Plaza, Arcos and Martin[Plaza, Arcos, &Mart´n1996] discuss two modes of cooperation among case-based reasoning (CBR) agents where an agent can leverage the learning capabilities or past experience of peer agents to achieve a task or solve a problem. These modes are developed within the Federated Peer Learning (FPL) framework [Plaza, Arcos, &Mart´n1996] that aims to study cooperative problem solving among agents possessing either same or different capabilities and incorporating potentially different knowledge and problem solving behaviors based on their individual learning and experience. Cooperative problem solving in such a system can result in bringing wide range of experience to bear on a task at hand in an agent. The approach taken here to achieve cooperation is through communication using the Noos representation language developed at IIIA for integrating learning and problem solving[Arcos &Plaza1996]. Plural Noos is an extension of Noos that allows communication and mobile (or ``migrating'') tasks and methods (to achieve these tasks) among agents that use Noos as a representation language. In particular, we will show two modes of cooperation among CBR agents: Distributed Case-based Reasoning (DistCBR) and Collective Case-based Reasoning (ColCBR). Intuitively, in DistCBR cooperation mode an agent delegates its authority to another peer agent to solve a problem - for instance when is unable to solve it adequately. In contrast, ColCBR cooperation mode maintains the authority of the originating agent: an agent can transmit a mobile method to another agent to be executed there. That is to say, uses the experience accumulated by other peer agents while maintaining the control on how the problem is solved.
Each of the cooperating agents in DistCBR and ColCBR is capable of solving the overall task by itself (most of time) unlike in Negotiated Retrieval where agents are specialists at specific subtasks.