William G. Lycan
University of North Carolina
The 1960s saw heated discussion of Eliminative Materialism in
regard to sensations and their phenomenal features. Thus directed,
Eliminative Materialism is materialism or physicalism plus the distinctive
and truly radical thesis that there have never occurred any sensations;
no one has ever experienced a sensation. This view attracted few
adherents(!), though to this day some philosophers are Eliminativists with
respect to various alleged phenomenal features of sensations.
But in the 1980s the Eliminativist focus shifted
to the propositional attitudes. Eliminativism in regard to those
is materialism plus the claim that no creature has ever had a belief, desire,
intention, hope, wish or the like. Though no one out-and-out accepts
this startling doctrine without qualification, Paul Churchland, Patricia
Churchland, Stephen Stich in some moods, and others have endorsed
it as a plausible conjecture or good bet.
1. Why would anyone take so outrageous an idea seriously
for thirty seconds, much less consider it a good bet? Well, there
are arguments, and some of them seem fairly powerful. Each of the
arguments presupposes the widely held "`Theory' theory" of folk or common-sense
psychological ascriptions. But each of the latter references needs
explaining.
"Folk psychology" (cf. "folk physics," "folk
meteorology") is the set of common-sense generalizations about the mind,
"platitudes" as David Lewis calls them, at least tacitly accepted by almost
everyone and expected by almost everyone to be accepted by almost everyone.
(E.g.: "If a person wants X, and believes that Y is a means of getting
X and has neither any better strategy nor any competing desire or motive,
the person will try to obtain Y," or "People don't like having their toes
stepped on.")
According to the "Theory" theory of folk psychology,
the folk generalizations are at least roughly causal-explanatory in purport;
we use them in the explanation and prediction of human behavior.
And the mental states and events that figure in them are the theoretical
entities posited by the proponents of these explanatory hypotheses.
Thus propositional attitudes are conceived as inner states of people that
supposedly play characteristic roles in causing those people's behavior.
2. Here, in brief, are some of the Eliminativist arguments
against the reality of the attitudes (the arguments overlap somewhat):
(E1) Folk psychology is a failed or at least
bad theory, superseded by a better cognitive science or neuroscience, and
so should be rejected as false.
(E2) Folk psychology will not be "vindicated"
by either cognitive science or neuroscience, and so should be rejected
as false.
(E3) Folk psychology itself posits things
of a kind that will simply not be found in human brains or anywhere else
in nature.
(E4) Folk psychology presupposes supervenient
causation of propositional attitudes by perceptual and other states, and
of behavior by the attitudes. But if the attitudes are supposed to
do their causing in virtue of their propositional contents, then there
is a problem of "explanatory exclusion," leading to a charge of epiphenomenalism
: Suppose a memory, say of singing a wrong note in public, causes
me to wince in embarrassment. The memory presumably causes the wince
by being identical with or at least supervening on a neurophysiological
state N which (together with background conditions) guarantees the wince
by biological law. But then N and the background conditions alone
suffice to cause the wince; the object or content of the memory itself--my
having sung the wrong note--plays no role, and is in that sense epiphenomenal.
(E5) Folk psychology characterizes propositional
attitudes as "wide" features of cognizers--i.e., as features that do not
supervene on their subjects' molecular composition at the time; attitude
contents are determined in part by causal, historical and social factors
outside the subjects' heads; and this makes the attitudes objectionably
un- or contrascientific, and possibly epiphenomenal again. (Cf. also
the more recent problems of "intentional causation.")
(E6) According to a competing line of attack,
attitude contents are determined neither by what is in subjects' heads
nor by causal, historical and social factors; in fact, they are entirely
indeterminate. Can we then still suppose that the attitude contents
are genuine properties of the owner/subject?
3. Unsurprisingly, there are also a number of common arguments
against Eliminativism. I will list those too, in order to emphasize
that my own antiEliminativist case will rely on none of them.
(F1) The "Theory" theory itself is a false
account of folk psychology. (That is not a direct argument against
Eliminativism, but it would undercut every known defense of the view.
Also, I myself accept the "Theory" theory, so I will have no truck with
this argument.)
(F2) We know there are propositional attitudes
because we introspect them in ourselves.
(F3) The posited attitudes are indispensable
to prediction, reasoning, deliberation and understanding, and/or to the
capturing of important macroscopic generalizations, and/or to various less
cognitive pursuits.
(F4) "Cognitive suicide" arguments show that
Eliminativism is somehow self-defeating or self-refuting.
4. I believe both (F2) and (F3) can be fleshed out into
powerful defenses of folk psychology; I think (F4) is hopeless. But
as I have said, my own antiEliminativist argument is distinct from all
these. It is a Moorean argument. Indeed it is a fairly obvious
Moorean argument, and I am surprised not to have heard it given by anyone
previously. (Naturally, I have heard no rebuttal either.)
Remember how G.E. Moore used to argue against
idealists and other anti-realists (as well as to skeptics). In considering
an anti-realist view, he would first draw from it a very specific negative
consequence regarding his own everyday experience. E.g., take the
idealist claim that there are no material objects, as defended by, e.g.,
Bradley and McTaggart. From it, Moore would deduce that he, Moore,
did not have hands at the ends of his arms--hands being near-paradigm cases
of material objects. (Some scholarly work was put in, in order to
assure us that the idealist claim was intended in such a sense as to have
that consequence.)
Now, consider the idealist's defense of the
antirealist claim about material objects. Let us charitably suppose
that the defense had taken the form of a deductively valid argument.
The argument must of course (on pain of regress) have had premises, themselves
undefended. So it is an argument that looks schematically like this:
(P1) ....
(P2) ....
.
.
[steps]
.
.
______________
... (C) There are no material objects.
QED
--to which we may add as a corollary,
... (C') I do not have hands.
By hypothesis, the argument is valid. But that is to say
only that each of the sets {P1,...Pn, ~C} and {P1,...Pn,
~C'} is inconsistent. The idealist of course wants us to accept
P1-Pn and therefore to accept C and reject ~C on the strength of them.
But nothing in the argument itself forces us to do that, since if we wish
to deny its conclusion we have only to reject one of the premises.
Any argument can be turned on its head.
More generally--in fact, throughout philosophy
and every other subject--a deductive "proof" can be no more than an invitation
to compare plausibility: Of the propositions P1,...Pn, and ~C, which is
the least plausible?
Applying that crucial question to any specific
argument for idealism concerning the external world, Moore thought it scarcely
left room for debate. Since ~C, the reality of material objects,
is directly entailed by something Moore already knows to be true (~C',
that he does have hands), the culprit must be one of the other members
of the inconsistent set; it must be one of the premises that is false.
It may be interesting to continue our plausibility survey and decide which
of the Pi is less plausible than the rest; in fact, surely it will be instructive
and illuminating to do that. But that is not necessary in order to
vindicate our common-sense belief in the reality of material objects.
For the latter philosophical purpose, it does not matter which of the Pi
is false. In fact, we do not even have to know what the argument's
premises are exactly; whatever they are, they cannot all be true.
The idealist was doomed from the start.
5. I have deliberately made Moore sound closed-minded, dogmatic,
pigheaded. And many philosophers have rejected his style of argument
on just that ground, finding it obvious that Moore is just begging the
question against his opponent and rather crassly too. But it is important
to see that Moore is doing no such thing. He is only modestly inviting
a plausibility comparison. The comparison is, in effect, between
(a) "Here is one hand and here is another" and (b) a purely philosophical
premise such as McTaggart's assumption that every existing thing has proper
parts that are themselves substances. Come, now: How could
a proposition like (b) be considered as plausible as (a)? How could
I possibly be more certain or more confident that every existent thing
has proper parts that are substances, than that I have hands?
We may wonder where metaphysical premises
(often called "intuitions") come from. Are they deliverances of the
lumen naturale? Does the Third Eye of the metaphysician's mind get
a rare look at a Platonic Form? Perhaps they just articulate features
of our ordinary ways of conceiving certain things. Whatever; their
epistemic credentials are obscure, and more importantly, they are shoddy.
A metaphysician who claims to "just know" that such an abstract premise
is true ("This is a very deep intuition") cannot be taken very seriously.
But Moore has excellent grounds for the competing proposition (a): He remembers
seeing and feeling his hands on millions of occasions, and he can do so
again at will. A forced choice between (a) and (b) has got to favor
(a).
6. Even so, is Moore clinging to his humdrum common-sense
beliefs come what may? Is common sense then sacred and utterly irrefutable?
Is that not precisely what our Founder Socrates taught us to leave behind,
indeed to snicker at? And, remember, it used to be "common sense,"
as obvious as practically anything, that the earth is flat and motionless,
and that the sun rises and gradually travels westward to pass over our
heads.
This is a second standard misinterpretation
of Moore. Moore did not hold that common sense is irrefutable.
Common-sense beliefs can be corrected, even trashed entirely, by careful
empirical investigation and scientific theorizing. (Literal) exploration
and astronomy have teamed up to show that, despite appearances, the earth
is a spheroid that moves around the sun and that "up" means only "away."
So too, chemistry and physics have shown that the most "solid" granite
boulder is actually a region of nearly completely empty space, populated
only by some minuscule and invisible particles racing through it at unimaginable
speeds.
But philosophers (especially idealist philosophers)
are not explorers or scientists. McTaggart provided no evidence
for his claim that every existent thing has proper parts that are substances;
it just seemed true to him, for some reason. Common sense must yield
to evidence, as I have said, but it need not yield to bare metaphysical
pronouncement. Moreover, as is notorious, a priori metaphysical views
historically have little staying power; one philosophical era's fundamental
principles are often rejected in the next era as ludicrous superstition.
No purely philosophical premise can ever (legitimately) have as
strong a claim to our allegiance as can a humble common-sense proposition
such as Moore's autobiographical one. Science can correct common
sense; metaphysics and philosophical "intuition" can only throw spitballs.
7. Which brings me (not before time) back to the admittedly
scientistic doctrine of Eliminative Materialism. I contend that the
Eliminativist is refuted by Moore's technique, in just the same way as
was the temporal idealist.
The argument will now be quite straightforward:
Numerous common-sense mental ascriptions, such as that Granny wants a beer
and believes there is one under the sofa, are individually more plausible,
and always will be more plausible, than are the purely philosophical premises
of any argument designed to convince us to the contrary. As Moore
saw, purely philosophical assumptions have very weak epistemic credentials
and cannot by themselves outweigh simple common-sense facts.
The Eliminativist may protest that her/his
case is not purely philosophical, but rests on scientific considerations
of cognitive psychology, neuroscience, connectionist modelling, and the
like. Indeed, that flaunted feature is what often makes Eliminativism
sound so hip. But this is to misunderstand the Moorean argument
a third time. Moore would not deny that arguments for Eliminativism
contain premises that are endorsed, perhaps simply established, by science.
The point is that each argument also contains at least one purely philosophical
premise. Make no mistake: In order to reach the staggering conclusion
that there has never been a belief, a desire, or any other propositional
attitude, any argument for Eliminativism will have to rest on one or more
a priori principles connecting scientific truths to negative ontology.
And it is terminally unlikely that any such principle could be more credible
for me than that Granny wants beer. Come to think of it, I
want beer.
8. Notice that my Moorean argument is immune to the customary
Churchland-Churchland counterblow, a comparison to alchemy, witchcraft,
and other folk but false theories. However beloved such theories
were to their proponents, they do not qualify as Moorean common sense.
To count as Moorean-commonsensical, a belief must be the sort of belief
that every normal human being holds every day of her/his life, such as
"Here is one hand and here is another" or "I had breakfast before I had
lunch" or "The sun is shining."
Thus, to address the Conference topic directly:
Whatever science can show about the mind, it cannot show that there is
none.
9. I pause to deal with a trenchant objection offered by Ann Wilbur MacKenzie, who reminded me that according to one school of linguistic theory, lexical semantics is an empirical science--in particular, it is claimed, there can be purely empirical evidence for propositions to the effect that a predicate F1 analytically implies another predicate F2. Now, suppose we were to get empirical evidence that, say, "S believes that p" entails something dubious, say, "S's brain has a live turtle in it." Then we would have a compelling argument for Eliminativism about beliefs, with no philosophical premises. (Up till now I have tacitly been assuming that lexical semantics is not empirical but philosophical, and that any such entailment claim is a tendentious philosophical thesis.) The argument would run as follows:
(1) [Empirical data e1...en]
... (2) "S believes that p" entails "S's brain has a live turtle in it." [Ampliative scientific inference from e1...en]
... (3) "If S believes that p, then S's brain has a live turtle in it" is true. [From 2, by Corresponding Conditional theorem]
... (4) If S believes that p, then S's brain has a live turtle in it. [3, Tarski]
(5) No one's brain has a live turtle in it. [Empirical fact]
... (6) S's brain does not have a live turtle in it. [5]
... (7) S does not believe that p. [4,6]
... (8) No one has ever believed anything.
[Universal Generalization from 9, since "S" and "p" were
arbitrary names]
QED
--and none of this argument's premises is contentious in a distinctively
philosophical way.
My own position is that lexical semantics
is not empirical, at least not to the degree intended, i.e., that there
is no uncontroversial form of ampliative inference from any robust body
of publicly observable data to semantic claims such as (2), and further,
that such semantic claims are indeed philosophical in my somewhat pejorative
sense. My reason is Quinean: I do not accept any principled distinction
between allegedly analytic truths and merely widespread common-sense beliefs,
or between purely lexical implications and material inferences.
There is no space here to rehearse the arguments for and against this skepticism.
Instead, for the sake of discussion I shall abandon the Quinean position
and grant the existence in principle of analytic lexical implications,
but argue that Eliminativism is in trouble nonetheless.
Suppose, then, that there are lexical implications
that can be discovered empirically by linguists. But my example is
obviously extreme, for surely "S believes that p" does not analytically
imply "S's brain has a live turtle in it." I chose that example to
illustrate how a well-established empirical fact could be used to support
Eliminativism without the aid of philosophical premises; the consequent
of the Eliminativist's corresponding conditional is supposed to be a proposition
that science (unaided) can discover to be false. But at the same
time it should not be so garishly false as "S's brain has a live turtle
in it." The conditional will presumably have the form "If S believes
that p, then S is in a state of type F," but the nonexistence of F-states
will not be so obvious or so easy for science to establish.
The trouble is that there is a tradeoff here,
to wit, an inverse proportion between the scientific impeachability of
the consequent and the apparent analyticity of the conditional itself.
In my turtle example, the consequent is eminently impeachable, but the
conditional is ludicrous. But at the other end of the spectrum, if
we have a plausible candidate for an analytic conditional, the conditional's
consequent will be pretty safe from science. Take "If S believes
that p, then if other things are equal, S is disposed to assert that p,"
which I suppose would be analytic if any such conditional were. It
is hard to imagine how science (alone) might falsify "If other things are
equal, S is disposed to assert that p"--and remember that on MacKenzie's
strategy, it is science alone that must falsify the consequent.
Consider an intermediate candidate: "If S
believes that p, then some state of S's brain has the content that p."
Though it is hard to see how science alone might impugn that conditional's
consequent, perhaps it is possible. But the conditional is hardly
analytic, for "S believes that p" does not mean anything about any brain.
Let us move back along the spectrum towards analyticity by deleting the
anatomical reference: "If S believes that p, then some inner state of S
has the content that p." Now it is harder to imagine science's falsifying
the consequent, because for all the conditional asserts, the "inner state"
may be global and utterly diffuse or distributed. But at the same
time, even this weaker conditional is not a plausible candidate for analyticity:
Behaviorists, Wittgensteinians and such who deny that beliefs are or require
inner states were not simply contradicting themselves.
Thus a dilemma: For any conditional
premise mobilized by the Eliminativist, if that conditional is plausibly
taken as analytic, its consequent seems immune to purely scientific refutation--and
contrapositively, if its consequent risks such refutation it is pretty
plainly not analytic.
Of course, the believer in analyticity need
not hold that all analyticities are obvious. Perhaps a conditional
like "If S believes that p, then some inner state of S has the content
that p" is indeed deeply analytic even though it does not seem so and the
Behaviorists et al. do not seem to be contradicting themselves. But
this possibility is of no use to the Eliminativist. For a deep, unobvious
analyticity has no epistemic power even if it makes for metaphysical necessity.
Even if there is a hidden contradiction in the idea of someone's believing
that p without being in an internal state of some scientifically vulnerable
kind, that does not help the Eliminativist use the relevant negative scientific
discovery to overturn common-sense belief ascriptions; if we are to be
convinced that we should give up on belief, the contradiction would have
to be brought out by further argument.
For these reasons, I do not think MacKenzie's
strategy can succeed.
10. My antiEliminativist argument has one sobering feature:
that of seeming to prove too much. If I am right, then it seems there
simply could not be any successful philosophical argument for Eliminativism,
and since there is unlikely to be any nonphilosophical argument for Eliminativism
at all, the doctrine is systematically indefensible. Yet (contra
the "cognitive suicide" arguments (F4)) it is perfectly coherent, indeed
empirical. How can a thesis be both empirical and seemingly impossible
to defend?
Theoretically, common-sense views can conflict
not just with philosophical assumptions but directly (neglecting the laws
of logic) with each other. (A candidate triple might be "People autonomously
perform many entirely free actions," "Actions are physical events," and
"Every physical event has a determining, nomologically sufficient physical
cause," though I myself do not think any of those three propositions qualifies
as commonsensical.) So perhaps some common-sense views must face
plausibility comparisons with each other, and one must lose, thus being
defeated not by science but by philosophy-plus-other-common-sense.
But it seems unlikely that our folk psychological
ascriptions would meet that unusual fate. Are we, then, stuck with
the odd contention that folk psychology is an empirical but also irrefutable
theory?
There is a halfway house. Recall "solid."
Chemistry and physics showed us something very startling and contra-commonsensical
about our granite boulder, something which can fairly be expressed in eliminative
style by saying that the boulder is not solid at all. Yet of course
the boulder is solid: It is impenetrable, hard, and obdurate, and if one
chips at it or splits it open one finds only more boulder inside, without
perceptible gaps. There is a difference between being solid in this
operational sense and being solid in the sense of being filled through
and through (to the real continuum) with stuff. Science has forced
us to distinguish those two paronymous meanings of the word.
So too, perhaps, science may force us to distinguish a fairly superficial,
operational sense of "believe" and the rest from a more commissive sense.
But no Eliminativist argument has yet been convincing enough to do that
either.
FOOTNOTES