2006 Dozier Award Winner / Agent-based models can provide paleoanthropologists with a view of behavioral dynamics and site formation processes as they unfold in digital caricatures of past societies and paleoenvironments. This paper argues that the agent-based methodology has the most to offer when used to conduct controlled, repeatable experiments within the context of behavioral laboratories. To illustrate the potential of this decidedly heuristic approach, I provide a case study of a simple agent-based model currently being used to investigate the evolution of Plio-Pleistocene hominin food sharing in East Africa. The results of this null model demonstrate that certain levels of ecological patchiness can facilitate the evolution of even simple food sharing strategies among equally simple hominin foragers. More generally, they demonstrate the potential that agent-based models possess for helping historical scientists act as their own informants as to what could have happened in the past.