76 Pieter E. Vermaas
defi nition it need not be true that x actually contributes in this way to s’s capacity. Yet it
should be true that x is capable of φ-ing, and this stronger requirement is captured in the
epistemic interpretation by the (more ontological) condition that x is capable of φ-ing.
The ontological interpretation of Cummins’s theory is a counterpart to the epistemic
interpretation since the ontological theory can be taken as providing support to the epis-
temic interpretation, provided it is the case that in the epistemic interpretation functions
are relations that items have. If Cummins’s functions indeed are such relations, and pre-
sumably they are, then the ontological interpretation seems to imply the epistemic one,
and one can take the epistemic interpretation as the epistemic associate of the ontological
interpretation.
Functions, including the technical ones, are in the ontological interpretation of Cum-
mins’s theory epistemically objective since the truth of the judgment of whether x is
capable of φ-ing in s, and by this capacity contributing to s’s capacity to ψ, depends on
physics, chemistry, and biology, and does not depend on the attitudes, feelings, and points
of view of agents. In the ontological interpretation, functions are also ontologically objec-
tive since their existence does not depend on mental states.
The formulation of the epistemic and ontological interpretations of Cummins’s theory
can be generalized by abstracting from the particular choices Cummins made for the
context c relative to which items x have functions φ, the evidential basis E agents use
for ascribing these functions φ, and the requirements that must hold for functional
descriptions.
An epistemic function theory Tep
Agent a justifi ably ascribes the capacity to φ as a function to x relative to context c
and relative to evidence E for R 2 (xφc) iff R 1 (xφc) and agent a is justifi ed to believe
on the basis of E that R 2 (xφc).
A counterpart ontological function theory Tont
Item x has the capacity to φ as a function relative to context c iff R 1 (xφc) and
R 2 (xφc).
For Cummins’s theory, the choice of c, E, and the requirements R 1 (xφc) and R 2 (xφc) are
the following:
c: the capacity to ψ of s;
E: the analytical account A of s’s capacity to ψ;
R 1 (xφc): x is capable of φ-ing in s;
R 2 (xφc): the capacity to φ of x in s causally contributes to s’s capacity to ψ.Yet other choices can now be considered. In effect, conformance to the generalized form
Tep of an epistemic function theory can be taken as a suffi cient condition for the existence
of a counterpart ontological function theory. If an epistemic theory can be brought in the