Professional Documents
Culture Documents
The central metaphysical question in the philosophy of explanation asks for the
explains q’, and if so, are these facts fundamental or are they grounded in something
the attitudes of the speaker. But setting such irrealist views aside, the world does
underlying laws of nature: the rock impact explains the window shattering because
‘Peter the elephant is colored because he is grey’ and ‘{Socrates} exists because
Socrates exists’. These explanations might arise from general laws which entail that
any grey object is also colored, and that any object is the member of its singleton
set (Glazier this volume). The role of such laws of metaphysics is to guide the
bottom-up development of facts, much like the role of laws of nature is to govern
There is a way to resist this analogy between causal and metaphysical explanation
in the explaining facts, in which case no law is required to account for metaphysical
explanations. Consider Peter the elephant again, whose greyness grounds, i.e.
1
contained in his being grey, no metaphysical law is required to take the explanans
simply accounted for by the fact that Peter is grey (Bennett 2011, deRosset 2013).
But without laws it will remain mysterious why similar grounds have similar
‘metaphysical effects’, and why dissimilar grounds have dissimilar effects. Why is
every grey object colored because of its being grey? This sort of generality of
with the status of a general law; manipulation requires that p explains q only if
q. Since such counterfactual variation requires modally robust laws, both the
We have been understanding metaphysical laws are general principles that support
metaphysical explanations. i But what could such principles be? This section asks
2
It is easy to see what the explanatory import of the singleton law is supposed to be:
‘{x} exists’:
An important but subtle point is that SL does not account for explanations like S,
then {Socrates} exists, which in turn is partly grounded {Socrates} exists. By the
In order to avoid any circularity, we should say that not SL, but the fact that SL is
SL* It’s a law of metaphysics that if x exists, then the singleton set {x} exists
If SL* is not grounded in instances of SL, it can explain such instances without
choose to express that notion. We will next introduce three candidates to constrain
our choice: truth-maker rules, essences, and sui generis metaphysical laws.
should then say that disjunctive representations are ‘made true’ by non-disjunctive
3
also appeal to truth-maker rules for other areas of discourse. On the extreme end of
the spectrum, even talk about ordinary objects is made true by a reality that contains
only atoms in the void (Williams 2010). Whether or not we accept such an extremist
essence or nature of things (Zylstra this volume). Boris Kment (2014: ch. 5) refers
2014b) offers examples like the following: the presence of the conference is
Perhaps all the metaphysical laws we ever need are truth-maker rules and essence-
truths. But if truthmaking and essence do not suffice to support all metaphysical
understood in terms of other notions from the metaphysics toolkit. The question of
whether there are such sui generis laws of metaphysics is currently among the most
If all metaphysical laws were truth-maker rules, reality would be flat as a pancake,
and the world’s layers would be mere artefacts of our representations. Since this
view is at odds with a rather sober form of realism, call it ‘realism about layers’, we
should expect that some metaphysical laws are either essences or sui generis. But
should we side with Dasgupta (2014a), who holds that no sui generis laws are
needed, or with Schaffer (2018), who rejects the need for essence? Or do we need
between essences and sui generis laws. Sections 3–5 discuss three arguments for
4
sui generis laws, and sections 6–8 raise three problems for such laws. The reader
should note that the following sections are meant to inspire further work, rather than
This section and the following two introduce reasons to go beyond essences to sui
generis metaphysical laws. The key consideration of this section is that essences
Consider again the explanation that Peter the elephant is colored because he is grey.
We would expect the underlying law to look something like the following (‘ML’ is
Essence-claims of the form ‘it’s essential to entity x that p’ require not only a
content clause (p), but also a bearer (x). If Color is a veiled essence-truth, we must
would need to pin the essence-statement on an entity. Any essence that supports a
answer them and because nothing seems to hinge on how we answer them. Consider
5
questions might be answered if, say, pain and goodness could be fully defined. But
in case they cannot, it is unclear, what considerations might even bear on these
These considerations are more compelling from the point of view of an essence-
skeptic. Schaffer (Ibid.: 312), for instance, argues that the bearer-aspect of essences
which have less structure than essences, but seem just as fit to do the job.
Proponents of essence, however, will deny that essences earn their keep merely by
accounting for metaphysical explanations. There is, after all, a live practice of
for relatively simple entities such as sets or events. These practices might justify
accepting the ideology of essence. Moreover, proponents of essence will point out
that some pinning questions have obvious answers: it is essential to {Socrates}, and
essential to you, and not to your parents, that you are your parents’ child.
But even proponents of essence may agree that pinning-questions for some
metaphysical laws appear so esoteric and difficult to answer that they better be
avoided. To develop this sort of argument further, one could try to isolate a class of
unanswerable. This class might include essences whose sole purpose is to connect
6
4. Anselmian Essences and Substantive Ontology
The second motivation for sui generis metaphysical laws is based on Gideon Rosen
(2006) who argues that there aren’t enough essences to ground all metaphysical
necessities. If successful, his argument also shows that essences do not support all
laws.
Rosen rejects so-called Anselmian essences, which entail the existence of their
that God’s existence is included in her essence. God’s essence thus conceived
would be Anselmian because it entails the existence of its bearer. According to the
cases involve essences that entail the existence of their bearers conditionally. To
illustrate, consider standard set theories and mereological theories, which entail that
if certain objects exist (and have certain features), then the associated set or
mereological fusion also exists. If these theories are essential to sets and fusions,
7
fundamental entities. AE, for instance, generates Socrates’s singleton set from
Socrates.
Rosen’s argument for a general ban on Anselmian essences rests on the following
intuitive claim: we could full-well understand what sets and mereological fusions
are supposed to be and yet deny their existence. IF AE was in fact a correct
what that singleton set is would require accepting that entity on the assumption that
Socrates exists. But it seems unfair to charge nominalists, who reject sets altogether,
say: “I reject your sets and fusions for what they are!” The nominalist might be
point). Sui generis metaphysical laws are the natural alternative: we can posit laws
to the effect that every object forms a set, without making that principle part of what
sets are. Correspondingly, a failure to know these laws would not reveal a lack of
understanding what sets are. To push back against sui generis laws, therefore,
meanings are determined by nominal definitions (Fine 1994). Assume that Rosen’s
ban on Anselmian essences and corresponding real definitions is lifted. Then there
could be definitions that conditionally entail the existence of all sorts of derivative
entities. There could, for instance, be definitions not only of mereological fusion,
8
but also of schmusion, which are like fusions, but which are composed only by
because it may entail that this real definition is ‘in effect’ and, hence, that there is
with a general ban on Anselmian essences, then what reason could there be for
rejecting it? After all, any phenomenon we can think of, including unicorns, dodos,
anyone has laid down a definition. Since real definitions are cheap in this way,
Anselmian essences run the risk of deflating derivative ontology: the threat is that
every derivative entity we can consistently define also exists. Sui generis laws of
metaphysics, in contrast, are robust principles that enable substantive debates about
A third argument in favor of sui generis laws of metaphysics runs as follows. Some
metaphysical explanations are productive in the sense that the explained facts are
laws. We need sui generis metaphysical laws because only such laws have the
power to govern.
To unpack this argument, let us first look at the notion of productive explanation.
Familiar causal explanations appear productive. When the rock-impact causes the
window to shatter, the resulting shattering is entirely new to reality and was brought
explanations, which subsume already existing facts and events under general laws.
9
For instance, so-called Humeans take causal explanations to be systematizing rather
of local matters of particular fact, just one little thing and then another” (Lewis
are lines drawn between facts that were already given; Humean laws, therefore,
2 + 3 = 5 can be derived from the axioms and suitable definitions. But it sounds odd
Rather, physicalists will say that the physical facts make it the case that the mental
facts are what they are, have the intrinsic natures they do. They will say that it all
unfolds ‘upwards’ from the physical. Rather, physicalists will say that the physical
facts make it the case that the mental facts are what they are, have the intrinsic
natures they do. They will say that it all unfolds ‘upwards’ from the physical.
(2011: 33) points out, broadly reductive views, such as physicalism about the
mental and naturalism about the normative, require that mental and normative facts
systematized a mosaic of facts that were already given, they would not be suitable
for sustaining such views. Hence, some metaphysical explanations are productive.iii
the governing operations of the laws of nature. For laws to govern is for them to
10
generate output-facts based on input-facts: laws of nature govern the shattering of
the window, just in case they take the rock-impact and generate the shattering on
directly explain the truths in their scope. For instance, your essence explains that
you are the child of your actual parents and the essence of {Socrates} explains that
Socrates is its sole member. As Glazier (2017) explains, ‘it is essential to x that p’
directly explains p. If essences directly explain the truths in their scope, then they
are not input-output operations, and are therefore not apt to govern.
To further illustrate this point, consider the essence-claim ‘It’s essential to color
that everything grey is colored’. How does this essence-claim facilitate the
explain the truths in their scope, then the assumed essence of color directly explains
the general truth ‘Every grey object is colored’. The explanation ‘Peter is colored
because he is grey’ then comes about derivatively by subsuming the pair <Peter is
11
This reasoning shows that essences don’t govern if they directly explain the
input-output mechanisms, which don’t establish the truths in their scope directly,
but which fundamentally take input-facts to output facts.iv A law of nature which
says that all hit windows shatter, for instance, might, in the first instance, take rock-
windows shatter is then derivative upon the law’s work on its instances. Since
essences don’t act as input-output mechanisms, but as direct explainers of the truths
in their scope, it is hard to see how essences could govern, and hence how
We saw over the past three sections that sui generis metaphysical laws may avoid
idle complexity, allow for substantive ontological debates, and account for the
productive aspect of metaphysical explanations. But there are also difficulties and
open questions arising for these laws. The following three sections present three of
Schaffer (2018) argues that some laws of metaphysics are fundamental. Assume
that ‘it’s a law of metaphysics that p’ is grounded in q. Then there is a second law
that supports this grounding-explanation. Since it cannot be the same law (‘it’s a
law of metaphysics that p’) on pain of circularity, a new law is required. If this new
Fundamental metaphysical laws, however, conflict with the claim that nothing
fundamental involves anything derivative. Ted Sider has dubbed a similar claim
12
‘the purity principle’ or purity for short (Sider 2011: ch. 7). Since metaphysical laws
connect entities of different ontological levels, they involve derivative objects and
either metaphysical laws or the purity principle. But purity is attractive, especially
Purity does not only conflict with fundamental sui generis metaphysical laws but
derivative entities, whether they are construed as sui generis metaphysical laws or
essences are autonomous, in the sense that essences are not apt for metaphysical
between essence and real definition. It makes little sense, Dasgupta argues, to
The autonomy of essence supposedly removes the tension with purity. For, the
purity constraint understood properly, says Dasgupta, is the claim that all
ungrounded truths which are apt for explanation mention only fundamental
Aside from rejecting purity or exempting metaphysical laws from its jurisdiction
(Schaffer 2017), the proponent of sui generis laws might try to push back against
Dasgupta’s use of autonomy (Glazier 2017, Raven forthcoming), or else claim that
their laws are also autonomous. But there are two other strategies worth presenting,
which don’t invoke the notion of autonomy: first, we might reduce all metaphysical
13
laws, pace Schaffer’s regress, and, secondly, we could develop a conception of
Consider first the attempt to reduce sui generis metaphysical laws to regularities in
an antecedently given mosaic: metaphysical laws are regularities that are important
Rosen was the first to sketch such a regularity theory. On his view, a regularity has
the status of a metaphysical law, if it is “governing how objects in general are put
Regularity theories face the following three difficulties. First, they must specify a
recipe, perhaps analogous to Lewis (1983)’s best system account of laws of nature,
which details the reduction of laws to regularities. Secondly, just like causal
address the regress-worry: if laws are grounded in regularities, don’t they require
These difficulties do not knock regularity theory off the table. There should be some
way to devise the relevant best system, and if Humeans can live without productive
explanations as well. Moreover, we might try to avoid the regress if we deny that
the explanation of metaphysical-law facts requires further sui generis laws. Instead,
these explanations might appeal to the essence of laws or to truth-maker rules: law-
14
A less radical response to purity worries offers fundamental laws that don’t involve
derivative object, property, or any other kind of derivative entity is constructed from
more basic entities. Some entities, like a mereological sum, a constituted artifact,
or a complex property, are constructed from their constituents; and some entities,
like a smile or a derivative part, are the constructed aspects of more expansive
entities like faces and organic wholes. According to the constructional conception
to entities and how the constructed entities ‘go together’ based on their
constructional profiles.
takes facts of the form of Fa and Ga to facts involving the conjunctive property
F&G: Fa, Ga → F&G(a). Does a law of this sort violate purity constraints? It
concern objects and properties that are constructed via fundamental constructors
A second difficulty for sui generis laws of metaphysics arises from the dictum that
metaphysical explanations are the “tightest” of all explanations (Fine 2001: 15). In
this vein, it is sometimes said that the metaphysically explained fact is ‘nothing over
15
and above’ the explaining facts, or that metaphysical explanations are reductive.
The challenge for proponents of sui generis laws is to explain how metaphysical
laws might be helpful. Laws of nature establish causal explanations, and emergence
laws establish non-causal explanations, which are however supposed to leave the
explained facts unreduced and independent from the facts in the emergence-base.
We can debate over whether there is room for the notion of an emergence-law, as
this volume.) But the challenge remains: what separates laws of metaphysics from
other laws, which establish explanations less tight than metaphysical explanations?v
We could answer this question with brute force by stipulating that this is just what
a law of metaphysics is: metaphysical laws are those laws that establish maximally
intimate explanations. The problem with this primitivist answer, apart from leaving
intimacy unexplained, is that it makes reductive enterprises look too easy. If you
wish to ground the normative in the descriptive or the mental in the physical, simply
states and be done. The intimacy-inducing nature of the postulated laws determines
that the mental (or normative) is nothing over and above the physical (or natural).
Doesn’t that make vindicating physicalism and normative naturalism too easy? vi
To avoid this consequence, we can constrain the contents of metaphysical laws: not
can impose constraints on the contents of laws that make it harder to postulate them,
and that also explain their connection to the tightness of metaphysical explanations.
One kind of constraint that may play the desired role comes from the constructional
16
conception (section 6). On that conception, metaphysical laws take physical states
to mental states, for instance, by specifying how the resultant state is constructed.
The first law is easy to postulate and does not render intelligible why resulting
from physiological properties or states. As we will see next, LAW 2 also explains
The constructional conception is naturally combined with the claim that talk of
‘nothing over and above’ does not really apply to truths, but rather to entities: the
composite entity is nothing over and above its parts, the statue is nothing over and
above its constituting matter, and the smile is nothing over and above the smiling
face. Generalizing from these cases, x is nothing over and above the ys, just in case
x is constructed from the ys. Since being nothing over and above is an ontological
relationship and not one between truths, we should say that two truths, p and q, bear
an intimate relationship to one another only insofar as some of the entities featuring
On this view, there is nothing about the laws of metaphysics per se which renders
17
kinds of explanation intimate. No explanation of mental truths in terms of physical
ones makes for a reduction; what is needed for physicalism, for instance, are
reductive ontological relationships between mental and physical entities like mental
To sum up, the reductive nature of metaphysical laws could either come from the
the content of the laws explains both why metaphysical laws don’t come for free,
conditional ‘if p then q’ is also necessary (although see Skiles 2015 and Skiles this
volume for critical discussions). But how do metaphysical laws give rise to
necessity and what kind of necessity do such laws give rise to? We will see that
On the received view, there is one absolute kind of necessity or necessity tout court,
several relative kinds of necessity, which are defined from the absolute one. The
that if the actual laws of nature hold, then p. Since necessity and possibility are
18
duals, the notion of a relative kind of necessity can also be understood as a
worlds at which the actual laws of nature hold (see Fine 2002 for disagreement).
The problem for sui generis laws of metaphysics is now as follows. On the one
hand, metaphysical laws seem necessary tout court. For instance, it just seems
absolutely impossible that Peter is grey and yet not colored, or that Socrates exists
without his singleton. But the case of laws of nature suggests, on the other hand,
that the necessity of laws is a relative kind. Since sui generis laws of metaphysics
If metaphysical laws are essences, this difficulty is easily avoided, because essences
are associated with absolute necessity. But there are also two responses available to
proponents of sui generis metaphysical laws. The first one denies that the necessity
of metaphysical laws is absolute, and the second response denies that laws only
give rise to relative kinds of necessity. As we will see, both responses raise difficult
questions.
The first response insists that the necessity of metaphysical laws is a relative sort of
necessity. Gideon Rosen (2006), for instance, suggests that there is an order of
necessity of laws of nature. On his view, the necessity of essence is absolute, but
the other two kinds of necessity are relative because metaphysical and natural
possible iffdef p is consistent with essences and metaphysical laws, and p is naturally
possible iffdef p is consistent with essences as well as metaphysical and natural laws.
19
On Rosen’s view, metaphysical and natural necessity are relative. But as Rosen
points out, relative kinds of necessity run the risk of deflation. For, we can use
essence-necessity together with any arbitrary set of truths to define a relative kind
of necessity. We could define, for instance, that a truth p has Peter-necessity just in
to us, the difference between them may not be sufficiently objective. (Rosen 2006
and Leech 2016 disagree. See also Fine 2002 and Wilsch 2017.)
relative kind. It follows that the necessity of metaphysical laws is absolute, and so
is the necessity of essence and the necessity of laws of nature. The view that
to address the following questions. First, is there one or are there multiple absolute
kinds of necessity? Secondly, what determines whether two sources exert the same
or different kinds of necessity? And what does it take for one absolute kind of
can be addressed, proponents of sui generis laws must explain whether the necessity
20
of metaphysical laws is distinctive, or whether it is the same as the necessity of
i
See Wilsch (2016) for an account on how metaphysical laws support explanations.
ii
Ted Sider (2012: ch.7) develops an important framework for truth-maker rules; not to be conflated with
truth-maker theory.
iii
Bennett (2011) accounts for the productive-ness of metaphysical explanations with her view,
discussed in section 1, that the grounded fact is contained in the ground.
iv
Glazier (2015)’s account of metaphysical laws as a variable-binding connective seems particularly
well-suited for a construal of metaphysical laws as input-output mechanisms.
v
There is a tension between the intimacy of metaphysical explanations and the claim from section 5 that
some metaphysically explained entities are ‘entirely new’. Although this tension may perhaps be resolved,
one could use it to argue against the productivity or the intimacy of metaphysical explanation.
vi
‘Grounding physicalism’ as developed in Schaffer 2017 is a potential target of this argument.
vii
An interesting exception to this ontological construal of ‘nothing over and above’ occurs with
applications of truth-maker rules.
viii
Many thanks to Nina Emery, Stephanie Leary, Mike Raven, Jonathan Schaffer, and Erica Shumener for
feedback on earlier drafts.
Related Topics
in this volume.
For a more thorough discussion of Humean reductions of laws, the reader should
Bibliography
10):557-592.
21
----- (2001) “The Question of Realism,” Philosophers’ Imprint 1 (1):1-30.
Glazier, M. (2015) “Laws and the Completeness of the Fundamental,” In Mark Jago
(ed.), Reality Making. Oxford, UK: Oxford University Press. pp. 11-37.
Press.
(2):158-180.
----- (1986) Philosophical Papers, Volume II, Oxford: Oxford University Press.
(12):3145-3165.
Maudlin, T. (2007) The Metaphysics within Physics, New York: Oxford University
Press
22
----- (2010) “Metaphysical Dependence: Grounding and Reduction,” In Modality:
University Press.
321.
Sider, T. (2012) Writing the book of the world. Oxford University Press.
751.
141.
172 (12):3293–3312.
23