Broad Content, Narrow Content, Phenomenological Content

Broad Content, Narrow Content, Phenomenological Content

Steven Horst
Wesleyan University (CT, USA)
[email protected]

" "
"This article is something of an intellectual orphan. The first "
"version of it was drafted while at an NEH Institute on Meaning at "
"Rutgers in 1993, and the present version updated while at an NEH "
"Institute on Consciousness and Intentionality at the University of "
"California at Santa Cruz in 2002. I have lost track of where and "
"when versions of it were submitted to journals (if indeed they ever "
"were), but at the time, the idea that phenomenology has anything to "
"do with content was so far out of the mainstream in analytic "
"philosophy that it tended to be greeted with great skepticism. I "
"became occupied with other projects, and did not make very great "
"efforts to get this published, in part because some of the main "
"ideas were already included in Chapter 9 of Symbols, Computation, "
"and Intentionality (1996). I have recently become aware that ideas "
"in this vicinity have been more thoroughly explored and better "
"received in recent years, including a collection by Tim Bayne and "
"Michelle Montague entitled Cognitive Phenomenology (2011, OUP). "
"Discovering this prompted me to search for what I had written on the"
"subject (this article). It has not been updated to reflect more "
"recent publications, and I am not sure whether I shall do so, given "
"the number and scope of other projects on my plate at the moment. I"
"am posting it online because it may be of at least historical "
"interest to some. It should be regarded in the category of draft "
"material that should not be cited without permission of the author, "
"though I suppose the citation information at present would be "
""unpublished paper, 2002". "


Abstract

Two different but similar-sounding distinctions are discussed. The first
is the familiar distinction between broad and narrow content, where narrow
content has come to be understood in terms of what physical and/or
functional doppelgangers would necessarily share. The second is the older
distinction between the "intentional character" or "phenomenological
content" of intentional states (which is fixed by conditions internal to
the experience of the individual) and its fulfillment in veridical
intentional states. It is argued (1) that phenomenological content is
distinct from both broad and narrow content, (2) that it is a useful idea,
and indeed (3) that it is possible to understand a notion of narrow content
characterized in functional terms as a notion of content only if it is
viewed as an abstraction from a pre-existing richer notion of content, such
as phenomenological content.

Broad Content, Narrow Content, Phenomenological Content

Steven Horst
Wesleyan University

Discussions of meaning and content have occupied a central place in
the philosophical spotlight, both in recent analytic philosophy of mind and
language, and in prior periods, such as the discussions of intentionality
among medieval philosophers and the phenomenological movement spawned by
Edmund Husserl.[1] Ironically, it has proven very difficult to pin down
the meaning of the word 'meaning'. Simple and familiar words such as
'meaning' and 'content' disguise a legion of subtly different technical
uses, and so misunderstandings and cross-purposes are all too familiar in
discussions of meaning. In this paper I shall discuss two similar-sounding
but ultimately orthogonal distinctions about meaning or content, both of
which might be seen as distinctions between individualist and non-
individualist interpretations of semantic ascriptions. The first is the
distinction between "broad" and "narrow" content that has been the focus of
so many recent publications in philosophy of mind and language. The second
is an older distinction between the "intentional character" or
"phenomenological content" of a mental state, which is fixed by the
phenomenology of first-person experience and invariant over epistemic
situations, and its "fulfillment" in veridical intentional states. I shall
argue (1) that this "phenomenological content" is equivalent neither to
broad nor to narrow content, (2) that it is a philosophically useful
notion, and indeed (3) that the notion of narrow content is useful (and
even intelligible) only on particular assumptions about its relationship to
phenomenological content.


1. Broad and Narrow Content

The notions of broad and narrow content have become commonplace in
philosophy of mind and language in the wake of articles by Putnam (1975),
Kripke (1971) and Burge (1979, 1986). The essential insight behind this
distinction is that there is a component of what we commonsensically think
of as "meaning" that is dependent on factors outside of the thinker's head.
According to this view, for example, when I use a natural kind term such
as 'water', the meaning of that term is fixed, not merely by the
qualitative properties I associate with it, but by a kind of ostensive act
(or perhaps a causal relation of a special sort) that picks out the natural
kind of that stuff, whether I know its nature or not. Thus (the argument
goes) my use of the kind-term 'water' picks out H2O, regardless of whether
I know the molecular structure of water, because I learn and use the term
in an environment where it picks out a certain kind of stuff, and that
stuff happens to be H2O. A being in a different environment that was
exposed to some other kind of compound that had the same phenomenal
properties as water but a different underlying makeup (call it XYZ) might
well call it "water*" (a homophone of 'water'), and it would reason about
water* the same way I reason about water, but the meaning of his word
'water*' would be different from that of my word 'water' because his word
would refer to XYZ while mine refers to H2O. Indeed, if we adjust the
example to refer to substances that do not make up most of our own body
weight, we can even image molecular duplicates A and B on different worlds
that are exposed to two different substances s1 and s2 that their senses do
not differentiate. They attach homophonic names (k1, k2) to what they
encounter, and their respective kind-terms are associated with other
knowledge in analogous ways (they are, after all, molecular duplicates),
but their kind-terms pick out different kinds of stuff due to the
essentially indexical nature attributed to kind-terms by this theory. 'k1
' refers to s1 because it was samples of s1 that produced A's thoughts,
and likewise for 'k2' and s2 , mutatis mutandis.
What arises from such thought-experiments is a conviction that
meaning and content—or at least certain components of what we call
"meaning" and "content"—depend upon factors that lie outside the
individual. You cannot simply look at thinkers/speakers A and B in
isolation from their histories and environments and know what their kind-
terms k1 and k2 "mean"—i.e., what natural kinds they refer to. Indeed, even
A and B themselves cannot look at their own experience to adjudicate the
matter.
But neither is it the case that everything we mean by 'meaning' or
'content' is determined by external factors. And in particular, it has
been argued that the portions of "meaning" that are important for a theory
of reasoning should confine themselves to factors internal to the organism,
since it is presumably only these that will play a role in inference.
(Putnam, 1975; Fodor, 1980, 1981) Thus it has become standard to
distinguish between "narrow content", which is confined to factors internal
to the organism and "wide" or "broad content", which includes the elements
of meaning dependent upon some combination of causation, ostension,
environment, and the use of socially-distributed semantic conventions.
Narrow content is sometimes described as what is "in the head" or internal
to the organism; and it is generally assumed within thought-experiments
involving broad and narrow content that narrow content is the portion of
semantics that is necessarily shared by molecular duplicates or physical
doppelgangers.[2] While discussions of broadness and narrowness have often
taken the form of debates over whether meaning "is (entirely) in the head,"
it is perhaps more helpful to cast the discussion in terms of broad and
narrow senses of the words 'meaning' and 'content', and as isolating
separate aspects of what had previously been thought of as unified ideas.


2. Phenomenological Content

A distinct yet similar-sounding distinction has been emphasized by
writers such as Brentano (1874), Husserl (1913) and Chisholm (1957). These
writers, whose chief interest was the phenomenon known as intentionality,
stressed the importance, when discussing intentionality per se, of
separating questions about the content of an intentional state from
questions about its relationship to extra-mental reality. I can make a
judgment that there is a dog in the yard even if there is no dog there.
Indeed, I can judge that there is a unicorn in the yard even if there are
no unicorns, Edison planned to make a lightbulb before there were any
lightbulbs, and so on. Likewise I can entertain hopes, fears,
expectations, and the like towards a broad panoply of possible states of
affairs regardless of whether those states of affairs obtain. Brentano and
Chisholm treat this feature of intentionality—the possible non-existence of
the intentional object—as the distinguishing feature of intentional states,
and Husserl's phenomenological method involves an epochê or bracketing of
all questions about extra-mental reality in considering the structure of
intentional experience as such.
Thus, like proponents of narrow content, these writers also find it
useful to distinguish between one kind of state that is "internal to" the
thinker from all concerns about extra-mental reality. The internalistic
property in question consists in the combination of the intentional
modality of the mental state (e.g., judging, doubting, hoping, desiring,
etc.) and the content of the state (e.g., that a dog is in the yard, that
Clinton is President, etc.) We may call this property the "intentional
character" of the state, as opposed to its fulfillment or veridicality when
it corresponds to the world in appropriate ways. A judgment that there is
a dog in the yard is just that—a judgment with that particular
content—regardless of the state of the external world. But it is only
veridical or fulfilled when there is in fact a dog in the yard, and is only
epistemically justified if the judgment is arrived at in appropriate ways.
This distinction has connections with certain other philosophical
discussions as well. It is this distinction that is pointed to by early
discussions of "methodological solipsism" (Putnam, 1975), and by the
Cartesian evil demon scenario. There are also resonances here with the
recognition by ordinary language philosophers that many verbs of perception
and cognition are ambiguous between senses that imply a certain kind of
success and senses that do not. For example, we might say of
schizophrenic Mr. Phelps, "he is hearing voices again," even if there are
no voices there for him to hear. When we do so, we are reporting the
intentional character of his experiences. (Sometimes this has been made
more explicit by the inelegant techniques of saying "Phelps is being
presented to voice-wise" or "Phelps seems to be hearing voices.") But it
is equally proper to use the verb 'hear' in a fashion that implies
successful hearing of real sounds: "Phelps can't be hearing voices, he
must be simply imagining them." Likewise, it makes perfect sense to say
"Johnny is afraid of minotaurs," (reporting the intentional content of his
fears after staying up late watching reruns of Xena, Warrior Princess );
but also to say, "He can't be afraid of minotaurs, because there are no
such things." We all "see things" in our dreams, and indeed "act" in them,
the composer "hears" the (yet unwritten, unperformed) piece of music, and
so on. There is in fact a systematic ambiguity of verbs of perception
here, and verbs of cognition are likewise ambiguous between interpretations
cashed out in terms of what Chisholm called intentional and non-intentional
contexts. These linguistic phenomena seem to map nicely onto the
distinction between intentional content and various kinds of fulfillment or
veridicality. Intentional contexts report intentional content alone; non-
intentional contexts report kinds of fulfillment as well.
Now the intentional character of a mental state involves both an
intentional attitude (such as judgment or hope) and the content of the
mental state. The "content" in question here, however, is isolated in a
different way from broad and narrow content. The intentional character of
a mental episode is what remains invariant about it under all alternative
assumptions about extra-mental reality, and so the kind of "content"
referred to by writers such as Brentano, Husserl and Chisholm must also be
precisely such as to remain invariant under those conditions. We may call
this notion of content phenomenological content. The phenomenological
content of a mental state is that portion of meaning that is fixed by the
experiences of the thinking subject as such, regardless of assumptions
about her environment or even her own body. It must be independent of such
assumptions because this is required by the fact that phenomenological
content remains constant under Putnam's methodological solipsism or
Husserl's phenomenological epoché, both of which set aside questions about
both the nature and even the very existence of the world and one's own
body. Broad content fixes the bodily state metaphysically;
phenomenological content fixes the experiential state epistemically. More
broadly, it is this notion of content that is needed for the classical
characterization of intentionality that makes it independent of whether any
real object corresponds to the intentional object.
Thus phenomenological content is fixed precisely in phenomenological
terms. It is defined by what is invariant for the thinking subject as
such, and is pointedly agnostic about any features that may depend upon
extra-mental reality, and even about the existence of such a reality,
including assumptions about the body of the thinker.


3. Phenomenological Content and Broad Content

Phenomenological content is not broad content. The phenomenological
content of my thought will remain the same under varying assumptions about
extra-mental reality that would change their broad content. For example,
we may imagine that I have a phenomenological doppelganger—a being who
shares all of my phenomenological properties—who is the victim of a
Cartesian demon or has lived his entire life hooked into a virtual reality
simulation. At least some of my thoughts about individuals are really
about individuals. And likewise at least some of my natural kind concepts
pick out genuine natural kinds. But my doppelganger's experiences that
represent individuals as being in particular situations are not veridically
connected to actual individuals in the fashions necessary for successful
reference. Hence the broad content of his thoughts is either vacuous, or
at least radically different form mine. (On some ways of interpreting
broad content, all of his thoughts would turn out to mean "demon"—although
this fact would no doubt come as great a great shock to him as it would to
you or me![3]) And if we accept an externalist theory of kind-terms,
those of his thoughts that have the same phenomenological content as my
thoughts about water do not refer to H20, because he has never interacted
with H20, and hence his thoughts do not have the same broad content as
mine.[4]
Of course, the fact that phenomenological content is not identical to
broad content does not imply that there are no essential relations between
them. For example, one thing that needs to be explained about broad
content is what it is that makes it the case that it is, say, the actual
molecular kind of water that determines the broad content of 'water'. One
plausible story would be that the concept 'water' either is or at least
contains a rule that ostensively picks out whatever extra-mental stuff one
is dealing with on particular paradigm occasions, and makes that kind of
stuff the referent of the concept. What stuff (i.e., H2O or XYZ) is picked
out is not spelled out in the rule, but the procedure for linking the
concept with whatever is picked out in a given act of ostension is spelled
out. What might be the source of such a rule, though? One possible
account would be that it was supplied by phenomenological content: that
when we point to water, the intentional content of our thought is something
like "that stuff, whatever it is." This account of what fixes broad
content may or not be correct; I present it merely as a way of illustrating
how broad and phenomenological content could have an essential relationship
while not being identical.[5]


4. Phenomenological Content and Narrow Content

There is little temptation to conflate phenomenological and broad
content, as one is an "internalist" notion while the other is
"externalist". It is much more tempting, however, to think that the kind
of content called for in classical analyses of intentionality—which bracket
or hold invariant how the world is outside of the mind—is identical to the
recent notion of narrow content. Indeed, the early discussions of
"methodological solipsism" seem to bolster this identification, as they too
bracket externalist factors, while the very term 'solipsism' suggests that
what is being held constant is what is invariant from the perspective of
first-person experience. Putnam (1975) introduces the notion of
methodological solipsism with the following words:

When traditional philosophers talked about psychological states (or
"mental" states), they made an assumption which we may call the assumption
of methodological solipsism. This assumption is the assumption that no
psychological state, properly so called, presupposes the existence of any
individual other than the subject to whom the state is ascribed. (In fact,
the assumption was that no psychological state presupposes the existence of
the subject's body even: if P is a psychological state, properly so called,
then it must be logically possible for a "disembodied mind" to be in P.)
(1975: page 136)

Thus the origins of this term were far removed from the assumption that
physical similarity was a necessary (or sufficient?) condition for having
the same narrow content. Whatever the reasons for the change in the
characterization of narrow content, we shall see that if narrow content is
understood as what would necessarily be shared by physically and/or
functionally identical beings (hereafter dubbed doppelgangers), narrow
content and phenomenological content diverge in counterfactual cases—and,
for all we know, in actual cases as well. In the remainder of the paper, I
shall attempt to spell out these differences, and to offer a diagnosis of
the situation.
I suspect (and indeed shall argue anon) that the deep motivations
behind the notion of narrow content are closely related to the
phenomenological notion of content. However, discussions of narrow content
have taken it in quite a different direction: by characterizing narrow
content as what would necessarily be shared by physical and/or functional
doppelgangers. Physical duplication and functional isomorphism are, of
course, two different notions, and it behooves us to ask of each of them
whether they are equivalent to phenomenological content. We shall thus ask
for each variant on narrow content (the physical and the functional)
whether identity of narrow content implies or is implied by identity of
phenomenological content.
First, must phenomenological doppelgangers be physical doppelgangers
as well? With respect to physical duplication, the answer seems clearly to
be no—there are any number of physical changes in an organism that would
not affect experience. This, indeed, is consonant with much of the
motivation behind functionalist approaches to the mind, and I shall take it
that the point needs no argumentation.
But must phenomenological doppelgangers be functional doppelgangers
as well? Again, the answer is clearly no if we include all functional
descriptions of the organism, as many of these have nothing to do with
conscious cognition. But suppose we restrict our functional properties to
those involved in cognition and particularly in reasoning, such as the
formal and material inferences a thinker is wont to make. Even here, it
seems clear that there can be functional differences that are not
transparent to the subject. For example, suppose that A and B are subject
to different fallacies of informal reasoning. As a result, they are
disposed to draw different inferences from a given proposition Px. This is
a functional difference that has impact upon the narrow content of P for
the two subjects. However, is this difference one that impacts upon what
it is like to think Px? Surely not. We may illustrate this by supposing
that, instead of two thinkers, we imagine ourselves to have different
logical shortcomings at different times of day: say, we accept fallacious
arguments in the early morning that we would reject in the mental lucidity
that dawns upon us after the second cup of coffee. This would seem to
imply that the narrow content of Px changes for us after a cup of coffee,
but need it change the phenomenology of thinking Px? Not unless our actual
inferential dispositions are phenomenologically transparent to us. And
this seems too strong an assumption to make without substantial
argumentation.

But perhaps functional differences of this sort—that is, of the sort
due to general features of reasoning and not due to particular
concepts—lead us astray. More interesting are functional differences
involving material inferences.[6] Bracketing functional differences
deriving from differences in formal reasoning leads to what is admittedly a
more limited form of functionalism, but also a more plausible one. And
here the relationship between phenomenological invariants and functional
organization is considerably more vexed. Can two people (or one person at
different times) be disposed to draw different material inferences from Px,
and yet have no phenomenological difference between the two thoughts? The
answer depends, in part, upon just what we count as a "phenomenological
difference" of the relevant sort. Let us motivate this concretely: there
are psychological studies that show that the material inferences a person
is likely to draw can be manipulated by context: for example, that in
context C1 a subject S will infer Qx from Px, while in context C2 she will
not do so. Now does this mean that there is a difference in the
phenomenological content of thinking Px in C1 and C2? If we restrict the
phenomenological element to the narrow slice of thought involved in merely
thinking Px, I think that the answer is probably no. But we might also
wish to include more than this in the phenomenological invariants of
thinking Px: we might wish these to include the assumptions and
implications that are routinely buried in a concept but not brought to
conscious awareness. If we go this route, the phenomenological invariants
of a content might include all that is invariant from the subjective
perspective in thinking more carefully about the concept as well. And this
would surely include many, though perhaps not all, material inferences that
the subject is likely to draw: namely, those that she rightly foresees that
she would draw in a particular context by imagining herself in that
context. If I believe whales are mammals and someone else believes them to
be fish, there may be no immediate difference in our phenomenologies when
we espy a whale, but the differences in the content of 'whale' are readily
available to phenomenological investigation: if asked, "Do you think whales
are mammals?" I quickly see that I do think this, while she sees that she
does not.[7]
This is not a case of differences that depend on the world, but
merely upon links between concepts. And here the gap between phenomenology
and functional structure seems to have been narrowed: we were looking for
cases of phenomenological invariance combined with functional divergence,
and it turns out that many cases of functional divergence in material
implication will be accompanied by a difference in phenomenological
possibilities—i.e., the differences in phenomenology that accompany
differences in functional relations between concepts may not be differences
in the occurrent phenomenology each time the concept is employed, but
differences in what facts about conceptual relations are phenomenologically
accessible.
The split between phenomenological and narrow content becomes much
more pronounced when one reverses the question, and asks whether functional
doppelgangers need be phenomenological doppelgangers. First and most
fundamentally, there is nothing about the functional description of an
organism (assuming this is done in terms that do not already presuppose a
phenomenology) that entails that the organism will have any phenomenology
whatsoever. Here we may appeal to familiar arguments of two different
sorts. On the one hand, writers like Block (1978) and Lycan (1987) have
shown that for any functionally-specified intentional system, one can
construct a trivial isomorphism to a system consisting of elements such as
molecules in a bucket of water that clearly do not have properties such as
meaning or enjoy any phenomenology. Horst (1996) has extended this
argument to show that there are also purely formal mathematical objects
that share such isomorphic structures. On the other hand, various writers
(Kripke (1971), Jackson (1982), Searle (1992), Nagel (1974), Horst (1996),
Chalmers (1996)) have argued for some variation upon what Chalmers calls
the "hard problem of consciousness"—that physical and formal descriptions
of a system do not entail the presence of a unique phenomenology, or indeed
of any phenomenology at all. Thus it seems conceptually possible that
there be beings that are functional doppelgangers of ourselves that are
completely bereft of a phenomenology. And of course it is a part of the
problem of other minds that we cannot know with certainty that others
around us are not indeed functionally like ourselves, yet phenomenological
zombies.
Other familiar thought-experiments can be used to show that, even in
cases where there is a phenomenology, functional isomorphism need not imply
sameness of phenomenology. Consider the case of spectrum-inversion: my
counterpart and I respond identically to light frequencies, yet he
experiences the same kind of qualia I call "green" when he is exposed to
red light, and vice-versa. Our concepts 'green' are thus functionally
identical yet phenomenologically different.
Nor do the problems disappear if we take narrow content to be what
physical, as opposed to functional, doppelgangers necessarily share. For
the thought-experiments that pose problems for identifying phenomenological
content with a functional property can also be adapted to physical
doppelgangers. First, the "hard problem" of getting the phenomenological
what-it's-like of conscious mental states out of naturalistic descriptions
is just as much a problem for physical description as for functional
description. It is conceptually possible that there be molecular
duplicates of ourselves with inverted qualia, or with no phenomenology at
all, because there is no way of forging metaphysically necessary (as
opposed to nomic but metaphysically contingent) connections between
physical and phenomenological properties.
In recapitulation, then, neither physical nor functional duplicates
need be phenomenological duplicates. Functional and physical properties
entail neither particular phenomenological properties, nor any
phenomenological properties at all—though we assume there to be strong
nomic yet contingent relations here. Phenomenological duplicates need not
be physical duplicates, for a variety of familiar and uncontentious
reasons. However, while there seems to be little reason to expect that
phenomenological sameness need imply thoroughgoing functional sameness
(e.g., being subject to the same fallacies of inference or associations),
it does seem to suggest a certain amount of functional sameness: i.e.,
insofar as I can discern what inferences I would make (by making or
simulating them in my mind), functional differences must be accompanied by
phenomenological differences: not differences in the occurrent
phenomenology of thinking Px, but in the fact that I can tell "from the
inside" whether or not I would infer Qx from Px. Of course the limitations
that undoubtedly exist in our ability to prognosticate how we would reason
in different circumstances would also place limitations on how far this
helps us to connect narrow and phenomenological content.
This would seem to establish that phenomenological content is not
identical with narrow content, at least if the latter is understood as
defined in terms of features that physical and/or functional duplicates
would necessarily share. However, once again, while phenomenological
content is not identical with another notion of content, it seems to have
close connections with it, in that relations of routine material inference
would seem to be systematically (though likely not universally or
incorrigibly) subject to being made phenomenologically transparent.


5. Is Narrow Content Phenomenological Content Gone Astray?

If narrow and phenomenological content are not identical, there is
nonetheless a close relationship between them. In this final section of
the paper I shall attempt further to clarify this relationship. In brief,
the now-standard formulation of narrow content in terms of what physical
and/or functional duplicates would necessarily share is best understood as
an abstraction from a pre-existing notion of content which is essentially
phenomenological. Viewing narrow content as an abstraction shows how it
can serve a methodological purpose in psychology while leading to vexed
philosophical problems such as zombie cases if taken as a metaphysical
truth: functional states, like point-masses, are useful abstractions, but
do not capture the full nature of the phenomena from which they are
abstracted.
First, note that the familiar formulation of narrow content is a
curious way to isolate any notion of content at all. Consider first the
case of functional isomorphism. Does it follow from the fact that A is
functionally isomorphic to B that A and B have the same content? The
answer is obviously no: it may be that neither A nor B has any content at
all. Two metronomes may be functionally equivalent, but neither has any
content. The mere notion of functional isomorphism is not sufficiently
rich to assure any content at all. Neither will it do to say that certain
kinds of complex functional structure assure the presence of content, again
hearkening back to the classic arguments of Block (1978) and Lycan (1987)
that any functional description has trivial realizations in nature that
clearly are not thinking beings and Horst's (1996) argument that every
functionally-defined system has a purely abstract (hence presumably non-
sentient) realization in mathematics.
There are also straightforward analogies here in the natural
sciences: certain parts of thermodynamics share a mathematical description
with information theory. The moral to be drawn from this fact, however, is
not that heat is really just negative information or vice-versa, but that
the two systems are isomorphic, and that their mathematical form does not
exhaust the nature of either. What makes heat and information different is
precisely what we abstract away from in rendering the mathematical
equations that describe their internal relations. If we treated the
mathematics of information or heat as its essence, forgetting that we had
performed an abstraction away from its defining features, we would reach
the paradoxical conclusion that temperature was nothing but (negative)
information. (Worse, we would expect all cases of receiving (negative)
information to be increases in temperature.) More fundamentally, there is
nothing about the mathematics itself that makes it about temperature or
information. The universe is not just mathematics—in older terminology, it
has a material as well as a formal nature. Likewise, if we abstract away
from thought in vivo to its form, we may well get a functional description
that is useful in, say, constructing a theory of reasoning, but we will err
if we then assume that this functional form is itself constitutive of
thought. Thought may have a functional form, but it is not reducible to
that form, any more than is heat or information or gravitation.
What might be right about functionalism in psychology is the idea
that a nomological science of cognition requires a functional (or, more
broadly, mathematical) description of "beings like us"—where that phrase is
meant to single out the features essential to our mental life. But it does
not follow that what is essential to beings like us is exhausted by the
functional description, any more than what is essential to gravity is
merely that it is well-described by the inverse square law. The notion of
'function' that is used in the context of, say, Turing tables is not itself
a notion of content. At best, it is a mathematical tool that is useful for
making explicit the relations between concepts with different contents in
reasoning. If Turing tables provide good models of the mind, the
conclusion we should draw is that Turing tables are to cognition as the
inverse square equation is to gravity. We start with some notion of
content that is not itself functional and abstract away from it in an
attempt to get a rigorous description of its structure.
What the functionalist is really after, I think, are (1) the
empirical hypothesis that thought does in fact have a functional
description, and (2) the transcendental claim that, if thought processes
are to be causal processes, and semantic values are causally inert, then
there must be some kind of coordination between the content of a thought
and something in the brain (sometimes misleadingly called "syntax") that
can be causally efficacious. This coordination need not be logically
necessary, and the "syntax" need not explain the semantics, as advocates of
computational theories like Fodor (1993) have gradually come clearly to
see.[8] Likewise the former hypothesis could turn out to be false, which
may not be a problem for mental states if these are not intended to be
functionally-describable in the first place, but would clearly present
problems for developing a nomological psychology. Now these aims are quite
compatible with viewing a functional description of the mind as an
abstraction that has bracketed the essential features of content. If we
are after a nomological description of, say, human reasoning, we do not
need to know what is essential to content per se, but only the relations
between thoughts that make a difference in reasoning. Nor need our theory,
as a theory of human reasoning, deny that there can be individual
differences, or that the same contents might give rise to very different
inferences in other species. Likewise, if we are looking for the
mechanisms that link semantics with causal powers in human brains, it is
not necessary that the relation between content and causal power be more
than an empirical generalization. Logically contingent connections between
content and causation are enough for empirical science, just as logically
contingent relations between variables are enough in physics. The
scientist, unlike the metaphysician, need not worry about zombie scenarios:
they are empirically irrelevant if they do not occur, and they are
empirically undecideable even if they do occur. Not all metaphysical
differences make an empirical difference.
Indeed, empirical generalizations are not imperiled even by the
existence of real counter-examples, because empirical generalizations and
laws are abstractions and idealizations and not universal claims.
Consider: if we take, say, Galileo's dynamic laws as universally quantified
claims about how objects move in nature, the laws are false. Even
Galileo's own measurements did not agree with his laws, because of the
influence of factors such as wind resistance and friction, as he rightly
recognized. His method was to take measurements in conditions that
progressively lessened the effects of these factors, and extrapolate to an
idealized case: i.e., a case in which we have idealized away from factors
that are at work in vivo. But the problem is not just that Galileo's laws
have exceptions. Taken as claims about how objects really move, they in
fact have no true substitution instances! (Compare Cartwright 1983.) For
subsequent physics has discovered that there are a variety of forces
(gravitation, electromagnetism, strong and weak force) that, unlike
mechanical force, are always present in nature and work over infinite
distances. Therefore if you take laws applying to any one of these forces
and treat them as universal claims about the behavior of objects, you will
get a howling falsehood: objects never behave exactly as though, say,
electromagnetism but not gravity were at work.[9]
The source of the problem here is not with the laws, but with their
philosophical interpretation: laws are not universally-quantified claims
about all token events, but descriptions of idealized cases—that is, of
real-world forces taken in idealization from factors that are present in
vivo. Thus it does not matter if nothing ever behaves exactly in
accordance with Galileo's dynamic laws.
Likewise, exceptions in psychology need not be seen as anomalous in
the sciences. There is both a philosophical and a psychological issue
here. Philosophically, it does not matter for a nomological psychology of
reasoning if there are some zombies walking around. Psychological laws
take our own phenomenologically-pregnant case and idealize away from some
of the features. The resulting description may as a result apply to
zombies as well, but this does not matter. The reason it does not matter
is that zombies are not in the domain that was originally singled out for
psychology. It is an artifact of the acts of abstraction and idealization
that phenomenologically-rich subjects, with their phenomenology idealized
away, are indistinguishable from zombies under the terms of the
idealization. The domain of psychology, however, is not everything that is
indistinguishable from us once you bracket our inner life. Zombies are not
a problem for nomological psychology because they are not part of
psychology's domain to begin with. (Zombies are, of course, a problem for
a theory that attempts to explain conscious phenomenology in terms of
something else. Here, however, the issue is reductive rather than
nomological explanation.)
Similar arguments apply in the case of physical duplicates. The fact
that A and B are physical duplicates does not assure that they have the
same content: two identical rocks do not have the same concepts because
neither has any concepts at all. Mere physical duplication does not assure
that content is in the picture at all. But perhaps this is unfair: the
doppelganger examples claim merely that if you take, not a rock, but one of
us—a being that we all agree has contentful mental states—and duplicate it,
then narrow content will be preserved. Moreover, in spite of the thought-
experiments that show that there is not a logically necessary relation
between phenomenological content and physical form, we all routinely assume
that beings even roughly like ourselves (e.g., other human beings, perhaps
even chimpanzees or gorillas) have roughly similar phenomenological
contents as well. Such assumptions of intersubjectivity are rampant in the
practice of psychophysics as well as in common sense reasoning about
others, and it is an assumption that is reasonable both as common sense and
as scientific methodology. (Not to mention perhaps being ethically
obligatory.)
However, notice the basis of this assumption. What we assume as a
safe empirical generalization is that there is a close and systematic
relationship between phenomenological content and physical comportment as
an empirical fact. Under the scope of this assumption, we are licensed to
make certain (empirical) inferences: for example, that two beings that are
physically identical will be phenomenologically identical as well. Notice
two things about this inference, however. First, its status is one that is
underwritten by the original assumption that there is such an empirical
generalization about phenomenology and physical structure. Thus it is not
a claim about metaphysical sufficiency—or even of material sufficiency. It
is a merely conditional claim. We are reasoning that if there is such a
generalization to be made, then a being like me physically will be like me
phenomenologically as well.[10] We have done nothing—and probably can do
nothing—to assure that this generalization in fact holds in all real cases,
unless perhaps we were to find a way to derive phenomenology from physical
structure the way we derive the gas laws from statistical mechanics.
Second, the generalization does not imply that content itself is a notion
that can be constructed out of physical terms. Given a notion of content
derived from our own experience (i.e., the phenomenological notion) it
makes sense to make an empirical (not a metaphysical) generalization to the
effect that a physical duplicate would in fact be a phenomenological
duplicate. But without this pre-existing phenomenological notion of
content, the idea of physical duplication does not give rise to any notion
of content at all.
In short, there is no independent notion of narrow content in either
physical or functional terms. There are indeed notions of functional and
physical duplication, but these are not themselves notions of content.
First, you can have functional and physical duplication in instances in
which there is no content at all. Second, the only way of picking out the
relevant instances of functional or physical form that count as contentful
is by having a prior notion of content, and the prior notion of content we
have is the phenomenological notion that we are familiar with from our own
case. In doing psychology we are interested in lawlike relationships and
causal mechanisms, and for these it is sometimes useful to look at
functional and physical properties in abstraction from phenomenological
content (though sometimes the phenomenology is essential, as in the case of
psychophysics (cf. Horst 1996)). But these physical and functional
properties do not supply notions of content in their own right. We get the
notion of "narrow content" by starting with a richer notion of content and
then abstracting away features that are irrelevant to particular
explanatory purposes. "Narrow content" is like "point mass"—it is an
artefact of abstraction and idealization that is methodologically useful so
long as we remember that it is a product of idealization, but
metaphysically suspect if we treat it as denoting a property of the world.

6. Summary
Recent discussions of individualism and non-individualism in philosophy
of mind have tended to conflate two distinct individualist notions of
'content' and 'meaning'. The more basic of these, I claim, is what I have
called "phenomenological content," understood as what remains invariant
under methodological solipsism and the phenomenological epochê. This may
be in original intuition behind the notion of "narrow content" as well, but
the latter term has been pushed in an independent direction, of signifying
what is necessarily shared by physical or functional duplicates. However,
the notions of functional or physical similarity are not themselves
sufficiently rich to amount to notions of content at all, and "narrow
content," thus understood, is best regarded as what is left over after a
certain kind of abstraction away from the phenomenology of human mental
states. Phenomenological content, as an individualistic notion, is also
distinct from broad content. However, it is arguably phenomenological
content that combines with ostensive and/or causal relations to fix broad
content.

NOTES
Bibliography


Block, Ned. 1978. "Troubles with Functionalism." In Perception and
Cognition. Issues in the Foundations of Psychology, Minnesota Studies
in the Philosophy of Science, vol. 9, ed. C.W. Savage, Minneapolis:
University of Minnesota Press, 1978.

Block, Ned, ed. 1980. Readings in the Philosophy of Psychology.
Cambridge, Massachusetts: Harvard University Press.

Brentano, Franz. 1874. Psychologie vom empirischen Standpunkt.
Leipzig: Duncker & Humblot. English edition, Psychology from an
Empirical Standpoint. Ed. Linda L. McAlister, trans. D.B. Terrell,
Antos C. Rancurello and Linda L. McAlister. London and New York:
Routledge & Kegan Paul, 1973.

Burge, Tyler. 1979. "Individualism and the Mental." In Studies in
Epistemology, Midwest Studies in Philosophy, vol. 4, ed. P.
French, T. Euhling, and H. Wettstein. . Minneapolis: University
of Minnesota Press, 1979.

Burge, Tyler. 1986. "Individualism and Psychology." Philosophical
Review 95: 1, 3—45.

Cartwright, Nancy. 1983. How the Laws of Physics Lie.
Oxford/Clarendon.

Chalmers, David. 1996. The Conscious Mind. Oxford University Press.
Chisholm, Roderick. 1957. "Intentional Inexistence." In Perceiving: a
Philosophical Study, by Roderick Chisholm. Ithaca: Cornell
University Press.

Fodor, Jerrold. 1980. "Methodological Solipsism Considered as a
Research Strategy in Cognitive Science." Behavioral and Brain Sciences
3:63-73.
Fodor, Jerrold. 1981. Representations. Cambridge, Mass.: Bradford
Books/MIT Press.
Horgan, Terrance and John Tienson. "The Phenomenology of Intentionality
and the Intentionality of Phenomenology." In David J. Chalmers
(ed.), Philosophy of Mind: Classical and Contemporary Readings,
Oxford University Press, 2002.
Horst, Steven. 1996. Symbols, Comptutation and Intentionality: A
Critique of the Computational Theory of Mind. Berkeley and Los
Angeles: University of California Press.

Husserl, Edmund. 1913. Ideen au einer reinen Phänomenologie und
phänomenologischen Philosohie. The Hague: Nijhoff. English edition,
Ideas: General introduction to pure phenomenolgy. Trans. W.R. Boyce
Gibson. Collier Books, 1931.

Jackson, Frank. 1982. "Epiphenomenal Qualia." Philosophical Quarterly
32: 127—136.

Kripke, Saul. 1971. "Naming and Necessity." In Semantics of Natural
Language, ed. D. Davidson and G. Harman, 253-355, 763-769. Dordrecht:
Reidel.

Lycan, William. 1987. Consciousness. Cambridge, Mass.: MIT
Press/Bradford Books.

Nagel, Thomas. 1974. "What is It Like to Be a Bat?" Philosophical
Review 83: 435—450.

Putnam, Hilary. 1975. "The Meaning of 'Meaning'." In Language, Mind
and Knowledge: Minnesota Studies in the Philosophy of Science, 7, ed.
Keith Gunderson. Minneapolis: University of Minnesota Press.

Searle, John. 1992. The Rediscovery of the Mind. Cambridge, Mass.:
MIT Press.


-----------------------

[1] This article was originally drafted while the author was a participant
in the 1993 NEH Summer Institute on Meaning hosted by Jerry Fodor and Ernie
Lepore at Rutgers University. It was shelved and largely forgotten by its
author until he heard Terry Horgan give a paper on a related topic at
another NEH Institute, at the University of California at Santa Cruz,
hosted by David Chalmers and David Hoy, in 2002. Thanks go to the NEH,
Rutgers, U.C. Santa Cruz, and my fellow Institute-goers for the impetus to
write and revise this article. Any errors, infelicities, and embarrassing
philosophical gaffes it may contain are, of course, the sole responsibility
of the author.

[2] It is worth pointing out the assumptions underlying the conclusions
generally drawn from the above thought-experiment. The idea is that
natural-kind terms are used to pick out some privileged kinds (e.g.,
molecular kinds) whose nature may be unknown to the thinker/speaker. It is
equivalent to saying that the rule for using, say, the word 'water' is on
the order of "the molecular kind, whatever it turns out to be, of that
stuff." This may strike the reader as a dubious theory of the ordinary use
of kind-terms. (One might doubt that, on this reading, anyone employed
natural kind terms before about 1850, for instance, and doubt that one's
tailor uses them today.)

[3] I am personally inclined to view this result as a reductio of
theories of content that entail it.

[4] Of course, in order for the state to refer to water, it is not
necessary that one be in a direct causal relation to water at that moment.
Even someone fooled at this moment by a Cartesian demon might have a mental
state whose broad content refers to water if she had been appropriately
related to water in the past.

[5] A similar point is made by Horgan and Tienson (2002), in their
discussion of "phenomenal intentionality." Horgan and Tienson, however, do
not distinguish between what I am calling phenomenological content and
narrow content understood as what is necessarily shared by physical or
functional duplicates. Their point is best understood if interpreted as
being about phenomenological content, and they are correct, in my view, in
holding that there is such an individualistic component at the core of
broad content.

[6] Here I use "material" in the Sellarsian/Brandomesque sense of
inferences based upon semantic relations rather than upon syntactic form.

[7] Husserl was aware of this distinction, and signaled it by pointing out
that not everything in is available to the phenomenological method is
"immanent" in the experience. Husserlian phenomenology (arguably unlike
"phenomenology" in the sense stemming from writers like Thomas Nagel
(1974)) includes everything that can be revealed through the technique of
imaginative variation.

[8] Contrary to widespread misunderstanding, Fodor's writings have never
implied that syntax explains semantics, but merely that it is coordinated
with it. Note the end of the Introduction to Fodor (1981), where, having
outlined the computational theory of mind, he then says that what is still
needed is a theory of how representations represent. This claim would be
nonsensical if the computational theory he had outlined in the previous
pages supplied such a theory. The contingency of the relation is only
explicitly endorsed in Fodor (1993), however, where he is concerned with
reconciling a computational theory of mental processes with a broad theory
of content.

[9] The locus classicus for these observations is Cartwright (1983).

[10] Better, we are working within an idealizing assumption that is
probably never made explicit. Were it made explicit, it would take the
form of such a conditional.