learning procedures. We consider the Stochastic Gradient-Algorithm as a boundedly rational learning procedure in an univariate … influence of the forecast feedback is limited the learning agents cannot destabilize the model and learn to form rational …