The viscous damping forces commonly added in earthquake engineering applications lack a sound physical justification. Consequently, whether or not they accurately represent what they are added for is uncertain. In this paper, this latter source of uncertainty is removed, appealing to the concept of discrepancy forces, in the analysis of the capabilities of a nonlinear reinforced concrete frame elements to represent actual seismic behavior. Frame elements are modeled as fiber elements accommodating uniaxial concrete and steel behavior laws; two different nonlinear cyclic concrete behavior laws are considered: a stochastic multi-scale concrete model along with a more classical one. Concrete parameters are identified from experimental data using Bayesian inference technique. Finally, the capabilities of the two models to represent the actual response of a reinforced concrete frame element tested on a shaking table are compared in an objective way, that is without any side effect resulting from the action of uncertain damping forces.