Overdet 3

Pluralism, Causation and Overdetermination
Author(s): Brian Jonathan Garrett

Source: Synthese, Vol. 116, No. 3 (1998), pp. 355-378
Published by: Springer
Stable URL: http://www.jstor.org/stable/20118086
Accessed: 19-03-2018 04:08 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and extend access to Synthese
This content downloaded from 132.174.250.186 on Mon, 19 Mar 2018 04:08:03 UTC
All use subject to http://about.jstor.org/terms
BRIAN JONATHAN GARRETT
PLURALISM, CAUSATION AND OVERDETERMINATION*
"This causal superfluousness is hard to believe in; it is

hard to believe that God is such a bad engineer."
Stephen Schiffer [1987].
0. INTRODUCTION
Metaphors such as Stephen Schiffer's have a strong hold on philosophical

imagination. Such metaphors could be expanded in the following manner:
God is the almighty engineer, with his eye upon the economics of construc
tion. When God arranges the world's causal facts He leaves no gaps, and
He doesn't waste his time with unnecessary causes: one entirely sufficient
cause, for each and every effect, is adequate for His purposes.
Such a metaphysical picture has natural consequences for our explana
tory and epistemological metaphors, as seen in dead metaphors like that
embedded in: The Principle of Sufficient Reason. There must be some
reason for each event's existence, some reason in God's mind. The intuition
that there is a reason for each and every event is transformed, in a secular
scientific context, into the idea that there is a full and complete expla
nation for each and every event. But where there is a full and complete
explanation for an event, we have answered the question: Why did this
event occur? Thus, having answered the relevant "why-question", there is
then no further need for a second answer; there is no reason to postulate
such "causal superfluousness". But where reason appears to deliver more
than one answer to the why-question regarding some event, reason must
be speaking in tongues: different scientific languages reveal the very same
cause or truth (Davidson, 1963).
We can see that token physicalism (Davidson, 1970) is the most suitable
metaphysical position to accommodate these intuitions and metaphors.
Every behavioral event must have only one cause, but reason may speak in
tongues and describe that cause in many different ways, either as mental
or as physical. If it is true that mental events are causes, then they must
be one with whatever else is mentioned by reason, when reason explains
behavior.
^M Synthese 116: 355-378, 1998.

P* ? 1998 Kluwer Academic Publishers. Printed in the Netherlands.
356 BRIAN JONATHAN GARRETT
Token physicalism can be approached from a different angle. God

leaves no "gaps" in His causal chains, but reason, when speaking in the
psychological tongue, leaves gaps, in the form of idealizations and ce
teris paribus qualifications. Thus, if God leaves no "gaps", those things
mentioned by psychology must also be those things mentioned by reason,
speaking in the language that doesn't require such qualifications. That
tongue, that language, so many have hoped, is the language of physics,
albeit the language of a future, perfected physics.
These metaphors and the philosophical and scientific articulation of
these metaphors are extremely powerful, but I think they are mistaken.
Gradually, we are beginning to see problems with these views. Arguments
regarding the failure of ontological and epistemological reductionism sup
port the idea that the causes mentioned in such autonomous scientific
theories are numerically distinct. Perhaps the various scientific theories
reflect a plural ontology, just as they appear to do. The plausibility of
genuine interdisciplinary causal explanatory practice, appears to indicate
that there is more than one cause for some particular effect. Combined with
suspicion over the success of ontological and epistemological reduction
ism, the existence of such interdisciplinary explanations for certain events
should lead us to reject the economic metaphor embedded in the token
physicalist's worldview. Perhaps God was not so frugal when He organized
the causal facts. Perhaps the world is somewhat excessive, meaning that
there are events that are overdetermined by more than one sufficient cause.
Event pluralism, causation and over/determination are thus the focus of this
paper.
Davidson's reasons for token physicalism rest, crucially, upon his con
ception of causation. The concept of causation is, more often than not,
the Archilles Heel of metaphysical theories. Token physicalism is the view
that mental events are physical events and that mental predicates are anom
alous, i.e. that there are no true law statements invoking such predicates;
where a law statement is an exceptionless (or "strict") counterfactual sup
porting generalization. Davidson argues for token physicalism by utilizing
the following assumptions:
(A) Mental events cause, and are caused by, physical events.
(B) Causation is nomological, i.e. causal relations between events

require the truth of at least one strict law statement that sub
sumes the events in question. That is, e\causes e2 only if there
are predicates P and Q such that e\ satisfies P at tq and e2
satisfies Q at tr and it is a law that For all x, (Pxtx -* There is
*y,(Qyty)l
PLURALISM, CAUSATION AND OVERDETERMINATION 357
(C) Psychological predicates (properties) are anomalous.
Given that mental events do interact with physical events, and given that
causal relations require strict laws covering such events and furthermore,
that there are no strict psychological laws subsuming these events, it
follows that mental events must fall under physical laws. If an event is
subsumed by a physical law, then it must be a physical event. Thus, all
mental events are physical events.
Two points require emphasis. First, psychological predicates are anom
alous, hence psychological laws are not the laws which satisfy the nomo
logicality of causation. Thus there cannot be causal relations between
events merely in virtue of a psychological generalization. That is to say,
we cannot conclude that two events are causally related from the mere
fact that they satisfy a psychological law (given that such laws are ce
teris paribus). Secondly, physical predicates are nomological, that is to
say, there are strict laws couched in a physical vocabulary; hence causal
relations may obtain in virtue of there being a true physical generalization.
We may conclude, therefore, that if two events are causally related, they
satisfy a strict law from physics. This asymmetry between the roles of
physical and psychological predicates marks Davidson's philosophy as a
form of nonreductive physicalism, and it is this asymmetry that has worried
his critics. The physicalist point is that only physical predicates can sat
isfy the necessary conditions on causation. Without this claim, Davidson's
argument for token physicalism is blocked.
But is this conception of causation plausible? Do causal relations be
tween events require that these events are subsumed by strict laws, or by
something less, say ceteris paribus laws: laws that admit of exceptions?
I am unconvinced. I see little reason for such a restriction on causation.
Certainly our folk-conception of causation does not entail such a claim.
Second, ceteris paribus laws appear to be invoked in our causal explana
tory practice. If a ceteris paribus law is invoked to justify a causal claim
regarding two events, in practice, that is sufficient reason for taking the
causal claim to be true (assuming, of course, you think that the law
statement is itself true). To justify a causal claim regarding two token
events, for example, the rise in the rate of poverty and the rise in the rate
of crime, we may need recourse to a ceteris paribus economic law that
poverty causes crime. There is little reason to think that, were we to fail in
discovering an underlying exceptionless law subsuming these two events,
we would have to retract our causal claim. In practice, economics provides
us with the law needed to secure belief in the causal claim, yet that law is
not an exceptionless law.
Furthermore, as Cartwright (1983) has argued, whether there are excep

tionless laws, physical or otherwise, is not so obvious. But if it is an open
question whether there are exceptionless laws, then it is an open question
whether causal relations exist. But that is pretty hard to swallow. I don't
think the existence of causal relations is an open question. The exact analy
sis of what is required for the existence of causal relations is certainly an
open question and it is Davidson's particular analysis that is problematic.
But without this constraint on causation, Davidson's argument for token
physicalism is incomplete.
In the remainder of this paper I shall explore some ideas regarding
causation within a pluralist ontology. I shall try to show that we need
not be uncomfortable with the idea of overdetermination and pluralism.
Continuing with the metaphors, I shall argue that it is not metaphysically
offensive to accept a conception of the world as excessive rather than
as economically frugal. Pluralist overdetermination, I shall argue, is as
common as our interdisciplinary causal explanatory practice leads us to
believe.
I shall limit myself to a discussion regarding mental and physical
events, although what I say is not specific to psychology. I shall assume
that a modest account of ontological pluralism (to be explained below) is
true, for the purposes of defending that account against Jaegwon Kim's
(1987, 1990 reprinted in Macdonald and Macdonald, 1995) anti-pluralist
"Exclusion" arguments. Second, in broad conformity with "the" traditional
picture, I assume that a cause is nomologically sufficient for its effect, such
that given the existence of the cause, the effect is nomologically guaranteed
to occur1 I want to defend pluralism and overdetermination. If I can defend
these two claims, even when causes are understood as nomologically suffi
cient for their effects, then these claims will be more acceptable to anyone
who has a weaker notion of causation at hand.
I shall not, however, provide detailed arguments for pluralism. Rather, I
want to defend pluralism against Kim's well-known "exclusion" argument,
that purports to show that dualism, along with several other assumptions,
leads to event epiphenominalism. The argument, put in a nutshell, is that
event dualists cannot hold that non-physical events cause behavior, for our
commitment to the "closed character of the physical domain" (Kim, 1995,
125) implies that all physical events (e.g. behavioral events) have physical
causes2. Given such causes, it appears that any putative non-physical men
tal cause is excluded from being the cause of the behavior. I shall argue that
in embracing an "excessive" view of reality, one that permits overdetermi
nation, we can defend pluralism. For those philosophers who take this to
be more a reductio ad absurdum of pluralism than a defense, I spend some
time arguing that overdetermination is not "metaphysically offensive". The

view to be defended, therefore, involves the following three claims:
Pluralism: Mental events (and other events) are numerically

distinct from physical events.
Supervenience: Mental events are (strongly) supervenient upon

physical events.3
Overdetermination: Both mental and physical events are causes

of behavioral (physical) events.
In section one I offer two interpretations of the exclusion argument.

The first version is an argument that can be constructed using Kim's
Explanatory Exclusion Principle. I shall argue that, under one interpreta
tion, this argument is consistent with the position defended above. That
would appear to show that Kim would be comfortable with the above
three claims. Not so, for Kim is a token (and type) physicalist. But the
exclusion argument as interpreted has no bite. In section two, I address a
second reading of the exclusion argument, one that focuses on overde
termination. I argue that causation is not necessarily exclusive, that it
is nomologically possible for an effect to have more than one sufficient
cause. In section three, I argue that to the extent we believe in interdisci
plinary causal explanations, we are committed to the existence (not just the
possibility) of overdetermination. Finally, in section four, I discuss overde
termination and counterfactuals. Here I defend the view against possible
objections. Overdetermination, combined with supervenience ("SOD" for
"supervenient overdetermination"), does not commit us to counterfactuals,
inconsistent with our explanatory commitments.
1. WHAT DOES THE EXCLUSION ARGUMENT EXCLUDE?
Kim (1987; reprinted 1993) introduces the Exclusion Principle (EP) as a

principle regarding causal explanation:
EP: "... there can be no more than a single and independent explanation of any one event
..."(Kim, 1987/1993,238).
Kim advocates explanatory realism: that token causal explanations require

the existence of some objective relation holding between the events. That
relation is the causal relation. A causal explanation of the form C explains
E (where 'C is a statement of the form 'event c occurred' and ? is a
statement of the form 'event e occurred') is true, only if the causal relation
obtains between c and e. The explanatory exclusion principle, along with

explanatory realism, makes plausible the ontological exclusion principle:
(1) OEP: If an event x is causally sufficient for an event y, then no event

jc*, distinct and independent from x is causally relevant to y.
This principle can then be used to form an argument against pluralism:
(1) OEP: If an event x is causally sufficient for an event y, then no event

jc*, distinct and independent from x is causally relevant to y.
(2) For every physical event y, some physical event x is causally sufficient
for y.
(3) For every physical event x and mental event jc*, jc is distinct and
independent from x*.
(4) Therefore: for every physical event y, no distinct and independent
mental event x* (distinct and independent of x) is causally relevant
to y.
Given the truth of (1) and (2) and the fact that (4) is unacceptable, unac
ceptable because it entails that no mental event is causally relevant to any
physical event, we are led to the view that we must reject (3). Pluralism, it
would seem, is false.
The first thing to note about this argument is that the target for the re
ductio is premise three, a claim regarding the independence of mental and
physical events. As stated, the exclusion argument works against "radical"
pluralists (and dualists like Descartes) who deny the dependence of mental
events upon physical events. This is not the version of pluralism that I wish
to defend. Rather, I am defending pluralism within the context of event
supervenience. Hence, this version of the exclusion argument is consistent
with pluralism and event supervenience. But pluralism is a more subtle
claim, one consistent with the dependence of mental events upon physical
events.
But this situation is rather puzzling. Kim seems to accept the possibility
that mental events are non-identical with physical events, thus permitting
a nonreductive event pluralism. Consider the following passage:
Cl is reducible to, or supervenient on, C2. This sort of relationship might obtain, for
example, on some accounts of the mind-body relation, which, though eschewing an out
right psychophysical identification, none the less recognizes the reductive or supervenient
dependency of the mental on the physical. In such a case, the causal relation involving the
supervenient or reduced event must itself be thought of as supervenient or reducible to the
causal or nomological relation involving the 'base' event. In this sense, the two explana
tions are not independent; for the one involving the reduced causal (that is, explanatory)
relation is dependent on the one representing the 'base' causal relation (Kim, 1987/1993,
240)
This passage is fairly explicit in permitting the existence of non
identical but supervenient mental events. Does this show that Kim is
comfortable with the position I am defending? Certainly not, for pluralism
is the claim that mental events are non-identical with the physical events
they supervene upon, and Kim is a token and type physicalist.4
What the above formulation of the argument reveals is that the exclu
sion argument does nothing, on its own, to attack the pluralism considered
here. Non-physical but supervenient events are not excluded from being
causes of the events their subvenient events also cause. This is rather
surprising, for others have certainly thought that the exclusion argument
was supposed to be an argument against this moderate form of pluralism.
Stephen Yablo reads the argument in the following way:
(1) If an event x is causally sufficient for an event y, then no event jc*,

distinct from jc is causally relevant to y.
(2) For every physical event y, some physical event jc is causally sufficient
for y.
(3) For every physical event x and mental event x*, x is distinct and
independent from jc*.
(4) Therefore: for every physical event y, no distinct mental event x* is
causally relevant to y.
Yablo then goes on to argue that ( 1 ) is false because there are relations be
tween mental and physical events such that the exclusion principle doesn't
apply to events so-related. Briefly and somewhat cryptically, Yablo holds
that supervenient events are "nested" in their subvenient events, in virtue
of the fact that the properties of the subvenient physical event are determi
nates of the determinable mental property. Such a relation between mental
and physical events (and the determinate/determinable relation between
physical and mental properties) is a dependency relation that captures
(event and property) supervenience.
Two things are worth pointing out here. First, Yablo simply doesn't
mention Kim's exclusion principle, for he fails to include in (1) the im
portant clause stating that the causes must be independent. Second, Yablo
proceeds to argue that when the causes are not independent (when the
properties of those events are related as determinate to determinable), there
is no problem of exclusion. But Kim's principle explicitly allows for this
kind of position. Given that Yablo concedes that mental events are super
veniently dependent upon physical events, he was never offering a view
inconsistent with Kim's stated exclusion principle.
The exclusion argument, by itself, is not an argument against those who

reject token physicalism, yet hold to supervenience. What then is the prob
lem with such a non-reductive metaphysical view? I think there are two
possible worries, the overdetermination problem and the causal powers
problem. The first concerns the fact that pluralism entails overdetermina
tion, albeit, overdetermination of an effect by causes that are themselves
not wholly independent of one another. In some passages Kim does seem to
think that overdetermination is a worry. In considering whether a property
instance, P* of jc at time t (an event in Kim's metaphysics) can be caused
by a physical property-instance P AND a mental property-instance M
(realized by P), he says:
Moreover, if we insist on m as a cause of P*, we run afoul of another serious difficulty,
"the problem of the causal-explanatory exclusion". For we would be allowing two distinct
sufficient causes, simultaneous with each other, of a single event. This makes the situation
look like one of causal overdetermination, which is absurd. (Kim, 1993, 354)
Here the problem appears to be the commitment to what "looks like"

overdetermination. Pluralism commits us to a form of overdetermination,
as such, it appears absurd. I think otherwise.
The second problem concerns how we are to understand the causal pow
ers of non-physical but supervenient events. Do supervenient events have
causal powers "over and above" their subvenient base events? If not, then
those properties that distinguish mental events from physical events are
non-causal properties. If so, then we need to know how this is consistent
with the closed nature of physical theory.
In the remainder of this paper I shall discuss only the first issue, that of
overdetermination, leaving the second for another paper. I shall argue that
overdetermination is not problematic for the pluralist.
2. EXCLUSION AND OVERDETERMINATION
In this section I shall introduce several versions of the exclusion argument.

Not all of these arguments can be attributed to Kim. The first version of the
argument is taken from Yablo ( 1992) (who is no defender of the argument).
I claim that this argument, and its revised form provides us little reason to
reject pluralism.
Second, I shall argue that causation is not necessarily "competitive"
and that overdetermination is physically possible. In the sections to follow
I shall argue that overdetermination is not merely possible, but ubiquitous.
(1) If an event x is causally sufficient for an event y, then no event jc*,
distinct from jc is causally relevant to y.
(2) For every physical event y, some physical event jc is causally sufficient
for y.
(3) For every physical event x and mental event jc*, x is distinct and
independent from jc*.
(4) Therefore: for every physical event y, no distinct mental event jc* is
Given the truth of (1) and (2) and the fact that (4) is unacceptable, unac
ceptable because it entails that no mental event is causally relevant to any
physical event, we are led to the view that we must reject (3).
I will argue that (1) is false, and that a weaker reading of (1) that makes
reference to overdetermination, is unmotivated. The weaker reading of (1)
is
(1*) If an event jc is causally sufficient for an event y, then unless

y is causally over/determined, no event jc*, distinct from x is
(1*), cannot be used to reject dualism, for overdetermination is common

place and metaphysically inoffensive. Being a cause for some event is not
a competition in which only one candidate can be the winner. As I shall
argue below, causal relations are not necessarily competitive; entirely suf
ficient causes of some particular effect do not, of necessity, exclude other
events from playing a causal role with respect to that effect. Second, our
commitment to supervenience and interdisciplinary causal explanations,
indicates, I think, that overdetermination is not merely possible, but actual
and ubiquitous.
My defense of pluralism and overdetermination involves an appeal
to the modal dependence of mental events upon their underlying sub
venient physical events. This dependence relationship is essential to the
metaphysical picture being defended. Mental events are non-identical to,
but supervenient upon, their underlying physical events. It is this fact
that enables us to see how causal overdetermination of a physical event
by both the supervenient and subvenient event, does not require revi
sionism regarding our explanatory practice, particularly with regard to
the counterfactuals that we take to be true in cases of mental causation.
Overdetermination is ubiquitous simply because mental events supervene
upon physical events.
Sally and Pedro are in competition for the gold medal. This entails that
Sally and Pedro cannot both receive the gold medal. Were both Sally and
Pedro able to receive a gold medal, then Sally and Pedro would not be
competing for the gold medal, although both are attempting to receive a
gold medal. Being in competition, is not to be confused with being in a
competition. (With this in mind I will use 'competing' to mean 'being in

competition with', not as 'being in a competition'). When a and b are in
competition, they are in competition for some prize, or at least, for the title
of 'winner'. Furthermore, when a and b are in competition, only one may
succeed. Thus, if a succeeds, then b is excluded from succeeding, and vice
versa.
Competition, in my sense of the word, permits only one winner. Some

activities are competitive in our restricted sense, like playing a game
of chess, in which only one player can win.6 Other activities are non
competitive, since multiple players may win. It is here that my use of the
word "competition" departs from everyday usage, which permits games
with more than one winner to count as competitive games e.g. when first
place is tied in a race. I use the concept of competition in my restricted
sense. Thus, horse racing, to the extent that it permits ties for first place, is
not competitive in my use of the word.
Furthermore, being competitive may be essential to one game, inessen
tial to other games. You must consult the practice of those who play the
game (or the codified rules of the game) to discover what kind of game
you are involved in. Obviously, not all games are competitive, accidentally
or essentially.7 The same epistemological point holds for theories of cau
sation and causal explanation. We cannot know a priori whether events are
causally competitive or non-competitive. Similarly, we shouldn't try to set
tle a priori whether causal statements are competitive or non-competitive,
and we shouldn't try to settle a priori whether causal explanations are
competitive or non-competitive. We must look to the practice of provid
ing causal claims and causal explanations to see whether they actually,
essentially, or accidentally involve competition.
Saying that events compete causally is to say that two events compete
for the title of 'cause' of a third event. Whether events compete causally
depends on what we take being a cause to involve. If causes are not nomo
logically sufficient for their effects, then distinct events need not compete.
Rather, there may be many causes of the effect; that is, many events may
succeed as cause. If causes are sufficient for their effects can many causes
be sufficient for a further event? If yes, then events do not compete for
sufficient cause; if not, then we have reason to believe that events compete
for sufficient cause. Competition, remember, entails exclusion; if e\ is a
sufficient cause of e2, and "sufficient cause" is a competitive relation, then
no further event, e^, can be a sufficient cause of ?2-8
The answer is clear. There are nomologically possible cases (and actual
cases) in which there is more than one event, causally sufficient for a fur
ther event. Two bullets simultaneously enter poor Pedro's heart, causing
his death. The firing of each bullet is sufficient, given relevant background
conditions, to kill Pedro. Such imaginary scenarios do not appear to vio
late any laws of nature, although they may not be common. Thus, there
are possible cases in which events are non-competitive with regard to be
ing sufficient cause. Thus, 'being a sufficient cause' is not a competitive
concept. If so, then it is false that, necessarily, if e\ is sufficient cause of
?2, then there is no further cause e^, that is sufficient for ?2- That is, the
exclusion principle (1) is false. Rather, it is nomologically possible for
e2 to have more than one distinct sufficient cause. The occurrence of ei
permits many winners in the causal-sufficiency game.
Perhaps, in defense of (1) we may weaken the letter of (1) but retain its
spirit. Admitting cases of overdetermination as special cases, as rare, meta
physical and scientific curiosities, is compatible with the general thrust of
the principle. Cases, such as those involving the unfortunate Pedro, will
be exceptions to the rule. It might be said, nevertheless, that the rule still
holds, despite these unusual cases. Overdetermination is rare.
Such a defense of (1) is plausible, but I argue below that even (1), ap
propriately re-interpreted is false. Certainly, cases such as Pedro's may be
hard to find. It does seem true that overdetermination of effects by events
from the same explanatory discourse, although nomologically possible, are
nevertheless hard to come up with. Overdetermination, however, is rarely
considered from the point of view of the various levels of explanation.
There may be many non-competitive causes for some effect, although we
may not find all these causes mentioned in the one science. We shall return
to this point below.
We have established that being a sufficient cause is not a competitive
concept. There are, no doubt, events which are causes that are competitive.
When this is so, it is not because these events are causes, but because of
some further property these events may possess. It may be that if e\ is a
sufficient cause of ^ then e^ cannot also be a sufficient cause of ei. This
fact, however, will not be traceable to the fact that they are sufficient causes
simplicitur. No doubt, when e\ does exclude e$ from being a sufficient
cause of e2, this is so in virtue of the causal properties possessed by e\
and e3. The sugar-cube's dissolution is caused by the cube's immersion in
water, and this excludes the cube's destruction from being caused by melt
ing from a flame of a relatively low heat, because the cube's immersion in
water is causally incompatible with the occurrence of the heat required to
melt a typical sugar-cube. Both causes cannot be simultaneously effective.
Put epistemologically, it is not that we already know a cause, suffi
cient for the effect, from which we conclude that the ice-cube was not
destroyed by melting. Rather, we already know a cause, sufficient for the
effect, the properties of which we know are incompatible with the proper
ties possessed by the postulated second cause.9 These properties may be
incompatible with the event being present, or they may be incompatible
with the event being a cause of the effect (perhaps because a required
background condition is prevented from occurring). But either way, we
cannot argue from sufficiency of cause, to there being no further fully
sufficient cause. There are nomologically possible cases in which an event
has more than one entirely sufficient cause. Having established the rather
weak claim that overdetermination is possible, I now turn to showing that
we have reason to believe in its actuality and relative ubiquity.
3. THE UBIQUITY OF OVERDETERMINATION
'Being a sufficient cause' is not a competitive concept, but how common is

the phenomena of non-competitive causes? Examples of non-competitive
sufficient causes (if causes are sufficient at all) are ubiquitous. They occur
in most cases in which one event supervenes upon another and both events
are causes of the same effect.10 For example, the sufficient cause of my
hand waving is my coming to possess a certain desire to do so, along with
my other background psychological states and the relevant psychologi
cal generalizations. My coming to have that desire supervenes on certain
complex neurophysiological events, taken together we can call them 'ep\
ep along with background conditions and the relevant neurophysiological
generalizations is, most plausibly, also causally sufficient for my hand
waving. Assuming that an explanation of an event consists in providing
the cause of that event under some suitable description, then both the psy
chologist and the neurophysiologist provide causes for the behavior when
they provide explanations for the behavior,11 and there is nothing special
about the example. We could construct similarly plausible examples for
any case in which an effect shares a supervenient cause with an underlying
subvenient cause. It is up to the sciences in question to tell us whether
the events they investigate, and hope to explain, are sufficient causes for
their effects. Given that 'being a sufficient cause' is a non-competitive
relation, we have no need to worry whether a science of events subvenient
to our favorite science claims to provide sufficient causes for the effect in
question. There are many winners in the game of discovering sufficient
causes.
But not everyone is happy with this view. How can we say that ment
events are sufficient for behavior and neurophysiological events are a
sufficient for the very same behavior? Isn't this simply overdeterminat
on a global scale?
First, we should commit ourselves to such claims only if we have gen

eralizations from the sciences that both causally explain the event in ques
tion. This is where epistemology enters our picture. If neurophysiology
and psychology both claim to have causal generalizations that subsume the
same effect, but which mention distinct causes, then we should take these
appearances seriously. The effect is overdetermined. So overdetermination
isn't necessarily global, however, it might well turn out to be, depending
on the future of interdisciplinary research. But what's the problem with
overdetermination? Kim (1990) says of the overdetermination option:
... we would be saying that B is causally overdetermined by the two events, since either
one singly would have been sufficient to bring about B. But again this seems like an absurd
picture of the situation; it implies, for example, that every bit of rationalizable behavior
is overdetermined by both a mental cause and a physical cause. This could also involve a
violation of the closed character of the physical domain. (Kim 1990/1995, 125)12
Coming from Kim this passage is surprising for, apart from the worry
regarding the closed character of the physical domain (to be dealt with
below) there is no forceful argument for the absurdity of such global
overdetermination. Furthermore, we should not jump to the conclusion that
overdetermination is global, unless we have adequate causal explanations
of the same event that mention distinct causes. But as we have seen, it is
nomologically possible that there is more than one sufficient cause for an
event. What is so absurd about this possibility being actual? That there are
supervenient events that appear, as Kim grants, to be causally sufficient for
their effects, shows that there is every reason to think that overdetermina
tion is actual and as commonplace as the success of interdisciplinary causal
explanations. If the actuality of overdetermination is a problem, even if it's
possibility is granted, then we need to be given an argument for that claim.
If we read the exclusion argument as involving:
(1*) If an event x is causally sufficient for event y, then unless

y is causally overdetermined, no event jc*, distinct from x is
then (4) must be modified to (4*):
(4*) Therefore, for every physical event y, unless y is overdeter

mined, no mental event x* is causally relevant to y.
But (4*) is not an absurd claim. If overdetermination is the normal sit

uation, as non-reductive compositional and interdisciplinary explanation
gives us every reason to believe, then such a conclusion is quite palatable.
The majority of supervenient events will not be excluded by (4*), hence
(4*) being of such limited scope, is perfectly acceptable. Overdetermina

tion, being commonplace, empties (4*) of any interest. (4*) is trivially true,
if overdetermination is commonplace. If (4*) is acceptable, then Kim has
failed to provide a reductio of one of the premises. Dualism is not, after
all, under threat by these considerations.
I suspect overdetermination has been considered a rare, or non-existent
phenomenon, because it looks hard to find actual cases in which the events
in question are from the same explanatory domain. There are few cases
where people are shot simultaneously. But if overdetermination is rare
when the events are from the same explanatory domain, it is ubiquitous
when the events are from distinct explanatory domains. If, as we have been
assuming, subvenient events (those events that constitute the supervenient
events) frequently cause the same events as their supervenient events, then
overdetermination is common and unobjectionable. If supervenient events
and subvenient events are never, or rarely, sufficient causes for the same
effect, then the problem for dualism does not arise. Either way, there is no
problem.
When the various sciences give causal explanations of the same effects,
the appearance of overdetermination we are left with should be respected.
These appearances require saving, not elimination or re-interpretation. The
massive appearance of overdetermination due to two or more theories
giving causal explanations of the same phenomena (and granting super
venience), is indicative of the fact that causation is a non-competitive
relation. It is the non-competitive nature of causation, along with the
supervenience of events on their subvenient events, that explains the
appearances.
4. OVERDETERMINATION AND COUNTERFACTUALS
Kim does raise a worry that is common among philosophers with physical
ist leanings. In the passage quoted above he says "This (overdetermination)
could also involve a violation of the closed character of the physical
domain". Kim does not expand on the point and it is not obvious how
overdetermination is incompatible with the closed character of the phys
ical. If physics is closed, then presumably every physical event has a
physical event as its cause, or every physical event is explainable by a
physical explanation. Global overdetermination is not in conflict with such
claims. Non-competitive overdetermination is the idea that there can be
more than one sufficient cause for an event, so it is perfectly consistent
with the idea that all physical events have physical causes; it is just that
they may have more causes beside. Believers in overdetermination and the
non-competitive nature of causation can hold on to the closed nature of

physics.
It is worth remarking that in Kim (1987) he takes it that cases of su
pervenience are not cases in which the causal relations of the events are
distinct. He says:
In such a case, the causal relation involving the supervenient or reduced event must itself be
thought of as supervenient or reducible to the causal or nomological relation involving the
'base' event. In this sense, the two explanations are not independent; for the one involving
the reduced causal (that is, explanatory) relation is dependent on the one representing the
'base' causal relation". (1987, 240).
If, due to its supervenience, the supervenient causal relation is identical

with the subvenient causal relation, then we lack overdetermination. But
this is just where pluralists and monists disagree regarding how super
venience is to be understood. The pluralist insists that supervenience of
events upon other events doesn't imply their numerical identity. So the
pluralist will say the same for causal relations. Unless the supervenient
dependence of one causal relation upon another entails their identity, we
are still left with a notion of overdetermination, albeit, one in which the
causes and causal relations are themselves dependent on underlying causes
and causal relations.
If there is a causal relation obtaining between two supervenient events,
we should see this causal relation as supervening upon the underlying
causal relation that obtains between the subvenient events. That is cor
rect. Furthermore, the supervenient causal relation is dependent on the
underlying causal relation only in that it must be present if the subvenient
causal relation is present. However, supervenient events can be multiply
constituted by various aggregates of subvenient events. There is no reason,
therefore, to doubt that supervenient causal relations are also multiply
realizable with regard to their subvenient causal relations. But it is just
such modal properties that pluralists seize upon to argue for ontological
irreducibility.13 Let me explain.
Causal relations have their identity conditions determined by the events
that are their relata. The difference between distinct token causal relations
is their relata; the events that instantiate the relation. Different events re
lated, different instance of the causal relation. There is nothing more to
a token instance of a causal relation than the fact that it is an instance
of a causal relation, and that it relates these token events. If we drop the
requirement that numerically identical instances of the causal relation re
quire the very same events as their relata, then it becomes mysterious as to
when we have the same causal relation surviving a change of relata, rather
than a new instance of the causal relation instantiated by distinct relata.
I suspect that all instances of relations, causal or not, possess their relata
essentially.14 If supervenient events can be multiply constituted by distinct
aggregates of subvenient events, then it follows that in such cases there
would be a distinct instance of the causal relation subvenient to the causal
relation obtaining at the supervenient level. The supervenient causal rela
tion can, in such cases, "survive" the counterfactual constitutive changes in
the events that constitute the relata of the relation. The relata remain, and so
too does the causal relation between them, even if, counterfactually, the ex
act composition of these relata were different. Overdetermination does not
disappear, when numerically distinct causal relations are superveniently
related.
Kim does not provide us with sufficient reason to reject overdetermi
nation however, overdetermination is not easy to understand. Consider
Pedro's death '?2', which is caused by two shootings 'e\ and '^3'. When
e\ and ?3 overdetermine ^ then we seem to be committed to the following
counterfactuals:
(Cl) Had e\ not occurred, then e2 would still have occurred
(C2) Had ?3 not occurred, then ?2 would still have occurred.
These counterfactuals are not in themselves problematic. They are exactly

what you would expect, given the situation. Had one of Pedro's killers
failed to fire his gun, Pedro would still be dead.
What is thought to be problematic is how these counterfactuals are
to be integrated into a theory of causation. Counterfactual theories of
causation will most obviously have difficulty in dealing with overdeter
mination. A crude counterfactual analysis would most likely require that
the effect be counterfactually dependent upon the cause such that: had the
cause been absent, the effect would have been absent. An analysis that
took this approach has to face cases of overdetermination as counterex
amples. Whether modified counterfactual analyses can accommodate such
counterfactuals is controversial (Lewis, 1986).
I do not wish to spend too much time on this problem, for the problem
stems from getting our methodology backwards. Metaphysical theories
of causation should take the phenomena of causation seriously. Although
some phenomena has to be discounted as "noise" in any discipline, I do
not think that the features of causation I am calling attention to can receive
such treatment. Although I do not have a completed theory of causation
up my sleeve, I am arguing that causation has at least these two features:
it is non-competitive and overdetermination occurs. A theory of causation
has to be consistent with these two features of causation. So, the coun
terfactuals that appear to be true in cases of overdetermination are not
the problem. Rather, these counterfactuals count as the data that a theory
of causation must accommodate and, perhaps, explain. Overdetermination
has been, to a great extent, ignored in much of the literature on causation.
The explanation for this is that overdetermination was thought to be rare, if
not simply a myth. I take overdetermination to be as common as interdisci
plinary causal explanations, for it is the success of these explanations that
lead me to postulate causes and causal relations. If the causes postulated
are distinct, and the events to be explained (by mentioning the causes)
are identical, then we have overdetermination. This should lead us to take
the phenomena of overdetermination more seriously in our metaphysical
theories of causation.
Nevertheless, there are still some nagging worries. For one SOD is
significantly different from the examples of overdetermination commonly
found in the literature. Pedro is shot simultaneously by two policemen; his
death is overdetermined by two distinct events. Had either one of the shoot
ings not occurred, the death would still have occurred. Typically, pairs of
counterfactuals like (C1 ) and (C2) (above) are true of such examples of
overdetermination (where e? and ?3 are shootings and ^2 Pedro's death).
However, in cases of SOD very different counterfactuals are true.
Consider a neurological event en and its supervenient mental event em.
Say we possess causal explanations that lead us to correctly believe that
both en and em are causes of some further behavioral event, e^. What coun
terfactuals are true of this situation? On analogy with Pedro-type cases of
overdetermination, we might expect the following:
(C3) Had en not occurred, then e?, would still have occurred.
(C4) Had em not occurred, then e^ would still have occurred.
Were SOD cases exactly analogous to cases like Pedro's shooting, then we
would expect (C3) and (C4) to be true in cases where both the mental and
neurophysiological events are causes of the same behavioral event-token.
However, the situation is not so simple. Were we to ignore the fact that em
supervenes upon eflm then (C3) and (C4), would seem to be true. Taking
account of supervenience is not easy, but it would appear that in virtue of
supervenience we must, in fact, accept the following counterfactual:
(C5) Had em not occurred, then en would not have occurred.
(C5) must be true, if em supervenes on en. Absence of the supervenient

event would require the absence of the subvenient event. Furthermore, we
need to recognize that the subvenient neurophysiological event necessi
tates, in accordance with supervenience, the existence of the mental event
(and multiple constitution will add a further twist to the situation, but I shall
ignore this further twist until later). Taking these features into considera
tion, the picture of overdetermination we possess is significantly different
from Pedro-type examples. After taking supervenience into consideration,
it looks as if we are not committed to the truth of (C3) and (C4) taken to
gether. Had em not occurred, neither would en, so it looks as if e^ would not
occur after-all (for now it appears that there is no second cause sufficient
to bring it about). Similarly, had en not occurred, then neither would em,
and hence it looks as if ey, would again lack a second cause, so ey, would
not occur.15
These considerations should be welcomed. Counterfactuals such as:
(C6) Had em not occurred, then e^ would not have occurred,
and
(C7) Had en not occurred, then ?/7 would not have occurred
are counterfactuals that we are currently committed to in our explanatory

practice; even though we do not yet know whether there are dual causes
(superveniently related) for the very same behavioral event. That is, we
commit ourselves to claims such as: Had I not believed that budget cuts
are morally wrong, I would not have voted against them. We typically
hold that such counterfactuals are true, while recognizing the fact that
epistemically, overdetermination is still a possibility. What the above con
siderations indicate, is that when we "add in" the information that the
events are superveniently related, and the events are common causes of
the very same behavioral effect, the counterfactuals do not "change" truth
value. That is, overdetermination and the supervenience of one cause upon
the other, preserves the counterfactuals that we quite reasonably commit
ourselves to in ignorance of the facts regarding overdetermination. Dis
covering, through our causal-explanatory practice, that we are committed
to SOD will not require us to radically revise our prior claims regarding
the counterfactual dependence of the behavioral event upon the mental
event (and of the dependence of the behavioral event upon the neuro
physiological event). This is a surprising and welcome outcome. If such
interdisciplinary explanations are forthcoming, then within a dualist meta
physics, we will not need to revise the counterfactual claims that we are
currently committed to in our everyday explanatory practice. Within this
metaphysical and explanatory framework we can still make claims like:
Had I not believed that budget cuts are immoral, I would not have voted
against them. Supervenient overdetermination does not require us to make
unpleasant, and in practice, very difficult revisions to our way of thinking

regarding the dependence of behavioral events upon mental events. This
feature of the metaphysical picture that I advocate makes the metaphysical
picture highly attractive.
But there is a prima facie difficulty that arises from these considera
tions. If we do not have counterfactuals like (C3) and (C4), counterfactuals
that we find in Pedro-type examples of overdetermination, then are we
really speaking of overdetermination? If we are not, so this worry goes,
then do we really have a reply to the exclusion argument?
These questions are fair, but I don't think they raise a serious prob
lem with the account. "Overdetermination" is a technical concept to some
extent, but at the core of the concept is the idea that an effect has two
(numerically distinct) sufficient causes. SOD satisfies that description;
there are two numerically distinct causes: the supervenient and subvenient
events. Thus, SOD really is a case, albeit one not frequently considered,
of overdetermination. Those who have difficulty accepting SOD as gen
uine overdetermination, focus on the counterfactuals that are true in virtue
of the overdetermination. But different counterfactuals may be true de
pending on the details of the overdetermination cases. Overdetermination
is, paradigmatically, a concept about causation, not a concept about the
counterfactuals true in virtue of the causal facts.
Pedro-type cases and SOD cases are two kinds of overdetermination,
but there are more. Consider this kind of case: Two events e* and ?** are
the effects of a common cause ec. Thus
(C8) Had e* not occurred, then <?** would not have occurred
is true, in virtue of their common cause.16 Now add to this scenario in

formation regarding the effects of e* and ?**. e* and e** are both causes
of some one effect, e^, thus e?\s overdetermined by ?* and e**, (that are
themselves the product of the common cause ec). Now, in this situation the
effect, e*, is overdetermined; but had either of the overdetermining causes
been absent, the effect would not have occurred. The counterfactuals true
of examples like this are very similar to the counterfactuals true of SOD
cases. But the cases are different. In SOD cases the overdetermining causes
of the effect are related by supervenience. In the case above, the overde
termining causes are related by a further cause, but they are not related
by supervenience. That they are similar should come as no surprise, for
supervenience does much of the same work as a common cause in keeping
the two overdetermining causes "together".
What these considerations show, is that there are many possible ways in
which an event could be overdetermined. That is, an effect can have more
than one sufficient cause, and there are various possibilities regarding how,
if at all, those overdetermining causes may be related. If the numerically
distinct causes are related in some such way, should we refrain from calling
these causes overdetermining causes? I don't see why we must, they are
both sufficient causes, and that warrants the description 'overdetermining
causes'. What counterfactuals are true of a particular case of overdetermi
nation will depend on the further facts regarding the relations that obtain
(if any) between the causes.
There is one further feature of SOD examples that is worth mentioning.
Supervenience permits the supervenient event to be multiply constituted
by subvenient events. For example, I may have the very same belief-state17
token in counterfactual scenarios in which that state is constituted by a dif
ferent neurophysiological state. Furthermore, a belief-state token may well
survive replacement, over time, of its underlying neurophysiological states
(my brain states change relatively quickly compared to my psychological
states). These features of event supervenience add a twist to the account of
counterfactuals of SOD examples. If we consider the phenomena of multi
ple constitution, we can see that examples of SOD are a little more closely
related to examples like that of Pedro's death, than we may have initially
thought. In Pedro-type overdetermination cases, we take it that, had either
shooting not occurred, Pedro would still be dead. In SOD cases, we have
the possibility, permitted by multiple constitution, that the absence of the
subvenient event is consistent with the presence of the effect. Consider
these counterfactuals a second time:
(C6) Had em not occurred, then e?? would not have occurred.
(C7) Had en not occurred, then e^ would not have occurred.
Above I argued that these counterfactuals are true of many SOD cases,
and so they are. However, there are cases of SOD in which (C7) may be
false. Such an example draws upon the possibility of multiple realizability
or constitution. Here is how (C7) may turn out false in such a "special"
case of SOD. In the actual world em causes e?7 and is constituted by, en. In
the closest possible world in which en does not occur, em is constituted by
some further neurophysiological event. Thus, counterfactual
(C7) Had en not occurred, then e^ would not have occurred
would be false, because em is still a sufficient cause for e??. Had I lacked
the neurophysiological state that constitutes my desire to run for the bus, I
would nevertheless run for the bus, for my desire would be constituted by
some further neurophysiological state.
What this shows is that some cases of SOD will look much more like
Pedro-type cases than we initially thought; and that should help quell wor
ries about whether these cases are sufficiently similar to the Pedro-type
cases to count as genuine overdetermination. However, SOD examples are
genuine examples of overdetermination because they are cases in which
some one effect has two distinct sufficient causes.
How often are cases of SOD like the one above? That is, how common
is it, that in the closest worlds in which the neurophysiological event is
absent, there is another to take its place? That is difficult to say, for we
would have to go through the examples one by one. I suspect that they
are common, but I shall not attempt to show this here. Suffice to say,
overdetermination is as common as are our successful, interdisciplinary
causal-explanatory practice leads us to believe.
5. SUMMARY AND CONCLUSION
In this paper I have argued that Kim's exclusion principle misses its mark
and that the weakened version of that principle is also no threat to event
pluralism. Furthermore, I have urged that overdetermination is not to be
seen as a metaphysical anomaly, but as a commonplace feature of the
world, and that this is not a problem for a metaphysical theory, since
the very possibility of overdetermination shows that causation is a non
competitive concept. This is, I take it, the important point. There can be
many sufficient causes for a single event. Second, I argued that overde
termination is not merely possible, but that we have reason to think it
is actual and ubiquitous. Overdetermination adequately accounts for the
many causal explanations we can provide for a single event.18 To the extent
that interdisciplinary causal explanations attribute distinct causes to the
same effect, (as they appear to) they offer us prima facie reason to take
the pluralist ontology seriously. These two features, taken together, may
provide challenges to a theory of causation, for they are part of the data,
part of the phenomena that a theory of causation must capture.
Overdetermination has usually been considered by offering examples
of events from the same explanatory domain: two doses of poison, two
deadly bullets and so forth. Cases like these are rare. But when we re
flect on how distinct sciences, (where one supervenes on the other, and in
which each science postulates numerically distinct events), attempt to give
explanations of the same phenomena, we can see that overdetermination is
common.19
Finally, although I have not provided thorough arguments for pluralism,
I have provided a defense of pluralism from the exclusion argument. There
are large issues remaining: what are those arguments for the rejection of
token physicalism? How should we understand the relationship between
subvenient and supervenient events, when these token events are non
identical? To the last question, I would say that physical events are the
parts of mental events. To the first question, I would answer with reference
to the second: part/whole relations are non-extensional, thus the whole is
not identical with the sum of its parts because they lack the very same
modal properties. Clearly, these issues must be left for another day.
NOTES
* Many thanks to the anonymous referees of Synthese for their critical comments on an
earlier draft of this paper.
1 It will become evident that a counterfactual analysis of causation will not be compatible
with my defense of overdetermination. Although some philosophers (Swain, 1978) have
tried to accommodate overdetermination within a counterfactual approach to causation, I
am not confident of their success.
2 I am granting, for simplicity, that behavioral events are correctly characterized as
physical, but this is by no means clear, given their intentional characterization.
3 Following Kim (1984) e* supervenes on e, if property F of e* supervenes on property
G oie.
4 Kim holds that mental properties are identical with their subvenient properties. Re
garding reductionism, Kim thinks that we can reduce human mental properties to human
physiological properties, while remaining skeptical regarding the possibility of reducing
mental properties generally.
6 Competition is compatible with both players failing to win. In chess, that is stalemate or
a draw.
7 Here I think of imaginative role playing in which children mimic, re-interpret, or invent
adult social relations. Given the imbalances in social roles, there is, I suppose, a sense in
which there are winners and losers in such games.
8 Someone may think that the principle should be interpreted as metaphysically necessary
since they may think that it is essentially constitutive of our concept of causality. If it is
constitutive, then cases of over determination appear to be counter-examples to the very
existence of causality. While these cases, or the possibility of such cases, cause headaches
for our theories of causality, we would be better advised to revised our concept than jettison
it. The concept is too embedded in practice and theory to be rejected without some very
fancy replacement.
9 To be more precise we should include the properties along with background conditions
and the relevant laws of nature.
10 Perhaps not in all cases, if we are willing to grant the nomological possibility that
events may be incompatible with their subvenient or supervenient events being causes.
For instance, my desires may be supervenient upon my brain state, but the properties of
certain chemicals may prevent those desires from being causes of behavior, e.g. when I am
intoxicated. Furthermore, there is a general question regarding "interactionism". Mental
events often cause and are caused by events that are not obviously physical; that is, they
often cause events that are best understood as events from domains other than physics.
For example, "Watching pornography made John react less sensitively to female subjects"
may be a supposed causal explanation, but we would be hard-pressed to reduce the events
in question to physical events, despite those events involving middle-sized objects. A form
of Parallelism may be more common than philosophers have recently allowed.
11 It is important to remember that some causal explanations may mention properties of
causes. Other causal explanations take the form of stating that some event is the cause of
the effect (and that cause is picked out under some description). It is the latter form of
causal explanation that I am concerned with, not the former.
12 Note that the absurdity does not lie in the idea that the effect is rationalized and caused,
for few people object to the idea that reasons can be causes. Rather, the worry is that there
are two causes, one of which can rationalize the behavior, another which cannot. This is,
however, the pluralist position I am defending.
13 Whether multiple constitution is compatible with (epistemological or theoretical) re
ductionism is not to the point here. The pluralist is more daring and ambitious. Multiple
constitution offers modal reasons for ontological pluralism.
14 Consider instances of "jc is taller than y" (and let us ignore nominalist worries). "A is
taller than B", picks out a different instance of Taller than, than that expressed in "A is
taller than C". The property is identical in both cases, but the property-instances are not.
15 As I shall discuss below, this will not always be the case. Multiple realizability allows
us, in some circumstances, to recognize that the absence of the neurophysiological event
may not make a difference to the effect being present, for the supervenient event is realized
by another neurophysiological event and remains the second cause.
16 All things being equal.
17 Alio wing states to count as events.
18 Explanations that explain by citing causes, not one's that merely cite the properties of
causes.
As common as such interdisciplinary explanations are successful.
REFERENCES
B?rge, T.: 1986, 'Individualism and Psychology', Philosophical Review 95, 3-46.
B?rge, T.: 1979, individualism and the Mental', Midwest Studies in Philosophy Vol. IV. P.
French (ed.), University of Minnesota Press, Minnesota.
Cartwright, N.: 1979, How the Laws of Physics Lie, Clarendon Press, Oxford.
Davidson, D.: 1993, Thinking Causes', in Heil (ed.), Mental Causation, Clarendon Press,
Oxford.
Davidson, D.: 1970, 'Mental Events', in L. Foster (ed.), Experience and Theory, University
of Massachusetts Press, MA.
Davidson, D.: 1963 'Actions, Reasons and Causes', Journal of Philosophy 60, 685-700.
Dupre, J.: 1993, The Disorder of Things; Metaphysical Foundations of the Disunity of
Science, Harvard University Press, Cambridge, MA.
Jackson, F and Pettit, P.: 1990, 'Program Explanation: A General Approach', Analysis
50(2), 113-115.
Jackson, F and Pettit, P.: 1988a, 'Functionalism and Broad Content', Mind97, 381-400.
Johnston, M.: 1990, 'Constitution is Not Identity', MindXClX, 89-105.
Kim, J.: 1990, 'Explanatory Exclusion and the Problem of Mental Causation', reprinted in
Macdonald and Macdonald, 1995, Cambridge, MA.
Kim, J.: 1987, 'Explanatory Realism, Causal Realism, and Explanatory Exclusion',
Midwest Studies in Philosophy 12, 225-239.
Kim, J.: 1984, 'Epiphenomenal and Supervenient Causation', in Rosenthal D. M. (ed.), The
Nature of Mind, Oxford University Press, Oxford and New York, 1991, pp. 257-265.
Lewis, D.: 1986, "Causation" and "Postscript" in Philosophical Papers Vol. II. Oxford
University Press, Oxford.
Macdonald, C. and Macdonald G.: 1995, Philosophy of Psychology; Debates on Psycho
logical Explanation, Basil Blackwell, Oxford UK and Cambridge MA.
Rudder-Baker, L.: 1995, Explaining Attitudes, Cambridge University Press, Cambridge.
Schiffer, S.: 1987, Remnants of Meaning, MIT Press, Cambridge, MA.
Swain, M.: 1978, 'A Counterfactual Analysis of Event Causation', Philosophical Studies
34, 1-19.
Department of Philosophy
McGill University
Montreal, Quebec
Canada
bgarrett@yorku.ca

Overdet 3

Uploaded by

Copyright:

Available Formats

Overdet 3

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Overdet 3

Uploaded by

Copyright:

Available Formats

Pluralism, Causation and Overdetermination

Author(s): Brian Jonathan Garrett

PLURALISM, CAUSATION AND OVERDETERMINATION*

"This causal superfluousness is hard to believe in; it is

Metaphors such as Stephen Schiffer's have a strong hold on philosophical

^M Synthese 116: 355-378, 1998.

Token physicalism can be approached from a different angle. God

(B) Causation is nomological, i.e. causal relations between events

(C) Psychological predicates (properties) are anomalous.

Furthermore, as Cartwright (1983) has argued, whether there are excep

time arguing that overdetermination is not "metaphysically offensive". The

Pluralism: Mental events (and other events) are numerically

Supervenience: Mental events are (strongly) supervenient upon

Overdetermination: Both mental and physical events are causes

In section one I offer two interpretations of the exclusion argument.

1. WHAT DOES THE EXCLUSION ARGUMENT EXCLUDE?

Kim (1987; reprinted 1993) introduces the Exclusion Principle (EP) as a

Kim advocates explanatory realism: that token causal explanations require

obtains between c and e. The explanatory exclusion principle, along with

(1) OEP: If an event x is causally sufficient for an event y, then no event

This principle can then be used to form an argument against pluralism:

(1) OEP: If an event x is causally sufficient for an event y, then no event

(1) If an event x is causally sufficient for an event y, then no event jc*,

The exclusion argument, by itself, is not an argument against those who

Here the problem appears to be the commitment to what "looks like"

2. EXCLUSION AND OVERDETERMINATION

In this section I shall introduce several versions of the exclusion argument.

(1*) If an event jc is causally sufficient for an event y, then unless

(1*), cannot be used to reject dualism, for overdetermination is common

competition. (With this in mind I will use 'competing' to mean 'being in

Competition, in my sense of the word, permits only one winner. Some

3. THE UBIQUITY OF OVERDETERMINATION

'Being a sufficient cause' is not a competitive concept, but how common is

First, we should commit ourselves to such claims only if we have gen

(1*) If an event x is causally sufficient for event y, then unless

then (4) must be modified to (4*):

(4*) Therefore, for every physical event y, unless y is overdeter

But (4*) is not an absurd claim. If overdetermination is the normal sit

(4*) being of such limited scope, is perfectly acceptable. Overdetermina

4. OVERDETERMINATION AND COUNTERFACTUALS

non-competitive nature of causation can hold on to the closed nature of

If, due to its supervenience, the supervenient causal relation is identical

(Cl) Had e\ not occurred, then e2 would still have occurred

(C2) Had ?3 not occurred, then ?2 would still have occurred.

These counterfactuals are not in themselves problematic. They are exactly

(C4) Had em not occurred, then e^ would still have occurred.

(C5) Had em not occurred, then en would not have occurred.

(C5) must be true, if em supervenes on en. Absence of the supervenient

(C6) Had em not occurred, then e^ would not have occurred,

are counterfactuals that we are currently committed to in our explanatory

unpleasant, and in practice, very difficult revisions to our way of thinking

is true, in virtue of their common cause.16 Now add to this scenario in

(C7) Had en not occurred, then e^ would not have occurred.

(C7) Had en not occurred, then e^ would not have occurred

5. SUMMARY AND CONCLUSION

As common as such interdisciplinary explanations are successful.

You might also like