Organizationally Adept Agents

An organizationally adept agent (OAA) is not only aware that it is part of an agent organization and about its role(s) in that organization, but it can also assess how well it is fulfilling its organizational responsibilities and can proactively adapt its behaviors to meet organizational needs better. OAAs evaluate their behaviors based not only on their (agent-centric) self-interests, but also on their (organizationcentric) responsibilities to each other and their (social-centric) willingness to perform activities requested by other agents. Agent organizations designed for less adept agents must specify detailed guidelines that, when blindly followed, will influence individual agents to work together in the expected environment. However, if the environment deviates from expectations, such detailed organizational guidelines can mislead agents into counterproductive or even catastrophic behaviors. Organizations designed for OAAs, on the other hand, can assume that agents will reason about organizational expectations, and will adjust their behaviors when the nominal guidelines misalign with those expectations. We describe an extended BDI reasoning architecture for an OAA that balances organizational, social, and agent-centric interests and can adjust this balance when appropriate. Then we show how we used the reasoning architecture with agents operating in RoboCup Rescue scenarios.