It is demonstrated in this paper that adaptive learning in least squares sense may be incapable to reduce, in a satisfactory way, the number of attainable equilibria in a rational expectations model. The model investigated, as an illustration, is the monetary approach to exchange rate...