[ Pobierz całość w formacie PDF ]
(here, b s firing) and the effect. If b s firing had not occurred, the firing of f would
62 Douglas Ehring
acd
e
bf g
Figure 5.3
not have; and, given that by the time f fires, the alternative process is already
doomed by the earlier firing of b, if the firing of f had not occurred, then both
processes would still have failed to run to completion (Lewis 1986: 171, 200).
Stepwise dependence is, thus, established. This approach does not work in Case A
since it requires that the blocking-initiator event dooms the preempted process
before occurrence of an intermediary event in the main line. In Case A, there is no
intermediary event after the blocking-initiator event. The blocking-initiator event,
g s firing, is also the direct cause of e s firing.6
In a later postscript to his original paper on causation, Lewis concluded that the
stepwise approach does not work for late preemption/late cutting. Consider the
following case (Figure 5.3) of deterministic late cutting (Lewis 1986: 203 4).7
The preempted line is blocked (d s firing is blocked) only by the effect event
itself and nothing earlier. For every event in the preempting process, it is false that
had that event not occurred, an earlier blocking event would still have occurred,
dooming the preempted process. Assuming that the final effect could have had
an alternate causal history and that it is not temporally fragile that it could
have occurred somewhat later stepwise counterfactual dependence fails. Lewis
revised his account as follows:
c causes e just in case there is a series of actual events x1, & , xn such that x1
depends or quasi-depends on c, x2 on x1, & , and e depends on or quasi-
depends on xn.
(Lewis 1986: 206)
Quasi-dependence is characterized basically as follows:
e quasi-depends on c just in case the intrinsic character of the process
connecting c and e is just like that of processes in other regions of this or other
worlds with the same laws and in the great majority of these regions these
processes display stepwise counterfactual dependence.
(Lewis 1986: 206)
Counterfactual theories, preemption and persistence 63
Lewis conjectured that the preempting cause satisfied this condition in cases
of deterministic late preemption (Lewis 1986: 206). Applied to the case in
Figure 5.3, in the great majority of regions with processes intrinsically just like
the one connecting b s firing with e s firing, there will be a chain of
counterfactual dependence since preemptive settings are rare. Lewis also seems
to presume that the same is not true of the preempted process. In the majority of
other-regional processes with processes much like a e, those processes will not
be exactly like that of a e since they will include an additional event in place of
the missing firing of d, for which there is no analogue in the a e process (the
missing event feature).
In order to test the quasi-dependence account against Case A, which is proba-
bilistic, let s now bring in Lewis s probabilistic notion of dependence required
for indeterministic situations in which there is still some chance that the effect
would have occurred spontaneously in the absence of the cause even without
preemption/overdetermination (Lewis 1986: 176).
For actual but distinct events c and e, e probabilistically depends on c just in
case the actual chance of e occurring is x (where the actual chance x of e is to
be its chance at the time immediately after c ) and if c had not occurred, e
would have had some chance y of occurring very much less than x.
(Lewis 1986: 176 7)
The quasi-dependence formulation, in effect, then becomes:
For actual and distinct events c and e, c causes e just in case there is a series of
actual events x1, & , xn such that x1 probabilistically depends or quasi-
probabilistically depends on c, x2 on x1, & , ande probabilistically depends on
or quasi-probabilistically depends on xn.
(Lewis 1986: 206)
The quasi-dependence approach does not work for Case A. Consider processes
in other regions that are just like the a c d e process (Figure 5.1). In a majority of
regions with the same laws in which there is realized a sequence which is event-
for-event intrinsically indistinguishable from that preempted line, the final effect
does stepwise probabilistically depend on events in that sequence. Those other-
regional processes do not differ from the a c d e sequence event-wise since
they do not include extra events . Although a transmission is blocked in the
preempted line, there is no cutting of events, no missing intermediary events.8
One might say that there is blocking but no cutting. The preempted line satisfies
the quasi-dependence account.
Consider a possible reply:
The failure of transmission/persistence in the preempted line is the relevant
intrinsic difference between the preempted process and its other-regional
64 Douglas Ehring
analogues in which there is no missing transmission. Missing transmission/
persistence takes over for the missing events.
I have two responses to this defence. First, this defence will not be acceptable to
Lewis. If the persistence of the particle or quantity of energy is itself to be
analysed as a matter of causally connected temporal parts of that particle/quantity,
that will be inconsistent with Lewis s reductionist counterfactual theory of causa-
tion. Or, if the persistence of this particle/quantity is analysed as a matter of that
particle/quantity being wholly present at each moment that it exists, that is not
[ Pobierz całość w formacie PDF ]