A CRITICISM OF NONREDUCTIVE PHYSICALISM by Jonathan Bartlett
Causal Relevance and Nonreductive Physicalism
-
Upload
jonathan-barrett -
Category
Documents
-
view
217 -
download
0
Transcript of Causal Relevance and Nonreductive Physicalism
![Page 1: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/1.jpg)
Causal Relevance and Nonreductive PhysicalismAuthor(s): Jonathan BarrettSource: Erkenntnis (1975-), Vol. 42, No. 3 (May, 1995), pp. 339-362Published by: SpringerStable URL: http://www.jstor.org/stable/20012626 .
Accessed: 21/06/2014 23:22
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
Springer is collaborating with JSTOR to digitize, preserve and extend access to Erkenntnis (1975-).
http://www.jstor.org
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 2: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/2.jpg)
JONATHAN BARRETT
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM
ABSTRACT. It has been argued that nonreductive physicalism leads to epiphenominal ism about mental properties: the view that mental events cannot cause behavioral effects
by virtue of their mental properties. Recently, attempts have been made to develop accounts of causal relevance for irreducible properties to show that mental properties need not be epiphenomenal. In this paper, I primarily discuss the account of Frank
Jackson and Philip Pettit. I show how it can be developed to meet several obvious
objections and to capture our intuitive conception of degrees of causal relevance. How?
ever, I argue that the account requires large-scale miraculous coincidence for there to be
causally relevant mental properties. I also argue that the same problem arises for two
apparently very different accounts of causal relevance. I suggest that this result does not
show that these accounts, on appropriate readings, are false. Therefore, I tentatively conclude that we have reason to believe that irreducible mental properties are causally irrelevant. Moreover, given that there is at least prima facie evidence that mental proper? ties can be causally relevant, my conclusion casts doubt on nonreductive physicalist theories of mental properties.
1. INTRODUCTION
Nonreductive physicalism is a widely held theory of the relation be? tween the mental and the physical. Its advocates seek a realist and
physicalist account of mental properties, while holding that most are
irreducible to, but in some way dependent upon, physical properties. They face two main problems. First, while eschewing reduction, can the nonreductive physicalist tie the mental closely enough to the physical in order that her theory may be counted as a genuinely physicalist theory? Second, given the claim that all causal phenomena are fundamentally physical, can the nonreductive physicalist secure our conviction that
mental events can cause physical events, notably behavior, by virtue of their mental properties? The latter, the problem of the seeming causal irrelevance of irreducible mental properties, is the topic of this paper.
I begin by making clearer the major commitments of nonreductive
physicalism and argue that irreducible mental properties, if real, cannot be causally relevant to behavior. I suggest a simple account of causal relevance as a solution, but I show it to be inadequate. Second, I argue
Erkenntnis 42: 339-362, 1995.
? 1995 Kluwer Academic Publishers. Printed in the Netherlands.
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 3: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/3.jpg)
340 JONATHAN BARRETT
that a related account, due to Frank Jackson and Philip Pettit, solves
the difficulty for the simple account. I develop this account into a fuller
account of causal relevance that accomodates our intuitions about when
particular types of properties are causally relevant. However, I go onto
argue that the Jackson-Pettit account leaves unexplained how distinct
physical properties could each realize a single mental property and
produce a behavioral effect of a particular kind. Finally, I show that
two alternative accounts of causal relevance fall victim to the same
problem. I tentatively conclude that nonreductive physicalism is com?
mitted to causally irrelevant mental properties and hence, on the as?
sumption that mental properties can in fact be causally relevant, that
there is reason to doubt nonreductive physicalist theories of mental
properties.
2. NONREDUCTIVE PHYSICALISM AND THE ARGUMENT
FOR EPIPHENOMENALISM
Nonreductive physicalism arises from the seeming failure of attempts to
show how mental properties could reduce to physical and/or functional
properties.1 If one is committed to physicalism and one wishes to
resist anti-realist (in the Dummett-Putnam sense) theories of mental
properties, then one has to choose between eliminativism about the
mental and a weaker physicalism that settles for something less than
reduction. Nonreductive physicalists take the latter route.
There is no one nonreductive physicalism. However, all nonreductive
physicalists are committed to three claims. The first two are as follows:
(1) Token Identity of Mental and Physical Events. Each mental
event token is identical to some physical event token.
(2) Property Dualism. In general, mental properties are distinct
from physical and/or functional properties.
By themselves, (1) and (2) are too weak to count as a form of physi?
calism, which requires that everything, all objects, events, processes and properties, are at bottom physical; they are either physical or are
in some way composed out of or dependent upon the physical. To meet
this demand, nonreductive physicalists add a third component. Here
they often diverge from one another.2 However, for present purposes we can operate with a very broad statement of this part of their ac?
counts, one which almost all nonreductive physicalists would accept:
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 4: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/4.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 341
(3) Priority of the Physical. Each mental property is either iden?
tical with a particular physical property or else is realized by some class of physical properties (with the emphasis firmly on the latter).
I use the vague term 'realized' deliberately, as a blanket term covering all nonreductive physicalisms.3 The most discussed realization relations are the many types of supervenience. However, since my discussion
applies to any nonreductive physicalism holding (l)-(3), it is unnecess?
ary for us to get into specific details. I will just assume that there is
an appropriate relation of realization, one strong enough to sustain
physicalism and at the same time weak enough to allow for the existence
of mental properties. Typically, the nonreductive physicalist will main?
tain that a given mental property M will be instantiated whenever one
of a class, perhaps an infinite class, of distinct physical properties Pt,. . . ,Pn is itself instantiated. Moreover, no guarantee is given that
these physical properties will be simple properties. Indeed, it is stan?
dardly assumed that they will not be. These physical properties are the
possible realizers of M and when one is instantiated at some place and
time it is the realizer of M at that place and time.
I should add that nonreductive physicalism is not a doctrine that only concerns the relation between the mental and the physical. Nonreduc?
tive physicalist theories have been defended for other kinds of proper? ties. Amongst these are moral properties, color properties, social pro?
perties and biological properties. My main concern will be mental
properties. However, I shall mention the possible application of my conclusions to other cases at various points.
We now turn to the causal relevance challenge. It is widely accepted that mental events can cause physical behavioral events. My believing that there is beer in the fridge causes me to open the fridge and reach inside. However, we want to say more than this. In addition, we are
inclined to think that my believing that there is beer in the fridge causes
my opening the fridge because it was my believing that there was beer in the fridge and not say, because it was the thought I had exactly two
minutes after the oldest camel in the world died.4 That is, sometimes a mental property, one instantiated in a mental cause, is required to
explain why that mental cause produced a certain physical effect. If so, mental properties must be capable of being causally relevant to be
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 5: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/5.jpg)
342 JONATHAN BARRETT
havior. This commitment, it is argued, cannot be secured by nonreduc?
tive physicalism. The argument closely parallels one directed against event dualism
the claim that mental and physical events are distinct. I begin with this
argument, as I consider one of the moves made in response to it in my discussion of the causal relevance objection to nonreductive physi? calism. It runs as follows:
(A) The Closure of Physical Causation. Every physical event is
caused by a physical event.
(B) If mental events are distinct from physical events, mental
events cannot cause behavior.
(C) Mental events can cause behavior.
(D) So, mental events are not distinct from physical events.
Putting aside the unattractive epiphenomenalist reply, the event dualist
might argue that (A), on its most plausible reading, allows that a
physical event can have both a physical and a distinct mental cause
(causal overdetermination). In cases where we think that a mental cause
is operative, such as in the production of behavioral events, we need
only maintain that the particular effect is causally overdetermined by a mental cause and a distinct physical cause to secure mental causation.
So, by denying that (A) entails (B), we block the route to (D), even
while accepting (A) and (C). The normal replies to the overdetermination move are as follows.
First, while overdetermination allows for mental causes, they are never
causally necessary for behavioral events. Had the mental event not
occurred, there still would have been a physical event that would have
been causally sufficient to produce the behavior. However, it seems
mental events often are causally necessary for physical events. If I want
some beer and so move to fridge and remove some, we ordinarily think
that had I not decided that I wanted beer, I would not have behaved
as I did. An event dualist opting for overdetermination cannot confirm
this, except by holding that the occurrence of the relevant physical event
necessitates the occurrence of the particular mental event. However, it
seems very unlikely that our psychology is so bizarrely causally con?
figured. The second problem with overdetermination is that it violates
the plausible methodological principle that the universe is causally con?
servative: events generally do not have more causes than are necessary. This principle requires us to reject hypotheses that postulate widespread
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 6: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/6.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 343
causal overdetermination. Event dualism plus overdetermination is such
a hypothesis. As stated earlier, there is a close parallel between the causation
argument against event dualism and the causal relevance argument
against nonreductive physicalism. We shall be able to exploit this fact
later when we come to see what responses we can make on behalf
of the nonreductive physicalist. The argument against nonreductive
physicalism runs as follows:5
(A) The Closure of Physical Causal Relevance. Physical causes
produce effects by virtue of their physical properties.
(B) Since mental events are physical events, if mental properties are distinct from physical properties, they cannot be causally relevant to behavior.
(C) Mental properties sometimes are causally relevant to be?
havior.
(D) So, mental properties cannot be distinct from physical pro?
perties (hence property dualism and nonreductive physi? calism are false).
Following the event dualist, the nonreductive physicalist might object that (A), on its most plausible interpretation, does not require that
there is just one causally relevant property per cause-effect relation. If we reject that assumption, we cannot get to (D) so easily. The nonre?
ductive physicalist can hold that when mental event c causes behavioral
event e, not only is some physical property P of c's causally relevant
to c's causing e, but so is some mental property M of c's. This is not
overdetermination at the level of causes or events, but at the level of
causally relevant properties. One property is physical and the other is
mental.
It is natural to respond as we did to the event dualist. First, mental
properties are not just causally relevant but are causally necessary for a mental event's producing a behavioral effect. Second, not only should
we believe superfluous causes are rare, but by similar reasoning, we
should believe that superfluous causally relevant properties are rare
too.6 However, these objections succeed only if nonreductive physi? calism really requires property-level overdetermination. It is unclear that it does, as long as the cause has the causally relevant mental
property M by virtue of the same facts F as it has the causally relevant
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 7: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/7.jpg)
344 JONATHAN BARRETT
physical property P. For then, on the relevant reading, the following counterfactuals are true:
Had c not been P (and so F did not obtain), c would not
have been M (because F did not obtain). Had c not been M (and so F did not obtain), c would not
have been P (because F did not obtain).
There is not then overdetermination at the level of causally relevant
properties because, on the relevant reading, the following counterf ac?
tuals are false:
Had c not been P, c would have caused A.
Had c not been M, c would have caused A.
We might object here that the nonreductive physicalist assumes without
warrant that c is P and c is M in virtue of the same facts F. However, claim (3) of nonreductive physicalism points to a solution. It tells us
that c's being M on some occasion is realized by c's having some
physical property. If this is property P, then c is P and c is M in virtue
of the same facts. This is not a full solution however, because we can
ask why we ought to believe that c's being P both realizes c's being M
and is causally relevant to c's causing A. I take it that the nonreductive
physicalist can currently offer no answer. However, in this respect, she
is in the same position as the reductionist. The nonreductive physicalist maintains that c's being M is causally relevant to c's causing some
effect, because the physical realizer of M in c is causally relevant to
the effect. The reductionist differs only in adding that the causally relevant physical property is identical to M. In each case, whether there
is a physical property that fits the bill is a contingent matter; neither
can rule out epiphenomenalism purely by a priori means.
So, the nonreductive physicalist is in a stronger position than the
event dualist. She can avoid overdetermination by making the following kind of claim (from here I use the term 'causally-relevantphy' to indicate
the uncontroversial causal relevance of physical properties and reserve
'causally relevant' for the alleged causal relevance of some irreducible
properties):
Mental property M of c is causally relevant to c's causing A
iff
(a) for some physical property P, c's being P realizes c's
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 8: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/8.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 345
being M and c's being P is causally-relevantphy to c's causing A.7
However, the proposed condition is not sufficient for causal relevance.
Suppose a single physical property of an event realizes each of two
distinct irreducible properties. According to our condition, both irred?
ucible properties are causally relevant to any effect to which their
physical realizer is causally-relevantphy, whereas intuitively we might
only count one of those irreducible properties as the properly causally relevant one. Frank Jackson and Philip Pettit offer the following exam?
ple.8 Suppose Mary lets her aluminum ladder touch the power line.
She is electrocuted and dies. Her ladder's touching the power line
caused her death, because it was a good conductor of electricity. The
physical property actually realizing this in Mary's ladder, its being made
of aluminum, realizes other properties of her ladder as well - its having a certain thermal conductivity, ductility, luster and so on. According to our condition, each of these properties is causally relevant to her
death too. Clearly, we do not want to say this. Therefore, the simple condition introduced is not sufficient for causal relevance.
3 . PROGRAM EXPLANATION
Having identified this problem for the simple account, Jackson and
Pettit have tried to overcome it by developing an account of program
explanation.9 They count an irreducible property as causally relevant to some effect, when there is "invariance of effect under variation of
realization."10 We can see how their account relates to the simple account by stating it as follows:
c's being M is causally relevant to c's causing A iff
(a) for any P, if P is a possible realizer of M, then had c
been P, c's being P would be causally-relevantphy to c's
causing^.
With this account we can dispose of the objection to our initial account.
Returning to Mary's ladder, no matter how her ladder's being a good electrical conductor is realized, that realizer is causally-relevantphy to her ladder's touching the power lines causing her death. However, some possible realizers of her ladder's being a good heat conductor
would not be causally-relevantphy to her death. These are those possible
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 9: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/9.jpg)
346 JONATHAN BARRETT
realizers that do not in addition realize her ladder's being a good electrical conductor. Therefore, only electrical conductivity and not
thermal conductivity are causally relevant to the effect in question. Moreover, the Jackson-Pettit account captures one of the main points of casting explanations in terms of irreducible properties, namely that
doing so allows us to appeal to lawlike generalizations that in many cases are missing at the physical level.
Despite these features, the Jackson-Pettit account needs fine tuning. I will begin with this and then look to more serious difficulties. First, consider cases in which we over-specify the causally relevant property, ones where c's being M is causally relevant to c's causing e, yet we cite
c's being M and A as causally relevant, even though A has no essential
role to play in explaining c's causing e. Suppose Strawberry hits the
ball 410 feet through center field, causing the crowd to go wild. His
swing's being a hitting of a home run explains its causing the crowd's
going wild, let us suppose. It did not matter precisely how he hit a
home run. Now, if I point to his hitting the ball 410 feet, along a certain
parabolic trajectory, at an angle of say 7? to a line running through home plate and the mound, as the causally relevant property, rather
than just his hitting a home run, have I succeeded in explaining his
causing the crowd's going wild? The Jackson-Pettit accounts says that
I have. Any realizer of that property would have caused the crowd to
go wild. Therefore, it is causally relevant to their going wild.
But this answer should not seem quite right. The best explanation of
Strawberry's causing the crowd to go wild is, in the circumstances, that
he hit a home run. By pointing to precisely what Strawberry did, we
seem to have missed the critical factor. We have instead added all sorts
of what are, in the circumstances, irrelevant extra details. However, this is not to say that the precise explanation offered fails completely to specify the causally relevant property of Strawberry's swing. Given
the explanation offered and the background conditions, it follows that
Strawberry hit a home run. Intuitively, our standards for explanation are not so demanding that we must discount explanations that include
some irrelevant information as well as the relevant information.11
This type of case suggests that we need to distinguish what we might think of as the maximally causally relevant properties from those pro?
perties, like in the Strawberry example, which contain the causally relevant property and some irrelevant properties. To accommodate
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 10: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/10.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 347
these intuitions I propose to revise the Jackson-Pettit account to count
conjunctive properties as causally irrelevant:
c's being M is causally relevant to c's causing A iff
(a) for any P, if P is a possible realizer of M, then had c
been P, c's being P would be causally-relevantphy, to c's
causing A.
(b) there is no property M', distinct from M, such that
c's being M' satisfies (a) and c's being M entails (in the
circumstances) c's being M'.
That is, I will take the Jackson-Pettit account not to be an account of
causal relevance in all cases, but only of certain core cases (CRC). Then we get an account o? full causal relevance (CRF) as follows:
(i) if c's being M is CRC to c's causing e, then c's being M
is CRF to c's causing e.
(ii) if c's being M is CRF to c's causing e, then, if c is M', c's being M and M' is CRF to c's causing e.
CRC properties then are what we might think of as the maximally
causally relevant ones. If we were to be more pedantic in our ascriptions of causal relevance than I think we are, these would be the only ones
we would count as causally relevant. CRF is more permissive. It allows as causally relevant any property of the cause that contains the CRC
property as a component. Thus CRC is sufficient for CRF, but, because
of conjunctive properties, not necessary. However, (ii) allows conjunc? tive properties to be counted as causally relevant. Naturally, we might feel disinclined to count all conjunctive properties with an irrelevant
component as causally relevant. But then the burden is on us to show what is so special about irrelevant conjuncts in some cases and not in others. Given the array of cases in which we think that a conjunctive property with an irrelevant conjunct is causally relevant, I am dubious that such a constraint could be articulated. However, even if a con?
straint of this sort were available, it would not undercut the Jackson Pettit account, for we could merely incorporate it into (ii) above.
These revisions might be thought to dissolve a different problem for the causal relevance of mental properties, one distinct from their
purported irreducibility. This is the problem posed by two theses. First, intentional properties are wide. They involve causal/historical relations
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 11: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/11.jpg)
348 JONATHAN BARRETT
to the thinker's environment as well as internal components of the brain
(a view made popular by Putnam-Burge thought experiments). Second,
only properties internal to the brain can be causally relevant to be?
havior.12 The consequence of these two theses is that intentional proper? ties are not CRC, because, at best, only the internal component of an
intentional property could be CRC. The problem is removed, it might seem, because if the internal components of intentional properties can
be CRC, even if the whole intentional property cannot be CRC, the
whole intentional property can be CR^. I do not think this is a solution, because the question arises of whether common sense can be satisfied
with intentional properties with causally irrelevant external compo? nents. The role of the latter is in fixing certain semantic properties of
beliefs and it is unclear that the commonsense view can tolerate these
properties being causally irrelevant. According to that view, it is the
contents of our thoughts that determines our behavior, they do not just ride on the back of internal and apparently largely contentless proper? ties. Without resolving this difficult problem, the causal relevance chal?
lenge posed by externalism about intentional properties and individ?
ualism about psychological explanation survives. This topic is too big to pursue further here.
We have looked at cases in which we specify the causally relevant
property in excessive detail. Now I want to examine cases where we
fail to give enough detail. I use these to argue for a concept of partial causal relevance, one not captured by the Jackson-Pettit account.
We often apparently explain behavior by attributing a de re belief to
the agent. In such cases, it is simply not true that any realizer of a
particular de re belief would produce the behavior we are explaining.13 For instance, suppose I say that Willy is cheering because he believes
that Superman has arrived. If, as seems quite plausible, I am attributing a de re belief to Willy, then it is likely that Willy would behave differ?
ently were the same de re belief realized differently. How Willy behaves
in the actual and in counterfactual situations depends in part upon how
he conceives of Superman. If in the counterfactual situation his belief
is realized so that his conception of Superman differs from the actual
case (Superman qua Clark Kent and not Superman qua Superman), then he might well not cheer. This counts against the necessity of
Jackson and Pettit's account, because it produces cases in which a
different realizer of a de re belief property, one that intuitively is
causally relevant, produces a different behavioral event.
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 12: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/12.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 349
Cases of this sort are very easy to generate. We might say that a
brain state causes my wincing because it is a pain state. However, there are different kinds of pain and I might only wince if I feel a sharp
jabbing pain. If the pain is a gentle throb perhaps, I might merely rub
the affected area. We can realize being in pain in different ways so that
different realizers have different kinds of effects. Nonetheless, being in
pain is still causally relevant. Second, how I behave on the basis of my belief states might depend not just upon the contents of those states, but also upon the means by which I represent those contents. Repre?
senting the information in one way, my believing that there are six
groups of nine children might cause me to utter, "There are fifty-four children." However, if my belief is represented by mental analogs of
Roman numerals, I might not make the same utterance, because I have no idea how to multiply Roman numerals.14 So, the same thought can
be represented in different ways and thereby produce different effects.
Third, there are degrees of believing and desiring. We might say that
I rented "Oklahoma" because I desired that I see "Oklahoma", without
specifing how much I desire to see it. However, were I to desire to see
it, but a little less than I in fact do, I might not rent it at all.
All of these cases work as counterexamples to the Jackson-Pettit account in the same way. The individuation of the relevant mental states
in these examples, while quite fine grained (it succeeds in identifying the
cause), is still too coarse grained (it fails to identify the CRC property). There are kinds of pain and different kinds produce different effects.
Thus, merely to point to being in pain as the causally relevant property is to specify a property that could not possibly satisfy the Jackson-Pettit account.
One view we might take is that commonsense psychology is simply mistaken in these cases. The properties it counts as causally relevant are actually irrelevant, as the Jackson-Pettit account shows. That seems
wrong. First, the counterexamples I have described are not particularly rare cases that can be set aside or given special treatment. Indeed, they are representative of our ordinary practices of psychological explana? tion. Second, it seems overly fussy to require that to specify a causally relevant property one must specify the entire property that fully explains the causal transaction in question. Our normal standards of explanation rarely require such precision. Third and most importantly, these proper? ties do partially explain the production of the effect and are essential to its full explanation. These are reasons to count properties that are
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 13: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/13.jpg)
350 JONATHAN BARRETT
only partially and not fully explanatory in this way as causally relevant
to some degree. Examples from outside of psychology suggest that this
is the natural line to take. Suppose the stereo caused the walls to
collapse because it was playing very loudly. Then it seems natural to
say that it caused the walls to collapse in part because it was playing
loudly. Not only is its playing very loudly causally relevant, but so is
its playing loudly. In contrast, since they have no role to play in the
explanation of the production of the effect, the stereo's being made in
Taiwan and its having a sleek black finish are causally irrelevant. Thus, it seems, our concept of causal relevance includes both full and partial causal relevance. So we need an account of causal relevance that cap? tures both kinds. Therefore, the Jackson-Pettit account is too strong.
In fact the required revisions are straightforward and do not undercut
the central account. We need only add an extra condition to capture
partial causal relevance (CRP):
(iii) c's being M is CRP to c's causing e iff for some property F of c's, c's being F and M is CRC to c's causing c.15
To this point, the Jackson-Pettit account has withstood the pressures we have subjected it to. Moreover, we have been able to revise the
account so that it rather simply captures different degrees of causal
relevance. Thus far then, it is in good shape. In the next section, I will
introduce what I take to be the real problem for the Jackson-Pettit
account.
4. MIRACULOUS COINCIDENCE
Tu2 cases of partial causal relevance we discussed above had a common
reature. The extra property needed to give the core causally relevant
prof? ty was a psychological property, a property that is part of com?
monsense psychology, or at least of philosophical refinements of that
theory. Now I want to look at cases in which commonsense psychology cannot itself supply the extra necessary information to give us the CRC
property. Here is a case in which such a sub-mental difference is crucial. I am
out one cold day. Feeling thirsty and in need of warmth, I stop at a
cafe to buy a hot drink. I do not care greatly whether I get tea, coffee
or hot chocolate. I like them all equally and believe that any one of
them will, on this occasion, satisfy my needs. When asked for my order,
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 14: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/14.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 351
I say, "Coffee please." However, at no point did I form a desire to get coffee specifically, I just desired a hot drink. Plausibly, being my desir?
ing a hot drink is causally relevant to some brain state of mine causing me to say, "Coffee please." However, it might be that for some alterna?
tive physical realizers of that desire, had it been realized by one of
them, I would instead have said, "Tea please", or "Hot chocolate
please." Moreover, it might be that these different possible utterances
of mine result directly from sub-mental physical differences in the
different realizations of my desire. Thus, different realizers produce different behaviors, yet each realizer realizes my desire for a hot drink
and intuitively that desire is causally relevant to my saying, "Coffee
please", in actuality. The case seems perfectly coherent. To explain how sub-mental differ?
ences might lead to different behaviors, we need only suppose, for
example, that the search order or mechanism for hot drink words is
partially dependent on which physical realizer of the desire is in?
stantiated. Indeed, I would have thought that we should in fact expect such cases to arise, given the complexity of our brains and the seeming indifference of commonsense psychology to neurophysiology. However,
my objection here only requires the mere possibility of such cases.
We have a case in which an apparently causally relevant mental
property's effect is realizer dependent and so the property does not
satisfy the Jackson-Pettit test. Now, if troublesome realizers are rela?
tively rare, if say for all but one possible realizer of my desiring a hot
drink I would say, "Coffee please", then maybe there is no problem.
Presumably, the Jackson-Pettit test allows some leeway, even if it does
not explicitly say so. It is widely accepted that where there are lawlike commonsense psychological generalizations, they do not hold without
exception and I see no reason why Jackson and Pettit should not be
allowed to relax their condition in recognition of this. However, to
avoid this reply, let us imagine that there are lots of troublesome
realizers in this case. Then it is not true that any realizer would produce a request for coffee (ceteris paribus).
In that case, it seems to me to be natural to say that the mental
property in question, my desiring a hot drink, is causally irrelevant to
my saying, "Coffee please." Rather it is some physical property, the
physical property that in fact realizes my desiring a hot drink on this
occasion, that is causally relevant. Recall that mere realization by a
causally relevant physical property, unless the realizer is identical to
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 15: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/15.jpg)
352 JONATHAN BARRETT
the realized property, is not sufficient for causal relevance. That is the
simple view we considered and rejected at the outset. This claim might seem to clash with our intuitions about the case. One thing I can say to ease the discomfort is that my desiring a hot drink is still causally relevant to my behavior under the description ordering a hot drink, for
every possible realizer of the desire does cause an ordering of a hot
drink. We could redescribe the case so that this was not so, so that some realizers cause my ordering coffee, some my killing the waiter, and yet others my doing a fine imitation of Madonna. But then it would seem even clearer that no mental property was causally relevant to my
saying, "Coffee please", and that counts as evidence in favor of my
interpretation of the case.
If I am right, the widespread troublesome realizer is not a problem for Jackson and Pettit, for they can plausibly argue that in these cases
the apparently causally relevant property is in fact causally irrelevant.
However, this is where the real problem for them begins. If troublesome
realizers are very common, then rarely, if ever, could an irreducible
mental property be causally relevant. Of course, a nonreductive physi? calist will simply deny that we have reason to believe that they are
common. However, if we look at the nature of nonreductive physi? calism, we quickly find a reason.16 Suppose c causes e and intuitively c's being M is causally relevant to its causing e. Suppose that M on this
occasion is realized by physical property Pi, but that any of properties Pi,. . . ,Pn would have realized M had one of them been present in c.
For M to be causally relevant the Jackson-Pettit account requires that
the presence in c of any otP1,. . . ,Pn would in the circumstances make c cause e. But, given that these realizers have nothing physical/func? tional in common (M is irreducible), we should expect there to be lots
of troublesome realizers for M. Thus, we should expect large scale
mental causal irrelevance.
Now the nonreductive physicalist will deny that we should have this
expectation. That she may do, but she then owes us an explanation of
how a set of physical properties with nothing physical/functional in
common realize a single mental property and produce effects of a
common type. Without such an explanation, we appear to have miracu?
lous coincidence on a large scale. That, I take it, is something we
cannot accept. In other words, if no explanation is forthcoming, then
either the Jackson-Pettit account is too strong or we are forced to
accept widespread mental causal irrelevance.
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 16: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/16.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 353
Can we explain away the apparent coincidence? One way is to view
the irreducible mental property as a dispositional property: a physical
property realizes M just in case it would cause an event of the appropri? ate kind in the appropriate circumstances. However, clearly this is a
reduction and so it is not a move that is open to the nonreductive
physicalist. Indeed, the suggested line is probably going to end up as
a commitment to a form of functionalism, a position which nonreductive
physicalists explicitly reject. How else are we to explain why all these distinct physical properties
produce effects of a common type? David Papineau suggests we employ a selectional explanation.17 All these different properties produce a
common effect because they were selected to do so. Selection here
need not be confined to natural selection, but can also include, for
example, the design of an intelligent agent or the selection that occurs
as part of the developmental process. The last of these is held by
Papineau to be particularly important to the psychological case. In the course of development, he hypothesizes, a neural pathway is strength? ened when its activation produces favorable outcomes and weakened
when its activation yields unfavorable outcomes. It is then no accident
that many different physical properties produce events of common
kinds, because, any available physical property can be recruited to
produce appropriate behavior, subject, of course to the limits of neuro
physiology. Thus, it is unsurprising that the developmental process selects different properties in different organisms and maybe in a single
organism at different times.
That the details of Papineau's selection account are sketchy does not
undercut his argument. All we can reasonably expect is a plausible
story of how different physical properties might produce effects of common kinds. However, even if we allow him this much slack, I do not see that selectional explanation solves our puzzle. He has explained how distinct physical properties might have common effects, but I
thought the task was to explain how distinct physical properties that are possible realizers of a single mental property might have effects of a common kind. An answer to the former question is part of an answer
to the latter, but it is not sufficient by itself. For all Papineau has said, we have no more reason to think that a set of distinct physical properties with a common effect each realizes a common mental property than to
think that they all realize different mental properties or no mental
properties at all. Similarly, he has given us no reason to think that all
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 17: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/17.jpg)
354 JONATHAN BARRETT
the possible realizers of a mental property will all be selected to cause
a common effect.
To see this, consider a number of devices each with a button on it.
When the button is pressed a buzzing sound issues from the device.
Even though these devices might be physically very dissimilar, their common response to having their button pressed can be explained,
perhaps as the result of an inventor's exploration of the different ways of building a buzzer. Such Papineau-style selectional explanations seem
perfectly acceptable in these cases. However, they only explain how
physically different devices produce the same effect. They do not ex?
plain in addition why each of these devices say, is red, or was once
owned by Louis XIV, or is coveted by several antique collectors (or whatever irreducible property you care to pick).
In the psychological case, I suppose Papineau's reply must be that the
distinct physical properties all brought about behavior with favorable
consequences and so were recruited to produce that behavior, because
they each individually realized a certain irreducible mental property. The idea here is that one's beliefs, desires and other mental states are
good indicators of when behaving in a certain way will have positive consequences or will avert negative outcomes. Such a view is initially
plausible. We all like to think that the hard work we put into acquiring new beliefs, checking their veracity, weighing our desires against one
another, discovering what our desires really are and so on is in some
way in our interests. However, this is not sufficient to secure the re?
quired claim. We need, in addition, a possible mechanism that explains the postulated correlation between the presence of certain beliefs and
desires and the beneficial effects of certain behaviors.
The obvious move here is to hold that physical states of types Pi,. . . , Pn, (realizers of mental property M) tend to be caused by environmental states of type C and that M's are recruited as causes of
behavior of type B just in case 2?-ing in conditions C is beneficial. Thus, the occurrance of C fixes both the instantiation of a particular mental
property and the causing of a particular effect by that mental property under its different realizations.
I believe this position is very implausible. There are three big assump? tions it must make. First, it has to assume that environments are rela?
tively static; static enough that C still obtains when the organism pro? duces B. Otherwise, there is no guarantee that 5-ing will be
advantageous to the organism and hence no reason to expect the link
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 18: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/18.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 355
between realizations of M and B-ing to be reinforced. One might
respond that one need not hold that C still obtains, but need only
require that the environment has evolved in a certain way that makes
B-ing beneficial. However, this merely redescribes the difficulty. Now
we have to hold that environments, while not static, evolve in orderly
patterns. But, the rich diversity of everyday experience strongly sug?
gests that this is not so. Environments change quickly and unpre?
dictably. So much so that it looks like it would be pure luck if selectional
processes made the mental fit with behavior in the way Papineau de?
sires.
Second, our account has to assume that the environmental factors
producing the particular mental state are identical to (or very intimately connected with) those environmental factors that bear on the likely value to the organism of producing a certain behavior. However, only one small piece of my current environment will often be salient to my
coming to hold some belief. It would seem extraordinary to claim either
that such a small part of my environment determines the likely value
of future behavior or that those parts of my environment that do
determine this are themselves dependent in some way on that small
part upon which I formed my belief.
Third, the part of the environment that causes a mental state can
itself take many forms. My believing that father has returned might result from seeing him come into the room, from seeing his identical twin come into the room, from hearing banging and cursing downstairs, from reading a letter and so on. It is hard to believe that any of these events favors a single beneficial behavior for the organism.
So, despite Papineau's suggested solution and my developments of
it, the miraculous coincidence problem still remains for the Jackson Pettit account. Furthermore, even if Papineau's selectional explanations can be made to remove the miracle of causally relevant irreducible
properties in psychological cases, there will still be a range of cases
where selectional explanations do not seem to be available. Take color
properties and moral properties for example. If these are irreducible
properties, as many nonreductive physicalists adamantly maintain, and their instantiations are sometimes causally relevant, it is hard to see
what sort of selectional explanation will explain the causal relevance of their various realizers, since these properties do not seem to have been
selected to cause anything. Suppose that Cindy's t-shirt's being white
explains her keeping cool. Seemingly, the t-shirt's being white is claimed
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 19: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/19.jpg)
356 JONATHAN BARRETT
to be causally relevant to Cindy's coolness. Yet nothing, it seems, has selected physical realizations of whiteness to be cooling.
I conclude that Papineau's selectional explanations do not explain away miraculous coincidences in either the psychological case or for
irreducible properties more generally. Consequently, it seems that the Jackson-Pettit account is too strong. It overcame the objection to our
initial simple account of causal relevance by introducing the strong requirement that other possible realizers of a causally relevant irreduc? ible property would have been sufficient to produce the effect in the circumstances. That lead to the miraculous realizers problem. Thus, if there is no way around the miraculous coincidence problem, and I do not see one, either we reject Jackson and Pettit's account or we hold that all irreducible mental properties are causally irrelevant.
5. ALTERNATIVES
If we opt to reject the Jackson-Pettit account, we really need an alterna? tive. Although it is unfortunate for the nonreductive physicalist, there is no obvious alternative to their account. Certainly, there are alternative
accounts, but they seem to me to be most plausible when interpreted to be variants on the Jackson-Pettit account. I will briefly look at two
of these, Jerry Fodor's account in terms of hedged causal laws and Brian LePore and Barry Loewer's counterfactual account.
Fodor has suggested an account of causal relevance that goes some?
thing like as follows:18
c qua F is causally relevant to e qua G iff
(a) c causes e, c is F and e is G,
(b) Fs cause Gs is a (possibly hedged) law.19
The contentious claim here is (b). Fodor has to explain why we should ever expect there to be such laws if we do not believe that mental
properties are reducible. Fodor's account is in terms of a language of
thought (Mentalese). Roughly, to believe that hockey is the thinking man's sport is to stand in the believe relation to a sentence in Mentalese
that means that hockey is the thinking man's sport. Thought processes,
according to Fodor, are computational operations upon syntactic pro?
perties of Mentalese sentence tokens. For example, the inference from
the belief that hockey is the thinking man's sport to the belief that
hockey is a sport requires a computational process that takes as its
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 20: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/20.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 357
input the Mentalese correlate of the former belief and produces as its
output the Mentalese correlate of the latter belief. Thus, the causal
powers of belief- and desire-properties are grounded in the causal
powers of the syntactic properties of their Mentalese correlates and
hence we can explain there being hedged psychological laws in terms
of there being computational laws. Now this account certainly takes a
lot of the apparent magic out of the idea of there being causally relevant
mental properties, but I do not see how it overcomes the problem of miraculous coincidence. Fodor is committed to many-one relations
between ontological levels at at least two points. First, he holds that
many different physical states can realize a single computational pro?
perty. Second, he holds that many different computational states can
realize a single intentional property. I am unsure whether this makes
Fodor a nonreductive physicalist. It will depend upon how exactly he
individuates properties at the computational and intentional levels (for
example, individuation by functional role or by computational role are
normally rejected by nonreductive physicalists). Whatever, for the sake
of argument, let us suppose Fodor is a nonreductive physicalist. In that
case, we are going to have to explain how any realizer of mental
property F will, ceteris paribus, produce a realizer of G, even though the various possible realizers of F have nothing physical and/or compu? tational in common. That is just the miraculous coincidence problem that I have argued that the nonreductive physicalist cannot explain and
there is nothing in the language of thought hypothesis, short of render?
ing it in reductionist terms, that offers any form of solution.
I should add that Fodor is somewhat dismissive of the problem of
mental causal relevance in his claim that hedged psychological laws are
sufficient for mental causal relevance. He argues that our success in
predicting behavior by appeal to mental properties is evidence that there are hedged laws linking mental properties to behavior and that these laws by themselves are grounds for thinking that mental properties can be causally relevant. Of course that is right, but it does not solve the problem for the nonreductive physicalist. Rather it highlights the
difficulty she faces, namely, how can irreducible mental properties be
causally relevant so that there can be hedged psychological laws? To
try to solve this problem solely by appeal to the existence of hedged psychological laws just begs the question.20
I now turn to LePore and Loewer. They have recently defended a
counterfactual account of causal relevance:21
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 21: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/21.jpg)
358 JONATHAN BARRETT
c qua F is causally relevant to e qua G iff
(a) c causes e, c is F and e is G,
(b) were c to be F, c would cause e (or cause e to be G),
(c) were c not F, c would not cause e or e would not be G.
A favorable portrayal of this account requires attention to how (b) and
(c) are to be read. On an uncharitable reading (b) is vacuous: since the
nearest world in which it is satisfied is the actual world, (b) is satisfied
by any property c actually has. I assume LePore and Loewer have in
mind a more interesting reading where we look at a range of cases in
which c is F, including ones where c lacks other properties it has in
actuality. Then not every property of c's will satisfy it, because in a
world in which c lacks the genuinely causally relevant property, c will
not cause e no matter which irrelevant properties it has. If, to satisfy
(c), we require no effect when c lacks only F, we get an intuitively attractive account of causal relevance whereby a property is causally relevant to some effect just in case its presence or its absence makes the
difference between, respectively, the occurrence and non-occurrence of
the effect.22
Whatever other criticisms we might want to make of their view23,
my concern here is only with whether LePore and Loewer's (b) can be
read in such a way that we do not have to consider counterfactual cases
in which the mental property is realized by a different physical property to the one that realizes it in actuality. If it can, then it looks like they can avoid the miraculous coincidence problem, for it would not matter
to them what the effects of alternative realizers are. If it cannot, then
LePore and Loewer's (b) implies sameness of effect across variation in
realization and thereby inherits Jackson and Pettit's problems. Recall that appeal to the effects of alternative realizers was invoked
by Jackson and Pettit to overcome the case in which the causally
relevantphy physical realizer realized more than one property of the
cause, the intuitively causally relevant property and an intuitively caus?
ally irrelevant property. If LePore and Loewer's (b) does not require variation in realizer in the counterfactual cases, then Mary's ladder's
being a good heat conductor will satisfy the condition. If we are to
avoid this property being counted as causally relevant to the ladder's
causing her electrocution when resting on the power lines, it must be
that it does not satisfy LePore and Loewer's (c). There must be an
appropriate reading of (c) on which were Mary's ladder not to be a
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 22: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/22.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 359
good heat conductor, Mary would still be electrocuted. Such a case
would be one in which the actual physical realizer of both the ladder's
electrical conductivity and its thermal conductivity is replaced by a
physical property that realizes only its electrical conductivity and not
its thermal conductivity. It appears that (b) does not require sameness of effect as the realizer
is varied, if we adopt this particular reading of (c). Thus, it does
not seem that LePore and Loewer inherit the miraculous coincidence
problem. In fact, it turns out that they do have the same problem. The
sameness of effect as realizer is varied assumption is not made in (b), but is required in (c) to preclude causally irrelevant properties that are
realized by the causally-relevantphy physical realizer in actuality, the
ladder's thermal conductivity, from being counted as causally relevant.
In the ladder case, we have to assume that there is an alternative
realizer of the causally relevant property, its electrical conductivity, and
not of the irrelevant property, that produces the relevant effect. If we
are to avoid this problem, then it looks as if the sameness of effect as
realizer is varied assumption has to built into (b). Either way, LePore
and Loewer face a miraculous coincidence problem just like Jackson
and Pettit do.
My suspicion here is that the LePore-Loewer account, like the Fodor
account, is best regarded as a variant on the Jackson-Pettit account.
Indeed, to conjecture further, I think that the miraculous coincidence
problem does not reveal that something is wrong with the Jackson
Pettit account or these other accounts. They all seem to be the kinds
of stories one would try to tell to show how irreducible properties could be causally relevant. Rather it strongly suggests that nonreductive
physicalism cannot make sense of the causal relevance of the irreducible
properties it postulates. If this is correct, then it counts against the
appeal of nonreductive physicalism, unless the nonreductive physicalist can show that she can get by with causally irrelevant mental properties. I tried to suggest that she cannot, by appeal to the claim that we
intuitively think that our mental states cause the behavior they do
because they are those states. My belief that it is raining causes me to
put on my hat precisely because it is my belief that it is raining. This sort of consideration has standardly been the reason for taking issues
of mental causal relevance to be important. However, it has always seemed to me to be an appeal to apparent intuition rather than a
clearcut argument. Therefore, in my view, the nonreductive physicalist
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 23: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/23.jpg)
360 JONATHAN BARRETT
has reason to see if she can deny the apparent intuition. That is too
large a task to pursue here.24
NOTES
1 See, for example, Schiffer (1987, pp. 19-47).
2 Until recently, most nonreductive physicalists stated the third part of their account in
terms of a relation of supervenience holding between mental and physical properties. Differences arose over which specific relation of supervenience was required to sustain
physicalism (strong versus weak, local versus global; see Kim, 1984, on supervenience and its varieties). Now, the relation of supervenience is increasingly regarded as being too
weak to sustain physicalism (e.g., Kim, 1989 and Charles, 1992). Stronger nonreductive
physicalisms are now appearing (e.g., Charles, 1992). 3
See Segal and Sober, (1991, pp. 9-10) on mereological supervenience for a discussion
of important constraints on realization in the context of issues of mental causation. The
commitments I attribute to the nonreductive physicalist are actually considerably stronger than some would accept. Some do not require that a particular instantiation of a mental
property must have a definite physical realization. They make the weaker claim that
mental properties of an object as a whole supervene on the physical properties as a whole
of that object. See Worley (1993) for an interesting defence of the causal relevance of
the mental given such a view. I do not intend my arguments in this paper to apply to
this sort of position and it is a substantial further question, which I shall not pursue, whether they could. I am inclined to think that very weak nonreductive physicalist views
are not properly physicalist. Again, that it is not a question that I can deal with here.
See the references in note 2 for some of the arguments for this view. 4
A detailed account of these reasons would sidetrack us hopelessly. One of the fullest
discussions, which links the causal relevance of mental properties to rationality, is Lennon
(1990, pp. 38 and 42-45). I think that the details of Lennon's argument are problematic.
However, they at least indicate the kinds of reasons people have for thinking mental
causal relevance is necessary. See Barrett (1994) for arguments, including a discussion
of Lennon, that a Davidsonian account of rationalizing explanation does not require
causally relevant mental properties. 5 The argument is at least hinted at, and sometimes given a fuller treatment, in, for
example, Fodor (1990), Honderich (1982, 1984), Horgan (1989), LePore and Loewer
(1987, 1989), and Sosa (1984). 6
In attacking dualism, Schiffer (1987, p. 151) makes the related move of arguing that
we should not believe that there are superfluous causal laws. 7
A distinct possibility is that the immediate (i.e. neurological) realizers of mental proper? ties are themselves irreducible and hence non-physical. The claim embodied in the
condition must then be that the ultimate realizers of the mental property are causally
relevantphy. 8 Jackson and Pettit (1990).
9 Jackson and Pettit (1988, 1990, 1992).
10 Jackson and Pettit (1990, p. 205).
11 A particularly compelling example in this respect is due to Segal and Sober (1991, pp.
14-16). The environment caused the struck match to ignite because it contained air, we
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 24: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/24.jpg)
CAUSAL RELEVANCE AND NONREDUCTIVE PHYSICALISM 361
might claim. But, in fact it was because it contained oxygen and air is only 30% oxygen.
However, learning this would hardly make us withdraw our original claim. 12
For a careful attempt to state this argument, see Fodor (1987, pp. 27-33). 13
This example was suggested by Brian Loar. 14
This problem is now part of the folklore of artificial intelligence (and partly motivates
connectionist alternatives to classical AI-see McClelland, Rumelhart and Hinton, 1986).
Supposedly, researchers originally thought that the architecture of representation was an
issue secondary to showing how representations are manipulated in thought. Soon how?
ever, it became obvious that the manner of representation adopted affects how easily one can construct an algorithm for a particular kind of reasoning. See, for example, the
debate over analog versus digital representations in discussions of mental imagery (Tye,
1991). 15
There is no danger of this condition counting irrelevant components of conjunctive
properties as CRP, because we stipulated that the Jackson-Pettit account counts conjunc? tive properties as irrelevant. Also, while this condition does count disjunctive properties in preemption cases as CRP, this is not a problem because these properties really are
CRP, though admittedly their importance is not great. 16
This problem is stated in Papineau (1985, 1992); see also Cussins (1992). The argu? ments there are not directed against the Jackson-Pettit account, but they have obvious
application to it. 17
Papineau (1992, pp. 60-68). 18 Fodor (1990). 19
As a side point, the conjunction of (a) and (b) is not sufficient, because (b) might be
true even though Fs do not cause Gs by virtue of being Fs. Contrary to Fodor's account, these Fs are not causally relevant, because we would in principle be able to explain Fs
causing Gs without appeal to Fs or its realizer's causal relevance. This point is due to
Segal and Sober (1991, pp. 4-5). They show how it can be overcome by adding to the
account. 20
My thanks to an anonymous referee for pushing me to discuss Fodor's view more
fully. 21
LePore and Loewer (1987, 1989). The account given in Horgan (1989) is closely related. 22
LePore and Loewer (1989, p. 190). 23
See Barrett (1994) for some other objections to counterfactual accounts of causal
relevance. 24
My thanks to John Dreher, Hartry Field, Janet Levin, Brian Lear and Barry Schein
for their criticisms of a distant ancestor of this paper, to two anonymous referees for
Erkenntnis for their helpful suggestions, and to Gawaine Batchelor and Harry Brighouse for discussions of causal relevance issues.
REFERENCES
Barrett, J.: 1994, 'Rationalizing Explanation and Causally Relevant Mental Properties',
Philosophical Studies 74, 77-102.
Charles, D.: 1992, 'Supervenience, Composition and Physicalism', in D. Charles and K.
Lennon (eds.), Reduction, Explanation and Realism, Oxford University Press, Oxford.
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions
![Page 25: Causal Relevance and Nonreductive Physicalism](https://reader036.fdocuments.net/reader036/viewer/2022081810/57509d0e1a28abbf6b04c783/html5/thumbnails/25.jpg)
362 JONATHAN BARRETT
Cussins, A.: 1992, The Limitations of Pluralism', in D. Charles and K. Lennon (eds.),
Reduction, Explanation and Realism, Oxford University Press, Oxford.
Fodor, J.: 1987, Psychosemantics: The Problem of Meaning in the Philosophy of Mind, MIT Press, Cambridge, Mass.
Fodor, J.: 1990, 'Making Mind Matter More', in J. Fodor, A Theory of Content and
Other Essays, MIT Press, Cambridge, Mass.
Honderich, T.: 1982, 'The Argument for Anomalous Monism', Analysis 42, 59-64.
Honderich, T.: 1984, 'Smith and the Champion of Mauve', Analysis 44, 86-89.
Horgan, T.: 1989, 'Mental Quausation', Philosophical Perspectives 3, 47-76.
Jackson, F. and Pettit, P.: 1988, 'Functionalism and Broad Content', Mind 97, 381-400.
Jackson, F. and Pettit, P.: 1990a, 'Causation in the Philosophy of Mind', Philosophy and
Phenomenological Research 50 (Supplement), 195-214.
Jackson, F. and Pettit, P.: 1990b, 'Program Explanation: A General Perspective', Analy? sis 50, 107-117.
Jackson, F. and Pettit, P.: 1992, 'Structural Explanation in Social Theory', in D. Charles
and K. Lennon (eds.), Reduction, Explanation and Realism, Oxford University Press, Oxford.
Kim, J.: 1984, 'Concepts of Supervenience', Philosophy and Phenomenological Research
45, 153-176.
Kim, J.: 1989, 'The Myth of Nonreductive Physicalism', APA Proceedings 63, 31-47.
Le Pore, E. and Loewer, B.: 1987, 'Mind Matters', Journal of Philosophy 93, 630-642.
Le Pore, E. and Loewer, B.: 1989, 'More on Making Mind Matter', Philosophical Topics
17, 175-191.
Lennon, K.: 1990, Explaining Human Action, Duckworth, London.
McClelland, J. Rumelhart, D. and Hinton, G. 1986, 'The Appeal of Parallel Distributed
Processing', in J. McClelland and D. Rumelhart, and the PDP Research Group, Parallel Distributed Processing: Explorations in the Microstructure of Cognition, vol.
1, MIT Press, Cambridge, Mass.
Papineau, D.:1985, 'Social Facts and Psychological Facts', in G. Currie and A. Musgrave
(eds.), Popper and the Human Sciences, Nijhoff, Dordrecht.
Papineau, D.: 1992, 'Irreducibility and Teleology', in D. Charles and K. Lennon (eds.), Reduction, Explanation and Realism, Oxford University Press, Oxford.
Schiffer, S.: 1987, Remnants of Meaning, MIT Press, Cambridge, Mass.
Segal, G. and Sober, E.: 1991, "The Causal Efficacy of Content', Philosophical Studies
63, 1-30.
Sosa, E.: 1984, 'Mind-Body Interaction and Supervenient Causation', Midwest Studies in
Philosophy 9, 271-282.
Tye, M.: 1991, The Imagery Debate, MIT Press, Cambridge, Mass.
Worley, S.: 1993, 'Mental Causation and Explanatory Exclusion', Erkenntnis 39, 333
358.
Manuscript submitted April 25, 1994 Final version received October 25, 1994
School of Philosophy University of Southern California
3709 Trousdale Parkway Los Angeles, CA 90089-0451 U.S.A.
This content downloaded from 188.72.96.189 on Sat, 21 Jun 2014 23:22:11 PMAll use subject to JSTOR Terms and Conditions