Hans Radder begins his introduction to this collection of twelve essays on scientific experimentation by lamenting the loss of ‘momentum’ amongst philosophers of science in the turn towards experiment that began in the early 1980s. While historical and sociological studies of experimental practice have flourished, philosophers of science have not generally given the experimental aspect of the scientific enterprise the sustained attention that it deserves. The Philosophy of Scientific Experimentation seeks to stimulate a resurgence of interest in experiment by collecting together works by some of those who have devoted significant effort toward a philosophical understanding of experiment. It succeeds by carefully balancing a diversity of topics with the coherence of recurring themes and problems that run through the collection. The book is well-suited for any philosopher of science seeking an introduction to the most pressing problems of the philosophy of scientific experiment and the current literature on those problems.
Since it will not be possible to comment on each of the essays in a brief review, I will confine myself to discussing some of the important problems and themes that emerge from the collection.
Many of the essays take up, in one way or another, the question of how the materiality of instruments and laboratories relates to the putatively nonmaterial knowledge gained by performing experiments. Davis Baird puts forward the boldest claim on this front. Baird advocates an extension of epistemology into the material realm, in which material things can act as bearers of knowledge. Frankly, I was initially skeptical. However, by showing that pieces of reliable technology can work very much in the way propositional knowledge works, and by showing how such an epistemology can make sense of the category of knowledge often called ‘know-how’, Baird effectively puts the burden on others to show why we should deny that material things can serve as bearers of knowledge just as legitimately as beliefs can.
Worries over the distinction between what is part of nature and what is not also recur in this collection. Peter Kroes argues that a tension exists within the conception of the nature of modern experimental physics. While physical theory depicts nature in nonteleological terms, physical experiment cannot be understood nonteleologically, insofar as every experiment uses devices that embody a design, and experiments yield scientific knowledge only insofar as such devices function as designed. Although the ‘structural’ descriptions of theoretical physics are not threatened by this, as the technological artifacts of the laboratory remain constrained by the laws of nature, modern physics is left unable to provide a foundation for distinguishing the natural from the artificial. Kroes argues that such a distinction is needed, however, for experimental physics as the analysis of experimental outcomes requires the separation between ‘natural phenomena’ and ‘experimenters’ artifacts’.
It is not clear that the distinction between the natural and the artificial merits so much concern. Arguably, what experimenters must be able to separate is not so much natural phenomena versus experimental artifacts as it is the phenomenon being studied versus confounding effects, whatever their source, that would, if left unaccounted, lead one to draw an erroneous conclusion. This same concern arises both in the study of putatively natural phenomena and in the testing of putatively artificial technological devices. Surely Kroes is correct in that the conduct and interpretation of experiment is guided by normative principles that cannot be explained by physical theory alone, but that does not make for news in philosophy.
In some cases, articles in this volume generate an illuminating dialectic on particular issues. For example, Jim Woodward here defends his intervention account of causation as vindicated by its ability to make sense of the way that experiment is used to identify causal relations. Furthermore, he points out the methodological value of an explicit theory of causation as a means to encourage specificity by researchers who make causal claims. That Woodward's theory is an improvement over von Wright's earlier intervention account is contested by Rainer Lange, who argues, pace Woodward, that von Wright's theory is not anthropocentric and that Woodward's theory is circular.
Another helpful exchange concerns the possibility of theory-free experiments. Ian Hacking famously claimed in Representing and Intervening that theory-free experiments are possible after all, but left his readers much to ponder regarding just what this meant. Essays by Michael Heidelberger and Hans Radder provide useful new insights and offer crucial distinctions. Heidelberger argues that, particularly in the exploration of a new phenomenal domain, when an instrument is used ‘constructively’ to produce a phenomenon that is relatively pure from confounding effects, experiment can proceed without drawing upon theory to situate the phenomenon within a symbolic or theoretical context. Radder disputes Heidelberger's claim in his essay (which also concerns the relation between experimental science and technology), maintaining that at the very least the judgment that repeated runs of an experiment yield consistent results must draw upon theoretical commitments about the variability of the phenomenon under investigation.
Essays by Mary S. Morgan, Evelyn Fox Keller, and David Gooding address, in different ways, the uses of models, and particularly of computer-based models, in experimentation. Morgan explores a taxonomy of experiments of different types, with material, laboratory-based experiments at one end, symbolic manipulations of mathematical models at another end, and different kinds of ‘hybrid’ experiments in between. Keller offers a different kind of taxonomy, distinguishing between the use of computer simulations to extract solutions from mathematically intractable theories (e.g., via Monte Carlo methods), the simulation of an idealized version of a physical system to produce computationally tractable and physically realistic models of the system, and the construction of models of phenomena lacking antecedent theoretical understanding (her example is cellular automata).
Following in the sprit of Francis Bacon's ‘idols of the mind’, Giora Hon proposes a scheme to identify major sources of error in the pursuit of experiment. Hon argues that his classification is superior, however, in that it attempts to correspond to ‘underlying principles of experimentation’. Knowing the ways that experiment goes wrong and knowing the essential elements of experimental procedure go hand in hand, in Hon's view. Hon identifies four stages of experiment, each yielding a distinct source of error: “the background theory, the apparatus, observation and measurement, and finally interpretation” (191). To these correspond what he calls the idols of, respectively, the script, the stage, the spectator, and the moral. Hon claims that his approach constitutes an improvement upon those of Deborah Mayo (whom he accuses of having a narrow understanding of error) and Allan Franklin (whom he accuses of producing an “ad hoc” list of epistemological strategies with an unsatisfying “etc.” at the end). In fact, his approach complements both of their views. Hon's typology of experimental errors is an example of the kind of typology of errors that Mayo explicitly calls for as part of the intelligent planning of an experiment, and provides a functional classification of the errors that can be ruled out by employing Franklin's strategies.
The Philosophy of Scientific Experimentation is an important contribution to this field of study and would make an excellent text for use in a graduate course in philosophy of science or one more focused on experimental practice.
