Basic HTML Version

Chapter 7
Robustness and Explanation
William Wimsatt
Robustness and explanation are multiply linked. Robustness is crucially connected, as physicist
Richard Feynman said, “to fundamental aspects of physical nature.” In his book
The Character of
Physical Law
(1967), Feynman argued that the fundamental principles of nature are remarkable
because they are derivable in multiple ways using multiple different assumptions. So in some sense
they are almost unavoidable. This he related to a “Babylonian” architecture of theory, in which
various elements of theory are multiply connected, redundant, and thus more reliable. For this
reason, fundamental physical laws do not depend on any particular assumptions; a fact that he
illustrated with two different independent derivations of the inverse-square law of gravitational
attraction. The great conservation laws in physics have this character, and much of the progress in
nineteenth-century physics progressed through the discovery of the interconvertibility of different
forms of energy. The conservation laws are, for this reason, deeply anchored in the explanatory
frameworks that we would use even for deciding what is a reasonable explanation and what is not.
However, this applies not only to the fundamental laws of physics, but also at the nitty-
gritty level of ordinary explanations. Ordinary objects are robust (Wimsatt 2007: Chapter 4). If I
am standing behind a podium when I lecture, I can use it to illustrate the point. I smack it on the
side and you hear it. I feel it with my hand. It is impenetrable. I put my computer on it, so it bears
weight. Visually, you cannot see the lower part of my body if I am behind the podium, so it is
visually opaque. Thus, I can detect the podium in multiple different ordinary ways. I could go
further with chemical analyses, and the other means of detection and analysis of modern science,
but you get the point. So, robustness is fundamentally connected with the objects that we see in
the world. Furthermore, the objects that are detectable in that multiplicity of different ways also
are things that are connected to other things through those modes of detection, which use different
of their properties. Thus, objects are central things from which to move out in organizing
explanatory schemes – they are going to be important to use. In terms of connectivity, they give a
big bang for a buck. A fundamental explanatory principle is therefore: Aim to explain that which
is less robust in terms of that which is more robust.
That applies not only to objects on different scales, but to the levels of organization
themselves (Wimsatt 2007: Chapter 10). Levels of organization are composed of robust objects
that are richly connected to one another, and thus play a role in their causal interactions and
explanations of their behavior. They are usually of roughly the same size and interact on similar
timescales. (An elementary particle can exist for 10
seconds, but any macroscopic object that did
so would be conceived of as an explosive illusion.) They hang together more strongly than objects
at that level do with objects at other levels, which they may compose or be composed of. You get
a bigger bang for a buck out of using the order found among those objects in explaining phenomena