Identity, Structure, and Causal Representation in Scientific Models - Mechanism and Causality in Biology and Economics - page 51

variable, determined by “deeper” parameters (Lucas 1972 , 1973 ; Hoover 1988 ,
Chap. 2). The monetary-policy system impounds these deeper parameters in the
causal field.
The monetary-policy rule in ( 3.13 ) offers another example. As written, we can
analyze the effects of different settings of the parameter

λ

. We can also consider an
“institutional” change in which the rule is altered to depend on different condition-
ing variables or in different ways. As it stands, these alternatives have been
impounded in the causal field. When released and represented in a model in
which ( 3.13 ) is a special case, we can consider which is the best rule within the
now wider class of rules, which contains the current rule as one parameterization
(see Woodford 2003 for an extensive discussion of optimal monetary rules).

3.2 Interventions and Identity

The structural account of causal order is similar to Woodward's account in a
number of ways. A key difference, however, is that direct cause is not defined
with respect to a token-level notion, such as Woodward's come-what-may inter-
vention. Direct cause is expressed instead entirely with respect to the type-level
relationship between a privileged set of parameters and a functional relation
representing the interrelationship among variables. The structural account, as we
showed in the last section, supports a notion of the identity of causal systems: two
causal systems are identical if, and only if, they differ at most by their parameteri-
zation (i.e., they differ only in the token settings of the parameters). This under-
standing of causal identity also suggests a different conception of an intervention.
The notion of a parameter developed in the structural account was inspired by
Simon's notion of direct control and the notion that the parameter space could be
thought of as the loci of direct control. Direct control is virtually indistinguishable
from Woodward's notion of intervention. The only question that separates them is
direct control of what? For Woodward, it is direct control of a variable; for the
structural account, it is direct control of a parameter. But a parameter was defined to
be a variable subject to some additional constraints; so the difference seems small.
I have no doubt that the experience of manipulation and control are the source of
our original intuitions about causal powers and, therefore, are important in the way
that we learn about causes and learn to use causal language. Nonetheless, the
structural account does not actually use the notion of direct control in any physical
or metaphysical sense to define causal order. For Woodward an intervention
involves a change to the bearer of a variable - a real entity. In contrast, the notion
of a parameterization does not require that we change any parameter in a temporal
or genetic sense, but merely that we consider different settings of parameters in
otherwise causally identical systems.
Woodward accepts that the relevant intervention could be hypothetical and
certainly need not be practically implementable (e.g., removing the moon to
discover its effect on the tides would be an acceptable, but hypothetical,