SINGULARISM
by Ann Whittle
I distinguish between two opposing intuitions about the nature of
the singular causal relation. The first stresses the nomological character of causation, while the second emphasises is seemingly local character. My question is
this: is it possible to formulate an account of causation which incorporates both
intuitions? Anscombe gives us reason to think that these intuitions could not be
jointly met in an account of causation. Foster and Tooley’s acount seems to
provide a counter-instance to her claim, but this proves to be a false start. I put
forward an alternative suggestion which, by utilising tropes, manages to reconcile both the local and nomological character of causation.
ABSTRACT
I
O
bservations on the nature of the singular causal relation
seem to pull us in two directions. First, we have what could
be called the Generalist’s Observation. This draws our attention
to the fact that causal relations are part of more general patterns.
If C causes E, this seems to imply something about similar situations in which C-type and E-type entities are present. If, for
instance, a balloon landed on something sharp and burst, we’d
expect similar balloons to burst if they landed on things equally
as sharp. The Generalist’s Observation can be articulated by the
principle of the nomological character of causation (NCC ). This
states that whenever there is a causal relation between two entities, C and E, there is a causal law that subsumes them (at least
under some description of C and E).1 The idea is that because
the relation between C and E will be part of a more general
pattern, it will follow from a more general causal truth or law.
The second observation draws our attention to a very different
aspect of the causal relation—its seemingly local character.
Whether or not two entities are causally related looks as if it
should depend solely on what happens between those two things.
The enjoyment caused by the chocolate Alice is now eating, for
instance, doesn’t seem to depend upon anything outside the
relation between Alice’s eating chocolate and her enjoyment. In
particular, it does not appear to rely on past or future events of
1. See, for example, Davidson 1970 p. 215.
372
ANN WHITTLE
a similar sort, nor upon events occurring at different places. It
seems only to concern what is going on at the time and place of
Alice’s eating the chocolate. This observation pushes us towards
the Singularist’s Thesis (ST ). This states that the causal connection between two relata doesn’t depend upon anything
extraneous to that relation. Rather, the truthmakers of singular
causal statements are entities which are local and intrinsic to
those relations.2
We have, then, two very different but plausible theses about
the nature of the singular causal relation. The question I shall
consider here is this: can we incorporate both ST and NCC in a
theory of causation?
II
Anscombe, who has been important in the development of singularism, suggests that we cannot satisfy both ST and NCC. She
argues that singular causal facts cannot be determined by facts
about causal laws, because nomic facts are unable to capture
what is essential to causation. She writes:
Effects derive or come out of their causes ... Now analysis in terms
of necessity or universality does not tell us of this derivedness of
the effect; rather it forgets about that. For the necessity will be
that of laws of nature; through it we shall be able to derive knowledge of the effect from its cause, or vice versa, but that does not
show us the cause as source of the effect.3
Anscombe thus endorses ST, because she thinks that an analysis
of the causal relation must be given in terms of particular causes
and effects. However, she suggests that accepting the claim that
singular causal facts are not determined by causal laws undercuts
our right to claim that causal relations are subsumed under laws.
For if we say that causal laws are not necessary for causal
relations (as the latter are not determined by the former), then C
can be the cause of E even if C and E are not subsumed under a
causal law. Hence, there is no reason to suppose that causal
relations will always be subsumable under causal laws.
2. On an intuitive understanding of intrinsicity, something is intrinsic to a relation if
it is determined by what lies within the confines of that relation, and so would remain
part of that relation regardless of what occurred outside it.
3. 1971, p. 92.
SINGULARISM
373
Anscombe is right to think that there is a natural connection
between ST and the denial of NCC. If we claim that the
truthmakers of causal statements are entities which are local and
intrinsic to those causal relations, then we have to deny that general causal facts determine the causal relations, as they will be
extraneous to these relations. This leads to scepticism about the
generalist’s claim that causal relations will be part of more general patterns and hence subsumable under causal laws. For why
should C causing E imply anything about other C-types causing
E-types if, as ST claims, C is the cause of E simply in virtue of
local, intrinsic facts of the relation between C and E? Anscombe
thus poses a challenge: why should NCC hold, if causal relations
are not determined by causal laws?
III
Foster and Tooley’s form of singularism may be thought to provide a response to this challenge.4 They endorse the Anscombesounding claim that not all singular causal facts are reducible to
(Foster) or supervenient upon (Tooley) facts about causal laws,
because there are possible situations in which indeterministic causal laws fail to determine which cause is paired up with which
effect.5 When this happens, they claim that there is an irreducible
singular causal fact which determines the pairing. Consequently,
it is wrong to say that causal laws (plus non-causal facts) determine all the singular causal facts.
This claim, however, is compatible with NCC. The fact that
extra causal facts may be required in order to determine which
cause is paired up with which effect, doesn’t demonstrate that
causal relations are not instances of causal laws. We might think
that despite this failure of supervenience, all relations have to be
subsumed under some law in order to be causal. This would be
the case, for instance, if we thought that laws, while not the sole
truthmakers of causal statements, were nevertheless among them.
Indeed, this sort of position seems to be one that Tooley
4. Their position is usually classified as a form of singularism due to its links with
Anscombe. However, if we take ST to be essential to singularism then, as we’ll see,
not all variations of this view will be singularist.
5. See Foster 1985, Part III, Section 7 and Tooley 1987, Ch. 6. All the scenarios
outlined depend upon the causal laws being indeterministic.
374
ANN WHITTLE
endorses.6 Thus, Foster and Tooley’s view does not force us to
reject NCC.
We have, then, an account which can combine NCC with the
thesis that causal laws fail to reduce the singular causal facts (call
this thesis FT ). Furthermore, FT is also compatible with ST, as
Anscombe’s analysis of causation clearly incorporates both of
these theses. This alone, however, does not suffice to show that
a version of FT could reconcile ST with NCC, as FT does not
entail ST. In order to see this, consider the case just given. There
laws were said to be part of the truthmakers of causal relations.
To make the example more concrete, suppose that laws are analysed via Lewis’s ‘best-system theory’.7 If these laws are part of
the truthmakers of causal statements (as they are on Lewis’s
counterfactual analysis), then ST will not hold. For Lewis’s laws
are just those global patterns of reoccurrence which appear in
the best scientific systems. Consequently, if they are among the
truthmakers of causal statements, causal relations will be partly
determined by entities which are extraneous to them.
Is there any reason to suspect that those accounts which
combine both FT and NCC, cannot also incorporate ST?
There are grounds for scepticism. FT theorists, such as Tooley,
are able to combine FT with NCC by utilising the generalist’s
rationalisation for NCC. In other words, they can say that causal
laws are among the truthmakers of causal statements. But
Anscombe’s challenge was to explain why NCC should hold,
given that no part of what it is to say that ‘A causes B’ is ‘comprised’ of the general causal truth ‘A-types cause B-types’.8 Now
Anscombe did have good reason to insist upon this, since it is
difficult to see how ST could hold even if laws are just one of the
truthmakers of causal statements. Laws concern facts about what
happens in all eons. Therefore, if they are responsible for the
truth of causal statements, it looks as if causal relations cannot
be determined solely by entities which are local and intrinsic to
6. At least in 1987. There, Tooley argues for the view that ‘underlying laws are necessary if there are to be causal relations’ but ‘laws together with relevant causal facts
are not always sufficient to determine that causal relations obtain’ (p. 203).
7. See Lewis 1986, p. 122.
8. See Anscombe 1971, p. 92.
SINGULARISM
375
those relations.9 This unresolved difficulty justifies the limited
conclusion that FT gives us no indication of how we might go
about reconciling ST and NCC.
IV
A more promising response to Anscombe’s challenge is suggested
by Ducasse’s brand of singularism. He writes:
A causal connection explains the regularity of the succession, but
it is not constituted by such regularity, which is but a corollary of
the causal connection.10
The thought here is that although regularities of E-types following C-types are not part of what makes the singular causal statement ‘C causes E’ true, these regularities are nevertheless implied
by the causal relation. But why is this? Ducasse suggests that it
is because causal laws are constituted out of ‘a class of resembling facts’, each of which ‘already happened to be a causal fact
individually and in its own right’.11 The position outlined by
Ducasse, therefore, is stronger and more positive than that
offered by either Anscombe or FT. For rather than just making
the negative claim that causal relations are not determined by
causal laws, Ducasse suggests that it is facts about causal
relations which determine facts about causal laws.
If we endorse this claim, we are making a radical break with
the generalist’s approach. However, it is still not clear why NCC
should hold. NCC doesn’t follow from Ducasse’s claim that causal relations determine causal laws, for there may still be some
causal relations which do not give rise to any causal laws and so
are not subsumable under them. What we require, is an account
which incorporates Ducasse’s thesis with the claim that the existence of a causal relation guarantees the existence of a causal law.
This would enable us to explain why causal relations are
9. I believe that there is one account of the laws of nature, namely Armstrong’s,
which is able to combine the generalist’s rationalisation for NCC with ST. I don’t
think it offers a convincing rapprochement of NCC and ST, however, as it relies on
the implausible thesis that spatiotemporal universals can be wholly present at different places at the same time. Due to the limits of space, I cannot pursue this line here,
but it does not count against the qualified conclusion I draw.
10. 1926, p. 130.
11. 1926, p. 129.
376
ANN WHITTLE
instances of more comprehensive causal laws, without thereby
undermining ST.
Is it possible to formulate such an account? I think it is. Suppose that we take causation to be a particular relation between
tropes. So, in any instance of singular causation, tropes are the
entities doing the causal work. This may be because causation is
reducible to certain facts about the tropes of the causal relation,
or because relations which are causal are related by an irreducible
causal necessitation trope. The details of the account don’t matter for these purposes, since as long as the truthmakers of singular causal relations are tropes, the account will be compatible
with ST. Why? Tropes are sui generis property instances, such as
‘the redness of this poppy’ or ‘the love of Bob for Katy’. Two of
their features make them suited for the role outlined here. First,
tropes are spatiotemporal particulars, so they have a unique, spatiotemporal location. Second, they are not analysable in terms of
instantiations of universals. This allows tropes to be wholly present and intrinsic to the entities which have them, as they are not
comprised of general entities which exist outside the confines of
the particular that possesses them.12 If tropes determine causal
relations, therefore, the truthmakers of singular causal statements will be entities which are local and intrinsic to those
relations, as ST demands.
Although tropes are particulars, trope theorists claim that sets
of tropes form natural or genuine universals. How are these
tropes grouped together into sets? This question is usually answered by appealing to the relation of exact resemblance, but this
is not as lucid as we might hope. First, it fails to offer an account
of what it is to belong to the set of tropes which stand for the
universal P-ness, rather than, say, the universal G-ness or F-ness.
Second, in some cases it may be unclear what it is for two tropes
to resemble each other exactly. Two tropes of charge will not be
identical in every respect since, according to trope theorists, these
entities are particular. Consequently, we have to say that their
12. Both traditions regarding the nature of tropes view them as intrinsic to the entities that instantiate them. Stout (1921) takes tropes to be independent parts of the
wholes they constitute. They are thus intrinsic to the entities they comprise, because
they literally lie within their confines. Husserl (1901) takes tropes (what he calls
‘moments’) to be entirely dependent on the wholes of which they’re part. Consequently, tropes are intrinsic to the wholes which possess them, because they are
entirely dependent on these wholes.
SINGULARISM
377
qualitative, non-particularised aspects are precisely similar. But
what does this amount to for tropes like charge? We can’t
just imagine what two precisely similar tropes of charge look
like, as we may be able to imagine tropes of red. Third, it is not
obvious that we should say that all tropes in a set have to
resemble each other exactly. If the tropes were property instances
of the determinable red, for instance, then this criterion would
be too stringent.
I think we can improve on this standard account by utilising
Shoemaker’s functionalist theory of properties (1981). Shoemaker argues that properties should be understood in terms of
their functional roles. However, we can also utilise his account
in order to provide a more precise specification of how tropes
(which stand for genuine universals) are grouped into sets. Those
tropes which implement the functional role definitive of a universal form a set, which grounds genuine resemblances between particulars. So the idea is that tropes belong to the same set, and are
thus instances of the same universal, if they meet certain causal
requirements laid down by the functional definition for that universal. Trope C1 and trope C2, for example, are members of the
set which stands for the universal of charge, if and only if both
C1 and C2 implement the causal role described by the functional
definition for the property of charge.
If we adopt this view, then causal relations will be guaranteed
to be part of more general patterns. Take, for instance, the causal
relation ‘Jack’s fall caused his crown to break.’ The events ‘Jack’s
falling’ and ‘his crown breaking’ possess certain tropes, the most
crucial ones being Jack’s trope of falling and his crown’s trope
of breaking. These tropes belong to sets, the members of which
all implement the same nexus of causal relations. Thus, if Jack’s
trope of falling and his crown’s trope of breaking are causally
related (given certain circumstances), then Jill’s trope of falling
and her crown’s trope of breaking will also be (given the same
circumstances), since Jill’s trope of falling and her crown’s trope
of breaking enter into the same causal relations as Jack’s trope
of falling and his crown’s trope of breaking.
The existence of the causal relation between Jack’s trope of
falling and his crown’s trope of breaking, therefore, guarantees
the existence of a more general pattern. This general pattern will
have a law-like description of the form: in circumstances C, the
378
ANN WHITTLE
universal of falling nomically necessitates the universal of breaking. Thus, we can say that the causal relation ‘Jack’s fall caused
his crown to break’ is subsumed under a law, because it is an
instance of this more comprehensive causal law. The same will
also hold for all other causal relations, but this won’t be because
causal laws are truthmakers of singular causal statements; rather,
it is because singular causal relations automatically give rise to
laws.
V
On this proposal, causal relations will be part of more general
patterns describable by causal laws, in accordance with NCC,
because relations between tropes give rise to more general
relations between properties. Furthermore, what determines the
causal relations will be local and intrinsic to those relations, as
the singularist claims, since causal relations are entirely determined by their tropes. Therefore, it is possible to reconcile NCC
and ST. This enables us to keep what seems plausible about
generalism, without thereby losing what seems right about
singularism.13
Ann Whittle
Department of Philosophy
Uniûersity College London
Gower Street
London, WC1E 6BT
a.whittle@ucl.ac.uk
REFERENCES
Anscombe, G. E. M., 1971, Causality and Determination (Cambridge: Cambridge University Press). Reprinted in E. Sosa and M. Tooley (eds.) Causation, 1993 (Oxford: Oxford University Press), pp. 88–104.
Davidson, D., 1970, ‘Mental Events’, L. Foster and J. W. Swanson (eds.) Experience and Theory. Reprinted in Davidson’s Essays on Actions and Eûents, 1980
(Oxford: Clarendon Press), pp. 207–227.
Ducasse, C. J., 1926, ‘On the Nature and Observability of the Causal Relation’,
Journal of Philosophy vol. XXIII, no. 3. Reprinted in Sosa and Tooley 1993,
pp. 125–136.
Foster, J., 1985, A. J. Ayer (London: Routledge).
13. Many thanks to Mike Martin, Joel Smith and audiences in Glasgow and Treaty
Street.
SINGULARISM
379
Husserl, E., 1900兾1901, Logical Inûestigations, trans. J. N. Findlay, 2001 (London: Routledge).
Lewis, D., 1986, ‘Postscripts to ‘‘A Subjectivist’s Guide to Objective Chance’’ ’,
Philosophical Papers Vol. II (Oxford: Oxford University Press), pp. 114–132.
Shoemaker, S., 1981, ‘Some Varieties of Functionalism’, Philosophical Topics,
vol. 12, pp. 83–118.
Stout, G., 1921, The Nature of Uniûersals and Propositions (British Academy
Lecture, Oxford: Oxford University Press).
Tooley, M., 1987, Causation: A Realist Approach (Oxford: Clarendon Press).
PRESIDENT: PROFESSOR BOB HALE
*Bell, Professor Martin
*Child, Dr. William
Cooper, Professor David E.
Cottingham, Professor John G.
Dancy, Professor Jonathan
Griffin, Professor James
Hamlyn, Professor D.W.
*Heal, Professor Jane
Holland, Professor R.F.
Hookway, Professor Christopher
Hornsby, Professor Jennifer
*Mackie, Dr. Penelope
Mellor, Professor Hugh
Morton, Professor Adam
O’Connor, Professor D.J.
O’Neill, Baroness O.
*Osborne, Dr. Catherine
*Price, A. W. (Editor)
Quinton, Lord Anthony
Skorupski, Professor John
Smiley, Professor T.J.
*Snowdon, Professor Paul
Sorabji, Professor R.R.K.
Sprigge, Professor Timothy
Strawson, Professor Sir Peter
*Sturgeon, Dr. Scott (Series Editor)
*Sullivan, Dr. Peter
Wiggins, Professor David
*Wolff, Professor Jonathan (Honorary Secretary)
Wollheim, Professor R.A.
*Executive Commitee
Additional Referees: Helen Beebee, Michael Bratman, Ursula
Coope, Dorothy Edgington, Sebastian Gardner, Ken Gemes,
Michael Otsuka, Timothy Williamson.