Adversarial Collaboration: An Overview of Social Quantum Logic

Collaborative Learning Agents

Volume

Papers from the 2002 AAAI Spring Symposium

Contents

Collaborative decision-making (CDM) to solve problems is an aspect of human behavior least yielding to rational theory. To simplify, early game theorists assumed that logical conceptions of cooperation and conflict in static configurations could represent the actual choices made by humans in an interaction, leading to the first stable solution of mutual competition (Nash equilibrium). Later, second stable solution of mutual cooperation was found by Axelrod to evolve in extensive form. But unary maps underdetermine reality, R; cooperation in the field to solve ill-defined problems produces suboptimal solutions; and a rigorous logical map from multiple individual preferences to a single group preference is not possible. More problematic for multiple agent systems or computational autonomy, as information (/) completeness produces knowledge (K), as the number of interactants approach an N 100 or more, or as agents depart from cooperation, computability decreases significantly. In contrast, adapting quantum logic to the interaction while difficult to understand produces a robust model of decision-making even as N increases. The evidence suggests that adversarial collaboration is superior.