We Must RE-MEMBER to RE-FORMULATE: The M System

The goal of this work is to develop a theory of mind that enables common sense reasoning to be applied in M. M was designed to observe situations and formulate beliefs about these situations regardless of the truth of the beliefs. It appears that humans observe and believe, and then over time continue to improve their knowledge of their beliefs while many types of computer programs just get stuck! I take the position that common sense learning is a time variant problem and that learning is a constant series of viewing similar situations from different points of view over time. You can not learn something until you learn about “it” from several points of view. Good common sense reasoning and learning results from the ability to perform reformulation on an idea, concept, or problem. Reformulation requires rich fluid representations, multiple modalities of reasoning, and a range of experiences over time. Minsky’s Society of Mind (SOM) Theory (Minsky 1985) is the essence of the study and implementation of M. M was initially designed and implemented in my work at AT&T Bell Laboratories. M functioned as a software process that recognized and classified objects and actions performed by humans in a multimedia desktop conferencing system we developed at Bell Labs, called the Virtual Meeting Service. In this Computer Supported Cooperative Work (CSCW) service, participants worked in a Virtual Meeting Room (VMR) on a series of tasks. Each participant has a personal M program that watches all legal actions performed on all legal objects in the VMR by all participants. Each personal M attempts to recognize and classify what all the participants are doing and then support its respective user to recall items and actions that might relate to the user’s current task and context.