Organizations as Decision Boundaries: How Structure Can Compensate for Flawed Mental Representations

Abstract

Because aggregation structures and mental representations have opposing effects on fallibility—mental representations are a source of errors while aggregation structures aim to overcome these—aggregation structures can be used to compensate for flawed mental representations. Yet given the multiple structures from which to choose and given that their effects depend on environmental factors, it is not clear which structure is best suited to what circumstances. To answer that question, this paper develops a formal model of group decision making among individuals who base their decisions on flawed mental representations. The model predicts the performance of three different aggregation structures (delegation, unanimity, and averaging) under different environments (defined by their munificence, uncertainty, complexity, and attribute dominance). We show that the concept of “decision boundary,” an idea we borrow from the machine learning literature, explains when and how aggregation structures compensate for flawed representations. This allows us to characterize the conditions under which it is preferable to use different aggregation structures as well as situations where all aggregation structures perform poorly. More generally, our paper provides a theoretical framework to understand how aggregation structure, mental representations, and the task environment jointly determine organizational performance.