[ Pobierz całość w formacie PDF ]

tion that (ii) when nonequivalent spells are cast the first comes true, but when only
equivalent spells are cast the last comes true. Using (ii) as a guide would make Morgana
efficacious. McDermott also argues that Schaffer s appeal to considerations of
simplicity will not have the consequence that (i) is a law, but (ii) is not, since both are
theorems and, on that model, simplicity considerations apply to axioms, not theorems.
4 Or more precisely, this example is a deterministic version of a indeterministic case from
Menzies (1989: 646) which itself is an indeterministic modification of a case from
Lewis (1986: 200). Figure 6 is borrowed from Menzies (1989: 646).
5 Had the preempted line not been blocked, it would have given rise to the effect at the
time the actual effect occurred by way of some additional events (missing intermediary
events).
6 Menzies (1989: 645 7) discusses a probabilistic version of early preemption and shows
that a probabilistic version of Lewis s original formulation is inadequate in that there
can be a chain of probabilistic dependence, as the latter is characterized by Lewis,
without causation. Menzies offered a revision of counterfactual theory which requires
that causal chains be spatiotemporally continuous in some sense to deal with this
problem. Menzies later (1996) gave this account up because it rules out temporal action
at a distance and because it cannot handle late preemption.
Counterfactual theories, preemption and persistence 75
7 I borrow Figure 5.3 from Menzies (1989: 652).
8 Ganeri, Noordhof and Ramachandran (1996: 223) complain that a  quasi-dependence
approach rules out brute singular causation in cases of preemption. Noordhof (1999:
101) also complains that this  quasi-dependence approach misclassifies the preempted
cause in certain cases of probabilistic preemption.
9 This variation in the case is based on cases found in Collins (2000: 231), Schaffer
(2001b: 16) and McDermott (2002: 92).
10  For any pre-empted cause, x, of an event, y, there will be at least one possible event &
which fails to occur in the actual circumstances but which would have to occur in order
for x to be a genuine cause of y & All genuine causes, on the other hand, do seem to run
their full course; indeed, they presumably count as genuine precisely because they do
so (Ramachandran 1997: 273).
11 It takes two units of time to travel from node to node.
12 Noordhof (1998a) also suggests that M-set analysis will fail for indeterministic causes
of effects that have a probability of occurring anyway.
13 Ramachandran (1998) proposes more than one revised M-set account.
14 Cases of frustration involve no  missing events , but they do involve  delayed effects ,
and if  Analysis #2 gets a grip, it does so, in part, because of this feature.
Ramachandran (1998: 467) does not think this solution will work in frustration cases in
which the connection between a s firing and d s firing is direct.
15 Noordhof s account descends from an account offered by Ganeri, Noordhof and
Ramachandran (1996, 1998).
16  All it relies upon is the existence of possible events which were actually suppressed due
to preemption (Noordhof 1999: 104).
17 With probabilistic late preemption the preempted cause while raising the probability of
the effect does not raise it at the time the effect occurred but only later.
18 This modification also incorporates Noordhof s conviction that the presence of a cause
makes the probability of the effect greater at the time of its occurrence than at any other
time if the cause were not present relative to the events in £.
19 Clause (II) is also modified in response to cases of probabilistic early preemption in
which the alternate line is blocked, not by the main line, but by a distinct inhibitory
causal process and in which the preempted cause cannot be ruled out by clause (II). The
modified clause reads as follows: (II)' for any superset of £, £*, (where £†"£*), if e
2
probabilistically £*-depends upon e , then every event upon which e probabilistically
1 2
£*-depends is an actual event (Noordhof 1999: 107).
20 On the other hand, Noordhof argues that the firing of b satisfies clauses (I III) (1999:
113).
21 The four clauses mentioned form a necessary condition for causation to be supple-
mented with an account of causal asymmetry (Noordhof 1999: 120).
22 g s firing is excluded from A: (a) If g s firing is a member of A, then
firing> satisfies (IV); and (b) If g s firing is not a member of A and we replace (IV)(1)
and (2) with (1*) If a s firing and g s firing were to occur, with none of the events in A
occurring, nor g s firing satisfying any of (I) to (III) regarding e s firing, then it would be
the case that p(e at t ) e" x; (2*) If g s firing were to occur with neither a s firing nor any
2 0
of the events in A occurring, nor g s firing satisfying any of (I) to (III) regarding e s
firing, then it would be the case that p(e at t ) d"y then would not
2 0
satisfy (IV) (Noordhof 1999: 117). In his (2000: 323), Noordhof weakens this test:  I
need not have formulated the test condition & in terms of an event failing to satisfy any
of conditions (I) to (III). Instead, I might have required just that the event fail to satisfy
all of condidions (I) to (III).
23 If that is not true, we can tweak the case by increasing the reliability of the a-process and
lowering the chance of g s firing inhibiting the a-process.
76 Douglas Ehring [ Pobierz całość w formacie PDF ]

zanotowane.pl

doc.pisz.pl

pdf.pisz.pl

dona35.pev.pl