2023年6月24日发(作者：)

Mind AssociationIf I Am a Brain in a Vat, Then I Am Not a Brain in a VatAuthor(s): Anthony BruecknerSource:

Mind,

New Series, Vol. 101, No. 401 (Jan., 1992), pp. 123-128Published by: Oxford University Press on behalf of the Mind AssociationStable URL: /stable/2254123Accessed: 12/09/2009 09:05Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at/page/info/about/policies/. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial contact the publisher regarding any further use of this work. Publisher contact information may be obtained at/action/showPublisher?publisherCode= copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with thescholarly community to preserve their work and the materials they rely upon, and to build a common research platform thatpromotes the discovery and use of these resources. For more information about JSTOR, please contact support@ University Press and Mind Association are collaborating with JSTOR to digitize, preserve and extendaccess to ://CUSSIONS

If

l Am a Brain in a

Vat,

Then

l Am Not a Brain in

a Vat

ANTHONY

BRUECKNER

argu-

of Hilary Putnam's

Massimo Dell'Utri

(1990) provides

a reconstruction

in a vat is

ment (1981, chapter

1) to show that

the hypothesis

that

we are

brains

Dell'Utri offers us is,

on the face

I will explain why the argument

self-refuting.

Then

I will provide

a way out of

the difficulty.

of it, quite

problematic.

premisses":

The argument has

two "background

<

does not have

any necessary

connection

with

A mental

wrong.

are

in a word,

of reference,

theories

< Magical

it stands

what

(Dell'Utri 1990,

p. 83)

in causal chains connecting

our

In order

to refer we

have to partake

are

intentions

that

our referential

provided

words with their

reference,

itself is knowable.

(Dell'Utri 1990,

p.

and

the reference

socially shared

84)

to have

which

causes

the brains

to a computer

connected

in a vat (BIV) are

Brains

from those

had

sense-impressions-"images"--qualitatively

indistinguishable

using the

above

According to Dell'Utri,

Putnam,

by normal

embodied brains.

premisses

shows that,

though

the BIV's causal

connections

could go beyond

the

only causal

connections

which

their

by the computer,

images produced

to those

very

are

the ones restricted

count as causal chains

of reference

in such a way that the

words in the

images [caused

by the computer],

sentence "We are

BIV" refer to objects in

the image. (Dell'Utri

1990,

pp. 86-7)

of the BIV's terms

is the backbone

the references

The foregoing claim

about

let us

note

that argument,

Before examining

argument.

of Dell'Utri's Putnamian

of a BIV's token

of

interpretation

on one reasonable

does hold that

Putnam

that

it refers to trees

in the image, i.e,

to the sense-impressions

as of

trees

"tree",

But he also holds that the token

could be

reasonably

caused by

the computer.

to those electrical impulses

which stimulate

the brain,

interpreted as

referring

the token could also be

reason-

as of trees;

causing

it to have sense-impressions

features

says Putnam,

as referring

to the computer

program

ably interpreted,

electrical impulses (Putnam

1981,

causally responsible

for the aforementioned

theory

of reference

is in fact

different

use of the

causal,

pp.14-5).

So Putnam's

it is

rea-

to Putnam,

according

of Putnam:

from

Dell'Utri's in his reconstruction

from

the images

to objects

distinct

refer

the BIV's terms

to suppose

that

sonable

Mind,

Vol. 101 . 401. January

1992

?

Oxford University

Press 1992 124 Anthony Brueckner

produced by the computer

(not trees, of course;

instead, objects

playing a causal

role analogous to that

played by trees in the production of

unenvatted

tokens of

"tree"). Even so, Putnam

agrees that it is reasonable to interpret the BIV's tokens

of "tree", "brain in a vat",

etc. as referring to objects

in the image.

So let us ignore

Dell'Utri's departure from

Putnam and

consider his Putnam-inspired argument:

(1) We are

BIV

[by hypothesis]

(2) If we are BIV, then we are

not BIV in the image

[explanation of hypothesis]

(3) We are

not BIV in the image

[(1), (2)]

(4) If we are BIV, then

every word

we utter

refers to objects

in

the image

[(1), (2)]

(5) Every word we utter refers

to objects

in the image

[(1), (4)]

A is true

(6) Every uttered sentence

iff A is true in the image

[(5)]

(7) The sentence

"We are

BIV"

is

iff it is true in the image

true

[(instantiation of (6)]

(8) The sentence "We are

BIV"

in the image iff we

is true

are

BIV in the

image

[thesis of

equivalence]

(9) The sentence

"We

are

BIV"

iff we are

BIV in

is true

the image

[(7), (8)]

(10) The sentence

"We are

BIV"

is false

[(3), (9)]

(1

1) The sentence

"We are BIV"

is false iff the sentence

"We are

not BIV" is true

[definition of falsity]

(12) The sentence

"We are not BIV"

is true

[(10), (11)]

(13) The sentence

"We are not BIV"

iff we are

not BIV is true

[thesis of equivalence]

(14) We are

not BIV

[(12), (13)]

A BIV's terms refer to

objects

in the

Let us work through this argument.

his sentence tokens do not have the standard, Tarskian truth

image and

therefore

conditions one

might

have expected.

For

example,

since a BIV's tokens of "tree"

refer to trees in the

image instead

of trees, it is not the case that: his tokens of

"Trees are green" are true

iff trees are

green. Instead,

his tokens of the

sentence

are true iff trees are

green

in the

image. The same holds for a BIV's tokens of

"We are BIV": these are true

iff he and

his cohorts are BIV in the

image.

Since

them as

instead

they are

not BIV in the

image (their

sense-impressions

represent

it follows that a BIV's tokens of "We are BIV"

are false and

that

his

embodied), Then I am not a Brain in a Vat 125

If I am a Brain in a Vat,

This reasoning is employed in Dell'Utri's

tokens of "We are not BIV" are true.

we are

(12). On the initial assumption (1) that

argument as it proceeds up through

BIV, Dell'Utri argues, our tokens of "We

are not BIV" are true, for the reasons

just outlined.

The truth are green" can be cor-

conditions

of a non-BIV's tokens of "Trees

rectly stated in my language (the language which "we", as Dell'Utri puts it,

speak) by using the device of disquotation. For

example, a non-BIV's tokens of

S as

green. In general, tokens of a sentence

"Trees are green" are true iff trees

are

rel-

L will have disquotational truth conditions

uttered in a given object language

L' iff there is a true sentence of L' which consists of S

ative to a metalanguage

"is

followed by an L'-translation of the phrase

surrounded by quotation marks,

in L iff', followed by S itself. When the tokens

of a sentence as uttered

in a

true

truth conditions relative to my lan-

given object language have disquotational

guage,

I will abbreviate this

by saying that

the tokens have disquotational truth

conditions, tout court, leaving implicit

the relativization to my own language as

metalanguage. Thus, a non-BIV's tokens of "Trees

are green"

have disquota-

of "We BIV" had

tional truth conditions. Now if a BIV's tokens are disquotational

ones specified in (9), then

truth conditions rather than the non-disquotational

these tokens would be true, not false, since a BIV and

his cohorts

are, after

all,

BIV. Thus, a BIV's tokens of "We

are not BIV" would be false, not true,

if his

sentence tokens had So it is crucial for Dell'Utri's

disquotational truth conditions.

argument up through (12) that a BIV's sentence

tokens have non-disquotational

Otherwise he cannot maintain that on the assumption (1) that we

truth conditions.

are

BIV,

our tokens of "We are

not BIV"

are true.

the contradic-

argument by deriving

Dell'Utri seeks to complete his reductio

for

tory of the initial assumption (1) that

we are

BIV.

Here is where the problem

the argument arises. In order to derive (14), the contradictory of (1), Dell'Utri

to this thesis, which

of equivalence"-(13)-to

(12). According

applies

a "thesis

are

not BIV" are true

iff we

Dell'Utri does not discuss at all, our tokens of "We

are not BIV. That is, according to the thesis, our tokens of "We are not

BIV"

initial assumption

conditions. But on the argument's

have disquotational truth

(1) that we are BIV (and given Dell'Utri's

first and second "background pre-

on reference),

our sentence

misses" concerning the causal-theoretic constraints

conditions as

specified

in

(6).

In

tokens in general have non-disquotational truth

truth

particular, then,

our tokens of "We

are not BIV" have

non-disquotational

conditions. On the argument's initial assumption

(1)

that we are

BIV, then,

our

iff we are not BIV in the

image. Thus,

given

are not BIV" are

true

tokens of "We

fol-

what

actually

in

(12) that

our tokens

of "We

are

not BIV" are

true,

the result

lows from the argument's

various

assumptions

is that we are not BIV in the

we

fails to contradict the initial

assumption (1)

that

image. But this consequence

As Dell'Utri notes

in

(2),

if we are

BIV,

then we are not BIV in the

areAMIV.

that

the

assumption

fails. Dell'Utri has not shown that

image. Thus,

the reductio

we are

BIV is self-refuting. 126 Anthony Brueckner

There is clearly something

odd going on in Dell'Utri's argument. Given the

initial assumption (1) that

we are BIV, it seems that Dell'Utri cannot consistently

hold, as he does, that (i) our tokens of "We

are BIV" have non-disquotational

truth conditions, while (ii) our tokens of "We

are not BIV" have disquotational

truth

conditions. Dell'Utri offers no explanation

whatsoever for this apparent

equivocation.

Here is a different

way of reconstructing Putnam, in which this problem of

is avoided:

equivocation

(A) If we are BIV, then

our

tokens of "We are BIV" are true

iff we are

BIV in the image

[(causal-theoretic semantics]

(B) If we are BIV, then we are

not BIV in the image.

(C) If we are BIV, then

our tokens

of "We are

BIV" are

false [(A),(B)]

(D) If we are not BIV, then

our

tokens of "We are

BIV"

are true

iff we are BIV

(E) If we are

not BIV, then our tok-

ens of "We are BIV" are

false

[(D)]

we are BIVor

(F) Either

we are

not BIV

tokens of "We are BIV"

(G) Our

are false

[(C),(E),(F)]

tokens of "We are not

(H) Our

BIV"

are

true

[(G)]

On this reconstruction, we derive (H) without

making any claims about

which

truth conditions our sentence tokens have. On either

possibility, our tokens of

A

problem

"We are

not BIV"

are true. with this argument, though, its con-

is that

in character. To

refute

it seems that

clusion is metalinguistic the skeptic, we need

to be able to claim the non-metalinguistic knowledge that

we are not BIV.1

to complete the argument by adding a disquotation

There is a great temptation

such as

principle

(*) In

general, our sentence tokens have disquotational

truth

conditions;

iff we are

not BIV.

thus,

our tokens

of "We

are

not BIV" are true

(*)

would allow us to move from (H) to

(**) We are

not BIV.

This expanded of which sorts

of truth

argument is not equivocal

on the question

But the addi-

conditions our sentence tokens

have,

unlike Dell'Utri's

argument.

if we are in fact

BIV,

then

for a different reason:

tion of (*) seems problematic

1

For a discussion of this reconstruction of Putnam,

see Brueckner (1986). Then I am not a Brain in a Vat 127

If I am a Brain in a Vat,

conditions even though things

our sentence tokens have non-disquotational truth

and uttering sentence

seem exactly as they would seem if we were unenvatted

tokens with disquotational truth conditions.

So what is it, the skeptic would ask,

tokens have disquotational (unenvatted)

that enables us to know that our sentence

conditions they possess

truth conditions rather than the hon-disquotational truth

How can we claim to know (*) if we do not

on condition that we are envatted?

know whether we are envatted?2

to the expanded argument is to insist

One way of responding to this objection

on the triviality of the claim that my sentence tokens have disquotational truth

conditions. For

example,

I know that

iff grass

is green is green" are true

(***) My tokens

of "Grass

is true as uttered by me so long as the metalanguage to which

my tokens of (***)

belong contains the object language to which my tokens of "Grass

is green"

of containment does hold. It

is not as though this relation

belong.

It is obvious that

from

the one I use is green"

belong to a different language

my tokens

of "Grass

to make semantic claims about these tokens, such as the claim in (***).

down version of the Putnamian

at a stripped

Using this point, we can arrive

argument:

(I) My sentence

tokens have

of the semantics

condit-

[knowledge

disquotational truth

of my language]

ions, as in (***)

(II) If I am a BIV, then my

sentence tokens have non-

conditions [causal-theoretic semantics]

disquotational truth

(III) Iam not a BIV

[(I),(II)]

as follows:

This argument could, of course, be put into the form

of a reductio

[assumption]

(i)

I am a BIV

of the semantics

[knowledge

(ii)

My sentence

tokens have

of my language]

conditions

disquotational truth

tokens have non-

(iii)

My sentence

conditions

[(i), causal-theoretic semantics]

disquotational truth

(iv) I am not a BIV

which is similar to Dell'Utri's.3 In both

So in the end we have an argument

to derive the claim that

my

of envattedness

we use the assumption

arguments,

In both

arguments,

we

conditions.

tokens have non-disquotational truth

sentence

take it as a premise that

rMy sentence tokens, instead,

have disquotational truth

infer the

contradictory

of

we immediately

conditions.

In the

simpler argument,

once we have derived the

foregoing

contradiction. In

the assumption for reductio

more

Dell'Utri's argument, things

are

complex.

The greater complexity,

we have

discussion.

See also Brueckner (1990) for a related

3

Unlike Dell'Utri's argument, the new argument is stated

in the first person. This is

I can

claim to know that

our

utterances have disquotational truth

that

it is not

clear

because

to as-

This obviously depends

upon who "we"

are (though

Dell'Utri appears

conditions.

refers to people who speak

a common language).

"we"

surie that

2

See Brueckner (1986) for a discussion

of this objection. For

a reply,

see Hill (1990). 128 Anthony Brueckner

now seen, is unnecessary. Further, we now have a way of rationalizing the baf-

in which we found ourselves when confronted

fling situation with the apparent

equivocation in Dell'Utri's argument.

It is unclear whether Dell'Utri would

endorse the approach

sketched here. The important question which remains

unsettled is whether the stripped

down Putnamian argument is ultimately suc-

cessful. This is a complex matter for discussion on another occasion.

Department of Philosophy

at Santa Barbara

University of California

Santa Barbara

93106

California

USA

REFERENCES

in a Vat".

Brueckner, A. 1986: "Brains

Journal of Philosophy,

83, 3, pp. 148-67.

1990: "Scepticism about

Knowledge of Content". Mind, 99, 395, pp. 447-

51.

Dell'Utri, M. 1990: "Choosing of Realism: The Case of the Brains

Conceptions

in a Vat".

Mind, 99, 393, pp. 79-90.

Hill, C. S. 1990: "Review of Pettit,

P. and McDowell, J. eds. Subject,

Thought,

and Context".

Journal of Philosophy,

87, 2, pp. 106-12.

Putnam,

H. 1981: Reason, Truth and

History.

Cambridge: Cambridge University

Press.

ANTHONY BRUECKNER