Vous êtes sur la page 1sur 12

Eliminative Materialism

First published Thu May 8, 2003; substantive revision Tue Apr 16, 2013
Eliminative materialism (or eliminativism) is the radical claim that our ordinary, common-sense
understanding of the mind is deeply wrong and that some or all of the mental states posited by
common-sense do not actually exist. Descartes famously challenged much of what we take for
granted, but he insisted that, for the most part, we can be confident about the content of our own
minds. Eliminative materialists go further than Descartes on this point, since they challenge the
existence of various mental states that Descartes took for granted.
1. A Brief History
2. Contemporary Eliminative Materialism
2.1 Folk Psychology and the Theory-Theory
2.2 Eliminative Theory Change
3. Arguments For Eliminative Materialism
3.1 General Theoretical Problems With Folk Psychology
3.2 Specific Problems With Folk Psychology
3.3 Eliminative Materialism and the Phenomenal
4. Arguments Against Eliminative Materialism
4.1 The Self-Refutation Objection
4.2 Rejecting the Theory-Theory
4.3 Defending the Virtues of Folk Psychology
4.4 Eliminativism Eliminated?
5. Concluding Comments
Bibliography
Cited Works
Further Readings
Academic Tools
Other Internet Resources
Related Entries

1. A Brief History
In principle, anyone denying the existence of some type of thing is an eliminativist with regard to
that type of thing. Thus, there have been a number of eliminativists about different aspects of
human nature in the history of philosophy. For example, hard determinists like Holbach (1770) are
eliminativists with regard to free will because they claim there is no dimension of human
psychology that corresponds to our commonsense notion of freedom. Similarly, by denying that
there is an ego or persisting subject of experience, Hume (1739) was arguably an eliminativist about
the self. Reductive materialists can be viewed as eliminativists with respect to an immaterial soul.
Nevertheless, contemporary eliminative materialismthe sort of eliminativism that denies the
existence of specific types of mental statesis a relatively new theory with a very short history. The
term was first introduced by James Cornman in a 1968 article entitled On the Elimination of
Sensations and Sensations (Cornman, 1968). However, the basic idea goes back at least as far as
C.D. Broad's classic, The Mind and its Place in Nature (Broad, 1925). Here Broad discusses, and
quickly rejects, a type of pure materialism that treats mental states as attributes that apply to
nothing in the world (pp. 607611). Like many future writers (see section 4.1 below), Broad argued
that such a view is self-contradictory since it (presumably) presupposes the reality of misjudgments
which are themselves a type of mental state.

Apart from Broad's discussion, the main roots of eliminative materialism can be found in the
writings of a number of mid-20th century philosophers, most notably Wilfred Sellars, W.V.O.
Quine, Paul Feyerabend, and Richard Rorty. In his important 1956 article, Empiricism and the
Philosophy of Mind, Sellars introduced the idea that our conception of mentality may be derived
not from direct access to the inner workings of our own minds, but instead from a primitive
theoretical framework that we inherit from our culture. While Sellars himself regarded this
theoretical framework as empirically correct, his claim that our conception of the mind is theorybased, and at least in principle falsifiable, would be influential to later supporters of eliminativism.
In articles such as Mental Events and the Brain (1963), Paul Feyerabend explicitly endorsed the
idea that common-sense psychology might prove to be radically false. Indeed, Feyerabend held that
practically any version of materialism would severely undermine common-sense psychology. Like
many of his contemporaries, Feyerabend argued that common-sense mental notions are essentially
non-physical in character. Thus, for him, any form of physicalism would entail that there are no
mental processes or states as understood by common-sense (1963, p. 295).
Like Feyerabend, Quine also endorsed the idea that mental notions like belief or sensation could
simply be abandoned in favor of a more accurate physiological account. In a brief passage in Word
and Object (1960), Quine suggests that terms denoting the physical correlates of mental states will
be more useful and, as he puts it, [t]he bodily states exist anyway; why add the others? (p. 264).
However, Quine goes on to question just how radical an eliminativist form of materialism would
actually be, implying no significant difference between explicating mental states as physiological
states, and eliminating mental state terms in favor of physical state terms. He asks, Is physicalism a
repudiation of mental objects after all, or a theory of them? Does it repudiate the mental state of
pain or anger in favor of its physical concomitant, or does it identify the mental state with a state of
the physical organism (and so a state of the physical organism with the mental state) (p. 265)?
Quine answers this question by rejecting it, suggesting there is no interesting difference between the
two cases: Some may therefore find comfort in reflecting that the distinction between an
eliminative and an explicative physicalism is unreal (p. 265).
Here we see a tension that runs throughout the writings of many early eliminative materialists. The
problem involves a vacillation between two different conditions under which mental concepts and
terms are dropped. The first scenario proposes that certain mental concepts will turn out to be
empty, with mental state terms referring to nothing that actually exists. Historical analogs for this
way of understanding eliminativism are cases where we (now) say it turned out there are no such
things, such as demons and crystal spheres. The second scenario suggests that the conceptual
framework provided by neurosciences (or some other physical account) can or should come to
replace the common-sense framework we now use. Unlike the first scenario, the second allows that
mental state terms may actually designate something realit's just that what they designate turn out
to be brain states, which will be more accurately described using the terminology of the relevant
sciences. One possible model for this way of thinking about eliminativism might be the
discontinuance of talk about germs in favor of more precise scientific terminology of infectious
agents. Given these two different conceptions, early eliminativists would sometimes offer two
different characterizations of their view: (a) There are no mental states, just brain states and, (b)
There really are mental states, but they are just brain states (and we will come to view them that
way).
These alternative ways of understanding eliminative materialism produced considerable confusion
about what, exactly, eliminative materialism entailed. Moreover, since it was difficult to see how the
second version was significantly different from various forms of reductive materialism (hence,
Quine's skepticism about the difference between elimination and explication) it also raised doubts
about the distinctiveness of eliminative materialism.
Much of this was brought to light in the discussion generated by an influential 1965 article by
Richard Rorty entitled, Mind-Body Identity, Privacy, and Categories. Rorty's so-called

disappearance theory appeared to openly endorse both conceptions of eliminative materialism,


suggesting that sensations do not actually exist and that they are nothing but brain processes (p. 28).
As one might expect, the ensuing discussion focused on getting clear on what Rorty's theory
actually claimed. For example, Cornman's article introducing the phrase eliminative materialism
claimed that Rorty was arguing that talk about sensations denotes brain states in much the same way
that talk about Zeus's thunderbolts (allegedly) denotes electrical discharges. Unfortunately, besides
suggesting a questionable perspective on reference, this interpretation raised further questions about
what distinguished eliminativism from reductionism. In one helpful article by William Lycan and
George Pappas (1972)entitled, appropriately enough, What Is Eliminative Materialism?the
authors convincingly argued that you can't have it both ways. You can either claim that common
sense mental notions do not pick out anything real and that mental terms are empty, in which case
you are a true eliminative materialist; or you can claim that mental notions can be, in some way,
reduced to neurological (or perhaps computational) states of the brain, in which case you are really
just a good-old fashioned materialist/reductionist. In a follow-up article, Steven Savitt (1974)
introduced the distinction between ontologically conservative (reductive) and ontologically radical
(eliminative) theory change, which helped to further clarify and distinguish the central claims of
eliminative materialism as it is understood today.
In more recent history, eliminative materialism has received attention from a broader range of
writers, including many concerned not only with the metaphysics of the mind, but also the process
of theory change, the status of semantic properties, the nature of psychological explanation and
recent developments in cognitive science. Much of this attention has been fostered by the husbandwife team of Paul and Patricia Churchland, whose writings have forced many philosophers and
cognitive scientists to take eliminativism more seriously. In his 1981 article, Eliminative
Materialism and the Propositional Attitudes, Paul Churchland presents several arguments in favor
of dropping commonsense psychology that have shaped the modern debate about the status of
ordinary notions like belief. Patricia Churchland's provocative 1986 book, Neurophilosophy,
suggests that developments in neuroscience point to a bleak future for commonsense mental states.
Another influential author has been Stephen Stich. His important 1983 book, From Folk Psychology
to Cognitive Science: The Case Against Belief, argues that even conventional computational
psychologywhich is often assumed to vindicate common-sense psychologyshould reject
taxonomies for cognitive states that correspond with belief-desire psychology. These authors' views
are discussed in more detail in Sections 3 and 4 below.

2. Contemporary Eliminative Materialism


Modern versions of eliminative materialism claim that our common-sense understanding of
psychological states and processes is deeply mistaken and that some or all of our ordinary notions
of mental states will have no home, at any level of analysis, in a sophisticated and accurate account
of the mind. In other words, it is the view that certain common-sense mental states, such as beliefs
and desires, do not exist. To establish this claim, eliminativists typically endorse two central and
controversial claims which we will examine below. Much of our discussion will focus upon our
notion of belief, since it figures so prominently in contemporary discussions of eliminative
materialism. However, many of the arguments presented below are thought to generalize to other
mental notionsespecially other propositional attitudes.

2.1 Folk Psychology and the Theory-Theory


The standard argument for eliminative materialism begins with the Sellarsian thesis that we employ
a theoretical framework to explain and predict intelligent behavior. Because this position claims that
we use a theory when employing mental idiom, it is often referred to as the theory-theory (see the
entry on folk psychology as a theory), and is endorsed not only by eliminative materialists, but by
many realists about mental states as well (like Sellars). Folk psychology is assumed to consist of
both generalizations (or laws) and specific theoretical posits, denoted by our everyday

psychological terms like belief or pain. The generalizations are assumed to describe the various
causal or counterfactual relations and regularities of the posits. For instance, a typical example of a
folk psychological generalization would be:
If someone has the desire for X and the belief that the best way to get X is by doing Y,
then (barring certain conditions) that person will tend to do Y.
Advocates of the theory-theory claim that generalizations like these function in folk psychology
much like the laws and generalizations of scientific theories. At the same time, many theorytheorists allow that the laws of folk psychology are learned more informally than scientific theories,
as part of our normal development (see, for example, P. M. Churchland, 1981 and Lewis, 1972).
According to theory-theorists, the posits of folk psychology are simply the mental states that figure
in our everyday psychological explanations. Theory-theorists maintain the (controversial) position
that, as theoretical posits, these states are not directly observed, though they are thought to account
for observable effects like overt behavior. Theory-theorists also claim that common-sense assigns a
number of properties to these states, such as causal, semantic and qualitative features. For instance,
the theory-theory claims common-sense assigns two sorts of properties to beliefs. First, there are
various causal properties. Beliefs are the sort of states that are caused in certain specific
circumstances, interact with other cognitive states in various ways, and come to generate various
sorts of behavior, depending on the agent's other desires and mental states. As functionalists have
claimed, these causal roles appear to define our ordinary notion of belief and distinguish them from
other types of mental states. Second, beliefs have intentionality; that is, they each express a
proposition or are about a particular state of affairs. This inherent intentionality (also called
meaning, content, and semantic character), is commonly regarded as something special about
beliefs and other propositional attitudes. Moreover, as we will see below, it is also a popular target
of eliminative materialists who challenge the propriety and explanatory value of beliefs.

2.2 Eliminative Theory Change


The second component of eliminative materialism is the thesis that folk psychology is profoundly
wrong about the actual nature of the mind/brain. Eliminative materialists argue that the central
tenets of folk psychology radically misdescribe cognitive processes; consequently, the posits of folk
psychology pick out nothing that is real. Like dualists, eliminative materialists insist that ordinary
mental states can not in any way be reduced to or identified with neurological events or processes.
However, unlike dualists, eliminativists claim there is nothing more to the mind than what occurs in
the brain. The reason mental states are irreducible is not because they are non-physical; rather, it is
because mental states, as described by common-sense psychology, do not really exist.
To see this a little better, it will help to return to the important distinction made by Steven Savitt
discussed in Section 1 between ontologically conservative (or retentive) theory change on the one
hand, and ontologically radical (or eliminative) theory change on the other hand. Ontologically
conservative theory change occurs when the entities and posits of the replaced theory are relocated,
often with some degree of revision, in the replacing theory. For example, as our theory of light was
gradually replaced by our understanding of electro-magnetic radiation, our conception of light was
dramatically transformed as we recognized ways in which our old conception was mistaken or
incomplete. Nevertheless, at no point did we come to say that there is really no such thing as light.
Rather, light was eventually identified with a form of electro-magnetic radiation.
By contrast, our notion of demons did not come to find a new home in contemporary theories of
mental disorder. There is nothing in the theories of schizophrenia, Tourette's Syndrome, neuropathology or any of the other modern explanations for bizarre behavior, that we can sensibly
identify with malevolent spirits with supernatural powers. The notion of a demon is just too far
removed from anything we now posit to explain behavior that was once explained by demonology.
Consequently, the transition from demonology to modern accounts of this behavior was

ontologically radical. We dropped demons from our current ontology, and came to realize that the
notion is emptyit refers to nothing real.
Eliminative materialists claim that an ontologically radical theory change of this sort awaits the
theoretical posits of folk psychology. Just as we came to understand that there are no such things as
demons (because nothing at all like demons appear in modern accounts of strange behavior), so too,
eliminative materialists argue that various folk psychological conceptslike our concept of
beliefwill eventually be recognized as empty posits that fail to correspond with anything that
actually exists. Since there is nothing that has the causal and semantic properties we attribute to
beliefs (and many other mental states) it will turn out that there really are no such things.
A somewhat similar framework for understanding eliminative materialism is provided by David
Lewis's discussion of functional definitions in psychology (1972) (see the entry on functionalism).
In Lewis's account, our commonsense mental notions can be treated as functionally defined
theoretical terms that appear in a chain of Ramsey-sentences. The Ramsey-sentences are a formal
reconstruction of the platitudes of commonsense psychology. They provide a set of roles or
conditions that more or less must be met for the instantiation of any given state. If nothing comes
close to actually filling the roles specified by this framework for a certain state, then we are
warranted in saying that the theoretical posit in question doesn't refer and there is no such thing.
Eliminative materialists claim that this is precisely what will happen with at least some of our folk
mental notions.

3. Arguments For Eliminative Materialism


Because eliminative materialism is grounded in the claim that common sense psychology is
radically false, arguments for eliminativism are generally arguments against the tenability of folk
psychology. These arguments typically fall into one of two major families. One family involves
arguments stemming from a broad range of considerations that pertain to the assessment of theories
in general. The second family focuses upon deficiencies that are unique to folk psychology and its
central posits.

3.1 General Theoretical Problems With Folk Psychology


Patricia and Paul Churchland have offered a number of arguments based on general considerations
about theory evaluation. For example, they have argued that any promising and accurate theory
should offer a fertile research program with considerable explanatory power. They note, however,
that common-sense psychology appears to be stagnant, and there is a broad range of mental
phenomena that folk psychology does not allow us to explain. Questions about why we dream,
various aspects of mental illness, consciousness, memory and learning are completely ignored by
folk psychology. According to the Churchlands, these considerations indicate that folk psychology
may be in much worse shape than we commonly recognize (P. M. Churchland, 1981; P.S.
Churchland, 1986). Another argument that appeals to general theoretical considerations offers an
inductive inference based on the past record of folk theories. Folk physics, folk biology, folk
epidemiology and the like all proved to be radically mistaken. Since folk theories generally turn out
to be mistaken, it seems quite improbable that folk psychology will turn out true. Indeed, since folk
psychology concerns a subject that is far more complex and difficult than any past folk theory, it
seems wildly implausible that this one time we actually got things right (Churchland, P.M. 1981).
These general theoretical arguments do not seem to have significantly undermined the intuitive
support that folk psychology enjoys. In response to the charge that folk psychology is stagnant,
many have argued that this assessment is unfair, and that folk psychology has actually stimulated a
number of fruitful research programs in scientific psychology (Greenwood, 1991; Horgan and
Woodward, 1985). Moreover, defenders of folk psychology note that it hardly follows from the
observation that a given theory is incomplete, or fails to explain everything, that it is therefore
radically false (Horgan and Woodward, 1985). Defenders of folk psychology object that these

theoretical considerations cannot outweigh the evidence provided by everyday, ordinary experience
of our own minds, such as our introspective experience, which seems to vividly support the reality
of mental states like beliefs.
Regarding this last point, eliminativists like the Churchlands warn that we should be deeply
suspicious about the reliability of introspective evidence about the inner workings of the mind. If
inner observation is as theory-laden as many now suppose outer perception to be, what we
introspect may be largely determined by our folk psychological framework. In other words,
introspecting beliefs may be just like people seeing demonic spirits or celestial spheres
(Churchland, P.M., 1988). This skepticism about the reliability of introspection is bolstered by
empirical work that calls into question the reliability of introspection (Nisbett and Wilson, 1977).

3.2 Specific Problems With Folk Psychology


The second family of eliminative materialist arguments focuses upon idiosyncratic features of folkpsychological posits and ultimately denies that these features will be accommodated by a scientific
account of the mind. The most widely discussed features are two associated with the apparent
linguistic nature of beliefs and other propositional attitudes. First, as a number of philosophers have
recently noted, propositional attitudes appear to have a form similar to public language sentences,
with a compositional structure and syntax. For example, a person's belief that, say, the president
dislikes terrorists appears to be composed of the concepts THE PRESIDENT, DISLIKES, and
TERRORISTS, and differs from the belief that terrorists dislike the president by virtue of
something analogous to syntactic arrangement. Second, beliefs resemble public sentences in that
they have semantic properties. Beliefs, like public linguistic representations, are about different
states of affairs. Both of these quasi-linguistic features of propositional attitudestheir alleged
sentential structure and their semantic (or intentional) propertieshave been used by philosophers
to mount arguments for eliminativism.
3.2.1 Challenging The Syntactic Structure Of Beliefs
Some writers have emphasized the apparent mismatch between the sentential structure of
propositional attitudes on the one hand, and the actual neurological structures of the brain on the
other hand. Whereas the former involves discrete symbols and a combinatorial syntax, the latter
involves action potentials, spiking frequencies and spreading activation. As Patricia Churchland
(1986) has argued, it is hard to see where in the brain we are going to find anything that even
remotely resembles the sentence-like structure that appears to be essential to beliefs and other
propositional attitudes.
In response to this line of reasoning, many have argued that it is mistake to treat folk psychology as
committed to a quasi-linguistic structure to propositional attitudes (Horgan and Graham, 1991;
Dennett, 1991). And even for those who find this reading of folk psychology plausible, there is a
further difficulty regarding the relevance of neuroscience for determining the status of folk
psychology. Some, such as Zenon Pylyshyn (1984), have insisted that just as the physical circuitry
of a computer is the wrong level of analysis to look for computational symbol structures, so too, the
detailed neurological wiring of the brain is the wrong level of organization to look for structures
that might qualify as beliefs. Instead, if we view the mind as the brain's program, as many advocates
of classical AI do, then folk posits exist at a level of analysis that is more abstract than the neurophysical details. Consequently, many realists about the posits of folk psychology discount the
importance of any apparent mis-match between neurological architecture and the alleged linguistic
form of propositional attitudes (Fodor & Pylyshyn, 1988; McLaughlin & Warfield, 1994).
3.2.2 Challenging the Semantic Properties of Beliefs
The second type of argument against beliefs focuses upon their semantic properties and concludes
that these sorts of properties make propositional attitudes ill-suited for even a computational theory

of the mind. Stephen Stich (1983) has emphasized that folk psychology individuates beliefs by
virtue of their semantic properties, e.g., we taxonomize states like beliefs by virtue of what they are
about. However, according to Stich, there are a host of reasons for rejecting a semantic taxonomy
for scientific psychology. Semantic taxonomies ignore causally salient aspects of cognitive states,
involve a high degree of vagueness, and break down in the case of the mentally ill or the very
young. In place of the semantic individuation method adopted by folk psychology, Stich argues for
a syntactic taxonomy that is based upon the causally relevant syntactic or physical properties of a
given cognitive state.
Yet, as Stich himself notes, even if it should turn out that folk posits do not belong in a scientific
psychology, more is needed to establish that they do not actually exist. After all, we do not doubt
the existence of several sorts of things (e.g., chairs, articles of clothing) that are defined in ways that
make them ill-suited for science. If our best scientific account posited states that share many
features with beliefs, such as similar causal roles, then even if the two taxonomies pulled apart in
certain cases, we may still regard folk psychology as, in some sense, vindicated. While the scientific
taxonomy may not list beliefs as basic cognitive states, it could conceivably still provide the
resources for developing a realist interpretation of these and other folk psychological states.
3.2.3 Eliminativism and Cognitive Science
One way to get a stronger eliminativist conclusion would be to argue that there is nothing posited in
a scientific account of cognition that shares the central properties we attribute to folk psychological
states, at any level of analysis. For example, Ramsey, Stich and Garon (1990) have argued that if
certain connectionist models of memory and inference prove successful, then this would form the
basis for eliminative materialism regarding states like propositional memories. Since some
connectionist models store information in a highly distributed manner, there are no causally
discrete, semantically evaluable data structures that represent specific propositions. It is not just that
these models lack the sort of sentential, compositional representations assumed in more traditional
(or language of thought) models. Rather, it is that in these networks there are no causally distinct
structures that stand for specific states of affairs. Consequently, there do not appear to be any
structures in these networks that might serve as candidates for beliefs and other propositional
attitudes. This is noteworthy since many critics of eliminativism claim it is virtually impossible to
imagine what a psychological theory would look like that doesn't invoke propositional attitudes to
explain cognition (Hannan, 1993). If Ramsey, Stich and Garon are right, certain connectionist
models may, for the first time, provide us with a plausible account of cognition that supports the
denial of belief-like states. More recently, Ramsey (2007) has argued that this earlier argument does
not go far enough, insisting that connectionist models of this sort not only fail to invoke inner
representations that are sufficiently similar to the posits of folk psychology, but that they don't
actually invoke inner representational states at all.
Ramsey, Stich and Garon's argument assumes that in highly distributed networks, it is impossible to
specify the semantic content of elements of the network that are causally responsible for various
cognitive episodes. Some have responded to their argument by suggesting that, with highly
sophisticated forms of analysis, it actually is possible to pick out causally relevant pieces of stored
information (Forster and Saidel, 1994). Others have argued that, like the Churchlands, Ramsey,
Stich and Garon have offered a mistaken interpretation of folk psychology, suggesting it requires far
less in the way of explicit, discrete structures than they suggest (Dennett, 1991; Heil, 1991). This is
a common criticism of eliminative materialism, and we will look at it more closely in Section 4.3.
Another development in cognitive science that has pushed some people in the direction of
eliminativism is the attempt to understand cognitive systems as neither classical nor connectionist
computational devices, but rather as dynamic systems, described using the mathematical framework
of dynamic systems theory (Beer, 2000; van Gelder, 1992; Port and van Gelder, 1995). This
approach is often conjoined with some version of embodied cognition, as both place a strong
emphasis on the way cognitive agents move about and interact with their environment. While

neither the dynamic nor the embodied approaches are inherently anti-representational in nature, at
least some authors have employed them to develop accounts of cognitive processes that abandon
inner representational states altogether. For example, Anthony Chemero has promoted what he calls
radical embodied cognitive science (Chemero, 2009). This theoretical framework treats the
cognitive agent and environment as a complex coupled system best explained by a mix of dynamics
and James Gibson's ecological theory of perception (Gibson, 1950). Chemero explicitly endorses
eliminativism by rejecting the traditional assumption that agents solve problems and navigate
through the world by consulting mental representations. He thus joins others in the cognitive
science community, like artificial intelligence researcher Rodney Brooks (Brooks, 1991), who have
tried to account for cognition without invoking representational entities. Of course, it is too early to
know how successful these non-representational approaches will ultimately be, and many defenders
of representationalism argue that these efforts are not likely to account for more sophisticated
representation-hungry tasks like planning (Clark and Toribio, 1994).

3.3 Eliminative Materialism and the Phenomenal


Although most discussions regarding eliminativism focus on the status of our notion of belief and
other propositional attitudes, some philosophers have endorsed eliminativist claims about the
phenomenal or qualitative states of the mind (see the entry on qualia). For example, Daniel Dennett
(1978) has argued that our concept of pain is fundamentally flawed because it includes essential
properties, like infallibility and intrinsic awfulness, that cannot co-exist in light of a welldocumented phenomenon know as reactive disassociation. In certain conditions, drugs like
morphine cause subjects to report that they are experiencing excruciating pain, but that it is not
unpleasant. It seems we are either wrong to think that people cannot be mistaken about being in
pain (wrong about infallibility), or pain needn't be inherently awful (wrong about intrinsic
awfulness). Dennett suggests that part of the reason we may have difficulty replicating pain in
computational systems is because our concept is so defective that it picks out nothing real. A similar
view about pain has been offered by Valerie Hardcastle (1999). Hardcastle argues that the
neurological basis for pain sensations is so complex that no one thing answers to our folk
conception. However, despite her own characterication of pain as a myth, Hardcastle's arguments
appear to be aimed not at showing that pain is unreal, but rather that it is actually a more
complicated phenomenon than suggested by our folk conception.
In another well-known article, Quining Qualia (1988), Dennett challenges not just our conception
of pain, but all of our different notions of qualitative states. His argument focuses on the apparently
essential features of qualia, including their inherent subjectivity and their private nature. Dennett
discusses several casesboth actual and imaginaryto expose ways in which these ordinary
intuitions about qualia pull apart. In so doing, Dennett suggests our qualia concepts are
fundamentally confused and fail to correspond with the actual inner workings of our cognitive
system.
Some writers have suggested an eliminativist outlook not just with regard to particular states of
consciousness, but with regard to phenomenal consciousness itself. For example, Georges Rey
(1983, 1988) has argued that if we look at the various neurological or cognitive theories of what
consciousness might amount to, such as internal monitoring or the possession of second-order
representational states, it seems easy to imagine all of these features incorporated in a
computational device that lacks anything we intuitively think of as real or robust consciousness.
Rey suggests that the failure of these accounts to capture our ordinary notion of consciousness may
be because the latter corresponds with no actual process or phenomenon; the inner light we
associate with consciousness may be nothing more than a remnant of misguided Cartesian intuitions
(see also Wilkes, 1988; 1995).

4. Arguments Against Eliminative Materialism


Like any theory that challenges our fundamental understanding of things, eliminative materialism
has been subjected to a variety of criticisms. Here, I'll discuss four that have received considerable
attention in recent years.

4.1 The Self-Refutation Objection


Many writers have argued that eliminative materialism is in some sense self-refuting (Baker, 1987;
Boghossian, 1990, 1991; Reppert, 1992). A common way this charge is made is to insist that a
capacity or activity that is somehow invoked by the eliminativist is itself something that requires the
existence of beliefs. One popular candidate for this activity is the making of an assertion. The critic
insists that to assert something one must believe it. Hence, for eliminative materialism to be
asserted as a thesis, the eliminativist herself must believe that it is true. But if the eliminativist has
such a belief, then there are beliefs and eliminativism is thereby proven false.
Eliminativists often respond to this objection by first noting that the bare thesis that there are no
beliefs is not itself contradictory or conceptually incoherent. So properly understood, the complaint
is not that eliminative materialism (qua-proposition) is self-refuting. Rather, it is that the
eliminativist herself is doing something that disconfirms her own thesis. In the above example, the
disconfirming act is the making of an assertion, as it is alleged by the critic that we must believe
anything we assert with public language. However, this last claim is precisely the sort of folkpsychological assumption that the eliminative materialist is suggesting we should abandon.
According to eliminative materialism, all of the various capacities that we now explain by appealing
to beliefs do not actually involve beliefs at all. So the eliminativist will hold that the self-refutation
critics beg the question against eliminative materialism. To run this sort of objection, the critic
endorses some principle about the necessity of beliefs which itself presupposes that eliminative
materialism must be false (P. S. Churchland, 1986; Cling, 1989; Devitt, 1990; Ramsey, 1991).
A more sophisticated version of the self-refutation ojection has been offered by Paul Boghossian
with regard to eliminativist arguments based on the content of psychological states. Boghossian
maintains that arguments for irrealism about the content of propositional attitudes work just as well
in support of irrealism about all forms of content, including the content of ordinary linguistic
expressions. Moreover, he argues that different forms of irrealism about linguistic content
presuppose robust semantic notions, such as realist conceptions of truth and reference. This leads to
the incoherent position that, for example, there are no truth conditions and yet certain sentences (or
beliefs) about content are false (Boghossian, 1990, 1991). In response, Michael Devitt and Georges
Rey argue that Boghossian's argument, despite its sophistication, nevertheless begs the question by
ascribing to the eliminativist some version of truth-conditional semantics, whereas many
eliminativists would reject such a view of linguistic expressions. While eliminativists would need to
construct some sort of non-truth-conditional semantics, Devitt and Rey argue that the challenge of
such a project reveals only that eliminativism is implausible, not that it is, as Boghossian claims,
incoherent (Devitt, 1990; Devitt and Rey, 1991).

4.2 Rejecting the Theory-Theory


In section 2, we saw that eliminative materialism typically rests upon a particular understanding of
the nature of folk psychology. The next criticism of eliminative materialism challenges the various
characterizations of folk psychology provided by its advocatesin particular the view set forth by
advocates of the theory-theory. This criticism comes from two very distinct traditions. The first
tradition is at least partly due to the writings of Wittgenstein (1953) and Ryle (1949), and insists that
(contra many eliminativists) common sense psychology is not a quasi-scientific theory used to
explain or predict behavior, nor does it treat mental states like beliefs as discrete inner causes of
behavior (Bogdan, 1991; Haldane, 1988; Hannan, 1993; Wilkes, 1993). What folk psychology
actually does treat beliefs and desires as is much less clear in this tradition. One perspective

(Dennett, 1987) is that propositional attitudes are actually dispositional states that we use to adopt a
certain heuristic stance toward rational agents. According to this view, our talk about mental states
should be interpreted as talk about abstracta that, although real, are not candidates for
straightforward reduction or elimination as the result of cognitive science research. Moreover, since
beliefs and other mental states are used for so many things besides the explanation of human
behavior, it is far from clear that our explanatory theories about inner workings of the mind/brain
have much relevance for their actual status.
Defenders of eliminative materialism often point out that folk theories typically have many
functions beyond explaining and predicting, but that doesn't alter their theoretical status nor
innoculate their posits from elimination (P.M. Churchland, 1993). Moreover, while eliminativists
have typically framed the vulnerability of commonsense mental notions in terms of a false folk
psychological theory, it is important to note that, at least in principle, eliminativism does not require
such an assumption. Indeed, eliminativism only requires two basic claims: 1) that we share concepts
of mental states that include some sort of requirements that any state or structure must meet to
qualify as a mental state of that sort, and 2) the world is such that nothing comes close to meeting
those requirements. The first of these claims is not terribly controversial and while the requirements
for beliefs might come as part of an explanatory theory, they don't need to. Hence, one common
criticism of eliminativismthat our invoking of beliefs and desires is not a theoretical or quasiscientific endeavorhas very limited force. Cherubs, presumably, are not part of any sort of quasiscientific theory, yet this alone is no reason to think they might exist. Even if it should turn out that
we do not (or do not simply) posit beliefs and other propositional attitudes as part of some sort of
explanatory-predictive framework, it may still turn out that there are no such things.
The second perspective criticizing the theory-theory is based on research in contemporary cognitive
science, and stems from a different model of the nature of our explanatory and predictive practices
(Gordon, 1986, 1992; Goldman, 1992). Known as the simulation theory, this alternative model
holds that we predict and explain behavior not by using a theory, but by instead running an off-line
simulation of how we would act in a comparable situation. That is, according to this picture, we
disconnect our own decision-making sub-system and then feed it pretend beliefs and desires (and
perhaps other relevant data) that we assume the agent whose behavior we are trying to predict is
likely to possess. This allows us to generate both predictions and explanations of others by simply
employing cognitive machinery that we already possess. In effect, the simulation theory claims that
our reasoning about the minds and behavior of others is not significantly different from putting
ourselves in their shoes. Thus, no full-blown theory of the mind is ever needed. Simulations
theorists claim that, contrary to the assumptions of eliminative materialism, no theory of the mind
exists that could one day prove false.
Both sides of this debate between the theory-theory and the simulation theory have used empirical
work from developmental psychology to support their case (Stich and Nichols, 1992; Gordon,
1992). For example, theory-theorists have noted that developmental psychologists like Henry
Wellman and Alison Gopnik have used various findings to suggest that children go through phases
that are analogous to the phases one would go through when acquiring a theory (Gopnik and
Wellman, 1992). Moreover, children appear to ascribe beliefs to themselves in the same way they
ascribe beliefs to others. Theory-theorists have used considerations such as these to support their
claim that our notion of belief is employed as the posit of a folk theory rather than input to a
simulation model. At the same time, simulation theorists have employed the finding that 3-year-olds
struggle with false belief ascriptions to suggest that children are actually ascribing their own
knowledge to others, something that might be expected on the simulation account (Gordon, 1986).

4.3 Defending the Virtues of Folk Psychology


Even among theory-theorists there is considerable disagreement about the plausibility of eliminative
materialism. A third criticism of eliminative materialism is that it ignores the remarkable success of
folk psychology, success that suggests it offers a more accurate account of mental processes than

eliminativists appreciate. Apart from the strong intuitive evidence that seems to reveal beliefs and
desires, we also enjoy a great deal of success when we use common sense psychology to predict the
actions of other people. Many have noted that this high degree of success provides us with
something like an inference-to-the-best-explanation argument in favor of common sense
psychology and against eliminativism. The best explanation for the success we enjoy in explaining
and predicting human and animal behavior is that folk psychology is roughly true, and that there
really are beliefs (Kitcher, 1984; Fodor, 1987; Lahav, 1992).
A common eliminativist response to this argument is to re-emphasize a lesson from the philosophy
of science; namely, that any theoryespecially one that is as near and dear to us as folk
psychologycan often appear successful even when it completely misrepresents reality. History
demonstrates that we often discount anomalies, ignore failures as insignificant, and generally
attribute more success to a popular theory than it deserves. Like the proponents of vitalism or
phlogiston theory, we may be blind to the failings of folk psychology until an alternative account is
in hand (P. M. Churchland, 1981; P. S. Churchland, 1986).
While many defenders of folk psychology insist that folk psychology is explanatorily strong, some
defenders have gone in the opposite direction, arguing that it is committed to far less than
eliminativists have typically assumed (Horgan, 1993; Horgan and Graham, 1991; Jackson and
Pettit, 1990). According to these writers, folk psychology, while indeed a theory, is a relatively
austere (i.e., ontologically non-committal) theory, and requires very little for vindication.
Consequently, these authors conclude that when properly described, folk psychology can be seen as
compatible with a very wide range of neuroscientific or cognitive developments, making
eliminative materialism possible but unlikely.
Of course, folk theories are like any theories in that they can be partly right and partly wrong. Even
writers who are sympathetic to eliminativism, such as John Bickle and Patricia Churchland (Bickle,
1992; P.M. Churchland, 1994) point out that the history of science is filled with with cases where
the conceptual machinery of a flawed theory is neither smoothly carried over to a new theory, nor
fully eliminated. Instead, it is substantially modified and reworked, with perhaps only some of its
posits being dropped altogether. Thus, full-blown eliminative materialism and complete
reductionism are end-points on a continuum with many possibilities falling somewhere in between.
The term revisionary materialism is often invoked to denote the view that the theoretical
framework of folk psychology will only be eliminated to a degree, and that various dimensions of
our commensense conception of the mind will be at least partly vindicated.

4.4 Eliminativism Eliminated?


One final argument against eliminative materialism comes from the recent writings of a former
supporter, Stephen Stich (1991, 1996). Stich's argument is somewhat complex, but it can be
presented in outline form here. Earlier we saw that eliminative materialism is committed to the
claim that the posits of folk psychology fail to refer to anything. But as Stich points out, just what
this claim amounts to is far from clear. For example, we might think that reference failure occurs as
the result of some degree of mismatch between reality and the theory in which the posit is
embedded. But there is no clear consensus on how much of a mismatch is necessary before we can
say a given posit doesn't exist. Stich offers a variety of reasons for thinking that there are
fundamental difficulties that will plague any attempt to provide principled criteria for distinguishing
cases of reference success from cases of reference failure. Consequently, the question of whether a
theory change should be ontologically conservative or radical has no clear answer. Because
eliminative materialism rests on the assumption that folk psychology should be replaced in a way
that is ontologically radical, Stich's account pulls the rug out from under the eliminativist. Of
course, this is a problem for the folk psychology realist as well as the eliminativist, since Stich's
skeptical argument challenges our grounds for distinguishing the two.

5. Concluding Comments
Eliminative materialism entails unsettling consequences not just about our conception of the mind,
but also about the nature of morality, action, social and legal conventions, and practically every
other aspect of human activity. As Jerry Fodor puts it, if commonsense psychology were to
collapse, that would be, beyond comparison, the greatest intellectual catastrophe in the history of
our species (1987, p. xii). Thus, eliminative materialism has stimulated various projects partly
designed to vindicate ordinary mental states and establish their respectability in a sophisticated
account of the mind. For example, several projects pursued by philosophers in recent years have
attempted to provide a reductive account of the semantic content of propositional attitudes that is
entirely naturalistic (i.e., an account that only appeals to straightforward causal-physical relations
and properties). Much of the impetus for these projects stems in part from the recognition that
eliminative materialism cannot be as easily dismissed as earlier writers, like C. D. Broad, had
originally assumed.
Of course, some claim that these concerns are quite premature, given the promissory nature of
eliminative materialism. After all, a pivotal component of the eliminativist perspective is the idea
that the correct theory of the mind, once discovered by psychologists, will not reveal a system or
structure that includes anything like common-sense mental states. Thus, for eliminative materialism
to get off the ground, we need to assume that scientific psychology is going to turn out a certain
way. But why suppose that before scientific psychology gets there? What is the point of drawing
such a drastic conclusion about the nature of mentality, when a central premise needed for that
conclusion is a long ways from being known?
One response an eliminativist might offer here would be to consider the broader theoretical roles
eliminative materialism can play in our quest for a successful theory of the mind. Various writers
have stipulated necessary conditions that any theory of the mind must meet, and on some accounts
these conditions include the explication of various mental states as understood by common sense.
According to this view, if a theory doesn't include states that correspond with beliefs, or provide us
with some sort of account of the nature of consciousness, then it needn't be taken seriously as a
complete account of real mental phenomena. One virtue of eliminative materialism is that it
liberates our theorizing from this restrictive perspective. Thus, the relationship between eliminative
materialism and science may be more reciprocal than many have assumed. While it is true that
eliminative materialism depends upon the development of a radical scientific theory of the mind,
radical theorizing about the mind may itself rest upon our taking seriously the possibility that our
common sense perspective may be profoundly mistaken.

Vous aimerez peut-être aussi