Fodor on Having Concepts
DANIEL A. WEISKOPF AND WILLIAM BECHTEL
Abstract:
Fodor offers a novel argument against Bare-bones Concept Pragmatism
(BCP). He alleges that there are two circularities in BCPs account of concept
possession: a circularity in explaining concept possession in terms of the capacity to
sort; and a circularity in explaining concept possession in terms of the capacity to draw
inferences. We argue that neither of these circles is real.
1. Introduction
As Fodor sees it, two views about the nature of concepts are fundamentally in
competition with each other. Pragmatism is the doctrine that concept possession is
constituted by certain epistemic capacities. On the other hand, according to the
kind of Cartesian view of concepts Fodor advocates, concept possession is an
intentional state but not an epistemic one. Having the concept DOG is just being
able to think about dogs (as such).
This distinction is problematic since its not entirely clear what the contrast
between Cartesian and Pragmatist accounts of concepts is supposed to amount to.
Presumably everyone who thinks that there are concepts thinks that one of
the things that they do is allow their possessors to think about or represent part
of the world. At least, every Pragmatist who isnt a behaviorist thinks this; certainly
the cognitive psychologists Fodor attacks do. Since Cartesians do as well, and
behaviorists are out, its unclear at the outset how the contrast is to be drawn.
Granting that having concepts lets us think about the world, its a further
question what the properties are in virtue of which someone has a concept,
hence has the ability to think about dogs as such. Having an account of those
properties (ideally, non-intentional properties) would be having an account of the
intentionality of thought. Conceptual or inferential role semanticists take it that
how one gets to think about Xs is by having a symbol with the right kind of causal
and inferential relations to other symbols and to Xs. Informational atomists, on the
other hand, think that how one gets to think about Xs is by having a symbol that is
nomologically locked to them, regardless of what mediates that locking or how
that symbol is in turn related to other concepts.
Hence its common ground that concept possession entails the ability to think
about Xs. The differences among theories of concepts are over what capacities an
organism must have to have concepts. Informational atomists claim that there are
Address for correspondence
: Department of Philosophy, University of South Florida, 4202
East Fowler Ave, FAO 226 Tampa, FL 33620, USA.
Email
: weiskopf@luna.cas.usf.edu
Mind & Language
, Vol. 19 No. 1 February 2004, pp. 4856.
#
Blackwell Publishing Ltd. 2004
, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. no particular cognitive capacities that are metaphysically necessary to become locked
to a property, hence there are no particular capacities that are metaphysically
necessary for concept possession (although it may be necessary that there be
some such capacities or other present). Conceptual role theorists disagree, propos-
ing a varying array of capacities that are constitutively linked to representation of a
content. The locus of the debate, then, is not between those who think concepts
are for thinking about things versus those who think concepts are for guiding
certain kinds of behaviors or mental acts. Rather, its between two views about
how to understand what it takes to be able to think about things.
With this clarification, consider the family of Pragmatist accounts Fodor criti-
cizes. According to Bare-bones Concept Pragmatism (BCP), concept possession is
constituted by. . .capacities for INFERRING and SORTING. Fodor levels three
objections against BCP. Two of these, the objections from analyticity and com-
positionality, are extremely familiar. Since they have been discussed extensively in
the concepts literature, we wont expend any more space commenting on them
here. Rather, well focus on what Fodor claims to be a novel objection against
BCP: the objection arising from circularity. This objection is leveled against both
of the capacities BCP proposes are constitutive of concept possession, sorting and
circularity. Well address these in turn. Our claim is that neither argument is
ultimately persuasive.
2. The Sorting Objection
Consider the circularity objection to the sorting condition first. The objection has
the form: (1)
Having concept C depends on being able to sort Cs; and (2) Being able to sort
Cs depends on having concept C (or something conceptually equivalent to it). And this
would be viciously circular. (1) is a premise of BCP, so the question is whether there
is a case saddling the Pragmatist with (2). Well argue that there is not, and that the
appearance to the contrary depends crucially on an ambiguity about sorting.
What is it to satisfy the sorting condition? As Fodor formulates the condition,
for a concept C to have a sorting requirement in its possession conditions is for
it to require the ability to sort things into those that C applies to and those that
it doesnt. He notes first that how subjects actually sort things doesnt distin-
guish contingently coextensional concepts from each other. So it matters how
subjects would sort their environment. But even counterfactual sorting alone
wont distinguish necessarily coextensional concepts like WATER and H
2
O, or
TRIANGLE and TRILATERAL. Its possible that subjects could sort the
same under all counterfactual circumstances, yet their sorting could be under
the control of different concepts. The result is this: possibly, a subject has two
distinct concepts, C
1
and C
2
, that are connected to necessarily identical sorting
behavior.
This suggests a dilemma for the BCP theorist: either deny that two distinct
concepts can produce necessarily identical sortings, or find another factor to
Remarks on Fodor
49
#
Blackwell Publishing Ltd. 2004 distinguish them. The second horn is, for good reason, the more popular one.
Indeed, it is the horn Fodor himself takes when he claims that WATER and H
2
O,
although identical in content in virtue of being locked to the same property, are
distinct concepts in virtue of their structural differencesand thus, incidentally,
in virtue of their possession conditions, since structural differences entail differences
in possession conditions (Fodor, 1990, p. 114). We will return to this point later in
our reply.
The argument turns on how sorting is to be understood. Here are two
possible ways to understand it. In the first sense, sorting is an intentional term.
It is something that a thinker does in virtue of representing her actions and
environment in a certain way. Call this sense of sorting intentional sorting, or
I-sorting. In this sense, whether or not I am I-sorting dogs depends on more than
whether I am making a pile of dogs and a pile of non-dogs. It depends on my state
of mind, in particular how I conceive of the entities Im sorting. If Im thinking
BARKERS as I pile up dogs, then I am in fact I-sorting barkers, not dogs, even if,
de facto, all and only the dogs are barkers. The same applies to the triangle and
trilateral case, which shows that thought can make distinctions finer than necessary
coextension.
On the other hand, sorting can be thought of in non-intentional terms. Any
device or system that covaries with some condition can be said to sort in this sense,
whether the response is purely internal (like changing some inner state) or external
(like moving an object onto a pile). To mark the distinction sketched here, call this
second kind of sorting mechanical sorting, or M-sorting. In the M-sorting sense
even a simple mechanical device lacking any representational states may be able to
sort its environment.
1
However, the sorts such devices can carry out are far more
coarse-grained than those that creatures with intentional states can enact. A photo-
electric switch may trigger when light hits it, in which case it is M-sorting light-on
states from lights-off states. If the lights go on only when class is in session, and are
off otherwise, then the switch may be M-sorting class-in-session states from class-
out-of-session states (and, for that matter, students-unhappy states from students-
happy states, and so on for other coextensional state description predicates). A
person who is I-sorting objects can be regarded as M-sorting by decreasing the
granularity of the distinctions she herself is making in I-sorting; so the subject who
is I-sorting barkers (not dogs) may be M-sorting both dogs and barkers (supposing
the set of barkers is the set of dogs).
Which sense of sorting is at issue in Fodors argument? Given that BCP is
intended to be a theory of concept possession, and thus cannot presuppose further
intentional notions, the appropriate sense seems to be M-sorting. The picture then
is the following. There is an internal state of a subject, call it S, that is causally
implicated in how that subject M-sorts her environment. Suppose it functions in a
1
For those with pansemanticist tendencies, substitute no robustly representational states, or no
mental representational states.
50
D. Weiskopf and W. Bechtel
#
Blackwell Publishing Ltd. 2004 causally central way in her M-sorting triangles from non-triangles (and hence
M-sorting trilaterals from non-trilaterals). If M-sorting were sufficient for concept
possession and it functioned to determine the content of the concept possessed,
then the subject would possess a concept that was ambiguous between TRI-