The Heavy Burden of Proof for Ontological

Naturalism

Georg Gasser & Matthias Stefan

University of Innsbruck

1. WHY NATURALISTS CANNOT DODGE ONTOLOGY

N aturalism enjoys almost a status of orthodoxy among contemporary

analytic philosophers. Unfortunately naturalism is not a clearly
defined philosophical position. Rather it is comparable with a broad stream
consisting of various philosophical approaches. It is not our aim to provide
an elaborated definition of contemporary naturalism. Following Mike
Rea’s characterization (Rea 2002, 50-73) we consider naturalism to be a
programmatic set of strategies to understand the world. Central to this
programmatic set of strategies is to analyse and present the world by
relying heavily on science (Forrest 1996, 89). Hence, naturalism implies
the attitude to consider science (at least) as the primary source of reliable
knowledge about reality. Science, according to naturalism, has shown to be
the most successful strategy for understanding the structure of our world
and its causal interaction (Löffler 1999, 36). This is the way we understand
naturalism. Such a characterization is very vague. It leaves room for many
different interpretations methodological, epistemological and ontological
alike. In our paper we focus on ontological issues. We aim at exploring
which ontological commitments come hand in hand with a naturalistic
outlook on reality. This exploration presupposes that naturalism and
ontological reflection are intertwined. An argument supporting this
presupposition goes as follows:
Ontology is the philosophical discipline investigating the ultimate
structures of reality. It aims at formulating what exists and what should be
deemed as real or unreal. If it is true that naturalists claim that the methods
of natural science exert a kind of hegemony over all strategies pursuing
160 Georg Gasser & Matthias Stefan

truth, then sciences are best in telling us what exists. Consequently

ontology—according to naturalism—depends on what is recognised as real
by sciences. A naturalistic minded philosopher, if asked what the ultimate
structures of our world are, should consistently base his/her answer on
current scientific research and the ontological commitments coming along
with it.1
Many contemporary naturalists, however, refuse to pursue studies in
ontology. They argue that naturalism should be interpreted neutrally from
any ontological point of view. Ontological discussions tend to
complicate—instead of enhancing—the interdisciplinary project between
science, philosophy and common sense respectively (e.g. Tetens 2000,
287f.; Clayton 2004, 142). If naturalism leaves ontological discussions
aside and concentrates on epistemology and methodology, for instance, it
might support interdisciplinary projects between science and philosophy. It
could start with concepts of folk psychology and take into account
empirical investigations of cognitive science. In such a way naturalism
might be helpful to analyse and better understand central concepts of
cognitive science (Koppelberg 2000).
Whatever the specific merits of epistemological or methodological
naturalism might be, they avoid the philosophical puzzles arising from
ontological thought. Such or similar the argument goes against the view of
taking ontological issues seriously within naturalism.
We concede that most scientists do not explicitly care for ontology. It is
alien to scientific practice to spell out what a certain theory implies
ontologically. This is rightly done. Scientists are not paid for doing
ontology. However, we deem an ontologically neutral or abstinent
naturalism unsatisfying from a philosophical point of view. If one accepts
scientific realism and the thesis that explanatory concepts in science come
along with ontological commitments, then a tension between different
causal claims and various scientific explanations becomes a problem at
some point. Psychology, for instance, makes causal claims about mental
states and neuroscience makes causal claims about neurological facts.

1
Papineau 2001, for instance, aims at pointing out how modern physics led many
philosophers to become ontological physicalists. Dupré 2004, 37ff., portrays a
similar route from materialism over modern physics to physicalism.
 The Heavy Burden of Proof for Ontological Naturalism 161

Now, if it is assumed (as most non-dualists would do) that mentality

depends upon neuronal activity in the brain, then the question arises
concerning the relation of these different causal claims. Are they just two
different descriptions of the same reality? The decisive question is who or
what is doing the causal work: Does mentality dispose of mental causal
powers? Or is mentality only causally efficacious in virtue of its
dependence upon neurological activity? This problem shows that it is not
unusual that various scientific explanations and its corresponding
ontological implications are in conflict with each other. There are causal
entanglements between the various levels of reality investigated by
different scientific disciplines. These causal entanglements raise all sorts of
issues which have been dubbed “problem of causal competition”, “problem
of causal exclusion” or “problem of cross-level causation”. These issues
will have to be evaded or answered. Because it has proven to be difficult to
provide a convincing way to evade the tensions between different causal
stories at the various levels of reality, we assume that an ontological
neutral position can only be provisional for naturalism: It might be useful
to leave ontological implications out of account for specific aims of
scientific research or philosophical reflection. However, if it is believed
that scientific concepts and hypotheses refer to something real, then
methodological and epistemological issues are closely intertwined with
ontological assumptions. Ontological questions are neither external to
scientific practice nor of no interest for science. Scientific theories make
statements about entities being causally efficacious. By doing so,
ontological questions are implicitly raised by scientific theories. If our
argument is correct, naturalistic minded philosophers should regard it as a
substantial topic of their research to work out which ultimate structures of
reality we are reasonably enabled to accept according to naturalism.
Ontology is not a marginal but a central issue for naturalism.

2. NATURALISTS’ COMPLAINT AND MEN OF STRAW

After having argued for the importance of taking ontology seriously in

naturalistic thought, in this section we discuss briefly which ontological
options should be preferred over others. In surveying philosophical
literature on naturalism we encounter again and again a complaint from the
162 Georg Gasser & Matthias Stefan

side of naturalists (see for instance Sukopp 2006, 28-33). Many naturalists
complain they are attacked from two sides: First, they are attacked because
naturalism is presented as being ultimately radically reductionist or
eliminativist. Secondly, they are attacked because naturalism is presented
in such a liberal way that almost everyone is willing to accept it.
According to the first line of attack naturalism coincides with reductive
physicalism. Based upon contemporary physical theories it is claimed that
our world contains only basic micro-physical entities. These entities
together with fundamental physical laws are sufficient to account for all
phenomena we encounter in our universe. Once the physical facts at the
bottom level of reality and the laws holding them are fixed, all facts of our
universe are fixed. There is nothing relevant above these fundamental
physical facts. Such a position implies the (in principle) reduction of our
macro-world to micro-physics, the (in principle) reduction of laws of
special sciences to the fundamental laws of physics and the (in principle)
reduction of ‘non-physical phenomena’ such as consciousness, subjectivity
or intentionality to physical phenomena. In such a reductionist world
ultimately only micro-physical entities and their causal interactions exist.
For many people such an ontology has unpalatable consequences because
the world we are familiar with is ontologically inferior or negligible. There
are no physical facts above micro-physics, no causal powers above the
powers of the ultimate constituent parts of reality. The world as we
conceive it, the causal powers we ascribe to the objects of our meso-
cosmos, in short, our Lebenswelt, drains away after all. In the final analysis
the ultimate level of reality—the level of micro-physics—is the only level
which truly has to be taken seriously from an ontological (and maybe on
the long run also from a scientific) point of view.
According to the second line of attack, ontological naturalism is so
widely defined that nearly every ontological thesis is part of it. If only
God, angels, immaterial substances or mythical creatures are excluded
from a naturalistic ontology, then naturalism does not seem to offer any
interesting ontological insights. Anyone who has not strong theistic,
animistic or obscurantist tendencies would subscribe to such a version of
ontological naturalism. A naturalism telling us that our world is material
based, that it evolved over time in a continuous process, that complex
systems consist of simpler parts, and that we do not have to postulate non-
 The Heavy Burden of Proof for Ontological Naturalism 163

material entities for explaining our universe, is a rather unimpressive

philosophical program. It might legitimately be asked whether such a view
is still worth to be named naturalism (Stroud 2004, 34f., Keil and
Schnädelbach 2000, 9f.).
Most naturalists consider the two options sketched above as mere men
of straw which easily can be torn into pieces. This complaint from the
naturalist’s side implies that any serious analysis (and critique) of
naturalism should look for positions being situated between a too strong
reductionist physicalism and a too liberal naturalism. We take these
complaints seriously. In the next section we aim at spelling out possible
compromises between the two extremes.

3. SCIENTISM AND ITS ONTOLOGICAL OPTIONS

We presented naturalism as being committed to the attitude that science

ultimately recognizes what is real and unreal. Any ontology faithful to this
attitude has to be developed under the authoritative guidance of science.
We label such an attitude ‘scientism’. Kornblith gives a succinct
expression of this view:
Current scientific theories are rich in their metaphysical implications. The task of
the naturalist metaphysician, as I see it, is simply to draw out the metaphysical
implications of contemporary science. A metaphysics which goes beyond the
commitments of science is simply unsupported by the best available evidence.
(Kornblith 1994, 40, our italics).

Kornblith’s view is programmatic in character. Hence, let us ask: How

shall an ontology be construed out of contemporary scientific theories? We
suggest that first of all we have to clarify which sciences are to be accepted
as providing relevant information for a naturalistic ontology. Second, we
have to explain how the relevant sciences relate to each other. It is easy to
see that these two problems are connected. A well-known example from
the history of science helps to explicate this interconnection. The bonding
problem in chemistry was a much debated topic at the edge of the 19th
century. As long as micro-explanations of chemical bonding were not at
hand, chemical theories assumed fundamental chemical forces of chemical
elements. After the development of quantum mechanics, the gap between
164 Georg Gasser & Matthias Stefan

chemistry and physics began to shrink. It was detected that quantum states
and not some emergent chemical properties are the reason for chemical
forces. Chemical bonding became explainable in terms of quantum
mechanics.2 The important point for our discussion is that in this case all
relevant information is provided by physical theory. Physics tells us
everything about the problem of chemical bonding. This scientific progress
had influential consequences for ontology. Before quantum states were
known, emergent chemical properties were assumed as being part of our
world. After the discovery of quantum states the assumption of proper
chemical properties became superfluous. As the example shows, it might
be the case that within the disciplines, scientism considers as relevant for
its ontological program, specific theories of a determinate discipline (e.g.
chemistry) are reducible to more basic theories in another (e.g. physics).
For a naturalist subscribing to scientism it is not only important to identify
those sciences which provide relevant information for a naturalistic
ontology; it is crucial to elucidate their interrelationship as well. We
propose three solutions for achieving this aim:

(i) Scientism pursues a reductionist strategy. Scientism assumes

that the entities of higher level sciences are reducible to
micro-physical entities. Biological entities, for instance,
ought to be reduced to chemical entities and these to physical
ones. Scientism then turns into physicalism, as all higher
level sciences are nothing more than special cases of (an
assumed complete) science of physics.
(ii) Scientism becomes a kind of conciliable naturalism.
Conciliable naturalism says we should accept everything as
relevant that we think we need to make sense of and which
we are convinced is part of our world (Stroud 2004, 33).
Conciliable naturalism accepts the whole range from natural
to social sciences, and humanities.
(iii) Scientism relies on some well established sciences, such as
physics, chemistry and biology. We call this position

2
McLaughlin 1992 discusses these scientific discoveries at length and relates them
directly to the rise and fall of British Emergentism.
 The Heavy Burden of Proof for Ontological Naturalism 165

‘naturalism of core sciences’. It might be claimed that this

version of naturalism is some kind of a compromise: It
avoids the openness of conciliable naturalism on the one
hand without subscribing to a too strong reductionist version
of physicalism on the other hand.

How shall we deal with the three options at hand? Position (i) is often
considered problematic. It seems to coincide with the position identified
above as our first man of straw because it claims that (micro-)physics alone
provides the relevant information for any ontology. Position (ii) seems to
be a non-starter due to its open-mindedness. Conciliable naturalism is not
more than “a slogan on a banner raised to attract the admiration of those
who agree that no supernatural agents are at work in the world.” (Stroud
2004, 35) Conciliable naturalism is identifiable with the position identified
above as our second man of straw.
The remaining candidate is position (iii). For our argument it is not of
further importance whether ‘naturalism of the core sciences’ considers
only physics, chemistry and biology as relevant sciences or whether the list
can be extended3. Central for our argument is that a well-defined notion of
sciences seems to be presupposed. Explicating the concept of science is a
necessary precondition for being able to say which sciences take part in the
ontological undertaking of naturalism. It is, however, anything but clear
what natural sciences are. As long as this problem remains unsolved the
problem of sciences’ interdependency cannot be tackled either. Without a
clear concept of science, scientism can hardly justify why it takes certain
sciences seriously for ontology, whereas others are seen as less important.
This gives rise to the impression that an envisaged science based ontology
amounts to a mere matter of taste—the groundless capriciousness of
certain people to favour certain disciplines over others.

3
Of course, the list cannot be extended arbitrarily. This extension has to be stopped
at some point before turning this position into a form of conciliable naturalism. But
this problem is of no further importance here.
166 Georg Gasser & Matthias Stefan

Naturalists have to answer the question about their concept of science:

Having declared that the methods of natural science provide the only avenue to
truth, the naturalist should be prepared to say what these methods are, or which
sciences qualify as ‘natural sciences.’ (Keil 2000, 148)

There are three possibilities to develop such a concept of science:

(a) Providing methodological criteria which separate sciences

from non-sciences.
(b) Providing a list of acceptable sciences.
(c) Demonstrating the unity of science.

Let us discuss the three possibilities in turn.

(a) It is a fact that in scientific progress methods of science develop.
Long established methods in scientific practice are factored out as not
being scientific enough anymore whereas other methods become integral
parts of current scientific practice. Standards what counts as scientific and
what as unscientific change with the course of science’s development. In
short, the methods of science cannot be determined a priori. Our
characterization of naturalism implies that naturalism finds its orientation
within science and in this respect it is at science’s mercy. As a
consequence, naturalism cannot impose a priori methodological criteria on
science. Otherwise naturalism is not a loyal companion of science anymore
but sets itself up as judge over it. Such a move is inconsistent with
naturalism’s commitment to follow and cooperate with science. If this
argument is sound, naturalism cannot develop criteria for distinguishing
sciences from non-sciences because such criteria would determine a priori
what has to count as science and what not.
(b) Providing a list of admissible sciences has also to be refused. It is
impossible to provide a non-arbitrary list of accepted sciences without
methodological criteria. If we look for methodological criteria, we are back
at (a), which has been already ruled out as a possible solution to the
problem at hand.
(c) Demonstrating the unity of science seems to be a philosophical
project most evidence speaks against. We have little reason to believe in
any kind of unity of science. If we look at science as practiced, then we are
 The Heavy Burden of Proof for Ontological Naturalism 167

unable to notice any methods of investigation which are characteristic for

all sciences. There are also naturalistic attempts to provide a unity of
sciences in content, which assumes that all sciences investigate entities that
are one and the same in nature. These entities might be, for instance,
physical in nature, so that all sciences are united in having physical
content. Such a move brings us, finally, to the doctrine of physicalism
which we have discussed as a first possible interpretation of scientism.
Whether this is a viable way we have to let others decide. Currently such
attempts lack any convincing philosophical basis and empirical support.
There are no signs of a realisation of the project of the unity of science in
terms of method and content (Dupré 2004, 51).
Scientism seems to be incapable of giving an adequate concept of
science. Without such a concept, however, it is impossible to construe an
ontology out of contemporary scientific theories. And even if we assume,
for the sake of argument, that scientism possesses such a concept of
science, an abiding problem for a science-based ontology still remains,
namely that of the interdependency of sciences. Most sciences imply ‘local
ontologies’. A local ontology is the ontology a scientific discipline (or even
a specific theory within a discipline) uses explicitly or implicitly for its
area of research. Biology, for instance, (to speak simplified) works among
other ontological categories with three-dimensional objects, such as
organisms for explaining biological phenomena. Many of these objects of
biological research correspond to objects we are familiar with from
common sense. Particle physics instead might carry out its research in a
four dimensional time-space system with fields, atoms and electron clouds.
It might not feel the urge to refer to three-dimensional objects familiar to
us from everyday experience.4 It is of no further importance for our
argumentation what entities exactly are assumed in different scientific
areas. It suffices to point to the fact that different local ontologies in
different sciences lead irreversibly to an unpleasant consequence: Various
sciences use different ontological categories, while their interrelationship is
everything but clear. Advocates of scientism have to explain how different

4
It should be noted that this assumed ontological framework for physics, which is
already simplified, is not at all undisputed. Within physics itself are many sites of
(epistemological and ontological) fracture (see Falkenburg 2006).
168 Georg Gasser & Matthias Stefan

ontological categories, drawn out of contemporary science, relate to each

other. If a clarification of these relationships cannot be provided, the
project of scientism is seriously threatened. It is as if we had many pieces
of a puzzle and did not know how to put them together to a unified picture.
A naturalist might reply that a thoroughgoing science-based ontology is
not available. There are gaps in a science-based ontology. As the various
sciences are not united so the local ontologies are not related to them.
These gaps have to be accepted as expressions of our ignorance within the
project of a science-based ontology. All we have are small puzzles, but a
great unifying picture is simply beyond our reach. We ought to live with a
fragmented ontology of our reality.
We see the main problem of such a proposal in the acceptance of gaps.
As argued before, scientism needs to draw a line between acceptable and
unacceptable sciences for being able to construe its ontology. Often this
line is drawn between natural sciences and other academic disciplines
which have the mental or the social as their primary research object
(Mellor/Crane 1995). Let us suppose there are reasonable grounds to draw
the distinction at the intersection of those sciences concerned with the
physical and those studying the assumed non-physical. If, for the sake of
argument, we accept this distinction we should be prepared to answer the
question why we are willing to tolerate gaps among the natural sciences
and still adhere to the drawn distinction. Why should we consider a
fragmentation between particle physics, atomic and molecular theory,
biology, physiology, or neurology as less problematic than the gap between
the physical and the mental? This assumption grounds on the
presupposition of the unity of the accepted sciences. It is presupposed that
one can smoothly go up the hierarchy of sciences from physics over
chemistry and biology to neurobiology without any change in content. It is
always the same realm that is investigated. The only problematic gap, then,
is lurking between the physical and the mental. Implicitly it is presupposed
that the mental stands alone in our physical world (Churchland 1981, 75).
This presupposition, however, is by itself not justified, as the unity of
science has to be provided first. And we saw already that the signs of a
realisation of the project of the unity of science are currently few and far
between.
 The Heavy Burden of Proof for Ontological Naturalism 169

If scientism goes another route instead and accepts gaps for its
ontological undertaking within sciences itself, as well as a further gap
between so called hard and soft sciences, then a science based ontology
does not have any reason to favour the subject matter of hard sciences over
the one of soft sciences. Scientism, then, would accept a plurality of
sciences as equally relevant for ontology. There would not be one ontology
but many ontologies depending on the respective (scientific) perspective
one takes up for getting an accurate account of the phenomena under
investigation. It seems likely to us that such a view turns into liberal
naturalism which has been ruled out as a non-starter at the very beginning
of the discussion. What our discussion should have made clear is the
following:

(i) If almost everything is assumed as being part of nature, then

naturalism becomes so liberal that it turns into triviality.
(ii) There is no generally accepted concept of science which
allows drawing a clear line between acceptable and non-
acceptable sciences. Criteria to distinguish between relevant
and irrelevant sciences seem to be difficult to obtain.
Without such criteria, however, the entire project of a
science-based ontology cannot be achieved. It remains
unclear to which sciences ontology should refer.
(iii) Even if it could plausibly be argued which sciences a
naturalistic minded philosopher should take into
consideration for his/her ontological studies, the
interrelationship between the accepted sciences and their
assumed entities has still to be clarified.

Drawing out the ontological implications of contemporary science, as

Kornblith demanded, has shown to be full of flaws. Scientism as a
philosophical project seems to be a failure—at least from the perspective
of its ontological implications. How shall we proceed then? We suggest
returning to one option we presented at the beginning of our discussion:
physicalism. It seems to be the only remaining option providing a solution
for the problems of the notion of science and the interdependency between
scientific disciplines. If this were true, naturalists should accept what they
170 Georg Gasser & Matthias Stefan

are generally loath to do: To argue that physicalism, and the reductionism
coming along with it, is the most promising route for presenting a
distinctive naturalistic ontology.

4. ONTOLOGICAL PHYSICALISM

How shall physicalism be characterized? We start with a list of core tenets

physicalism5 entails (see e.g. Pettit 1993, 213-223; Beckermann 2000, 128-
143; Kim 2005, 149f.):

(1) The world is constituted out of microphysical entities which

physics is in the best position to identify. Every entity in the
world is either a microphysical entity itself or constitutes of
microphysical entities.
(2) Microphysical entities are subject to law-like regularities
described by physics. Since microphysical entities constitute
everything, macro-entities are subject to laws which are
constituted by microphysical laws.
(3) Once microphysical entities and law-like regularities holding
them are fixed, all facts are fixed—metaphysically speaking
(Loewer 2001a, 39). This is implied by (i) and (ii).
(4) Higher level entities exist by being identical to or by
supervening upon physical entities. This thesis itself remains
tacit whether we can still be realistic about higher level
entities (Kornblith 1994, 42; Loewer 2001a, 46;
Hüttemann/Papineau 2005, 34).

5
Physicalism is an ambiguous term. Some philosophers call themselves physicalists
but in fact they reject only the acceptance of non-material substances in our world.
Such a version of physicalism is identifiable with what we call ‘liberal or
conciliable naturalism’. We think it is ill-founded to label such a view ‘physicalism’
as it creates more confusion than clarifications. Others, however, propose some
constrained notion of physicalism, which we aim at defining in points (1) to (6). We
leave it open which commitments the single philosophical tenets entail. The point is
rather, that everyone accepting tenets (1) to (6) faces the problem we expose in what
fallows.
 The Heavy Burden of Proof for Ontological Naturalism 171

(5) Causation of non-fundamental entities exists supervening

upon or depending on physical facts and laws. This thesis
does not rule out that higher level causation exists as well.
(6) Physicalim’s commitment to the thesis that higher level
entities are sums of or supervenient upon physical entities
leaves open whether higher level entities are only adequately
studied and interpreted via a reductionist methodology and
ontology.

Tenets (1) to (6) leave room for a variety of physicalistic positions. A

certain prevalence for microphysical entities and the objects studied by
physics can be noticed as general commitment of them all. But at this point
physicalism is still open to various interpretations reductionist and non-
reductionist alike.
Reductionists claim that all higher level entities are (in principle)
reducible to physical ones. According to reductionism, higher level entities
present no domain of their own but are reducible to and identical with
entities in the physical realm. In a final analysis, all that exists are physical
entities—whatever they may be—and sums of them.
Non-reductive physicalists reject a strong reading of (1) to (6). They
accept the existence of higher level entities in a genuine sense as well.6
According to non-reductive physicalists there are facts in the world that
simply cannot be stated or noticed in terms of lower level entities. It is the
failure of reduction because of the incompleteness of lower level ontology
that justifies the acceptance of irreducible higher level entities. These
higher level entities are asymmetrically dependent on the physical level.
This dependency-relation is mostly dubbed as supervenience relation.
Basically it says: No changes at the higher levels without changes at the
lower level. Two systems exemplifying exactly the same physical states
exemplify the same higher level states as well, but not the other way
around.
What should be noted at this point is that reductionists and non-
reductionists share a common worry: To leave out important features of
our world which we care about. If reductionists aim at reducing higher

6
For a detailed version of non-reductive physicalism see e.g. Poland 1994.
172 Georg Gasser & Matthias Stefan

level entities to lower ones then they do so for being realists about higher
level entities. By providing a reduction, higher level entities no longer
‘hover over’ the physical world but become a part of it. The existence and
causal efficacy of higher level entities is thus guaranteed by reducing them.
The worry of most reductionists is that entities which cannot be reduced
will prove to be epiphenomenal or, even worse, unreal altogether (Kim
1995). Exactly the same worry drives non-reductionists as well. They hold,
however, the opposite view of what counts as legitimizing higher level
entities. For non-reductionists reduction amounts to elimination of higher
level entities: If mental states are reducible to physical states, then there are
only physical states. Mental states become superfluous, ontologically
speaking. To protect the ontological status of higher level entities it has to
be shown that they are indispensable and irreducible. Any ontology leaving
them out or not granting them the full right to exist would be incomplete:
Important parts of our world would be missing.
The reason we stress this common worry of reductive and non-reductive
physicalists alike is to avoid a ‘straw man argument’ in the discussion.
Often anti-naturalists argue that naturalists do not ascribe great importance
to such crucial features of our self-conception as subjectivity or the first-
person-perspective. Such a reproach misses the mark. Many naturalists aim
at naturalizing these features via reductionist strategies. Only few consider
them as entirely eliminable or superfluous. Thus, reduction should not be
confused with elimination (see Kim 2005, 160). Naturalization via
reduction means to provide a home for ontologically disputable entities
such as mental entities within an ontologically undisputed realm such as
the physical.
Probably most philosophers fancying physicalism, subscribe to a
version of non-reductive physicalism. According to them one can go up the
hierarchy of levels and consider entities at higher levels as real without
being forced to assume new kinds of obscure metaphysical ingredients like
vital forces, entelechies or souls. Higher entities do not consist of physical
parts and something non-physical. All the entities being there are “physical
in nature”. Being physical in nature, however, does not imply that higher
level entities and their properties are reducible to the sums of physical
particles and their properties.
 The Heavy Burden of Proof for Ontological Naturalism 173

Thus, one might want to say that higher-level entities, such as human beings, are
real—as real as the entities that compose them—and at the same time reject all
sorts of vitalism or dualism. (Murphy 1999, 130)

Higher level entities supervene upon their physical basis but the
ontological significance of these entities is acknowledged because a
comprehensive reduction to their basis is excluded.
With the distinction of reductive and non-reductive physicalism in mind
it is easier to notice the conflict between (1) to (6). The conflict is to be
located between the ontological primacy of the physical level on the one
hand and a realist understanding of higher level entities on the other hand.
Perplexities arise when we ask how higher level entities ought to be
characterized ontologically. The discussion should have made clear that
non-reductive physicalists have to solve a major problem of their account:
The assumed dependency of higher level entities on the physical oscillates
between reductionism and breaking the seal of the physical domain: Either
dependency and supervenience is defined in such a way that higher level
entities are identical and consequently reducible to physical ones or
dependency is so weakly defined that the higher level domain gets a life on
its own (Kim 1995). Then it is hard to keep the higher level domain in
check within a physicalist framework. The problem for non-reductive
physicalism can be stated in the form of a dilemma which is similar to the
one of scientism in general: Non-reductive physicalism aims at interpreting
the world of our common sense realistically. For achieving this purpose it
considers mere dependency or supervenience relations on the physical as
sufficient. Thereby the physical level loses in importance and definiteness.
Non-reductive physicalism runs the risk to burst the physical realm
altogether and thus turning into a version of liberal naturalism. If, on the
contrary, the domain of the physical is kept restrictive, then non-reductive
physicalism seems to slide into reductive physicalism. In what follows, we
present an argument that tightens this assumption.

5. ONTOLOGICAL PHYSICALISM AND REDUCTIONISM

In various articles Kim argued at length that non-reductive physicalism is

an unstable house of cards. It is a promissory note between the poles of
174 Georg Gasser & Matthias Stefan

open mindedness and reductionist physicalism which has not been cashed
out yet. Kim reminds non-reductive physicalists to be consequent:
[…] if you have already made your commitment to a version of physicalism
worth the name, you must accept the reducibility of the psychological to the
physical […]. (Kim 1995, 134)

Kim’s argument concerns not only the mental but can be generalized for all
higher level entities since the mental is just one kind of higher level entity
among others (Loewer 2001b; Sparber 2005; especially Kim 2005, 52-56).
Before focusing on Kim’s argument itself, something needs to be said
about its premises: It is crucial for the argument that physicalists subscribe
to the principle of causal closure of the physical (CCP). CCP has been
stated in different constructions (e.g. Papineau 1993, 16f. and 29-32;
Armstrong 1995, 38; Papineau 2001; Kim 2005, 15f.). CCP says
something like the following: “At every time at which a physical state has
a cause, it has a fully sufficient physical cause” (Lowe 2000, 27). Every
physical state P which is caused at a certain time t has as its cause a set of
other physical states existing at this certain time t, such that: (i) each of
these states is a cause of P and (ii) together they are causally sufficient for
P (ibid.). Physicalists “worth the name” accept CCP equally. If CCP is
rejected, various kinds of non-physical entities could be accepted as causes
of physical states. One major problem with this assumption is that these
entities are not accessible to physics. According to most physicalists CCP
is a presupposition which is part of a reasonable interpretation of our
physical theories about the world. Otherwise physics could not be applied
to certain domains of our reality (e.g. Armstrong 1995, 38; Beckermann
2000).
Additionally physicalists accept a supervenience relation of higher level
entities to lower level ones. The supervenience relation is supposed to
grant that higher level entities are ‘bound’ to lower level ones. The
assumption of identity of higher level phenomena to physical ones is
perfectly compatible with the assumption of supervenience.
If these two premises are accepted, Kim’s argument briefly goes as
follows (Kim 1995, Kim 2005): For the easiness of illustration, we call any
arbitrary higher order state Mn and any arbitrary lower level state Pn: First,
higher order states are supervenient on lower level ones. If there are two
 The Heavy Burden of Proof for Ontological Naturalism 175

arbitrary higher level states M1 and M2, then there must also be two lower
level states P1 and P2. P1 and P2 are the supervenience bases of M1 and M2.
Suppose now, that M1 causes M2. If this is assumed, M1 would have to
cause P2 as well in virtue of which M2 appears. However, because of the
causal closure principle, P2 must also be caused by P1, while the
appearance of P2 allows M2 to exist. As the domain of the physical is
closed (according to the physicalist) M1 cannot by itself cause P2. Thus, P1
must cause P2. Therefore the causal chain from P1 to P2 and to the
supervenience of M2 is sufficient for M2 to appear. Since a physicalist
accepts the causal closure principle, the causal chain from P1 to P2 is not
only sufficient but furthermore the only acceptable one for him. If M1
really causes M2, and the causal chain leading to M2 starts with M1’s
supervenience base P1, then M1 must be identical with P1. Thus, taking
higher level causation seriously within a physicalist ontological
framework, this embraces the thesis that only reducible higher level
entities possess causal efficacy. The main purpose of the argument is to
show that the assumption of the supervenience relation together with CCP
lead to a determinate ontological commitment. This commitment states that
higher level phenomena make a difference only if—via the supervenience
relation—they are identifiable and hereby reducible to physical states.
Such a conclusion follows if causal over-determination (M1 and P1 each
fully sufficiently cause M2 at the same time t) on the one hand and partial
causation (M1 and P1 together partially cause M2 at t) on the other hand are
excluded. Assuming a (constant) causal over-determination would lead us
to a highly fragmented understanding of reality because various causal
histories for a single event would be equally true. Apart from this
undesirable result there is no empirical evidence for it (for a further
assessment of this assumption see Sparber 2005). Partial causation, on the
contrary, is excluded by CCP itself, as it states that every physical event
has a sufficient physical cause. Thus, a physicalist—by accepting CCP and
some kind of supervenience relation—faces pressing ontological reasons
for embracing as well the thesis that ontological relevant states at higher
levels are identical and thereby reducible to the states of the bottom level
of reality. CCP and supervenience seem to be intertwined insoluble with
the ontological commitment that all facts are fixed on the bottom level of
reality.
176 Georg Gasser & Matthias Stefan

What CCP does not rule out are epiphenomenal states. Epiphenomenal
states are higher level states which are not reducible to physical ones but
which are causally powerless. As such, epiphenomenal states remain
outside the realm of the physical. Nevertheless they are often considered to
be no serious threat to a physicalist framework as they are causally
ineffective.7 It might, however, be asked whether epiphenomenalism truly
can be dispatched so easily from physicalism. If epiphenomena are
causally irrelevant, how can we know anything about them? As we have
knowledge about them, they must somehow directly or indirectly affect us:
If a thing lacks any power, if it has no possible effects, then, although it may
exist, we can never have any good reason to believe that it exists. (Armstrong
1995, 40)

Any good reason for assuming the existence of an entity is due to its causal
efficacy, by whose virtue we get knowledge of it. So either we have a good
reason for assuming the existence of an entity, but then we must accept that
it affects us in some way and is no epiphenomenon anymore; or it is an
epiphenomenon but then we can really have no good reason for assuming
its existence. Epiphenomenalism seems to be a too cheap way out for
physicalism. A thoroughly coherent form of physicalism is pushed towards
complete reduction.
We do not want to dwell on this issue any longer because it launches a
discussion on its own. The decisive point we wanted to raise is whether it
can be shown that higher level phenomena are reducible to physical
entities or not. If not, we have to assume the existence of irreducible higher
level phenomena. Then, a thorough physicalist ontology is unable to
capture certain features of our world. Physicalism as a comprehensive
ontological program has failed.8 If it can be shown instead that no such
irreducible phenomena exist, reductionism is on its move. It is our
impression that many physicalists postpone the pressing answers how their
7
Kim for instance takes qualia as epiphenomenal states; see Kim 2005, 22-29, and
170-173.
8
An alternative possibility is simply to deny those entities which do not fit into a
physicalist picture. Melnyk 2003, 42f., explicitly denies all entities that cannot be
reduced. Whatever the merits of such a strategy are, Melnyk agrees with us that
reductionism must be comprehensive and thorough.
 The Heavy Burden of Proof for Ontological Naturalism 177

acceptance of CCP and/or their conception of supervenience are

reconcilable with an ontologically realistic understanding of a layered
model of reality. If our arguments are correct they pinpoint that a non-
reductive physicalist faces unavoidably conflicts between the assumptions
of different levels of reality: Questions concerning the causal closure of the
physical domain and the ontological overlapping or competition of higher
and lower level entities call for an answer.
At this point a proponent of the reductionist program might refer to
science’s future success of accomplishing complete reduction. Although
we are currently still remote from such a state in science’s development, a
glance at the history of science sheds hope: Successful programs of
reduction already have been carried out and further ones will follow. Thus,
it is legitimate to suppose that new forms of reduction will once be as
successful as it already has been proven to be in other realms of science.9
We do not think that referring to successful examples in the history of
science is really a convincing argument for a general reductionist outlook.
There are many phenomena tenaciously resisting reduction as
contemporary debates in philosophy of mind or philosophy of nature show.
Nor is it satisfying to talk about ‘reduction in principle’ or ‘reduction being
possible in the long run of science’. To refer to a future point in science’s
progress (when the program of reduction finally will be carried out) runs a
risk of becoming a mere strategy of immunisation in the face of
unsuccessful attempts of reduction. The same holds for the claim that
reduction is possible in principle, but cannot be executed in practice
because the entities which are to be reduced are too complex. The physicist
Falkenburg plausibly demands that successful reductionism has to be
carried out in a double way: It needs to analyse higher level entities into
physical ones (top down reduction) on the one hand and it has to be shown
how higher level phenomena result from its physical constituents (bottom
up reduction) on the other hand (Falkenburg 2006, 61-68).

9
See, for instance, Kim 2005, 68, citing McLaughlin 1992.
178 Georg Gasser & Matthias Stefan

6. CONCLUSION

Naturalism is a very popular philosophical position. We share its

underlying conviction that modern sciences help us to see many things
clearer, also in the field of ontology. But the problem is that there is no
direct way from the subject matter of sciences to ontology. The aim of our
paper was not to present new arguments in favour of naturalism or against
it. Its aim was to map the landscape of the current discussion of ontological
naturalism and to portray the ways a naturalist might want to go within it.
We pointed out that naturalism has no clear concept of science.
Therefore it is hard to tell from which sciences naturalists should or could
derive their ontology. Even if a list of acceptable sciences were provided,
serious questions would remain: How do the ontological implications of
the accepted sciences relate to each other? And do some sciences have a
higher priority in determining ontology than others?
According to our analysis the most promising way for naturalists to
elaborate a science-based, uniform, and coherent ontology is physicalism.
Many of the contemporary physicalistic positions, however, accept CCP
and supervenience—the core assumptions of physicalism—on the one
hand but reject reductionism on the other hand. We do not see how this
wish-list of non-reductive physicalism can be brought together
consistently. A physicalist is committed to reductionism, as Loewer10
unwittingly summarizes our conclusion:
[…] philosophers true to their physicalism will have to swallow reductionism.
Those who find reductionism impossible to swallow will have to find a way of
living without physicalism. (Loewer 2001b, 315)

Some naturalists might have the impression to face another ‘straw man
argument’ of naturalism being fought here. But we aimed at interpreting
naturalism benevolently. We neither rejected its allegiance to science nor
did we claim naturalism to be a failure as a philosophical program. We
rather wanted to press the naturalist to consider more carefully the

10
Loewer himself wants to avoid this conclusion—against his line of argument see
Sparber 2005.
 The Heavy Burden of Proof for Ontological Naturalism 179

ontological commitments she enters by taking science as authoritative

guidance for philosophy in general and ontology in particular.
We do not know where science will lead us and whether naturalism will
have a better standing in the future. Contemporary naturalism, however,
carries a heavy burden: If it wants to be successful, it either has to show
how reductionism is possible, or it has to point out convincing ways for
manoeuvring between reductionism and too liberal versions of naturalism.

7. REFERENCES

Armstrong, D. 1995: “Naturalism, Materialism, and First Philosophy.” In:

Moser, P. K. and Trout J. D. (eds.) Contemporary Materialism. New
York: Routledge, 35-47.
Beckermann, A. 2000: “Ein Argument für den Physikalismus.” In. Keil, G.
und Schnädelbach, H. (eds.) Naturalismus. Frankfurt am Main:
Suhrkamp, 128-143.
Clayton, P. 2004: Mind and Emergence. From Quantum to Consciousness.
Oxford: Oxford University Press.
Crane, T. and Mellor, H. 1995: “There is no question of physicalism.” In:
Moser, P. K. and Trout J. D. (eds.) Contemporary Materialism. New
York: Routledge, 65-89.
Dupré, J. 2004: “The Miracle of Monism.” In: De Caro, M. and Macarthur,
D. (eds.) Naturalism in Question. Cambridge, MA: Harvard
University Press, 36-58.
Falkenburg B. 2006: “Was heißt es, determiniert zu sein? Grenzen
naturwissenschaftlicher Erklärung.” In: Sturma, D. (ed.) Philosophie
und Neurowissenschaften. Frankfurt am Main: Suhrkamp, 43-74.
Forrest, P. 1996: God without the supernatural: A defense of scientific
theism. Ithaca: Cornell University Press.
Hüttemann, A. and Papineau, D. 2005: “Physicalism decomposed.”
Analysis 65, 33-39.
Keil, G. 2000: “Naturalism in the philosophy of mind—and what is wrong
with it.” In: Nannini, S. and Sandkühler, H. J. (eds.) Naturalism in the
Cognitive Sciences and the Philosophy of Mind. Frankfurt am Main:
Peter Lang, 145-155.
180 Georg Gasser & Matthias Stefan

Keil, G. and Schnädelbach, H. 2000: “Naturalismus.” In: Keil, G. and

Schnädelbach, H. (eds.) Naturalismus. Philosophische Beiträge.
Frankfurt am Main:Suhrkamp, 7-45.
Kim, J. 1995: “The Myth of Nonreductive Materialism.” In: Moser, P. K.
and Trout J. D. (eds.) Contemporary Materialism. New York:
Routledge, 133-149.
Kim, J. 2005: Physicalism or something near enough. Princeton, NJ:
Princeton University Press.
Koppelberg, D. 2000: “Forms of Naturalism in the Philosophy of Mind.”
In: Nannini, S. and Sandkühler, H. J. (eds.) Naturalism in the
Cognitive Sciences and the Philosophy of Mind. Frankfurt am Main:
Suhrkamp, 11-29.
Kornblith, H. 1994: “Naturalism: Both Metaphysical and
Epistemological.” Midwest Studies in Philosophy 29, 39-52.
Loewer, B. 2001a: “From Physics to Physicalism.” In: Loewer, B. and
Gillet, C. (eds.) Physicalism and its Discontents. Cambridge:
Cambridge University Press, 37-56.
Loewer, B. 2001b: “Review of Jaegwon Kim, Mind in a Physical World.
An essay on the mind-body problem and mental causation.” Journal
of Philosophy 98, 315-324.
Löffler, W. 1999: “Naturalisierungsprogramme und ihre methodologischen
Grenzen.” In: Quitterer, J. and Runggaldier, E. (eds.) Der neue
Naturalismus. Köln: Kohlhammer, 30-76.
Lowe, J. E. 2000: Introduction to the philosophy of mind. Cambridge:
Cambridge University Press.
McLaughlin, B. 1992: “The Rise and Fall of British Emergentism.” In:
Beckermann, A.; Flohr, H. and Kim, J. (eds.) Emergence or
reduction?: Essays on the prospects of Nonreductive Physicalism.
Berlin, New York: De Gruyter, 49-93.
Melnyk, A. 2003: A physicalist manifesto: thoroughly modern materialism.
Cambridge: Cambridge University Press.
Murphy, N. 1999: “Nonreductive Physicalism. Philosophical Issues.” In:
Brown, W. S., Murphy, N. and Malony, N. (eds.) Whatever Happened
to the Soul? Scientific and Theological Portraits of Human Nature.
Minneapolis, MN: Fortress Press, 127-148.
 The Heavy Burden of Proof for Ontological Naturalism 181

Papineau, D. 1993: Philosophical Naturalism. Oxford: Oxford University

Press.
Papineau, D. 2001: “The Rise of Physicalism.” In: Loewer, B. and Gillet,
C. (eds.) Physicalism and its discontents. Cambridge: Cambridge
University Press, 3-36.
Pettit, P. 1993: “A definition of Physicalism.” Analysis 53, 213-223.
Poland, J. 1994: Physicalism. The Philosophical Foundations. Oxford:
Clarendon Press.
Rea, M. 2002: World Without Design: The Ontological Consequences of
Naturalism. Oxford: Oxford University Press.
Sparber, G. 2005: “Counterfactual overdetermination vs. the causal
exclusion problem.” History and Philosophy of the Life Sciences 27,
479-490.
Stroud, B. 2004: “The Charm of Naturalism.” In: De Caro, M. and
Macarthur, D. (eds.) Naturalism in Question. Cambridge, MA:
Harvard University Press, 21-35.
Sukopp T. 2006: Naturalismus. Kritik und Verteidigung
erkenntnistheoretischer Positionen. Frankfurt am Main: Ontos.
Tetens, H. 2000: “Der gemäßigte Naturalismus der Wissenschaften.” In:
Keil, G. and Schnädelbach, H. (eds.) Naturalismus. Philosophische
Beiträge. Frankfurt am Main: Suhrkamp, 273-288.