Open Journal of Philosophy
2012. Vol.2, No.2, 75-83
Published Online May 2012 in SciRes (
Copyright © 2012 SciRes. 75
Two Myths of Psychophysical Reductionism
Ricardo Restrepo
Escuela de Constitucionalismo y Derecho, Instituto de Altos Estudios Nacionales, Quito, Ecuador
Received January 30th, 2012; revised February 22nd, 2012; accepted March 9th, 2012
This paper focuses on two prominent arguments claiming that physicalism entails reductionism. One is
Kim’s causal exclusion argument (CEA), and the other is Papineau’s causal argument. The paper argues
that Kim’s CEA is not logically valid and that it is driven by two implausible justifications. One is “Ed-
ward’s dictum”, which is alien to non-reductive physicalism and should be rejected. The other is by en-
dorsement of Papineau’s conception of the physical, immanent in Papineau’s causal argument. This argu-
ment only arrives at the physical property-property identities by using a conception of the physical that
licenses anything to be reductively physical, including putative core anti-physical entities; thus, leaving
Papineau’s causal argument and Kim’s CEA without a reductive physicalist conclusion of philosophical
Keywords: Causal Exclusion; Mental Causation; Physicalism; Supervenience; Reduction
Jaegwon Kim and David Papineau have each developed influ-
ential arguments to justify reductive physicalism: the causal
exclusion argument (CEA) and the causal argument, respec-
tively. Both approaches purport to show that physicalism en-
tails reductionism. The thesis of reductive physicalism is that
each mental property is itself identical with a physical property.
There will be more to say in this paper about the nature of this
claim, of course.
Using causal considerations, reductionists argue that physi-
calists can’t be “soft and cuddly non-reductionists” (Melnyk,
1995: p. 370, 2008), rejecting with their characteristically cool
rationality the “myth of non-reductive materialism” (Kim,
1989), and with stoic discipline “swallow” the “unpalatable im-
plications of our assumptions and presumptions” (Kim, 2010: p.
104). The reductionist pill is not without its rewards, however,
for its conclusions are of “great philosophical interest” (Pap-
ineau, 2002: p. 41). Or so we are told.
Despite the mentioned arguments to the contrary, physical-
ism, I think, does not entail reductionism.1 In this paper, I be-
gin by looking at the premises under which Kim’s argument is
formed. Kim’s intention is to reduce non-reductive physicalist
theses to absurdity, and derive reductionism from the apparent
contradiction. I present a basic non-reductive physicalist read-
ing of the premises, however, and find no such contradiction.
The idea is not to argue for one particular way in which to be a
non-reductive physicalist, but to show the existence of the logical
space available to this position, compatible with the premises.
The existence of this logical space has the consequence that the
causal exclusion argument is not logically valid. The premises
may all be true but the conclusion false.
Because one way Kim (2005) justifies his reductive physi-
calism is by endorsing Papineau’s (2002) conception of the
physical, it will be convenient, from an expository point of view,
to address Papineau’s causal argument for reductionism in the
process of addressing Kim’s. What I think is Papineau’s myth
of psychophysical reductionism is embedded in Kim’s argu-
ment. I argue that Papineau advocates a position that does not
in substance contradict non-reductive physicalism, because it
allows for higher-level causal properties, and these are pre-
cisely what the non-reductionist thinks mental properties are.
The deeper reason, however, is that Papineau’s position li-
censes calling any metaphysics “reductive physicalist”, include-
ing positions which endorse the existence of things clearly out-
side physicalist metaphysics, so it is trivial. Thus, Kim’s at-
tempt to use Papineau’s conception of the physical in the CEA
to yield the reductionist conclusion is unsuccessful, since using
that conception still does not generate the required contradic-
tion between the premises.
The other way Kim justifies reductionism is by endorsing
Edward’s dictum, which says that basic properties exclude non-
basic properties from making causal contributions (Kim, 2005:
p. 36). This justification fails, I argue, for several reasons. Ed-
ward’s dictum is not part of the official CEA. Kim does not
identify it as a premise, but rather introduces it as an aside, even
though he acknowledges that it is the driver of the reductionist
conclusion of the CEA. Edward’s dictum, however, is not part
of non-reductive physicalism and consequently cannot be taken
to be an internal generator of the required contradiction be-
tween the non-reductive physicalist premises. Edward’s dictum,
I add, is in serious need of justification.
Physicalism and the Causal Exclusion Argument
Kim’s CEA is well-known. It purports to take the form of a
reductio ad absurdum of non-reductive physicalism. Thus, typi-
cal theses of non-reductive physicalism are identified, a contra-
diction is said to be found, and a premise is rejected. Kim opts
for rejecting the thesis which distinguishes the non-reductionist
from the reductionist: the one which claims that mental proper-
ties are not reducible to physical properties. The premises of the
argument are as follows:
1Melnyk’s reductionism is outside the scope of this paper, though it is cer-
tainly a view worth careful analysis.
Supervenience: Mental properties supervene on certain
physical properties. That is, if any system s instantiates a
mental property M at t, there necessarily exists a physical
property P such that s instantiates P at t, and necessarily
anything instantiating P at any time instantiates M at that
Irreducibility: Mental properties as such are not reduce-
ble to the physical properties on which they supervene.
Mental Causation: Mental properties have causal effi-
cacy and relevance—that is, mental properties and their
in- stantiations causally contribute to bring about other
events with their properties.2
Exclusion: No single event can have more than one suffi-
cient cause occurring at any given time—unless it is a
genuine case of overdetermination.3
Closure: Each physical event has a sufficient physical
cause (insofar as causes of events are sufficient) (Kim,
2005: pp. 34-43).4
The causal argument says that since each physical effect has
a sufficient physical cause, each mental cause must itself be
physical (Papineau, 2002). The premises of the CEA can be
seen to be those of the causal argument for physicalism, plus
non-reductionist views of mental causation. If indeed a contra-
diction were to be found, then physicalism would be income-
patible with non-reductionism, and reductionism would be a
consequence of physicalism.
Suppose Sally picks up a glass of beer, which by Mental
Causation, her intention to do so causes. By Supervenience,
her intention is supervenient on a physical property. By Irre-
ducibility, her mental intentional property does not reduce to a
physical property upon which it supervenes. By Closure, there
was a sufficient physical cause for her picking up of the glass of
beer. Now Exclusion rules out the mental from making a causal
contribution on pain of degenerate overdetermination. So Mental
Causation can be true only if some other premise is false.
Kim’s final solution is to reject Irreducibility, and with it goes
non-reductive physicalism. Physicalism must consequently be
The argument seems compelling, but I believe it is mislead-
ing. One should first note that it is to be expected, given physic-
calism, that mental properties, like any other property of our
world, must be physical in some significant sense, since every
property inhabiting our world is physical. So, mental properties
are identical with certain physical properties. This follows just
from the fact that our world is completely physical and that
mental properties have identities, and consequently, that every
property is completely physical. The bare statement identifying
mental properties with physical properties should itself be un-
There is a standard distinction to be made between the two
ways in which a property can be physical. One is that it belongs
to the realm of fundamental physics. Call these physical1 prop-
erties. Various branches of physics are called in to inform us of
what that realm is like. It should be noted that physical science
can also posit and be interested in non-fundamental physical
properties. For example, astrophysics deals, amongst other
phenomena, with non-basic objects such as stars and galaxies,
which are governed by aggregates of arranged physical1 proper-
ties and laws. Such arranged aggregates of the basic constitu-
ents are also physical. There is no part or aspect of them which
is not physical. Distinctive properties of such arranged aggre-
gates of fundamental elements, to mark the distinction, are
Jackson (2006) provides a useful illustration of this distinc-
tion by use of an analogy with geometry, and identifies the
higher level sciences as gatherers of information about aspects
of physical2 properties.
We need an extended sense (of the physical) because
the patterns that economics, architecture, politics and very
arguably psychology, pick out and theorise in terms of,
include many that do not figure in the physical sciences.
The reason is no mystery: it is that aggregation creates
new properties because aggregations fall under patterns,
kinds, etc. that the items they are aggregations of do not
fall under Physicalists must allow that the world con-
tains aggregations that have properties that are not physi-
cal1 properties for the same reason, when all is said and
done, that someone who holds, rightly, that a triangle is an
aggregation of straight lines must allow that the triangle is
not itself a straight line (Jackson, 2006: p. 234, parenthe-
ses added).
It appears attractive to hold that at least some of the proper-
ties and instances physics aims to discover are the most basic of
all, and are regulated by a set of laws which allow things to be
composed of them in various ways and not in others, and that
those higher level objects and the causal patterns they fit into
are tracked by higher level sciences. As Jackson says, with
good reason, mental properties are of the higher level physical2
This is surely a physicalist view. Dualist arguments hinge on
the supposed lack of de re logical entailment from physical1
properties to mental properties, such as consciousness. For
example, Chalmers (1996) argues that the very possibility of
zombies, creatures which are physically identical with humans
but do not have conscious experience, refutes physicalism. In
such a case, consciousness would be neither intrinsically physic-
cal1 nor physical2. Having the relevant physical properties a
being with consciousness has would not be logically sufficient
for being conscious. To duplicate the conscious experience of
seeing bright red or feeling a sharp pain, one would need to add
something not present in beings merely physically identical to
beings who have the experiences. Non-reductive physicalism is
incompatible with this view. It would say that indeed, any
minimal physical duplicate of our world includes the conscious
experience of seeing red and of feeling pain, to use Jackson’s
(1994) schema. Both the experiences of seeing red and of feel-
ing pain are bona fide physical properties present in our world.
2In a recent reply to Shoemaker’s non-reductive physicalism, Kim (2010, p.
107) says that “For a positive resolution of the mental causation problem
(the CEA), I believe we should reach the statement that mental properties
are causally efficacious as a conclusion, not start with it as an assumption.
What troubles me is that Shoemaker’s procedure seems the opposite; for him
mental causal efficacy is a ‘starting assumption’, as he puts it” (parentheses
added). Notice that this shifts the goal posts of the CEA. For the problem o
mental causation is supposed to be one which arises from within non-reduce-
tive physicalism, a view which, as Kim himself formulates it, includes the
claim that mental properties have causal efficacy.
3It seems correct that systematic overdetermination could not be genuine, as
Kim supposes.
4The parenthesis is to make sure Closureis compatible with the possibility
of quantum indeterminism.
What the non-reductionist denies is that physical2 properties
(of which mental properties are an example) and physical1 prop-
Copyright © 2012 SciRes.
erties per se are identical. Reductionism requires mental and
physical1 properties per se to be identical. It is only when prop-
erties as such are identical that a property reduces to “another”;
not when they merely coincide nor when they have a contingent
identity relation. The minimal, but sufficient, claim of the non-
reductionist is that the relation between mental properties and
physical1 properties as such is not identity. There is a family of
ways this non-reductive relation can be. Baker (2002) and
Pereboom (2002), for example, have claimed that the relation is
constitutional coincidence without identity. A decision coin-
cides constitutionally with the instantiation of certain physical1
properties at a time, though they are not the same properties.
This is in the same way in which, in terms of properties, being
the Ship of Theseus coincides constitutionally at a time with
having certain physical1 properties. In terms of tokens, Pere-
boom (2002) argues that the fact that the Ship of Theseus and
the basic physical entities of which it is composed at a particu-
lar time have different temporal and modal properties is enough
to establish that the two items are not reductively the same. It
should nevertheless be clear that the Ship of Theseus is a per-
fectly physical entity all of whose properties are completely
bona fide physical properties in the context of physicalism.
Similarly, physical1 properties might coincide constitutionally
with the property of being a decision, but have temporal and
modal differences sufficient for them not to be identical with
each other.
Constitutional coincidence without identity could be extended
to events, like the sinking of the ship. That token event has many
smaller lower-level events composing it: the many sinkings
“experienced” by its parts. Are the part-sinkings identical with
the ship’s sinking? According to the constitutionalist, they are
not. For one, that same ship-sinking could have been composed
of different part-sinkings, for example, if there were certain
repairs done on it before leaving port, but which did not actu-
ally take place. Thus, the sinking of the ship and the part-sink-
ings have different modal properties. In this sense, the event
that is the sinking of the ship is distinct from the events that are
the part-sinkings. The same reasoning for non-identity that
applies to the property and token case, applies to the event case.
However, one of the ways in which higher level objects are
seen to be distinct from lower level objects is that they, at least
sometimes, maintain their identity past a time when their for-
mer constitution does not, or vice versa. In the event case, this
might seem to be pre-empted by the apparent fact that events
are dated particulars and that the lower level events coinciding
with the higher level event have the same temporal coordinates
This apparent fact, constitution theorists may contend how-
ever, is not really a fact. It is a live possibility that in fact the
temporal coordinates of lower level constituting events do come
apart from the coordinates of higher level constituted events in
that sometimes the sinking of the ship and the part-sinkings
coincide in time only partially. Such is the case of a ship that is
being sunk and is fully submerged but has not reached the bot-
tom. As the ship descends a powerful bomb detonates within it
and its parts scatter, destroying the ship, but its parts survive
and continue to sink. The part-sinkings have temporal coordi-
nates that outstrip the temporal coordinates of the ship’s sinking.
Thus, temporal properties for lower level and higher level
events can differ as well, and the reason for advancing a rela-
tion between lower level and higher level events to be constitu-
tion and not identity would hold.
Another way in which the relation between mental and physic-
cal1 properties might be non-reductive is through contingent
identity. Reductive identities are stronger than contingent iden-
tities. A lump of clay might be identical with a statue (Gibbard,
1975), but being a statue does not reduce to being a lump of
clay. A lump of clay and a statue, if identical, are contingently
identical. Being a statue and being a lump of clay are two dis-
tinct properties, which at times are had by the same object, but
not always, and sometimes the same statue can change compo-
sition and the same lump of clay can stop being a statue. One
may wonder whether this is merely a perspective game. From
the perspective of snake or bat, the lump of clay and the statue
may not seem to be so different. This may well be true, but it
does not prove anything. Bats and snakes are not sensitive to
many facts humans are sensitive to (and we are probably insen-
sitive to others they are sensitive to). For instance, Carbon and
Iron may not seem so different from the perspective of a bat or
a snake, but certainly these are two chemical kinds. The same
goes for many such things human legitimately differentiate and
other species do not. Of course, one could subsume statues
under the category of lumps of clay. A statue might just be “a
lump of clay that has been artistically worked upon.”5 However,
while it is true that some statues have this property, the point is
that they have this property contingently. One such statue might
have been, instead, a lump of metal that has been artistically
worked on, for instance, and certainly, being a lump of clay that
has been artistically worked on is a distinct property from being
a lump of metal that has been artistically worked on. Not all
statues are lumps of clay. The point is that these two properties
could be had by identical objects, but since they can and some-
times do come apart, being an object with both these properties
is a contingent fact—not one that reduces statues to lumps clay.
To take another example, suppose the winner of the race is
identical with David (Lewis, 1972). Does the property of being
the winner of the race reduce to the property of being David?
Are the two properties per se identical? To the non-reductionist,
asserting that they are is untenable. These are two properties of
the same object, and they are both physical in a sense compati-
ble with physicalism. An identity between the properties as such
does not have to be asserted in order to keep being a physicalist.
It takes different things to be David and to be the winner of the
race, though David satisfies both. Being David and being the
winner of a race should not pose a problem for physicalist
There are also “closer” contingent identities, short of reduc-
tion, namely contingent property identities. Such cases fall into
a generalized version of the Supervenience schema nicely. The
general supervenience schema says:
General Supervenience: Physical2 properties supervene
on certain physical1 properties. That is, if any system s in-
stantiates a physical2 property M at t, there necessarily
exists a physical1 property P such that s instantiates P at t,
and necessarily anything instantiating P at any time in-
stantiates M at that time.
For analogy, the property of weighing more than seven kilo-
grams supervenes on the property of weighing ten kilograms,
and even though these are two properties, they can be co-in-
stanced in a single token (Macdonald & Macdonald, 2006).
Furthermore, one might say, x’s having the property of weigh-
5Thanks to a reviewer at the Open Journal of Philosophy for these objec-
Copyright © 2012 SciRes. 77
ing more than seven kilograms is identical with x’s having the
property of weighing ten kilograms. These are two different
properties in themselves, but the properties might be identical
in certain instances. Consider another example: an animal’s
being a mammal is its being a dog, though by no means can we
say that being a dog and being a mammal is the same property
per se, which is what reduction requires. Copeland (2000) de-
velops a formalism for this Aristotelian conception of certain
property relations. This kind of non-reductive identity can make
sense of Kim’s claims emphasizing that in particular instances,
mental and physical properties might be identical, but shows
that more is required for reduction.
Reductive identities require complete property coincidence.
This is, at least in part, how one property per se is identical to
“another”, and this is a requirement the reductionist does not
deal with very well because of the phenomenon of multiple
realizability. Take Putnam’s (1975a) twin earth example, that if
someone came back from twin earth bringing the transparent
liquid in twin earth’s rivers with the XYZ chemical structure,
this substance would not be water. The exciting original finding
was that water seemed to be reductively identical with H2O
because it could not be identical with something without this
chemical structure. Water cannot be XYZ (Funkhouser, 2007).
Or so it seemed. Of course, if XYZ is heavy water, D3O, then
water could be something other than H2O. This, however, only
shows that water is not necessarily identical to H2O, and con-
sequently not reductively identical to H2O, even though, again,
in particular instances, it might be that a particular sample’s
being water is its being H2O.6
Of course, to this argument Kim responds with the “disjunc-
tive move” (Kim, 1998, 2005; Jaworski, 2002). If M is multiply
realizable by P1 and P2 (and only those) then M is reductively
identical to (P1 or P2). The disjunctive move is that mental
properties are reductively identical with the disjunction of their
possible realizers. This is a remarkably ad hoc move in need of
justification in my view.
One can say that having a negative charge is disjunctively
identical with having a negative charge before yesterday or
today or after, for instance. But a move such as this one is
baseless. One wants to ask, what is the basis of the disjunctive
move? The fact that it may theoretically be performed certainly
does not justify that it be performed. Physical1 properties, such
as having a negative charge, are not disjunctive in a compelling
way, and for the same reason physics2 properties, of which
mental properties are a subset, are not compellingly disjunctive.
Though mental properties can be broken up in a disjunction just
as the physics1 properties can, there is little by way of motiva-
tion to think this way. On the contrary, such moves miss what
is common to all the possible realizations in virtue of which
they fall under the same kind, and in virtue of which they have
similar effects.
The sufficient physical causes relevant to Closure are basic
properties and instances physical science aims to uncover, and
what they logically amount to (physical2 properties). Thus, for
example, we can know that triangles logically result from put-
ting lines (putatively basic properties in geometry) together in
certain admissible ways. This is where Putnam’s (1975b: p. 296)
well-known peg example is relevant: it identifies geometrical
properties in a peg, let’s say, being a triangle of particular size,
which determine the peg’s causal capacity to fit in certain holes
while other pegs of a different shape and size do not. The basic
physical properties arranged in a certain way logically add up to
the object’s being a triangle (a physical2 property), which cor-
responds to certain capacities which contribute to certain of its
actual and potential activities.
So Closure is fine as long as we know it counts in physical2
properties, and this opens the door for mental properties to be
of this kind. Spin, charge, and the rest of the physical1 proper-
ties make a contribution, but capacitate the peg to go through
certain holes only when they add up to being a triangle within a
range of certain sizes.
Kim frequently refers to Alexander’s dictum to support his
case (e.g. 1998: p. 119). This dictum says that only causal
properties exist or should be believed to exist. But if we agree
that physical2 properties such as having a particular macro-
scopic shape exist, then we see that by Alexander’s dictum we
should believe in the causality of such properties. Of course,
Kim says that he does but in a reductive way. If all this
amounts to is saying that, to take the case of the peg for con-
creteness, that a triangular peg is identical to physical1 proper-
ties arranged as a triangular peg of a certain size, then who
could disagree? The condition that the physical1 be arranged in
certain ways, however, is exactly, according to the non-reduce-
tionist, what raises the situation to one where physical2 proper-
ties are instanced, enabling new causal patterns of interaction.
There is famous complication here (Dretske, 1988). A singer
shrieks “break” and a glass breaks. Now, the sufficient physical
cause of the breaking includes the amplitude and frequency of
the generated sound waves. However, in this case, the sufficient
physical cause, by non-reductionist standards, appears to in-
clude the semantics of “break”. But this clearly is not a property
that makes a causal contribution to the breaking. The qua prob-
lem remains, it might seem.
But in my view this is not the case. The proposal of this pa-
per involves the idea that sufficient physical causes are reliably
tracked through the sciences using the experimental method.
With the experimental method, we can manipulate independent
variables to see their effects. Thus, take Dretske’s case to be
representative of the control group. A word, with a certain se-
mantics, pitch and volume, is shrieked, and the glass breaks. In
another case, we can vary the semantics (the independent vari-
able) while keeping the pitch and volume fixed. For example,
the singer might agree with her audience that when she shrieks
“break” it will mean that it is time to take a seat, and she might
faithfully intend to mean that proposition when she does. The
glass would break without dependence on the independent
variable (the semantics). Alternatively, she might decide to
shriek “Amen” with the same pitch and volume. Again, the
glass would break without dependence on the semantics of the
word, thus confirming that this is not a part of the sufficient
physical cause.
Analogous experiments yield different scientific results with
respect to mental properties generally, however: their causal
contributions are detected. Their manipulation results in ob-
served differences. Consider Sally’s intention to pick up the
beer again. It is a physical2 property, the logical result of certain
physical1 properties put together in a certain way. Furthermore,
it is verifiable that without the intention, Sally would not have
picked up the glass and that by varying physical1 properties
within the intention’s range of possible realizations, the same
result of picking up the glass comes about. However, if you
vary the independent variable, that is, Sally’s having the inten-
6For another argument that “water = H2O” does not express a necessary
identity see Barnett (2000).
Copyright © 2012 SciRes.
tion to pick up the glass of beer, another effect will occur,
thereby verifying the causal contribution of her mental proper-
This is the non-reductionist reading of the premises of the
CEA, and, as we have seen, under such a reading there appears
to be no contradiction and the argument seems, consequently,
to not be logically valid. We might then conclude that since it is
the non-reductive physicalist position which was supposed to
be reduced to absurdity, it is the non-reductionist reading that
counts. However, while this is true to a significant extent, it is
not complete. For it might be that there is some other reading
which we somehow nevertheless ought to give the theses. So it
is worth seeing what other reading Kim suggests. One of his
two suggestions involves Papineau’s conception of Closure and
the causal argument, and the second suggestion involves Ed-
ward’s dictum, the additional unofficial assumption of the CEA,
mentioned earlier. I consider these in turn.
Papineau’s Reductive Physicalism
Kim (2002: p. 671) indicates that reductive physicalism is
the only consistent and robust form of physicalism there is.
However, he considers many things physical which are not
explicitly a part of the theoretical universe investigated by fun-
damental physical theory, such as neural activity, outward bod-
ily movement, and so on. All these things are physical2 phe-
nomena, which the ideas raised thus far place on a par with
mental phenomena.
So what could Kim mean by “sufficient physical causes”?
Kim (2005: p. 43) tips us off by pointing to Papineau (2002).
The first thing to notice is that Papineau (2002: Ch. 2) claims to
be committed to a reductive form of physicalism, so it is ques-
tionable that Kim should be able to say that he is reducing
premises representing non-reductive physicalism to absurdity,
when his reading of the premises are exactly those of a self-pro-
claimed reductionist instead of a non-reductionist. There is a
whiff of circularity in this move.
Nevertheless, Papineau’s (2002) physicalism does not entail
that “sufficient physical causes” must be non-mental.7 We will
see that some of the principles at work in Papineau’s account
allow for mental properties to do irreducibly mental causal
work, and thus are able to be referred to as “sufficient physical
causes”. Papineau’s (2002) physicalism asserts the two follow-
ing components. First, the physical is to be understood as what
is inorganically identifiable, or “identifiable non-mentally-and-
non-biologically”. That is, the physical just is what can be re-
ferred to “independently of this specifically mental conceptual
apparatus” (Papineau, 2002: p. 41). If physicalism is true, this
means that those concepts which do not operate in terms like
seeing and believing are sufficient to refer to anything that ex-
ists in nature and participates causally in the world.
The second part is the supposedly reductive part, advancing
the idea that “materialism is to be understood as a matter of
property identity” between mental and biological properties, on
the one hand, and material properties, on the other. Mental
properties, and particularly, “conscious properties are identical
to material properties—that is, they are identical either to strictly
physical properties, or to physically realized higher properties”
(Papineau, 2002: p. 47). But this, of course, can easily be fit
into the physical1/physical2 scheme, with the non-reductionist
Recall that the causal argument says that since every physical
effect has a sufficient physical cause (with the fine print about
quantum indeterminism), every mental cause must itself be
physical. By what we have seen in the previous section about
the conception of the physical, this is not itself a reductionist
position. But let’s suppose we insert Papineau’s two claims.
Then, the causal argument looks like this: since every non-men-
tally-and-non-biologically identifiably effect has a non-men-tal-
ly-and-non-biologically identifiable cause, mental causes are
non-mentally-and-non-biologically identifiable. In fact, the rea-
soning continues, supposing such a mental cause to be a prop-
erty, then that property is identical to one which is non-men-
tally-and-non-biologically identifiable.
This conception of the physical provides an interpretation of
the premises of the CEA without generating a contradiction.
Mental properties are non-mentally-and-non-biologically iden-
tifiable. Take Sally’s intention, for example. It is identifiable as
a state resulting from an arrangement of fundamental particles
in a particular space-time region. But this fact certainly does
not imply that the intention does not cause, which is what is
required for a contradiction.
Furthermore, that the mental is identical with the physical, in
his sense, is what Papineau thinks is of “great philosophical
interest” (2002: p. 41). Given that conception of the physical, it
is of course easy to derive the thesis that any property or cause,
including mental ones, will be identical with properties or
causes thus identifiable. However, this conception is not rec-
ommendable. For, how significant is it to say that one is a re-
ductive physicalist when any world would qualify as such? Any
putative property or entity I can think of, physical or not, I can
identify non-mentally-and-non-biologically. Let us consider things
which are clearly outside the physicalist worldview: God, emer-
gent properties, Cartesian souls, and Platonic numbers. Sup-
posing theism is true, God is non-mentally-and-non-biologi-
cally identifiable as the thing that created the universe. Sup-
posing emergentism is true, non-physical emergent properties
are non-mentally-and-non-biologically identifiable as non-linear
effects of certain arrangements of matter. The immaterial Car-
tesian soul is non-mentally-and-non-biologically identifiable as
one of the things that interact causally with certain particles
(coincident with the pineal gland). The Platonic number eight is
non-mentally-and-non-biologically identifiable as the number
of planets orbiting the Sun.
By Papineau’s standard, any possible world, including those
containing core anti-physical things, are worlds where physic-
calism is true, since anything can be non-mentally-and-non-bio-
logically identified. Thus, it follows that under his conception,
“physicalism entails reductionism” is true, but it is trivially so.
There is no philosophical interest in this assertion, since any-
thing, including any property and any possible cause is non-
mentally and non-biologically identifiable. Furthermore, given
that any property can be identified in this way, then any world
with any kind of property is a world where reductive physical-
7In places other than Papineau (2002), he argues that the physical is the
non-sui-generis-mental (Spurrett & Papineau, 1999; Montero & Papineau,
2005). It is worth pointing out here that this account is also compatible with
the non-reductionist idea that mental properties are organized aggregates o
basic physical properties, which can be referred to by “sufficient physical
causes”. For a property to be non-sui-generis-mental, and consequently
physical by the standards of these theorists, it is sufficient that it be com-
letely determined by basic physics. This is a condition that the mental
satisfies. This is not a thesis I treat further here because it is different from
the one Papineau advocates by himself and from the one Kim alludes to.
Both theses are addressed by Restrepo (2012).
Copyright © 2012 SciRes. 79
ism is true. But that position faces a choice between having
reductive physicalism be trivial and rejecting the proposed con-
ception of the physical (Restrepo, 2012). The attempt to get
reductionism from physicalism, by trivializing the notion of
reductive physicalism, is Papineau’s myth of psychophysical
reductionism. This is one supposed avenue to reductionism that
Kim endorses, but it is not the only one.
Edward’s Dictum
As mentioned before, Kim says that the idea that drives the
CEA is Edward’s dictum. This idea purports to generate a con-
ception of sufficient physical causes that prevents Closure from
including mental causes. Edward’s dictum says that:
There is a tension between vertical determination and
horizontal causation. In fact, vertical determination ex-
cludes horizontal causation (Kim, 2005: p. 36).
This is a very strong assertion. Assuming that supervenience
is a vertical determination relation (as Kim does), together with
Supervenience, Edward’s dictum is sufficient to result in the
conclusion that mental properties are causally excluded by the
properties on which they supervene. The rest of the premises of
the CEA would be superfluous. However, Kim does not for-
mulate this assumption as a premise. Rather he acknowledges
its strong and necessary role as something of an aside. But why
formulate the rest of the premises of the CEA then, and not just
have Edward’s dictum together with Supervenience as the
official argument?
The answer, I think, is that no non-reductive physicalist
would endorse Edward’s dictum—and not without reason—so
it has to be smuggled in. It is worth noting that the dictum is not
a proposition that looks attractive to, nor entailed by, non-re-
ductive physicalism. Consequently, it appears that it cannot
form a part of a successful reductio of non-reductive physical-
ism. In being a supposed reductio ad absurdum of non-reduce-
tive physicalism the argument was supposed to be driven by a
tension arising from within that position, which we can now see
it is not. But this is perhaps too quick a judgement to make.
Perhaps this principle should in the final analysis be endorsed
by non-reductive physicalists for some good reasons. Let’s see
if Kim’s are.
Being a Yellow Lump and Time
Suppose a lump is yellow at t. What determines this fact?
One option is that the lump has a microstructural property M at
t that vertically determines its being yellow. Another option is
the lump’s being yellow at t-, which we may suppose it was.
Kim (2005) reflects that “[a]nything that happened before t
seems irrelevant to the lump’s being yellow at t; its having M at
t is fully sufficient to make it yellow at t” (pp. 36-37), and con-
cludes that because the lump’s having M at t is sufficient for the
lump’s being yellow at t, that the lump’s being yellow before t
makes no causal contribution.
The initial problem with this idea is that Kim is considering
determinative facts that happen at different times (t- and t),
and consequently the two options need not compete. In general,
to each of two times in the history of a fact there will corre-
spond a complete set of conditions that contribute causally to
the determination of the fact in question. No conflict arises out
of this supposition because we know from the beginning that
considering facts determinative of another do not necessarily
compete if they happen at different times. Picture a horizontal
column of standing dominoes. The domino falling at t- does
not causally compete with the domino falling at t. It might be
said that the falling domino at t is self-sufficient in the sense
that if we erased its causal history, while maintaining the falling
domino at t, that time-slice of the domino would still exist. But
this supposition should not lead us to believe that the falling of
the domino at t was not caused by the prior event of another
domino falling at t-. This is why the Exclusion premise in the
CEA is time-relative—it says that there cannot be more than
one sufficient physical cause at any one time, not in general.
To bring this out more, suppose a painter painted the lump
yellow prior to t and you wish to know what caused the lump to
be yellow at t. Then it would be quite causally relevant that a
painter painted it prior to t, even if the lump’s microstructural
property M vertically determines this at the later time, t. Kim’s
view would say, to the contrary, that this is not really the case
simply because the lump’s having M at t is fully sufficient to
make it yellow at t.
Kim’s offered model has the implausible implication that a
person’s painting a lump yellow does not causally determine
the lump’s being yellow at a later time just because the lumps
being yellow at the later time has a microstructural base. It
should be noted that by this argument, there are no prior causes
of the lump’s being yellow at t. The lump’s being yellow at t
would be completely causally undetermined by prior facts just
because it has M at t, which is implausible. By the same stan-
dard, the lump’s having M at t, being self-sufficient at t, was
not caused by prior facts. By extension, no physical fact at a
time has prior causes. This, of course, is no longer a specific
problem for mental causation, but expresses a general sceptic-
cism about causation, which is not supposed to be in question.
A common response to this argument is that the painter
causes the lump to be yellow by causing a difference in its mi-
cro-structural base. While this may be true, the point is that the
principle that Kim uses to justify Edward’s dictum in order to
drive the CEA implies that no properties of prior events are
causally relevant to latter events. He just assumes, on the basis
of somehow being characterizable as an isolated event at t, that
no prior event caused x’s being B at t. If this applies to any
event that is not the first event in the universe, then it applies all
events. This causal eliminitavist implication of Kim’s reasoning
is of course very uncompelling and therefore cannot provide
support for Edward’s dictum.
Causing at the Same Time
Let’s focus on a case that does not make the mistake of try-
ing to pit two candidate causes from two times against each
other. Take some physical effect P2 which happens at t and the
candidate physical cause P1 which occurs at t- and the candi-
date vertically determined cause M1 which happens at t-.
Does P1 exclude M1 in the causation of P2?
Let’s test whether vertically determined properties are caus-
ally excluded by their realization bases. Consider the Boat:
Suppose a boat has the property of being made out of metal
arranged in a certain way (P1). The boat’s being made of solid
material (M1) is a supervenient property of its being made of
metal. Being made of solid material is multiply realizable. The
boat’s property of being made of solid material could also be
realized, for instance, by being made of wood. Notice the fact
Copyright © 2012 SciRes.
that the boat’s being made of solid material is a supervenient
property with a realization base and that this does not rule this
property out from being one of the causally contributing factors
of floating. A causal power of being made of solid material is
that it makes the boats that have this property float.
Is the boat’s being made of solid material excluded by its
being made of metal in the causation of floating? Intuitively, it
is hard for it to even seem that it is or for one to think how or
why one should believe it so. Reasserting Edward’s dictum to
support the exclusion at this point would be unconvincing,
since it is this principle that needs explanation and justification
in the first place.
One of the biggest “temptations”, I think, in the debate about
the CEA is to think that physical bases of supervenient proper-
ties are sufficient, without the supervenient property, to cause
the putative effect of the mental: that P1, without M1, is caus-
ally sufficient for P2.
In order to evaluate this claim, one might consider the fol-
lowing. It is an indicator of X causing Z while excluding Y,
that X be able to cause Z in the absence of Y. But this is exactly
what X in this case is unable to do. The boat’s being made of
metal could not cause the boat to float unless the boat was made
of solid material. It is beyond question that ceteris paribus the
boat floats when it has the property made of metal and of solid
material. However, suppose you minimally modified the boat
so that you make the material of which it is made not solid.
You could do this by subtracting the metal itself or melting it.
The result would be that the boat stops floating. Of course,
this procedure would involve changing the microstructural base.
Still, the point is that there is no scenario whereby the micro-
structural base is held fixed and does the causing of the event in
question, while the supervenient property is not present. And, if
you change the supervenience base (say, by substituting the
metal with wood) while maintaining the supervenient property
of being made of solid material, the boat would still float. This
is evidence that the supervenient property is not causally super-
fluous. The base is not able to do any causing of the relevant
effect, without the supervenient property. The fact that every
time you subtract the supervenient properties from a cause, it
fails to bring about the otherwise realized effects, together with
the fact that you can change the base while keeping the super-
venient property and generate the same effect, should lead one
to think that supervenient properties have causal powers. This is
how we generally detect causal powers: we see the difference
between having certain factors present and not having them
The realization base was supposedly sufficient, but we see
that if the supervenient vertically determined property is taken
out, the subvenient physical cause is made insufficient. This
provides evidence that supervenient properties make causal
contributions, and Edward’s dictum is false. This also indicates
that there is no degenerate overdetermination. These factors are
more aptly understood as contributing causes which form the
sufficient physical cause of the boat’s floating.
It is worth addressing at this point Gillet and Rives’ (2005)
belief that determinables (a variety of supervenient properties)
with their own causal powers don’t exist. Being red, being a
determinable of being scarlet (or some other preferable physical
base), does not exist. By the same reasoning, we might suppose
that supervenient properties with their own causal powers don’t
exist. Gillet and Rives believe this based on what they call the
“Parsimony Worry” and the “Causal Power Concern”. The Par-
simony Worry is that:
[I]t is not clear that we should also take determinable pro-
perties such as being a mass and being charged to con-
tribute powers in addition to the determinates that always
accompany them. For to do so would be a kind of “double
counting”, to use David Lewis’ phrase, of the causally ef-
ficacious properties (pp. 486-487).
The Causal Power Concern is that:
Once we distinguish determinates and their corresponding
determinable properties the question is whether determi-
nable properties really contribute any powers at all. By
this we do not mean to again press the criticism that all
the powers of individuals can be accounted for simply by
positing determinates. Rather, the concern is whether be-
ing a mass or being charged, where the latter are not to be
confused with some determinate mass or charge, actually
contribute any causal powers to individuals. For example,
ask yourself exactly what the property of being charg-
ed—again, not to be confused with some determinate
charge—contributes to individuals by way of causal pow-
ers? The suspicion is that there are no such properties (p.
Were Gillet’s and Rives’ thinking to apply to supervenient
properties generally, their view would imply that causal super-
venient properties don’t exist. As Gillet and Rives recognize,
however, the Subset View developed by Sydney Shoemaker
(2003) provides a basis of an account that responds to these
eliminativist worries, though they reject it. The Subset View
says that determinable properties are distinct but partially over-
lapping with determinate properties. Causal powers of deter-
minable properties coincide with a (non-empty) subset of causal
powers of their determinate properties.
In an analogy with parts and wholes, it is worth pointing out
that wholes don’t eliminate their parts or the causal contribu-
tions of their parts on the basis of parsimony or because of the
fact that parts do not outstrip wholes. The analogous claim is
that causal powers of supervenient properties overlap with a
subset of causal powers of their supervenience bases; but by no
means does this entail their elimination.
To deal with their specific examples, I do think that having
charge, any charge, has a causal power all positive and negative
charges have: namely, the ability to interact through the elec-
tromagnetic force. All and only those things with charges have
this ability, so it is not a trivial truth. Think about it this way.
Sally has ten apples. Anybody who has ten apples has five ap-
ples. By implication, Sally has five apples. Sally should not say
that because she has ten apples, that she does not have five
apples. It is evident that she is not double-counting her apples if
she asserts that she does have five apples when she has ten. She
has ten apples, of which five are a subset. Now, those five ap-
ples surely have causal powers. For example, they might cause
Sally to enjoy their taste, or be nurtured, or if she throws them
down a building, they might break a car’s windscreen. The fact
that the five apples are a subset of ten apples does not eliminate
this fact. The fact that having the ability to interact through the
electromagnetic force is a subset of the causal powers of having
some determinate charge does not eliminate that fact (see also
Pereboom, 2002).
Now suppose that Sally did not have the five apples, then
ceteris paribus the apples left to her are going to nurture her
Copyright © 2012 SciRes. 81
less than the ten apples she might have had to eat when she had
five apples. There will be less things she has which interact
with the electromagnetic and gravitational forces, and so on,
which makes prior sufficient causes for certain effects, insuffi-
cient. Having five apples is vertically determined by having ten
apples. However, Gillet and Rives’ eliminativism gets this
wrong, I think, as does Edward’s dictum. Eliminating Sally’s
having five apples, like eliminating her intentions, changes the
world (see also Yablo, 1992).
The Analogy with Fake Causation
Taking the lead from Jonathan Edwards, the second intuitive
model Kim offers to support Edward’s dictum is that of a mir-
ror image which never causally depends for its existence on the
existence of prior mirror images, but rather on being produced
at each time by the thing it is an image of (Kim, 2005: pp.
36-37). The first thing to notice is that this model does not ex-
hibit the truth of Edward’s dictum, where vertical determination
relations are at issue. Rather, the model consists of two things
which might seem at first glance are such that one causally
determines the other (the two images at different times), but
this appearance is false. The model is an instance of “faux cau-
sation” (Kim, 2003: p. 171). But surely, this could not mean
that just any putative causal relation is fake. So what is special
about mental causation that makes it especially vulnerable or
suspicious? Why does the existence of fake causation imply
that vertically determined elements are always fake causes?
Further, if the existence of fake causation implied that vertically
determined causes must be fake, why would this not imply that
all causes are fake? These questions are never addressed.
Just as there are plenty of examples of the fake causation that
Kim points out, there are plenty of examples where there is real
causation. By the same standard, these would show that verti-
cally determined elements are real causes. Kim’s argument here
seems to be like saying that because Pluto is a fake planet (not a
real planet in the current understanding anyways), that there are
no real planets, which is evidently false.
General scepticism about causation is not at issue, nor is Kim
supposed to be such a sceptic. That there is a distinctive threat
for the mental is precisely what the CEA was supposed to show.
But to point out that fake causation exists is insufficient for this.
If it, by itself, were to work against the mental, it would work
against any other kind of physical causation, which again re-
sults in general scepticism about causation.
The CEA can only validly proceed by official incorporation
of Edward’s dictum. Without it, there is no reductive physical-
ist conclusion of philosophical interest. Once this fact is uncov-
ered, it can be seen that the position validly reduced to absurd-
ity is not non-reductive physicalism, but some other position no
one should hold. If the premises are real representations of the
non-reductive physicalist position, then there is no contradic-
tion between them and the irreducible causal powers of the
mental can retain their rightful ground in the physical world.
Furthermore, if we give the premises the trivial understanding
of “reductive physicalism” that Kim and Papineau endorse, we
see that it is a trivial conclusion they are defending—one not
worth calling by that name and standing up for, since it is un-
able to differentiate itself from a position asserting the exis-
tence of anti-physical things.
Baker, L. (2002). On making things up: Constitution and its critics.
Philosophical Topics, 30, 31-52.
Barnett, D. (2000). Is water necessarily H2O? Philosophical Studies, 98,
99-112. doi:10.1023/A:1018636505551
Chalmers, D. (1996). The conscious mind: In search of a fundamental
theory. New York: Oxford University Press.
Copeland, J. (2000). Indeterminate identity, contingent identity, prop-
erty identity, Aristotelian-style. Philosophical Topics, 28, 11-25.
Dretske, F. (1988). Explaining behavior: Reasons in a world of causes.
Cambridge, MA: MIT Press.
Funkhouser, E. (2007). A liberal conception of multiple realizability.
Philosophical Studies, 132, 467-494.
Gibbard, A. (1975). Contingent identity. Journal of Philosophical
Logic, 4, 187-221. doi:10.1007/BF00693273
Gillet, C., & Rives, B. (2005). The non-existence of determinables: Or,
a world of absolute determinates as default hypothesis. Nous, 39,
483-504. doi:10.1111/j.0029-4624.2005.00510.x
Jackson, F. (1994). Finding the mind in the natural world. In R. Casati,
B. Smith, & G. White (Eds.), Philosophy and the cognitive sciences.
Vienna: Holder Pichler-Tempsky.
Jackson, F. (2006). On ensuring that Physicalism is not a dual attribute
theory in sheeps clothing. Philosophical Studies, 131, 227-249.
Jaworski, W. (2002). Multiple-realizability, explanation, and the dis-
junctive move. Philosophical Studies, 108, 289-308.
Kim, J. (1989). The myth of non-reductive materialism. Proceedings
and Addresses of the American Philosophical Association, 63, 31-47.
Kim, J. (1998). Mind in a physica l world. Cambridge: MIT Press.
Kim, J. (2002). Responses. Philosophy and Phenomenological Re-
search, 65, 670-679. doi:10.1111/j.1933-1592.2002.tb00231.x
Kim, J. (2003). Blocking causal drainage and other maintenance chores
to do with mental causation. Philosophy and Phenomenological Re-
search, 67, 151-176. doi:10.1111/j.1933-1592.2003.tb00030.x
Kim, J. (2005). Physicalism, or something near enough. Princeton:
Princeton University Press.
Kim, J. (2010). Thoughts on Sydney Shoemaker’s Physical Realization.
Philosophical Studies, 148, 101-112.
Lewis, D. (1972, 1980). Psychophysical and theoretical identifications.
Australasian Journal of Philosophy, 50, 249-258.
Macdonald, C., & Macdonald, G. (2006). The metaphysics of mental
causation. The Journal of Philoso phy, 103, 539-576.
Melnyk, A. (1995). Two cheers for reductionism: Or, the dim prospects
of non-reductive materialism. Philosophy of Science, 62, 370-388.
Melnyk, A. (2008). Can physicalism be non-reductive? Philosophy
Compass, 3, 1281-1296. doi:10.1111/j.1747-9991.2008.00184.x
Montero, B., & Papineau, D. (2005). A defense of the via negativa for
physicalism. Analysi s, 65, 233-237. doi:10.1093/analys/65.3.233
Papineau, D. (2002). Thinking about consciousness. Oxford: Oxford
University Press. doi:10.1093/0199243824.001.0001
Pereboom, D. (2002). Robust nonreductive materialism. Journal of
Philosophy, 99, 499-531. doi:10.2307/3655563
Putnam, H. (1975a). The meaning of “meaning”. In K. Gunderson,
(Ed.), Language, mind, and knowledge. Minneapolis: University of
Minnesota Press.
Putnam, H. (1975b) Mind, language and reality (Vol. 2). Cambridge:
Cambridge University Press. doi:10.1017/CBO9780511625251
Restrepo, R. (2012). Thinking about physicalism. Open Journal of
Philosophy, 2.
Shoemaker, S. (2003). Realization and mental causation. Identity, cause,
Copyright © 2012 SciRes.
Copyright © 2012 SciRes. 83
and mind (Expanded ed.). Oxford: Clarendon Press. 427-451.
Spurrett, D. & Papineau, D. (1999). A note on the completeness of
“physics”. Analysis, 59, 25-29. doi:10.1093/analys/59.1.25
Yablo, S. (1992). Mental causation. Philosophical Review, 101, 245-
280. doi:10.2307/2185535