gaze with humans, it increased the amount of evidence accumulation required by humans for decision-making (decision boundary) and prolonged the non-decision time in the human decision-making process. This suggests that these parameter changes may reflect psychological processes related to interpreting the robot’s intentions or suppressing attentional interference caused by the robot’s gaze.

Regarding game strategies, in the gaze avoidance condition, humans tended to adopt a self-centered strategy, such as sticking with the previous strategy if they won and switching strategies if they lost (win-stay-lose-shift). However, in the case of gaze contact, the frequency of using self-centered strategies decreased, giving way to other-oriented strategies, such as tit-for-tat or mimicking the opponent’s strategy (stay-shift-imitation).

Furthermore, even though the experiment used a simple level 0 strategy (win-stay-lose-shift) as the preset for the robot’s game strategy, most humans believed that the robot was using a level 2 strategy (assuming that the robot followed a level 0 strategy when humans made decisions). In terms of research methodology, this study employed the Drift-Diffusion Model and electroencephalography (EEG) to mutually support the data analysis results, demonstrating the credibility and reliability of the findings.

Extended Discussion

This research holds practical significance for the development of human-machine interaction. Even when interacting with robots, humans exhibit similar reactions as they do in human-human interactions. This behavioral pattern will impact our thinking assumptions about the application scenarios of human-machine interaction and their corresponding effects.

Application Scenarios:

Using robotic teachers in digital education with gaze interaction with human students.
Employing robot shop assistants in virtual shopping experiences with gaze interaction with human customers.
In high-stakes communication situations where robots are involved, such as negotiations at crime scenes, communication with individuals awaiting rescue at disaster sites, or providing guidance to non-professionals during emergency medical situations.

Research Extensions:

The main conclusion of the above study is that robot gaze affects human decision-making processes. For practical applications, a direct question arises: Is this influence long-lasting? In long-term interactions, will the emotional impulses triggered by gaze and the behavioral paradigm of strategy shifting change due to learning? Therefore, exploring long-term effects is necessary.

In the experiments, participants were required to engage in gaze with the robot. If this condition were relaxed, meaning that in practical scenarios, gaze could be actively avoided, how would human-robot interactions differ from the current setup? Under what circumstances would humans choose to actively avoid or engage in gaze with the robot?

Since robots can be programmed, after understanding the behavioral patterns of humans in gaze interactions, would the robot’s strategic use of gaze affect human performance in games? Is the long-term impact of this strategic gaze positive or potentially detrimental? Throughout the study of human-machine interactions, vigilance against strong interference effects should always be maintained.

Reference:

Belkaid, M., Kompatsiari, K., De Tommaso, D., Zablith, I. and Wykowska, A., 2021. Mutual gaze with a robot affects human neural activity and delays decision-making processes. Science Robotics, 6(58), p.eabc5044.