Académique Documents
Professionnel Documents
Culture Documents
Scientific
Composition and
Metaphysical Ground
Editors
Kenneth Aizawa
Department of Philosophy
Rutgers, The State University of New Jersey
Newark, New Jersey, USA
Carl Gillett
Department of Philosophy
Northern Illinois University
DeKalb, Illinois, USA
(eBook)
vi
Acknowledgments
We would like to thank Dr. Jan Lewis, Dean, Faculty of Arts and Science,
Newark, and Dr. Raffaella De Rosa, Chair, Department of Philosophy,
both of Rutgers University, Newark, for financial support for the
Composition and Ground workshop held at Rutgers University, Newark,
April 1011, 2015. The papers at that workshop formed the nucleus of
the present volume. We would also like to thank Kit Fine and Ned Block
for contributing papers and commentary to the workshop, though they
could not contribute to this volume.
vii
Contents
39
41
75
91
ix
Contents
123
141
143
171
205
249
271
Index
301
Notes on Contributors
xii
Notes on Contributors
calism, realization and multiple realization, reduction and emergence, and other
topics related to scientific composition.
Jens Harbecke is Professor of Theoretical Philosophy and Philosophy of the
Social Sciences at Witten/Herdecke University, Germany. He holds a PhD from
the University of Lausanne under the supervision of Michael Esfeld with a dissertation on the problem of mental causation. Subsequently, he worked as an
assistant professor and visiting scholar at the universities of Bern, Tel Aviv,
Jerusalem, and Washington University in St. Louis. He works on constitutive
explanations, causality in the metaphysics of mind, and counterfactual and regularity theories of causation. He is the project coordinator of a research project
funded by the European Commission on the philosophy of social science and
neuroscience with researchers at Witten/Herdecke, Helsinki, and Louvain-laNeuve. He also collaborates as a principal investigator within a philosophical
research project funded by the German-Israeli Foundation on causation and
computation in neuroscience with partners in Jerusalem and Cologne.
Andrew Melnyk is Professor of Philosophy at the University of Missouri,
where he has taught since 1991. He is interested in all aspects of the philosophy
of mind and in many aspects of philosophy of science. Much of his work is unified by his attempt to formulate, explore, and argue for a comprehensive thesis
of physicalism that invokes a carefully defined relation of realization. He has also
written about other minds, naturalism in philosophy, conceptual analysis, and
the inference from conceivability to possibility. His work has appeared in Journal
of Philosophy, Nos, Philosophy and Phenomenological Research, Mind, Australasian
Journal of Philosophy, Philosophical Studies, Philosophy of Science, and Synthese,
among others. His book, A Physicalist Manifesto: Thoroughly Modern Materialism,
was published in 2003. Melnyk was educated at St. Pauls School, London, and
Oxford University.
AlyssaNey is Associate Professor of Philosophy at the University of California,
Davis. She works in metaphysics, the philosophy of physics, and the philosophy
of mind. She is the author of Metaphysics: An Introduction (2014) and co-editor
of The Wave Function: Essays in the Metaphysics of Quantum Mechanics (2013).
DerkPereboom is the Susan Linn Sage Professor of Philosophy and Ethics at
Cornell University. He works mainly on philosophy of mind and on free will.
The physicalist position Pereboom proposes in philosophy of mind, initially
defended in a number of articles, is set out in detail in Consciousness and the
Prospects of Physicalism (2011). There he develops two responses to the leading
Notes on Contributors
xiii
arguments against physicalism. The first exploits the open possibility that introspective representations fail to represent mental properties as they are in themselves; more specifically, that introspection represents phenomenal properties as
having certain characteristic qualitative natures, which these properties might
actually lack. The second response draws on the proposal that currently unknown
intrinsic properties provide categorical bases for known physical properties and
also yield an account of consciousness. The books final theme is an anti-reductionist account of physicalism. In the view Pereboom defends, the fundamental
relation between the mental and the microphysical is constitution, where this
relation is not explicated by the notion of identity. Perebooms views on free will
are developed in Living without Free Will (2001), Four Views on Free Will (2007),
Free Will, Agency, and Meaning in Life (2014), and in his articles on this issue.
His overall position is that due to general facts about the nature of the universe,
we lack the sort of free will required for moral responsibilitythat is, for our
deserving, in a fundamental sense, to be blamed or punished for immoral action,
and to be praised or rewarded for morally right action. At the same time, he
contends that a conception of life without this type of free will would not undermine morality or to our sense of meaning in life, and in certain important
respects it may even be beneficial.
Jonathan Schaffer is a Distinguished Professor of Philosophy at Rutgers
University, New Brunswick. He holds a PhD from Rutgers in 1999, and has
since worked at the University of Houston, the University of MassachusettsAmherst, the Australian National University, and St. Andrews, before returning
to Rutgers in 2011. He works mainly in metaphysics, with interests in epistemology, language, mind, and science. His On What Grounds What (2009) is
widely regarded as one of the main sources of current interest in metaphysical
grounding.
Jessica Wilson is Associate Professor of Philosophy at the University of
Toronto. Her primary research interests are in general metaphysics (especially
metametaphysics, modality, and indeterminacy) and the metaphysics of science
(especially inter-theoretic relations). Wilsons recent publications include What
is Humes Dictum, and Why Believe It? (Philosophy and Phenomenological
Research, 2010), Fundamental Determinables (Philosophers Imprint, 2011), A
Determination-based Account of Metaphysical Indeterminacy (Inquiry, 2013),
and No Work for a Theory of Grounding (Inquiry, 2014). Her book
Metaphysical Emergence is forthcoming with Oxford University Press.
1
Introduction: Vertical Relations in Science,
Philosophy, and the World: Understanding
the New Debates over Verticality1
1
KennethAizawa andCarlGillett
For comments on an earlier draft, we are grateful to Geoff Pynn, Jonathan Schaffer, and Craig
Warmke.
2
One might think we could use the term non-causal relations, but as we outline below this would
beg some questions since one approach to verticality in nature takes it to be causal or of a kind with
causation. As we outline below, we also do not use the terms Grounding or realization, and so
on, because each of these terms is associated with one of the competing accounts of the nature of
verticality. We therefore plump for the neutral expression vertical relation.
K. Aizawa ()
Department of Philosophy, Rutgers University, Newark, NJ, USA
C. Gillett
Northern Illinois University, DeKalb, IL, USA
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_1
To take just one example, there is the tradition in analytic metaphysics built around formal mereological systems adapted from work on sets. The latter tradition, like the Grounding traditions,
allows that the relata of vertical relations may be causally inert, so the points we make below for the
Grounding tradition also apply to this approach and show it offers one more competing kind of
V-framework for certain object phenomena involving verticality.
6
The workshop took place at Rutgers, Newark, in April 2015, and in addition to the papers collected in this volume also included presentations by Ned Block and Kit Fine.
approaches all make competing claims, for example, about scientific conceptions of verticality and/or the vertical relations in nature.
Inter-twined with the latter disputes, we also have competition between
accounts pitched as treatments of the genus of verticality, such as some
Grounding V-frameworks, and other accounts focused on particular species of verticality, like many neo-Functionalist V-frameworks. Initially, one
might suppose that accounts of genus and species are pursuing different,
though interconnected, set of issues. However, in our new debates various
accounts focused on genus and species are each often intended by their
proponents to answer the same questionsthus still ending up as rivals.
For example, neo-Functionalists take their V-framework to articulate the
species of vertical relation that is best taken to back compositional explanation in science or underpin the formulation of physicalism. But proponents
of Grounding V-frameworks take their accounts to illuminate a genus relation of verticality that they in turn argue is the relation best taken to back
compositional explanation or formulate physicalism. So, we have debates
not just between V-frameworks offering different accounts of the vertical
relations in same object phenomena, but also differing over whether species or genus relations provide the best accounts of such phenomena.
Furthermore, each of these competing V-frameworks has also used to
offer distinct accounts of various issues connected to verticality. For example, with regard to verticality in nature, in our new debates we have distinct treatments of physicalism, reduction versus emergence, and more.
In the remainder of the Introduction, we focus primarily on the differing
V-frameworks themselves, but we highlight where disputes over applications
are important and a number of coming Chapters address such applications.
Finding that we have competing V-frameworks leads to a range of
meta-methodological changes. Since theory appraisal is comparative, for
instance, one can only ultimately justify ones favored account of verticality after comparing it to relevant rivals. It is thus no longer the case that
metaphysicians can only read other metaphysicians, letalone only writers
in their research tradition; or that philosophers of science can stick to
work by other philosophers of science; and so on. To justify ones favored
account of verticality with regard to some object phenomenon one must
critically engage rival accounts from different areasand ones account is
not justified until one shows ones V-framework is better than these rivals.
the same vertical relations, and/or offer conflicting answers to some of the
same questions about phenomena involving verticality, and these distinct
V-frameworks are rivals.
We should mark that some V-frameworks from different areas of philosophy, when used to pursue certain projects, may well be compatible
or even complementary. Below we give some examples. But we focus on
COMPETITIVE, since it corrects unhealthy assumptions or practices in
earlier debates and has substantial consequences.
One implication of COMPETITIVE is that researchers on verticality
can no longer continue to pursue the siloed approach. Theory appraisal
and justification are always comparative where a theory is assessed by
comparison to its relevant rivals. So defenders of V-frameworks need to
acknowledge and engage rival accounts from other areasand then show
their V-frameworks are better accounts of the relevant object phenomena
than these competing treatments. So a second thesis is also plausibly true
about the new phase of discussion:
(COMPARATIVE) A V-framework, offered as an account of certain object
phenomena involving verticality, is justified only if it is shown to provide a
better account of the object phenomena than its relevant rivals regardless of
the area of philosophy in which these rivals are offered.
and so on, while other accounts focus on vertical relations in the world or
some portion of it, and some accounts seek to do both. Whatever object
phenomena a V-framework is intended to cover, the primary work of this
account is to accommodate the nature, and various features, of the object
phenomena and related evidence such as the characteristics of associated explanations and formal systems. A corollary of this requirement is
that prior work is also necessary, since one must consequently provide a
detailed treatment of the nature of the specific vertical relations involved
in the object phenomenon under consideration, the characteristics of
associated explanations, and so on.
The new debates therefore address specific object phenomena in a
range of important ways. And we are therefore led to a third thesis about
our new discussions:
(FOCUSED) A central, and necessary, element in the assessment, and justification, of a V-framework offered as an account of certain examples of
object phenomena involving verticality is whether this account does, or
does not, capture the features of the relevant vertical relations and/or associated characteristics of the successful explanations that posit them and/or
other relevant evidence.
The basic idea of the approach framed in META-SUCCESS is the contention that successful explanations are best explained by these explanations
being true. Consequently, the conceptions of verticality in the successful
explanations about area X are taken to provide a veridical picture of the
vertical relations in area X of the world itself. In coming sections, we will
fill out the nature of META-SUCCESS, and our other theses, and offer
support for these claims and our overall picture of the new debates over
verticality.
10
I. What was the genesis of the research tradition offering this kind of
V-framework?
II. Who are some of the main proponents of this kind of V-framework
and what are some of the competing options within the research
tradition?
III. What object phenomena involving verticality are the intended focus of
the V-frameworks in the tradition?
And:
IV. What are some of the central features of verticality as it is characterized by the V-frameworks offered in this research tradition?
With these questions in mind, we start with philosophy of science, in
2.1, by sketching the rise of the New Mechanism and what we term the
neo-Causal research tradition that has grown along we dub neo-Functionalist V-frameworks in the metaphysics of mind, and philosophy of
mind/psychology, that now serve as both accounts of verticality in nature
and also increasingly figure in work on various issues in the philosophy
of science. Lastly, in 2.3, we look at the Grounding V-frameworks in
analytic metaphysics intended to provide universal accounts of all vertical
relations. We briefly summarize our findings, in 2.4, focusing in particular on how they support COMPETITIVE.
11
12
few accounts offered within the New Mechanism of these key vertical
relation taken to posit in this way in mechanistic explanations.7
The latter situation has been increasingly addressed by writers in what
we term the neo-Causal research tradition on verticality whose driving
idea is that accounts of causation can be tweaked and applied to scientific
composition, or that techniques that offer sufficient conditions for causal
relations can be tweaked to supply sufficient conditions for scientific
composition.8 The central idea of neo-Causal V-frameworks for scientific
composition is thus that, very crudely put, compositional or constitutive
relations are either identical to causal relations or of a kind with causal
relations. Let us briefly note some of the versions of neo-Causalism.
The most prominent neo-Causal approach is offered in Craver (2007)
who is struck by the success of recent interventionist accounts of causation (Woodward 2003) that Craver takes to offer a way to avoid what
he thinks are the intractable ontological debates over the nature of
causation. Craver thus takes the manipulability relations outlined in
interventionism to offer a sufficient condition for causation, while leaving its ontology open. Consequently, Craver suggests that manipulability
relations drawn from the interventionist framework, with a number of
alterations, can provide a sufficient condition for the existence of constitutive or compositional relations between processes posited in mechanistic explanations, while again avoiding what Craver apparently takes to be
intractable ontological debates.
Strictly put, Craver is therefore not offering the standard kind of
V-framework, but only a sufficient condition for the vertical relations
posited in mechanistic explanations. Nonetheless, Craver still claims that
manipulability relations, in fact relations of mutual manipulability, suffice for vertical relations between processes of the kind posited in successful scientific explanations. Although amended in various ways, Craver
thus explicitly contends that just as one species of manipulability relation
7
See Kaiser and Krickle (2016) for a recent survey of such work.
A number of neo-Causal accounts often include conditions that compositional relations are not
causal, or hold between entities that are not logically distinct, but these accounts then all still go on
to use the machinery developed for causal relations with these conditions added to the machinery.
We therefore call these neo-Causal accounts and contend they treat causation and scientific composition as being of a kind.
8
13
suffices for causation, so another kind of manipulability suffices for scientific composition relations between processes.
A variety of other neo-Causal accounts of scientific composition
have subsequently sprung up. However, unlike Cravers account, these
V-frameworks are explicitly intended to provide ontological accounts
of either the conceptions of composition posited in compositional
explanations and/or the vertical relations found between entities in
nature. These versions of neo-Causalism take existing accounts of
causation, tweak these accounts in various ways, and then offer the
resulting machinery as a V-framework for scientific composition. For
example, Jens Harbecke (2010, 2014a, 2014b) has developed a neoCausal V-framework by adapting regularity accounts of causation, and
Mark Couch (2011) has used Mackies INUS-based treatment to produce another neo-Causal V-framework. Still more explicit neo-Causal
approaches, such as that of Totte Harinen (2014), also now exist which
claim that scientific composition relations are quite literally identical to
causal relations.
To summarize, Neo-Causal approaches to verticality have grown out
of debates over the nature of a species of compositional explanation in
mechanistic explanation. Neo-Causal accounts are therefore intended
to offer V-frameworks for the conceptions of verticality used in certain
successful scientific explanations, and/or vertical relations in nature, but
are not intended to cover all vertical relations such as those posited in
mathematics or found in the abstract realm. The motivating thought
behind neo-Causal V-frameworks is that the vertical relations we find in
the sciences, and/or nature, are of a kind with causal relations, although
neo-Causalists diverge over what this involves. Neo-Causal accounts
therefore take vertical relations in scientific explanations, and/or nature,
either to be identical to relations of causation, regularity, mutual manipulability, sufficiency or counterfactual dependence, or take such relations
to suffice for such vertical relations.
14
15
16
17
10
For introductory expositions, see Bliss and Trogdon (2014), Clark and Liggins (2012), or Raven
(2015).
18
ers like Schaffer take its relata to be unconstrained and hence also to hold
between entities such as individuals, properties, and more.
It is important to note many writers take Grounding V-frameworks to
be articulating a genus relation covering various species of vertical relation. However, it is equally significant that proponents of Grounding
V-frameworks, such as Fine or Schaffer, also take this genus relation, in
Grounding, to be the relation that backs all metaphysical or constitutive explanationhence implying Grounding relations underpin the
compositional explanations we find in science. Proponents of Grounding
have also claimed it is the relation that underpins the formulation of
physicalism.
To summarize, Grounding V-frameworks have been developed
by writers in analytic metaphysics who have taken vertical relations
between entities of logic or mathematics as their exemplars, developed
V-frameworks for such relations, and then extended such Groundingbased V-frameworks to all vertical relations. Though Grounding
V-frameworks are often pitched as general accounts, articulating a
genus relation, it is also argued that all constitutive explanations are
best understood as being underpinned by this general relation. The
Grounding V-frameworks are thus intended to cover vertical relations
wherever they are found from logic and mathematics across to science
and nature. Among other characteristics, Grounding V-frameworks are
marked by allowing vertical relations to have causally inert entities as
relata.
19
11
As will become clearer below, Schaffer (2016, this volume) offers a Grounding V-framework that
may be a species of neo-Causalism. Nonetheless, other Grounding accounts still conflict with neoCausal views accounts of the relation backing compositional explanation, and all Grounding
accounts offer rivals to neo-Functionalist views.
20
on. Let us therefore push on to explore the nature of the new discussions
that consequently ensue.
21
12
Given the latter points, there a number of different projects a Global Account (see below) could
be intended to pursue. One Global Account defends the claim that a V-framework captures the
nature of vertical relations as they appear in all successful explanations positing verticality and that
this V-framework also covers all vertical relations found in the world. However, one could also
offer a Global Account as a V-framework just covering all vertical relations in the world, but be
neutral about whether ones account covers conceptions of verticality or even accept that other
V-frameworks offer the best accounts of the concepts of verticality used in some successful
explanation.
22
23
24
25
26
27
28
The central idea is the one we briefly marked earlier, namely the idea
that the best account of the success of an explanation is that the explanation is true or likely to be true. Consequently, successful explanations
about area X that posit vertical relations are likely true and so the conception of vertical relations used in these explanations either are, or are likely
to be, veridical representations of verticality in area X.We can therefore
see that there is a clear and compelling meta-justification for constructing accounts of verticality in some part of the world using our successful explanations about this portion of realityhence bringing us to the
approach we earlier framed in META-SUCCESS. We can thus begin
to see why it is no surprise to find V-frameworks increasingly offered as
Local accounts of both the conceptions of verticality offered in successful
explanations about domain X and also as accounts of the vertical relations
in domain X in the world.
We should note that META-SUCCESS is a meta-justification of
accounts of vertical relations in the world. V-frameworks offered in pursuit of other projects, such as illuminating this or that conception of
verticality, will likely pursue different methodologies and have different
meta-justifications. We should also note that we have framed METASUCCESS solely as a sufficient condition for a meta-justification of an
account of verticality in the world. However, we know of no other viable
meta-justifications of such accounts at present. So it is very much a live
question whether META-SUCCESS, or some version of it, should be
framed as a necessary and sufficient condition.
29
30
als (or entities) and their processes (or activities). However, Aizawa
argues that the relata of the relevant vertical relations, and hence also
list of explananda and explanantia of mechanistic explanation, should
be expanded to include properties as well as individuals and processes.
Aizawa makes his case, in a manner following FOCUSSED, by looking at concrete scientific cases where individuals have properties that are
explained by properties of their parts. In making his case, Aizawa assumes
that the vertical relations between posited in the mechanistic explanation of properties are better characterized by the theory of Dimensioned
realization (Gillett, 2002, 2003), but Aizawa does not argue for this as
an alternative account of the compositional relation between properties.
As well as these contrasts with the New Mechanism, Aizawa presses the
point that scientific composition is not merely a logical or quasi-logical
relation. Aizawa contends instead that scientific composition involves
a non-logical, non-causal form of natural necessitation of the NeoFunctionalist variety outlined in Aizawa and Gilletts V-framework. And,
again following FOCUSED, Aizawa uses concrete cases of compositional
explanation to highlight how Grounding accounts do not accommodate
key features of scientific composition and also compositional explanation. In particular, Aizawa focuses on what he terms the love this expression of compositional explanation, in its feature of explaining entities
of one kind using qualitatively different kinds of entity, and argues that
Grounding frameworks fail to accommodate this key feature.
Jens Harbeckes Chapter, Is Mechanistic Constitution a Version of
Material Constitution, lays out Harbeckes sophisticated neo-Causal
V-framework for the vertical relations deployed in compositional
explanations in the sciences (Harbecke 2010, 2014a, b). Like Craver,
Harbecke begins with the thought that constitution is like causation,
but rather than starting with manipulability Harbecke begins with a
Humean regularity conception of causation and then transforms this
causal foundation into a constitutional relation by adding conditions
from classical mereology. The main work of Harbeckes Chapter is to
explore how his account of such mechanistic constitution relates to the
relation, which Harbecke terms metaphysical constitution, that we find
in puzzle cases in metaphysics such as the statue and the clay. Harbecke
defends a number of interesting substantive theses about the relations of
31
mechanistic and metaphysical constitution and draws out their implications for a number of long-standing debates and positions in philosophy.
Broadening the discussion of accounts of vertical relations in science
and nature still further, Derk Perebooms paper, Anti-Reductionism,
Anti-Rationalism, and the Material Constitution of the Mental, makes
positive and negative contributions to the ongoing disputes between
competing Neo-Functionalist V-frameworks outlined briefly in 2.2
above. On the negative side, Pereboom further develops his important
critique of Subset/Flat accounts of vertical relations between properties
in nature. In past work, Pereboom (2011) has argued that under the
Flat/Subset view of Sydney Shoemaker (2001) or Jessica Wilson (1999)
the powers of realized and realizer property instances must be identical.
Perebooms paper now extends his critique by arguing that latter feature
means that realization under the Flat/Subset view is not the kind of asymmetric vertical relation that we plausibly find in nature.
On the positive side, Pereboom outlines his favored approach to the
vertical relations between individuals in nature developed in his previous
work (Pereboom 2011). Pereboom defends his account against a number of objections ranging from in-principle concerns about his primitive
made-up-of relation to Andrew Melnyks concern that Perebooms position is committed to physicalistically unacceptable posits. In response,
Pereboom outlines why his position does everything a naturalist would
require without falling into what he argues are the overly strong demands
of his critics.
As outlined above, the Chapters of Part II are all focused, to lesser or
greater degrees, on Grounding accounts. To begin the section, there is a
pair of papers that continue one prominent thread of the new debates
started by Wilson (2014). The final three Chapters consider, among other
issues, whether Grounding V-frameworks are suited to certain projects.
Jonathan Schaffers paper, Ground Rules: Lessons from Wilson,
explicitly engages Wilson (2014) which, as we noted earlier, presses
a broad critique of the utility and informativeness of Grounding
V-frameworks. Schaffer draws out what he takes to be two important
lessons from Wilsons critique, but then argues that, suitably modified,
his favored Grounding V-framework has learned these lessons and that
Wilsons favored account has not. The first lesson Schaffer draws from
32
Wilson is that an account of Ground must not settle for bare Grounding
claims, such as X grounds Y, but must also make sense of the follow-up
inquiry into how a vertical relation makes sense of a certain explanatory
connection. The second lesson Schaffer draws from Wilson is that the
best account of certain object phenomena involving verticality should
follow a formalism that successfully unifies the putative features of verticality. We consequently see Schaffer offering ideas about how to articulate both FOCUSED and META-SUCCESS. Furthermore, Schaffer
argues that structural equation modelling, a technique drawn from the
sciences, provides the best formalism for verticality and then argues that
his Grounding V-framework is the account favored by this formalism.
On the other side, Schaffer argues that Wilsons favored account fails to
live up to her two lessons.
Jessica Wilsons Chapter challenges the adequacy of Schaffers new
Unity Argument for Grounding, then revisits a further Priority
Argument for Grounding. She summarizes Schaffers Unity Argument
as follows:
1. If some phenomena are aptly formally unified, then this provides
strong (albeit defeasible) reason to posit a unifier.
2. The diverse (small-c) causal relations are aptly formally unified by the
SEM framework.
Therefore, there is strong (albeit defeasible) reason to posit a general
notion of causation.
3. The diverse (small-g) grounding relations are just as aptly formally
unified by the SEM framework as the diverse (small-c) causal
relations.
Therefore, there is strong (albeit defeasible) reason to posit a general
notion of Grounding.
Wilson then provides multiple arguments that each of the premises
in Schaffers argument is false. One of her points against premise 1 is
that determinables are metaphysically and formally unified, but this has
not blocked the common view that determinables are schematic for or
reducible to determinates. Wilson thus challenges Schaffers approach to
33
34
35
References
Aizawa, K. (2007). The biochemistry of memory consolidation: A model system
for the philosophy of mind. Synthese, 155, 6598.
Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative.
Studies in History and Philosophy of Science Part C: Studies in History and
Philosophy of Biological and Biomedical Sciences, 36(2), 421441.
36
37
38
Part I
Scientific Composition
and the New Mechanism
2
New Mechanistic Explanation
andtheNeed forExplanatory
Constraints
L.R.Franklin-Hall
Introduction
In the past decade and a half, a new movement (Glennan 2005: 443)
has arisen in the philosophy of biology, one called a revolution (Bechtel
2006: 280) with broad implications (ibid: 2) and which has met with
broad consensus (Campaner 2006: 15). On this hot topic (Robert
2004: 159), a vast literature has developed, within it one of the most
cited papers in Philosophy of Science (viz. Machamer etal. 2000).
What is the subject of such attention? It is the new mechanistic philosophy (Skipper and Millstein 2005: 327), articulated by a group of
philosophersincluding William Bechtel, Carl Craver, Lindley Darden,
Peter Machamer, and Stuart Glennaninterested in the nature of mechanisms, complex systems characterized most prominently as entities or
L.R. Franklin-Hall ()
Department of Philosophy, New York University, New York, NY, USA
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_2
41
42
L.R. Franklin-Hall
activities organized such that they are productive of regular changes from
start or set-up to finish or termination conditions (Machamer et al.
2000: 3). Mechanisms are said to be worthy of attention largely because
they are central to a new and superior approach to scientific explanation, one truer to scientific practice than the long defunct deductivenomological (DN) view. It is also claimed that the mechanistic approach
has implications beyond explanation, as it transforms how one thinks
about a host of other issues in the philosophy of science (Bechtel and
Abrahamsen 2005: 426), including causality, laws, kinds, reduction, discovery, and scientific change.
Philosophical movements can be judged by their fruits. We can ask
of them: what problems does a movement offer solutions to? Judging
by both the language of the new mechanists and the influence of their
work, it would appear that the mechanistic approach had served up a
bevy of solutions. Yet I argue here that, at least with respect to its core
projectthat of elucidating the nature of scientific explanationappearances are deceptive: the mechanisms movement has not yet yielded the
advertised results. This is not because mechanisms advocates are committed to claims that are false. My critique is motivated instead by concerns that mechanistic explanatory accounts offered to dateeven in
their strongest formulationshave failed to move beyond the simple and
uncontroversial slogan: some explanations show how things work. In
particular, I argue that proposed constraints on mechanistic explanation
are not up to the task required of them: namely, that of distinguishing
acceptable explanations from those that, though minimally mechanistic,
are uncontrovertibly inadequate.
Sections The Mechanistic Explanatory Framework and Formulating
Explanatory Constraints sketch a version of the new mechanistic
explanatory account, one constructed by combining the most promising proposals from across the mechanistic corpus. After articulating three
principles at the heart of this pictureconcerning causation, parts, and
explanatory levelsections The Causal Standard through The Levels
Standard argue that these principles remain promissory notes. The chapter concludes in section Conclusion with an evaluation of the mechanistic explanatory program.
43
Beyond token capacities and events, mechanists also aspire to treat regularities. Though the details
are rarely made explicit, a given regularity can be explained via a mechanistic model jointly applicable to all of the particular systems underpinning a regularitys instances; to do this, such a model
must be at least somewhat abstract. For a discussion of how this might work, see Strevens First
Fundamental Theorem of Explanation (2008: Chap. 7).
44
L.R. Franklin-Hall
45
accountsthe other three will appear flawed. They each make a distinct
variety of explanatory error and can illustrate in the breach the constraints
that a mechanistic model must fulfill to be explanatorily acceptable.4
According to the Standard Model, a neurons capacity to release neurotransmitters when exposed to them is explained by describing the neuron as composed of a variety of somewhat organized macromolecular
parts, including membranes, channels within them, and ionic concentrations in the internal and external environmentall of which interact
according to dynamic principles, such as one stating that neurotransmitter
binding is followed by channel opening.5 Though these details could be
communicated in a variety of ways, they are most often presented in narrative form, as follows: neurotransmitter exposure leads neurotransmitter molecules to bind to ligand-gated receptors located in the dendrite
membrane. Upon binding, these channels open. Then sodium ions rush
into the cell, depolarizing the membrane locally. Next, a population of
voltage-gated membrane channels, located in the same region, also open
and more sodium enters. This begins a cascade of channel opening, depolarization, and further channel opening, that moves up the neuron until
it reaches the neurons axon terminal where voltage-gated calcium channels open and calcium enters the cell. Finally, vesicles containing neurotransmitters located near the axon terminal bind with the membrane,
releasing neurotransmitters to the extracellular environment.
To formulate a second kind of model that applies to the same explanandum, consider any regular side-effect of neurotransmitter binding, such
as the mild vibration of the cell membrane surrounding the receptor.
Presume that whenever the neuron is exposed to neurotransmitters, this
4
All four candidate models maintain that a neuron behaves thus because it is constituted in such a
way that (1) it does not release neurotransmitters absent neurotransmitter exposure, and (2) exposure initiates a cascade of events, one of which is neurotransmitter release. Yet, the first condition is
customarily taken for granted, and explanatory presentations focus on the second by describing the
relevant features of the constitution of the neuron, and how exposuregiven this constitution
has the specified result.
5
Just as the overall phenomenon might be treated either probabilistically or deterministically, so it
goes with this dynamic principle. Though I will not worry about the details, which sort of treatment is most apt will depend on how the channels are individuated. If single channels are separately
represented, a probabilistic treatment is most appropriate; if large collections of channels are treated
together, deterministic treatment will be preferred.
46
L.R. Franklin-Hall
Some might suggest that this model isnt mechanistic at all, insisting that to be mechanistic a
model must satisfy a causal constraint. This would be to cut up the project slightly differently than
I have, but with no consequences for the overall argument. The task facing the new mechanist
would still be to cash out the causal constraint; it matters not whether that constraint is appealed
to in the definition of mechanistic models simpliciter, or (as in my exposition) in the characterization of explanatorily adequate mechanistic models.
47
The fourth model characterizes both real causal connections and appeals
to natural, rather than gerrymandered, parts. Its distinctive explanatory shortcoming is that it describes the system at the wrong level, in
terms of organized atomic parts changing according to dynamic principles (in this case, principles aptly called laws) describing atomic interactions. Such a model will be so complex that, in contrast with the three
rehearsed already, it is not possible here to sketch the course of events it
would describe as following from neurotransmitter exposure. Yet such
a low-level model will still satisfy the requirements of the mechanistic
framework above: it describes organized parts that change over time
according to dynamic principles, collectively bridging inputs and outputs. By depicting the neuron in such detail, it makes what I call a zooming error, and should, as above, be censured by any explanatory account
that takes actual explanatory practice as its touchstone.7
The three flawed accounts just sketched were easy to design, and
equivalent alternatives are readily available for any explanandum you
might choose; they require no real creativity or insight. One starts with
the inputoutput relationship for which the mechanistic model must
account. These inputs and outputs, as the mechanists rightly emphasize,
will be underpinned, in any particular system, by a complex set of connections between that systems parts. To produce a model that errs causally,
describe at least some portion of the system underlying the explanandum
behavior in terms of correlationalnot causalprinciples. To produce
one that makes a carving error, describe that underlying system veridically, but use a peculiar set of terms, those that individuate the system in a
non-traditional way.8 Finally, to produce a model at the wrong level, either
zoom in on the parts of the system more than is explanatorily appropriateby describing, for example, the inner working of entities usually
treated as wholes by scientists accounting for the focal phenomenonor
fail to break the system into parts, thereby producing a global model.
A zooming error is a species of carving error, and they are separated largely for rhetorical purposes.
The first prototypically concerns using gerrymandered parts, while the second concerns otherwise
natural parts at too fine (or coarse) a grain, considering the explanandum phenomenon.
8
Though many peculiar sets will exist, not any will do: they must still be sufficiently expressive that
they can be used, in concert with some set of dynamic principles, to bridge inputs and outputs.
48
L.R. Franklin-Hall
49
50
L.R. Franklin-Hall
mechanists approach was a successor project, writing that while philosophers of science typically associate the causal-mechanical view of scientific explanation with the work of Railton and Salmon, [.I] shall
argue that the defects of this view arise from an inadequate analysis of the
concept of mechanism (S342).
Yet a clear contrast exists between the old mechanists and the new, and
it may be misleading to see their projects as continuous. The key difference
concerns the relationship between cause and mechanism. The old mechanists were trying to reduce causation to mechanism; however, most new
mechanists use accounts of causation to understand the relations between
parts (or, properties of parts) of mechanisms. Speaking metaphorically,
old mechanisms were the causal glue, while new mechanisms are glued
together by causes. Along these lines, recent commentary calls for abandoning the idea that causation can be reduced to mechanism. On closer
inspection, it appears that the concept of mechanism presupposes that of
causation, far from being reducible to it (Kistler 2009: 599).
Given that mechanisms dont reduce causation but instead require an
account of it, what account should that be? Clearly, it must differentiate dynamic principles that reflect relations of correlation from those
of causation. To this end, two paths have been taken. The first is to tie
the mechanistic approach to an independent account of causation, one
that may lack any interestingly mechanistic character, for instance, to
Woodwards interventionism or Lewis counterfactual account. Craver
(2007), Glennan (2005), and Leuridan (2010) have pursued this strategy, adopting Woodwards (2003) account of causation, according to
which causal relations are those potentially exploitable for the purposes
of manipulation and control (Woodward 2003: 17). The second is to
develop an account of causation with mechanistic contexts in mind. For
example, Bogen (2008) and Machamer (2004) have pursued this option,
developing an activities view of causation.
Though the first approachthat of adopting an independent, nonmechanist account of causationis perfectly reasonable, I will not
explore it. Given the uncontroversial nature of the basic mechanistic conditionsat least for fans of causal explanationthose who fill out the
mechanistic picture by adopting a self-standing account of causation are
not much advancing the explanatory project. Needless to say, outsourcing
51
causation may well be the right move for mechanists to make, and those
who do so may still contribute to our understanding of scientific explanation; however, their contributions must come from elsewhere, presumably from their elucidations of the other two constraints on mechanistic
explanationson parts and levelwhich will be explored in due course.
Some mechanists have attempted to make sense of the causal relation
via the notion of activities (see Bogen 2005, 2008; Machamer 2004;
Waskan 2011). Here is an early statement of the view:
An entity acts as a cause when it engages in a productive activity. [] A
mechanism is the series of activities of entities that bring about the finish
or termination conditions in a regular way. These regularities are nonaccidental and support counterfactuals to the extent that they describe
activities (Machamer etal. 2000: 68).
52
L.R. Franklin-Hall
53
that philosophers might produce of the activities, and that it is the job of
scientists, in any case, to compile it.
The central feature of this account of causationthe activityis, from
a philosophical perspective, brute. Scientists identify activities, but they
have nothing generally in common; short of listing those taken seriously
by scientists at a given time, we cant say anything about what they are. It
remains possible that the quest to find a general account of causality like
Humes, Hempels, or Woodwards is misguided, and that wed be better off talking only of particular activities (Bogen 2008: 214). Yet, if we
take these claims seriously, the content of the first restriction on explanatory modelsthat they call on causal dynamic principlesis completely
opaque. Were I to offer a model containing a dynamic principle which
(intuitively) reflected relations of correlationsuch as the model above
that referred to membrane vibrationall that could be said is that such a
model is bad because it doesnt reflect activities, and that activities themselves were just the things that competent scientists talk about.
54
L.R. Franklin-Hall
In particular, in addition to potentially addressing the carving problem, these conditions are
offered as standards for distinguishing models that appeal to real parts from those that describe
fictional posits(Craver 2007: 128133).
10
I focus on Cravers presentation because it is the most systematic available, but it is characteristic of
the new mechanist literature. For instance, compare Dardens (2008: 961962) discussion of working entities and Machamer etal.s (2000: 56) comments on individuation of entities and activities.
55
1. Robustness: components should be detectable with a variety of causally and theoretically independent devices (2007: 132).
2. Manipulability: it should be possible to manipulate the entity in such
a way as to change other entities (2007: 132).
3. Plausibility: components should be physiologically plausible (2007:
132).
4. Stability: components should have a stable cluster of properties
(2007: 131) and should be loci of stable generalizations (2007: 190).
The first standard is that components be robust. Though some discussions of robustness have a more metaphysical cast, the variety of robustness at issue here is epistemic. To say that a component is robust is simply
to say that it is detectable by different kinds of devices, optimally those
operating on different principles. This standard is inspired by the usefulness of multi-device detection in helping scientists to distinguish genuine
features of a system from artifacts (Culp 1994).
Yet, robust detectability will not address the carving problem. First,
no device detects individuated parts as such, and consequently no part
component or otherwisecan be detected by more or fewer devices than
another. To illustrate, consider an electron micrograph of a cell. Such a
micrograph is (roughly) a representation of the electron density of material
in different regions. Patterns in the density revealed by electron microscopy
can provide evidence about the features of particular components, such as
the shape of a membrane channel. The micrograph itself, however, does
not detect which of the pieces are components; a carving into components is
something that the scientist brings to the micrograph to interpret it.
An alternative to insisting that components be detectable by different
devices is to suggest that the properties of components, as opposed to
parts, be so detectable. The problem with this alternative is that components and gerrymandered pieces will pass the test equally: we can detect
the properties of protein channels as well as quarter-neurons using a variety of normal neurophysiological devices. Thus, it does not appear that
robustness will contribute to solving the carving problem.
The second standard is manipulability. This standard requires that a
good part be itself manipulable in the service of affecting something else,
a constraint inspired by Ian Hackings (1983) famous call for entity
56
L.R. Franklin-Hall
57
58
L.R. Franklin-Hall
ties of the partin particular, those of its properties that underpin the
mechanisms behaviorare stable, at a minimum, throughout the range
of conditions over which the mechanisms overall behavior is stable.
Why might one want a parts property stability range to be determined
by that of the stability of the overall systems behavior? Arguably, because
it is only a part with this characteristic that could actually underpin the
behavior to be explained. After all, mechanism-level behaviorssuch
as the inputoutput relationships that are the target of most mechanistic explanationsthemselves have modal scope, holding in at least
some range of background conditions. If a mechanistic model is to fully
account for such a modally robust explanandum, the parts appealed to
in the model must themselves survivemaintaining their property clustersover that same range.11
In illustration, recall the explanandum behavior discussed above that
neurons release neurotransmitters when exposed to neurotransmitters.
This behavior holds of neurons over a range of conditionsin different
temperatures, different ionic environments, and so on. Among the neuronal components critical for the behavior are the ligand-gated ion channels located in the dendrite membrane. The channel properties relevant
to the overall mechanisms behaviormost notably, their disposition to
open in response to neurotransmitter bindingmust be stable over a
range of background conditions in order for the mechanistic model to
account for the stable systems behavior. Imagine, for instance, that in
some condition in which the system behavior was maintained, the ion
channel was denatured, and thus no longer possessed the property relevant to the behavior under analysis. Were this to be the case, one could
not model the behavior in terms of these parts.
How might this standard reject gerrymandered parts? The contrast
between the quarter-neuron model and the Standard Model can illustrate. Consider the range of background conditions over which the parts
11
There are situations more complicated than this. If a mechanism contains a variety of redundant
subsystemseach of which has a different range of stable functioningthe overall mechanism
behavior could have a range of stability greater than that of any particular component, or component pathway. Yet, this possibility doesnt undermine the more generic suggestion that some identifiable relationship exists between the stability of a mechanisms parts properties and the
mechanisms systems-level behavior, and that this connection might be used to determine the relevant stability range required of mechanism parts.
59
represented in the quarter-neuron model would maintain their properties, as well as the range over which the macromolecules in the Standard
Model would do so. At first glance, in contrast to the macromolecules, it
may appear that the quarter-neuron will fail to possess properties as stable
as required. Its properties will change in a broad range of circumstances,
as the quarter-neuron will be modified in some way just in case any of its
proper parts is so modified. Thus, this revised standard maybe effective,
and the carving problem solved.
Unfortunately, the proposal just described is not strong enough to distinguish good parts from mere pieces, and can only be used to rule out
non-veridical models, not those reflecting inferior carvings. The problem
is that many gerrymandered pieces, correctly characterized, will in fact
possess properties that are just as stable as required by the constraint
that is, as stable as the behavior of the overall mechanism. This is because
only the properties that underpin the mechanisms behavior need to be
so stable, according to the standard under consideration here. Although
it is true that a relatively large partgerrymandered or otherwiselike
the quarter-neuron, will change in some ways in the face of a wide array
of background circumstances, it will not change as often with respect to
the properties that underpin mechanism behaviorthose determining
its capacity to bridge the relevant inputs and outputs. In fact, with the
caveat noted above, it will maintain these properties at least over the range
for which the system-level behavior is stable. One might be tempted to
reject such properties as peculiar or gerrymanderedand thus not those
whose stability is relevant for determining component-hood. However,
this would be to make ones account of good parts dependent upon a
substantive account of good properties, which mechanists dont provide. Thus, the tactic shows little promise. A stability constraintat least
in the version Ive proposedcannot solve the carving problem.
60
L.R. Franklin-Hall
12
Craver sometimes presents the standard, quite reasonably, using his own symbolism. For instance,
another version of (A) requires that there is some change to Xs -ing that changes Ss -ing
(2007b: 153). Though these alternative statements are compatible with the interpretation I give of
the MM standard, and have informed my presentation, I do not use Cravers notation because it
would require too much space to adequately explain.
13
Though this statement is from Cravers (2007a), in explicating the view I am very influenced by
Cravers presentation in his (2007b). In correspondence, he reports that his presentation of the
standard there is particularly careful.
61
to change the component? Note that use of the term intervention here,
though clearly inspired by its use by causal interventionists, should not be
understood in the precise technical sense defined by them (e.g., Woodward
2003) but instead as another sort of in-principle causal manipulation,
sometimes glossed simply as wiggling (Craver 2007b: 153; 2007a: 15).
With this in mind, there are two genres of change that might be intended.
First, the manipulation might change the input to the component. For
instance, in the case of a part like a ligand-gated ion channel, a change
might involve exposing the channel to neurotransmitters, something that
would have a variety of downstream effects, the most direct of which is the
opening of the channel. Second, such a change might be made to the features underpinning the inputoutput regularity realized by the component
itself. Again, focusing on the ligand-gated ion channel, a wiggling of the
inputoutput relationship could involve a modification of the channels
disposition to open upon neurotransmitter binding. A parallel ambiguity
faces the second half of the (A) conditionthat involving the resulting
change to the behavior of the mechanism as a whole. This could involve a
change (from some default) of the output produced in a particular circumstance, or a change to the overall inputoutput relationship that the
mechanism underpins.
In light of these alternative versions of the conditionboth with
respect to the feature intervened upon and the consequent changeI
distinguish between two versions of Cravers condition (A).
(Ai) intervening to change the input to a component (from a default
input)
changes the output of the mechanism as a whole (from a default
output).
(Aii) intervening to change the inputoutput relationship realized by the
component changes the inputoutput relationship realized by the
mechanism as a whole.
Some examples used to illustrate the MM standard fall under (Ai),
while others align more with (Aii). For instance, indicating the relevance
of the first version, Craver (2007a, b) suggests that what he calls activation experiments can (sometimes) test the fulfillment of the condition,
62
L.R. Franklin-Hall
63
64
L.R. Franklin-Hall
65
66
L.R. Franklin-Hall
67
For a critique of the mechanists most promising response to the stop problem, that offered by
difference-making accounts of causal explanation as articulated by Woodward (2003, 2010), and
adopted explicitly by Craver (2007), see Franklin-Hall (2016). For my own positive proposal on
the stop problem, see Franklin-Hall (forthcoming). A recent paper on this problem that came out
too late for me to consider is Harbecke (2015).
15
For a detailed account of the different things philosophers have meant by level, see Craver
(2007, Chap. 5).
68
L.R. Franklin-Hall
features of the world would be, by my lights, to overreach. This follows from the fact that mechanisms themselvesand their componential
specificationsare only well defined (if at all) relative to some behavior.
And no behavior is delivered to us by the world, but must be picked
out by us. Third, even relative to a chosen behavior, questions about the
relative level of any two things can be ill-posed. Such questions are only
kosher when both entities in question are components (either immediate
or otherwise) of the mechanism in question.
The suggestion just sketched offers a more scientifically plausible, and
nuanced, understanding of our folk conception of levels than do the
global, flat stratifications advanced by Oppenheim and Putnam (1958).
And compared to levels defined in terms of the philosophically esoteric
laws, properties, and causeslevels of composition can appear innocent
and straightforward. Furthermore, and of central importance here, levels
of mechanistic composition can be naturally recruited to provide constraints on proper mechanistic explanation, as follows: for any phenomenon that one might want to explain, there is a mechanism responsible
for it, positioned at level n. To explain the phenomenon, an explanatory
mechanistic model should describe entitiesthat is, the immediate component parts of the mechanismat one level down, at n1.
Yet does this proposal address the levels problem, characterizing what
it means for one thing to be one level below another? If so, it is only by
way of a substantial promissory note. The problem follows immediately
from the difficulties already encountered in distinguishing components
or good parts, thus this discussion can be brief. Levels of mechanistic composition are only well defined if linked to an account of what
is required for a part to be an immediate component of a mechanism for
a behavior. Immediate components must themselves meet two conditions. First, they must be genuine components, not gerrymandered parts
or pieces. Second, these good parts must be, in some sense, just below
the mechanism as a whole (level n1), and not components of components (level n2). If supplied with a standard that, for any mechanism for
a behavior, specified all of its nested components, one could make sense
of which components were immediate; however, lacking a distinction
between parts and components, the immediacy requirement is impotent,
having no material on which to work. In light of this lacuna, even those
69
willing to grant a response to the stop problem, and who see the cascade
view as normatively superior to explanatory fundamentalism, should not
yet consider it to be a genuine alternative; the levels standard cannot,
from the surplus of minimally adequate mechanistic models, tell the
good explanations from the bad.
Conclusion
Though attractive at first glance, none of the new mechanists explanatory guidelines have survived scrutiny, successfully discharging the work
assigned to them. This work, it is worth emphasizing, is extremely difficult. So, even granting that I am right that mechanists have yet to complete it, this hardly shows that their general framework, and particularly
their commitment to causal explanation, is mistaken. Rather, it suggests
that the mechanistic account is but a story half-told. Thus far, proponents
have labeled some important distinctionssuch as between causal and
correlational relationships, between components and mere pieces, and
between appropriate and inappropriate explanatory levels. But the task
of filling them out remains.
As I see it, the present shortcomings of the mechanistic explanatory
account are the flip side of an admirable feature of the mechanism movement, one which has had a salutatory influence on contemporary philosophy of biology (and science): that of taking science (and particularly
biology) seriously. I conclude by recalling the origins of the mechanists
explanatory project, in doing so noting both its merits and its limits.
From early writings to the present day, the new mechanists have been
struck by what appears to be an evident mismatch between the DN analysis of explanation and explanatory practice in the life sciences. On the DN
view, explanations are deductively valid arguments, in which a statement of
the explanatory target is derived from true sentences, including one stating
a law of nature. Reasonably enough, mechanists have found it difficult to
make this jive with what scientists actually did. Where were these supposed
arguments in scientific articles and textbooks? What were these strict laws
in a science like biology, where exceptions are more than just distracting
litter on a landscape of regularity? Something seemed to have gone wrong.
70
L.R. Franklin-Hall
As Lindley Darden explains in her overview of the movement, [t]his work on mechanisms in
biology originated (primarily) not as a response to past work in philosophy of science but from
consideration of the work of biologists themselves, especially in molecular biology and neurobiology
and biochemistry and cell biology (2008: 958959). Similarly, Bechtel writes that these accounts
of mechanistic explanation attempt to capture what biologists themselves provide when they offer
explanations of such phenomena as digestion, cell division and protein synthesis (2007: 270).
71
Inparticular, rather than opening the black boxes of the scientific enterprisewith respect to causation, part individuation, and explanatory
levelphilosophers have (largely) taken those practices for granted.17
Perhaps this results from a too-successful enculturation of philosophers
into the scientific mindset, making it difficult to achieve the critical distance needed to philosophize about science. If so, while mechanists may
be right that advocates of the DN account were too far from science to
say anything true about it, perhaps the new mechanists have remained too
close to science to say anything surprising about it.
References
Bechtel, W. (2006). Discovering cell mechanisms: The creation of modern biology.
Cambridge: Cambridge University Press.
Bechtel, W. (2007). Biological mechanisms: Organized to maintain autonomy.
In F.Boogerd (Ed.), Systems biology: Philosophical foundations. Amsterdam:
Elsevier.
Bechtel, W. (2008). Mental mechanisms: Philosophical perspectives on cognitive
neuroscience. NewYork: Routledge.
Bechtel, W. (2011). Mechanism and biological explanation. Philosophy of
Science, 78(4), 53357.
Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanistic alternative.
Studies in History and Philosophy of Biological and Biomedical Sciences, 36,
42141.
Bogen, J. (2005). Regularities and causality; generalizations and causal explanation. Studies in History and Philosophy of Biological and Biomedical Sciences,
36, 397420.
Bogen, J. (2008). Causally productive activities. Studies in the History and
Philosophy of Science, 39, 11223.
Campaner, R. (2006). Mechanisms and counterfactuals: A different glimpse of
the (secret?) connexion. Philosophica, 77, 1544.
17
There is one mildly ironic exception to my general diagnosis. The only putative black box that
mechanists have opened is the scientists concept of mechanism. On reflection, this focus was
imprudent. Not every concept used by scientists is meaty, and not every term reflects a genuine
black box; mechanism is not a theoretical term within the science, but is a mere pointer, or placeholdersimilar perhaps to the philosophers term conception.
72
L.R. Franklin-Hall
73
74
L.R. Franklin-Hall
3
Compositional Explanation:
Dimensioned Realization,
New Mechanism, andGround
KennethAizawa
75
76
K. Aizawa
Compositional Explanation...
77
The New Mechanists in the philosophy of science have devoted considerable attention to the explanation of biological processes, such as phototransduction. (See, e.g., Bechtel and Abrahamsen 2005, p. 423; Craver
2007; Craver and Darden 2013; Glennan 1996, 2002; Machamer et al.
2000; Thagard 2003.) How are photons impinging on the photoreceptors
converted into changes in neurotransmitter release? The answer, according
to many New Mechanists, involves a multitude of entities that engage in
activities. In the case of phototransduction, these entities (or individuals)
include molecules of rhodopsin, transducin, and cGMP phosphodiesterase. During phototransduction, the rhodopsin molecules capture photons,
change conformation, and dissociate into two components. These are the
activities of rhodopsin molecules. One of the components of the rhodopsin molecule, an opsin molecule, activates numerous transducin molecules.
Each transducin molecule next activates many cGMP phosphodiesterase
molecules, which in turn hydrolyze cGMP. In a familiar accounting, the
ontology of compositional explanations includes processes, entities, and
activities, with entities and activities explaining how processes are implemented.1 Properties are typically mentioned only in passing.
The New Mechanist reticence about properties is surprising, given the commonalities between compositional explanations of processes and properties, on
the one hand, and their dierences from causal explanations, on the other. 2
1
Kaiser and Kriekel (forthcoming), document a number of dierent proposals that have appeared in the
New Mechanistic literature regarding what the explananda of interlevel explanations are supposed to be.
2
The New Mechanists typically use the term mechanistic explanation, whereas the term compositional explanation is used here. There are two reasons for this terminological shift. They are,
somewhat paradoxically, that the term mechanistic explanation might be both more restrictive
and more expansive than what is intended here. This is not to say that any single philosopher
simultaneously uses mechanistic explanation both more restrictively and more expansively.
Instead, these usages are more like competing tendencies.
On the more restrictive side, some philosophers may wish to propose that mechanistic explanations just are what, say, Machamer et al. (2000), say they are. It is something like a conceptual
or analytic truth that a mechanistic explanation is an explanation of a process in terms of entities
and activities. Given this, there just cannot be a mechanistic explanation of, say, the properties of a
whole in terms of the properties of the wholes parts. To think otherwise is just a conceptual confusion. The use of compositional explanation is meant to sidestep this concern.
On the more expansive side, there is the fact that mechanistic explanation is now sometimes
used to include causal explanations. As one illustration, Craver and Darden (2013, Chap. 5),
claims that mechanisms produce, underlie, or maintain their phenomena. Producing and maintaining a phenomenon appear to be species of causing that phenomenon, whereas underlying a
phenomenon seems to be composing that phenomenon. (N.B., this use of producing is not the
technical sense of production from Hall 2004.)
78
K. Aizawa
First, recall that causes must precede their eects, but realizer properties
are contemporaneous with their realized properties. Similarly, the entities
and activities that compositionally explain a process are also contemporaneous with the explained process. Phototransduction begins with photon
capture by the photopigment molecules and ends with the change in
release of neurotransmitters. The entire cascade of biochemical reactions
plus neurotransmitter release need not run its course within the photoreceptor cell before phototransduction begins. Second, recall that causes
must be wholly distinct from their eects, whereas realizer properties are
not wholly distinct from the realized property. In like manner, the entities and activities that compositionally explain phototransduction are not
wholly distinct from the phototransduction. It is this lack of distinctness
that sometimes tempts reductionists to claim that a process, such as phototransduction, is nothing more than such and such entities engaging in
such and such activities.
Dimensioned realization, and indeed the whole of the New Mechanistic
literature, has essentially been ignored in the expansive Ground literature. It has been ignored even by advocates of Ground who conceive
of Ground as a generic non-causal determination relation that might
include as species interlevel relations among individuals, properties, processes, and so forth, in nature. Moreover, Ground has been claimed to be
the relation invoked in non-causal explanations. (Of course, Ground has
also been largely ignored in the New Mechanist literature as well.) So,
for example, Clark and Liggins (2012), mention that the brittleness of a
cup results from the conguration of its constituent atoms, then proposes
that Ground is closely related to explanation. Because the brittleness
of a cup results from the conguration of its constituent atoms, we can
explain why the cup is brittle by pointing out its atomic structure. This
is at least pointing in the direction of interlevel property relations of the
sort schematized by the theory of Dimensioned realization. Nevertheless,
no mention of Dimensioned realization or New Mechanism appears in
the literature on Ground. For example, Schaer 2016, proposes that
A second cluster of cases [of Ground] is that of the dependence of the
higher-level on the lower-level. So consider the physical state of Socrates
and his mental state. For the physicalist, the physicalist state realizes the
mental state. And this backs an explanation: Socrates is in this mental
Compositional Explanation...
79
80
K. Aizawa
See, for example, Machamer et al. (2000), Craver and Darden (2013).
Strictly speaking, we should consider whether scientic explanation invokes both properties and
property instances, but for present purposes of bringing Dimensioned realization to the attention
of New Mechanists, we may forebear.
5
Compositional Explanation...
81
These points suggest that Machamer, Darden, and Craver, are, at least in
theory, open to an ontology that includes properties that gure in compositional explanations.
Of course, making passing references to properties and being open to
including them in mechanistic explanations is not the same thing as admitting compositional explanations based on a many-one non-causal determination relation from realizing properties to a realized property. One might
believe there are properties, but doubt that they stand in this sort of relation or that such relations form the basis for a distinct type of explanation.
Nevertheless, there is reason to embrace Dimensioned realization as a basis
for a type of compositional explanation that diers from a well-known
picture of the way in which compositional explanations work.
Craver (2007) has produced an extremely popular schema of the way
in which the process of an entity Ss engaging in an activity of -ing
is explained. One appeals to, say, an entity X1s engaging in an activity
of 1-ing causally inuencing an entity X2s engaging in an activity of
82
K. Aizawa
Compositional Explanation...
83
Hooke oered three properties of the cork to be explained: the cork does
not absorb water, it oats, and it has the capacity to stop a bottle. The
property of the cork is explained by the properties of its parts, the cells.
Each of the cells contains air and its walls are perfectly enclosed or
impermeable to air and water. The oating of the cork, for example, does
not depend on causal interactions between the impermeability of one
cell and the impermeability of another. There may, however, be causal
interactions between the impermeability of the individual cells and other
properties. And, of course, causal interactions between other properties
of the cells, such as their masses. What appears not to be relevant to the
explanation of the oating of the cork are causal interactions among the
properties invoked to explain the corks oating.
The upshot is that, in theory, New Mechanists should be willing
to accept that scientists believe in properties. Moreover, they should
be willing to accept that there are cases in which scientists use properties in compositional explanations wherein a property of a whole is
explained by (typically) many properties of the parts. The (small) price to
pay for accepting compositional explanations of the sort envisioned by
Dimensioned realization is that we must accept a picture of interlevel
compositional relations that diers somewhat from the one that is most
familiar in the New Mechanist literature.
84
K. Aizawa
Fine does not explicitly consider any instances of compositional explanations in the sciences, and then reject them as not being instances of
Ground. Nevertheless, his map of the landscape seems to exclude the possibility of a non-causal, but nevertheless scientic, form of constitutive
(or compositional) determination found in compositional explanations.
What to say here? In principle, one might simply expand the category
of scientic explanations to include both causal and compositional explanations and continue to distinguish them from metaphysical explanations in terms of Ground. Dimensioned realization and compositional
explanation would then merely be part of a more expansive scientic and
philosophical project tangential to Ground.
One waybut, of course, not necessarily the only wayto defend this
position would be by making the case that scientic causal and compositional explanations invoke a natural necessity, whereas Grounding explanations invoke a metaphysical necessity.6 One might think that there is a
kind of natural necessity according to which the HF molecule has a particular dipole moment in virtue of the electronegativities of the hydrogen
and uoride components and their bond length, but still wonder why
it is that having these electronegativities and bond length leads to this
dipole moment. Thus, there would be a kind of explanatory gapno
a priori connection, saybetween the atomic facts about the hydrogen
and the uorine components and the molecular fact about HF. The interesting or important work for a theory of Ground would come from its
wielding a metaphysical necessity corresponding to the strictest form of
in virtue of relation.
The foregoing example suggests that, in theory, one might set aside the
investigation of compositional explanation and Dimensioned realization
as orthogonal to the investigation of Grounding explanation. In addition, there are, in practice, philosophers who are likely to take this option
in practice. In a nuanced regimentation of a theory of Ground, Audi
(2012), postulates a number of features of Ground that appear to hold
of Dimensioned realization. Both Ground and Dimensioned realization
See Fine (2012, p. 38) for comments that might invite this argumentation. The point here is not
to attribute or recommend this position to Fine, or anyone else, for that matter. Instead, it is merely
to note that there are options for the theorist of Ground.
Compositional Explanation...
85
So, for Audi, the Grounding project in metaphysics and the compositional project in the philosophical of science are orthogonal.
The foregoing two examples only touch the surface of the diversity
of views regarding Ground. This lack of consensus notwithstanding,
there are at least important segments of the Grounding project for which
Dimensioned realization and compositional explanations are quite germane. These are the more ambitious strains of Ground that attempt to
understand compositional explanations as a species of Grounding explanation. This is the strain apparently embraced by Schaer and more
explicitly by Clark and Liggins.
How might attention to Dimensioned realization and compositional
explanations in the sciences inform the theory of Ground? One thought is
that, in scientic explanations, there is more to having an explanationor
a good explanationthan merely having a putative explanans that determines an explanandum. There are principles of good explanations that
must also be respected. So, for example, some explanations are thought
to be in some sense defective because they are ad hoc. Roughly speaking,
such explanations rely on hypotheses that provide for a determination of
the explanandum by an explanans, but the explanans is defective since it
is tailored specically to handle only a single explanandum. As a second
86
K. Aizawa
As noted in footnote 3 above, Schaer, for his part, is apparently inclined to the latter option.
The following analysis was inspired by Gillett (2010), which draws some important consequences
from the (typical) qualitative distinctness of the properties invoked in compositional explanations
for at views of realization. Credit should also be given to Gillett for helping me rene my articulation of how this account bears on the example from Hooke.
8
Compositional Explanation...
87
Gillett (2002, 2010), notes that the qualitative distinctness of realization properties from a realized property. In various lectures and Gillett (forthcoming), Gillett has described the dispelling of
the mystery idea in terms of the piercing explanatory power or PEP of compositional explanations. Compositional explanations get their PEP by appealing to qualitatively distinct properties.
88
K. Aizawa
Coincidentally, Craver and Darden (2013, Chapter 6), discuss homuncular explanation drawing
attention to the fact that, in such explanations, a property of the whole is explained by appeal to
the very same property of a part and that such explanations invite a further explanation of why the
part has that property.
Their analysis of what is wrong with homuncular explanations, however, diers from the one
oered here. They treat homuncular explanations as phenomenal models and oer the following
analysis of what is wrong with them:
Phenomenal models are supercial because they specify neither the internal components of
the mechanism nor the organizational and productive features by which the mechanism
works. Mechanistic models have depth. They reveal the internal structure of a mechanism.
This is why they are explanatory, not merely descriptive. (Craver and Darden 2013; Kindle
location 2011f )
By Craver and Dardens lights, what is missing in a homuncular explanation is a specication of the
internal components and the organizational and productive features by which the mechanism
works. Yet, in explaining the oating of a cork by appeal to the oating of its cells, one is, in fact,
specifying the internal components and the organizational and productive features by which the
oating comes about. In theory, however, Craver and Darden could embrace the analysis oered by
the Dimensioned realization. They accept the qualitative distinctness of the explaining processes
from the explained process.
Compositional Explanation...
89
Conclusion
The goal of this chapter has been to introduce Dimensioned realization
as a friendly addition to the toolbox of philosophers of science working
on mechanisms and metaphysicians working on Ground. Dimensioned
realization provides an account of non-causal, compositional relations
among properties that are invoked in scientic explanations of the properties of a whole in terms of the properties of their parts. This is analogous
to what many New Mechanists envision, namely, a non-causal, compositional relation that might be invoked in explanations of the processes of a
whole in terms of processes among its parts. Dimensioned realization also
provides one with the tools to analyze a particular kind of problematic
explanation in the sciences, namely, homuncular explanations. The existence of decient explanations suggests that, in order to explicate what
is decient in some explanations, Grounders need to move beyond bare
claims of Ground. Moreover, Dimensioned realization provides some of
the tools needed to analyze some of these decient explanations.11
References
Audi, P. (2012). Grounding: Toward a theory of the in-virtue-of relation. Journal
of Philosophy, 109(12), 685711.
Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative.
Studies in History and Philosophy of Science Part C: Studies in History and
Philosophy of Biological and Biomedical Sciences, 36(2), 421441.
Clark, M. J., & Liggins, D. (2012). Recent work on grounding. Analysis, 72(4),
812823.
Craver, C. (2007). Explaining the brain. Oxford: Oxford University Press, USA.
Craver, C. F., & Darden, L. (2013). In search of mechanisms: Discoveries across the
life sciences. Chicago, IL: University of Chicago Press.
11
I have beneted in numerous ways from many discussions of these issues with Carl Gillett,
though any mistakes herein are my responsibility. Thanks to Carl and Kelly Trogdon for helpful
comments on earlier drafts of this chapter.
90
K. Aizawa
4
Is Mechanistic Constitution aVersion
ofMaterial Constitution?
JensHarbecke
Introduction
Recent philosophy of neuroscience has made substantial effort to develop
a conceptually sound and descriptively adequate account of mechanistic
constitution. The referent of this term is supposed to be the relation
that successful neuroscientific explanations declare to hold between the
to-be-explained cognitive or neural phenomenon and the explanatory
neural mechanisms underlying the phenomenon. Neuroscientists have
also referred to this relation by terms such as is responsible for (Bliss
and Lmo 1973, 331), gives rise to (Morris etal. 1986, 776), plays a
crucial role in (Davis etal. 1992, 32), contributes to, forms the basis
of (both Bliss etal. 1993, 38), underlies (Lmo 2003, 619; Frey etal.
1996, 703), or is constitutively active in [the phenomenon] (Malenka
etal. 1989, 556).
J. Harbecke (*)
Witten/Herdecke University, Witten, Germany
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_4
91
92
J. Harbecke
93
between the two notions. In a final step, I demonstrate that, once the
metaphysics presupposed by the regularity account is accepted, puzzles
such as the statue-and-lump case disappear. They receive an eliminativist
solution, which may be attractive in various respects.
The investigation proceeds as follows. In a first step, I reconstruct the
context in which the question about mechanistic constitution arises (section The Question of Mechanistic Constitution). I then review the
philosophical enquiry associated with mechanistic constitution (section
The Mechanistic Approach), which includes a discussion of the regularity theory (section What Is Mechanistic Constitution?) and of identity statements about phenomena and mechanisms (section Identity).
Subsequently, I review the problem of material constitution and the
grounding problem (section The Question of Material Constitution).
I then show that material constitution is to be distinguished from
mechanistic constitution (section Is Mechanistic Constitution Material
Constitution?) while there are various logical and conceptual connections between the two notions (section Connections). In a final step,
I suggest that the ontology presupposed by the regularity approach to
mechanistic constitution offers an informative eliminativist solution
to the problem of mechanistic constitution and grounding (section
Mechanisms and the Grounding Problem). The last section summarizes the argument and raises some open questions that the present chapter was unable to answer (section Conclusion).
94
J. Harbecke
From the 1990s onwards, this general picture of explanation in the special sciences has been challenged, notably through the works of authors
such as Bill Bechtel and Bob Richardson (1993), Machamer etal. (2000),
Bill Bechtel and Adele Abrahamsen (2005), and Carl Craver (2002,
2007). What unites these authors is their interest in explanations in the
neurosciences. In their view, neuroscientific explanations essentially focus
on the mechanisms constituting a to-be-explained phenomenon. Lawlike
regularities subsuming the phenomenon play a secondary role in an explanation, if they play a role at all (however, cf. Fazekas and Kertsz 2011).
The views of these authors were inspired by close examinations of various cases of successful neuroscientific inquiry. The following case from
the cognitive neuroscience of memory is an example in this sense.
95
96
J. Harbecke
enon via a general law. Rather, the explanation is essentially based on the
identification, location, and analysis of the mechanisms that constitute a
to-be-explained phenomenon. The mechanists have taken this observation as the basis both for the formulation of a descriptive, or naturalist,
theory of successful and satisfactory explanations in neuroscience and for
the formulation of an explanatory ideal that is declared to be normatively
binding for future research in neuroscience.2
A conceptually transparent statement of both the descriptive theory
and the normative ideal requires a prior clarification of the following two
concepts: It has to be determined what a mechanism is, and it has to be
specified what it means for a mechanism to constitute a phenomenon.
It is here where the genuine philosophical work begins. A large part of
the recent debate on the mechanistic approach has focused on these two
issues, accordingly.
97
What Is aMechanism?
The most popular definitions of the term mechanism are the following
ones. Machamer etal. describe it as consisting of entities and activities
organized such that they are productive of regular changes from start or
set-up to finish or termination conditions (Machamer etal. 2000, 3).
Bechtel and Abrahamsen extend this definition by describing a mechanism as a structure performing a function in virtue of its component
parts, component operations, and their organization. The orchestrated
functioning of the mechanism is responsible for one or more phenomena (Bechtel and Abrahamsen 2005, 423).
Both formulations have their weaknesses. For instance, the distinction
between an entity and an activity can be tricky in scientific contexts as
many scientifically interesting entities are primarily defined over their
activities. Electrons, for instance, are often described as particles. At the
same time, in physics, they are primarily characterized by what they do.
But even if the absence of a clear distinction introduces some vagueness
into the definitions, they are generally considered as serving their purpose
well enough. In fact, they do seem to capture sufficiently well the metaphysical nature of the NMDA-receptor activity that is invoked to explain
head direction representation (cf. section The Question of Mechanistic
Constitution). And so, I will accept the definitions for the remainder of
the chapter.
98
J. Harbecke
Not having distinct instances does not imply that the instances are identical; it merely says that
the spacetime regions instantiating the relata of mechanistic constitution overlap. Note that two
spacetime regions can overlap without being identical in the case that they do not perfectly overlap (i.e. if they are not both a mereological part of the respective other).
4
The term composition has been used by Machamer etal. (2000, 13), Bechtel and Abrahamsen
(2005, 426), and Craver (2007, 164); constitution occurs in Craver (2007, 153); constitutive
relevance is found in Craver (2007, 139). It is safe to say that the authors intend these terms
widely synonymously. For the sake of terminological unity, from now on, I will use the term constitution to denote the relation that is referred to by these expressions.
5
As mentioned in footnote 1, Gillett (2007, 2013) has developed a similar analysis for the notion
of realization.
3
99
See also Kaiser and Krickel (2016) for the idea that the relata of mechanistic constitution are
hybrids between entities and activities.
7
This certainly is a non-standard way of speaking of mechanisms. Craver, for instance, likes to
characterize mechanisms as an activity -ing of one or more objects/entities denoted by x, y
and so on. In other words, mechanisms, according to Craver, are (kinds of ) events or processes
(where processes are spatiotemporally extended events). Nevertheless, since science is typically
interested not in events but in kinds of events, the commitment to an ontology consisting of space
time regions and mechanistic types may be more adequate to reconstruct pertinent explanations.
6
100
J. Harbecke
The main idea underpinning this definition is that mechanistic constitution is a relation between mechanistic types that are regularly, but
not redundantly, co-instantiated such that their instances are mereologically related. The mereological theory presupposed here is General
Extensional Mereology (GEM) as explicated by Varzi (2009). Since often
no single mechanism is sufficient for the occurrence of a given phenomenon, the definition makes reference to complex mechanisms involving
a range of mechanistic properties. Additionally, since sometimes more
than one mechanism can secure the occurrence of the given phenomenon, the definition also allows for alternative constitutive conditions.
The mereological requirement is introduced in order to ensure that the
phenomenon must occur (at least partially) in the same place and time
as the mechanisms that constitute it. All of these ideas are expressed by
conditions (i)(iii, a).
Condition (iii, b) is added for the following three reasons. First, it captures an intuitive relationship between the mechanisms realizing a given
phenomenon in the sense that an instantiation of the former is demanded
to occupy no less space or time than the phenomenon induced. It would
be strange to think that when a set of neural mechanisms constituting a
learning process by the rat occurs, that learning process partly takes place
outside the rat. Or if it actually does, condition (iii, b) demands that also
further lower-level mechanistic aspects outside the rats body must be
relevant. In particular, the condition excludes that there is such a thing as
a phenomenon partially or wholly occurring in the void (unless it itself is
a fundamental mechanism).8
It is an interesting question whether Craver (2007), as one of the most famous mechanists, would
accept conditions (iii, a) and (iii, b). He does say that mechanisms sometimes transcend or transgress the clear boundaries of a system. For instance, his view of the action potential is that it relies
crucially on the fact that some components of the mechanism are inside the membrane and some
are outside (141). At the same time, he points out that [o]ne cannot delimit the boundaries of
mechanismsthat is, determine what is in the mechanism and what is notwithout an account
of constitutive relevance (141). Since his definition of constitutive relevance explicitly demands a
mereological relationship of the mechanisms instance and the phenomenons instance, it seems
clear that Craver as well believes that no phenomenon can occur unless there is a mechanism that
occurs wherever the phenomenon occurs. Or in short, if the mechanism for a phenomenon transgresses the boundaries of a system, so does the phenomenon, and vice versa. Hence, the textual
evidence in Craver (2007) suggests that he accepts (an analogous version of ) condition (iii).
8
101
PHDR : ( FX1 Y1 c G ) ( GX 2 Y2 c H ) ( HX 3 Y3 c I ) ,
102
J. Harbecke
Identity
An important aspect of the regularity theory of mechanistic constitution
is the fact thatunder the presupposition of a nomological coextensiveness principleit provides an empirical criterion for the identity of constitutively related mechanistic types. This is a valuable result in light of
the fact that, for instance, in the philosophy of mind and the philosophy
of the special sciences, it has sometimes been argued on purely conceptual grounds that higher-level types are (non)-reducible and (non)-identical to lower-level types.
The following example demonstrates how the regularity theory provides an empirical criterion for the identity of types. Suppose that the
mechanistic type F stands (as before) for the activation of NMDA-
receptors, F1 stands for calcium/calmodulin-dependent protein kinase
II (cf. Malenka etal. 1989), F2 stands for calcium-dependent protease,
and G stands for the phenomenon of the firing of HD cells. Moreover,
103
G
PHDR
: FF1 F2 X1 Y1 c G
Suppose further that it is determined on the basis of further experiments that the type conjunction FF1F2 is c-minimally sufficient for G and
that, hence, X1 is in fact empty. In this case, the initial material theory
PHDRG will be transformed into the following one:
G
PHDR
: ( FF1 F2 Y1 c G )
G
PHDR
: FF1 F2 c G
It should be mentioned that, since philosophers are part of the scientific community, it is not
entirely likely that the community will reach this conclusion for any phenomenon. Philosophers
such as Jerry Fodor might insist that there is always an open disjunctive list of possible constituters
for any phenomenon (cf. for instance 1974). Consequently, Y1 is never empty, even if in this world
no further complete constituting condition is found. In my view, this modal argument is irrelevant
for science as we know it. In particular, the proponents of this argument can still agree that the
emptiness of Y1 supports at least what can be called actual, or this-worldly, reduction. The latter, I
submit, is the only reduction of interest for science, and it is the one characterized here.
9
104
J. Harbecke
105
Pluralists have voted for a non-identity on the basis of the fact that,
even though the two objects share all their categorical properties, David
seems to have certain properties that Lump lacks. Among these may be
what are often called sortalish properties, which include (i) persistence
conditions, particularly modal properties like being essentially shaped
about like so, (ii) kind or sortal properties, and (iii) properties that things
have partially in virtue of their instantiation of properties in categories (i)
or (ii) (Bennett 2004, 341). In particular, David seems to instantiate the
properties is essentially a statue, could have partially been made out of
granite, did not exist on Monday, and so on, while Lump lacks these.
According to Leibniz law, a necessary condition for identity is the sharing
of all properties, and so the non-identity of the two objects follows.
The pluralists argument for a non-identity has the following general
form (where s and l are individual constants, x and y are individual variables, P(_,_) stands for the parthood relation, and is a
property variable):
Premise 1: Object s and object l share all of their parts, but there is at
least one property that s has and l lacks. (in short:
"x ( Pxs Pxl ) $f ( fs fl ) )
Premise 2: If two objects are identical, then they share all properties (in
short: "x"y ( x = y "f ( fx fy ) ) ).
Conclusion: Object s is not identical to l, even though the two share all
of their parts. (in short: s l "x ( Pxs Pxl ) )
The argument is formally valid. Hence, if the premises are true for
some s and l, the general result is established that it is possible for two
objects to occur in the same place and time and yet be non-identical. The
fact that one object is materially constituted by the other may imply that
the constituted object is not over and above the constituting object.
Nevertheless, the two are not the same thing. The general result obviously
contradicts the extensionality theorem that is part of the mereological
system, extensional mereology (EM), (and all systems containing it; cf.
Varzi 2009): "x"y ( x = y "z ( Pzx Pzy ) ) , and so it asks for a particular kind of mereology.
106
J. Harbecke
Different versions of the above argument use modal or temporal operators, such that s and l share all properties in this world or now, but one
out of s and l could lack, or will lack, a property that the other one has
and will continue to have. The conclusion remains the same. Moreover,
the argument can be developed for many other puzzles such as the persons
and their bodies puzzle or the ship of Theseus puzzle (cf. Wasserman 2015).
An interesting aspect of the metaphysical theory supported by versions of the above argument is the asymmetry that is typically associated
with material constitution. The sameness-of-parts relation as expressed in
Premise 1 is symmetric, and Lump clearly has certain sortalish properties that David lacks. But then why doesnt David materially constitute
Lump? And what is the further condition that one of the two objects
must satisfy to receive the primacy underlying the asymmetry?
These questions hint at the fact that the sentence The lump of clay
called Lump constitutes the statue called David is misleading. The pluralist cannot possibly mean that the relation of material constitution holds
between Lump and David. Rather, she (presumably) means that material
constitution holds between a set or collection of material clay parts, atoms
perhaps, and David, as well as between the set of clay parts and Lump. The
asymmetry then stems from the fact that both David and Lump can cease
to exist in this world, but the set or collection of material parts cannot.10
Monists have attacked the pluralists picture from several angles. First,
they have argued that it seems strange for two things to share all their
parts and yet fail to be identical. In virtue of what, so the monist asks,
does David have those further properties that Lump, or the collection
of clay parts, lacks? What grounds the sortalish properties if it is not the
categorical properties shared with Lump, or with the collection of material clay parts? If it is further material parts or properties not shared by
the collection, then the statue would fail to coincide with the collection
of clay parts. However, if no material parts or properties ground Davids
sortalish properties, then the collection of clay parts should not have any
different ones than the statue. Unless, of course, the sortalish properties
are simply projected onto the world by humans, or unless they are primitive properties.
See also Paul (2010, 583) and what she calls the mereology puzzle for material constitution.
10
107
If sortalish properties are projected onto the world, they dont seem
to be sufficiently ontic. There would not be a statue out there. Rather,
the properties would be in the eye of the beholder. On the other hand, if
Davids sortalish properties are primitive, the existence of a statue in the
same place and time as the collection of clay parts cannot be established
by any independent argument (but, see Bennett 2004).
Moreover, monists have claimed that the argument equivocates at least
one term, which would render the argument inconclusive (but, see Fine
2003). Finally, they have argued that, once we accept statues, the universe
will be overpopulated quickly. The point is that the statue and the collection of clay particles do not seem to be the only inhabitants of that space
time region. Also the arms-torso-head-legs object seems to be coincident
with the clay. Moreover, the forearm-upperarm-cribs-calves-feet-torso-head
object seems to coincide with it. But all of these objects are distinguished
from the others by at least some sortalish properties. In short, it is difficult for the pluralist to avoid becoming a bazillion-thinger (Bennett
2004, 358).11
As a reaction, monists have sometimes denied the existence of macroscopic objects altogether. What exists, they claim, are fundamental material things, perhaps atoms, perhaps more fundamental entities. These
entities can be configured and combined in various ways. But no such
configuration or combination gives birth to a new macroscopic object
(cf. Unger 1979). This position has sometimes been called mereological
nihilism or eliminativism.
A deflationary approach may open up a third way in between pluralism and monism. Such an approach points out that both pluralists and
monists agree on all the non-statue facts, such as the fact that there is a
particular collection or set of material entities present where the statue
is supposed to be present. Their argument over the existence and non-
existence of the statue on top of the collection of material entities is therefore mostly a verbal one. It changes nothing about the basic facts.
So, in short, material constitution is a relation that exists in an ontic
sense only in the eyes of pluralists. From this perspective, it is a relation
that holds between a set or collection of material things and an object
11
But see Rea (1998) for the position that this consequence is unproblematic.
108
J. Harbecke
such that the object has no material parts not contained in the collection.
The following definition captures the idea (where, x, y, z, x, and y are
individual variables, and P(_,_) signifies the parthood relation):
Material ConstitutionA collection of material things or particles x
materially constitutes an object y (written as Cxy) if, and only if:
(i) for all z, if z is a part of y, then it is an element of x, and
(ii) x is not identical to y, and
(iii) there is at least one collection of material things or particles that is x
and at least one object that is y.
In short
Cxy def "z ( Pzy z x ) ( x y ) $x $y ( ( x = x ) ( y = y ) )
109
110
J. Harbecke
111
112
J. Harbecke
113
Connections
The previous section focused on the differences between mechanistic constitution and material constitution as they have been defined in sections
What Is Mechanistic Constitution? and The Question of Material
Constitution. This section will highlight some interesting analogies
between the two relations.
The most striking one is clearly the fact that the metaphysical theory associated with Mechanistic Constitution faces a similar problem
as the one brought forward against pluralists in the debate on material
constitution. It seems difficult for the pluralist to resist the conclusion
that, for any collection of n fundamental entities or particles, there are
at least 2n objects (partially or wholly) materially constituted by it. For
instance and as pointed out already in section The Question of Material
Constitution, not only the statue seems to be constituted by the collection of clay particles. Also the arms-torso-head-legs object seems to be
coincident with the clay. But all of these additional objects are distinguished from the others by at least some sortalish properties. In short, it
is difficult for the pluralist to avoid becoming a bazillion-thinger.
Mechanistic constitution as defined above faces a similar problem of
ontological inflation. It seems that for almost any phenomenon p and any
mechanism m constituting p, another more fine-grained mechanism can
be described that constitutes both m and p. Consequently, a proponent of
Mechanistic Constitution and the metaphysical theory associated with
it will rapidly become a bazillion-mechanism theory with respect to
any investigated phenomenon.
The only solution to this problem seems to be a balance between different ontologies in terms of their simplicity and explanatory strength.
However, it is not clear at this point whether the problem can be
adequately addressed by the mechanistic approach based on Mechanistic
Constitution.
114
J. Harbecke
A second correspondence between mechanistic constitution and material constitution has already been mentioned in section Is Mechanistic
Constitution Material Constitution?. For both relations, the notion
of a spacetime region plays an important role. Phenomena and their
constituting mechanisms are characterized as sharing the same space
time region by definition. An object materially constituted by a collection of material entities is typically described as sharing the same place
and time with the latter. Mechanistic Constitution actually presupposes the applicability of GEM to spacetime regions. Above, it was left
open whether material constitution could follow along with this idea,
and thereby accept the extensionality theorem for spacetime regions at
least. Unfortunately, however, it was not even clear whether, for instance,
statue-and-clay pluralists would accept the characterization of spacetime
regions as individuals in the first place.
A third correspondence consists in the fact that both mechanistic
and material constitution are to be distinguished from supervenience.
As I have shown elsewhere (Harbecke 2014), mechanistic constitutional
claims are not equivalent to supervenience claims. Since material constitution is a relation between individuals, whereas supervenience is a relation between sets of types, it seems clear that the former is not equivalent
to the latter as well.
115
does not recognize objects as primary at all. Objects are essentially projections on persisting mechanisms, and they do not enjoy any ontic status.
Perhaps surprisingly, such an ontology offers an eliminativist solution to
the problem of material constitution that still allows to make sense of the
pluralists and monists main claims. Whereas the compositional nihilists
(Paul 2010, 586) deny the existence of macroscopic objects and recognize
the existence of fundamental objects, an ontology as the one described
denies the existence of objects toto imperio. Not even the fundamental
objects exist qua objects.
Such a picture has sometimes been brought forward as an alleged
reductio argument against the eliminativist or nihilist stance (cf. Sider
1993; Zimmerman 1996; Schaffer 2003). However, the belief that our
everyday objects including statues and persons are in fact processes is less
absurd than it may seem at first. For one thing, metaphysical analysis
has found it notoriously difficult to clarify what an object actually is qua
object, since all objects that we encounter seem to be inseparable from at
least some of their properties.
Second, the boundaries of non-fundamental objects are typically
vague, and determining where a statue begins and ends is almost an
impossible task. Third, the traditional human hope to find at the bottom
of things atoms or some other kind of fundamental objects has been continuously disappointed by twentieth-century physics. As a consequence,
some philosophers of physics have reverted to a structural realist position
that accepts only relations as the fundamental building blocks of things
(cf. French and Ladyman 2003; Esfeld and Lam 2008). Hence, accepting
objects into ontology looks less appealing in the twenty-first century than
it might have looked 100 years ago.
The point is that, if statues, lumps of clay, and even clay particles considered as objects are merely more or less arbitrarily delineated chunks of
processes, the conceptual problem of material constitution disappears.
Whether the statue-process type is then identical to the collection-ofclay-particles-process type are identical becomes a problem open for
empirical investigation (cf. section Identity). Moreover, should the
strategy for choosing the best system out of a set of multilevel ontologies be successful, it may be possible to answer the question whether the
statue-process type has ontic status at all.
116
J. Harbecke
Conclusion
The main aim of this chapter was to analyze the potential similarities
and differences of the regularity account of mechanistic constitution and
the standard accounts of material constitution. In a first step, a descriptive explanatory project from the neurosciences was reviewed in order to
frame the context in which the question about mechanistic constitution
is believed to arise. Subsequently, the regularity theory of mechanistic
117
References
Baumgartner, M., & Gebharter, A. (2014). Constitutive relevance, mutual
manipulability, and fat-handedness. British Journal for the Philosophy of
Science. Advance online publication. doi:10.1093/bjps/axv003.
Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative.
Studies in History and Philosophy of Biological and Biomedical Sciences, 36(2),
421441.
118
J. Harbecke
119
120
J. Harbecke
121
Taube, J., Muller, R., & Ranck, J. (1990b). Head-direction cells recorded from
the postsubiculum in freely moving rats. II.Effects of environmental manipulations. The Journal of Neuroscience, 10(2), 436447.
Unger, P. (1979). There are no ordinary things. Synthese, 41(2), 117154.
Varzi, A. (2009). Mereology. In E.N. Zalta (Ed.), Stanford encyclopedia of philosophy (Summer 2009). http://plato.stanford.edu/entries/mereology/
Wasserman, R. (2015). Material constitution. In E.N. Zalta (Ed.), The Stanford
encyclopedia of philosophy (Spring 2015). http://stanford.library.usyd.edu.au/
entries/material-constitution/
Wiener, S., & Taube, J. (2005). Head direction cells and the neural mechanisms of
spatial orientation. Cambridge, MA: The MIT Press.
Zimmerman, D. W. (1996). Could extended objects be made out of simple
parts? An argument for atomless gunk. Philosophy and Phenomenological
Research, 56(1), 129.
5
Anti-Reductionism, Anti-Rationalism,
andtheMaterial Constitution
oftheMental
DerkPereboom
Ive set out and defended an account of the vertical relations between
the mental and more fundamental levels in terms of a theory of material constitution (Pereboom 2002, 2011). A controversial feature of
this account is that it rejects identity as the distinctive interlevel relation, by contrast with standard reductive positions and, perhaps surprisingly, with the rival nonreductive subset view. Instead, I appeal
to a fundamental made up of relation. Critics have argued that this
account is objectionable because this relation is obscure. Such a criticism is fueled by a rationalist presumption that accounts of this sort
can only appeal to conceptual analysis and logical relations such as
identity. I reject the obscurity criticism, and more generally, the rationalist presumption.
D. Pereboom ()
Cornell University, Ithaca, NY, USA
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_5
123
124
D. Pereboom
Introduction
Physicalism about the mental requires that all mental entities be appropriately founded in microphysical entities. Supposing the truth of physicalism, it remains an open question whether the relation between the
microphysical and the mental is reductive or nonreductive. By contrast
with the mid-twentieth century, currently, most nonreductivists maintain that the main reason for accepting the nonreductive option is not
methodological but metaphysical. On the position I endorse, mental
natural kinds are not identical to natural kinds in microphysics because
mental causal powers are not identical to microphysical causal powers.
The fact that mental kinds are multiply realizable at the level of microphysical kinds provides an important reason to believe this is so. The view
I propose departs from other contemporary nonreductivisms insofar as
it rejects the token identity of mental and microphysical entities of any
kindincluding causal powers.
The most fundamental relation between the mental and the microphysical is material constitution, understood in such a way that it does
not feature identity (Pereboom 2002, 2011). Material constitution will
be a relation between material entities, and thus entities that can be
causally efficacious. It can be thought to qualify as a kind of grounding
(Fine 2001; Schaffer 2009) or building relation (Bennett 2011) on the
supposition that such relations are characterized in a general way, just by
asymmetry, irreflexivity, one-way necessitation, and as giving rise to generative explanations of the less fundamental by the more fundamental.
My particular conception of material constitution essentially involves
the made up of relation, which holds, for example, between a statue
and the lump of clay (Pereboom 2011, 2013a). The made up of relation
is itself asymmetric and irreflexive: the lattice is not made up of the
diamond, and the diamond is not made up of itself. It has a particular
directedness: the less fundamental is made up of the more fundamental,
and not vice versa. Crucially, at its core, the made up of relation is primitive and not characterizable in terms of more fundamental relations.
Here is a formal characterization of this notion of material constitution. Suppose that x and y are concrete physical entities. Entities x and y
125
are materially coincident just in case they, at some level, are made out of
the same parts. Then,
(C1) x materially constitutes y at t if and only if
(a) y is made up of and materially coincident with x at t;
(b) necessarily, if x exists at t, then y exists at t and is made up of and
materially coincident with x at t; and
(c) possibly, y exists at t and it is not the case that y is made up of and
materially coincident with x at t.
The last provision (c) rules out the identity of x and y (on the assumption of the necessity of identity), as does clause (a), since the made up of
relation is irreflexive.
Carl Gillett (2002, 2003) distinguishes between a flat notion of realization, in which properties of a thing are realized by properties of that
same thing, and a dimensioned conception, in which properties of a thing
are realized by properties of a distinct thing from which it is constituted,
and one might adapt this distinction to the notion of material constitution as Ive characterized it. Sydney Shoemaker (2003, 2007, 2014)
invokes a similar distinction between a variety of property realization in
which property F of X is realized by property G of the same thing X, and
microrealization, in which property F of X is realized by a microphysical
state of affairs, which consists in microphysical entities having certain
properties. Material constitution allows for Gilletts dimensioned view
and Shoemakers notion of microrealization. If M is constituted by P,
P might be a microphysical state that consists in microphysical entities
having certain properties. In my view, there will also be states of affairs at
a level higher than the microphysicalthe chemical and neural levels, for
examplethat constitute M.
Lynne Bakers discussion of constitution features several counterexamples that pose a challenge to clause (b) of this characterization, the necessitation of the constituted entity by its constitutor (Baker 2007: 1113,
106110). The existence of the dollar bill in my wallet is not necessitated
by its underlying microphysical arrangement, because its existence also
requires the United States Federal Reserve Bank and the laws governing
126
D. Pereboom
127
128
D. Pereboom
play the roles at issue. Similarly, the various physical realizations of a dogs
and a humans belief that this fire is dangerous may feature a particular
compositional property. This property would be more abstract than any
specific neural compositional property, since it can be realized in different kinds of neural systems. Moreover, it may be that this compositional
property can also be realized in a silicon-based system, which could then
have the belief about danger. Imagine that computer engineers built such
a silicon-based system that mimicked the capacities of and interconnections among neurons in a human brain as exactly as could be, and then
activated it to replicate what happens in that human brain in the case
of the belief about danger. Its possible that the silicon-based state that
ensues would constitute that belief and have an internal structure similar
enough to the internal structure of the neural system for each to count as
instantiating the same compositional property.
129
130
D. Pereboom
Identity
Despite their allegiance to nonreductivism, Jessica Wilsons (1999) and
Sydney Shoemakers (2003, 2007) subset view endorses a token-identity
thesis for mental and lower-level causal powers, although it does avoid
131
reductive type- and token-identity claims for mental states. The term
causal power refers to that by virtue of which an effect is produced
(Baker 2007: 98). Im attracted to the view that property instances are
the entities by virtue of which effects are produced, as L.A.Paul (2000)
advocates, and thus causal powers would be property instances. Causal
powers would then be, in the first place, token entities rather than types,
as is sometimes supposed.1
On Wilsons and Shoemakers view, the mental is realized by the neural
and in the microphysical because the forward-looking causal powers of
a mental state are a proper subset of the forward-looking causal powers
of the lower-level state. Here is Shoemakers most recent statement of his
position:
It is compatible with the claim that the instance of the higher-order property and that of its realizer are not identical that the former is part of the
latter. And that seems the right conclusion to draw from the fact that the
causal powers of the former are a proper subset of those of the latter. And
then it is open to us to say that while it is true that the instance of the realizer property causes the various effects we attribute to the realized property,
it does so because it includes as a part the instance of the realized property.
(Shoemaker 2014)
Paul argues for this view partly on the ground that aspects or property instances feature the right
fineness of grain, by contrast with events, at least on some accounts. I prefer to construe property
instances just as ways particular things are, as in David Robbs (2007) characterization, which he
contrasts with abstracta. Token causal powers will then also be ways particular things are.
132
D. Pereboom
of realization, it will never be the case that a given higher-level property has
a conditional causal power different from any of those of its realizer base
property. (Wilson 1999: 50)
Thus, at the most basic metaphysical level, when it comes to causal powers of properties, the relation between the mental and the microphysical
is identity. Types and tokens of mental states are not identical to types
and tokens of microphysical states. Yet at the level of causal powers, the
mental and the physical are identical.
Donaldson points out that on Stephen Yablos (1992) position according to which the relation between the mental and the physical is the
determinable/determinate relation, the most basic relation also turns out
to be identity. For on Yablos view, the relation between the determinable
and the determinate is also the subset relation, and the relation between
metaphysically basic mental features and physical features will be identity
as well. The relation between red and scarlet is that possible instances
of scarlet are a subset of possible instances of red. Within that subset,
such instances of scarlet are identical with instances of red. Similarly,
the relation between pain and a neural realization of pain is that possible
instances of the neural realization are a subset of possible instances of
pain, and within that subset, such instances of the neural realization are
identical with instances of pain.
However, if modal multiple realizability arguments count against
token-identity claims for mental states and properties, they will count
against token-identity claims for mental causal powers as well, no matter
what, metaphysically, one thinks causal powers are. Here is a modal multiple realizability argument that targets mental causal powers directly
(Pereboom 2011). Consider Annes belief at some particular time that
her parents live in Manhattana mental token, an instance of a mental
propertyand the causal power that it has or (in my preferred view)
with which it is identical. Suppose Anne is threatened with an illness that
would damage a small part of her brain that has a crucial role in realizing
this belief (but other parts of her brain have important roles as well).
Before this part is damaged by the illness, a neurosurgeon could remove
it and replace it with a sophisticated electronic microprocessorlets call
it a silicon prosthesis. Imagine that the illness never actually materializes, and Anne does not undergo the operation. Still, this token belief
133
would have retained its token mental causal power had she undergone
the operation, and had it thus at that time been realized by the token
neural-and-silicon causal power instead of the token neural power that
actually realizes it. As a result, the token mental causal power cannot be
identified with a token neural causal power, in particular not with the
token neural causal power that actually realizes it.
A temporal variant of this argument can be constructed if we are
allowed the supposition that belief tokens can persist over a significant
duration. Imagine that Annes illness does continue to threaten, and
before the part of the brain is damaged, the neurosurgeon removes it and
replaces it with the silicon prosthesis. After the operation, Anne retains
her token belief about where her parents live, and it possesses the token
mental causal power it had before the operation. This mental token causal
power is, however, no longer realized by the pre-operation neural token
causal power but rather by a neural-and-silicon token causal power. For
this reason, the token mental causal power cannot be identified with any
neural token causal power, specifically not with the token neural causal
power that realized it just before the operation.
For this reason, the dependence relation between the mental and the
physical, even at the basic metaphysical level, wont be identity. In my
account of this relation, I invoke the made up of relation instead. Theres
a connection here with the recent literature on grounding. The grounding relation, its advocates typically argue, is a general relation, of which
constitution is just one type, one on which the ground necessitates what
is grounded (perhaps given certain circumstances, as in my C2), and the
ground exists in virtue of the grounded (Schaffer 2009; Rosen 2010; Fine
2012). Jessica Wilson (2014) argues that there is no work for a theory
of ground to do, given that there are many metaphysical dependence
relations, such as type or token identity, functional realization, classical
mereological parthood relation, the set membership relation, the proper
subset relation, the determinable/determinate relation that can do this
work. Wilson makes a plausible case for her claim. But note that if the
grounding relation is conceived as irreflexive and asymmetric in addition (e.g. Bennett 2011), identity doesnt qualify. And, as Ive argued, the
proper subset relation and the determinable/determinate relation feature
identity at their core. So if the mental and the physical are related in
these sorts of ways, the relation between the two will not at its core be a
134
D. Pereboom
grounding relation. So it may be that if grounding is conceived as asymmetrical and irreflexive, grounding isnt so commonly invoked after all
as the relation between the mental and the physical, and that theres a
controversial philosophical issue at stake.
Two Objections
To this account, Andrew Melnyk (2014) objects that the primitive made
up of relation is not physicalistically acceptable, and he argues that the
same is true for the brute metaphysical necessitation that the account
appears to invoke, that is, necessitation that is not explicable in other
terms, such as the necessity of identity or analyticity. In response, even
if these two notions do not meet the strict sort of test that he has in
mind, and that he sets out in his (2003) book,2 there he himself suggests
a weaker test, that the posited entity in question and the role it plays
provide no encouragement whatever to any familiar sort of antiphysicalist (2003, p.26). I claim this weaker physicalist test-passing status for
the made up of relation. When a physics teacher says that diamonds are
made up of carbon atoms, toying with some familiar sort of antiphysicalism doesnt come to mind. And suppose that I am indeed forced to
say that the necessitation in question is brute and not explicable by the
necessity of analyticity or the identity of necessity. The reason Melnyk
cites for the incompatibility of brute necessitation and physicalism is
that if the necessitation were brute, it wouldnt be anything physical that
would make higher-level property ascriptions true: because this neces2
Here is Melnyks (2003: 2032, 2014) account of realization, which, in his view, secures
physicalism:
Let p name a particular actual physical state token, and m a particular actual mental state token.
Then p realizes m (in the intended sense) only if
(i) m is a token of a mental state type with a certain higher-order essence, that is, m is a token of a
mental state type M such that for a token of M to exist just is for there to exist a token of some
(lower-order) state type such that tokens of that (lower-order) state type play role RM, the role distinctive of M;
(ii) p is a token of a physical state type P such that, necessarily, given the physical laws, tokens of P
under physical circumstances C play role RM; and
(iii) the physical laws hold and physical circumstances C obtain.
135
136
D. Pereboom
By contrast, the ideas of the moral sciencesand he later adds the metaphysical sciencesare subject to an absence of clarity, in the form of
ambiguity, and this lowers the quality of reasoning in those sciences:
137
But the finer sentiments of the mind, the operations of the understanding,
the various agitations of the passions, though really in themselves distinct,
easily escape us, when surveyed by reflection; nor is it in our power to recall
the original object, as often as we have occasion to contemplate it.
Ambiguity, by this means, is gradually introduced into our reasonings:
Similar objects are readily taken to be the same: And the conclusion
becomes at last very wide of the premises.
Hume concedes that the moral sciences have the advantage of shorter
and less complex inferences, and thus reasoning in those sciences is in
one sense easier.
The chief obstacle, therefore, to our improvement in the moral or metaphysical sciences is the obscurity of the ideas, and ambiguity of the terms.
The principal difficulty in the mathematics is the length of inferences and
compass of thought, requisite to the forming of any conclusion.
Hume then moves to the application of this methodological perspective to the idea of causal power. He claims that there are no ideas,
which occur in metaphysics, more obscure and uncertain, than those of
power, force, energy or necessary connexion, of which it is every moment
necessary for us to treat in all our disquisitions. And the point of the
section of the Enquiry on causation is to fix, if possible, the precise
meaning of these terms, and thereby remove some part of that obscurity, which is so much complained of in this species of philosophy.
Hume judges the idea of a fundamental causal power obscure, and
offers two influential proposals as substitutes, strict regularity: A causes
B just in case As are always followed by Bs; and counterfactual dependence: A causes B just in case: if As hadnt occurred, B wouldnt have
occurred.
However, we might ask: What is it, specifically, that recommends these
notions over the idea of causal power? Humes argument that the idea
of causal power has no corresponding impression in sensory experience
doesnt currently move many, and, moreover, he seems too quickly to set
aside the impression of ones own exertion of physical force as a source of
the idea. Many have argued that irreducible causal powers play an indispensable explanatory role, and all of this without the idea of causal power
138
D. Pereboom
being even broadly mathematical or logical. Can we say the same for the
idea of being made up of?
Thus there is a response to the obscurity charge against one genuinely
nonreductivist relation between the mental and the physical, namely,
material constitution construed as featuring the being made up of relation.
For there is a way to elucidate the difference between the being made up
of relation and mere dependence. One might agree that being made up of
is more obscure than identity, while resisting the claim that it is irremediably obscure. And a good case can be made that this relation has an indispensable explanatory role. Any mental/physical dependence relation that
issues in identity at its root is subject to multiple realizability arguments,
which are the standard type of argument against identity claims in this
general area of inquiry. This suggests that a relation other than identity is
required, and the material constitution construed as featuring the being
made up of relation is my candidate.3
References
Baker, L.R. (2000). Persons and bodies. Cambridge: Cambridge University Press.
Baker, L. R. (2007). The metaphysics of everyday life. Cambridge: Cambridge
University Press.
Baker, L. R. (2013). Perebooms robust nonreductive physicalism. Philosophy
and Phenomenological Research, 86, 736744.
Bennett, K. (2011). Construction area: No hard hat required. Philosophical
Studies, 154, 79104.
Boyd, R. (1999). Kinds, complexity, and multiple realization. Philosophical
Studies, 95, 6798.
Burge, T. (1978). Individualism and the mental. Midwest Studies in Philosophy
6, ed. P. French, T. Uehling, and H. Wettstein, Minneapolis: University of
Minnesota Press, 73121.
Donaldson, J. (2015). The superoverdetermination problem. Doctoral thesis,
University of Glasgow.
Fine, K. (2001). The question of realism. Philosophical Imprint, 1, 130.
3
Thanks to Louis DeRosset, David Christensen, Jessica Wilson, Ted Sider, Laurie Paul, Karen
Bennett, Nico Silins, Sydney Shoemaker, Lynne Baker, and Carl Gillett for valuable comments and
discussion.
139
140
D. Pereboom
Part II
Grounding, Science,
and Verticality in Nature
6
Ground Rules: Lessons fromWilson
JonathanSchaffer
Wilsons No Work for a Theory of Grounding (2014) offers an insightful critique of grounding-based approaches to metaphysical inquiry. She
argues that the notion of grounding is uninformative, disunified, and in
the end unhelpful. She then sketches a rival approach, which eschews the
notion of grounding, in favor of a plurality of small-g grounding-type
notions alongside a primitive notion of absolute fundamentality.
I think that Wilson is right to criticize many extant grounding-based
approaches for not being sufficiently informative. I just think that it is
possible for the grounding theorist to do better, and that my own (forthcoming) treatment in terms of structural equations (which are formal
models developed for understanding causal structure) does better in the
needed ways. I also think that her rival approach deserves serious consideration in its own right. But I argue that her approach is open to serious
criticisms, including every one of the criticisms she levels at the grounding theorist.
J. Schaffer (*)
Rutgers University, New Brunswick, NJ, USA
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_6
143
144
J. Schaffer
My subtitle is Lessons from Wilson since I am saying that all theoristsincluding Wilson herselfshould draw the lesson that one needs
more informative conceptions of metaphysical structure, of the sort I
take structural equation models to provide.
My main title is Ground Rules, which I offer as a three-way pun.
First, one of the underlying issues between Wilson and myself is whether
the notion of grounding is sufficiently unified for useful work. Such issues
of unity arise not just with the notion of grounding, but with virtually all
interesting philosophical concepts. So I want to articulate some general
ground rules for unity debates. Second, I want to discuss a formalism to
model grounding, which will display the rules of grounding. These first
two ideas are related, insofar as I think that the general way to adjudicate
unity debates is by looking at the best formalism, and seeing whether or
not it enfolds the notion in a unified set of rules. Third, I want to express
my continued enthusiasm for the notion of grounding as one that can
be framed in a unified, informative, and helpful way, and so I exclaim:
ground rules!
Overview: In section A Brief Introduction to Grounding, I offer
a brief introduction to the notion of grounding. In sections Are
Grounding Claims Informative? and Are Grounding Claims Helpful?, I
take up Wilsons two main objections to grounding-based approaches
that bare grounding claims are uninformative, and that such claims are
unhelpfuland extract two main lessons. In section Wilsons Pluralistic
Framework, I critique Wilsons rival pluralistic approach for, among
other things, not taking up Wilsons own lessons. I conclude in section Structural Equation Models to the Rescue by explaining how an
approach based on structural equation models for grounding has a special
claim to adequacy.
145
mental entities and thereby imposes structure over what there is. Some
entities are more fundamental than others (for instance, particles are
more fundamental than chemicals, and chemicals are more fundamental
than animals). Once one distinguishes more from less fundamental entities, it is natural to posit a relation linking certain more fundamental entities to certain less fundamental entities which derive from them. Grounding
names this directed linkage.
Grounding may be understood as the relation of dependence which
philosophers tried but failed to understand via the modal pattern of
supervenience. As Kim (1993: 167) notes:
Supervenience itself is not an explanatory relation. It is not a deep metaphysical relation; rather, it is a surface relation that reports a pattern of
property covariation, suggesting the presence of an interesting dependency
relation that might explain it.
Supervenience, after all, is a reflexive and (a fortiori) non-asymmetric relation, as well as a merely intensional relation that cannot distinguish features
of reality found at all the same possibilities. So grounding may be understood as the deep relation of dependence which shallow supervenience
analyses unsuccessfully targeted. In my view, one of the morals of the failure of the supervenience analysis is that the notion of metaphysical dependence is needed but unanalyzable, and hence best treated as primitive.1
Grounding then serves to back a distinctive sort of metaphysical explanation. If one wants to understand, for instance, why there is an H2O
molecule present, then one perfectly good sort of explanation for this fact
would involve the fact that an H, another H, and an O atom are arranged
and bonded in the right way. This is not a diachronic causal explanation, citing previous causes. (A diachronic causal explanation might for
instance cite the previous events in which hydrogen and oxygen gasses
were combined and exposed to a spark.) It is rather a synchronic metaphysical explanation, citing the more fundamental basis at the time. Just
as causation provides the direction and the linkage needed for causal
Of course I cannot prove the negative existential that there is no reductive analysis of the concept
of grounding to be found (though when has reductive conceptual analysis ever succeeded?); I only
mean to say that it is legitimate to use the concept regardless, without any such analysis to hand.
1
146
J. Schaffer
147
about whether normative goings-on exist; about whether, if they exist, they
are reducible or rather irreducible to (though still nothing over and above)
naturalistic goings-on; about how exactly normative goings-on are related
to naturalistic goings-on; about whether normative goings-on are efficacious and, if so, whether they are distinctively efficacious (that is, efficacious qua normative); and so on. Hence it is that naturalists almost never
rest with the schematically expressed locutions of metaphysical dependence, but rather go on to stake out different positions concerning how,
exactly, the normative or other goings-on metaphysically depend on the
naturalistic ones.
So Wilson (2014: 545) thinks that our friend Natalie has managed to tell
us almost nothing about how, exactly, normative and intentional goingson stand to naturalistic goings-on.
I think that there is something right and insightful here, but that it is
hard to identify exactly what. Or at least, I found Wilsons objection initially puzzling. For imagine a scientistlet us name him Sigmundwho
utters a bare causal claim, such as smoking causes cancer. Wilsons main
concerns about Natalie could equally be raised about our new friend
Sigmund. After allto mimic what Wilson saysscientists do not only
care about whether smoking causes cancer; they also care about whether
cancer exists, about how exactly smoking is related to cancer, and about
what the more fundamental physicochemical conditions underlying cancer are, and so on. So what? Surelywhatever the ultimate status of causation may bethese observations alone do not show that the notion
of causation is uninformative and unhelpful! So how could Wilsons
concerns possibly show that the notion of grounding is uninformative
and unhelpful?
Clearly, both Sigmund and Natalie have told us something informative. Sigmund has said something that rules out alternatives such as that
smoking is causally unrelated to cancer, or only related as a correlate of a
common cause (indeed decades of careful medical research, countered by
costly corporate propaganda, went into establishing his claim). Likewise
Natalie has said something that rules out alternatives such as those given
by certain forms of moral realism and by divine command theory (centuries of philosophical debate have centered on assessing her claim).
148
J. Schaffer
Both Sigmund and Natalie have also told us something helpful, insofar
as causation and grounding both serve to provide explanatory handles on
the world. Sigmund has said something that might help us understand,
for example, why a particular smoker has cancer, and likewise Natalie
has said something that might help us understand, for example, why a
particular natural situation is morally impermissible.
Of course neither Sigmund nor Natalie has said everything, but so
what? Given anything short of a maximally specific description of reality,
there will always be more information to add. What is the problem with
informative and helpful claims that merely leave some further questions
open? So when Wilson says that Natalie has said almost nothing, I want
to ask back, do you mean that Natalie has said nothing, or that she has said
something but should just say more? The former option strikes me as false
and the latter true but unobjectionable.3
Of course if Sigmund refuses to say anything further about the smokingcancer connection beyond smoking causes cancer, then s omething
has gone wrong. But the problem here is not with bare causal claims, nor
with the concept of causation they involve, but only with the strangely
silent theorist who refuses to do anything more than make such bare
causal claims. Likewise if Natalie should refuse to say anything further
about the naturalnormative connection, then something has gone
wrong. Wilson (2014: 549) speaks of the perversely uninterested metaphysician who only makes bare grounding claims and says nothing further. But the problem here is likewise not with bare grounding claims, nor
with the concept of grounding they involve, but only with the perversely
Two puzzling passages: Wilson (2014: 5445) says that it is not just that Grounding (failure of
Grounding) claims leave some interesting questions open; rather, it is that such claims leave open
questions that must be answered to gain even basic illumination about or allow even basic assessment of claims of metaphysical dependence, or associated theses such as naturalism. But I find this
puzzling since she does not say what she means by distinguishing merely interesting questions
from those that must be answered, or relatedly what she means by basic illumination. And I
think she is just wrong that bare grounding claims cannot be assessed. Natalies claim, for instance,
rules out alternatives such as divine command theory. So if divine command theory could be shown
to be true, Natalies claim would thereby be shown to be false.
Wilson (2014: 553; also 575) also takes up the analogy with bare causal claims. She allows that bare
causal claims are informative for entailing that their relata exist as distinct and causally connected
events, but admits no comparable value to bare grounding claims. But I find this puzzling as well
since, at least by my lights, grounding claims are informationally comparable: they entail that their
relata exist as non-distinct and grounding-connected entities.
3
149
150
J. Schaffer
151
So I agree with Wilson that Natalies bare grounding claim leaves open
how the grounding pattern works, and think that there is an important
lesson to be drawn from this. But I think that the lesson is not to discard
the notion of grounding but to develop it further, in ways that allow one
to go beyond bare grounding claims and add even more information
about the underlying pattern. Thus, I think that Wilson is best understood as offering the following lesson:
Wilsons first lesson: An account of grounding must give one more than just
the bare ideology of this grounds that; it must also allow one to make
sense of follow-up inquiry into how the connection runs, in terms of the
specific rule mapping the more basic inputs to the less basic output.
152
J. Schaffer
relations that are capable of answering these crucially basic questions about
the existential, ontological, metaphysical, and causal status of metaphysically
dependent goings-on.
153
154
J. Schaffer
155
allows us to infer Brynners cancer risk given that he smoked 100 cigarettes per day, to counterfactually consider what Brynners cancer risk
would have been had he smoked 0 cigarettes per day (or 20, or 200),
and to underwrite explanatory claims as to why Brynner developed lung
cancer. Such a framework would display the relevant connections between
causation, inference, counterfactuals, and explanation, which constitute
some of the generalizations that make causation worth positing. Likewise,
assuming that de Sades acting to inflict pain grounds his acting wrongly,
one should want a framework for grounding that allows us to infer the
wrongness of de Sades action given that he acted to inflict pain, to counterfactually consider the normative status of alternative courses of action,
and to underwrite explanatory claims as to why de Sade has acted wrongly.
My second answer as to why genus notions may still be helpful is that
without the genus notion one may be unable to enumerate the species. For
instance, a theorist who refused the general notion of causation would have
no clear way to enumerate her own small-c causal relations. To illustrate, let us imagine that she starts off by invoking some more specific
causative notions like baking, making, waking How can she continue?
She cannot say and all other species of causation because that would
be cheating (explicitly invoking the very notion of causation that she has
foresworn). And she cannot just say and so on because what could
that mean for her (besides serving as a device to implicitly invoke the very
concept of causation that she has foresworn)? Likewise, the theorist who
refused the general notion of grounding would have no clear way to enumerate her own preferred menu of small-g grounding relations. Wilson
herself (2014: 535) resorts to and so on when listing her own open-ended
plurality of small-g grounding relations, and so one must wonder how
she understands her own list to continue, if not in terms of listing further
species of the very genus notion that she has foresworn, namely grounding.
So I think that what is right and insightful in Wilsons complaint is
that the best way to determine whether a genus notion is helpful is to embed
the notion in a formalism which treats the notion in a unified way, and
reveals the generalizations one would miss without the notion. For without
such an embedding, one has no rules governing the notion. Accordingly,
I think that Wilson is best understood as offering the following further
lesson:
156
J. Schaffer
But she (2014: 576) claims that the notion of dependence is best understood as schematic, merely standing in for some yet-to-be-specified one
of her many small-g grounding relations:
[S]uch idioms should rather be taken just to advert, schematically or otherwise, to one or other of the specific metaphysical relationstype and
token identity, the functional realization relation, the classical mereological
part-whole relation, the causal composition relation, the set membership
relation, the proper subset relation, the determinabledeterminate relation,
etc.already on the scene.
So far it might seem as if Wilson was not positing anything metaphysically new, but on the contrary advocating that the new questions of
dependence should be interpreted schematically against the backdrop of
a plurality of old relations.
Not so: there is a twist. For Wilson (2014: 55862) considers an argument for groundingwhich she credits to Fine and Hellie, and labels the
157
So Wilson also posits her own primitive structuring concept of metaphysical inquiry, to help orient her plurality of small-g grounding relations in the right direction.
I pause to note howrhetoric asideWilson winds up largely agreeing with friends of grounding. The underlying point of agreement is that
the metaphysician needs a new primitive hyperintensional notion to go
beyond merely listing what exists so as to characterize the structure of reality. The main residual disagreement is whether this primitive should be
one of being-absolutely-fundamental, or the relative and linking notion of
grounding (/being-relatively-more-fundamental-than-and-linked-to). These
notions are respectively analogous to being-causally-initial and causing
(being-relatively-causally-earlier-than-and-linked-to).10 This is not to say
that there is no disagreement between Wilson and friends of grounding
Both grounding and causation are notions of a directed linkage, which is why they are both apt
to back explanation (section A Brief Introduction to Grounding). Note that the and-linked-too
bit is needed. This H atom on Earth is relatively more fundamental than that H2O molecule on
Mars just because atoms are generally more fundamental than molecules, even though this atom is
not linked to that molecule (cf. Bennett forthcoming: Chap. 5). Likewise this event on Earth yesterday is relatively causally earlier than that event on Mars today just because of the overall causal
10
158
J. Schaffer
such as myself (cf. Wilson 2014: 5623), but only that there is a merely
internecine disagreement between those friends of primitive hyperintensional notions of metaphysical structure such as Rosen and I who opt
to take grounding as primitive, those friends such as Wilson (also Sider
2011) who prefer to take being-absolutely-fundamental as primitive, and
those such as Fine who claim a need for multiple primitive notions in the
neighborhood.
That saiddegree of real disagreement asideI think that Wilsons
framework is clearly worthy of serious consideration. I have three objections, however, the first of which is that I think Wilsons framework is
impoverished compared to the grounding framework. It seems to me that
absolute fundamentality can easily be defined in terms of grounding (the
fundamental is that which has no deeper grounds), and so a framework
using grounding as a primitive can easily be used to say everything one
wants to say via absolute fundamentality. But there is no obvious definition to be found in the other direction, and so it is not at all obvious that
using absolute fundamentality as a primitive will allow one to say everything one wants to say in terms of relative fundamentality, or in the even
stronger linking terms of the grounding connection.
This impoverishment makes trouble for Wilson in scenarios in which
there is no fundamental level at all, but just a limitless descent of ever-
deeper structure. If such a scenario is metaphysically possible,11 it is trouble for Wilson, for her framework can attribute no metaphysical structure
to it. After all, when nothing is metaphysically fundamental, her primitive gives no guidance. But the friend of relative fundamentality can still
make sense of metaphysical structure in such scenarios, including the
guiding idea that things are getting ever more fundamental without limit.
This relative impoverishment also makes trouble for Wilson, with
respect to making sense of structure among non-fundamental entities.
temporal order, even though these two events are not linked. (My thanks to Ross Cameron for
insightful comments which prompted these clarifications.)
11
I myself have wavered over whether such a scenario is metaphysically possible. If parts are always
more fundamental than wholes, and if gunky structures with limitless descending chains of parthood are possible, then one seems to get scenarios with no fundamental entities at all (Schaffer
2003). But if one does not assume that parts are always more fundamental than wholes, then no
such argument looms, and one may be able to respect the intuition that there needs to be an ultimate ground of being (Schaffer 2010: Sect. 2.4).
159
Suppose that what is fundamental are just particles in the void, and consider the following three non-fundamental entities: my whole body, my
whole body minus my left shoulder, and my heart. Holding fixed that
particles in the void are fundamental, and holding fixed the mereological
and other small-g relations among these three entities, there still seems
to be a residual question as to the direction of fundamentality (and one
not so different in spirit from the question of whether the ultimate parts
or the ultimate whole is basic, which inspired Wilson to add a primitive
notion of fundamentality in the first place). Again Wilsons view seems to
give no guidance.12 So overall I do not think that Wilson has successfully
blocked the fix the direction of priority argument for grounding.
My second and third objections concern whether Wilsons own framework is equally liable to the criticisms she herself levels against grounding
theorists. It seems to me that Wilsons own framework does not adequately take up her own lessons, which (to repeat) were:
Wilsons first lesson: An account of grounding must give one more than just
the bare ideology of this grounds that; it must also allow one to make
sense of follow-up inquiry into how the connection runs, in terms of the
specific rule mapping the more basic inputs to the less basic output.
Wilsons second lesson: An account of grounding should be embedded in
a formalism that outfits the notion with unified rules and so reveals useful
generalizations one would miss without the notion.
As to Wilsons first lesson, while she goes beyond the bare ideology of
this grounds that, she also stops short of saying exactly how the grounding connection works, and instead settles for some in-between resting
point involving her small-g relations. So by Wilsons lights, if Natalie
just says that the natural grounds the normative, then what she has said
Wilson herself (2014: 5646) takes up a similar example, but I am afraid that I do not understand her reply. I read her as saying that the answer turns on whether one treats the entities involved
as fusions or as functionally defined entities. But I do not see how either treatment makes a difference within Wilsons framework, unless one also has some general principle of relative fundamentality for fusions or for functionally defined entities (etc.) For suppose that my whole body, my
whole body minus my left shoulder, and my heart are all understood as fusions, and that particles
are fundamental. I see no way to extract any conclusion as to relative fundamentality for these
fusions, without some general principle connecting parthood to relative fundamentality.
12
160
J. Schaffer
161
of her preferred concepts. So I cannot guess why she thinks that her preferred concepts are any better off than grounding. But it seems to me that
every concern she raises against grounding being too uninformative and
needing to be superseded by more specific relations could have been raised
with equal force against virtually every notion that she herself deploys.
Consider Wilsons primitive posit of fundamentality. Could the metaphysician rest with bare claims of the form this is fundamental (/this
is not fundamental)? Of course not. With the posit of fundamentality
will come the need to settle certain framework questions (e.g., does fundamentality entail existence?), and to integrate the machinery of fundamentality into the machinery of Wilsons small-g relations (e.g., Can
entities related by proper parthood both be fundamental? Can entities
related by set formation both be fundamental?) Andperhaps most relevantly given the current dialecticthere will be the question (one which
Wilson especially should face) as to whether there is a single unifed notion
of fundamentality, as opposed to a merely schematic notion standing in
for some yet-to-be-specified small-f status, such as being mereologically
atomic and being set theoretically elemental.
Or consider causation, as involved in Wilsons own notion of causal
composition. As I have argued throughout the preceding discussion, parallel issues of unity arise for both grounding and causation. Overall I find
it puzzling that Wilson dismisses the notion of grounding as disunified,
but then goes on to deploy notions such as fundamentality and causation
with no concern as to their unity. At minimum, she owes a reason for
thinking that her preferred notions are any better off.
So I conclude that Wilsons interesting and original view is not just
impoverished but also fails to heed her main lessons. Wilsons framework does not succeed in saying how exactly the grounding connection
works (at most it says something slightly more specific, using some intermediate determinables instead of specifying precise rules), and Wilsons
framework is not associated with any formalism by which the unity of
concepts is judged in a principled way (I see no stable general conception
of conceptual unity behind it, nor any reason to regard Wilsons preferred
notions such as fundamentality and causal composition as being any better off). I conclude that Wilsons alternative view is a step in the wrong
direction, even by her own lights.
162
J. Schaffer
Indeed my main criticism of these views (Schaffer forthcoming: Sects. 4.14.3) is that they conflate grounding with metaphysical explanation, which is tantamount to conflating causation with
causal explanation.
13
163
cies of dependence, and structural equations are our best technology for
modeling dependence.)
In what remains, I briefly sketch the structural equation models I now
favor for understanding grounding (Schaffer forthcoming), and explain
how this approach fully respects Wilsons lessons (in a way already known
to work quite well for causation) and thereby has a special claim to adequacy. I am not saying that such models are perfect or that there is no
further work to be done. I am only saying that the technology already
exists to provide a fairly informative conception of grounding, and so
absorb Wilsons lessons.14
It is useful to think of a structural equation model as constructed in
three stages (Halpern 2000). First, one is trying to model some portion of
reality, so one sets up some variables to represent the system under study. In
a structural equation model, one starts off by dividing these variables into
exogenous (/independent) variables representing the basis conditions, and
endogenous (/dependent) variables representing the resulting conditions,
with all of these variables allotted a contrast space of values serving as the
options under consideration. Second, one adds in a dynamics for the system, whichin the deterministic caseconsists of specifying dependence
functions which say, for each endogenous variable, what value it takes
as output given input values for certain other variables (which thereby
count as parent variablesno parenthood loops are permitted). Third,
one adds in an assignment, whichin the deterministic casespecifies
a unique value for each exogenous variable. Once one has specified the
assignment (/set the initial conditions) and the deterministic dynamics,
the value of every other variable is uniquely determined.
To illustrate, suppose that one is trying to model how the truth-value
of a conjunction p&q depends on the truth-values of its conjuncts p and
q. Then a natural classical model would take the system under study S*
to consist of a pair of exogenous variables P and Q, and one endogenous
variable R, each allotted 0 and 1 as options to represent falsity and truth.
The dynamics L* would say that the value of R is determined by the min
See Koslicki forthcoming for criticism of my use of structural equation models for understanding
grounding. Wilson (2014: 5705) claims that self-grounding and other grounding loops are possible, which would also constitute a line of criticism to structural equation models (at least in the
form I present them). I am not convinced but I lack the space to engage with the examples here.
14
164
J. Schaffer
function on the pair {P, Q}. And the assignment A* would set P and
Q to their actual truth-values, which I will suppose is P=1 and Q=1.
Formally, this may be stated as:
Or suppose that one is trying to model how the mass of an H2O molecule depends on the masses of its atomic parts: the H, the other H,
and the O.Working in a Newtonian regime and approximating a bit, a
natural model would take the system under study S** to consist of three
exogenous variables, H1, H2, O, and one endogenous variable, H2O,
each mapped to the positive reals (R+) to represent Daltons of mass. The
dynamics L** would say that the value of H2O is determined by the addition function on {H1, H2, O} (mass is additive in Newtonian systems).
And the assignment A** would set H1 to 1, H2 to 1, and O to 16 (these
are the approximate atomic masses). Formally, this may be stated as:
{ }
{ }
{ }
L * * = < S * *, {H 2O = H1 + H 2 + O} >
M * * = < L * *, {< H1, 1 >, < H 2, 1 >, < O, 16 >} >
165
166
J. Schaffer
given the actual masses of its atomic parts, one can reason counterfactually as to what mass the molecule would have had if the atomic parts had
had different masses, and one can underwrite an explanation as to why
the H2O molecule has a mass of 18Da.
Notice that the formalism itself, and the connections to matters such
as inference, counterfactuals, and explanation, is indifferent as to which
species of causal relation is present. One does not need red arrows for
pushing or production, and blue arrows for pulling or dependence or
some other type of causal connection. The mathematics works the same
regardless, and the connections to inference, counterfactuals, and explanation are the same regardless. In this way, structural equation models
justify causal monism, by outfitting the notion of causation with uniform
rules and thereby allowing one to say something informative and worthwhile about causal relations generally.16
Exactly the same case can be made for grounding monism, given a
structural equations model treatment. The mathematics does not care
which of Wilsons small-g grounding relations is present. One does
not need red arrows for composition and blue arrows for realization
or some other type of metaphysical connection. The mathematics works
the same regardless, and the connections to inference, counterfactuals,
and explanation are the same regardless. And so:
Second lesson learned: Structural equations models are embedded in a formalism that outfits grounding with unified rules and so reveals useful generalizations one would miss without the notion.
Corollary: the notion of grounding has exactly the same claim to unity as the
notion of causation.
Putting this together, in the case of causation the technology already
exists to go beyond bare causal claims, and to articulate exactly how cause
and effect are connected, as well as to embed causation in a formalism
which treats the notion in a unified way and reveals useful generalizations
about inference, counterfactuals, and explanation which one would miss
Pearl (2010: 72) offers exactly this style of reply to the pluralist Cartwright (2007), challenging
her to cite a single example that does not fit his unitary structural equations formalism.
16
167
if one refused the notion. This technology smoothly extends to grounding. And so one finds a ready-made way to take up Wilsons lessons and
to reach an informative conception of grounding, while making sense
of the deep analogies between causation and grounding as relations of
directed dependency, and while understanding the special power both
relations have of giving us explanatory handles on the world. So I conclude that Wilson has lessons to teach the grounding theorist, but equally
the grounding theorist has ways to learn these lessons.17
References
Anscombe, G. E. M. (1975). Causality and determinism. In E. Sosa (Ed.),
Causation and conditionals (pp.6381). Oxford: Oxford University Press.
Audi, P. (2012). Grounding: Toward a theory of the in-virtue-of relation. The
Journal of Philosophy, 109, 685711.
Bennett, K. (2011). Construction area (no hard hat required). Philosophical
Studies, 154, 79104.
Bennett, K. (forthcoming). Making things up. Oxford: Oxford University Press.
Boolos, G. (1971). The iterative conception of set. The Journal of Philosophy, 68,
21531.
Cartwright, N. (2007). Hunting causes and using them: Approaches in philosophy
and economics. Cambridge: Cambridge University Press.
Correia, F. (2005). Existential dependence and cognate notions. Munich:
Philosophia Verlag.
Daly, C. (2012). Scepticism about grounding. In F. Correia & B. Schnieder
(Eds.), Metaphysical grounding: Understanding the structure of reality
(pp.81100). Cambridge: Cambridge University Press.
Fine, K. (2001). The question of realism. Philosophers Imprint, 1, 130.
Fine, K. (2012). Guide to ground. In F. Correia & B. Schnieder (Eds.),
Metaphysical grounding: Understanding the structure of reality (pp. 3780).
Cambridge: Cambridge University Press.
Thanks especially to Jessica Wilson, and also to Kenneth Aizawa, Ross Cameron, Janelle Derstine,
Kit Fine, Kathrin Koslicki, Jon Litland, Meghan Sullivan, and audiences at the 2014 Eastern APA,
Fordham, and the Composition and Ground Workshop (2015) at Rutgers-Newark.
17
168
J. Schaffer
169
7
The Unity andPriority Arguments
forGrounding
JessicaWilson
Introduction
Grounding, understood as a primitive posit operative in contexts where
metaphysical dependence is at issue, is not able on its own to do any substantive work in characterizing or illuminating metaphysical dependenceor
so I have argued (Wilson 2014). Such illumination rather requires appeal to
one or other of the specific metaphysical relationstype or token identity,
functional realization, the determinabledeterminate relation, the mereological partwhole relation, and so ontypically at issue in these contexts.
In that case, why posit big-G Grounding in addition to the small-g
grounding relations already in the metaphysicians toolkit? The best reasons for doing so stem from the Unity argument, according to which the
further posit of Grounding is motivated as an apt unifier of the specific
relations, and the Priority argument, according to which Grounding is
J. Wilson (
)
Department of Philosophy, University of Toronto, Toronto, ON, Canada
171
172
J. Wilson
Preliminaries
The Dialectical Import ofGrounding
Why posit Grounding? Certain of the original proponents, including Fine (2001), Schaffer (2009), and Rosen (2010), initially motivate
Grounding as a neo-Aristotelian corrective to overly Quinean or empiricist approaches to metaphysical theorizing: rather than ignore metaphysical dependence or treat it, unsuccessfully, in empiricist-friendly
terms like entailment or supervenience, we should return to a traditional
Aristotelian concern with what is fundamental, and what depends on
what, understood in metaphysically substantive terms as involving a distinctive, primitive notion or relation of Grounding operative in contexts
where metaphysical dependence is at issue.
The rhetoric of revolutionary revival here is inspiring, and frequently
reproduced. But it is misleading, in ways that obscure Groundings dialectical import. To start, the initial motivation for Grounding is enthymematic. As Ive previously pointed out:
Attention to metaphysical dependence is not new: many, perhaps most,
contemporary metaphysicians have spent their careers investigating forms
of such dependence, typically assumed to go beyond merely modal or
173
Nor does the rhetoric of a neo-Aristotelian revival of concern with metaphysical dependence
make sense, for Aristotle operated with a variety of small-g relations, differently applied in different
cases, rather than with a primitive big-G conception. The point here isnt merely (anti-)rhetorical,
but also indicates that no ready appeal to an Aristotelian notion in good historical standing is available to proponents of primitive Grounding.
174
J. Wilson
physicalreductive physicalism, non-reductive physicalism, eliminativism, epiphenomenalism, strong emergentismare defined in terms of
their answers to these questions. Moreover, specific versions of these views
crucially appeal to features of specific small-g relations (or the lack of any
appropriate such relation) as motivating the answers at issue. Hence it is,
to focus just on a few non-reductive physicalist accounts, that Putnam
(1967) argues that taking mental states to be functional states accommodates the multiple realizability of the mental, that Yablo (1992) argues
that taking mental states to be determinables of physical determinates
accommodates the distinctive efficacy of the mental, and that Wilson
(1999) argues that taking mental states to have a proper subset of the
token powers of their physical realizers guarantees the distinctness and
physical acceptability of the mental.
As a primitive posit, however, Grounding is too abstract, on its own,
to provide answers to such questions, much less illuminating answers.
Suppose that the mental is Grounded in the physical. Does the mental
exist? Is it distinct from the physical? Is it epiphenomenal or not? If it is
efficacious, is it distinctively efficaciousefficacious qua mental? As is
reflected in the discussions of the original proponents, who express inclinations toward realism (Schaffer), anti-realism (Fine), and agnosticism
(Rosen) about Grounded goings-on, no clear answer even to the question
of existence follows from a Grounding claim. Similarly for failures of
Grounding claims. Suppose that the mental is not Grounded in the physical. Is this due to the mentals being a case in point of strong emergence,
substance dualism, eliminativism, expressivism, or what? Here again, no
answers follow just from attention to Grounding. Nor is there hope of
overcoming this underdetermination by supplementing Grounding with
general presuppositions entailing specific answers to such questions, not
just because such presuppositions will fail to accommodate various live
accounts of metaphysical dependence (a point to which I will return
down the line), but because such presuppositions will necessarily fail to
provide the fine-grained explanations of how these answers are generated
that attention to small-g relations is able to provide.
We are now in position to see past the rhetoric to the real dialectical
import of Grounding. The debate over Grounding is not over whether
metaphysicians should be concerned with dependence and prioritythey
clearly already are. Nor is it over whether investigations into dependence
175
The Relata
Proponents of Grounding differ somewhat as regards the metaphysical
category of this notion. Most commonly, Grounding claims are taken to
express the holding of a relation, but there is disagreement about whether
the relata of Grounding are facts, understood as states of affairs or
Russellian propositions (Fine 2001; Rosen 2010; Audi 2012), or rather
entities of diverse ontological categories (Cameron 2008; Schaffer 2009).
There is also disagreement about the adicity of the Grounding relation
(see Jenkins 2011; Schaffer 2012), and about whether there is one or
rather multiple primitive relations of Grounding, associated with metaphysical, nomological, and normative areas of inquiry (see Fine 2012).2
Accounts on which the relata of Grounding are broadly representational, or which aim to neutrally regiment claims about metaphysical
dependence via appeals to sentential or propositional operators, reflect
a conception of Grounding as entering into explanations, suited to be
reasoned with (as in Fines logic of ground).3 As I discuss in my (2014),
my view is that in specifying the relata relevant to grounding explanations, metaphysicians should talk about the worldly goings-on directly:
compare causation and causal explanation, where theorizing cuts to the
metaphysical chase; Schaffer (2012) makes a similar point. Hence, I will
follow Schaffer in characterizing Grounding as a relation whose relata are
worldly entities, perhaps supplemented (though here too I think there is
2
Note that the specific Grounding relations here, as well as the constituent and feature-based
forms of dependence discussed in Koslicki 2012, are distinct from the specific small-g relations
Ive flagged.
3
Hence Fine (2001, 15) says, We take ground to be an explanatory relation.
176
J. Wilson
Ideology andOntology
As well see, Schaffers pitch is sometimes presented in terms of our possessing a general concept of Grounding, rather than in terms of there
being a general relation of Grounding, even though his ultimate aim is
to motivate the latter posit. This may reflect the supposition that general
concepts bring general metaphysical posits in their wake. I think representation and reality can come apart, howeverin particular, I think that
even if there were a general concept of Grounding, it wouldnt immediately follow that there is a correspondingly general metaphysical posit,
since the concept might be given a deflationary treatment, as schematic
for or reducible to some specific metaphysical posit(s). Im interested in
the metaphysical question, so Ill pitch my remarks accordingly. Down
the line I will revisit the question of whether and when a general concept
(or associated general term) should be taken to motivate a correspondingly general metaphysical posit.
177
some evidence of real unity. At the very least, I would think it incumbent
on the objector to provide further reason for thinking that the general
term grounding denotes no unified notion. (377)
Here the primary motivation for unity stems from taking the smallg grounding relations to share the formal structure of a strict partial
orderthat is, to each be irreflexive, asymmetric, and transitive.
Schaffer has come to believe that a better Unity argument is needed.
This reflects, in part, that he now thinks that some cases of metaphysical
dependence are not transitive, and asymmetry and irreflexivity seem too
thin a reed upon which to hang a unified general posit of Grounding.4
This also reflects that, in response to my previously stated concerns, he
has come to appreciate the need for Grounding to provide a basis for
answering certain core questions about the status of Grounded entities
(or to say why a general notion of Grounding doesnt need to answer
them), and moreover to provide an account (though presumably one
more general than those provided by specific small-g relations) of how,
exactly, some goings-on metaphysically depend upon some others.
Schaffers new unification strategy, as per his contribution to this volume, as described in Schaffer (this volume), is threefold. First, he offers a
rulelet the best formalism decidefor determining when considerations of unity (defeasibly) support positing a general concept:
It seems to me that the best principled way to decide [whether to posit a general
notion] is to construct the best formalism one can for the concept. If there is no
meaningful concept, this should show up in a lack of any clear formalism, and
if there are many, this should show up in a need for a formal distinction. But if
one winds up with a clear and precise formalism that embeds the concept in a
unified way, then this is a good sign that there is a single unified concept. I offer
this as a general ground rule for unity debates: let the best formalism decide
(Rosen 2010; Schaffer forthcoming, Sect. 4.4). (153; emphasis in original)
4
See Schaffer (2012). Interestingly, Schaffers main case illustrating intransitivity (whereby a balls
surface being dented partly grounds the balls having a specific shape, and the balls having a specific
shape grounds its being more-or-less spherical, but the balls surface being dented doesnt ground
the balls being more-or-less spherical) involves mixing two different small-g relations (mereological
parthood and the determinable-determinate relation). A more straightforward case (see Wilson
2014) adverts to set membership: sets metaphysically depend on their members, but set membership is not transitive.
178
J. Wilson
Note that here Schaffers remarks target a general relation of causation as opposed to (just) a general concept of causation. Schaffer goes on to discuss two other reasons for endorsing genus-level
notions of causation and Grounding, respectively, according to which, first, the SEM formalism
enables one to make useful explanatory and predictive generalizations, and second, that such general notions enable one to speak open-endedly about the species-level notions or relations. Ill discuss these further motivations down the line when considering whether and how considerations of
formal or other forms of unity motivate an associated general metaphysical posit.
179
180
J. Wilson
Against Premise 1
If some phenomena are aptly formally unified, does this in itself provide
strong (albeit defeasible) reason to posit a unifier? It seems not.
The cases of determinables and determinates and of special science
entities are illustrative. Diverse determinates are formally unified in ways
181
that would be constitutive of general determinables, were determinables to irreducibly exist; moreover, we have terms for, and concepts of,
determinables. But these considerations alone are not taken to provide
strong (albeit defeasible) reason to posit determinables. On the contrary,
the most common treatments of determinables are along deflationary
anti-realist or reductionist lines, according to which (on an anti-realist
view) terms for/concepts of determinables are taken to be schematic for
determinate terms/concepts, or (on a reductionist view) determinables
are taken to be identical with disjunctions of determinates. Similarly for
special science entities: diverse lower-level physical goings-on are formally
unified in ways that would be constitutive of special science entities, were
special science entities to irreducibly exist; moreover, we have terms for,
and concepts of, special science entities. But these considerations alone
are not taken to provide strong (albeit defeasible) reason to posit special
science entities. On the contrary, the most common treatments of special
science entities are in deflationary anti-realist or reductionist terms.
Why is it that formal unity alone isnt typically seen as strongly (albeit
defeasibly) motivating the posit of a general unifier? The obvious reason is that parsimony considerationsgood old Ockhams Razorpush
toward give deflationary treatments of formal unity, wherever possible. Moreover, the methodological force of parsimony is not that of a
defeaterit is not as if unity considerations first motivate general metaphysical posits, which are then potentially defeated by considerations of
parsimony. Rather, parsimony considerations are first on the scene: thou
shalt not posit entities beyond necessity. So premise (1) is false.
Similar remarks apply to a version of premise (1) understood as incorporating two other broadly unity-based considerations: first, that the
SEM formalism enables one to make useful generalizations, pertaining
to prediction, counterfactual reasoning, and explanation; second, that
a general notion or relation provides a basis for speaking open-endedly
about the species-level notions or relations.6 Here again, attention to
the standard treatments of determinables and special science entities is
6
As Schaffer (this volume) says, the theorist who refused the general notion of grounding would have
no clear way to enumerate her own prefed menu of small-g grounding relations. Wilson herself
(2014: 535) resorts to and so on when listing her own open-ended plurality of small-ggrounding
relations, and so one must wonder how she understands her own list to continue, if not in terms of
listing further species of the very genus notion that she has foresworn, namely grounding (155).
182
J. Wilson
There are other motivations for deflationary strategies in these and other cases, including George
Berkeleys concerns about the coherence of general concepts or abstract ideas, and Jaegwon Kims
concerns about causal overdetermination.
8
See Wilson (1999) and (2009). The irreducibility at issue here is compatible with determinables
being posterior to determinates, as is usually assumed.
183
Against Premise 2
Are the diverse (small-c) causal relations aptly formally unified by the
SEM framework? It seems not.
To start, we need to get clear about what sort of unification of the
small-c relations is needed, if the SEM-based motivation for a unified notion of causation is going to serve as an analogue for a unified
posit of Grounding. Schaffer suggests that the SEM framework formally
unifies diverse causal relations, such as baking, making, waking. But
these sorts of small-c causal relations are not relevantly analogous to
small-g grounding relations. The diverse small-g relations are of importantly different forms of dependence that might be at issue in a given
caseagain, type and token identity, the determinable/determinate relation, the set membership relation, and so on. As such, if the unity associated with the SEM formalism is to provide a model for the unity of the
small-g relations, what is in the first instance required is that the SEM
formalism unify any diverse forms of causal relation that we have reason
to think exist. Here there is room for dispute, since some accounts of
causation are presented, rightly or wrongly, as competitors. As such, we
might not require that the SEM framework formally unify all candidate
9
See Wilson 2010. The irreducibility at issue here is compatible with the special science entities and
laws being posterior to the lower-level (e.g., quantum mechanical) entities and laws, as physicalists
assume.
184
J. Wilson
forms of causation, which include (among other contenders) regularity or nomological sufficiency accounts (on which causation is a matter
of instantiation of a causal law), dispositional essentialist accounts (on
which causation involves the manifestation of a disposition), transference accounts (on which causation involves the transfer of a conserved
quantity), and counterfactual accounts (on which causation is tracked by
certain counterfactual dependencies). After all, some of these might be
wrong.
Still, independent of the end of metaphysical causal inquiry, thanks
to Hall (2004) we have good reason to think that there are (at least) two
fundamentally different, and incompatibly applicable, forms of causal
relation: first, causation as production (covering regularity, transference, and powers-based accounts), and second, causation as (counterfactual) dependence. Both sorts of accounts are needed, Hall compellingly
argues, to accommodate various theses about causationthat causation
is transitive (Transitivity), that cause and effect are connected by spatiotemporally continuous processes (Locality), that the character of a causal
relation is determined by its intrinsic features in combination with the
laws (Intrinsicness), that counterfactual dependence between wholly
distinct events is sufficient for causation (Dependence), and that omissions can be causes and effects (Omissions)that are individually true
but jointly incompatible, as is illustrated by what theses are required in
order to handle (in particular) cases of double prevention. Halls somewhat conservative interpretation of this incompatibility is that there are
at least two notions of causation associated with different of the true
theses, which are operative in different cases of causal relation:
[T]he five theses I have mentioned are, I claim, all true. Given the deep and
intractable tensions between them, that can only be because they
characterize distinct concepts of causation. Events can stand in one kind of
causal relationdependencefor the explication of which the counterfactual analysis is perfectly suited (and for which omissions can be perfectly
suitable relata). And they can stand in an entirely different kind of causal
relationproductionwhich requires an entirely different kind of analysis (and for which omissions are not suitable relata). Dependence and
Omissions are true of the first of these causal relations; Transitivity, Locality,
and Intrinsicness are true of the second. (226)
185
Now, the SEM formalism does not unify these two fundamentally
different small-c forms of causal relation. On the contrary, as James
Woodward (see, e.g., his forthcoming) and others have noted, the SEM
formalism aims to accommodate causation as counterfactual dependence,
and does not aim to accommodate causation as production. This is no
surprise, since SEM accounts incorporate and model the driving intuition
behind counterfactual dependence accounts, according to which causes
make a difference to their effects: counterfactually wiggle the cause, and
the effect wiggles, too. As Schaffer put it in a previous draft of his paper,
structural equation models are our best technology for understanding
difference-making relations.10
Since the SEM framework models (at best) counterfactual dependence
accounts of causation and clearly does not model production accounts
of causation (much less diverse forms of such accounts), this framework
does not aptly formally unify the diverse small-c relations, in the relevant
sense. Premise (2) is false.
Against Premise 3
The falsity of premises (1) and (2) undermines the sub-conclusion of
Schaffers argument, according to which there is strong (albeit defeasible)
reason to posit a general notion of causation: the SEM framework does
not in fact formally unify the relevant small-c causal relations, and even
if it did, Ockhams razor would push toward trying to accommodate
such unity in deflationary terms, antecedent to positing a general unifier.
These results in turn technically undermine the value of premise (3)
according to which the diverse small-g grounding relations are just as
aptly formally unified by the SEM framework as the diverse (small-c)
causal relationsas generating the desired unity-based motivation for
Grounding, since at this point the truth of premise (3) is compatible with
the SEM frameworks not aptly formally unifying the small-g relations.
10
As such, the formal unity afforded by the SEM framework in modeling, for example, baking,
making, and waking is at best a unification of different applications of a counterfactual dependence account. But the formal unity associated with different applications of a single small-c relation is beside the point of motivating Grounding as a unifier of the diverse small-g relations.
186
J. Wilson
Let us put aside the analogy to small-c relations, however, and independently ask: does the SEM framework aptly formally unify the small-g
relations? It seems not.
We can start by observing that for the SEM framework to properly
model grounding relations, grounded goings-on must counterfactually
depend on grounding goings-on: wiggling the ground must wiggle the
grounded. Thats the whole point of the SEM-based approachto identify
or model dependence relations, of whatever sort, as reflected in counterfactual dependencies. But, for reasons Ill discuss shortly, many grounded
goings-on are not counterfactually dependent on grounding goings-on.
For example, and to start, suppose one wants to model the dependence
of determinables on determinates, against the backdrop assumption that
fundamental reality is maximally determinate. More specifically, suppose
one wants to model the metaphysical dependence of a shirts being red, at
a time, on the shirts being maroon, at that time. On the face of it, there
is no counterfactual dependence in this case; for both intuitively and on
every similarity-based account of counterfactuals (i.e., on nearly every
live account), the counterfactual if this shirt werent maroon, it wouldnt
be red is false, since in the closest worlds where the shirt isnt maroon, it
is some other determinate of red. Koslicki (2016) precisifies this observation by constructing an SEM model for this case, on which the exogenous
variables represent determinate states of the shirtMaroon, Crimson,
Navy, and so on; the endogenous variables represent determinable states
of the shirtRed, Blue, and so on; and the structural equations connect
these variables in the obvious ways (Maroon = Red, Navy = Blue, and so
on). Having done so, she notes:
There is now reason to doubt whether [] the model at hand actually
encodes how the shirts determinate shade sets its determinable color, as
Schaffer claims []. Given that Maroons being set to 0 leaves open, for
example, whether Crimson should be set to 1in the scenario in question
[it] would therefore be incorrect to define [the relevant structural equation] in such a way that it assigns 0 to Red whenever 0 is assigned to
Maroon; for the scenario in question may be one in which the shirt is
nevertheless red, only in some other way, for example, by being crimson
rather than maroon. This result presents a counterexample to Schaffers
187
slogan, wiggle the ground, and the grounded wiggles (Schaffer 2016,
Sect. 3.2): for in a case in which we wiggle the ground by imagining the
shirts color to be changed from maroon to crimson, say, it is not the case
that thereby the grounded wiggles as well, since the shirt continues to be
red, only in a different way. (107)
I agree with Koslicki both that the SEM formalism does not appropriately model the determinable/determinate case, and that the difficulty here is of a piece with notorious difficulties that difference-making
accounts of causation have in modeling cases of causal preemption. In
what follows, I want to defend, develop, and generalize these concerns.
To start, Schaffer would presumably reject Koslickis specific model
of the determinable/determinate case, on grounds that, like causation,
grounding is a contrastive notion, such that grounding claims have the
general form a rather than b grounds c rather than d. To be sure, relative
to the fine-grained contrast class of the exogenous variables in Koslickis
model, where Maroon contrasts not just with Navy but with other determinates of red, the counterfactual dependence between determinate and
determinable will not be in place, but, Schaffer might say, thats not a failure of the SEM modelits a failure of the modeler to correctly specify
188
J. Wilson
Nor, I think he would say, need we require (as Koslicki seems to assume) that the variables in the
determinable/determinate case exactly mirror the variables in the throw/shattering case that he
originally discusses, in representing the onoff obtaining of events or states of affairs.
189
contrast class to which one might appeal in order to set up any counterfactual dependencies, for whatever determinate shape properties the shirt
has, it will still be shaped.12
Ultimately, then, the SEM framework does not aptly model the metaphysical dependence of determinables on determinates, since gaining the
requisite counterfactual dependence requires an appeal to coarse-grained
contrast classes that may be either conceptually unavailable or metaphysically impossible. But, as Schaffer grants, one might well maintain that
determinables metaphysically depend on determinates.13 Hence, the
SEM framework does not formally unify (all) the small-g relations, and
premise (3) is false.
This result can be pressed further by pinpointing and generalizing the
underlying reason why the SEM framework goes wrong in the determinabledeterminate case. As above, Koslicki observes that the failure
of determinables to counterfactually depend on their associated determinates is analogous to the failure of effects, in cases of preemption, to
counterfactually depend on their associated causes: in both cases, the
dependence at issue is strongly immune to counterfactual variation,
effectively due to there being a manyone structure between dependence
base and dependent goings-on. In the determinabledeterminate case,
however, we can say more; for here the manyone structure is generated by the determinabledeterminate relations being, on the operative
understanding, an abstraction relation taking more specific to less specific
goings-on. It is the washing away of determinate-level details that makes
it the case that wiggling the determinate often and sometimes necessarily fails to wiggle the determinable.
That the underlying concern in the determinabledeterminate case is
generated by this relations taking more to less specific goings-on spells further trouble for the claim that the SEM formalism aptly unifies the smallg relations. For many paradigm cases of small-g relations are such that the
associated grounded goings-on are ontologically and causally less specific
than their grounding goings-on. This is true, for example, of many of the
12
Indeed, even if the shirt is a vague object, it will still be (indeterminately) shaped.
Indeed, one might well maintain this even if, as I argue in Wilson 2012, determinables can be
fundamental; for as I discuss in Wilson 2014, metaphysical dependence can be symmetric.
13
190
J. Wilson
191
Perhaps some small-g relations are able to fix the direction of priority on their own; in cases of set
membership, for example, perhaps members are always prior to their containing sets.
192
J. Wilson
16
193
which the fundamental goings-on are self-grounding (as per, e.g., a selfsustaining god) or mutually grounding (as per, e.g., Leibnizian monads).
Rather, we should metaphysically characterize the fundamental in primitive, metaphysically neutral termsafter all, if anything is fundamental,
its the fundamental! Though the fundamental is primitive, we can say
more about this notion; namely, that it follows from what goings-on are
fundamental at a world that these, individually or together, provide a
ground (nota bene: in one or other small-g fashion) for all goings-on at
the world. Such a conception encodes the intuitive, commonly registered
understanding of the fundamental in terms of all God had to do to create the world. Perhaps, in drawing to attention that what is fundamental
sets the valence for certain priority relations, I may be seen (as Schaffer
suggests) as here introducing a new hyperintensional primitive notion;
but really, I think I am just making explicit the presuppositions of standard metaphysical methodologyas in, for example, Schaffers (2010)
descriptions of monism and pluralism:
The monist holds that the whole is prior to its parts, and thus views the
cosmos as fundamental, with metaphysical explanation dangling downward from the One. The pluralist holds that the parts are prior to their
whole, and thus tends to consider particles fundamental, with metaphysical explanation snaking upward from the many. Just as the materialist and
idealist debate which properties are fundamental, so the monist and pluralist debate which objects are fundamental. (31)
Indeed. And the way the monist and the pluralist go about debating
which objects are fundamental is, again, to first assume (as a working,
speculative, or antagonistic hypothesis) one or the other fundamental
base, and then go on to explore which such base best accommodates the
rest of the reality, by appeal (in sometimes complex fashion) to various
small-g relations understood as holding between fundamental and nonfundamental goings-on.
Moving now to the second case: what about priority relations between
goings-on each or all of which are non-fundamentalsay, between
hands and bodies? A specification of the fundamental wont, in itself,
always fix the direction of priority between such non-fundamenta: for
194
J. Wilson
195
Expressive Power
Schaffer claims that the primitive fundamentality framework is expressively impoverished as compared to the Grounding framework:
It seems to me that absolute fundamentality can easily be defined in terms
of relative fundamentality (the fundamental is that which has no
deeper grounds), and so a framework using relative fundamentality as a
primitive can easily be used to say everything one wants to say via absolute
196
J. Wilson
I respond by restating my (2014) position that the notion of the fundamental17 should not be defined in terms of relative fundamentality (the
fundamental is that which has no deeper grounds); for we should not
metaphysically define or characterize the fundamental in non-basic, theoretically loaded terms. In particular, it is not true that a framework using
relative fundamentality as a primitive can easily be used to say everything
one wants to say via absolute fundamentality, since the relative fundamentality framework will not allow one to express a number of currently live
metaphysical theses, including views positing a self-sustaining God, mutually dependent monads, and so on. And while it is true that on my account
there is no definition of relative fundamentality just in terms of fundamentalityor, more to the metaphysical point, that the facts about relative fundamentality are not generated just by the facts about fundamentalityon
my view the facts about relative priority, if such exist in a given case, are
generated by facts about what is fundamental, coupled with off-the-shelf
resources about small-g relations and their features and implications in specific circumstances. Since my framework can say everything Schaffers can
say and then some, it is not a fundamentality framework but a Grounding/
relative fundamentality framework that is expressively impoverished.
Contra Schaffers exegesis, I do not use the expression absolute fundamentality, or appeal to any
such notion, for reasons that I discuss in my 2014 (note 64), and upon which I will expand shortly.
197
18
The possibility broached here might also be seen as a metaphysical version of the temporal supertask discussed in Cameron 2008, 9.
198
J. Wilson
Such a level acts as a fundamental base for what lies above; hence, for
example, the physical level might operate as a fundamental level for purposes of understanding priority relations among broadly scientific phenomena, even if the physical entities are non-fundamental relative to
some deeper level of reality.19 The existence of this sort of structure would
also provide a principled basis for fixing the valence of priority; hence
it is that Montero (2006) suggests that the physicalist commitment to
the priority of the physical over the mental can be accommodated in an
infinitely decomposable world as the thesis that all mental properties
are eventually determined by non-mental properties such that all further
determinations of these properties, if any, are non-mental (187).
Again, these possibilities indicate that associating a direction of priority with a given small-g relation (assuming that the relation does not do
this on its own) does not require an absolutely fundamental level; rather,
an appropriately principled metaphysical asymmetry in the structure of
reality, enabling some goings-on to act as a fundamental level, will do.
But what if there is no fundamental level, no convergence on a fundamental level, and no level at which deeper archeology ceases to matter?
In that case, one may reasonably deny that it makes sense to posit any
priority relations between non-fundamenta (besides those fixed just by
the relation alone, as might be the case with the set membership relation).
As Leibniz said in his correspondence with Arnaud,
Where there are only beings by aggregation, there are no real beings. For
every being by aggregation presupposes beings endowed with real unity,
because every being derives its reality only from the reality of those beings
of which it is composed, so that it will not have any reality at all if each
being of which it is composed is itself a being by aggregation, a being for
which we must still seek further grounds for its reality, grounds which can
never be found in this way, if we must always continue to seek for them.
(1686/1989, 85)
19
199
Here, as Cameron (2008) puts it, the thought is that if everything were
dependent, there would be no grounding to being (67). In further
support of this thought, consider again the all God had to do to create
the world heuristic that underlies the conception of the primitively fundamental in my framework. In cases where there is no convergence, and
no level of goings-on that fix the priority relations at higher levels, then
to create the world God would have to bring into being all the levels, and
all the goings-onthat is, God would have to do, or create, everything.
Hence, on the operative understanding of the fundamental (and again,
modulo any fixed directions of priority there might be) everything would
be on a par, priority-wiseand thats just to be expected. Indeed, Schaffer
(2010) agrees: There must be a ground of being. If one thing exists only
in virtue of another, then there must be something from which the reality
of the derivative entities ultimately derives (37).20
Relative Fundamentality
Schaffer claims that a primitive fundamentality framework has difficulty
accommodating priority relations between non-fundamenta:
[Wilsons framework has trouble] in making sense of structure among nonfundamental entities. Suppose that what is fundamental are just particles in
the void, and consider the following three non-fundamental entities: my
whole body, my whole body minus my left shoulder, and my heart. Holding
fixed that particles in the void are fundamental, and holding fixed the
mereological and other small-g relations among these three entities,
there still seems to be a residual question as to the direction of fundamentality (and one not so different in spirit from the question of whether the
ultimate parts or the ultimate whole is basic, which inspired Wilson to add
a primitive notion of fundamentality in the first place). [] Wilsons view
seems to give no guidance. (1589)
20
Schaffers commitment to a fundamental level opens the door to a fourth response to his objectionnamely, to deny that it makes sense to posit a world without any fundamental base. Hence
it is that in his (2010) he offers as an advantage of monism that it can accommodate both infinite
decomposition and the reasonable assumption that dependence relations require a ground of
being.
200
J. Wilson
Schaffer notes, more specifically, that I provide an account of relative fundamentality, which aims to treat the sort of case he mentions, but says:
I am afraid that I do not understand her reply. I read her as saying that
the answer turns on whether we treat the entities involved as fusions or
as functionally defined entities. But I do not see how either treatment
makes a difference, within Wilsons framework, unless we have some
general principle to hand of relative fundamentality for fusions or for functionally defined entities (etc.). For suppose that my whole body, my whole
body minus my left shoulder, and my heart are all understood as fusions,
and that particles are fundamental. I see no way to extract any conclusion
as to relative fundamentality for these fusions, without some general principle connecting parthood to relative fundamentality. (159, note 12)
201
Here it is also worth pointing out that what priority relations hold
between non-fundamenta is further complicated by its being the case that
the dependence of a given non-fundamental entity on another is typically
wrapped up in not one, but a number of relations. For example, even
supposing that there is a clear sense in which my hand qua functionally
specified entity depends on my body qua functionally specified entity,
but not vice versa, there is an equally clear sense in which the weight of
my body, at least at the present moment of writing, is in part metaphysically dependent on the weight of my hand, in that the weight of my body
is clearly an additive function of the weight of its parts, and certainly the
things that my body, even if abstractly functionally specified, can doits
powers, so to speak, depend to some extent on the powers of my hand.
That relations of relative fundamentality are not properly seen as subject to uniformly applicable general principles, for even a single small-g
relation, much less (as the proponent of Grounding supposes) for all such
relations, spells deep and to my mind insuperable trouble for a framework that appeals to primitive Grounding/relative fundamentality. As I
previously rhetorically asked:
Is all this complexity supposed to involve numerous Grounding relations,
primitively pointing in different directions? The idea is just plain silly, and
suggests that, even if there were some problem (which there is not) with the
specific relations not being themselves up to the task of fixing directions of
priority among non-fundamental goings-on, the posit of additional
Grounding relations would not be of any help. (Wilson 2014, 566)
Non-rhetorically: no. Nor, given that the SEM framework does not formally unify the small-g relations, can this framework be appealed to as
a general basis for rendering the primitive pointings of Grounding any
more substantive. We can do no better, in such investigations, than to
work closely with the diverse relations that are plausibly taken to hold
between non-fundamental goings-on, as informed by the accounts of
these non-fundamenta in terms of fundamental goings-on (or goings-on
that properly serve as fundamental), making explicit in the process what
assumptions are guiding our claims that one or another of these is, in a
given case, operating as a grounding relation (or not).
202
J. Wilson
Summing up: the primitive fundamentality framework is not subject to any of Schaffers three concerns: it is not expressively impoverished, it has resources for dealing with the absence of a fundamental
level, and it accommodates relative fundamentality in an appropriately
fine-grained and articulate way. This is no surprise, since the framework
encodes the usual suppositions and strategies of standard metaphysical
investigations into dependence and priority. Moreover, and by way of
contrast, the primitive Grounding/relative fundamentality framework is
both expressively impoverished andfor all that Schaffer has yet establisheddeeply inarticulate as regards characterizing the diverse and
complex network of relations of relative fundamentality. I conclude that
considerations of priority provide no reason to posit a general relation
of Grounding.
Acknowledgements Thanks to Jonathan Schaffer, an interlocuter par excellence, and to the Dr. Martin R. Lebowitz and Eve Lewellis Lebowitz Foundation,
along with the Phi Beta Kappa Society and the American Philosophical
Association, for making possible the fruitful debate that led to this paper. Thanks
also to members of audiences at the Eastern APA and Fordham Lebowitz Prize
lectures, the University of Buffalo, the University of Notre Dame, the University
of Tennessee, the University of Edinburgh Workshop on Grounding, and the
Rutgers Newark Workshop on Composition and Ground, for helpful comments
and questions.
References
Audi, P. (2012). Grounding: Toward a theory of the in-virtue-of relation. Journal
of Philosophy, 109, 685711.
Cameron, R.P. (2008). Turtles all the way down: Regress, priority and fundamentality. Philosophical Quarterly, 58, 114.
Fine, K. (1999). Things and their parts. Midwest Studies in Philosophy, 23,
6174.
Fine, K. (2001). The question of realism. Philosophers Imprint, 1, 130.
Fine, K. (2012). Guide to ground. In F. Correia & B. Schnieder (Eds.),
Metaphysical Grounding (pp. 3780). Cambridge: Cambridge University
Press.
203
204
J. Wilson
Spirtes, P., Glymour, C., & Scheines, R. (1993). Causation, Prediction, and
Search. NewYork, NY: MIT Press.
Wilson, J.M. (1999). How superduper does a physicalist supervenience need to
be? The Philosophical Quarterly, 49, 3352.
Wilson, J. M. (2009). Determination, realization, and mental causation.
Philosophical Studies, 145, 149169.
Wilson, J. M. (2010). Non-reductive physicalism and degrees of freedom.
British Journal for the Philosophy of Science, 61, 279311.
Wilson, J. M. (2012). Fundamental determinables. Philosophers Imprint, 12,
117.
Wilson, J. M. (2014). No work for a theory of grounding. Inquiry, 57, 145.
Wilson, A.(in progress). Metaphysical causation.
Woodward, J. (forthcoming). Causation in science. In P. Humphreys (Ed.), The
Oxford Handbook of the Philosophy of Science. Oxford: Oxford University
Press.
Yablo, S. (1992). Mental causation. The Philosophical Review, 101, 245280.
8
The Metaphysics ofNature, Science,
andtheRules ofEngagement
CarlGillett
The Scientific Revolution was powered, at least in large part, by explanations that pierced the manifest image of common sense by explaining its
level of everyday individuals, properties, and processes using qualitatively
distinct, lower-level entities taken to compose them. And such explanations have now been iterated through all the levels of nature. For example,
we take the corrosive action of glaciers to be explained by the movement
of the ice molecules that we take to compose glaciers. We explain the
motility of cells using the properties and relations of the molecules that we
take to compose them. We understand why kidneys clean blood in terms
of the properties and relations of the cells taken to compose them. And we
could easily go on, and on, through such explanations across the sciences.1
This paper makes explicit the methodology used in Gillett (2016) and overlaps with some of the
arguments offered in Chap. 2 of the latter.
C. Gillett ()
Northern Illinois University, DeKalb, IL, USA
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_8
205
206
C. Gillett
2
Analytic metaphysicians sometimes complain that I am using composition in a way different
from their usage, but it is important to emphasize that I am simply following a long-standing, and
continuing, scientific usage of the terms composition and compose that stretches back at least
to Sir Isaac Newton. Throughout the paper, unless I indicate otherwise, my usage of composition
therefore solely refers to the relations and concepts deployed in compositional explanations. If there
are still further concepts of composition used in the sciences (Healey 2013), then I do not discuss
them here.
3
Contemporary work on compositional explanation, including the species of it relating processes
in so-called mechanistic explanation, goes back at least to early work by Dennett (1961) and
Fodor (1968), through Wimsatt (1974) and Cummins (1983), down to more recent work such as
Bechtel and Richardson (1993), Glennan (1996), Machamer et al. (2000) and Craver (2007),
among many others.
207
There is a whole lot more to the ontology of nature than what I am terming the metaphysics of
nature that is limited to issues about verticality, but framing this limited area will facilitate my
discussion here. Many of my points can plausibly be extended to other ontological issues ontological
issues about nature falling outside of the metaphysics of nature as I have defined it here.
208
C. Gillett
Building on the results of Step 1, we can then turn to the next phase
which is that of constructing theoretical accounts for scientific notions
of composition whose success is judged by how well they do at capturing
the features of such concepts, and the explanations using them, illuminated in Step 1. We can roughly frame this phase as follows:
(Step 2Constructive Engagement) Formulate theoretical accounts of scientific notions of composition, whether the partwhole relations of individuals, realization of property/relation instances, or implementation of
processes, to accommodate their features and other evidence outlined in
Step 1, where such frameworks are judged by how well they succeed or fail
at accommodating such characteristics illuminated in Step 1.
209
Scientific reductionists argue that reflection on such explanations shows there are no composed
entities or compositional entities in nature, since the scientific reductionist argues that the truthmakers for compositional explanations are very different in character. So a wide array of positions
on metaphysics of nature can be, and are, adopted in Step 3.
210
C. Gillett
211
I do not deny that on the lists of Grounding relations in some articles on Grounding we occasionally find what appear to be scientific examples. However, what I do deny is that we ever find a
detailed examination of such compositional explanations in the sciences informing accounts of
Grounding.
212
C. Gillett
covering SLAM entities and their relations, formal features have been
outlined to capture the resulting notion of Grounding and associated
concepts.
Given the latter approach, Grounding accounts of scientific composition are pretty clearly Appropriational and Unengaged because they
were not directly developed by looking at compositional explanations
in the sciences, but were developed for vertical relations in the abstract
realm and then extended to other vertical relations. This approach is
less surprising when one realizes that many defenders of Grounding
appear to accept a division among the types of explanation, and relations underpinning them, across science and metaphysics. Thus, we find
Fine telling us Ground, if you like, stands to philosophy as cause stands
to science (Fine 2012, p. 40). Furthermore, Fine elaborates on this
picture:
A number of philosophers have recently become receptive to the idea that,
in addition to scientific or causal explanation, there may be a distinctive
kind of metaphysical explanation, in which explanans and explanandum
are connected, not through some sort of causal mechanism, but through
some form of constitutive determination. (Fine 2012, p.38)
The resulting view is one that takes the sciences and their explanations
to always focus on relations of causation, while taking metaphysics (philosophy?) to focus on constitutive explanations based around
Grounding. It is therefore unsurprising that in formulating their accounts
of vertical relations, proponents of Grounding frameworks do not use
compositional explanations from the sciences because many proponents
of Grounding are unsure there are any such explanationsa stance we
see in the next section is also increasingly common in the philosophy of
science as well. However, proponents of Grounding take this relation to
be the vertical relation underpinning all constitutive explanations, thus
presumably including compositional explanation in the sciences.
Accounts of Grounding are not easy to articulate in detail, since
Grounding is taken by its proponents to be a primitive notion that is not
213
open to definition using other terms. And there are a range of differences
between proponents of Grounding over further issues such as the exact
relata of Grounding, whether Grounding is a worldly relation or not,
and more. However, what we can safely say is that Grounding has the
features of the relations between the SLAM entities used as exemplars of
such relations, including allowing causally inert entities as their relata,
and that Grounding is taken to have certain formal features captured by
logical systems.
To make our discussion less abstract, I focus on the work of Schaffer
who allows the relata of Grounding relations to include entities like
properties, individuals, and processeshence offering the most promising approach to the kinds of relation we will see in the next section
are posited in compositional explanations in the sciences. Schaffer takes
Grounding relations to be well-founded and to take the form of a partial
ordering. And Schaffer also explicitly states that Grounding relations do
not require causal entities as relata. In addition, Schaffer is like other
writers on Grounding in claiming his account of Grounding covers all
vertical relations wherever we find them, including those in the natural
world. More pertinently, Schaffer claims that Grounding is the relation
that backs constitutive explanations wherever one finds them (Schaffer
this volume)thus taking scientific composition, the relation that underpins or backs compositional explanation, to be identical to Grounding.
Furthermore, in another important development of his account Schaffer
(2016, this volume) claims that we should favor an account of vertical
relations that fits the best formalism, that structural equation modeling drawn from the sciences is the best formalism for verticality in various explanations, and that such structural equation modeling best fits his
Grounding account.
In coming sections, I am going to assess whether Grounding, and
Schaffers account in particular, provides an adequate account of the
vertical relations underpinning compositional explanations in the
sciences. For, as we have seen, its proponents claim Grounding is
the vertical relation that backs all constitutive explanation wherever
we find it.
214
C. Gillett
215
cases of compositional explanation, but the neo-Causal approaches to scientific composition are again still Appropriational. Neo-Causal accounts
are plausibly not constructed following Steps 1 and 2 to build frameworks tailored to capture the nature of scientific composition after examining actual cases. Instead, the neo-Causal approaches take machinery
already developed for causation and then revise this machinery to putatively accommodate scientific composition, so this is plausibly another
Appropriational strategy.
I cannot sketch all of the neo-Causal approaches, so let me briefly note
a number of these accounts and then focus on the most prominent example of such a view. For instance, some versions of neo-Causalism take
existing accounts of causation, alter these accounts in various ways, and
produce frameworks for scientific composition. Jens Harbecke (2010,
2014a, b) uses regularity accounts of causation in this way and Mark
Couch (2011) takes Mackies INUS-based account to underpin his treatment. And writers like Totte Harinen (forthcoming) claim that scientific
composition relations are quite literally identical to causal relations.
However, the most prominent example of such an account is found in
the work by Carl Craver (2007) so I consider his framework in detail, but
the critical assessment I later provide of Cravers account plausibly carries over to the other neo-Causal approaches. Craver has popularized the
strategy of using interventionist approaches to causal explanation, and
their manipulability relations, as a way to characterize the compositional,
or as Craver terms them constitutive, relations between processes that
underpin compositional explanations. For my purposes, the interesting
feature of Cravers approach is that he is charitably interpreted as providing a sufficient condition for the composition of processes that is built
primarily around the mutual manipulability of processes.
Craver is quite explicit that his approach is an extension to compositional relations and explanations of a framework developed for causal
relations and explanations. And Craver takes this feature to be a merit of
his view. Craver tell us:
The mutual manipulability account is a plausible condition of constitutive
relevance because it fits well with experimental practice and because it is an
extension of the view of etiological [causal] relevance. (Craver 2007, p.162)
216
C. Gillett
Craver outlines the heart of his account of the composition of processes in this passage (where -ing and -ing are processes grounded
by the relevant individuals X and S):
My working account of constitutive [i.e. implementational] relevance is as
follows: a component is relevant to the behavior of a mechanism as a whole
when one can wiggle the behavior of the whole by wiggling the behavior of
the component and one can wiggle the behavior of the component by wiggling the behavior as a whole. The two are related as part and whole and
they are mutually manipulable. More formally: (i) X is a part of S; (ii) in the
condition relevant to the request for explanation there is some change to
Xs -ing that changes Ss -ing; and (iii) in the condition relevant to the
request for explanation there is some change to Ss -ing that changes Xs
-ing. (Craver 2007, pp.1523. Original emphasis)
The first condition demands partwhole relations between the individuals that ground the composed and composing processes, that is, between
the - and -ing individuals. The other two conditions require a relation of mutual manipulability. It is important to emphasize that Craver
deploys the sophisticated machinery of interventionism to articulate
these core ideas and that Craver also adds a number of further nuances to
his final account in order to address phenomena he highlights in scientific practice.7 However, I am going to leave these nuances to one side in
my discussion, since it is the adequacy of the core mutual manipulability
conditions that I am interested in.8 The basic idea is that mutual manipulability, suitably qualified, suffices for the composition of processes in the
sciences.
In this case, Craver is not importing a metaphysical framework developed outside the sciences in order to characterize scientific phenomena.
Nonetheless, Cravers, and the general neo-Causalist, strategy is
For example, Craver adapts his conditions so they take account of redundancy. See Chap. 4 of
Craver (2007), and particularly section 8, for the details.
8
The type of objections to Cravers account that I offer in the next chapter also apply to Cravers
more sophisticated, amended account.
217
218
C. Gillett
219
10
Dennett utilized these distinctive features in articulating his influential methodology for a scientific psychology. Consider Dennetts example of how we explain a capacity for face recognition as
composed by a team of lower-level entities. Famously, in such cases, Dennett clarifies that we can
understand richly intentional entities as composed by teams of entities with more rudimentary, and
hence qualitatively distinct, intentionality, which in turn can be understood as composed by teams
of entities with still more rudimentary intentionality.
11
Fodor (1968, 1974).
12
See Piccinini (2004) for a survey of these developments and their results.
220
C. Gillett
See Gillett (2002, 2003, 2010) for an outline, and critical evaluation, of the basic features of the
Subset/Flat view of realization. The latter critique is extended below.
221
222
C. Gillett
this way the neuron slowly moves. Given its complexity, I am going to
focus solely on the first step of the neurons movement in its protrusion.
The basis for our molecular explanation of cellular movement is complex, but at its core is the behavior of the actin molecules found in the cell
and particularly its cytoskeleton. Narrowing down just to focus onto our
molecular explanation of cellular protrusion, the crucial molecular processes are easily outlined. Crucially, the cell is filled with monomers of
globular actin (G actin) in the form of unchained actin molecules. One
important feature of actin is that it can polymerize swiftly in long filaments (F actin). And this is what we find in neuronal movement. When
stimulated in a certain direction, many filaments of actin are all formed
within the neuron pushing the molecules composing the cell membrane in
a certain direction, given the rigidity of the molecules in the cells cytoskeleton, until other molecules attach the protrusion to the surface. We consequently explain the protrusion of the cell in large part using these directed
polymerizations of monomers of G actin into many filaments of actin that
press on the membrane in the direction in which the cell is travelling.
Notice that in this case scientists take actin and other molecules to be
parts or constituents of the neuron. Furthermore, we have compositional explanations of the property of motility of the cell using properties and relations of the constituent molecules taken to compose this
property instance. And we have a compositional explanation of the cells
protruding using the molecular processes of polymerization, and other
processes, taken to compose this cellular process. So we not only have
compositional relations posited between individuals, or partwhole
relations, but also compositional relations between property instances
in realization relations, and between processes in relations of implementation. Table 8.1 summarizes these compositional relations and my
favored terms for them. There are all manner of interesting characteristics
of such compositional relations that I note in the next section, but let me
briefly outline a related group of compositional explanations.
The same molecular phenomena are also used to explain a property, and
associated process, of adult neurons that can swiftly grow dendritic spines
in the direction of an electrical stimulus. Our compositional explanation,
at the molecular level, of how the neuron grows dendritic spines, and so
quickly, is based upon the same components. The neuron has receptors
223
Table 8.1 Table of the various kinds of compositional relation, the category of
entity they relate, and my terms for them.
Relata
Compositional Relation
Processes
Individuals
Properties
Powers
224
C. Gillett
225
stituent individuals bear spatio-temporal, powerful, and/or productive relations to each other and hence form collectives;
(xiii) relations always involving comprising powers;
(xiv) relations always involving realizing properties;
and:
(xv) relations always involving implementing processes or the potential
for them.
This list is somewhat overwhelming and is likely not exhaustive. But it
should really be no surprise that the compositional notions used in some
of our most successful scientific explanations should be complex, sophisticated, and nuanced.
Given space limitations, we cannot assess how well the three kinds of
account we surveyed in Part 1 do with all of these features. And it might
also reasonably be argued that we should only be focused upon explanatorily salient characteristics. So I propose to focus on a handful of key
features from this list in features (i)(iii) and (vii)(ix), since the latter
are important general characteristics of scientific composition and also
the features that plausibly underlie the PEP and OUP of compositional
explanations themselves. Let me therefore briefly outline characteristics
(i)(iii) and (vii)(ix), and their links to PEP and OUP.
To begin, it is important to note a quite simple, but also foundational,
feature of the entities that are the relata of scientific notions of composition framed in (i). The relata of scientific relations of composition are
what I term working entities. That is, the various kinds of compositional relations in the sciences all relate entities that are individuated, at
least partially, by the processes with which they are associated and hence
by what I term roles. As we shall see shortly, this simple point has wideranging implications.
Given its centrality to the nature of compositional explanations, it is
important to mark another foundational feature of scientific composition: under the conditions, components naturally necessitate the composed entitythat is, the components suffice for the composed entity
in the relevant circumstances. This feature of compositional relations,
226
C. Gillett
227
228
C. Gillett
229
230
C. Gillett
that there are brute laws of nature that when there are changes in the
B-properties and B-processes of organs or organisms, then there are
changes in the properties and processes of the cellular or molecular individuals contained within the relevant organ or organism, and vice versa,
even though the B-properties and B-processes are not composed by these
cellular and molecular properties and processes.
Although rather elaborate, this kind of Vitalist hypothesis has the
virtue of highlighting important concerns about Cravers or other
manipulability-based accounts of scientific composition. Notice that
under the Vitalist hypothesis, cells and molecules plausibly are parts of
the organ and organism under a spatial containment account of parthood that Craver apparently favors. So Cravers first clause demanding
partwhole relations between the relevant individuals involved in the
processes is satisfied. And so too are Cravers other conditions demanding
mutual manipulability. Given the brute laws of nature at play, changing
the properties and processes of the spatially contained cells or molecules
results in changes in the B-properties and B-processes of the organ or
organism, and vice versa.
Given the latter points, Cravers manipulability-based sufficient condition for the composition of processes is satisfied in the scenario outlined by
the Vitalist hypothesis and counts the B-processes as composed, and implemented, by the processes of molecules or cells. The obvious problem is that
the Vitalist hypothesis frames a paradigm example where we lack any such
scientific composition of processes, so Cravers manipulability criterion
offers a mistaken account of the composition of processes in the sciences.
Simply offering counter-examples to an account is rarely fully satisfying or convincing, but our earlier work also illuminates the deeper
flaws of the manipulability account that lead to these mistaken attributions. As I highlighted earlier, scientific composition involves relata that
are in some sense the same, noted in (ii), where such relations are also
mass-energy neutral as (iii) frames. But a manipulability-based condition can be satisfied by entities that are wholly distinct and fails to guarantee that we have relations that are mass-energy neutral. This is why
Cravers manipulability-based condition implies there is a compositional
relation between the B-processes, and the processes of the cells or molecules, which are wholly distinct processesthus failing to accommodate
231
232
C. Gillett
A second lesson is that although manipulability, regularity, counterfactual dependence, sufficiency, or supervenience relations may underpin
methods for discovering, or illuminating, the existence of specific relations
of scientific composition, it is a mistake to take compositional relations
in the sciences to be relations of causation, manipulability, regularity,
counter-factual dependence, sufficiency, or supervenience, or to take
such relations to suffice for scientific composition. Scientific composition
is a singular, and ontologically richer, and more complex, relation than
the latter relations. Unsurprisingly, appropriating frameworks for these
different relations consequently, fails to provide an adequate account of
scientific composition or compositional explanation.
233
234
C. Gillett
I suspect that Grounding accounts also fail to cover feature (ii) and and do not entail entities that
are in some sense the same as their relata. However, I do not rely on this objection since I am unsure
about the commitments of Grounding accounts with regard to this kind of characteristic of their
relata.
235
236
C. Gillett
4.3 Why Standard functionalist Accounts Fail: RolePlaying, Qualitative Distinctness, andaLack ofPEP
In marked contrast to the previous two accounts, standard functionalist
treatments from the philosophy of mind appear to offer more promising treatments of features (i)(iii) because they have the great virtue of
being role-based. Let me briefly highlight how standard functionalist
accounts successfully accommodate the troublesome features of scientific
composition that trip up both neo-Causal and Grounding accounts.
Recall that the Flat/Subset view of the realization of properties is based
upon what I termed role-playing in a oneone relation between property
instances that are qualitatively the same, or similar, in either matching
or overlapping in their contribution of powers under some condition.16
Notice, first, that standard functionalist accounts take properties to be
individuated by their contributions of powers which are entities that
when triggered manifest in certain processes. So the Flat/Subset view
plausibly relates working entities and so satisfies (i). Second, the realizer
and realized properties instances under the Flat/Subset view are plausibly
entities that are in some sense the same, since the powers and processes of the one just are powers and processes of the other, so the view
accommodates (ii). Third, role-playing is also plausibly a mass-energy
neutral relation, covering (iii), because the processes that result from
the powers of the realized property instance just are, or are among, the
processes that result from the powers of the realizer property instance
hence the work of the realized instance just is the work of the realizer. Finally, and fourth, role-playing is a species of natural necessitation
16
If the arguments of Pereboom (2011, this volume) are correct, then the powers of realized and
realizer property instance are actually identical under the Flat/Subset view.
237
because, under the conditions, the role-playing suffices for the property/
relation instance whose role it plays, thus satisfying (ix).
Given the failures we have seen other approaches have had with features (i)(iii) one can consequently better appreciate the recent popularity of standard functionalist accounts from the philosophy of mind
in application to various topics in the philosophy of science involving scientific composition. And it bears emphasis that the Flat/Subset
account also unsurprisingly accommodates the OUP of compositional
explanation. Unfortunately, once we delve more carefully into the success of the Flat/Subset view with the other central features of scientific
composition, and compositional explanation, then we find a different
set of problems.
To begin, notice that role-playing is a oneone relation and hence fails
to accommodate the manyone character of scientific composition and
feature (viii). And this is linked to a still more troubling problem. By its
nature, role-playing cannot have qualitatively distinct relata and hence
accommodate feature (vii). Role-playing works through powers, processes, and roles that match exactly or are overlappingso role-playing
necessarily has relata that are qualitatively the same or similar and also
fails to cover feature (vii).17
Sometimes proponents of the Subset/Flat view seek to save their frameworks by challenging my descriptive account of cases of compositional
explanation and their compositional notions as involving manyone relations or even qualitatively distinct relata. I contend such alternative interpretations are descriptively defective and lead to all manner of problems
which I have documented in various ways elsewhere. However, I can most
swiftly highlight the deeper difficulties here by accepting for arguments
sake that scientific composition is neither a manyone relation nor one
with qualitatively distinct relata, for we immediately face a glaring problem.
The resulting concern is that we are left committed to a PEP-less
characterization of compositional explanation. If we take compositional
explanations to be based upon realization relations that do not relate
qualitatively distinct entities, then these explanations do not explain
17
If Perebooms arguments noted above are correct, then we have identity of powers, processes, and
roles.
238
C. Gillett
I count the framework, or parts of it, in Sydney Shoemakers (2007) if it is interpreted as involving manyone relations of role-filling. On the other hand, if Shoemaker (2007) is taken to focus
239
types of account in what I term joint role-filling relations, since the nature
and significance of this type of relation has not been properly appreciated
and because this suffices to highlight the promise of this Engaged approach.
This type of account is constructed by reference to the features of scientific composition relations we find in actual cases. Given the finding
that entities in the sciences are plausibly working entities, it assumes that
such entities are all individuated by roles that, in various ways, connect back to their associated processes. Slightly different kinds of role
individuate the different ontological categories of entity, in individuals,
properties, powers, and processes, given their differences from each other,
but I proceed below with a generic notion of role to articulate the general framework.
Given that the entities that are components and composed in the
sciences are all working entities, this type of account also assumes that
their componency is based around the work or role of component
entities sufficing for the individuative work or role of the composed
entity. Putting the general idea of working components in terms of roles,
it appears that working components are entities whose roles mean that
these entities suffice, under the circumstances, for the role that is individuative of the composed entity. However, as we have begun to see, we
now have reason to conclude that a role-based account using role-playing
provides an inadequate account of scientific composition.
However, focusing on the features we find in concrete cases in the sciences, and especially their manyone character, this alternative Engaged
account takes scientific composition to involve what I am terming joint
role-filling where scientific components are members of collectives, or
teams, of working entities spatially contained within the relevant composed entity and inter-related such that they jointly result in the role
of the composed entity by jointly filling it. So, for example, a team of
inter-related molecules, including actin, jointly fills the role of the neuron. Or various processes of polymerization, and other molecular processes, jointly fill the role of protruding in the cell.
Notice the contrasts between joint role-filling and the role-playing that
underlies the Flat/Subset account of realization and related accounts. In
role-playing, the component entity plays the very role of the composed,
on oneone relations, basically role-playing relations, then Shoemakers account taken as a treatment of scientific composition is flawed in the ways laid out above.
240
C. Gillett
and hence we have a oneone relation between entities that are qualitatively identical or similar. However, in joint role-filling, none of the
components itself plays the role of the composed entity, since in this
manyone relation the components together fill the role of the composed.
Consequently, the entities jointly filling the relevant role are qualitatively
distinct from the composed entity.
Although abstractly framed, I now want to detail why these differences
between role-playing and joint role-filling have important implications
for accommodating the features of scientific compositionand both the
Ontological Unifying Power and also the PEP of compositional explanations. So let us consider whether an account of scientific composition
built around joint role-filling accommodates the characteristics (i)(iii)
and (vii)(ix) we have targeted in our discussion here.
To start, the relata of a joint role-filling relation are working entities
that do have mass-energy. Satisfaction of (i) by the joint-filling accounts
is therefore built into them given their focus on working entities. And
joint role-filling is also a non-productive relation passing (ii), since it is
synchronous, has relata that are in some sense the same, does not involve
the transfer of energy and/or mediation of force, and is not identical to
the manifestation of powers. Furthermore, the work of the composed
entity, in its individuative role, just is the combined work of the rolefillers. And the mass-energy of the entity whose role is filled just is the
combined mass-energy of the role-fillers. Consequently, joint role-filling
is plausibly a mass-energy neutral relation and covers feature (iii) that so
troubled neo-Causal, and Grounding, frameworks.
We thus accommodate three important features of scientific composition by taking it to be a joint role-filling relation between working
components. But it should also be obvious that we can cover another
central feature scientific composition, in the manyone character framed
in (viii), since we saw that composition is a manyone relation and this is
central to joint role-filling which is also a manyone relation. The latter is
important because it allows the joint role-filling framework to accommodate qualitatively distinct relata and hence (vii). Joint role-fillers may all
each be qualitatively distinct in their roles from the entity whose role they
fill. That is, none of the role-fillers itself plays the role of the composed
entity. Instead, the role of the composed entity is only jointly filled by the
241
role-fillers where their roles together suffice for the qualitatively distinct
role of the composed entity. Consequently, two more central features
of scientific composition are accommodated under the joint role-filling
framework.
Next it is important to note that joint role-filling is a non-productive
relation of natural necessitation satisfying (ix). Under certain conditions,
we have inter-related entities that can serve as role-fillers and which hence
suffice for, and as we saw above non-productively result in, the role that is
the composed entity. Spatio-temporally arrayed teams of entities related
by powerful and/or productive relations are such that their relations mean
that these entities do, or would, jointly fill the productive role of the
composed entity. But this role is individuative of the composed entity.
Hence the existence of these lower-level entities, under the conditions at
this time, non-productively determines that we also have the composed
entity at this time, so joint role-filing accommodates (ix) because it is a
relation of natural necessitation.
Given the foregoing points, the framework of working entities bearing
joint role-filing relations plausibly accommodates the key features of scientific composition we have focused upon in this chapter. And elsewhere,
I have shown the framework covers the other features of scientific composition as well.19 We thus already have good reasons to think the Engaged
framework of joint role-filling does better than the Appropriational
accounts. At this point, let me therefore step back and outline why the
account also offers a promising account of compositional explanation.
Let me assume what I term ontic representationalism about compositional explanation in the view that some scientific explanations work by
representing ontological relations between entities in the world. Given
this assumption, first, I have already outlined how joint role-filling is
a relation of natural necessitation. So, when one has joint role-filling,
under the conditions, then this naturally necessitates that one has the
composed entity whose role is filled. When we have certain molecular
processes of polymerization, alongside other molecular processes, under
the conditions, then one must have a cell protruding. Thus, we see how
the explanans of a compositional explanation explains its explanadum
under the joint role-filing account. Second, the joint role-filling account
19
242
C. Gillett
243
244
C. Gillett
References
Aizawa, K. (2007). The biochemistry of memory consolidation: A model system
for the philosophy of mind. Synthese, 155, 6598.
Aizawa, K., & Gillett, C. (unpublished): The parts of Science: Scientific composition and compositional explanation.
Anderson, P. (1972). More is different: Broken symmetry and the nature of the
hierarchical structure of science. Science, 177, 39396.
Baumgartner, M. (2010). Interventionism and epiphenomenalism. Canadian
Journal of Philosophy, 40, 359384.
Bechtel, W., & Richardson, R. (1993). Discovering complexity. Princeton:
Princeton University Press.
Couch, M. (2011). Mechanisms and constitutive relevance. Synthese, 183,
375388.
Craver, C. (2007). Explaining the brain. Oxford: Oxford University Press, USA.
Dennett, D. (1969). Content and consciousness. London: Routledge Kegan Paul.
Dennett, D. (1978). Brainstorms. Montgomery, VT: Bradford Books.
Fine, K. (2001). The question of realism. Philosophers Imprint, 1(1), 130.
245
246
C. Gillett
247
9
Grounding andtheFormulation
ofPhysicalism
AndrewMelnyk
There are many sciences, and each science, to the extent that it gets things
right, uses its own characteristic theoretical vocabulary to describe a characteristic domain of entities. But how are the many sciences related to
one another? And how is the domain of entities proprietary to each science related to the domains of entities proprietary to the others? To try to
answer these questions is to address what I once called the problem of the
many sciences (Melnyk 1994, 222224; 2003, 12).
The problem of the many sciences looks like a promising candidate for
the sort of philosophical problem that naturalistic metaphysics should
addresswhere naturalistic metaphysics seeks to answer questions
that (i) ask what the world is like, albeit at a very high level of abstraction, that (ii) apparently dont fall within the province of the sciences
(as traditionally understood), but that (iii) creatures like us are capable in
A. Melnyk (
)
University of Missouri, Missouri, USA
The Author(s) 2016
K. Aizawa, C. Gillett (eds.), Scientific Composition
and Metaphysical Ground, DOI10.1057/978-1-137-56216-6_9
249
250
A. Melnyk
principle of answering (Melnyk 2013). Because I aspire to be a naturalistic metaphysician, I ask in this chapter whether an appeal to the relation
of grounding posited recently by certain philosophers might be useful
in one kind of approach to the problem of the many sciencesa physicalist approach. Jonathan Schaffer has explicitly proposed appealing to
grounding to formulate physicalism, albeit very briefly (Schaffer 2009,
364). Gideon Rosen has suggested formulating naturalism, a close relative of physicalism, by appeal to grounding (Rosen 2010, 111112). And
Shamik Dasgupta has recently tried to remove one obstacle to formulating physicalism by appeal to grounding (Dasgupta 2014).1 The prospects
of a grounding formulation of physicalism are also worth investigating
simply because of the remarkable level of current philosophical interest
in the putative relation of grounding.
The putative grounding relation that my question concerns is not
meant to be a generic relation under which such familiar relations as
supervenience, realization, and composition fall as species. Rather, it is
supposed to be a relation on a par with such relations; and it might be
posited either in addition to, or as a replacement for, such relations and
their kin (Wilson 2014, passim). It is also supposed, at least by three of its
leading proponents, to be a primitive relation (Schaffer 2009, 364; Rosen
2010, 113114; Fine 2012, 7879).2 Not all philosophers sympathetic
to grounding take it to be primitive. Dasgupta, for example, identifies
grounding with a certain sort of explanation: to say that some facts
ground another is just to say that the former explain the latter, in a particular sense of explain (Dasgupta 2014, 558). He therefore leaves open
the possibility that the particular sense of explain could be spelledout, yielding an account of what grounding is; indeed, he states what is
in effect a non-trivial sufficient condition for the grounding relation to
hold.3 In this chapter, however, I shall only consider a supposedly primitive grounding relation.
1
Daniel Stoljar devotes a subsection of his Stanford Encyclopedia of Philosophy entry on physicalism
to what he calls grounding physicalism, as if it were a standard approachwhich it isnt (Stoljar
2015, 10.3).
2
Rosen says we must accept it as primitive at least for now.
3
It is this: It is essential to ground that for any Xs and any Y, if the Xs obtain and if a fact about
the essence of a constituent of Y implies that the Xs are materially sufficient for Y, then the Xs
251
252
A. Melnyk
253
intended to demonstrate this failure of entailment; it invites us to consider a series of three cases.6
Suppose, first, that a state-token x of one state-type nomically necessitates a later state-token y of an entirely different state-type. For vividness,
think of x as a neural state and y as a pain state:
(1) x at t1 nomically necessitates y at t2.
Suppose also that this nomic necessitation is brute, not in the sense
that it has no explanation at all (for it may have a theistic explanation in
terms of a divine will), but in the sense that it has no explanation in terms
of more basic nomic generalizations: it has no same-level explanation in
terms of states of other types that intervene between x and y; and it has
no lower-level explanation in terms of underlying states that constitute x
and y. Clearly, the brute nomic necessitation of y by x does not entail that
y is nothing over and above x.
Now consider a second case exactly similar to the first except that the
necessitating state x and the necessitated state y are now simultaneous,
so that the brute nomic necessitation of y by x is synchronic rather than
diachronic:
(2) x at t1 nomically necessitates y at t1.
Surely, the brute nomic necessitation of y by x still doesnt entail that y
is nothing over and above x. For it didnt entail this in the first case, and
the second case differs from the first only in the changed relation between
the time of x and the time of y. It is very implausible to think that we
could move the time of x arbitrarily close to the time of y while y continues to be something over and above x, but that the moment we make the
times identical, y becomes nothing over and above x. To think that would
be to attribute magical powers to time.
Consider, finally, a third case which is exactly the same as the second, except that now the brute necessitation is not nomic but rather
metaphysical:
(3) x at t1 metaphysically necessitates y at t1.
For earlier arguments with the same goal, see (Melnyk 2003, 5770; Wilson 2005).
254
A. Melnyk
255
Hence I think Wilson concedes too much when she writes that A Grounding claimeffectively
stipulates nothing over and above-ness (Wilson 2014).
256
A. Melnyk
257
explanation of the observed regularity than, say, the rival hypothesis that
the Xs are lawfully sufficient for the Ys, because it is more parsimonious
than this rival? No. The hypothesis that the Xs ground the Ys would
be more parsimonious than the rival only if it entailed that the Ys were
nothing over and above the Xs. But, I have argued in this section, there is
no reason to think that it does, and one reason to think that it does not.
To a first approximation only, because the formulation leaves various questions unanswered.
Should the entities quantified over include abstracta? Or necessary existents? To what categories
should the entities belongstates, events, properties, objects, facts, truths? See Melnyk (2003,
611; 2032).
258
A. Melnyk
259
260
A. Melnyk
In a recent paper, Dasgupta appears to raise exactly the same problem for a grounding formulation
for physicalism (Dasgupta 2014, 561562). But a careful reading reveals that in fact he doesnt. The
appearance arises because he raises his problem by presenting an argument that Xs grounding Y
has no purely physical ground (Dasgupta 2014, 571). But the reason Xs grounding Y has no
purely physical ground, for Dasgupta, is that it is partly grounded in some ungrounded connection between Xs and Ys (Dasgupta 2014, 569); the reason is not that the grounding relation itself
is problematically non-physicala possibility, indeed, that the paper nowhere mentions. And
Dasguptas solution to his problem is to argue that physicalism can allow the sort of connection
between Xs and Ys that he has in mind to have no ground at all, and hence no physical ground
(Dasgupta 2014, 575).
261
10
The hence is justified because in this paragraph I am assuming that every instance of grounding
would be made consistent with physicalism in the same way, by being grounded in something narrowly physical.
11
Dasgupta argues that an infinite series of grounded grounding facts is harmless (Dasgupta 2014,
587589). But he is not talking about an infinite series of physically grounded grounding facts, as I
am; and in any case an infinite series of grounded grounding facts is harmless on his account only
if grounding is not primitive.
262
A. Melnyk
12
Committed to the grounding relation cannot just mean the same as logically entails that the
grounding relation has instances. In the sense I intend, a complete physical description of the
world is committed to grounding iff (i) the complete physical description is possibly true, (ii) the
claim that there exist instances of grounding is possibly false, and (iii) it is logically necessary that,
if the complete physical description is true, then there exist instances of grounding. Conditions (i)
and (ii) serve to rule out degenerate cases of entailment, in which the complete physical description
is necessarily false or the conclusion necessarily true.
13
For example, physicalism is formulated too strongly if it is formulated as saying that all facts or
truths hold in virtue of physical facts or truths. This formulation is too strong because, if physicalism is true, then its a fact that physicalism is true; but the fact that physicalism is truethe fact
that nothing exists that is neither narrowly nor broadly physicaldoesnt hold in virtue of physical
facts alone. See Melnyk (2003, 2526; 97 n.17).
263
14
15
264
A. Melnyk
(Kripke 1980, 109). Since this is so, realization physicalism does not have
to require that instances of identity be narrowly physical or realized by
something narrowly physical; and so instances of identity are consistent
with physicalism even if they are neither narrowly physical nor realized by
something narrowly physical.
At this point, enthusiasts for a grounding formulation of physicalism
might ask whether what I have just said about identity is true also of
grounding, so that grounding too is consistent with physicalism (contrary to my earlier contention). The answer to this question is affirmative,
of course, only if its true that, if X actually grounds Y, then theres no
possible world in which X and Y exist but X does not ground Y.But I see
no reason to think that this is true. The two standard arguments for the
necessity of identity, which appeal, respectively, to the necessity of selfidentity and to the claim that non-descriptive referring expressions are
rigid designators (Kripke 1980, 104), look most unlikely to carry over to
support an analogous thesis of the necessity of grounding. Also, as noted
in the section Grounding and the Broadly Physical, the grounding relation, because it is primitive, doesnt hold between two items in virtue of
any other facts; a fortiori, it doesnt hold in virtue of other facts about
the relata; a fortiori again, it doesnt hold solely in virtue of other facts
about the relata. So we cannot reason that, just because in some world
w the relata exist, in w the grounding relation must hold between them.
Finally, even if it is granted that grounding is a species of metaphysical
necessitationin a sense which implies that, necessarily, if X grounds
Y, then X metaphysically necessitates Yit doesnt follow that in every
world in which X and Y exist X grounds Y.It does indeed follow that in
every such world X metaphysically necessitates Y, given the transitivity of
inter-world accessibility; but there is (we are assured) more to grounding
than metaphysical necessitation.16
16
Might what I said about identity be true also of realization? Not if the first relatum is taken to be
physical state-token p, or p plus physical conditions C.For the actual worlds laws of physics dont
hold in all possible worlds in which p, or p plus C, exists. But if the first relatum is taken to be p
plus C and the holding of the actual worlds laws of physics, then perhaps yes. I dont know which
view of the first relatum is correct.
265
What about the entities said by claims (i) and (iv) to be (self-)identical?
Claim (iv) speaks of mental state-token m, which is by hypothesis physically realized and therefore consistent with physicalism. Claim (i) speaks
of the mental state-type M. But because (one might reasonably suppose)
there are no untokened types, the existence of state-type M just is the
existence of its state-tokens. So for M to be consistent with physicalism,
it is enough if each of its tokens is physically realizedwhich they are
if realization physicalism is true. In short, realization physicalism says
that every contingent entity-token is either narrowly physical or realized
by something narrowly physical, and tokens of realization are consistent
with physicalism because they are in part narrowly physical and in part
realized by something narrowly physical.
My argument that the existence of instances of realization is consistent with physicalism relies on the tacit premise that the existence of X
is consistent with physicalism if (i) the existence of X just is (i.e., =) the
existence of Y1, Y2, Y3,Yn, and (ii) each of the Yi is narrowly physical
or realized by something narrowly physical. It might therefore seem as if
I have modified the formulation of realization physicalism by introducing a new way in whicha new relation in virtue of whichan instance
of a property (or relation) that isnt, or isnt wholly, narrowly physical can be rendered consistent with physicalismwhich would then
immediately prompt the same question that we asked about instances of
realization, mutatis mutandis, namely, the question of how instances of
this new relation can be rendered consistent with physicalism. But this
appearance is illusory. To be sure, one could express realization physicalism as follows:
Every contingent instance of a property (or relation) is
either narrowly physical
or realized by something narrowly physical
or is one and the same as the existence of instances I1, I2, I3,In (of
properties or relations P1, P2, P3,Pn, respectively), every one of which is
either narrowly physical or realized by something narrowly physical.17
17
This rough formulation is good enough, I hope, for the present purpose.
266
A. Melnyk
When the R-relation is realization, I have tried to divide and conquer, taking it to be partly among the Ps and partly among the Ms. To
the extent that it is among the Ms, I have acceptedof coursethat it
depends on and is determined by the Ps, and have said specifically that
it is realized by the Ps. Polger thinks that a regress must now ensue, but
18
267
does not say why. He may think a regress inevitable because we need a
second R-relation to make the first R-relation consistent with physicalism,
and a third to make the second consistent with physicalism, and so forth.
And in response to the contrary suggestion that further R-relations are
not needed because the first R-relation could make itself consistent with
physicalism, he may mean to object that, even so, there would still ensue
a regress of tokens of the first R-relation.
What should we make of Polgers charge of regress? Here, for convenience, are the four crucial claims from above that together define realization in my sense:
(i) m is a token of a mental state-type M with a certain higher-order
essence: for a token of M to exist just is for there to exist a token of
some (lower-order) state-type such that tokens of that (lower-order)
state-type play role RM, the role distinctive of M;
(ii) p is a token of a physical state-type P such that, necessarily, given the
physical laws and physical circumstances C, tokens of P play role
RM; and
(iii) the laws of physics hold and physical circumstances C obtain.
(iv) the token of mental state-type M whose existence is entailed by
claims (i) through (iii) = m.
I have claimed two things: (1) the holding of the realization relation
between physical state-token p and mental state-token m just is the
holding of the four conditions described by claims (i) through (iv); and
(2) everything required for claims (i) through (iv) to be true is physical or physically realized, hence consistent with physicalism (when formulated by appeal to realization). But in so claiming, have I appealed
to a second R-relation? If I have, then it is with (1), and the second
R-relation is identity; but I have already argued, in two ways, that
identity is consistent with physicalismand neither way appeals to a
third R-relation. In fact, however, I need not be construed as having
appealed, in (1), to identity as a second R-relation. For, as noted two
paragraphs ago, I am at liberty simply to cease speaking of realization as
such, while retaining the substance of my realization physicalismto
replace p realizes m with the claim that the four conditions described
by claims (i) through (iv) hold.
268
A. Melnyk
References
Dasgupta, S. (2014). The possibility of physicalism. The Journal of Philosophy,
111(9), 557592.
Fine, K. (2012). Guide to ground. In F. Correia & B. Schnieder (Eds.),
Metaphysical grounding: Understanding the structure of reality (pp. 3780).
Cambridge: Cambridge University Press.
Kripke, S. (1980). Naming and necessity (p.1980). Cambridge, MA: Harvard
University Press.
19
For helpful comments on an earlier draft I thank Ken Aizawa, Carl Gillett, and the other participants in the Composition and Ground Workshop at Rutgers University, Newark, April 1011,
2015.
269
10
Grounding inthePhilosophy
ofMind: ADefense
AlyssaNey
Introduction
One of the major trends in metaphysics in recent years has been in the
development and application of novel conceptual frameworks for representing facts about realism, fundamentality, and metaphysical priority.
Of particular interest have been the concepts of grounding (proposed by
Paul Audi (2012), Kit Fine (2001), Gideon Rosen (2010), and Jonathan
Schaffer (2009), among others)1, the concept of the real (proposed by
Fine (2001)), and that of metaphysical structure (proposed by Ted Sider
(2011)). All of these have been proposed as new primitive concepts, and
Witmer et al. (2005) defend a related in virtue of notion. Bennett (2011) speaks of
building.
A. Ney ()
Department of Philosophy, University of California, Davis, CA, USA
271
272
A. Ney
10
273
This is a principle frequently appealed to by Jaegwon Kim and others in the metaphysics of mind.
It says that for something to be real it must possess causal powers.
4
In other work, I apply these resources to debates in the philosophy of causation and mental
causation.
274
A. Ney
Grounding andAnti-Realism
In order to see what may be added by an appeal to grounding, lets start
by noting something about the metaphysical relations that are typically
discussed in the philosophy of mind such as type- or token-identity, constitution, and property- or event-realization.5 We are apt to talk about the
obtaining of one of these relations when we have some entities (or types
or ways of conceiving some entities or types) that are already assumed to
exist. Our interest isnt in whether or not these entities or types exist, but
rather in characterizing the metaphysical relationship of one to the other.
5
Of course, there exists a diverse variety of ways of understanding the constitution and realization
relations. The differences among them will not matter for what follows. Note that I will not discuss
supervenience and necessitation as these notions have been widely recognized for years in the philosophy of mind to be insufficient to characterize the sense in which mental phenomena may be
ontologically dependent on physical phenomena. See Kim (1984) and Wilson (2005).
10
275
Here is one of the main examples Fines paper takes up. Suppose some
philosopher is a nominalist about abstract entities. Then it looks like she
will hold, as one of her main philosophical claims:
(1) Numbers do not exist.
And yet, even if she is a nominalist, it seems clear that she should not
want to deny something all of us accept, the simple mathematical fact
expressed by:
Here I am using dependence to indicate a relation such that when it obtains, it need not imply
that one of the relata is more fundamental than the other. (Ontological dependence is not an asymmetric relation.) When I speak of the obtaining of an ontological priority relation, what is prior is
thereby implied to be more fundamental than what it is prior to.
276
A. Ney
Fines official view is that this operator (and the grounding operator to be described shortly)
should be taken to apply to sentences (2012, p. 46), but he sometimes speaks of propositions
grounding other propositions. I will sometimes speak loosely as well of facts grounding other facts.
This should be understood as indicating the grounding of a sentence describing one fact in some
sentences describing some other facts.
10
277
the sort of claim the nominalist intends. Recognizing this, we can say that
really there is no tension between the nominalists main claim and (2) (or
even (3)). For what the nominalist intends to assert is not (1) that numbers do not exist, but rather that in reality, numbers do not exist.8 And as
long as we do not conflate this reality locution with quantificational idioms like there exists, we can see that there is no contradiction between:
(4) In reality, numbers do not exist, and:
(3) There are numbers. (Numbers exist.)
Thus, the tension introduced by non-skeptical anti-realism is resolved.
To this, some have objected that they dont have a grip on what Fine
means by in reality.9 As noted, the reality operator is officially introduced as a primitive, but this shouldnt lead one to worry we have no grip
on what it adds to a sentence. Fine gives a positive characterization of it
as follows:
One might think of the world and of the propositions by which the world
is described as each having its own intrinsic structure; and a proposition
will then describe how things are in themselves when its structure corresponds to the structure of the world. Thus it is this positive idea of the
intrinsic structure of reality that should be taken to inform the relevant
conception of what is fundamental or real. (2001, p.25)
The real propositions are those that describe the intrinsic structure of
reality. This isnt to say that propositions that do not describe the intrinsic structure of reality may not be true. They of course may be. But it is
the goal of the metaphysician (at least some of the time) to make claims
that are not just true but also real, that do describe the intrinsic structure
of reality.
Some will complain that this positive characterization of the reality
operator doesnt address the concern since it relies on a further esoteric
notion, the intrinsic structure of reality. Since it is difficult or impossible
278
A. Ney
10
279
reality, there are such things as messes and one of them is in the kitchen.
Lets pause a moment to see a third account according to which (5) may
express an objective truth and yet still not be real.11
On a functionalist understanding of mess, a claim like (5) is true
when there are some things or other in the kitchen capable of playing
a certain causal role, whatever is the causal role associated with our concept, mess. Ill tentatively work with: being a collection of things that
persists in a location without good reason that is apt in the circumstances
to cause obstruction and annoyance. Suppose in our imagined situation
what plays that role is a pile of dirty dishes in the sink. Functionalists may
take different approaches when they consider (5)s connection to reality
in this situation. One approach would be to adopt a functionalist realism
about messes. Then one will say that in reality there are messes, one such
mess is the pile of dishes in the sink, and since the sink is in the kitchen,
the sentence expresses a truth about what there is in reality.12 However,
one might be concerned about this approach for several reasons. One
is that the concept of mess permits multiple realization and so it seems
wrong to think that messes just are piles of dishes in sinks. Ill however
focus on another reason for rejecting this functionalist realism about
messes. This is that a pile of dishes in a sink is only properly counted
as a mess in a particular context, namely one in which the dishes are
there without good reason and apt to cause annoyance and obstruction.
In other circumstances, such as when one places a pile of dishes in a sink
in order to promptly clean them, a pile of dishes doesnt count as a mess
but a means to an end. So it isnt right to think of the pile of dishes itself
as the mess. Piles of dishes arent the right kinds of things on their own
to be messes. Nor would it be correct to think of the larger mereological
sum consisting of the pile of dishes, the sink, and all of the things apt to
be annoyed orobstructed in the situation as the mess. Thats not a mess
11
In outlining this third way for how it may be that a sentence is true, yet not true in reality, I am
departing from Fines official view. Fines descriptions of cases involving sentences that are not true
in reality generally involve subjectivity such as we find in the first two accounts above. However, as
I will now argue, this third way also constitutes a way in which a sentence may be true while not
correctly describing the intrinsic structure of a given situation.
12
This would be to endorse what is usually called an occupant or realizer functionalism about
messes. Messes are the things that occupy the mess-role.
280
A. Ney
either. (Anyway, if that were the right account, it would make (5) false,
since that object is too big to be in the kitchen.) Better, one sympathetic
to a functionalist approach to (5) should deny that (5)s truth depends
on the existence of any one kind of thing, a mess, but rather depends
on a particular kind of situation being instantiated.13 There being a mess
amounts to a situation that may involve not only dishes but also a variety
of other kinds of objects, but only in the larger circumstance in which the
objects play the causal role associated with the concept of mess. This is
an account in which the truth of the sentence (5) is objective in the sense
that its truth does not depend in any way on someones perspective. It
depends merely on what kinds of things there are in the world and how
they are arranged. But the sentence is still not true in reality because interpreted as a claim about what there is in reality, it would make a false claim
that in reality there are such things as messes that are located in kitchens.
According to this anti-realist functionalist account, sentences like (5) latch
onto the world in a more complicated way, by referring to a more spatially
extended situation, a causal network. Is there a mess in the kitchen? Yes,
but not because in reality there are messes and one of them is located in
the kitchen. Rather, there is a mess in the kitchen because there are in
reality many kinds of things, dishes and sinks and people, interacting in
the right way to make this sentence true.
So there are many ways it could turn out that when someone asserts
(5), they are not making a claim accurately tracking the kinds of things
there are in reality. This isnt to say that one is thereby speaking figuratively or not expressing a fact or saying something that isnt true, interesting, or justified. Not all assertions, not even all true, justified, and
interesting assertions need to mark out the kinds of things there are in
reality in a way that would interest a metaphysician. I hope Ive indicated
some ways this could work out for the everyday case in which one says
someone has made a mess.
We can now see where the notion of grounding enters in Fines framework. It is precisely here, to show how those sentences that are not tracking
the intrinsic structure of reality may yet be true. Formally, grounding is a
13
10
281
To say that some true sentences may not themselves be real, yet nonetheless be grounded in the real is to say that while these truths do not
themselves track metaphysical structure, they have an explanatory basis
in truths that do. The details of this basis might itself be complicated
(involving facts about individual perspectives or, as Ive argued, causal
networks). But ultimately a true statement will have a set of facts that
explains its connection to reality. Once these are provided, there is no
longer any explanatory gap left over regarding why the grounded sentence is true, or why the fact it describes obtains.
In addition to the primitive notions of reality and ground, Fine also
introduces a third notion that is defined in terms of the notions of the
real and ground. This will be useful in what follows. It is the notion of
reduction:
The true proposition P reduces to the propositions Q, R, iff (i) P is not
real; (ii) P is grounded in Q, R, ; and (iii) each of Q, R, is either real
or grounded in what is real. (2001, p.26)
282
A. Ney
real, the question is rather what relation they bear to physical phenomena. For example, in a situation in which Tom is in pain, the reductionist,
nonreductive physicalist, and emergentist will typically all agree the mental state is real. Their debate concerns rather whether Toms being in pain
is identical to a physical state, is realized by, but not identical to a physical
state, or is instead caused by rather than constituted by a physical state
of Toms. In a debate that has this form, we may note that Fines notion
of reduction will not be particularly useful since it doesnt capture any of
these three options. As one might already guess, I do not agree that the
debate should be understood in this way, as limited to a choice between
these three options. One of the main points I want to make is that the
grounding framework gives us a way of framing views on the mindbody
problem that are (in at least some domains) more reasonable than those
that have previously been articulated. We will come back to this when we
examine the case of phenomenal and other psychological states in more
detail momentarily.
As a final exegetical point, note that this framework and all that has
been said up until now leaves open the possibility that a sentence may be
true, grounded in other sentences, and yet itself be real. This would be
a case of grounding without reduction, where what is grounded reflects
metaphysical structure as well as its grounds. The distinction between the
case of grounding with reduction and grounding without reduction will
play a role in the applications below.
10
283
Suppose again we are discussing the status of the fact that Tom is in
pain. In Audis framework, if this fact is grounded in some fact about
physical states (say that Toms C-fibers are firing), this is to say that there
is a kind of noncausal determination relation obtaining between these
facts, one that arises due to an essential connection between the properties that constitute these facts. In this way of thinking about grounding, grounding isnt even compatible with reduction in Fines sense. Pain
has to be real for there to be an essential connection between it and the
physical property figuring in the grounding claim. In Schaffers framework as well, we find that where a grounding relation obtains, there is
no presumption that the grounded is not real. Instead for Schaffer, quite
the opposite, anything that is grounded must be real. Grounding is a
relation that obtains, not like for Audi between facts or for Fine between
sentences, but instead between entities of any ontological category. So,
Schaffer might speak of Toms pain being grounded in some physical feature of Toms brain or body. To say for some entity that it is grounded is
just to say that it has the status of a derivative entity (2009, p.373), which
entails that it is an entity and hence real. I propose that it is an advantage
of the Finean framework that it does not have this consequence of Audi
and Schaffers frameworks that what is grounded is automatically real or
an element of ones ontology.
But why? Why should it be an advantage of the account rather than a
cost that it permits this flexibility? Wilson (2014, pp.244248) has argued
this is a liability for grounding approachesthere is what she calls the
metaphysical underdetermination problem for theories of grounding. This
is that knowing a grounding fact obtains leaves completely open all of the
interesting questions we care about when we raise questions of existence,
ontological dependence, priority, and fundamentality. This is a problem
because if the argument for introducing a new grounding primitive was
that the logical, mereological, and modal relations metaphysicians were
previously using were inadequate to capture the metaphysical relations
between (say) mental phenomena and physical phenomena or mathematical phenomena and observable phenomena, then an appeal to grounding
doesnt offer an improvement in this respect and so is unmotivated.
This is an important worry. To respond, we may start by noting that
nothing in the framework I have discussed here suggests that a bare
284
A. Ney
appeal to grounding can answer the question of the precise nature of the
metaphysical relation between (for example) mental and physical phenomena by itself, nor even questions about their existence. When we just
say a truth is grounded, we do not say what its grounds are. But although
simply saying a truth is grounded wont answer all of the metaphysical
questions that interest us (and so close the explanatory gaps a grounding
claim is supposed to close, according to Fine), stating what those grounds
are will. To see this return to:
(7) Tom is in pain.
Here are four proposals for the grounds of (7):
(8) Toms C-fibers are firing.
(9) Toms C-fibers are firing. The firing of ones C-fibers is typically
caused by tissue damage and typically causes withdrawal behavior. One
is in pain if one is in the kind of state that in the relevant circumstances
is typically caused by tissue damage and typically causes withdrawal
behavior.
(10) Tom is in an internal state of the kind that in the circumstances
is typically caused by tissue damage and typically causes withdrawal
behavior. One is in pain if one is in the kind of state that in the relevant
circumstances is typically caused by tissue damage and typically causes
withdrawal behavior.
(11) Tom believes he is in pain.
We may also consider a fifth possibility that while (7) is true and real,
the event it describes is caused by the event described by (8), but (7) is
not grounded in anything.14
What we see here are different candidate grounds for (7) (or the denial
of a ground altogether) that correspond to different ways of answering the question of the relation between pain facts and physical facts.
These correspond to five canonical views in the metaphysics of mind:
(brute, i.e., nonfunctionalist) type identity theory, occupant functionalism, causal role functionalism, subjectivism, and emergentist dualism. This of course by no means exhausts the range of available options.
14
A complication arises here in that (7) refers to Tom, a human being, and that we should not think
of facts about human beings as generally ungrounded. Lets postpone this issue and just ask the
question of whether (7) is grounded relative to its ascription of pain.
10
285
But it suffices to demonstrate how even though merely saying that something is grounded does nothing to eliminate metaphysical underdetermination, saying in what it is grounded will.
The question that next naturally arises and is indeed pressed by Wilson
is then why the concept of grounding is needed when we already have at
our disposal these concepts of identity, realization, causation, as well as
mereological notions.15 Cant we accommodate all of the options mentioned above without also using the notion of ground? We may first
remark that what we have in effect shown is how (using the grounding
framework), we may bypass any explicit mention of identity, realization,
etc. while still seeing a diversity of metaphysical options via the variety of
grounds possible for (7). But the more important point is again to insist
on being careful about which grounding framework we are considering.
For some frameworks (e.g. Schaffers), Wilsons concern would be justified. We may dispense with the grounding notion in favor of identity,
realization, and the like (assuming we also have available a way of saying
which entities are real or fundamental). Grounding is just a less specific
way of describing the ontological dependence of some entities on some
other entities. But Fine is interested as we have seen in accommodating a
form of non-skeptical anti-realism. In the Tom case, this would amount
to a view according to which (7) is true, but not real. The view is antirealist in the sense that its proponent is denying the reality of mental states
without claiming that sentences like (7) are false. A view like that cannot
be accommodated using the frameworks of identity, realization, or emergence. The view rejects the existence of identity, realization, causal, and
mereological relations between pain and physical phenomena, because it
rejects the reality of mental phenomena. Yet, it is not the eliminativism
of Paul or Patricia Churchland (1981, 1986) either. According to their
eliminativism, it is not just that mental states are not real, but that all psychological claims that would appear to be about them are false. Using the
grounding framework, we can say that many psychological statements are
true even though they are not real, so long as they are grounded in what is
real. This is accommodated because grounding is not a relation between
15
There is a question about whether we need the concept of real as well. Wilson allows that we need
at least something like this, a concept of fundamentality.
286
A. Ney
10
287
And recall Putnam (1967) on pain: Consider what the brain-state theorist has to do to make
good his claims. He has to specify a physical-chemical state such that any organism (not just a
mammal) is in pain if and only if (a) it possesses a brain of a suitable physical chemical structure;
and (b) its brain is in that physical-chemical state.
288
A. Ney
10
289
290
A. Ney
may be intended to track the intrinsic structure of reality (as Ill discuss
below), the point of emphasizing the autonomy of the special sciences
from fundamental physics in Fodors work seems precisely to emphasize
that many psychological claims are not intended as claims of fundamental metaphysics, about what kinds of things there are, but rather intend to
track causal patterns that may be instantiated by a broad variety of things.
Fines framework allows us to say that even if this is so, the special science
claims in which Fodor is interested may be true. And this seems to be precisely what Fodor most wants. Greshams law, good money drives out bad
money, can be true. But for this to be so, the world need not be carved
up into little things that are money.17 As in the third account of mess talk
above, the truth of such claims may be explained by facts that do track
reality, but not what is suggested by the grammar of the financial claim.
If functionalism is correct, we may see monetary truths as grounded not
in facts about a particular ontic kind that is instantiated in wallets and
banks, but rather truths about a complex web of causal facts. Then we
should give up the claim that money is a kind of thing altogether. But
again, this doesnt undermine the truth of the claim, just its success at
reflecting metaphysical structure.
Seeing things this way also gives Kim most of what he wants. To say
that a special science claim is true and justified does not require saying its
predicates refer to real kinds. So it does not require we posit the existence
of additional higher level kinds any more than the truth of There is a
number between 2 and 5 (in any normal mathematical context) requires
the existence of numbers. As such, there is no threat of overdetermination
or epiphenomena. Special science claims, when true, will be grounded in
real claims (seeing how is an important first-order project in the metaphysics of science in which the various conceptual tools of realization and
constitution may be brought to bear), but to be so grounded does not
require the existence of potentially overdetermining higher-order kinds.
17
It is clear that many who have followed Fodor in adopting nonreductive physicalism want more,
want to say that many special sciences claims arent just true and justified but also that they refer to
additional higher-level special science kinds. However, there does not seem to be any justification
for this further ontological claim and there are reasons (those noted by Kim) against it.
10
291
18
292
A. Ney
19
Here, I am considering proposals for the full, as opposed to partial, grounds for (7). See the distinction in Fine (2012).
20
This represents a position in the ballpark of what is proposed in Dennett (1991). The fact that
this view shows how Dennetts position is able to accommodate true phenomenal claims is also a
virtue of the account, but one I do not have the space to explore here.
10
293
We may allow that (9) and (10) are both real and also that either may
in principle explain the truth of (7). But there is a question of which of
(9) or (10) gives the best explanation of (7) as it is asserted in a given
context. Which is the correct grounding explanation for (7) has ramifications for whether one should take a realist or anti-realist position about
pain. (9) does, while (10) does not, explain the truth of (7) in terms of
the existence of a particular kind of state that is a pain state.21 (10) reveals
pain talk to be metaphysically grounded not in the instantiation of a
particular kind of state, but rather a broad causal nexus. Those who take
(10) to be the correct view about what grounds (7) will thus (on the picture I have sketched here) adopt a non-skeptical anti-realism about pain.
This is what I have argued is the reasonable position if one accepts with
Fodor (and Putnam and the very many other nonreductive physicalists)
that psychological statements using the concept pain do not refer to a
homogenous physical kind, but accepts the metaphysical points of Kim
(and Lewis and other reductionists).
A Comment onReality
As a sidebar, it is worth acknowledging a bit of awkwardness in Fines
terminological framework as applied to the mindbody debate. In the
framework I am proposing, only psychological statements that track univocal physical kinds make claims about reality. Those that do not may be
true, justified, and important, but not real. Some have asked whether we
really want to say that in all cases in which functionalism is motivated,
that is, all attitude ascriptions or (if Fodor is right about pain) pain
ascriptions, these claims are not real. In every sense of real that matters to us for most but the most esoteric purposes, one can see how this
is really not very helpful. The use of not real here is too easily confused
with a way of rejecting a statement. After all, does a metaphysician really
want to offer the diagnosis that someones pain is not real?
21
294
A. Ney
10
295
Other Frameworks
After all of this, one might still be wondering why we need to use this new
framework of grounding to make the distinctions Ive wanted to make
in this chapter, in particular in order to introduce the conciliatory position I discussed above. Several have asked why we cant simply say that
we should analyze psychological statements in terms of statements about
causal networks and leave it at that. Talk of analysis is something metaphysicians (and philosophers of mind) have long been comfortable with
and doesnt require introducing new terminology like real and ground.
This is prima facie a reasonable point, but there are many problems
with this approach that metaphysicians are by now well familiar with.
First, it is not possible in many cases to provide the relevant analyses.
But more generally, even if we had them, analyses dont tell us in virtue
of what in the world a claim is true, what our ontological commitments
ought to be if we accept that claim, only in what circumstances it is true.
This point was articulated years ago by William Alston in a paper critical of Quinean approaches to ontological commitment (Alston 1958).
Suppose a nominalist about universals wishes to allow that sentences like
Patience is a virtue may be true while denying the existence of universals. She may then analyze Patience is a virtue in terms of some sentence
that doesnt quantify over universals, something like Patient people are
virtuous people. The trouble Alston noted is that the result of the analysis is simply the claim that these two sentences mean the same thing. But
agreeing the sentences are semantically equivalent doesnt entail anything
about the reality or unreality of universals. Rather we then see the sentences assert the existence of patience as much as they assert the existence
of patient people. What Alston argued, and what Fine and Rosen and I
are pressing, is that if one wants to say that Xs are not real, but Ys are,
and that the X-truths that are grounded in Y-truths, then one should just
come out and say this and use this language, since such metaphysical
claims cannot be replaced by talk of analyses.
One interesting view in many ways similar to what I am proposing
here but formulated without a grounding framework was developed in
2007 by Carl Gillett. This view, which he calls compositional reduction-
296
A. Ney
ism, also aims to reconcile the different insights of reductionism and nonreductive physicalism. The central idea of compositional reductionism is
that for reasons of ontological parsimony we should reject the existence
of psychological kinds that are not identical to causally univocal physical
kinds, but we should also allow that the claims made by psychologists are
not intended to track physically univocal kinds. Gilletts compositional
reductionist accommodates these points by saying that a sentence like my
Tom is in pain, may have associated with it two sets of truth conditions:
Tom is in pain is true iff Tom instantiates a particular physical type (say,
his C-fibers are firing).
Tom is in pain is true iff Tom instantiates the higher-order property of
instantiating a physical type that in the circumstances plays the pain-role.
10
297
298
A. Ney
that might allow one to state the kind of non-skeptical anti-realist view
I defend here. I havent argued against Siders framework in this chapter either. My goal rather has been to argue that a desirable approach
to solving the mindbody problem should be able to capture situations
in which a sentence is true, its truth is grounded or made true by facts
about the world, and yet it misleads on matters of ontology. If one prefers to adopt Siders framework or perhaps the truthmaking framework
to accommodate this, fine. One is still thereby acknowledging that one
must move beyond the tools for presenting metaphysical positions that
philosophers of mind have traditionally allowed themselves. And this is
what needs to be recognized.
Conclusion
I hope to have shown here how at least one grounding framework may be
useful in the philosophy of mind, providing us especially with a range of
anti-realist views that do not reject the truth, factuality, importance, or
justification we have for claims in psychology. Psychological claims may
possess all of these honorifics without undermining the search for a unified, sparse, and nonredundant underlying metaphysics.24
References
Alston, W. (1958). Ontological commitments. Philosophical Studies, 9, 817.
Audi, P. (2012). Grounding: Toward a theory of the in-virtue-of relation. Journal
of Philosophy, 109, 685711.
Bechtel, W., & Mundale, J. (1999). Multiple realizability revisited: Linking
cognitive and neural states. Philosophy of Science, 66, 175207.
Bennett, K. (2011). Construction Area (No Hard Hat Required). Philosophical
Studies, 154, 79104.
Carnap, Rudolf. (1950). Empiricism, semantics, and ontology. In Meaning and
necessity. Chicago: University of Chicago Press.
24
I thank Ken Aizawa, Louise Antony, Jamin Asay, Carrie Figdor, Kit Fine, Carl Gillett, Jens
Harbecke, Kerry McKenzie, Kelly Trogdon, and especially Jessica Wilson for comments and criticism that led to substantial improvements of this chapter.
10
299
300
A. Ney
Index
A
absolute fundamentality, 143, 157,
158, 195, 196, 196n17
activation experiments, 612
activities account of causation, 503,
81, 97
actualist-mechanist theory, 51
agnosticism, 174
Alexanders Dictum, 273
analytic metaphysics, 2, 3, 3n5, 10,
1718, 206n2, 209, 21113,
228, 235
angular head velocity (AHV), 95
anti-realism, 174, 27482, 285, 289,
291, 292, 296
anti-rationalism, 12338
anti-realists, 181, 275, 276, 278,
281, 285, 293, 298
functionalist account, 280
anti-reductionism, 12338
Appropriational accounts, 209,
238
assessing, 22132
manipulability-based and neoCausal accounts, 21417
philosophy of mind, standard
functionalist machinery,
21721
appropriation of machinery, 33
A-realization relations, 14n9
B
bazillion-mechanism theory, 107,
113
big-G Grounding relations, 171,
173n1
B-processes, 22931
301
302
Index
B-properties, 22930
broadly physical relation, 34, 2527,
261, 264
C
carving error problem, 29, 46
carving standard models, 48, 5364
cascade view of explanatory
mechanism, 656
causal composition relation, 152,
156, 160, 161, 173
causal explanation, 49, 50, 67n14, 69,
77, 77n2, 83, 85, 129, 145,
165n15, 175, 212, 215, 288
causal power, 124, 1367
dened, 131
forward-looking, 127, 1312
microphysical, 12930
token, 131n1,1323
causal preemption, 187, 190, 191
causal relation. See relations
causal standard models, 4853
causation, 12, 13, 19, 303, 48, 49,
137, 1458, 1525, 157n10,
1603, 1667, 178180,
1835, 187, 190, 214, 215,
262
activities account of causation,
503, 81, 97
error, 46
distinguished from Dimensioned
realization, 76
C-bers, 284, 292
classical mereological parthood
relation, 152
classical mereological partwhole
relation, 156, 171, 173, 191,
207, 208, 216, 222, 230, 232,
242
D
deductivenomological (DN) model
of explanation, 11, 42, 93, 95,
146n2, 154n9
deationists, 108, 182
dependence, 49, 51, 78, 130, 133,
135, 145, 154, 156, 160,
1636, 17980, 190, 194,
197, 201, 202, 233, 275n6
counterfactual, 33, 136, 137, 178,
1846, 189, 190
metaphysical, 145, 147, 148n3,
151, 152, 156, 1718, 186,
188, 189
ontological, 273, 275, 275n6,
283, 285, 286
depolarization, 45, 46, 48, 81
Descriptive Engagement, 208
determinable/determinate relations,
33, 132, 133, 151n1, 152,
154n9, 156, 171, 173,
177n4, 1801, 183,
18691
dialectical importance of scientic
composition, 1820
Index
E
eliminativism, 107, 114, 174, 285
eliminativists, 93, 108, 11517, 151,
281
emergentism, 174, 273
Engagement, 33, 2079, 217, 221,
238, 2434
phases of, 208
rules of, 210, 242, 244
entailment, 149, 172, 178, 253,
262n12
entity, entities, 81, 97
microphysical, 207
realism, 556
epiphenomenalism, 35, 174
exclusion problem, for nonreductive
physicalism, 12830
explanandum phenomenon, 11, 435,
47, 48, 57, 58, 85, 212
explanans, 11, 83, 85, 93, 212, 226,
234, 241
explanation
causal, 49, 50, 67n14, 69, 77,
77n2, 83, 85, 129, 145,
165n15, 175, 212, 215, 288
303
F
lament actin (F actin), 222, 223
Flat/Subset view of the realization,
15, 2369
formalism, 1536, 159, 161, 162,
166, 177, 213
structural equation modeling,
323, 79, 144, 150, 1627,
17881, 178n5, 18391, 201,
214, 2345
formulation of physicalism, 250, 251
consistency of grounding, 2608
dispensability of grounding,
25760
grounding and broadly physical,
2527
forward-looking causal power, 127,
131
304
Index
G
General Extensional Mereology
(GEM), 100, 112
Global account, 212, 23, 24
globular actin (G actin), 222
good parts
as components, 549
as mutually manipulable, 5963
as scientically approved, 634
Greshams law, 287, 290
Grounding, 1n2, 2, 3n5, 1446,
171202, 2714
and anti-realism, 27482
and broadly physical relation, 34,
2527, 261, 264
consistency of, 2608
dialectical import of, 1725
dispensability of, 25760
H
head direction cells (HD cells),
945, 102, 103
head direction representation
(HDR), 945, 1014
higher-order state-type, 2579
homuncular fallacy, 86, 88n10
homuncular functionalism, 219n10
hydrogen uoride (HF) molecule,
75, 76, 82, 84
I
identity
mechanistic constitution, 1024
material constitution of the
mental, 12338
Index
ideology, 176
inference, 137, 162, 165, 166
inaming, 160
informative claims, grounding and,
14651
interference experiments, 62
intervention, 60, 61
interventionism, 12, 50, 216
intrinsicness, 184
J
joint role-lling, 23942
L
lateral mammillary nuclei (LMN)
HD cells in, 95
Leibniz law, 105
levels standard models, 48, 649
Local account, 212, 23, 25
locality, 184
L-realization relations, 14n9
M
made-up-of relation, 31, 1236,
130, 1336, 138
manipulability relations, 5566,
21416, 22832
manipulationist theory, 98
mass-energy neutrality, 226, 227
material constitution
formal characterization of, 1245
and mechanistic constitution,
connections between,
11314
and mechanistic constitution,
dierences between, 10813
305
306
Index
N
narrowly physical relation, 34, 251,
252, 2546, 257, 2605
necessitation, 30, 49, 124, 125,
1345, 232, 236, 241, 2524,
259, 260, 264, 274
neo-Causal approach, 1213, 22, 26,
28, 21415, 22832
neo-Causal research tradition, 2, 4,
12, 16
neo-Functionalist account, of vertical
relations, 35, 19, 26
neo-Functionalist V-frameworks,
1516, 22, 25
new mechanistic philosophy, 1112,
4171
carving standard, 5364
causal standard, 4953
Dimensioned realization, 7983
explanatory constraints,
formulating, 448
levels standard, 649
mechanistic explanatory
framework, 434
nihilism, 107, 154
nihilists, 115, 153
nomic necessitation, 253, 254
non-causal explanation, 756, 789
non-causal relations, 1n2
non-reductive physicalism, 174, 273,
290n17, 291, 292, 296
exclusion problem for, 12830,
288
nonreductivism, 124, 127, 130
Index
O
Ockhams razor principle, 1813,
185
omissions, 184
ontic representationalism, 241
Ontologically Unifying Power
(OUP), 206, 210, 2257, 231,
234, 240
ontology, ontological, 1213, 14n9,
29, 67, 77, 7981, 93, 98,
99, 99n7, 101, 109,
11115, 117, 175, 176,
182, 188, 189, 207n4, 219,
220, 226, 2389, 241, 243,
266, 273, 274n5, 275,
275n6, 283, 285, 286,
2948
overdetermination, 129, 130
redundant, 129
super-overdetermination, 130
P
particles
role in mechanism, 97
phenomenal models, 88n10
philosophers of mind, 273, 274
Philosophical Engagement, 208
philosophy of mind, 2, 11, 1416,
102, 209, 211, 243
Grounding in, 27198
standard functionalist machinery
of, 21721, 2368
philosophy of science, 2, 3, 1014,
19, 70, 77, 80, 93, 20810,
212, 214, 21820, 228, 235
manipulability-based and neocausal accounts, 21417
307
phototransduction, 77, 78
physicalism, 124, 250, 251
broadly physical relation, 34,
2527, 261, 264
consistency of grounding, 2608
dispensability of grounding,
25760
formulation of, 345
grounding and broadly physical,
2527
narrowly physical relation, 34,
251, 252, 2547, 2605
non-reductive, 12830, 174,
273
reductive, 174, 273
physicalist approach, 250, 251
physical state-token (p), 2579, 266,
267
piercing explanatory power (PEP),
30, 87n9, 206, 210, 219, 225,
23742
plausibility, 55, 56
pluralism, 107, 154, 193
pluralists, 92, 1057, 11117, 153,
156, 160, 193
Positivist(s)
Nagelian model of reduction, 10
philosophy of science, 1011
power
causal (see causal power)
expressive, 1956
Ontologically Unifying Power,
206, 210, 2257, 231, 234,
240
piercing explanatory power, 30,
87n9, 206, 210, 219, 225,
23742
powers-based accounts, 184
308
Index
Q
qualitative distinctness, 86n8,
87n9, 23641
Quinean approaches, 272, 295
R
realism, 174
anti-realism, 174, 27482, 285,
289, 291, 292, 296
entity, 556
functionalist, 279
moral, 147
reality, 145, 146, 148, 163, 176,
182, 186, 188, 193, 198,
199, 263, 272, 273,
27581, 285, 289, 290,
2937
realization
A-realization, 14n9
Dimensioned, 15, 7589, 125
Index
309
S
scientic composition, 2, 1113, 30,
20712, 214, 221, 22544
dialectical importance of, 1820
Scientic Revolution, 205
semantic, logical, abstract, and/or
mathematical (SLAM)
entities, 21113, 233, 234,
243
set membership relation, 133, 152,
156, 173, 177n4, 183,
191n15, 198
small-c causal relations, 153, 155,
160, 180, 1835, 186
small-g grounding relations, 143,
1512, 1557, 15961, 166,
171, 173, 174, 17680,
177n4, 181n6, 185, 18996,
198, 200
spacetime region
role in mechanistic and material
constitution, 114
special sciences debate, 28693
stability, 557
standard functionalist framework,
21721, 23642
standard model, 4469
carving, 48, 5364
causal, 4853
levels, 649
stipulation, 254, 255
stop problem, 66, 67, 67n14, 69
structural equation modeling (SEM),
323, 79, 144, 150, 154,
1627, 17881, 183, 18591,
201, 213, 214, 2345
Subset/Flat view of realization, 220,
228, 237
310
Index
T
theory of object eliminativism, 114
token causal powers, 131n1, 1323
token identity, 152
topic-neutral Ramseycation, 219
transference accounts, 184
transitivity, 184
type identity, 104, 111, 133, 173,
284
typetype reductionist proposals,
127
U
Unengaged approach, 21013, 238
unexplanatory phenomenological
models, 44, 44n2
unicationist model of explanation,
93
unity argument, for Grounding,
17691
V
verticality. See vertical relations
vertical relations, 1, 1n2
general characterization of, 25
neo-Functionalist account of, 35
V-frameworks, 36, 11
comparative, 7, 20, 24, 26
competitive, 6, 9, 10, 18, 19
focused, 8, 20, 24, 26, 30, 32, 33
Grounding, 4, 5, 1719, 2223,
25, 312
meta-justifying, 278
meta-success, 9, 20, 27, 33
neo-Causal, 4, 12, 13
neo-Functionalist, 1516, 25
types of, 203
Vitalism, 22831, 233, 235
W
Wilsons pluralistic framework, 15661
working parts. See good parts
Z
zooming error problem, 29, 47,
47n7, 65