might say that a relation is metaphysically primitive 3 just in case there is a fact F such
that (i) F has the grounding relation as constituent and (ii) every fact that either
grounds F or is among the facts that ground F itself has the grounding relation as a
constituent.^7 Perhaps there are other notions of metaphysical primitiveness worth
articulating, but these three will suffice for now.
None of these construals of metaphysical primitiveness is equivalent to ideological
primitiveness or methodological primitiveness. A feature might fail to be perfectly
natural and yet be ideologically primitive.Beingis probably like this. Similar remarks
apply to the second and third construals. DeRosset (2013) thinks that every (atomic)
grounding fact is grounded in a non-grounding fact, but maintains that grounding
is still ideologically primitive.^8 Conversely, we might have a definition in our
language for an expression thatshouldbe taken as primitive. (Recall the speakers
of the Gruesome Tongue from section 1.3, for whom “grue”is a semantically
primitive expression from which their expression “green”is defined.) A non-
natural feature might also be methodologically primitive. I have my doubts about
whether set-membership is perfectly natural since I suspect that sets are beings by
courtesy. Yet there is nothing methodologically improper in mathematicians or
metaphysicians employing set-theoretical notions in their theories withoutfirst
defining “set.” (As Lewis (1991: 29–31) notes, when mathematicians attempt
definitions of“set,”the definitions produced are often inapt, unclear, or involve
mere picture thinking.)
We’ve distinguished three kinds of primitiveness. The questions of whether
grounding is ideologically or metaphysically primitive will loom large in the sections
that follow. So here I will focus on the question of methodological primitivism.
Let’sfirst note that some proponents of grounding embracegrounding monism,
the view that there is exactly one grounding relation, while others embraceground-
ing pluralism, the view that there is more than one grounding relation. Given
grounding pluralism, we must assess the case for the methodological primitiveness
ofeachgrounding relation. All versions of pluralism generate a similar sort of
question. Here, the question is: in virtue of what is a given relation a grounding
relation?
One way to try to answer this question is to attempt adefinitionof what it is to be a
grounding relation. It might be that no such definition is forthcoming, however.
Recall the discussion of compositional pluralism in section 2.3. There are many
and worth pursuing. Here, however, I will focus on whether an element is in some way a constituent of an
ungrounded fact.
(^7) That grounding is metaphysically primitive 3 seems to be a consequence of the view defended in Kang
(ms). See also Raven (2016). Alex Skiles has suggested to me a weaker requirement: instead of every
grounding claim being grounded in a grounding claim, merely require that there is some grounding claim
whose only grounds are further grounding claims. 8
I imagine that, given deRosset’s other commitments, he would accept that every grounding fact is
grounded in some non-grounding fact. See also Skiles (2015).