Cooperation with autonomous machines through culture and emotion

Abstract

As machines that act autonomously on behalf of others-e.g., robots-become integral to society, it is critical we understand the impact on human decision-making. Here we show that people readily engage in social categorization distinguishing humans ("us") from machines ("them"), which leads to reduced cooperation with machines. However, we show that a simple cultural cue-the ethnicity of the machine's virtual face-mitigated this bias for participants from two distinct cultures (Japan and United States). We further show that situational cues of affiliative intent-namely, expressions of emotion-overrode expectations of coalition alliances from social categories: When machines were from a different culture, participants showed the usual bias when competitive emotion was shown (e.g., joy following exploitation); in contrast, participants cooperated just as much with humans as machines that expressed cooperative emotion (e.g., joy following cooperation). These findings reveal a path for increasing cooperation in society through autonomous machines.

The interplay of emotion expressions and strategy in promoting cooperation in the iterated prisoner’s dilemma

“…Given that this nonverbal signal is pervasive in nature [29][30][31] , it is important to shed light on how strategy and emotion expressions interact with each other to promote cooperation, as we do here. Consistent with research indicating that emotion expressions serve important social functions [25][26][27][28] and influence others' decision making [12][13][14] , our results report a moderating effect of emotions on a zero-determinant generous strategy, similarly to what had been shown for tit-for-tat 12,19 . In contrast, with a zero-determinant extortion strategy, emotion signals had no effect, thus, revealing a limitation for extortionists; in this case, given the highly competitive nature of the strategy, participants appear to be reluctant to believe the emotional expressions of extortionists were genuine.…”

“…To estimate sample size, we followed the power calculations proposed by Cohen 44 and implemented in G*Power 45 -a software that is often used by behavioral researchers. Based on earlier work 12,19 , we predicted a small to medium effect size (Cohen's f = 0.20). Thus, for α = 0.05 and statistical power of 0.95, the recommended total sample size was 327 participants.…”

“…These findings provide insight into the importance, relative influence, as well as the limits, of behavioral strategies and emotion signals for emergence of cooperation. The results also have important practical applications for the design of increasingly pervasive autonomous machines-such as robots, self-driving cars, drones, and personal assistants-which will inevitably rely on cooperation with humans for their success [15][16][17][18][19] .…”

“…The effects of emotion expressions in extortion and generous strategies, however, have not been studied so far. When engaging with counterparts that follow a tit-for-tat strategy-i.e., only cooperate if the other cooperated in the previous round-de Melo and Terada 19 showed that participants cooperated more or less according to whether the emotion expressions signaled a cooperative (e.g., joy following mutual cooperation) or competitive intention (e.g., joy following exploitation). Tit-for-tat is an interesting strategy as it strikes a balance between rewarding cooperation by the other player and punishing if the other player defects 4,5 .…”

“…Participants were instructed they would engage in the task with other participants but, to increase experimental control and implement these strategies precisely, participants engaged with a computer script. Similar methods have been used in previous research 15,19 , all experimental procedures were approved by the Gifu University IRB, and participants were fully debriefed at the end.…”