Basic HTML Version

“Context-dependency does not stop it being science but, acknowledged and dealt
with, it may open up new approaches allowing us to push the envelope of scientific
knowledge a little further.” (Busch, 2010)
Attempts by Persson, Ylikosk, Steel, Mitchell, Reiss, and others to modify Woodward’s
(2003) model of causal explanation implicitly reflect such orthogonality. Woodward claims:
“One restriction in this claim is explicitly a realist one: namely that the manipulations be of
an ‘invariant’ (non-contingent, non-context-dependent) model of the action, event or
situation.” Daniel Steel
proposes to broaden this view by “suggest[ing] that the
concept of a ‘ceteris paribus model’ is helpful for understanding how, from a manipulationist
perspective, a causal model could be deemed explanatory despite failing to be invariant.”
“A ceteris paribus model may fail to be invariant yet be useful for guiding conditional
plans. Doing this involves two things: (1) the associated unless clause lists the most
likely contextual factors with which the intervention can interact and thereby disrupt
the generalization and (2) the generalization, or model, indicates some format that
could potentially incorporate those factors and examine their effects. Consider a first
step in a conditional plan that generates an ‘unpleasant surprise,’ that is, a result that
was contrary from what was expected and hoped for, and suppose that this has
happened because the intervention unexpectedly altered some contextual factor. If (1)
obtains, then it is likely that this contextual factor is included in the purview of the
associated unless clause, and if (2) obtains, the causal generalization or model upon
which the original action was premised suggest a means for explicitly including this
omitted contextual factor into the analysis. In this case, a model may fail to be
invariant with respect to a particular intervention, but may include resources for using
the information inherent in the ‘unpleasant surprise’ to make a model that is invariant
with respect to some further set of interventions or at least more nearly so.” (Steel,
What Woodward’s causal explanation model wants to stress – invariant intervention – is
perhaps impossible in the Science 2 world. If so, then the realist interpretation of Woodward
needs modification – not only in the
ceteris paribus
manner described above, but in a more
blatantly constructivist manner as well.
“People are self-interpreting beings who can learn from and change their
interpretations so that they can act and respond in novel ways, thereby producing
novel stimuli for subsequent actions. In other words, their causal powers and
liabilities are considerably more diverse and changeable (even volatile) than those of
non-human objects. While they are influenced by material circumstances, their
actions do not stand in fixed relations to them, precisely because they are mediated by
the ways of seeing available to them, and these can vary enormously. … On the other