We investigate the extent to which agents can learn to coordinate onstationary perfect-foresight cycles in a general-equilibrium environment.Depending on the value of a preference parameter, the limiting backward(direction of time reversed) perfect-foresight dynamics are characterized bysteady-state, periodic, or chaotic trajectories for real money balances. We relax the perfect-foresight assumption and examine how a population ofartificial, heterogeneous adaptive agents might learn in such anenvironment. These artificial agents optimize given their forecasts offuture prices, and they use forecast rules that are consistent with steady-state or periodic trajectories for prices. The agents' forecast rules areupdated by a genetic algorithm. We find that the population of artificialadaptive agents is able eventually to coordinate on steady state and low-order cycles, but not on the higher-order periodic equilibria that existunder the perfect-foresight assumption.