A Cross-Cultural Study of Playing Simple Economic Games Online with Humans and Virtual Humans

We compare the simple online economic interactions between a human and a multimodal communication agent (virtual human) to the findings of similar simple interactions with other humans and those that were run in the laboratory. We developed protocols and dialogue capabilities to support the multi modal agent in playing two well-studied economic games (Ultimatum Game, Dictator Game). We analyze the interactions based on the outcome and selfreported values of possible factors involved in the decision making. We compare these parameters across two games, and the two cultures of US and India. Our results show that humans’ interaction with a virtual human is similar to when they are playing with another human and the majority of the people choose to allocate about half of the stakes to the virtual human, just as they would with another human. There are, however, some significant differences between offer distributions and value reports for different conditions (game, opponent, and culture of participant).