In order to tackle virtual worlds’ increasing complexity, it becomes increasingly important to develop agents that are socially aware and able to react to their social environment. Because the presence of others and group memberships can have influence on how someone thinks, feels and behaves, it is important that agents be able to incorporate this influence in their decisions.

With that in mind, project INVITE’s aim is threefold:


Within this project the tool developed is a multi-player 3D video-game platform that allows the investigation of how factors such as social identity, discontinuity effect, group size and decision framing can influence the behavior of individuals in different scenarios. This research tool is fully parameterizable, providing the creation of several different scenarios and case studies, from the classical prisoner’s dilemma to more complex team games, such as two-level public good games, where conflict can be present at both in-group and out-group level.

This game places the players (who can be humans or autonomous agents), after a plane crash, on a deserted island which they must evade before an active volcano begins to erupt. Players are assigned into camp sites, with each one having the collective goal of constructing a raft in order for its members to survive. To build the raft, players should each turn (day) collect wood for their camp site. If the team of a camp site is able to build a raft before the volcano’s eruption, its members can use the raft to escape. However, although survival is the main purpose of the team, gold can also be found scattered all over the island and, in the end, the player that survives with more gold wins the game. Players are then faced with the dilemma of either collecting wood to escape (team’s interest) or gathering gold (free-riding) to become rich and thus win the game if they escape (personal interest).

Since more similarities with a group leads to more cooperation within that group due to social identification process, agents in this framework were developed to take these influences into consideration.  Strategically, agents and players have to try to maximize their personal gains to win the game, but the process of social identification could lead to a bias in their decisions. In situations where team members share more characteristics, this bias should result in replacing their own personal interests by the group’s interests. This is an example of one of the hypotheses being currently tested with controlled experiments using this framework.

Developer: INVITE Project

Genre: Strategy

Web page:

Platform: Windows / OSX

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.