The Emergence of Complex Exchange Networks

The celebrated îsmall-worldî and îscale-freeî network structures have recently been promi- nently proposed as models of large networks in a variety of substantive realms. However, it remains unclear to what degree the mechanisms of îpreferential attachmentî and îrandom rewiringî that have been argued to underlie the emergence of these networks are sociologically plausible. First, these mechanisms require the sociologically implausible assumption of globally available knowl- edge about structural positions. Second, individual goals and cognitions that motivate actors' deci- sions to make or break ties remain implicit. In this paper we present a computational model of the emergence of complex networks (such as small-world or power-law) that is grounded in social exchange theory and assumptions of bounded rationality. More specically , we draw on previous work by Flache and Hegselmann and assume that agents seek to nd and keep attractive exchange partners in a population where actors differ in attractiveness, for example in their expertise in knowledge exchange. Agents pursue their goal under imperfect, local information without initial knowledge about others or the global network structure. Moreover, agents in our model are strongly adaptive, i.e. knowledge is only acquired in the course of interaction and simple search heuristics are applied. Our results suggest that under a large range of conditions a segregated network arises in which ex- change partners are highly similar in attractiveness. Moreover, we nd that the topological structure of the emergent social network depends heavily upon the harshness of the exchange game (i.e. the potential damage agents can suffer from exploitation and the restrictiveness of access to potential exchange partners) and the learning capacities of the agents (memory size, exploration probability, knowledge exchange mechanism). In particular, we show that it depends on the combination of these conditions whether star-like, small-world or scale-free structures emerge.