Species

304 Species

Boyd defines the three core ideas of HPC in this way:

Three ideas characterize the early literature on HPC kinds. First, there are important

natural kinds such that membership in them is a matter of participating in (types or

tokens of) natural kinds of processes. Second, in many important cases the relevant

processes will involve imperfect homeostatic clustering, so the kinds defined in terms

of those processes will have indeterminate boundaries.

Finally, because the HPC conception was developed so as to apply to species defini-

tions, there was an emphasis on the dynamics of HP cluster(ing) (so that the relevant

homeostatic clustering mechanisms were to be counted as elements in the HP cluster)

and on variability: on the possibility that the components of a single HP cluster would

vary over space and time. Such HP cluster(ing)s were seen as historically individuated

in ways that reflected the contributions that reference to them makes to the reliability

of inductive and explanatory practices (i.e., in order to make the corresponding kinds

causally grounded...^67

This sense of “essence” is, as I have argued before, relatively benign. Assuming
that HP clusters can change over time by substitution (i.e., that the homeostasis is not
rigid all the time), HPC kinds function both as ampliative, or projectable, foundations
for inference and explanation, and overcome some of the issues that modern evolu-
tionists have about essentialism. This is not the place to do an in-depth analysis,^68 but
one thing we might say is that, like the monophyly-based phylospecies definitions,
HPC kinds do not specify how much clustering or what kind of properties must
cluster in order to identify a species (or for that matter, higher taxa) as that category
of kind. In short, one already needs to know what is a species, in order to causally
account for the homeostasis that makes it one. How that is done I shall argue below.
Another observation: If the developmentally entrenched properties, HPC or not,
that causally generate species are the basis for being a species, and I have no reason
to doubt that this is one of the property clusters that does so, then the developmen-
tal system itself is an autapomorphy of that species (or, if species hybridize in that
group, a synapomorphy of the clade). In short, the HPC kind account doesn’t solve
the phylospecies problem.
However, as a metaphysics of biological kinds, the HPC account has much to
recommend it. For a start, the account of kinds is drawn from biology (and social
kinds) rather than from logic and language. It deals with, as it was intended to deal
with, polytypy. If one can discover an HPC kind, then one can make projectable
inferences about unobserved members of that kind. Indeed, that is one of the vir-
tues of phylogenetic classification—because traits are phylogenetically conserved
on the whole, monophyletic groups share properties of all sorts. Hence inferences
from one species to another, or one species member to another, are warrantable. But
if species themselves are not monophyletic, and organisms do vary, that license is
reduced accordingly. This is why monophyly or coalescence are often proposed as a
definiendum of species.

(^67) Boyd 2010, 690.
(^68) For which, see the treatments by Richards 2016, Slater 2016.