We study adaptive learning behavior in a sequence of n-period endowment overlapping generations economies with fiat currency, where n refers to the number of periods in agents' lifetimes. Agents initially have heterogeneous beliefs and seek to form multi-step-ahead forecasts of future prices using a forecast rule chosen from a vast set of possible forecast rules. Agents take optimal actions given their forecasts of future prices. They learn in every period by creating new forecast rules and by emulating the forecast rules of other agents. Computational experiments with artificial adaptive agents are conducted. These experiments yield three qualitatively different types of outcomes. In one, the initially heterogeneous population of artificial agents learns to coordinate on a low inflation, stationary perfect foresight equilibrium. In another, we observe persistent currency collapse. The third outcome is a lack of coordination within the allotted time frame. One possible outcome, a stationary perfect foresight equilibrium with a relatively high inflation rate, is never observed.