I develop a behavioral macroeconomic model in which agents have cognitive limitations. As a result, they use simple but biased
rules (heuristics) to forecast future output and inflation. Although the rules are biased, agents learn from their mistakes
in an adaptive way. This model produces endogenous waves of optimism and pessimism (“animal spirits”) that are generated by
the correlation of biased beliefs. I identify the conditions under which animal spirits arise. I contrast the dynamics of
this model with a stylized DSGE-version of the model and I study the implications for monetary policies. I find that strict
inflation targeting is suboptimal because it gives more scope for waves of optimism and pessimism to emerge thereby destabilizing
output and inflation.