Professional Documents
Culture Documents
DOI: 10.1111/j.1468-0378.2011.00492.x
Perceptual Nonconceptualism:
Disentangling the Debate Between Content
and State Nonconceptualism
Laura Duhau
Abstract: In this paper I argue, against recent claims by Bermdez (2007) and
Toribio (2008), that within the debate about whether perceptual experiences are
nonconceptual, state nonconceptualism (or the state view) can be a coherent
and plausible position. In particular, I explain that state nonconceptualism and
content nonconceptualism, when understood in their most plausible and moti-
vated form, presuppose different notions of content. I argue that state noncon-
ceptualism can present a plausible way of unpacking the claim that perceptual
experiences are nonconceptual once the notion of content it should presuppose
is taken into account; and once this notion of content is clearly distinguished
from the one usually presupposed by content nonconceptualism, the criticisms
that Bermdez and Toribio place against state nonconceptualism become
ineffective.
It has been almost 30 years since the debate about whether perceptual experi-
ences1 are in some sense wholly or partly nonconceptual started. The issue is
whether perceptual experiences are independent (and to what extent) of the
perceiving subjects conceptual capacities (if any). A few years ago, philosophers
started distinguishing between two different ways of understanding the claim
that perceptual experiences are nonconceptual: in terms of the kind of content
they have and in terms of the kind of states they are2 (Byrne 2005; Crowther
2006; Heck 2000, 2007; Speaks 2005). Thus, what has come to be known as
content nonconceptualism (or the content view) is the view that perceptual
experiences have nonconceptual content, which is a different kind of content than
the sort that conceptual states (such as beliefs, desires and other propositional
attitudes) have. In contrast, what has come to be known as state nonconcep-
tualism (or the state view) is the view that perceptual experiences and con-
ceptual states have the same kind of content, but are different kinds of states.
What is usually understood by the claim that perceptual experiences and
conceptual states are different kinds of states is that instantiating the latter
requires the possession of concepts while instantiating the former does not. In
particular, according to state nonconceptualism, instantiating conceptual states
requires subjects to possess the concepts we use in accurately characterizing the
content of such states, while instantiating perceptual experiences does not
European Journal of Philosophy 22:3 ISSN 0966-8373 pp. 358370 2011 John Wiley & Sons Ltd
Perceptual Nonconceptualism 359
I will first present what I think is Toribios argument for the claim that SNC is
either incoherent or false,4 and then will explain how she defends its premises.
The argument goes like this:
(1) Content has to capture the way the subject grasps the world as being.
(2) The content of beliefs satisfies (1) by being composed of concepts.
(3) The content of perceptual experiences satisfies (1) either by (i) being com-
posed of concepts, or by (ii) having some other kind of content.
(4) If (i), then perceptual experiences are conceptual, against the assumption of
SNC that they are nonconceptual.
(5) If (ii), then CNC obtains.
Thus,
In favour of premise (1), Toribio claims that, for all contenders in the debate,
[t]he two main concerns are to elucidate the justificatory relation in
which perceptual experiences stand to perceptual beliefs and the
explanatory relation in which they stand to perceptual discriminative
abilities. But for both perceptual conceptualists and nonconceptualists,
only facts about how a subject grasps the world in experience can justify
a subjects perceptual beliefs and explain the subjects intentional behav-
ior. (Toribio 2008: 352)
Premise (3) is just a statement of the ways in which the content of perceptual
experiences could satisfy (1). Since Fregean content is a kind of content that
satisfies (1), a way for perceptual experiences to satisfy (1) would be to have
Fregean content. This is part (i) of premise (3). But of course, the possibility is
open for the content of perceptual experiences to satisfy (1) by having some
other kind of content (part (ii) of the premise).
Premise (4) states that SNC cannot consistently claim that the content of
perceptual experiences is Fregean, since understood in neo-Fregean terms, this
would imply that in perceptual experiences the subject possesses the concepts
that compose their contents, which is precisely what SNC denies. Therefore,
Fregean content cannot be the common currency among beliefs and perceptual
experiences.
Premise (5) is a consequence of (2) and (4) together. If the content of beliefs
is Fregean (1), but the content of perceptual experiences cannot be Fregean (4),
then the content of perceptual experiences has to be of some other kind. But then
CNC obtains, since beliefs and perceptual experiences have different kinds of
content. The only way to try to make SNC consistent, claims Toribio, would be
to deny (2) and claim that the kind of content that is the common currency
between beliefs and perceptual experiences is some kind of non-Fregean content,
such as Russellian or possible world content. However, claims Toribio, none of
these kinds of content capture the way the subject grasps the world as being, so
they are inadequate as characterizations of both the contents of beliefs and the
contents of perceptual experiences.
Bermdez (2007) gives us three reasons for thinking that state nonconceptu-
alism should not be taken seriously. The first one is related closely to Toribios
argument. On the one hand, like Toribio, Bermdez understands Fregean
content in Neo-Fregean terms, and thinks that one cannot really make sense of
the idea that perception has Fregean content but nonetheless being in a per-
ceptual state does not require one to possess the concepts that constitute the
content of that state. On the other hand, also like Toribio, he believes that the
notion of perceptual content is supposed to reflect how the distal environment
perceptually appears to the perceiving subject (ibid.: 67), and believes possible
world content is inadequate to fulfil this role (he does not talk about Russellian
content, but it is likely that he agrees with Toribio in thinking that it is also
unfit to account for how the distal environment appears to the perceiving
subject).
Bermdezs second reason to reject SNC has to do with what he thinks is the
most powerful motivation for thinking that perceptual experiences have non-
conceptual content. He says that this motivation is a claim about the direction of
explanation between accounts of what it is to possess an observational concept,
and what is to be perceptually capable of discriminating objects and properties
Both Toribio and Bermdez have a specific understanding of what the notion of
content is supposed to do. In particular, they both think that our attributions of
content should reflect how things appear to the subject, or the way that the
subject grasps the world as being. This is why they are particularly fond of a
Fregean notion of content. They think that only a Fregean notion of content
makes content fine-grained enough to allow for the explanation of certain
particular pieces of intentional behaviour. They are motivated by what I call,
following Schiffer (1992), Freges Constraint. According to Freges Constraint,
our notion of content has to serve to explain Frege Cases.
Frege Cases are cases in which a subject believes or desires Fa but not Fb
despite a and b referring to the same thing. The subject does not know that a and
b refer to the same thing, and this affects her behaviour (or her inferences
involving a/b), in such a way that the subjects behaviour does not conform to
our intentional generalizations.
A classical example is the famous case of Oedipus, who wanted to marry
Jocasta but did not want to marry his mother, despite Jocasta and his mother
being the same person. Oedipuss problem is that he has two ways to represent
the same person and he does not realize that they refer to the same thing. This
makes Oedipuss behaviour to constitute an apparent counter-example to the
following generalization:
(M) Ceteris paribus, if people believe that they should not marry their
mothers and wish not to marry their mothers, they avoid marrying their
mothers.
Oedipus satisfies the antecedent of (M), but he does not satisfy the consequent
because he marries Jocasta, who is his mother.
According to both Toribio and Bermdez, the notion of content has to serve
to explain what is going on in Frege Cases.5 Content has to be individuated in
such a way that Oedipuss desire to marry Jocasta has a different content than
a desire to marry his mother. Frege Cases are taken as one of the main
motivations to claim that contents should be individuated by something more
than truth conditions, and that the constituents of contents should be something
like modes of presentation, and not just referents. If one were to individuate
contents only by truth conditions, then one would have to accept that Oedipuss
belief that he wants to marry Jocasta has the same content than a belief that he
wants to marry his mother, in which case we would be right to attribute to
Oedipus the belief that he wishes to marry his mother. But if we did this, then
we would have to claim that Oedipus is being irrational, since he both believes
that he wishes to marry his mother and believes that he does not wish to marry
his mother. But, since it is agreed that, though Oedipus is making a mistake, he
is not being irrational, it seems that we should claim that Oedipuss belief that
he wishes to marry his mother and Oedipuss belief that he wishes to marry
Jocasta are beliefs with different contents, and that the reason Oedipus fails to act
according to (M) is that he has beliefs with the same truth conditions (or
conditions of satisfaction), but different contents, since the contents are consti-
tuted partly by two different modes of presentation. Oedipus fails to associate
his two modes of presentation; he does not realize that they have the same
referent, and this explains his behaviour. But in order to be able to explain his
behaviour in this way, we need to individuate content by more than referents
and more than truth conditions. Fregean propositions, constituted by modes of
presentation or senses, are considered, therefore, to give us the finely individu-
ated content we need to meet Freges Constraint. If we take content to be Fregean
content, we can claim that two contents with the same truth conditions are
nonetheless different contents. Therefore, under this view of content and its
world in many different ways, they grasp it in many different ways, but despite
this, they succeed both in their interaction with the world and in their interaction
with others, and it seems that for this to happen it is enough, in general, that
people have representational states with the same truth conditions as those of
other people.
We can see that the two explanatory roles, the one associated with Freges
Constraint and the one associated with the Publicity Constraint can be in
tension, since they pull in different directions with respect to how content
should be individuated. Freges Constraint seems to require a notion of content
that is individuated finer than truth conditions, while the Publicity Constraint
seems to require a notion of content that is individuated as coarse as truth
conditions. The two explanatory roles assigned to content result on two differ-
ent ways of individuating content: one that slices contents finer than truth
conditions, the other that does not. But many have thought that we somehow
need them both. For some the solution has been to accept the existence of two
different kinds of content for every representational state (see Fodor 1990, Block
1986), while others have tried to meet one of the two constraints by appealing
to something other than content. What Fregeans usually do, for example, is
claim that content has truth conditions. They just place content at a different
level than truth conditions and use it to distinguish representational states with
the same truth conditions. This allows them to explain Frege Cases at the level
of content. But they still need to explain publicity, and it seems that they will
have to do it at the level of truth conditions. Conversely, those worried about
the Publicity Constraint place content at the level of truth conditions, thus
explaining publicity at the level of content. But they have to explain Frege Cases
at another level (for example, by claiming that holding two beliefs with the
same content may result in a Frege Case when the subject has different
representational states towards them, since she grasps the same content in
different ways).
One needs to have an account of both conceptual and perceptual states that
can explain both Frege Cases and publicity, but it seems that one does not
have to, and from certain perspectives one in fact cannot, explain both Frege
Cases and publicity at the level of content, since there is no one notion of
content that can fit both explanatory roles. There is no one notion of content
that can be individuated finely and coarsely at the same timealthough one
can, as indeed some philosophers have done, postulate that all representa-
tional states have two different kinds of content, one coarsely individuated,
one finely individuated.
reason to avoid trying to see the plausibility of SNC from the framework given
by the other, coarse-grained, notion of content. And from this other perspective,
things may look quite different for the defender of SNC.
So, how does SNC explain the difference between perceptual experiences
and conceptual states? SNC claims that perceptual experiences and conceptual
states have the same kind of content. Under the framework given by the
coarse-grained notion of content, SNC has to be understood as claiming that
both perceptual experiences and conceptual states have truth-conditional
content. Therefore, for the defender of SNC, the difference between the two
kinds of states has to be explained by appealing to something other than kinds
of content. While CNC explains the difference between conceptual states and
perceptual experiences at the level of content, SNC can explain it at the level
of mental representations. This involves claiming that what determines whether
a state is conceptual or nonconceptual is the kinds of representations that are
deployed in it. This explanation assumes that conceptual and nonconceptual
representations are not constituents of contents, but instead are entities that have
content. Therefore, under this understanding of SNC, conceptual and noncon-
ceptual states can have the same kind of content, viz. coarse-grained truth-
conditional content. The way the subject grasps the world as being is
accounted for at the level of representations of content, not at the level of
content itself. But it is accounted for. This strategy is familiar from other
debates. Fodor (1998, 2009), for example, uses this strategy to explain Frege
Cases under his theory of concepts. He claims that what explains the failure of
Oedipus to realize that he has two representations with the same content is
that these two representations are different in form. Hence, his explanation of
Frege Cases uses a coarse notion of content, individuated in terms of referents
and truth conditions, together with a fine-grained individuation of represen-
tations, which includes both the content of the representations and their non-
semantic properties.
What I am suggesting is that SNC is a plausible and consistent position under
a coarse-grained notion of content. What SNC does is account for the way the
subject grasps the world as being i.e., satisfy Freges Constraint, at the level of
representations, claiming that the same content can be represented in different
ways. SNC amounts to the claim that while propositional attitudes represent
content via conceptual representations, perceptual experiences represent content
via nonconceptual representations. The difference between these two kinds of
representations is accounted for in terms of the formal or syntactic properties
that they have, and not in terms of the contents they represent. SNC can then be
defended by explaining the differences between propositional attitudes and
perceptual experiences in terms of a difference among the kind of representa-
tions that they involve, instead of a difference in the kind of content that they
have. Propositional attitudes involve concepts, and perceptual experiences do
not. Concepts are a kind of representation of content, not a kind of constituent
of content.6
So, summarizing, this is the difference between CNC and SNC:
CNC claims:
(1) Perceptual experiences are nonconceptual, propositional attitudes are
conceptual.
(2) The conceptual/nonconceptual distinction is a distinction at the level of
content, so (1) implies that perceptual experiences have a different kind of
content than do propositional attitudes.
(3) The content of propositional attitudes is conceptual (or Neo-Fregean), so the
content of perceptual experiences must be something different.
(4) Both the content of propositional attitudes and the content of perceptual
experiences have to be fine-grained (have to meet Freges Constraint).
SNC claims:
(1) Perceptual experiences are nonconceptual, propositional attitudes are
conceptual.
(2) The conceptual/nonconceptual distinction is a distinction at the level of
states, so (1) implies that perceptual experiences are different kinds of states
than propositional attitudes. They are different kinds of states because they
involve different kinds of mental representations. The conceptual/
nonconceptual distinction applies at the level of mental representations, not
at the level of content. Two mental representations can be of different kinds
(because they are syntactically or formally different) while having the same
content.
(3) Propositional attitudes are conceptual mental representations, while percep-
tual experiences are nonconceptual mental representations, but they share
the same kind of content, because they can have the same truth conditions.
In other words, you can represent the same content in two different ways: via
concepts and via nonconceptual mental representations.
(4) Both the content of propositional attitudes and the content of perceptual
experiences have to be coarse-grained (have to meet the Publicity
Constraint).
We can see that, when the notion of content in use is made explicit, SNC and
CNC are very different theses. Both make sense of the claim that perceptual
states are nonconceptual, but they do it in different ways, because they rely on
different notions of content.
From the perspective of someone working with the notion of content that SNC
presupposes, Toribios argument and Bermdezs reasons for disregarding SNC
become ineffective. Let us look at Toribios argument first. Premise (1) is denied
by the defender of SNC, since it is not the case that content has to capture the
way the subject grasps the world as being; the way the subject grasps the world
as being can be accounted for by appealing to other (non-semantic) properties of
representations. It is not the content of beliefs that is composed of concepts, as
Toribios premise (2) claims, but beliefs themselves. Beliefs are conceptual mental
representations, and are thus composed of concepts. Perceptual experiences are
mental representations that are not conceptual, and therefore not composed of
Laura Duhau
Instituto de Investigaciones Filosficas
Universidad Nacional Autnoma de Mxico
Mxico
lauraduhau@yahoo.com.mx
NOTES
1
Perceptual experiences are supposed to be perceptual states at the personal level, as
opposed to subpersonal perceptual states, such as images in the retina, to which the
subject has no conscious access.
REFERENCES
Byrne, A. (2005), Perception and Conceptual Content, in E. Sosa and M. Steup (eds),
Contemporary Debates in Epistemology: 23150. Oxford: Blackwell.
Crowther, T. (2006), Two Conceptions of Conceptualism and Nonconceptualism, Erken-
ntnis, 65: 24576.
Duhau, L. (2009), Conceptuality and Generality: A Criticism of an Argument for Content
Dualism, Crtica, 41: 3963.
Fodor, J. (1990), A Theory of Content and Other Essays. Cambridge, MA: MIT Press.
(1998), Concepts: Where Cognitive Science Went Wrong. Oxford: Oxford University
Press.
(2007), The Revenge of the Given, in J. Cohen and B. McLaughlin (eds),
Contemporary Debates in the Philosophy of Mind: 10516. Oxford: Blackwell.
(2009), Lot 2: The Language of Thought Revisited. Oxford: Oxford University Press.
Heck, R. (2000), Nonconceptual Content and the Space of Reasons, The Philosophical
Review, 109: 483523.
(2007), Are There Different Kinds of Content? in J. Cohen and B. McLaughlin
(eds), Contemporary Debates in the Philosophy of Mind: 11738. Oxford: Blackwell.
McDowell, J. (2009), Avoiding the Myth of the Given, in J. Lingaard (ed.), John McDowell:
Experience, Norm and Nature: 114. Oxford: Blackwell.
Millikan, R. (2000), On Clear and Confused Ideas: An Essay on Substance Concepts. Cam-
bridge: Cambridge University Press.
Schiffer, S. (1992), Belief Ascription, Journal of Philosophy, 89: 499521.
Speaks, J. (2005), Is There a Problem about Nonconceptual Content? The Philosophical
Review, 114: 35998.
Toribio, J. (2008), State versus Content: The Unfair Trial of Perceptual Nonconceptualism,
Erkenntnis, 69: 35161.