Adaptive Learning in In nite Horizon Decision Problems