Identity, Structure, and Causal Representation in Scientific Models - Mechanism and Causality in Biology and Economics - page 35

everything that could be truly said and short of saying everything that we might like
to say. Recognition of the gap between the representation and the world leads
various philosophers - among them, Paul Teller ( 2001 ) and Ronald Giere ( 2006 )-
to reconceptualize scientific knowledge as perspectival. Part of the reconceptuali-
zation is a repudiation of the view that omni science sets the standard for the
worthiness of scientific knowledge.
The truth in a once-common vision of science is that the only fully adequate
scientific knowledge trades in exceptionless, universal generalizations - scientific
laws. All more specific knowledge is, in principle at least, derivable from these
laws. Recognizing - as indeed any serious philosopher or scientist must - that we do
not, in fact, possess all the laws simply meant that what we did possess was a
slightly shabby, deficient version of what we wanted. We do not stand on Olympus,
but science was nonetheless to be judged from the Olympian heights.
An alternative vision of science championed by Giere and Teller, as well as by
Nancy Cartwright ( 1999 ), and William Wimsatt ( 2007 ), among others, starts lower
and builds upward. The standards of good or successful science are partial and
local, and science itself is constructed, to use an apt term from the subtitle of
Wimsatt's ( 2007 ) topic, in a piecewise manner.
The local knowledge that grounds science in this vision is often causal knowl-
edge. Yet, like other parts of science, causation has often been analyzed top down.
Many accounts of causation - for example, those of David Lewis ( 1973 ) and Daniel
Hausman ( 1998 ) - explicate causes against a background of universal laws. In
contrast, piecewise accounts of science typically take the causal relation as primi-
tive or, at least, built from something more local and specific than universal laws.
A piecewise approach is especially suited to economics and other social
sciences, biology, and areas of physical sciences, such as climatology - fields that
would be hard to analyze from a small set of universal laws on the model of
Newtonian mechanics. Economists, for example, increasingly conceptualize eco-
nomics causally, as evident in the work of Clive Granger in time-series economet-
rics, James Heckman in microeconometrics, recent developments in “natural
experiments” in economics, and counterfactual analysis. 1
Causal realism is the doctrine that causal relationships exist in the world and that
the role of causal models is to represent them adequately for some purpose. Not all
scientists (nor all philosophers) who talk about causes are realists, but that is a
question for another day. Here, I want to focus on representation of causes and not
on fundamental ontology or epistemology. It is a commonplace that different
representations or notational schemes allow us to see different things and that
some schemes are more effective than others - consider Arabic numerals or
Feynman diagrams. The main goal of this chapter is to develop a scheme for
representing causal relationships and to consider the light that it sheds on how we
1 See Hoover ( 2008 and 2012a ) on the place of causal analysis in economics and Reiss ( 2007 )on
natural experiments and counterfactual analysis. Hoover ( 2004 ) documents the fall and rebirth of
causal analysis and language in economics.