WHAT DOES IT TAKE TO BE A TRUE BELIEVER?
David Henderson and Terry Horgan
Eliminative
materialism, as William Lycan (this volume) tells us,
is materialism plus the claim that no creature has ever had a belief, desire,
intention, hope, wish, or other “folk-psychological” state. Some contemporary
philosophers claim that eliminative materialism is very likely true. They
sketch certain potential scenarios, for the way theory might develop in
cognitive science and neuroscience, that they claim are fairly likely; and they
maintain that if such scenarios turned out to be the
truth about humans, then eliminative materialism would be true.
Broadly
speaking, there are two ways to reply to such arguments, for those who maintain
that eliminative materialism is false (or that the likelihood of its being true
is very low). One way is to argue that the scenarios the eliminativists
envision are themselves extremely unlikely—that we can be very confident, given
what we now know (including nontendentious scientific
knowledge), that those scenarios will not come to pass. The other is to argue
that even if they did come to pass, this would not undermine
common-sense psychology anyway. People would still have beliefs, etc.
The two
strategies are not incompatible; one could pursue them both. But the second
strategy attacks eliminativism at a more fundamental
level. And if it can be successfully carried out, then the dialectical state of
play will be strikingly secure for folk psychology. For, then it will turn out
that folk psychology simply is not hostage to the kinds of potential
empirical-theoretical developments that the eliminativists
envision. It doesn’t matter, as far as the integrity of folk psychology is
concerned, whether or not such scenarios are likely to come to pass.
Eliminativist arguments inevitably rely, often
only implicitly, on certain assumptions about what it takes for a
creature to have beliefs, desires, and other folk-psychological
states—assumptions about some alleged necessary condition(s) for being a true
believer (to adapt this colorful usage from Dennett 1987). With some such
assumption in play, the eliminativist then envisions
a scenario in which the putative necessary condition is not satisfied. Since
that scenario might very well come to pass, it is argued, eliminativism
is very likely true.
To pursue
the second strategy of reply against eliminativist
arguments is to argue that the assumptions about putative necessary conditions
for true-believerhood that are operative in eliminativist arguments are themselves very likely false.
We advocate the second strategy. Lycan, we take it,
advocates this generic strategy too, and offers an
argument of his own in an effort to implement it—his “particularly compelling” Moorean argument. Our way of implementing the strategy
differs from his in significant ways, however. In this paper we will set forth
our own anti-eliminativist argument, in a way that
highlights the differences between our approach and Lycan’s.
1. What Eliminativist
Arguments Need to Assume.
For
concreteness, let us focus on one common eliminativist
argument. (The key points we make will generalize to others.) Eliminativists often assume that a necessary condition for
being a true believer is possession of a so-called language of thought
(for short, LT)—i.e., a system of internal mental representations that (i) possess language-like syntactic structure, and (ii)
possess the propositional content of putatively attributable beliefs, desires,
and other folk-psychological states. They then argue that it is entirely
possible, or even likely, that mature cognitive science will not posit an LT.
They conclude that it is entirely possible, or even
likely, that humans are true believers.
In
what sense of ‘necessary condition’ must the eliminativist
claim that possession of an LT is a necessary condition for being a true
believer? In considering this question, two distinct notions of necessary
condition need to be distinguished. First is the idea of a conceptually
grounded necessary condition—i.e., one that, in some fairly robust sense,
is “built into” the very concept of belief, or anyway the very concept of a
true believer (i.e., a creature that has beliefs, desires, intentions, and other
folk-psychological states). Second is the idea of a de facto necessary
condition—i.e., one that, as a matter of scientific fact, is a prerequisite for
having beliefs and for being a true believer.
Given
this distinction, the question we are raising can be sharpened and reformulated
this way: Does it suffice for the eliminativist’s
purposes to claim that possession of an LT is merely a de facto necessary
condition for being a true believer, or must the eliminativist
make the stronger claim that LT possession is a conceptually grounded necessary
condition for true-believerhood?
We will
address this question by considering two different kinds of staunch advocate of
the LT hypothesis, who we will call Zenon and Jerry.[1]
Zenon, let us suppose, has the following views.
First, he holds that a conceptually grounded necessary condition for being a
true believer is susceptibility to a robust folk-psychological
interpretation—that is, a coherent overall assignment of beliefs, desires,
etc. that conforms to the generalizations of folk psychology, and that allows
for systematic folk-psychological explanations of a vast and diverse range of
the creature’s behaviors and behavioral capacities. (Robustness excludes
creatures and systems whose behavioral capacities are too narrow, parochial,
and constrained for such entities to count as true believers—e.g.,
chess-playing computers.) Prototypical true believers must actually
possess and exercise the relevant behaviors and behavioral capacities—call them
true-believer-indicating capacities (for short, TBI capacities).
(Non-prototypical true believers, such as total paralytics, must have internal
states that are sufficiently relevantly similar to prototypical ones—indeed,
states that would subserve TBI capacities were it not
for the lack of a properly functioning body.)
Second is a
conceptually grounded constraint on how beliefs are realized: they must be
realized by autonomous internal states of the creature. Thus, humanoid
robot-bodies who behave systematically just like ordinary humans, but whose
behavior is completely remote-controlled by Martians who are deliberately
causing these bodies behave in a way that is susceptible to robust
folk-psychological interpretation, would not be true believers.[2]
Third,
Zenon believes that it is a conceptually grounded sufficient
condition for being a true believer that a creature satisfies, clearly and
unequivocally, these conceptually grounded necessary conditions. Although there
can be borderline true-believers—creatures for whom there is no clear fact of
the matter whether or not they satisfy the requisite conditions, nevertheless
those creatures who clearly do satisfy these conditions thereby qualify as
full-fledged true believers.
Fourth,
he holds that there is overwhelmingly strong evidence that humans
satisfy all these conditions. Hence, he maintains, there is overwhelmingly
strong evidence that humans are genuine true believers. His reasoning here is
as follows. There is no serious doubt that humans have the relevant TBI
capacities; indeed, the behavioral capacities of normal humans are
paradigmatic. Nor is there any serious doubt that normal humans are susceptible
to robust folk-psychological interpretation (since we routinely interpret
ourselves and one another as true believers). Moreover, we certainly know
enough about the etiology of human behavior to be very confident that normal
behavior is the product of autonomous inner states--rather than the product of
remote control by Martians, or anything of the sort.[3]
So there are bound to be some states of ordinary humans that realize
beliefs, desires, intentions, etc.; the realizing states are ones temporally
coincident with the states assigned by folk-psychological interpretation.
Fifth, he maintains that FP states are realized in humans by states of the kind described in cognitive science, whose essence involves cognitive “functional architecture.” (We will call this ‘psychotechtonic realization’, borrowing from Colin McGinn (1989) the term ‘psychotechtonic’ for states and processes described at the level of cognitive “engineering.”) The states that psychotechtonically realize beliefs, etc. are themselves realized physically in the brain.
Sixth,
he holds that there is very good evidence, including evidence from cognitive
science, that the psychotechtonic realizing-states in
humans involve language-like mental representations; they are states in an LT
cognitive architecture. Indeed, seventh, he holds that as a matter of
scientific fact, the only way that beliefs, etc. could be psychotechtonically realized, at least in creatures whose
constitution is physical and whose range of entertainable
beliefs and associated TBI capacities is as vast as that of humans, is by
states in an LT cognitive architecture. This seventh claim is the contention
that having an LT is a de facto necessary condition for being a true believer.[4]
For Zenon, the LT hypothesis is no part of the concept
of belief, or the concept of a true believer. Although he is staunch and
confident in his advocacy of the LT hypothesis, this is not because he thinks
that the hypothesis is somehow built into the notion of belief itself; on the
contrary, he would deny this. Rather, he holds that as a matter of scientific
fact, the only kind of cognitive architecture that can subserve
TBI capacities—at any rate, TBI capacities as extensive and subtle as those of
humans—is an LT cognitive architecture. For him, then, possession of an LT is
not a conceptually grounded necessary condition for true-believerhood—even
though he adamantly argues that it is a de facto necessary condition.
Let us
suppose that Jerry, in contrast to Zenon, holds that
possession of an LT is indeed a conceptually grounded necessary condition for
true-believerhood. I.e., Jerry maintains that the LT
requirement is built right into the concept of a belief (and into
folk-psychological concepts generally). Jerry also agrees with Zenon in these claims: (i) that
mature cognitive science will certainly posit an LT cognitive architecture;
(ii) that beliefs, etc., are psychotechtonically
realized in humans by states in this LT architecture; and (iii) that as a
matter of scientific fact, no non-LT cognitive architecture could subserve TBI capacities.
Now, both Zenon and Jerry would vigorously take issue with the eliminativist’s premise that it is entirely possible, or
even likely, that mature cognitive science will not posit an LT. In doing so,
they would be pursuing the first kind of anti-eliminativist
strategy we mentioned at the outset—viz., arguing that the eliminativist’s
favored future-science scenario is simply not going to be a part mature
science, and has no serious likelihood of being a part of it.
With this fact duly noted,
however, it is crucial to appreciate that Zenon and
Jerry would part company with respect to the following subjunctive claim:
(S) If we were to obtain good
evidence that humans do not possess an LT, we would then have good evidence
that humans are not true believers.
For Zenon,
this subjunctive conditional is just false. Even in the envisioned hypothetical
scenario, he says, there would still be enormously good evidence that all
relevant conceptually grounded necessary conditions for true-believerhood are satisfied; hence, there would still be
overwhelmingly strong evidence that humans are true believers. Thus, the epistemically appropriate conclusion to draw would be, not
that humans fail to be true believers, but rather that beliefs are psychotechtonically realized in a way that does not involve
language-like representations. The evidence against an LT would point, not
to eliminative materialism, but rather to the falsity of Zenon’s
erstwhile beliefs (i) that possessing an LT cognitive
architecture is a de facto necessary condition for being a true believer, and
(ii) that humans possess an LT cognitive architecture.
For
Jerry, however, having an LT is a conceptually grounded necessary condition for
being a true believer. By virtue of the very concept of a belief, he maintains,
beliefs can only be psychotechtonically realized by
states in an LT cognitive architecture. Thus, a putative psychotechtonic
realization relation that satisfied all other conceptually grounded
constraints, but did not involve an LT, would not be a genuine realization
relation for beliefs and other FP states; it would violate a conceptually
grounded necessary condition for true-believerhood.
So for him, the subjunctive claim (S) is true.
This
parable about Zenon and Jerry has two important
morals, concerning what the eliminativist needs to
assume about the relation of the LT hypothesis to folk psychology. First, it is
not enough to claim that possessing an LT is a de facto necessary condition for
being a true believer, and that we have good evidence for this. For, some who
accept these claims could also maintain (with Zenon)
that our evidence for them would simply dissolve in the hypothetical scenario
where we obtain evidence that humans lack a language of thought; in that
scenario, they contend, we would have evidence that beliefs are psychotechtonically realized some other way in
humans. So the eliminativist must claim, in
opposition to this, that even if we were to obtain
good evidence that humans do not possess an LT, we would still have good
evidence that possession of an LT is a prerequisite for being a true believer.
The
second moral (and our main point in the present section) is this: the eliminativist must claim, with Jerry,
that having an LT is a conceptually grounded necessary condition
for being a true believer. For, in order for the hypothetical senario to be one in which the evidence points toward eliminativism, rather than pointing (as Zenon
maintains) toward the conclusion that beliefs are psychotectonically
realized by non-LT states, there would have to be some reason why the
non-LT states, despite satisfying all the other conceptually grounded
constraints, would not qualify as legitimate psychotechtonic
realizers of beliefs. But, since non-LT states would
be otherwise eligible as belief-realizers (apart from
the fact that they do not involve language-like representations), evidently the
only reason why they could nonetheless fail to be genuine belief-realizers is that the requirement of LT-realization is
built into the very concept of a belief.
Having
thus clarified the assumption needed by the eliminativist
argument, we should add some observations about what the argument is not
necessarily presupposing. First, the eliminativist
need not assume that there is a tenable distinction between analytic and
non-analytic truths. Many in contemporary philosophy think not. Lycan is among them, and we take it that various eliminativists are among them too. As far as we can tell,
this is a matter about which the eliminativist can
remain neutral. Nevertheless, there are some things one might say about belief
that would be just plain wrong in virtue of how our concepts work—that
would constitute changing the subject, rather than changing what one claims or
believes about the original subject. Conditions of true-believerhood
whose denial would constitute changing the subject are conceptually grounded
necessary conditions. Presumably there can be such conditions, whether or not
the analytic/synthetic distinction is viable.
Second,
the eliminativist need not assume that the claim that
being a true believer requires LT possession is a purely conceptually
grounded truth, untinged by any empirical claims.
Certain conceptually grounded truths might be conditional in form, with
a conditional “slot” filled by some empirical claim that is not itself
conceptually grounded. Here is a plausible example:
Given that
the stuff we call water on Earth is composed of H20 molecules, superficially
water-like stuff can be real water only if it too is composed of H20 molecules.
Likewise, the eliminativist
could perhaps rely upon a putatively conceptually grounded truth of following
form:
Given that…, a creature can
have beliefs only if it has an LT.
As long as ‘…’ were filled in by
some relatively nontendentious empirical claim, a
conceptually grounded conditional statement of this form could still serve the eliminativist’s dialectical purposes. Of course, this
empirical claim, whatever it is, needs to be one whose epistemic warrant is not
only very high, but would remain very high even if we were to obtain
good evidence that humans do not possess an LT. Here is an example of a
conditional statement, containing such an empirical constituent, that an eliminativist might rely upon as a putative conceptually
grounded necessary truth:
Given that beliefs are
realized by physico-chemical processes, a creature
can have beliefs only if it has an LT.
Finally,
it need not be assumed that the relevant conceptually grounded truths are
knowable a priori, where we understand ‘a priori’ in the standard way, as a
non-empirical way of knowing. Epistemic status is a different matter from
semantic/modal status. This leads to our next section.
2. Philosophy and
Ideological Inquiry.
We
will use the term ‘ideology’ for inquiry into the workings of human concepts,
and into the semantics of the terms that express these concepts, and also for
that facts that such inquiry seeks to discover (cf. Horgan 1993, Graham and
Horgan 1994, Henderson and Horgan 2000). Ideology as an area of inquiry is a
broadly empirical, interdisciplinary, enterprise encompassing such
fields as psychology, linguistics, social anthropology, and philosophy. In our
view, even the philosophical dimension of ideology is empirical, rather than
being a priori in the traditional sense.[5]
The data that philosophers employ is relatively close at hand—data that
includes, but need not be limited to, one’s own conceptual/linguistic
intuitions about how to describe various concrete scenarios, actual or
hypothetical. Philosophical thought experiments really are experiments:
they generate empirical data, in the form of such intuitions. Such data
constitutes powerful, albeit defeasible, empirical
evidence vis-à-vis questions of ideology, because ceteris paribus, the relevant
intuitive judgments are likely to reflect the workings of one’s own conceptual
and semantic competence.
In order to appreciate that the kind
of ideological inquiry typically pursued in philosophy is indeed a broadly
empirical enterprise (even though the data employed is largely available from
the armchair), it is illuminating to compare such inquiry to the methodology
typically employed by linguists in constructing and evaluating theories of
natural-language syntax. The empirical data for syntactic theory includes
certain judgments and judgment dispositions of competent language users--in
particular, judgments and dispositions concerning the grammaticality or
ungrammaticality of various sentence-like strings, and concerning grammatical
ambiguity or nonambiguity of various sentences. Such
judgments are relevant simultaneously to psychological theories of human
language processing, and also to linguistic theories about the syntax of
language itself. Native speakers, after all, can be expected to have judgment
dispositions about these matters that reflect a solid mastery of their own
language (or their own regional dialect, at any rate). So, when native speakers
are intersubjectively consistent and also uniformly
confident about such syntactic judgments, then normally the best psychological
explanation will be that these judgments reflect the natives' syntactic
competence, their mastery of the syntactic norms or syntactic structures
underlying their language. And this psychological hypothesis, in turn, has a
direct implication for linguistic theory--viz., that under an adequate theory
of syntax for the natives' language (or dialect), those syntactic judgments
will turn out correct.
Similar observations
hold with respect to hypotheses or theories concerning ideology. Certain robust
patterns of judgment among competent users of concepts and language will be
plausibly explained as manifesting the users' conceptual/semantic competence.
Here too, as with grammaticality judgments, much of the relevant data is close
at hand, some of it in the form our own introspectively accessible linguistic
intuitions about how to describe various actual and envisioned scenarios. Since
the evidence this data provides is empirical, it is of course defeasible. Nevertheless, it can be very strong, comparable
in epistemic weight to the empirical evidence that syntactic judgments provide
for theories of syntax.
Consider,
for instance, the above-mentioned thesis
about the ideology of our concept water:
Given that
the stuff we call water on Earth is composed of H20 molecules, superficially
water-like stuff can be real water only if it too is composed of H20 molecules.
Hilary Putnam (1975) convinced virtually the entire philosophical
community of this thesis, by asking us to consult our intuitions about how to
describe his Twin Earth scenario. (The influential arguments
of Saul Kripke (1972) for related ideological theses
work similarly.) We were right to be convinced, because the deliverances
of our descriptive intuitions very likely reflect the proper workings of our
own conceptual/semantic competence with the notion of water, and hence are very
likely correct.
Philosophers
have not often been explicit about the nature of ideological investigation in
philosophy. And when they have been explicit, typically they have conceived it
as a non-empirical enterprise, pursued by employing reason alone. But such
inquiry is more credibly construed, especially within the framework of a
broadly naturalistic approach to human cognition and to epistemology, as
broadly empirical—as argued at greater length elsewhere (Graham and Horgan
1991, 1994, Horgan 1993, Henderson and Horgan 2000).
3. The Ideology of Folk Psychology: Opulent or Austere?
Recall that Jerry and the eliminativist hold that possession of a language of thought is a conceptually grounded necessary condition for being a true believer. Zenon, on the other hand claims that the conceptually grounded prerequisites for being a true believer are much more modest: the putative requirement of an LT cognitive architecture is not built into folk-psychological concepts; nor are any other putative requirements that could turn out, under any remotely likely scenario for the future development of cognitive science or neuroscience, not to be satisfied by humans. Zenon’s conception of the ideology of folk psychology we will call austere; competing conceptions that treat LT possession, and/or various other scientifically tendentious features, as conceptually grounded prerequisites for true-believerhood, we will call opulent (cf. Graham and Horgan 1991, 1994, Horgan 1993).
It is an empirical question whether the ideology of FP is opulent or austere. Very well, then; here are some broadly empirical arguments in defense of the austerity hypothesis, over against the opulence hypothesis. All involve data so close at hand we can obtain it from our armchairs (die vom Armchair aus zuhandenen Daten). This fact does not, however, prevent them from being empirical and hence epistemically defeasible. (The arguments, as here formulated, will be directed explicitly against the claim that possessing an LT is a conceptually grounded necessary condition for being a true believer; but they are really broader in scope, and apply mutatis mutandis against other putative conceptually grounded prerequisites for being a true believer sometimes embraced by eliminativists.)[6]
First is the argument from recalcitrant
intuitions. When we envision a scenario in which the LT hypothesis turns
out to be false, and then ask ourselves whether it seems intuitively
appropriate to describe this scenario by saying "Humans turn out not to
have beliefs and desires," the answer is negative. On the contrary, it
seems natural to say things like the following, about this hypothetical
situation: "Humans will have acquired grounds to believe that the
LT hypothesis is false." If the LT hypothesis is really built into the
ideology of FP, however, then this fact ought to reveal itself in our own
descriptive intuitions about the envisioned scenario: describing the people in
this scenario as having beliefs and desires should seem semantically mistaken
to us, just as it seems semantically mistaken to describe the stuff in the
oceans and lakes of Putnam's Twin Earth as water. But it doesn't. So,
since these judgments very probably emanate from our conceptual/semantic
competence with respect to the concepts and terms of FP, it is very likely that
our descriptive intuitions about the envisioned scenarios are correct--and
hence that the falsity of the LT hypothesis would not falsify FP at all. I.e.,
it is very likely that ideology of FP is austere vis-a-vis the LT hypothesis,
not opulent.
Second is the argument
from ideological conservatism. Notions like action, assertion, having
reasons, and epistemic warrant all are folk psychological: they presuppose that
humans are true believers.[7] These
notions play certain essential roles in human life that would surely persist
even if we discovered that the LT hypothesis is false; hence, if FP were
ideologically committed to an LT, this commitment would go directly contrary to
certain central purposes for which FP concepts and terms are employed. But since
human concepts and terms evolve in a broadly pragmatic way, in general they are
not likely to have conceptually grounded satisfaction conditions that are more
severe or restrictive than is required by the purposes they serve. So FP is not
likely to exhibit any such gratuitous, counter-pragmatic, features. Hence the
ideology of FP is very likely austere vis-a-vis the LT hypothesis, rather than
opulent.
Third is the argument
from conceivability. Although we humans can readily conceive discovering
that the LT hypothesis is false, we cannot even conceive of ourselves then
dropping folk-psychological notions like action, assertion, and epistemic
warrant, and thus we cannot conceive of ceasing to regard ourselves and one
another as true believers. For, to drop these notions on these grounds--or even
to try dropping them--would be actions, performed for a reason;
and notions like action, and having a reason, are themselves thoroughly
folk-psychological. This conceivability mismatch between dropping the LT hypothesis
on the one hand, and dropping FP on the other, is naturally accommodated under
the ideological hypothesis that FP is conceptually austere: for, in that case
the envisioned scenario is one in which the FP concepts still would
apply to humans. But if FP is ideologically committed to an LT cognitive
architecture, however, then no such accommodation is possible; instead, it
remains a puzzle why we should find ourselves unable to conceive dropping
concepts whose putative ideological commitments we can fairly easily conceive
ourselves discovering to be false. So the conceivability mismatch provides
evidence for the austere conception of FP over against the opulent conception
vis-a-vis the LT hypothesis, since the former accommodates the mismatch whereas
the latter renders it puzzling.[8]
We again emphasize that the arguments just set forth are
empirical arguments, and hence defeasible: in each
case, the claim is that the austere conception of FP's ideology accords better
with the adduced empirical data than does the opulent conception. Here, as with
abductive empirical reasoning in general, the fact
that several distinct forms of evidence converge on the same conclusion means
that their net epistemic import is even greater than the "sum" of
their respective individual epistemic "weights." In our view, the
empirical case for the austere conception, over against an opulent conception
that treats the FP hypothesis or scientifically tendentious hypotheses as
conceptually grounded prerequisites for true-believerhood,
is overwhelmingly strong.
As remarked above, these arguments against the specific claim that LT possession is a conceptually grounded necessary condition for true-believerhood can be readily adapted against the various other putative conceptually grounded necessary conditions that sometimes figure in eliminativist arguments. The arguments thus provide strong empirical support for the claim that folk psychology is ideologically austere tout court (and not merely austere vis-a-vis the LT hypothesis). The upshot is that it is highly unlikely that science can yield eliminativist results. For, it is highly unlikely that any conceptually grounded necessary conditions for true-believerhood would fail to be to be satisfied in any remotely plausible scenario for how future science would go.
This
leaves it open whether or not various features that eliminativists
treat as conceptually grounded necessary conditions for true-believerhood are de facto necessary conditions for true-believerhood. Perhaps some are. But debates about such
matters just do not threaten the integrity of folk psychology. The right way to
look at it is this: No scenarios that are even remotely likely to be true of
humans are ones under in which any conceptually grounded necessary
conditions for true-believerhood would fail to be
met. So at most, what could reasonably be concluded, if we were to acquire good
evidence that some such scenario were true for humans, would be that certain
thought-to-be de facto necessary conditions for true-believerhood
are not really de facto necessary conditions after all.
4. Comments on Lycan.
As
we understand Lycan’s anti-eliminativist
argument, in effect he too construes eliminativists
as needing, and implicitly relying upon, putative conceptually grounded
necessary truths. He describes his own Moorean
argument as at root a comparison of plausibilities:
the plausibility of common-sense mental ascriptions (e.g., that Granny wants a
beer and believes there is one under the sofa), vs. the plausibility of such eliminativist assumptions (e.g., that possession of a
language of thought is a conceptually grounded necessary condition for being a
true believer). In effect, the Moorean argument
relies upon the contention that claims of the former kind will always seem
intuitively more plausible than claims of the latter kind.
We
are inclined to agree with this contention, but its dialectical force against eliminativism certainly can be questioned. Why, one might
well wonder, should comparative intuitive plausibility have epistemic trumping
power, in the context of debates about eliminativism?
After all, comparative intuitive plausibility can be a pretty shaky reed epistemically, especially when what’s under comparison is a
specific common-sense claim on the one hand vs. a general theoretical
hypothesis on the other. It is not intuitively plausible that these hands I am
looking at are mostly empty space, or that simultaneity is relative to a
reference frame, but so much the worse for intuitive plausibility in these
cases! When common-sense claims are being compared with theoretical hypotheses
that call those very claims into question, defending the former by direct
appeal to comparative intuitive plausibility excessively privileges common
sense. Higher intuitive plausibility does not have automatic epistemic trumping
power.
We
take it that Lycan would agree. “Common-sense
beliefs,” he says, “can be corrected, even trashed entirely, by careful
empirical investigation and scientific theorizing” (p. 8). But for him,
such epistemic trumping power over common sense can only accrue to empirical
scientific hypotheses, ones that are well supported by empirical evidence. They
cannot accrue to the “purely philosophical” premises that he thinks eliminativist arguments invariably rely upon. He says:
Philosophers...are
not explorers or scientists.... Common sense must yield to evidence...but it
need not yield to bare metaphysical pronouncement.... 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 “intuitions” can only
throw spitballs. (p. 9)
Thus, Lycan’s
Moorean argument turns out to depend crucially upon
the supposition that the eliminativists need premises
that are non-empirical. These are the kind of premises that he claims
can never epistemically trump common-sense beliefs.
One
is tempted to become embroiled in arguments concerning whether any kinds of allegedly
non-empirical claim can trump common-sense beliefs (rather than being “bare
metaphysical pronouncement”)—and if so, which kinds can possess this privileged
epistemic status, and why.[9]
But we can set these issues aside here. From our own point of view, Lycan’s argument is flawed at a more fundamental level:
viz., it falsely assumes that eliminativist arguments
employ non-empirical premises. We claim, on the contrary, that although these
arguments do need premises about putative conceptually grounded necessary
conditions for true-believerhood, such premises are empirical
ideological hypotheses. In principle, these hypotheses could receive enough
evidential support to become very well warranted—enough so to epistemically trump common-sense claims, like the claim
that Granny wants a beer and believes there is one under the sofa.[10]
Lycan’s Moorean argument
therefore commits a straw man fallacy, since it rests crucially upon a misconstrual of the eliminativist
reasoning toward which it is directed.
Our own anti-eliminativist approach recognizes the eliminativist assumptions for what they are—viz., broadly empirical contentions about the ideology of folk-psychological concepts—and then gives broadly empirical arguments against these assumptions. Certain intuitive, common-sense, judgments enter here too: for instance, judgments about what would be the right way to characterize various actual or hypothetical scenarios. But now such common-sense judgments figure not as direct and automatic trumpers over putatively non-empirical philosophical claims, but in a different way, viz., as data for ideological theorizing—data that defeasibly may be presumed to emanate from our conceptual/semantic competence, and thus defeasibly may be presumed to reflect the nature of the folk-psychological concepts themselves.
Lycan’s Moorean argument, despite its flaws, can be seen as gesturing toward the kind of reasoning we have offered here: a broadly empirical, albeit still philosophical, argument for the ideological austerity of folk psychology. We and Lycan share the goal of effectively implementing the second of the two anti-eliminativist strategies we described at the outset. Our argument does so successfully, we submit, whereas his does not.
References
Baker, L.
R., 1987. Saving Belief: A Critique of Physicalism.
Brand, M.,
1984. Intending and Acting: Toward a Naturalized Action Theory.
Davidson,
D., 1963. “Actions, Reasons, and Causes,” Journal of Philosophy 60,
685-700.
Dennett,
D., 1987. “True Believers: the Intentional Strategy and Why it Works,” in his The
Intentional Stance.
Fodor, J.,
1980. “Propositional Attitudes,” in his Representations.
Goldman,
A., 1970. A Theory of Human Action.
Graham, G. and Horgan, T., 1988. “How to be
Realistic About Folk Psychology,” Philosophical
Psychology 1, 69-81.
Graham, G. and Horgan, T., 1991. “In Defense of Southern Fundamentalism,” Philosophical Studies
62, 107-34.
Graham, G. and Horgan, T., 1994. “Southern Fundamentalism and the End of Philosophy,” Philosophical
Issues 5, 219-47.
Henderson, D. and Horgan, T., 2000. “What Is A
Priori and What Is It Good For?” Southern Journal of
Philosophy 38, Spindel Conference Supplement.
Horgan, T., 1993. “The Austere Ideology of Folk Psychology,” Mind and Language
8, 282-97.
Horgan, T., and Tienson, J., 1996. Connectionism
and the Philosophy of Psychology.
Kripke, S., 1972. “Naming
and Necessity,” in G. Harman and D. Davidson (eds.), The
Semantics of Natural Language.
Lycan, W., this
volume. “A Particularly Compelling Refutation of
Eliminative Materialism.”
McGinn, C., 1989. Mental Content.
Putnam, H.,
1975. “The Meaning of ‘Meaning’,” in K. Gunderson (ed.,), Language,
Mind, and Knowledge.
Pylyshyn, Z., 1984. Computation and Cognition: Toward a Foundation for Cognitive Science. Cambridge MA: Bradford Books/MIT Press.
[1] Jerry Fodor
and Zenon Pylyshyn are
perhaps the most prominent philosophical defenders of the LT hypothesis as one
that cognitive science needs. It also figures centrally in their account of
belief and other folk-psychological states. But we do not mean to suggest that
the actual Zenon Pylyshyn
is the first kind of LT-fan we will describe, or that Jerry Fodor is the second
kind. As far as we can tell, it is not clear from their writings how to
classify either of them on this matter (although the discussion of common sense
psychology in Pylyshyn (1984) perhaps suggests a
tendency toward the first kind of position concerning LT, whereas the treatment
of propositional attitudes in Fodor (1980) and in other of Fodor’s writings
perhaps suggests a tendency toward the second). We borrow their names mainly
just for vividness.
[2] A possible third condition which Zenon might perhaps embrace as another conceptually grounded prerequisite for being a true believer is that the creature have a certain phenomenology, a certain “what it’s like” of belief—for instance, a phenomenology that distinguishes believing “that’s a rabbit” from believing “that’s a collection of undetached rabbit parts.”
[3] Conceptually grounded phenomenological aspects of true belief, if any, are accessible through introspection, and so there is no serious doubt that any such aspects are also instantiated in humans.
[4] An important mode of reasoning, to support this kind of claim, is what Horgan and Tienson (1996 chapter 5) call “inference to the only available explanation.” They themselves employ it in defense of an LT cognitive architecture. The basic idea is this: the range of potential beliefs, etc. that humans can possess is so vast that the only way to implement such states, in a physical system with resources on the scale of a human brain, is via a system of representations with language-like syntactic compositionality.
[5] A non-traditional, partially empirical, notion of the a priori is described in Henderson and Horgan (2000), where it is argued that the philosophical dimension of ideology typically is a priori in this non-traditional sense (but not in the traditional sense).
[6] See Graham and Horgan (1991) and Horgan (1993), where these arguments are given in more general form, and also are applied explicitly against several other putative conceptually grounded prerequisites for true-believerhood sometimes invoked by eliminativists. The next several paragraphs are slight modifications of the formulations in Horgan (1993).
[7]. Cf. Baker (1987), Graham and
Horgan (1988, 1991). It is widely accepted in philosophy of mind that
the concept of an action is the concept of an item of behavior that is caused,
in a certain characteristic way, by FP states like belief, desire, and
intention; cf. Davidson (1963), Goldman (1970), Brand (1984). Eliminativists, in our experience, tend to grant that
action is a folk-psychological notion.
[8]. It should be stressed that this argument is
empirical, not transcendental. Maybe non-FP-tinged successors of FP concepts
could be devised, even though we presently have virtually no idea what such
replacement concepts would be like. But even those philosophers
who, like us, do not buy transcendental arguments, should acknowledge that
conceivability considerations can constitute important empirical data
about matters of ideology.
[9] Such a challenge to Lycan might include the idea that non-empirical knowledge of conceptually grounded necessary truths is provided by a specific kind of intuitive apprehension that is a byproduct of conceptual competence. It is ironic that Lycan so quickly dismisses intuition as a potential vehicle of a priori knowledge, given the central role of intuitive comparative-plausibility judgments in his own Moorean argument.
[10] This is not to say, of course, that eliminativists have actually provided such evidence. On the contrary, often they do not even articulate their needed ideological premises explicitly, let alone argue for them.