Validation of simulation models

The reconstruction of the search for valid knowledge in science shows that standard methods for validation are based on a number of misunderstandings of the concepts Popper et al. introduced in the logic of science. To criticise are: (1) the naive principle of falsification; (2) the idea that only a cause influences the effect, and not the effect the cause; (3) and the prejudice that linear models are always better. From the practice of research in natural and socials sciences a plea is made for a better logic of validation, particularly concerning nonlinear models. A number of starting points from mathematics, system dynamics, cybernetics and statistics are presented for the development of new methods of qualitative and quantitative validation for nonlinear models.