<< . .

( 29)

. . >>

same referential symbol is encountered again? The reaction depends,
rather, on the context in which the representation is found, reactions
being, paradigmatically, to pairs of strings, which the reader combines to
yield a third. Such combinings invariably require an overlap in the two
strings, a “middle term.” The middle term has to be duplicated in the
two premises for a rule of mediate inference to apply.
Nor do we need the image of a language of thought in order to
grasp the role that reidentifying plays in amplifying information. Imag-
ine a creature that carries mental maps of various places it has been
about in its head. It has a map of the locale in which it last found wa-
ter, and another of the locale in which it last saw lions. On each of these
maps its den is marked. Now imagine that it overlaps these maps, using
its den as a pivot, and arrives at a third map showing the proximity of
lions to the source of water. Guided by this new map, it seeks a new
source of water rather than going back to the lion-infested source on its
map. As is characteristic of all mediate inference, two vehicles of infor-
mation have been combined, pivoting on a middle term, an overlap, so
as to produce a third vehicle containing new information. Thus our
creature exhibits a grasp of the sameness in content of the two repre-
sentations of his den that were on the two original maps.
More basic even than the involvement of identifying in theoretical in-
ference is its involvement in practical inference, action, and learning. It is

only through recognizing the identity of an item currently perceived
with an item known or perceived earlier that what was learned earlier
can be joined with what is perceived now to yield informed action. Sup-
pose that I wish to congratulate A on his engagement and that I see that
B is in the lounge talking to C.This seeing will be of no use to me un-
less I grasp whether A = B or A = C. Consider learning. Suppose baby
has noticed that A scolded her when she cried but that B, C, and D
kindly picked her up.Whether she learns anything from this will depend
on which if any of these four she takes to be the same person again.
Returning to Evans™ speculations on Molyneux™s question and be-
havioral space (Section 9.3), it is not a person™s ability to be motorically
guided in the same way by perceptions from different sensory modalities
that would manifest grasp of sameness of content represented through
these modalities. Rather, such a grasp would be manifested in the abil-
ity to combine information obtained through these different modalities
to yield behavior or thought guided by both put together. Or, taking a
different sort of example, consider a duckling that has imprinted on its
mother. The result of imprinting is that whenever the duckling sees its
mother, a certain set of behavioral dispositions emerges. The duckling
has stored away a “template” matching its mother™s appearance so as to
“recognize” her. Despite our natural use of the term “recognize” in this
context, it does not follow (though it may of course happen to be true)
that the duckling reidentifies its mother (for example, that it has a sub-
stance concept of her). Only in so far as the duckling is capable of
learning things about its mother on some encounters to apply on other
encounters with her does it identify her. Just reacting the same to her
time after time does not indicate identification.
Every mediate inference, every recognition of a contradiction, every-
thing learned either from perception or inference and applied in action,
every belief or behavior issuing from coordination among sensory
modalities, for example, eye-hand coordination, even such subpersonal
activities as the use of images from two eyes in depth perception, de-
pends upon recognition of content sameness. Grasp of identity is the
pivot on which every exercise of perception and thought must turn that
collects together different pieces of information from different percep-
tual modalities, or from different contexts, or over time, and effects its
interaction. Every act of identifying is thus implicitly an act of reidenti-
fying, consisting in the use of two or more representations or pieces of
information together. Described on the level of content, on the level of

the visagings or believings-that involved, we call these “acts of identi-
fying” or “reidentifying.” Described on the level of the vehicles or men-
tal bearers of information involved, we can call them acts of “coidenti-
fying.” In an act of coidentifying, two representational vehicles are
employed together in a manner that assumes, that is, requires for cor-
rectness, an overlap or partial identity in content, thus effecting an act
of reidentifying of content.


The thesis of Section 10.2 can be put as follows:
For a perceiver or cognizer to reidentify something JUST IS to be disposed, or
for some subsystem of theirs to be disposed, to pair representations of that thing
in perception and/or thought as a middle term for mediate inference, or other
amplificatory information-processing, and/or for guiding action.

That will do for a first pass over the phenomenon of recognizing
A second pass must take into account that where valid mediate in-
ferences are made, or correct content-sameness pairings or groupings
are made for other information-using purposes, this result must follow
from some kind of indication in the initial or “premise” representations
of where sameness of reference is occurring. It must result from a sys-
tem or systems of sameness marking in perception and thought “ per-
haps using Strawson-style markers, and/or duplicates markers, and/or
Christmas light markers, and so forth.
What makes a marker a sameness marker is that the perceptual/cognitive sys-
tems use it to control the mediate inferences and other content pairings that
they make in guiding amplificatory information-processing and action.

Derivatively, then, the mere occurrence of an appropriate sameness
marker connecting two perceptions or thoughts can count as an “un-
derstanding” that the marked representations are representations of the
same. It is, as it were, a “first act understanding of sameness,” where a
“second act understanding” is an actual process of mediate inference,
amplificatory information processing, or action guidance controlled by
these markers. First act identifications, sameness markings, prepare for
second act identifications.
Suppose that in the case of thoughts of substances, we were to take
first act identifications to be “applications of substance concepts.” That

is, “applying a substance concept” would be marking incoming infor-
mation in such a way that its bearers will be ready for coidentification
with certain other information bearers. “Applying a substance concept”
will be readying bearers of incoming information for interaction with a
restricted set of other information bearers “ those bearing information
about the same substance. Then there will be, after all, a sense in which
“applying the same substance concept” counts as an act of identifying
or recognizing sameness. But this sense of “substance concept” will be
that in which the abilities that are substance concepts are counted or in-
dividuated by their ends, not their means. Substance concepts must be
scrupulously distinguished from conceptions of substances (Sections 1.9,
4.8, and 6.3) in this context.
A third pass over the question what it is for content sameness to be
recognized in thought should take error into account. Under unfavor-
able conditions, even simple perceptual identification tasks can be mis-
managed. For example, there is a way of crossing your fingers so that
the identity markings that bridge between tactile and visual percepts
become mixed.The finger one sees being touched does not seem to be
the finger one feels being touched.When looking through a stereoscope
your visual systems misidentify portions of two pictures as portions of
the same, thus producing the illusion that you are looking at a three di-
mensional scene. The skill of sleight of hand artists depends largely on
their ability to fool your visual systems into failing to track objects cor-
rectly, thus inducing perceptual misidentifications.
Such misidentifications do not occur commonly, and may require
specially designed apparatuses or other circumstances of perception to
induce them. Conceptual responses to the data of sense, on the other
hand, are more tenuously correlated with affairs in the world than are
perceptual responses. Failure to mark sameness correctly in thought is
quite common. We often fail to recognize a thing, or we confuse two
things together, say, mistaking Jim for Bill or failing to distinguish be-
tween mass and weight. Consider, then, a mediate inference that is made
over two premises containing information in fact derived from different
sources. The premises do not carry information concerning the same
thing, and as a result, let us suppose, the conclusion arrived at is false.
Should such an erroneous move count as a mistake in inference? Or
should it count merely as a mistake in data collection and labeling?
Which internal moves should count as valid inferences would seem
to depend on how sameness of origin is marked during data collection.
But how sameness of origin should be marked during data collection

surely depends on what sameness markers the inferencing systems will
recognize. There will be nothing wrong, for example, with representing
two different objects with identical representations so long as duplica-
tion is not the identity marker. Does it follow that which structures re-
ally are the sameness markers is well defined only for a system that
never makes mistakes?
This kind of problem is classic, of course, for theories of naturalized
thought-content. These theories typically take cognitive abilities to be
some kind of dispositions (Chapter 4), or to rest on ceteris paribus laws.
The problem is then taken to concern “idealization.” How far away
from a certain ideal can a system™s actual practice or actual dispositions
be while still counting as an example of a given ideal type? What do we
say about content when the system hovers indeterminately between or
among alternative ideal types?
My own preference is to refer instead to evolutionary design on this
sort of question (Millikan 1984, 1993a, in press b). There will be ways
that our perceptual-cognitive systems worked when they operated such
as to be selected for by natural selection. There will be a way or ways,
that is, that they were “designed” to mark and to recognize sameness.
With enough knowledge of the internal mechanisms controlling cogni-
tion, what these normal ways are should be no harder for us to distin-
guish than, say, how the human eye is designed to work, even though
many human eyes function poorly. The distinction between having
gone wrong in collecting the data and having gone wrong in inference
may then be a perfectly objective distinction.
Thus there is room to distinguish two kinds of error, either of which
might be called an error of misidentification. There could be error in
performing a first act identification, that is, an error in the labeling of
incoming data. Or there could be an error in second act identification,
an error in mediate inference, or an analogue, of the general sort tradi-
tionally labeled “the fallacy of the fourth term.” The first would be an
error in the fixation of belief, the second an error in inference or an
analogue of inference. I will not try very hard to keep these possibilities
distinct in the chapters that follow, though occasionally it will be help-
ful to recall their difference.

In Search of Strawsonian Modes of

§11.1 THE PLAN

There are many alternative ways that a mind or brain might represent that
two of its representations were of the same object or property “ the
“Strawson” model, the “duplicates” model, the “equals sign” model, the
“synchrony” model, the “Christmas lights” model, the “anaphor” model,
and so forth (Section 10.1). In the last chapter I discussed what would con-
stitute that a mind or brain was using one of these systems rather than an-
other in order to mark identity. In this chapter, I discuss the devastating
impact of the Strawson model of identity marking on the notion that
there are such things as modes of presentation in thought. I will then ar-
gue that Evans™ idea that there are “dynamic Fregean thoughts” has exactly
the same implications as the Strawson model. In Chapter 12, I will claim
that, in fact, all of the other models of identity marking we have discussed
are strictly isomorphic to the Strawson model, hence have exactly the
same devastating results for modes of presentation. There is no principled
way to individuate modes of presentation such as to achieve any sem-
blance of the set of effects for the sake of which Frege introduced them.


Suppose that our minds/brains used Strawson markers for marking
identity. Keeping clearly in mind that the project here is neither exege-
sis of Strawson™s text nor exegesis of Frege™s, let us ask what, on this

1 Portions of this chapter are revised from “Images of Identity” (Millikan 1997b) with the
kind permission of Oxford University Press.

model, would correspond most closely to the Frege-inspired notion that
the same object can be thought of by a thinker under various different
“modes of presentation.”
Gareth Evans tells us that different modes of presentation are, just,
different ways of thinking of an object (e.g., Evans 1982, Section 1.4).
Suppose that we take this statement completely naively. On a Strawson
model it appears that, so long as we always recognized when we were
receiving information about the same object again, each of us would
end up having only one way of thinking about each object. No matter
what attributes the Strawson-style cognitive system thinks of an object
as having, as long as it does not fail in the task of reidentifying, it always
thinks of the object the same way, with the same dot. Two modes of
presentation of the same might occur, for example, as the system col-
lected information about a person seen in the distance prior to recog-
nizing them, or about a person being discussed by gossipers before find-
ing out about whom they were talking. But this sort of situation is
usually temporary. Either the person seen or discussed is soon identified,
or the information collected about the unknown person is easily for-
gotten. For example, we do not usually retain memories of people we
pass on the street unless we recognize them. On this model, it would
usually be so that all your beliefs concerning the same object were be-
liefs entertained under precisely the same mode of presentation.
On this naive reading of “modes of presentation,” moreover, no two
people could think of an object under the same mode of presentation.
To do so they would have to have numerically the same dot in their
heads! On a Strawson model, there is no kind of similarity between two
minds, either in internal features or in external relations, that would
constitute their thinking of the same “in the same way.”There might be
relevant similarities between the ways you and I think of a thing, con-
ceivably we might even have exactly the same beliefs about a thing, as-
sociate with it all the same identifying descriptions and so forth. But on
this interpretation this would not bring us any closer to thinking of it
under the same mode of presentation.
Interpreted this way, “modes of presentation” obviously would bear
scant resemblance to Fregean senses, the very first job of which was to
correspond to shared meanings of words and sentences in public lan-
guages. For example, Frege supposed that the very same senses are
grasped first by the speaker and then the hearer when communication
is effected through language. Also, on the Strawson model the different
identifying descriptions that you attach to the dot representing a given

man are not different ways of thinking of him, but merely various
things you know about him, some of which might sometime come in
handy in helping to reidentify him as the source of some incoming in-
formation. Correspondingly, the differences between various kinds of
referring expressions “ descriptions versus proper names versus indexicals
“ would not parallel differences between various kinds of thoughts. On
this model there are, for example, no indexical thoughts or ideas, al-
though there would, of course, be times when the thinker used percep-
tual tracking abilities to collect various bits of incoming information to-
gether next to the same dot in his head.
And, of course, sentences expressing nontrivial identities could not
be analyzed Frege™s way on the naive Strawson model. Accordingly,
Strawson™s description of the semantics of identity sentences (Strawson
1974) differed radically from Frege™s. The public meaning of the iden-
tity sentence does not correspond to a particular sharable thought. It
concerns what the sentence conventionally does to hearers™ heads. What
it does is not to impart information but to change the mental vocabu-
lary, altering the mental representational system. As such, its function is
different, in one important sense, for every hearer. Both the affected
dots and, barring weird coincidences, the information in the structures
attached to these dots, will be different for each hearer.
Perhaps most critical of all, on this model, should the thinker make a
mistake in identifying, the result will be the creation of an equivocal
mode of presentation, one that has two referents at once. Nor will the
subject who grasps the equivocal mode of presentation have direct ac-
cess to this flaw. Suppose that you are confused about the identity of
Tweedledum, having mixed him up with Tweedledee, so that whenever
you meet either you store the information gathered next to the very
same dot.Which man does this dot represent? Which man is it that you
misidentify, thinking he is the other? Rather, the dot must stand for
Tweedledumdee, an amalgam of the two. Further, if systematic misiden-
tifications occurred, or if misidentifications were frequent and random,
it seems that a dot™s reference might focus on no object at all, hence
reasonably be considered quite empty. Similarly, it seems that a dot
might undergo massive yet invisible shifts in reference. Consider, in this
light, the suggestions offered on the epistemology of substance concepts
in Chapter 7.
An interesting corollary would be that negative identity sentences
have no determinate meaning, not even for individual persons. For ex-
ample, on this model you have no separate ideas Cicero and Tully, nor

even the man called “Cicero” and the man called “Tully.” Your way of
thinking of the referent of each of the four corresponding public terms
merges them irretrievably together. They are different for you only in
that they are recognizably different packages in which information
about the same thing can enter when you are among speakers speaking
a language you know. Suppose then that a historian now informs you
that there has been, in fact, an unaccountable confusion among philoso-
phers and that Cicero was not in fact Tully. How are you to understand
this negative identity claim? What you™ve got in your head is one dot,
attached to which is a variety of (presumed) information, including the
information . . . is called “Cicero” and . . . is called “Tully.” But how
will you divide the rest of the information into two piles? This could
only be accomplished through a major job of reconstruction, as you
tried to remember or to guess how you had acquired each separate bit
of information, hence from which of these separate men it was most
likely to have originated.
It is hard to imagine anything further from Frege™s intention than
these various results. What has gone wrong? Later (Section 12.4“6), I
will tease apart several strands that are woven together to produce the
peculiarities of this “naive” Strawson-inspired image of “modes of pre-
sentation,” and I will try to articulate the underlying principles that di-
vide it from Frege™s own vision. But there is also another interpretation
possible of what modes of presentation might be for a mind that used
Strawson markers.


In our “naive” image above, Strawson™s dots are taken to be modes of
presentation because they are “ways of thinking of things,” a phrase
most easily interpreted in this context to mean kinds of mental repre-
sentations of things. In interpreting modes of presentation this way, we
parted from a very important strand within, anyway, the contemporary
neo-Fregean tradition. Gareth Evans, for example equates the way one
is thinking about an object with the way in which the object is identi-
fied (Evans 1982, p. 82, McDowell™s formulation for Evans). Similarly,2
Dummett takes Fregean sense to be a method or procedure for deter-

2 Similarly enough that is. Evans is at pains to distinguish his views from Dummett™s here, but
not in ways that affect what is at issue for us.

mining a Bedeutung, paradigmatically, for determining the presence of
the Bedeutung (e.g., Dummett 1973, pp. 95 ff). Evans and Dummett
agree, for example, that grasp of a particular way of recognizing a refer-
ent encountered in perception corresponds to a mode of presentation
of the referent. Now the Strawson image of sameness marking seems to
pry apart the way one thinks of a thing from the various ways one
knows to recognize it. Perhaps, then, if we identify modes of presenta-
tion with the latter instead of the former we will find them to be more
as Frege intended.
Suppose then we take modes of presentation, on the Strawson
model, to be not ways of thinking about a thing but ways of identifying
it, in particular, ways that a thinker knows to recognize incoming infor-
mation, arriving via perception, language (Chapter 7), or inference, as
being about that thing.That is, given the terminology developed in ear-
lier chapters of this book, we take modes of presentation to correspond
to various aspects of the conception that a person has of an object, rather
than to the concept itself. Modes of presentation will thus describe
people™s conceptual reidentifying abilities by their means rather than by
their ends (Section 4.5).
On this reading, it seems that a person might grasp not just one but
many modes of presentation for a given object. Also, perhaps different
people might grasp the same mode of presentation, for they might be
able to recognize the object in the same way. Moreover, suppose that
understanding a certain sort of linguistic expression as referring to an
object is, just, grasping a particular way to reidentify the object, through
its manifestations in the speech of others (Section 6.2) and/or, in the
case of definite descriptions, through prior identification of certain of its
properties. Then it might seem that some modes of presentation would
correspond directly to meanings of referring expressions. Certainly
many philosophers have supposed something like this to be true. The
results look much better at first than on the “naive” interpretation of
Strawson-model modes of presentation.
But trouble is not far away. On the Strawson model, the terms in the
various beliefs that a person has will not then be characterized by deter-

<< . .

( 29)

. . >>

Copyright Design by: Sunlight webdesign