Planning with Others in Mind
Nastaran Arfaei, WeiJi Ma, NYU, United States
Session:
Posters 3B Poster
Presentation Time:
Sat, 26 Aug, 13:00 - 15:00 United Kingdom Time
Abstract:
People usually plan in environments inhabited by other planning agents, making the possibility of their future actions limited in predictable ways. Here we investigate whether people account for the possible future actions of their partners during a collaborative dyadic task with shared environment, costs and rewards. We find that almost all dyads perform better than the best performing agent who does not consider their collaborator, meaning that they predict and consider the future actions of their partner during their own planning. We investigate possible biases that lead to suboptimal performance. We find a bias toward scoring more equally and one for moving toward each other. This study provides a foundation for investigating the algorithms of complex planning in collaborative environments.