Tools for Evaluating the Consequences of Prior Knowledge, but no Experiments. On the Role of Computer Simulations in Science
|Table of Contents|
|2 Common features of simulations and experiments|
|3 Distinguishing features of experiments and simulations|
|4 Borderline cases|
|4.1 Experimentum crucis and analog simulations|
|4.2 Simulation-like experiments|
|4.3 Experiment-like computer simulations|
|4.4 Hybrid simulation-experiments|
|5 Summary and conclusion: Computer simulations as a tools for drawing conclusions from prior knowledge|
The fact that the categories of simulation and experiment overlap in cases where none of the three distinguishing features become relevant, may lead to the mistaken conclusion that simulations and experiments cannot be clearly distinguished or separated as research methods or that it is at least hard to tell the difference (Morrison 2009, Parker 2009). The confusions arising from this mistaken conclusion can best be clarified by considering the border case of experiments that fall clearly outside the overlap region of the two categories and the other border case of experiments that obviously fall inside the overlap region.
The border cases that need to be considered in this respect are on the one hand the experimentum crucis that can never be replaced by a computer simulation, and the analog simulation that can in principle always be replaced by a computer simulation on the other hand.
An example for an experimentum crucis is Young's double-slit experiment to demonstrate the wave nature of light (Wikipedia double_slit). At the time when it was conducted it would for principle reasons have been impossible to replace this experiment by a computer simulation, since the outcome of a computer simulation would depend on which of the competing theories, wave theory or corpuscular theory, had been built into the simulation.
The opposite case of an analog simulation can best be highlighted by a quotation by John von Neumann, because it describes this case in ideal-typical form:
“The purpose of the experiment is not to verify a proposed theory but to replace a computation from an unquestioned theory by direct measurement. Thus wind tunnels are used as computing devices to integrate the nonlinear partial differential equations of fluid dynamics.” (quoted by Winkler et al. (1987, p. 28) and Winsberg (2010, p. 35))
In the particular case that von Neumann speaks of, the experimental setup performs the function of an analog computer. It is trivial that in this case the experiment can - without any loss of epistemic power - be replaced by a computer simulation if the computers are powerful enough. It is important to keep in mind that this example covers only one particular kind of experiment.