Sunteți pe pagina 1din 18

This article was downloaded by: [Arizona State University]

On: 27 October 2014, At: 14:03


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

Australasian Journal of
Philosophy
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/rajp20

The teleological theory of


content
a b a b
David Braddon-Mitchell & Frank Jackson
a
University of Auckland
b
Australian National University
Published online: 02 Jun 2006.

To cite this article: David Braddon-Mitchell & Frank Jackson (1997) The
teleological theory of content, Australasian Journal of Philosophy, 75:4, 474-489,
DOI: 10.1080/00048409712348051

To link to this article: http://dx.doi.org/10.1080/00048409712348051

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all
the information (the “Content”) contained in the publications on our
platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy,
completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of
the authors, and are not the views of or endorsed by Taylor & Francis.
The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and
Francis shall not be liable for any losses, actions, claims, proceedings,
demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in
relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/terms-and-
conditions
Downloaded by [Arizona State University] at 14:03 27 October 2014
Australasian Journal of Philosophy
Vol. 75, No. 4; December 1997

THE TELEOLOGICAL THEORY OF CONTENT

David Braddon-Mitchell and Frank Jackson

The teleological theory of content is an attractive package. The first part is the plausible
contention that we are a biological part of nature - biology is the science that studies us
qua human beings. The second part is a plausible analysis: the aetiological cure evolu-
tionary analysis of function in biology - roughly, a biological state's function is what it
was selected for; or, in Ruth Millikan's terminology, function in biology is proper func-
Downloaded by [Arizona State University] at 14:03 27 October 2014

tion.~ These parts get combined with a thesis about intentional states, namely, that the
right approach to their content is via the recognition that they have a function in the
sense of a telos: if you want to find the content of an intentional state, look to what it is
for. But intentional states are biological states, so, from the thesis that function in biolo-
gy is proper function, we end up with the teleological theory of content (or teleonomy, as
we will sometimes say for short), the thesis that an intentional state's content is given by
its proper function.
Here is how the story often runs in the case of belief. Belief is a state whose function
is to represent how things are. This means that the content of a belief is the particular
way things are which the belief has the function of co-varying with. But, then, by the
aetiological analysis of function in biology, it follows that the content of a belief is the
way things are or the state of affairs with which the belief was selected to co-vary. For
example, belief that there is water nearby is, roughly, the state whose presence in us is
explained by the way its induction by nearby water was selectionally advantageous, and
it gets to have the content that there is water nearby because of this. In similar fashion,
starting this time from the premise that the function of desires is to represent how things
should be, we reach the conclusion that the content of a desire is the state of affairs it
was selected to make obtain. Thus, roughly, desire to be near water is the state whose
presence in us is explained by the way its tendency to make us be near water was selec-
tionally advantageous, and this is how it comes to have the content it does have.
Which aspect of selectional history determines content varies from one version of
teleonomy to another. Famously, certain frogs respond equally to flies and to any small,
dark, moving things. This is because, although it was important to the frogs' survival that
they detected flies, the vast majority of small, dark moving things that they came across
were in fact flies, and so it did not matter that their fly detectors could not tell the differ-
ence. On some teleological theories, this means that their fly detectors, strictly speaking,
only represent that there is a small, dark, moving thing within tongue range. 2 On most
teleological theories, though, their fly detectors really do represent that there is a fly

t Ruth Millikan, Language, Thought and Other Biological Categories (Cambridge, MA: MIT
Press, 1984).
See, e.g., Karen Neander, 'Misrepresenting and Malfunctioning', Philosophical Studies 79
(1995) pp.109-141.

474
David Braddon-Mitchell and Frank Jackson 475

within tongue range, because it mattered for survival that what got ingested were flies.
Again, teleological theories differ over whether it is evolutionary history alone, or
whether (as in 'neural Darwinism') quite recent selectional history also matters for con-
tent, and hence over what to say about beliefs and desires concerning kinds of things that
were not around when we were shaped by evolution. Again, some teleological theories
focus on the relational proper function of the mechanism that gives rise to a state rather
than on the state's own proper function: a state is about such and such if it is produced by
a mechanism whose proper function is to produce that state when such and such.
We will abstract from these important details. Our initial concern is with the status of
the view that understands content in terms of selectional history. Are teleonomists offer-
ing us a conceptual analysis, or are they making a claim about essential properties, or are
Downloaded by [Arizona State University] at 14:03 27 October 2014

they putting forward a scientific identification or theoretical reduction of intentional con-


tent? In the first half of this paper, we argue that they must be doing the third. In the
second half, we argue, on the basis of this understanding of their view, that it faces two
serious objections. Our arguments in both parts are independent of the details about
which aspects of selectional history have the claimed special relationship to content; our
focus will be on various views about what that special relationship might be and the
problems attendant on these views. (Also, we will frame matters in terms of teleonomies
that focus on the selectional histories of the contentful states, but the same points apply
mutatis mutandis to versions that focus on the relational proper functions of the mecha-
nisms that produce the states.)
Further, we will collapse the distinction between selectional views and teleological
views. Strictly, it is biological function that teleonomy holds to be crucial for the deter-
mination of content, and selectional history enters the picture only inasmuch as a
selectional-cum-aetiological account of biological function is correct; but the most inter-
esting and discussed versions of the teleological account of content assume, or argue for
in an earlier chapter or paper or book, a selectional-cum-aetiological account of function
in biology - the function as proper function account - and we will be concerned only
with this style of teleonomy.

It is not in dispute that there is a close, non-accidental connection between content and
selectional history. The belief that there is water nearby is important for survival in part
because of its connection with their being water nearby. This means that, in light of the
overwhelming evidence in favour of the theory of evolution, we can be confident that
this fact played a major role in our tending to have the belief that there is water nearby
when appropriate. Likewise, the desire for food is important for survival because of its
connection with the consumption of food; so we can be confident, on evolutionary
grounds, that this connection played a major role in our sometimes having the desire for
food. In short, it is very plausible that, in very many cases, the states of affairs that
beliefs and desires are about, their contents, played an important role in our having the
beliefs and desires in the first place.
David Papineau expresses his version of teleonomy by saying that, roughly, the belief
that p is the state selected to co-vary with p, and the desire that p is the state selected to
476 The Teleological Theory of Content

bring about p ? This is not merely the claim that we can, on evolutionary grounds, be
confident that, by and large, there is a non-accidental connection between the obtaining
of p and having the belief that p, and there is a non-accidental connection between the
bringing about o f p and having the desire that p : Teleonomy only becomes an interesting
theory when it goes beyond this common ground observation. But how might it do so?
One answer is to treat it as a matter of conceptual analysis that the content of beliefs
and desires is determined by selectional history (whether in the way Papineau favours or
notS). Now, most teleonomists insist that they are not advancing a conceptual analysis.
Some, Millikan notably, do this as part of a general scepticism about the project of offer-
ing c o n c e p t u a l a n a l y s e s - ' " c o n c e p t u a l a n a l y s i s " . . . is a c o n f u s e d p r o g r a m , a
philosophical chimera, a squaring of the c i r c l e . . . ,.6 Others, Karen Neander, for exam-
Downloaded by [Arizona State University] at 14:03 27 October 2014

ple, grant the propriety of conceptual analysis - indeed see it as having an important
place in the overall case for teleonomy - while insisting that the claim that content is
given by selectional history is not itself a piece of conceptual analysis. 7
We think conceptual analysis is essential to much work in philosophy, but agree that
teleonomists had better not be offering their selectional account of content as pure con-
ceptual analysis. Despite the agreement, we need to argue the matter, as the points that
come up will be important later.
To offer a conceptual analysis of K-hood is, as we understand it, to offer an illuminat-
ing account of what, implicitly or explicitly, guides those who understand the term 'K' in
classifying something as a K. The role of the reflections on possible cases associated
with conceptual analysis is to help articulate what is guiding us in our classificatory prac-
tice. The famous illustration of this role is the debate over the analysis of knowledge. It
is tempting to think that the term 'knowledge' is a short expression covering cases of
true justified belief; however, by describing possible cases of true justified belief which
are true by accideilt but justified all the same, Edmund Gettier convinced us that classify-
ing by knowledge is not classifying by true justified belief? This account of conceptual
analysis may make it sound a purely descriptive exercise - it is simply a matter of find-
ing the best description o f what in fact guides our classifications - but there is a
normative dimension to conceptual analysis. The terms and concepts whose conceptual
analysis is worth taking time over - like the concepts of free action, content, and knowl:

3 David Papineau, Philosophical Naturalism (Oxford: Blackwell, 1993), p.94.


4 We talk here of 'by and large, non-accidental connections' rather than typical ones in order to
side-step the important, but here irrelevant, debates over whether and to what extent evolution-
ary considerations support our beliefs being typically true and our desires being typically
satisfied.
5 Some supporters of teleonomy favour a more complex account that sees the contents of parts of
intentional states - the words of mentalese - as determined by selectional history, and then
accounts for the contents of the intentional-states themselves compositionally. As far as we can
see, the added complexity of the LOT approach does not buy a reply to the objections we raise
in the second half of this paper.
6 Ruth Millikan, White Queen Psychology and Other Essays for Alice (Cambridge, /vIA: MIT
Press, 1993), p.15.
7 According to Neander, the place of conceptual analysis in the overall story is much as we
sketched at the beginning. Puzzlingly, Millikan, White Queen Psychology and Other Essays,
p. 15, fn. 1, remarks that Neander offers 'a brilliant defence of the "etiological" account of func-
tion while remaining within the tradition of conceptual analysis'. We are not sure how one
pursues 'a confused program, a philosophical chimera, a squaring of the circle' brilliantly.
8 Edmund L. Gettier, 'Is Justified True Belief Knowledge?', Analysis 23 (1963) pp.121-123.
David Braddon-Mitchell and Frank Jackson 477

edge, and unlike the concept of a bicycle (who cares much if we decide to count a tricy-
cle as a special sort of bicycle?) - play important roles in our theories. In consequence,
we need to ask of any proffered analysis whether it is fit to play the needed theoretical
role: an analysis of free action should make sense of the connection between acting
freely and being liable for punishment, an analysis of content should make sense of its
role in our theories of mind and language, and so on. This gives us a 'one sentence'
answer to critics of conceptual analysis. How can it be wrong (or confused, or chimeri-
cal, or a squaring of the circle) to seek to make explicit what guides us in the intentional
and considered activity of classifying things as falling under one or another interesting
term or concept, especially when it is informed by the desideratum of making good sense
of the term or concept's theoretical role?
Downloaded by [Arizona State University] at 14:03 27 October 2014

It will now be clear why teleonomy cannot be a conceptual claim. Creationists under-
stand content clauses perfectly well. So did people before the theory of evolution was put
forward and the notion of a selectional history became common currency in biology. It is
absurd to suppose that these people could come up with the theory of evolution simply
by careful enough reflection on when they counted someone as believing that snow is
white or desiring coffee, and, more generally, on what underlies their classifying some-
one as having one or another belief or desire. Their classifying Jones as believing that
snow is white - as one of the things that believe that snow is white, as opposed to one of
the things that do not - has nothing to do with their holding, implicitly or explicitly, that
Jones' belief state is in some way selected for by the whiteness of snow, and yet that is
what would have to be the case if the selectional account were a conceptual truth.

II.

The second suggestion is that we should read teleonomy's distinctive claim as that selec-
tional history is an essential property of contentful states - much as being H 2 0 is an
essential property of water, and having atomic number 79 is an essential property of
gold. According to this reading, it is an a posteriori scientific discovery about the nature
of contentful intentional states that selectional histories of a certain kind are essential
properties of them.
Now, we can think of essential properties in the transworld identity way principally
associated with Saul Kripke, or in the counterpart way principally associated with David
Lewis .9
On the transworld identity way of thinking about essential properties, an essential
property of something is simply a property which that very thing has in every world
where it occurs. Construing essential properties in this way makes life difficult for the
teleonomist. What has the right kind of selectional history are certain physical states in
our brains. '° They are those particular states and structures that evolution has bequeathed

9 See, e.g., Saul Kripke, Naming and Necessity (Oxford: Blackwell, 1980), and David Lewis,
'Counterparts of Persons and Their Bodies', The Journal of Philosophy 68 (1971) pp.203-211.
,0 We talk of physical states rather than neurological or brain states because teleonomists often
insist that for a state to be properly described using the terms 'neurological' or "brain" it must
have a suitable selectional history, and it makes it easier to grasp the point that follows to use a
term we can all agree lacks any implications about any particular selectional history. We take
'physical' to be such a term, but anyone who disagrees can of course plug in their own candidate
to be the needed neutral term.
478 The Teleological Theory of Content

to us in order to help us find our way around and get what we need to survive. And it is
not an essential property of any physical state, S, that it has a certain selectional history:
S might have had a different selectional history from the one it actually has. This is true
even if S's causal origin is one of its essential properties. In this case the different selec-
tional history must start at the same place, but this is consistent with the selectional
history itseff being different. Different selectional histories set in different environments
can take one from a given starting point to the very same S.
This point stands independently of a claim about biological kinds often made by
teleonomists. They argue that in biology we type by proper function: a kidney is that
which has a certain proper function, not that which meets a certain anatomical specifica-
tion (or, for that matter, performs a certain current function)." Nevertheless, each and
Downloaded by [Arizona State University] at 14:03 27 October 2014

every (natural) kidney is a certain anatomical structure, namely, the anatomical structure
with the right proper function, and it is not an essential property of that anatomical struc-
ture that it has the selectional history it in fact has. Of course, on the view about
biological kinds in question, if that structure lacked the right selectional history, it would
not count as a kidney - in which case being a kidney would not be an essential property
of kidneys, just as being Prime Minister is not an essential property of Prime Ministers.
In the same way, what has the selectional histories that are, according to teleonomy, dis-
tinctive of intentional content are certain physical structures, but it is not an essential
property of these structures that they have this history; and this is true consistently with
their having these histories being what matters in biological theory and what makes it
right to describe them as content-bearing states.
So, on the Kripkean way of thinking about essential properties, it is false that any par-
ticular selectional history is an essential property of intentional states. However,
teleonomists sometimes talk of the very same physical structure having different
essences relative to the different 'states' that it can be said to be in: thus, a physical struc-
ture qua neurological state is said to have different essential properties from the same
physical structure qua contentful state. This suggests that if we want to read teleonomy
as a claim about essential properties, we should adopt the alternative, counterpart way of
thinking about essential properties; for, on that approach, something's essential proper-
ties are not an all or nothing matter, but are relative to one or another counterpart
relation.
The counterpart approach is set against a background that denies transworld identity.
According to it (albeit roughly, since the details of various versions of counterpart theory
are irrelevant here), an essential property of x is a property of x that every counterpart of
x has in every world in which it exists, where a counterpart of x in a world is something
very similar to x, not x itself. But what counts as very similar depends on which respects
of similarity are salient; in consequence~ a thing's counterparts can vary depending on
which similarity is salient, or, as it is put, a thing can have different counterparts under
different counterpart relations. This is why, according to counterpart theory, a thing's
essential properties are not an all or nothing matter. For example, the counterparts of
some physical structure, S, which is simultaneously a certain neurological state, a certain
contentful state, and a certain physical configuration, will vary depending on whether we

~t See, e.g., Karen Neander, 'Swampman meets Swampcow', Mind and Language 11 (1996)
pp.118-129, for a recent defence of this view.
David Braddon-Mitchell and Frank Jackson 479

look to what, in other worlds, counts as very similar going by neurological nature, by
content, or by physical structure; in consequence, what counts as S's essential properties
depends on the counterpart relation in play. This complicates the discussion of the read-
ing of teleonomy as the view that selectional history is an essential property of contentful
states. For when teleonomists say that a certain selectional history is an essential proper-
ty of a contentful state, they say, according to the counterpart way of understanding
essential properties, that every counterpart of it shares the certain selectional history. But
which counterpart relation or relations should they be taken as working with?
It would obviously be a mistake to be too liberal and say that any reasonable answer
as to a contentful state C's counterparts in other possible worlds will do. C will, for
example, be a particular physical structure, and so, under the physical-structure counter-
Downloaded by [Arizona State University] at 14:03 27 October 2014

part relation, anything physically exactly like C will be a counterpart of C, but physical
duplicates in other worlds can, and do, have different selectional histories.
The right counterpart relation for teleonomists is, we suggest, the same-content rela-
tion; that is, the best way to read teleonomy as a claim about essential properties is as the
claim that a certain selectional history is an essential property of contentful states under
the same-content counterpart relation. But, to see why this is so, we need to have before
us the scientific reduction reading of teleonomy.

1II.

The final reading of teleonomy is as a scientific reduction or identification of content -


the identification of content with selectional history is of a piece with the identifications
of lightning with an electrical discharge, of water with H 2 0 , and of heat in ideal gases
with molecular kinetic energy. This is the suggestion most often found in the writings of
teleonomists and will be the one discussed critically in the second half of this paper. 12
(Indeed, except when explicit note to the contrary is made, from now on we mean by
'teleonomy' teleonomy so understood.) But first we need to detail briefly this way of
understanding teleonomy.
Although the identification of lightning with an electrical discharge is not a piece of
conceptual analysis, the identification did not, of course, come from nowhere. There was
a case to be made for it that involved observations about what we use the word 'light-
ning' for. (What follows abbreviates a much told story, so we say it quickly.) Roughly,
the case starts from the point that 'lightning' is a word for a phenomenon that plays a
certain role - not in the sense that the word means that which plays the role, but in the
sense that the role does the reference-fixing - and concludes with the point that, as a
matter of fact, what plays the role is an electrical discharge. ~3 This is how the sentence
'Lightning is an electrical discharge' gets its famous necessary a posteriori status. It is a
posteriori because it is a posteriori that it is an electrical discharge that plays the role. It

David Papineau is especially explicit that he is offering a reduction, see Philosophical


Naturalism, p.93.
To put the matter in terms of Kdpke's distinction between giving the meaning and fixing the
reference in Naming and Necessity. You might well regard an account of what reference-fixes as
part of an account of meaning in a wide sense, but what matters here is that there is an important
distinction, not the words used to mark it. We will follow Kripke's usage because of its familiar-
ity.
480 The Teleological Theory of Content

is necessary because playing the role reference-fixes on an electrical discharge, and this
means that 'lightning' refers to an electrical discharge in all worlds (in which it exists,
but we will suppress complications arising from the possibility of non-existence in a
world), regardless of whether or not electrical discharge plays the role in those wodds.
The result is that 'Lightning is an electrical discharge' is true in all worlds.
In the same way, a scientific identification of content with selectional history has, or
should have, two parts. The first says something about what we folk use words like
'believes that snow is white' and 'desires a cup of coffee' for, in the sense of the usages
we need to master to understand these phrases. This is given by what are sometimes
called the folk markers for the phrases; TM or we can talk in terms of reference-fixing, as
we just have in the case of lightning. The second part says that what we reference-fix on,
Downloaded by [Arizona State University] at 14:03 27 October 2014

what the folk markers pick out, are, as it happens, states-with-a-certain-selectional-histo-


ry. This means that what counts as a state with a certain content in some possible world
is the state with right selectional history in that world, regardless of whether or not it sat-
isfies the folk markers in that world. This then delivers an interesting, necessary a
posteriori identification of content states with states with the right selectional history,
which goes well beyond the common ground observation that there is a non-accidental
association between content and selectional history.
How should the folk markers or reference-fixers be characterised? In the cases of
temperature, water, and lightning, there are reasonably non-controversial answers. The
situation is, as so often, more controversial in the case of words for intentional states. We
will offer what seems to us the most promising approach to the characterisation of the
folk markers or reference-fixers for intentional content. As far as we can see, it does not
beg any important questions to come; the points we make later could be made against the
background of any reasonable account of the folk markers.
Language is primarily a convention-generated system of communication. We use it to
tell one another how things are, by means of implicit agreements concerning when to use
various words and sentences. Coming to understand a language is coming to understand
when to use the various terms - 'rouge' is the word to use in French when confronted
with something you take to be red, and so on. Now what did we learn when we came to
understand phrases like 'believes that snow is white' and 'desires a cup of coffee'? As
we observed when rejecting the reading of teleonomy as a piece of conceptual analysis,
we did not learn to use these words for creatures with states having a certain selectional
history. For we learnt something that could be accepted by creationists and those who
have never heard of the theory of evolution. W e learnt, rather, to use these terms for
organisms having states that lead to the display of certain complex interaction patterns
with their environment. For it is the observation of the complex interaction patterns
which prompts in speakers (and writers) claims about what people believe and desire. It
is after learning what people do (and say in the case of creatures with a language) that
the folk come to opinions about what subjects think and want. W e can sum this up by
saying that teleonomists do best to hold that the folk markers or reference-fixers for
intentional states are folk functional roles. ~5

t4 See, e.g., Michael Devitt, Coming to Our Senses: a naturalistic program for semantic localism
(New York: Cambridge University Press, 1996).
,5 And Papineau does exactly this; see his response to Andrew Woodfield on pp.93-94 of
Philosophical Naturalism.
David Braddon-Mitchell and Frank Jackson 481

We can now set out their argument for the a posteriori identification or reduction of
intentional content to proper function, as follows.

P1 Belief that p is the theoretically interesting state that actually plays the folk func-
tional roles distinctive of belief that p (from reflection on what we master when
we master intentional vocabulary).

P2 The theoretically interesting state that actually plays the folk functional roles dis-
tinctive of belief that p is the state selected to do what belief that p is for qua state
so selected (from the fact that biology is the relevant theory for intentional crea-
tures like us, combined with the etiological account of function in biology).
Downloaded by [Arizona State University] at 14:03 27 October 2014

Con. Belief that p is the state selected to do what belief that p is for.

The conclusion will have the status typical of interesting scientific identifications; it will
be necessary a posteriori.
A view of this general kind seems to us to be the most interesting and plausible style
of teleonomy, and to be closest to what teleonomists say about how their view should be
understood. It entails a claim about essential properties under a suitable counterpart rela-
tion. If it is necessary a posteriori that belief that p, in particular, and contentful states, in
general, have certain selectional histories, then everything that has the same content as
some contentful state has the same selectional history in the relevant respect; that is to
say, this selectional history is an essential property of it under the same-content counter-
part relation. This is why we said, at the end of the previous section, that this is the right
version of the essential properties reading of teleonomy.
Teleonomy read as a scientific identification or reduction seems to us to face two
serious objections, and, as heralded early on, the second half of this paper is concerned
with these objections.

IV.

The first turns on the claim that teleonomy cannot be offered as the theory of content. At
best, it is a theory of one kind of content.
In the previous section, we outlined the Lockean conception of language as a system
of communication, as something in the same general family as morse code and sema-
phore. Accordingly, one good thing to mean by the term of art 'content' in connection
with belief is that it is how we take things to be which we often succeed in communicat-
ing to one another by using a language we both understand. When Jones says to Smith,
'There is a tiger nearby', the content of what Jones believes is what Jones is typically
seeking to convey to Smith by using these words. Likewise, the content of desire is what
we often convey to one another about how we would like things to be with sentences we
both understand like 'Let's have coffee' and 'Please turn on the air conditioner'. We will
call this role of content its informational role. It concerns the information we transfer by
means of language about how we take and want things to be - the information about the
states of affairs which we take and want to obtain, or the truth conditions which we take
482 The Teleological Theory of Content

and want to be satisfied, or the possible combination of objects and properties which we
take and want to be the actual combination, or the set of possible worlds which we take
or want our world to be a member o f - to mention just some of the options for character-
ising content. Now it is obvious that informational content is not identified in the sense
of being picked out and thereby made known to those seeking to transfer the informa-
tion, via selectional history. The information we transfer to one another with the kinds of
sentences understood equally by believers in evolution, creationists, and those who have
never heard of the theory of evolution or of a selectional history, is not identified by
selectional history. This is because they do not know or believe the relevant selectional
history. When Shakespeare talked to his friends, he was not using a convention-based
system that identified certain states of affairs (to settle on that way of characterising-con-
Downloaded by [Arizona State University] at 14:03 27 October 2014

tent in order to formulate the point), via the kind of selectional facts that feature in
teleonomy, as the ones he took or wanted to obtain, and was anxious to tell his friends
about - and the same goes for his friends. He and his friends obviously had some way of
making known to each other how they took and wanted things to be, but it was not via
their (non-existent) knowledge of selectional histories. Teleonomists are well aware of
this point, and often mention it, or something like it, when they explain that they are not
offering a conceptual analysis.
Although we are contrasting selectional and informational approaches to identifying
content, we are not committing ourselves to the view that there are two distinct notions
of content in the sense that they concern different states of affairs (sets of worlds, truth
conditions, etc.). It may well be that the selectional account often picks out the very same
states of affairs as those Shakespeare and his friends picked out, in whatever way they
did, when they sought to tell each other how they took and wanted things to be. It is use-
ful to have a name for the way they came to know (when they did) the states of affairs
they were telling each other about as ones they took to obtain, or ones they wanted to
obtain. We will call it the folk way, to mark the fact that most of us have mastered it, and
did so without knowing about the theory of evolution. In these terms, our point is that
informational contents are the states of affairs identified in the folk way, and, though the
folk way is not that of going by selectional history, the states of affairs identified in the
folk way may, nevertheless, overlap with the states of affairs identified selectionally.
(The extent to which they do depends on the matters of important detail we set aside at
the beginning of the paper.)
Moreover, there is an obvious way for teleonomy to incorporate the informational
side of content. We argued that teleonomy, in its most plausible version, is in part a view
about reference-fixing: we reference-fix by folk functional roles. But folk functional
roles are common knowledge. That is what makes them folk. They are, thus, plausible
candidates to be what underpins what we have just called the folk way of knowing about
content; the way of which we, creationists, and Shakespeare share a mastery. Hence,
although people in Shakespeare's time did not have a clue about the selectional histories
of their intentional states, and so could not be drawing on this non-existent knowledge
when they swapped information about how they took and wanted things to be, they did
know about their states' typical causes and effects and, more generally, their folk roles.
Hence, teleonomists can explain the informational role of content in terms of the states
of affairs on which the folk roles reference-fix. The point here is similar to a point often
David Braddon-Mitchell and Frank Jackson 483

made about the information conveyed by using a word like 'water'. What information is
conveyed by a sentence like 'There is water nearby' to those who understand it well
enough but do not know that water is H 2 0 ? Not that there is H 2 0 nearby. An attractive
answer is obtained by drawing on the reference-fixers for 'water'. The information con-
veyed is that the clear, potable stuff of our acquaintance which falls from the sky, fills
the oceans and is known as 'water' in our language community, is nearby.
Where is the objection in all this? Why should acknowledging informational content
alongside selectional content be a problem for teleonomists? The trouble is that their
remarks are evangelical rather than ecumenical. They argue, for example, that only the
teleological theory can capture the normative dimension of content. This claim is promi-
nent in Millikan's presentations of teleonomy - famously, her choice of the phrase
Downloaded by [Arizona State University] at 14:03 27 October 2014

'proper function' in describing her view is no accident - but, in the interests of widening
the net, we will consider Colin McGinn' s presentation of the claim.

We cannot, it seems, recover the content of a representation (conceptual or linguistic)


simply from the way it is actually used or is disposed to be used. This is fundamental-
ly because representations can always be used i n c o r r e c t l y . . . The general point here
is that we can partition the totality of uses (actual or potential) into two sets, the cor-
rect and the incorrect; but this partitioning cannot be effected without employing a
notion not definable simply from the notion of bare use, actual or dispositional, viz.
the normative notion of using a representation as it ought to be used given its content.
It is thus a condition of adequacy upon any account of content that it provide for this
distinction among u s e s . . . Here the teleological theory scores valuable points. For it
is similarly true that proper function is not definable dispositionaUy: what a trait or
organ is supposed to do cannot be deduced from what it is causally disposed to do.
You cannot deduce a norm f r o m a fact. 16

This is misleading advertising. The teleological theory does seek to deduce a norm from
a fact: something's selectional history is a fact about it. The difference between theories
that focus on, for example, actual, current functional roles (which we can think of as the
most plausible way to extend the roles McGinn has in mind when he talks of current
actual and potential uses) and teleological theories that bring selectional history into the
picture, is not that the latter refuse to deduce a norm from a fact; it, rather, concerns the
kinds of facts from which the norms are deduced: aetiological facts have a special place
in the latter that they do not have in the former. McGinn is fight that there is a problem
raised by the normativity of content, but the problem is not especially easy for teleono-
mists. Moreover, because informational content clearly has a normative dimension -
there is an important difference between right and wrong ways of conveying informa-
tion, and right and wrong information, about how things are - and yet informational
content is not, as we have seen, delivered by selectional history, there had better be a

Colin McGinn, Mental Content (Oxford: Blackwell, 1989), pp.159-160. The final emphasis is
ours. McGinn's phrasings here might suggest that he is primarily interested in the content of
(public) language, but in fact he takes the point he is making to apply quite generally. We (and
most teleonomists) take the content of intentional states to be the primary notion, and hold that
language is contentful inasmuch as it is used to tell about the content of various intentional
states.
484 The Teleological Theory of Content

non-aetiological way of accounting for its normativity; it had better be the case that
selectional history is not the only way to account for the normativity of content.
Teleological theorists also argue that only they can handle a well-known 'embarrass-
ment of riches' problem that arises for certain functionalist treatments of content, a
problem that derives from the fact that one and the same state has many.different, possi-
ble and actual, causes and effects. Here is a typical passage, this time from Papineau's
critical discussion of the idea that some version of 'arms length' or commonsense func-
tionalism might explain how our intentional states might be about happenings in the
world around us• He frames matters in terms of truth conditions of beliefs and satisfac-
tion conditions of desires.
Downloaded by [Arizona State University] at 14:03 27 October 2014

• . . why not [in the account of functionalism] simply extend our causal net to allow
more distal causes of perception, on the input side, and more distal effects of behav-
iour, on the output side? This would allow us to analyse the truth conditions of beliefs
as those distal circumstances which cause them, and the satisfaction conditions of
desires as those distal states of affairs they give rise t o . . .
This move, however, is fatally afflicted by the disease known as 'disjunctivitis'. The
belief that there is an ice-cream in front of you can be caused not only by a real ice-
cream, but also by a plastic ice-cream, or a hologram of an ice-cream, or and so on.
So, on the current suggestion, the belief in question ought to represent either-a-real-
ice-cream-or-a~plastic-one-or-any-of-the-other-things-that-might-fool-you. Which of
course it doesn't.
Similarly with desires. The results which follow any given desire include not only the
real object of the desire, but also various unintended consequences• So the current
suggestion would imply that the object of any desire is the disjunction of its real
object with all those unintended consequences. Which of course it isn't.
So even if we widen functionalism's causal roles to include distal causes and effects,
we still need somehow to winnow out, from the various causes that give rise to
beliefs, and the various results that eventuate from desires, those which the beliefs are
about, and which the desires are for.
This is where an appeal to teleological considerations seems to yield a natural and
satisfying answer. W e can pick out a desire's real satisfaction condition as that effect
which it is the desire's biological purpose to produce. And, similarly, we can pick out
the real truth conditions of a belief as that which it is the biological purpose of the
belief to be co-present with. t7

Papineau is right that there is a major problem here for the simple version of common-
sense functionalism that he cites, and it may be that one way of solving it is to appeal to
selectional history. But we can be certain that it is not the only way. We do succeed in
conveying to one another useful, not hopelessly disjunctive information about how
things are. Informational content is no chimera. But, as we have noted, informational

David Papineau, PhilosophicalNaturalism, p.58. Kim Sterelny, The Representational Theory of


Mind (Oxford: Blackwell, 1990), pp. 124-125, also sees teleological considerations as providing
'more discriminatory machinery', but gives them a much more circumscribed role than does
Papineau.
David Braddon-Mitchell and Frank Jackson 485

content is not identified via selectional history. Shakespeare successfully conveyed and
received information about how things are. He often knew not hopelessly disjunctive
things about what he and his friends thought - he knew the states of affairs (or truth con-
ditions, o r . . . ) that they had in mind - despite not knowing about selectional history, and
so not being able to use it to tell him what these states of affairs were. There must, there-
fore, be some other, folk way of solving the problem Papineau points to (though it would
be beyond the remit of this paper to speculate in any detail about what it might be).
Further, we know that the winnowing proceeds in a systematic way without reference to
selectional history. Suppose that yesterday we thought that there was an ice-cream in a
certain shop window, and today we discover that the 'ice-cream' is in fact plastic; we
move rationally to the judgement that we were mistaken yesterday. We do it, and things
Downloaded by [Arizona State University] at 14:03 27 October 2014

like it, every day of our lives. We grant that it is hard to give a theory of exactly how it is
done - to say that we do it by appeal to a folk theory is to say little more than that we
folk can do it - but we know that doing it does not depend on using information about
selectional history, because very many of those folk who do it lack the relevant informa-
tion about selectional history. Maxim: if something is done without appeal to such and
such, then it can be done without appeal to such and such.
There is a general point to b~ made here, not limited to the (important) problems of
normativity and winnowing. Once we acknowledge that there is informational content as
well as selectional content, and that both are properly to be counted notions of content, it
cannot be that there is some problem for content as such that only teleonomy can handle.
Teleonomy is not an approach to content that solves certain famous difficulties that
defeat all other approaches.
Our first objection, then, is that teleonomy cannot be thought of as the one true theory
of content, but, at best, as identifying a notion of content that exists alongside informa-
tional content (and maybe other kinds as well, but we will not pursue that issue here).
It is worth noting that we could have made our key point against Papineau on the
winnowing problem without invoking the informational role of content. Although we
made the point in terms of it, we did not have to. This means that denying (heroically, as
we see things) informational content does not buy a reply to our objection to Papineau.
We saw that the best version of teteonomy sees it as a theory reduction that delivers
an a posteriori scientific identification of the content of a state with the states of affairs
that play some specified role in the state's selectional history. W e also noted that a poste-
riori scientific identifications rest on views about reference-fixing (or folk markers). The
identification of heat in gases with kinetic energy goes via a view about how we refer-
ence-fix on heat - the usual view being that we do so via its filling a certain functional
role; and, in the same way, the path to the identification of content with, say, proper
function goes via a view about how we reference-fix on content. We suggested, in agree-
m e n t w i t h P a p i n e a u , t h a t the m o s t p l a u s i b l e v i e w for t e l e o n o m i s t s is t h a t we
reference-fix on content via folk functional roles; but, in any case, it is clear that we do
not reference-fix via proper function, because we are able to talk about content in igno-
rance of proper function.~8

'g Moreover, if we did reference-fix by proper function, the identity of content with proper func-
tion would be a priori true, contrary to the views of virtually every teleonomist.
486 The Teleological Theory of Content

This means that there must be a solution to the winnowing problem that does not
appeal to selectional history. Although Papineau is right that a functionalist approach to
content needs somehow to winnow out, from the various causes that give rise to beliefs,
and the various results that eventuate from desires, those which the beliefs are about, and
which the desires are for, and that the simple version of commonsense functionalism he
sketches cannot do the job without supplementation, we can be sure that the winnowing-
out job can be done without appeal to selectional history. For we are able to reference-fix
on content - and, on pain of losing the best version of their theory, teleonomists must
grant that we are able to reference-fix on content - but we do this, as Papineau himself
allows, without appeal to selectional history. If, as we and Papineau agree, the folk roles
serve to reference-fix on content, then, somehow or other, we can do the needed.win-
Downloaded by [Arizona State University] at 14:03 27 October 2014

nowing out without appeal to selectional history.

V.

Our second and final objection to teleonomy calls for a modicum of stage setting.
According to a tradition associated with Descartes, a person's identity over time is not
secured by the kinds of physical and psychological continuities much discussed in the
recent literature, the kinds that ground our best opinions about personal identity; rather, it
is secured by the numerical identity through time of a certain subject of experience, the
subject Hume found so elusive. One objection to this view is that we have no good rea-
son to believe in the existence of this mysterious subject of experience, but the most
influential objection is the one associated with Locke, the 'who cares' objection. The
point behind this objection is that questions of personal identity matter; our personal
relations, the institutions of punishment and reward, and our assignments of responsibili-
ty are structured around how we individuate persons. If personal identity is, somehow or
other, a matter of the much-discussed continuities, we can make good sense of why ques-
tions of personal identity matter. We have, for example, at least the beginnings of an idea
of why it is typically right to punish the person who did the evil deed rather than some-
one else altogether. But it is deeply obscure why we should care about the numerical
identity over time of the notoriously elusive subject of experience.
Similar considerations apply against G. E. M o o r e ' s view that goodness is an
unanalysable, non-natural property. ~9 One objection is that we have no good reason to
believe that this mysterious property is ever instantiated, but the most influential objec-
tion is that it is deeply obscure why we should care whether or not it is instantiated. We
would not take seriously someone who said, 'How lucky it is that happiness, honesty and
truth are associated with this special, non-natural, unanalysable property goodness, oth-
erwise we could not approve of happiness, honesty and truth'. What we care about is the
presence of happiness, honesty, and truth, and the absence of misery, dishonesty, and
error; the presence or absence of some further unanalysable, non-natural property is nei-
ther here nor there.
We think a similar point applies against the selectional account of content. It does not
cohere with the way we care about content. Our first objection to teleonomy was that

'~ G.E. Moore, Principia Ethica (Cambridge, Cambridge University Press, 1929), §§6ff
David Braddon-Mitchell and Frank Jackson 487

reflection on the informational role of content shows that we can and do identify content
independently of selectional history, and that this fact undermines the arguments
advanced for teleonomy. Our second objection is that we do not value selectional history
in the right way for it to determine content in the sense in which we care about the con-
tent of what we believe and desire. It at best determines the content we do not care
enough about.
It has been widely noted that prosthetic surgery for the brain and central nervous sys-
tem is no more problematic in principle than prosthetic surgery in general. Just as it is
sensible to replace a defective heart valve with an artificial one that does the job the orig-
inal valve did before the deterioration, so it would be sensible to replace defective brain
circuits with silicon chips if they did better what was wanted and needed. A common
Downloaded by [Arizona State University] at 14:03 27 October 2014

thought experiment is to imagine that, as parts of someone's brain degenerate, they are
progressively replaced by silicon implants, and to note that, provided the implants fill the
pre-degeneration functional roles of the parts they replace, the surgery counts as success-
ful. The key point for us is that what a patient facing such an operation will care about is
essentially that what is about to be inserted in her brain will do the job done by the rele-
vant brain states before things started to go wrong.
Our objection is that teleonomists have to deny this, or else embrace the conclusion
that content does not matter. According to them, the selectional history of an implant is
of vital importance to content. This means that if the patient cares about the content of
her beliefs and desires, she should be vitally concerned not merely with whether or not
the implant will fill the pre-degeneration input-output function of the brain structure that
it replaces, but also with the selectional history of the implant; for that is crucial to the
beliefs and desires that it is possible to have as a result of the implant, according to her.
But it is very implausible that our patient should regard the selectional history of the
implant as an additional thing to worry about over and above what the implant will do,
how long it will last, how much the operation will cost, the appearance of the scar, and
such like.
(We have been asked whether we would extend this objection to the proper function
account of function in biology. Would we argue that because we would all choose an
artificial heart that did the right things here and now ahead of one with an impeccable
pedigree but which had a tendency to malfunction on occasion, there is something wrong
with the proper function account of function in biology? We would not, and the reason is
that there is a crucial difference between the two cases. The contents of our beliefs and
desires are central to our whole conception of what we are like, what is important and
worth caring about, and our place in the world - and also to the more mundane business
of getting around from day to day - in ways that function in biology does not begin to
approach. The idea of becoming like Hitler horrifies us precisely because of what Hitler
believed and desired; scientists devote themselves to finding the grand unified theory or
a cure for cancer, and that is to devote themselves to acquiring certain kinds of belief;
and, on the more mundane level, the passage we carve through the world day by day is
shaped by how we take things to be and how we want them to be.)
It might be objected that, in the case of silicon implants, we should not focus on the
history of the physical structures themselves, but, rather, on the history of the relevant
input-output functions. We should be concerned with ensuring that the patient ends up
488 The Teleological Theory of Content

with input-output functions that have the right evolutionary history, even if in an indirect
way. These indirect ways might include the intentions or history of the makers of the
chips (they designed them with the input-output function they have because of the mak-
ers' own evolutionary histories), or the intentions or history of the surgeon as they bear
on the input-output function. But switching to the selectional history of the input-output
function of the implant, rather than that of the implant itself, does not affect our basic
point. It is just as implausible that our patient should regard the intentions of, say, the
implant's maker concerning the input-output function as a crucial additional thing to
worry about, over and above what the implant will do, how long it will last, and the like,
as it is to regard the creator's intentions regarding the implant as an additional thing to
worry about. And it would be no more sensible for a patient to ask for an implant to be
Downloaded by [Arizona State University] at 14:03 27 October 2014

removed and done over again in exactly the same way because of a concern about the
intentions of the surgeon with respect to the input-output function, as it would be to ask
for the replacement because of a concern about the surgeon's intentions with respect to
the implant itself.
Teleonomists who insist that we should take a holistic, anti-language of thought,
approach to what has content - bits of the brain as such do not have content, but instead
a large part of the brain represents a great deal about how things are and should be -
might well object that what matters for them is not the history of bits of the brain or of
particular implants, but instead the history of the brain as a whole. As a result, for them
the history, or lack of it, of some particular implant or its input-output function is unim-
portant. They can, therefore, agree that patients should take a don't care attitude to the
history of the implants. But these teleonomists cannot take a don't care attitude to a
sequence of changes, and we can as easily make our point in terms of such a sequence.
Suppose that aseries of operations is planned by a surgeon to replace someone's neu-
rones because of a degenerative brain disease. At the end of the sequence, little or none
of the original structure will be left. The patient, a good teleonomist, is assured that all
will be fine, because the operations will be done with the right intentions, and, crucially,
from a blueprint with the right causal history involving her mind-brain to ensure a
derived function of the kind that confers content according to her version of teleonomy.
Our teleonomist agrees, and the operation proceeds. But, we may suppose, a 'swamp'
blueprint, a random aggregation of molecules which exactly resembles the real blueprint,
materialises overnight on top of the real blueprint, and is accidentally picked up and used
by the surgeon. How plausible is it that, after the sequence of operations is performed,
the patient - or her family, perhaps convinced by teleonomists that she now believes and
desires nothing - should insist that the operation be repeated using the original blueprint?
It seems to us that it would absurd to ask for the operation to be done again; but then the
teleonomist is committed to saying that content does not matter - just as we said before,
and this time it does not help to deny the language of thought

VI.

We conclude by considering an interesting response that involves a big departure from


usual versions of teleonomy.
It might be urged that what really matters is that the input-output profile be of a kind
David Braddon-Mitchell and Frank Jackson 489

with those with the right evolutionary history, even if the reason for the profile's pres-
ence in a given case has nothing to do with this history. This kind of teleonomy can
w e l c o m e D o n a l d D a v i d s o n ' s S w a m p m a n into the c o m p a n y o f r e p r e s e n t o r s , for
Swampman's input-output profile is of the kind which has evolved in the right way in
humans? ° On this version, it might seem, there would be no reason for our patient or her
family to ask for the operation to be done again, because the input-output profiles would
be of a kind with those with the right histories.
The trouble is that the input-output profile might have too many histories. There is no
guarantee that a particular profile evolved only once, and if it evolved a number of times,
there is no guarantee that on each occasion the same kind of history was behind it. 21 But
then there would be no answer as to the content associated with the profile, as the kind it
Downloaded by [Arizona State University] at 14:03 27 October 2014

belonged to would have two quite different, putatively content-conferring histories.


Moreover, and worse, this version of teleonomy suggests that what we think and want
may depend crucially on how things are with creatures totally removed from us, both
causally and in space and time. Whether or not an input-output profile is of a kind that
has a certain history could be settled by the evolutionary histories of creatures in other
solar systems, before the dinosaurs, after humans have died out, and so on. Suppose, to
sharpen the point, that what matters according to the revised proposal is that there has
already been a system which evolved to have a certain input-output profile. Moreover,
there are two Swampmen: one early in the history of the universe, one late. Between
them in time a species evolves with the same input-output profile as both. It seems arbi-
trary to suppose that the later Swampman gets to be a representational system but not the
earlier one due to an accident of timing. On the other hand, if what matters is that at
some point in history a system evolves to have the right profile, we get an even more
unpalatable result. If a Swampman arises and there has not yet been an evolved system
with the same profile, its representational status awaits what will happen in the possibly
distant future? 2

University o f Auckland Received December 1996


Australian National University Revised April 1997

20 Donald Davidson, 'Knowing One's Own Mind', Proceedings and Addresses of the American
Philosophical Association 60 (1987) pp.441-458.
2, The same point applies mutatis mutandis to the suggestion that what really matters is that the
neurological state be of a kind which has evolved in the right way in humans.
22 We are indebted to dissenting comments from Karen Neander, and to dissenting and supporting
comments from referees.

S-ar putea să vă placă și