Confusion is corruptive belief in false identity
Elmar Unnsteinsson
Postprint. Please cite published version:
Canadian Journal of Philosophy (2016) 46(2): 204–227
doi:10.1080/00455091.2016.1153994
Abstract
Speakers are confused about identity if they mistake one thing for two or two
things for one. I present two plausible models of confusion, the Frege model
and the Millikan model. I show how a prominent objection to Fregean models
fails and argue that confusion consists in having false implicit beliefs involving
the identity relation. Further, I argue that confused identity has characteristic
corruptive eects on singular cognition and on the proper function of singular
terms in linguistic communication.
1 Introduction
What is it to be confused about the identity of an object? People often confuse things.
They think one thing is two or two things are one. Also, ever since Frege argued
for the distinction between sense and reference, this type of mistake has been the
driving force behind a host of inuential arguments and theories in philosophy of
language and mind. Until recently, however, theorists did not pay much attention to
the metaphysics of confused mental states. The point of this paper is to argue for and
present a new metaphysical theory of confusion, called the ‘Frege model.’
In §2 I describe two independently plausible models of confused identity. The rst
is the Frege model already mentioned and the second is called the ‘Millikan model.’ In
§3 I present an argument against the Frege model—called the objection from unavailable
∗
(B) elmar.geir@gmail.com
∗
I would like to thank Hrafn Ásgeirsson, Daniel Deasy, Michael Devitt, Cressida Gaukroger, Daniel
Harris, Rachel McKinney, Eliot Michaelson, Stephen Neale, Gary Ostertag, Ben Phillips, David Plunkett,
Jesse Prinz, François Recanati, and David Rosenthal for comments and suggestions. I also want to thank
the responsible editor and anonymous reviewers for this journal.
Two models of the mental state of confusion 2
representation—and show in some detail why it fails. In addition, I argue that the model
is superior both in terms of descriptive and explanatory adequacy.
Interestingly, however, many of the insights of the Millikan model can, and should,
be incorporated into our nal theory of confused identity. As Ruth Millikan has argued
extensively, confused identity is a type of corruption of the proper function of our
capacity to reidentify particulars. In §4, I explain how Millikan’s notion of proper
functions can be applied to the case of confused identity, rst in thought and then in
communication. Particularly, I show how confusion essentially disrupts the proper
function of singular terms in natural language, even if the Frege model is correct.
2 Two models of the mental state of confusion
Confused identity is a mental state of two basic types. Combinatory confusion is when
an agent takes two things to be a single thing. Separatory confusion is when an agent
takes a single thing to be two things. In principle an agent can be confused by taking
n things to be m things for any distinct natural numbers n and m. For convenience,
I focus only on the more simple cases.
1
A dog who holds a bone in his mouth and
takes its reection in a pond to be a dierent bone suers, albeit momentarily, from
separatory confusion. Another dog who chases two similar rabbits, only ever seeing
one of them at any given moment, might suer from combinatory confusion. If the
rabbits switch roles behind a rock, appearing to the dog as if a single running rabbit
disappears from sight only for a moment, he seems to have some kind of mental
representation that is supposed to help him track a single rabbit but really tracks two
rabbits.
2.1 Frege vs. Millikan
How exactly should confusions like those attributed to the two dogs be explained and
modeled? There seem to be two plausible ways to go here (cf. Lawlor 2005).
1
Note that this is not supposed to be an analysis of the meaning of the word ‘confusion’ in English.
The characterization is an attempt to describe what confusion itself consists in. It seems like the word
‘confusion’ is partly negative and is not properly applied, in English, unless the agent is taken to bear
some responsibility for their sorry epistemic state. This would arbitrarily disqualify many cases that a
general theory of confusion ought to capture. Further, theorists who have written specically about
confusion tend to think only of combinatory confusion (cf. Camp 2002; Lawlor 2007). This may simply
reect common beliefs about the etymology and morphology of the lexical item ‘confusion’ in English.
Translations into other languages do not always give rise to the same tendency (German ‘Verwirrung,’
Icelandic ‘ruglingur,’ etc.). But of course there is a strong tradition of discussing separatory confusions
under the head of Frege’s puzzle.
2.1 Frege vs. Millikan 3
1.
Frege model. According to this model confusion is a mental state that consists in
someone’s having identity beliefs that are false. It is on this basis that Fregeans
have postulated modes of presentation or senses as semantic values for linguistic
or mental representations. A mode of presentation can be thought of as an
identifying condition that is specied by a description. Most famously, Frege
proposed that someone might think of Venus via the identifying condition of
being the morning star or of being the evening star. And it is not necessary that
they realize that these conditions determine the very same object. Modes of
presentation, however, are an optional feature of the Frege model. It is ‘Fregean’
only because of its role in some of the most inuential arguments for postulating
such modes. But let us state the model more precisely.
Frege model of confusion
Speaker A suers from confused identity at time t i there are singular
terms a and b and propositional attitudes V and W that A explicitly holds
at t, such that (i) A V s that Fa, (ii) A W s that Gb, and only one of (iii–iv)
is true; (iii) a
=
b and A believes that a
6=
b (or lacks the belief that a
=
b),
(iv) a 6= b and A believes that a = b (or lacks the belief that a 6= b).
2
‘F’ and ‘G’ are arbitrary predicate expressions of the speaker’s language, stand-
ing for any two properties A can cognitively represent. They can, for example, be
identical or mutually exclusive. Similarly, V and W can be, for example, the same
(e.g. belief), completely dierent (e.g. hope and fear respectively), or mutually
exclusive (e.g. belief and disbelief respectively). Finally, the denition employs
the terminology of explicit propositional attitudes ‘at a time.’ What this means
is just that there is some specic time span in which A has two explicit mental
states whose contents represent a and b. In many cases, the two states can be
collapsed into a single mental state whose content represents a relation holding
between a and b, i.e. (i–ii) can be: A V s that Rab. But the terminology is used
here, rstly, to exclude cases where the mental states occur at two very distant
points in time and, secondly, to contrast them with the identity beliefs (iii–iv)
that A may have implicitly.
2
To be clear, the denition is intended in such a manner that A V s or W s the proposition associated
with the expression ‘that Fa.’ That is, A does not stand in propositional attitude relations to particular
sentence types or tokens containing singular terms a and b, but only to the propositions, whatever
they turn out to be, associated, whatever that relation really is, with the sentences. Also, by ‘singular
term’ I mean any expression that the speaker intends to stand for an object, it need not be a referring
expression; it can be a proper name, demonstrative, indexical or a denite description used referentially.
See also §4.1 below. Note, nally, that, in the terminology introduced in the next section, the Frege
model is couched entirely in the t-language, not in the c-language.
2.1 Frege vs. Millikan 4
2.
Millikan model. Ruth Millikan (1994, 1997, 2000) develops a subtle theory of
confused thoughts. On her view, confused identity cannot consist in having
a false belief about an identity, rather, it is an error of its own kind (1994: 96,
2000: 173). In the ‘central cases,’ misidentifying the object of one’s thought is ‘an
act that muddies the thought involved, corrupting the inner representational
system’ (1994: 75). This is taken to mean that confused representations involve
concepts that are themselves confused. Here is how I propose, in more detail,
to esh out the Millikan model. To be clear, the formulation is inuenced by
Millikan, but it is not intended as a perfect exposition of her broader view. I
want to focus on the model in its own right. A confused singular representation
is one where the number of objects it actually applies to diers from the number
of objects it is, in eect, taken to apply to by its user. Consider the dog chasing
two rabbits. He, let us suppose, possesses a singular representation or quasi-
representation C
R
such that it actually applies to a spatially discontinuous object
that consists of two individual rabbits, Rabbit A and Rabbit B. The representation
C
R
is corrupt, however, since its proper function is to apply to a single spatially
continuous object and, one may add, the dog seems to quasi-believe that C
R
really performs this function.
3
Now consider separatory confusion. Suppose that the dog carrying the bone
in his mouth has a singular representation or quasi-representation C
B
of that
particular bone. On the Millikan model, as soon as the dog confuses the reec-
tion of the C
B
-bone in the pond for some dierent bone, his tokening of the
C
B
-representation applies to nothing. It applies to nothing because there is no
object which consists of that-bone-in-the-mouth without consisting of that-pond-
reected-bone at the same time. So the C
B
-representation actually applies to no
object, although its proper function is to apply to a single spatially continuous
object.
It should be noted here, in particular, that the Frege model implies that cases that
David Kaplan (1968, 2013) and others have described in terms of suspended judgment
are in fact cases of confusion. This may seem implausible but the following argument
indicates otherwise.
Here is a typical scenario in which a speaker (call her ‘Lois Lane #1’) satises
the Fregean denition of confusion. Lois #1 believes that Superman can y. She also
believes that Clark Kent cannot y. As a matter of fact (assuming this is fact rather than
ction), Clark Kent is Superman. And, nally, Lois #1 does not believe that Clark Kent
is Superman. Now consider a case that is less typical (relative to the massive literature
3
The notion of ‘proper function’ is explained in §4 below.
2.1 Frege vs. Millikan 5
on paradigm confusion cases). Another Lois—‘Lois #2’—is likewise confused, on the
proposed denition, if she hopes that Superman saves her and wonders whether Clark
Kent has read Shakespeare while not believing that Superman is Clark Kent, possibly
even suspending belief.
Many philosophers have argued that someone like Lois #1 is confused but not
irrational. On one account of directly referential singular terms, however, it would
follow that Lois #1 is being irrational, because she has inconsistent beliefs about Su-
perman/Clark Kent. Let us call this account ‘Millianism.’ According to Millianism the
content of a belief expressed by referring directly to object o will have an o-dependent
truth condition.
What about Lois #2? On my denition she is also confused about the identity of
Superman/Clark Kent. Here is my reason for taking this view. At time t, let us assume,
Lois #2 tokens the hope that Superman saves her and the wondering-whether Clark
Kent reads Shakespeare. But it is not the case at t that Lois #2 believes explicitly that
(1) λxh
she tokened the hope that x saves her and she tokened the wondering-
whether x reads Shakespeareio.
Where o
=
Superman
=
Clark Kent. Still, this very belief can be derived from the set
of her beliefs at t and is then, arguably, one type of implicit attitude. The derivation
takes a very strong form of Millianism for granted and, plausibly, assumes that con-
scious, reective individuals can reliably form true higher-order beliefs about their
own propositional attitudes. Using only variables, at t, Lois #2 tokens a V-ing of Fa and
a W -ing of Gb. Assuming that she knows some symbolic logic, she can then always
derive the proposition that, at t, she believes that
λ
x[
λ
y
h
she tokened a V -ing of Fx
and a W -ing of Gy
i
a]b. And since in this case it is true that a
=
b the imagined Millian
theorist must predict that there is only a single object o to which Lois #2 refers in
deriving the
λ
-sentence. Thus, given these assumptions, the logician Lois #2 must
derivatively and implicitly believe (1) or something very much like it.
Lois #2 does not explicitly believe the negation of (1) at t, yet this can also be
derived from her beliefs at t. To make the point briey, if one were to put (1) to her
in the form of a question, she would respond negatively if she responds honestly and
in accordance with her actual beliefs. Here is the question I have in mind: ‘Do you
believe that having the property of being an x such that you have a token hope that
x saves you and you have a token wonder whether x reads Shakespeare applies to
a single object o, o being either the object to which you refer by ‘Superman’ or the
object to which you refer by ‘Clark Kent’?’ Again, assuming that she knows some
logic, Lois #2 ought to understand such contorted questions perfectly. It seems, then,
that Lois #2 implicitly believes a contradiction.