What if? Dealing with uncertainity in Repage’s mental landscape

This paper is focused on cognitive decision-making about how to solve inconsistencies and incompleteness in social evaluations (e.g. about potential partners in exchange). We propose a development of Repage, a computational system for forming and updating social evaluations. The system draws upon a fundamental difference between reputation and image as a way out from the trade-off between agents' autonomy and their need to adapt to social environment. A full exploitation of its potentialities includes the activation of a special module, the analyzer, aimed at solving possible inconsistencies, uncertainties and incompleteness in the output of lower level modules by means of inner simulation. In this work, Repage's analyzer architecture is described; some representative examples of problems posed by the planner to the analyzer is discussed and hypothetical simulations are run within this module to find a solution to uncertainty, avoiding at the same time the exceeding complexities of rule-based reasoning and the costs of reinforcement learning.