Recognizing the New : A Multi-Agent Model of Analogy in Strategic Decision-Making

In novel environments, strategic decision-making is often premised on analogy, and recognition lies at its heart. Recognition refers to a class of cognitive processes through which a problem is interpreted associatively in terms of something that has been experienced in the past. Despite recognition's centrality to strategic choice, we have limited knowledge of its nature and its influence on strategic decision-making in individuals, much less in the multi-agent settings in which these decisions typically occur. In this paper, we develop a model that extends neural nets techniques to capture recognition processes in groups of decision-makers. We use the model to derive some fundamental properties of collective recognition. These properties help us understand how the intensity of communication among group-members and some select structural characteristics of the group affect recognition outcomes in novel and structurally ambiguous worlds. In particular, we demonstrate that communication pressure can lead agents to converge to shared interpretations or recognitions that are new to each of them, thereby helping them recognize problems that are genuinely new. We also show that when communication is too intense, its beneficial aspects give way to the pathologies of "groupthink." We conclude by discussing how our results are relevant to strategic choice, as well as how our model complements both other theories of choice that view the role of experience as central and recent work in population ecology that emphasizes cognitive processes