ABM simulations generate novel hypotheses testable through later lab experiments

Evidence

Yet another important feature of the agent-based modeling approach is its usefulness for theory development. Decades of behavioral laboratory research form the basis for the rules governing the individuallevel behavior of agents in the model. However, ABM simulations frequently produce novel and unexpected outcomes that could not have been predicted from individual-level research. These outcomes can then form the basis for further traditional behavioral research, suggesting novel hypotheses to be tested at the individual level. For example, Kalick and Hamilton’s (1986) model of mate selection generated the novel prediction that mate pairs formed later in time will be less attractive than those that formed earlier. This, of course, can be directly tested with human participants in a laboratory. Similarly, it is likely that agent-based models that are developed to study group formation and change will generate new hypotheses about how different individual-level motives interact under different conditions. These hypotheses can then be tested in the laboratory. In this way, research moves back and forth between models and empirical investigations (Smith & Conrey, 2007). — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 860

Categorization

• SEED: The Kalick and Hamilton result — mate pairs formed later are less attractive — is a striking counterintuitive prediction worth chasing down for its mechanism and implications.
• XREF: Connects to complexity-science literatures on emergence and to the law-merchant/private-judges work: both show aggregate outcomes diverging from what individual-level rules would suggest.
• QUESTION: Why would later-forming mate pairs be less attractive? The mechanism matters for whether this generalizes beyond mate selection to any matching/formation process.