When a principal and an agent operate with simple contracts, at equilibrium, renegotiation will occur after the action is taken. Also, since renegotiation makes incentive contracts non-credible, the principal may prefer non-renegotiable monitoring options. Current literature does not fully reconcile these predictions with the observation of simple non-renegotiated incentive contracts. We model a principal-agent interaction in a social learning framework, and assume that when renegotiation is not observed, players may forget its feasibility, with infinitesimal probability. The unique stable state of our model predicts that the second-best simple incentive contracts occur with non-negligible positive frequency.