Tools for Evaluating the Consequences of Prior Knowledge, but no Experiments. On the Role of Computer Simulations in Science

Table of Contents

4 Borderline cases

The above described distinguishing features between simulations and experiments justify making a difference between the categories of simulations and experiments. At the same time they represent optional features of experiments that do not become acute in every single instance. Therefore, there exist instances of experiments that do not differ in any epistemically relevant sense from simulations which means that the two categories overlap. In the following, it will be argued that this does not imply that there are single instances where the distinction is blurred up to a point at which it cannot be determined any more whether some scientific procedure is a simulation or an experiment. Rather, a distinction can always be made, only in some cases the distinction is not epistemically relevant. In these cases computer simulations can (at least in principle) act as replacement for experiments.

In order to clarify the situation, I will first give examples of extreme cases of experiments that do not fall into the overlap region of simulations and provide examples of experiments that fall squarely inside the overlap region. Then, the borderline cases of simulation-like experiments and experiment-like computer simulations will be described on a more abstract level. Finally, I discuss the question of hybrid simulation-experiments, i.e. scientific procedures like magnetic resonance imaging where empirical raw data is post-processed by simulation-like procedures so that on the first sight it appears hard to tell whether it is a simulation or an experiment. Nevertheless, it is claimed, a reasonable distinction remains possible.

t g+ f @