OTHER VOICES, OTHER MINDS
Theodore
J. Everett
I.
Solipsism
Solipsism is the view
that only myself, or my mind and its contents, is real. There are no other minds; there is no
external world at all - just me and my sensations, plus their strictly internal
relations. It may well be, as has often
been remarked, that no one genuinely believes in solipsism. What makes the theory important is that it
has been claimed to represent the limits of what can actually be known about
the world. (Thus we can distinguish
between metaphysical and epistemological statements of solipsism, where the
epistemological may be identified with the claim that the metaphysical cannot
be refuted. I will not fuss about this
technicality.) We can know what our
thoughts are like, hence we can know what we think the world is like,
but we can never find out if it is actually that way. As far as each of us can tell subjectively,
after all, he might be a proverbial brain in a vat, being fed his subjective
experiences through a set of wires.
Nothing rules this out a priori, and no feature of subjective
experience in itself can distinguish between this or another 'demon
hypothesis', and the hypothesis that our experiences represent reality fairly
accurately.
A main goal of
traditional epistemology has been the refutation of this view, based only on
the internal, subjective features of one's own experience. Descartes's Meditations is the prime
example of such an attempt to prove that the way things really are is basically
the way they seem to us to be. But
epistemologists are nearly unanimous, these days, in viewing Descartes's
efforts as a failure, and many now despair of Descartes's project as a
whole. Instead, they view any effort to
refute solipsism 'from the inside' as either inherently futile, or, worse,
completely misguided. Externalists say
that any conclusive argument against solipsism must depend on objective
considerations, such as the idea that natural selection favors reliable mechanisms
for belief production. If such arguments
plainly beg the question that traditional epistemologists thought they were
asking, then so much the worse for those epistemologists, since no other kind
of answer can be given. Others, such as
Wittgenstein and Ryle, have argued that traditional questions about other minds
and the outside world rely on a mistake about the essentially public meanings
of words like 'real' and 'actual', and of mental terms in general. Solipsism cannot be refuted, they claim, because
it makes no sense in the first place.
I do not intend to
attack either of these views. Instead, I
will show that a successful internalist response to solipsism can, in fact, be
made. If the traditional problems of other
minds and the external world can be shown to have plausible solutions, as I
claim they do, then there will be somewhat less reason, at least, to adopt
alternative approaches to epistemology in general.
I should also note
that I will not claim by my arguments to have refuted skepticism, if
that is identified with the proposition that all things are doubtful. In particular, my arguments take both the
reliability of inductive inference and the veracity of memory for granted, as
well as pure, deductive reasoning. This
is a limitation, of course, but not a weakness in my view. For surely, everything that we believe can be
doubted; and if everything were doubted at the same time, then nothing could be
said with confidence at all. The refutation of such extreme, global skepticism
is plainly impossible. Solipsism, on the
other hand, or skepticism with respect to other minds and the external world,
is a definite view which allows that some few things are known. It is refutable in principle, then, provided
one can make sufficient use of the few tools and materials at hand.
II.
Second-order induction
I claim that new
solutions to the problems of other minds and the external world are possible,
by way of a technique that I call second-order induction. This is not a different kind of argument, but
just a special case of regular induction.
Where regular induction is about the properties of objects in general,
second-order induction is based specifically on the truth or falsity of
propositions. Instead of reasoning from this or that raven's being black, say,
to the assertion that all ravens are black, one might reason from the fact that
a certain group of statements is true to the conclusion that all similar
statements are true. So if, for example, all those statements found in the Encyclopedia
Britannica which I have verified so far have turned out to be true, then I
have inductive reason to believe that all of the statements in that work are
true. If I then discover the statement,
'all ravens are black', in the same encyclopedia, I will have an inductive
reason to believe that this new statement is also true - hence, that all ravens
are black. This inference succeeds, even
if I have never seen a single raven myself - indeed, even if I am blind, and
even if I do not know exactly what a raven is.
In this way, second-order induction can function as an indirect means of
confirming propositions which are otherwise unverifiable.
Consider the examples
below:
First-order Induction
All observed ravens
are black.
Therefore, all ravens
are black.
Therefore, this
(unobserved) raven X is black.
Second-order Induction
All verified
statements in the Encyclopedia Britannica (EB) are true.
Therefore, all
statements in the EB are true.
Therefore, this
(unverified) statement X in the EB is true.
Therefore, X.
Note that the first three sentences in
the second-order case are identical in form to the entire first-order
case. They just make reference to true
statements instead of to black ravens.
So far, then, there is nothing obviously special about the second-order
case. What is special is that
second-order cases of induction have a fourth step, where one can reason from
the fact that some statement is true, to the statement itself. (This is just by way of Tarski's famous
definition: The statement 'X' is true, if and only if, X.) So substitute the statement 'All ravens are
black' for X in the second argument. If
it says in the Encyclopedia Britannica that all ravens are black, and this
gives us reason to think that the statement is true, then we now have a reason
to believe something about ravens, based on evidence not about ravens.
This technique is
straightforwardly applied to the refutation of solipsism. Here is how I am justified in believing in
other minds. First, other people often
tell me things that I can verify subjectively, for example, that I am about to
feel some pain. I can discover in this
way that certain other people are reliable sources in general. Next, these same other people happen to tell
me that they have minds (or have their own pains, etc.). By induction on their prior testimony, I now
have reason to believe that this new statement is also true. Hence, I have reason to believe that other
minds exist. The same is true for the
external world in general. If other
people tell me that the world is external to, or independent of, my thoughts,
and I have prior reason to believe that those people are reliable, then I have
reason to believe that what they tell me this time will also be true. Hence, I have reason to believe in the external
world.
Does this argument
even make sense? In order to trust what
someone else tells me, do I not need to know already that they exist, and that
they mean what they mean by what they say?
No. The perceptible statements of
others form a part of my own stream of sensory experience. These perceptions of testimony can be
correlated with other experiences, as a child learns to associate sounds like
'mommy' and 'dog' with certain clusters of visual and other sensations. I can learn, gradually, by generalizations
based solely on such regular conjunctions, that the statements of certain
others (e.g. my parents) are reliable.
Thereafter, I am justified (to some extent, at least), upon experiencing
tokens of any statement, in the perceptible voice of a reliable other
person, to believe that that statement is true.
Thus it is not, after all, necessary that I know in advance that others
have minds, or even that they exist outside of my imagination, in order for me
to have reason to believe what they say.
If someone ordinarily reliable says 'here is your dinner', I should
believe him, and expect to experience some food. If that person says 'there is water on the
planet Mars', there may be nothing in particular that I should expect by way of
a confirming experience. But I have
reason to believe it anyway, since I have reason to rely on the rule
that whatever this person says is true.
When the same person says 'I am in pain', or 'I have a mind', or 'the
world is independent of your thoughts', I have reason to believe these
statements as much as any other in the same voice, by virtue of the same
inductive rule. That there can be no
directly confirming experiences for such beliefs is irrelevant to my
justification.
Perhaps not every
reader is convinced. Some may feel that
this quick argument to an important conclusion simply must be circular, or must
involve some other kind of unfair trick.
So let me go through the whole argument again, more slowly and a bit
more thoroughly.
III.
Observational Reliability
Here is my main
preliminary argument. I want to show
that it is possible for one person to find out that another's testimony is
reliable, prior to any solution to the problems of other minds and the external
world. I say that one can learn this,
very gradually, by ordinary inductive reasoning, through observing correlations
between people's utterances and other observable events.
Assume for the moment
that I have sensible knowledge of ordinary physical events, but know nothing as
yet about other minds and their contents, or about the intended meanings of
their words and sentences. (These are
the traditional assumptions governing the problem of other minds, as distinct
from the problem of the external world.)
I claim that I can still make reasonable inductive inferences from and
about the utterances of other people.
(1) Denny Stampe says, 'there is a
chicken in the truck.'
In order to make sure that this is
understood according to my minimal assumptions, I will replace it with:
(2) o-Denny Stampe o-says, o-'there is
a chicken in the truck.'
Here, the 'o-' (observational) versions
represent the sensible surfaces of the ordinary referents of their terms. O-Denny Stampe could be a robot, for all I
know at this point. O-saying is the mere
production of sounds, meaningful or not.
And an o-sentence is only a string of such sounds, in a certain
recognizable pattern, but with nothing semantic built in.
So I hear a complex
sound coming from this thing o-Denny Stampe, the sound o-'there is a chicken in
the truck'. I happen to look in the
truck, and there I find a chicken. He
o-says the same thing again later, and I find another chicken in the
truck. If inductive knowledge is
possible at all, it must be possible for me to associate such sounds coming
from this source with the appearance of chickens in trucks. Similarly for other o-utterances and events:
a system of patterns in these sounds turns out to match up with a variety of
other features of the observable world.
In this crude, Quinean way, through many thousands of such correlations,
I can come to 'understand' the code in which o-Denny Stampe's o-statements
appear. This only means that I learn
that I can use these utterances predictively with some success, if I interpret
them as statements in such a code. What
is important is that ultimately, I can learn the following general fact about
this source of sounds:
(3) o-Denny Stampe is o-reliable.
That is, I can learn that his
o-statements are associated with true propositions in a certain reliable way -
more simply, that what he o-says, as I interpret it, is usually true. This o-reliability is plainly an adequate
basis for second-order induction. If
this o-person's o-statements are usually true, then his future o-testimony will
count as evidence for the truth of whatever propositions I have learned
inductively to associate with it.
IV.
Other Minds
I claim that
o-reliability is a proper basis for inductive inference, not just regarding
observable, physical events, but also regarding mental events, or anything else
that one can talk about. Consider
(4) o-Denny Stampe o-says o-'I am in
pain.'
The truth of (3) means that (4) is
evidence for the truth of the proposition I associate with this o-utterance of
'I am in pain', because (3) means that I have (some) reason to believe that whatever
this o-person o-says is probably true, according to the scheme of
interpretation I have learned to impose on it.
And if whatever he o-says is probably true, then the o-statement in
apparent reference to his private mental life is probably true, as well.
This is not an
abnormal induction. If we want some
evidence about the dark side of the moon, for example, we can look at the
visible side, and that gives us reason to believe that the whole thing is a
certain way, e. g. covered with craters.
If the whole thing is probably a certain way, then probably the part we
cannot see is that way, too. This is
just how any instance of induction works.
The only 'trick' to
this argument lies in its applying induction, not to the first-order facts
themselves in question, but to the truth of statements of them. Ordinarily, there is nothing to choose
between the two: the statement, 'all ravens are black', is true, after all,
just in case all ravens are black. But
here, in the case of other minds, I have no direct access in principle to most
of the facts that I am interested in. So
it makes all the difference to induce over the observable statements other
people make, given that I can find out empirically whether these sources can be
trusted in general. By going up a
logical level to considerations of reliability, and then back down to
first-order conclusions (e. g. that Denny Stampe is in pain) I am able to jump
over the wall of unobservability which separates my mind from every other.
Compare this to the
traditional argument from analogy, which is a first-order inductive argument
for the existence of other minds.[1] The idea is that in my own case, I can
associate mental states with physical events - feeling pain when something
falls on my foot, followed by my jumping up and down, etc. This gives me some kind of reason to believe
that other people are also experiencing pain whenever they are jumping up and
down after dropping things on their own feet, etc. But not much reason, unfortunately, since I
am forced to generalize from only one observed case of a body with a mind,
namely mine, to the conclusion that minds accompany human bodies in general. By
contrast, in my second-order argument (as in most standard inductions) I can
build up as much evidence as I want to for the general claim (i. e. that other
people are reliable) first, before drawing any conclusions about unobserved
cases.[2]
V. The
External World
My inductive argument
can be pushed a little deeper. In
addressing just the problem of other minds, I have assumed that the major claim
of solipsism is false, and taken as given the reliability of my senses and the
existence of the physical world. But
suppose that I do not yet know that there are physical objects, or an outside
world at all. As far as I can tell, my
senses might be radically deceptive, or even totally unconnected to any outside
source of information. Still, as long as
I am rationally capable of forming 'objects' out of patterns of sense-data (or
whatever else is held to be epistemically immediate), I can make inductive
inferences about the phenomenal sounds that accompany the presence of various
other such patterns, such as the clusters of phenomena that I usually take to
represent trucks, chickens, Denny Stampe, etc.
Statement (2) above can then be replaced with
(5) p-Denny Stampe p-says p-'there is
a chicken in the truck'.
Although he may be an hallucination, the
phenomenal object that I want to call Denny Stampe can still be associated with
the sounds that accompany his presence in my mind. As before, I can learn from sufficiently
varied experience that this thing is a reliable source of information in
general. Whenever I hear sounds in the
pattern, 'there is a chicken in the truck', I find phenomenal chickens in
phenomenal trucks. I discover that a different sound, 'here is an apple',
predictively appears before this other round, red sort of phenomenon. In the same way, I learn to distinguish 'here
is an apple' from 'there is an apple', 'there are no apples', 'I will bring you
an apple' and so on, to the point where I have cracked inductively the bulk of
the code in which these messages seem to appear. The phenomenal 'meanings' I attach to
individual terms will be highly indeterminate, no doubt, especially at
first. But I can still eventually gain,
by this process, sufficient reason to believe that
(6) p-Denny Stampe is p-reliable.
That is, I can find out inductively that
the sounds I associate with this apparent person, as I have come to
understand them, usually represent true propositions.
Now let my source
again say something useful:
(7) p-Denny Stampe p-says p-'I exist
independently of your thoughts'.
I can understand well enough what this
means. The terms 'I' and 'your thoughts'
can be learned subjectively, by way of fairly simple ostensions. The ideas of existence and dependence, if
they are not held to be innate, can be demonstrated with some further trouble
(through experience with plenty of statements like 'the chicken in your truck
does not exist - I was only joking', 'the size of these apples depends on the
condition of the tree', and so on). Now,
even if I have been supposing that Denny Stampe is merely an hallucination, my
justified belief in (6) gives me reason to believe that the proposition I
understand from (7), to the effect that he is not just an hallucination,
is true. From this point on, he can continue to enlighten me about the nature
of the outside world with greater and greater efficiency. As long as he continues to perform reliably,
as far as I can tell, I will have reason to believe it all.
I
should note a limitation of this second argument, as it applies to the
distinctly physical (i. e. non-mental) side of the
external world. Ordinarily, we think of
our knowledge of physical reality as coming before our knowledge of other
minds, and being stronger. But in my
view this order of priorities should be reversed. According to my argument, we know of the external
physical world through our knowledge of other minds, or at least of other
truth-telling external things.[3] The argument here for a distinctly non-mental
outside reality is thus dependent on, and turns out to be weaker than, the
argument for other minds.
From the
point of view of an ordinary small child, there will be little difference
between the second-order evidence for other minds and the second-order evidence
for physical objects, because he is not in any position seriously to doubt what
the adults around him say in either case.
In normal instances, the parents of a small child serve not just as
reliable sources of information, but as a set of epistemic authorities - that
is, sources that he knows to be more reliable than himself (i. e. than all of
his first-order epistemic resources combined).
Thus a child, up to a certain age, will find it rational to believe in
Santa Claus, say, merely on his parents' say-so, despite whatever reasons of
his own he may have for doubt. When the
child grows up, his parents' statements will normally still count as evidence
for him, but not much more than those of other competent adults; they will not
automatically trump his independent judgement. This is because he comes to
realize that his own, first-order opinion on most things is about as reliable
as his parents' or anybody else's, with the exception of eyewitnesses, or
others in an especially good position to know something.
It is
similar with one's second-order evidence for the existence of an external
world, apart from other minds. A child
or lay adult may well have good second-order evidence of the physical world in
the ordinary course of life, but one who considers the matter carefully will
come to realize that other voices in general have a limited evidential value
with respect to metaphysics, both because most people are not very reliable on
such issues, and because their reliability in this respect is very hard to
gauge. This will not matter for as long
as one is inclined to agree with his fellows, anyway. But if one encounters
skeptical arguments, and finds them persuasive, things may change. In the case of other minds, every other
person seems ideally placed to know whether he has thoughts and feelings
(though some things about one's mental life may take an outside expert to
know). So an adult can reestablish the
likelihood of the existence of other minds - that aspect of an external reality
- more or less at will, by following the
main argument in this paper. But why
should an adult believe in the physical realm on the basis of other people's
say-so, when the others are in no better epistemic position with respect to the
issue than oneself? Perhaps, if the
other is an able and experienced philosopher, he would be better placed to know
about this than the layman; but the layman can easily learn that there are many
such experts, and that they disagree on the matter. I suppose it ought to give one heart that most
philosophers are not convinced of skepticism, even if no individual philosopher
stands out as an authority. But the fact
that this is seriously controversial, with great thinkers on both sides, leaves
the second-order question rather murky.
VI. A
Problem About Meanings
It may be suggested
that I cannot fully understand, within the limits of my solipsistic
predicament, the statements that Denny makes by producing all these noises,
etc. How can I claim justifiably to
believe such statements, if I cannot even determine precisely which
propositions are being asserted? How can
I know, for example, prior to an understanding of the actual semantics of the
English word, that it is pain that I am thinking about, when I say to
myself that I am in pain? How can I know
that it is pain that I wish to attribute to others, when I hear from them the
mere sound (not the English sentence) 'I am in pain'?
My response is that
we must allow beliefs to be vague, or else it will turn out that most people
have no beliefs at all. Even for someone
in the primitive epistemic position solipsism entertains, a requirement that
beliefs should be fully articulate, and fully determinate in meaning, is
plainly unreasonable. It is surely
sufficient that beliefs should be articulated in a general, ballpark way at
first, subject to clarification through further research. In the case of learning to attribute the word
'pain' to myself, for example, this is initially a matter of the crudest
induction. I hear a type of sound, 'pain', and associate it with a type of
feeling that occurs at the same time.
The boundaries of such types will be vague, of course, but if I am to
rely on any faculty of internal discrimination at all, I must be allowed to
reidentify these same rough types, at least provisionally. To attribute pain to others - and still mean
pain, not something about their behavior - I need only intend that they are in
a state that is somehow importantly similar to the states that I am in when I
hear others say that I am in pain - that is to say, in a state like this,
where this is what the word 'pain' brings to mind.
Moreover, as soon as
I am in a position to use someone else's testimony as a prima facie
ground for belief, I will also be able to receive explicit semantic information
from the same source. Thus I can be
guided toward more and more refined understandings of my sources' statements by
their own explanations of what they mean.
Even if Denny Stampe is, for all I know, a robot (or even an
hallucination) he can still seem to point to things and seem to give me
definitions. If he says, 'when I say
that I am in pain, I mean that I am in a private state which is caused by the
same things that cause you to experience what you experience when you are
inclined to say that you are in pain, under certain normal conditions,
etc.', this may be harder for me to
figure out than what he tells me about chickens. But it is not different in principle. And again, I do not have to figure out exactly
what he means by such statements. I can
derive inductive benefit from them as soon as I have the roughest, purely
ostensive idea of how they are being used.
The Wittgensteinian
concern about private meanings can certainly be pushed to more radical lengths,
into a general critique of internalism, which is beyond the scope of this
paper. So let me phrase my answer
hypothetically. If solipsism is held to
make sense as a doctrine - that is, if the problems of other minds and the
external world are to be taken seriously - then there is no special
problem about meanings that results from my solution. The thought that solipsism might be true is,
after all, one of my thoughts - and by hypothesis, transparent to me. If I can know what I mean when I ask myself
if there are other minds, or an external world, then presumably I can know what
I mean when I answer.
Second-order
induction does not necessarily depend, after all, on information coming from an
outside source. I can make inductive
inferences about my own beliefs, as well as about other people's
statements. Suppose that I discover that
my own mere inclination to believe a proposition is a reliable predictor of its
being true. That is, suppose that when I
notice in myself an inclination to believe something, this is usually followed
by some other kind of evidence for its truth. (This is what it would be like, I
suppose, to find out that one is psychic.)
Now suppose that I simply find myself, one day, believing in other
minds. Do I not now have a reason to
believe in other minds? It seems I must,
if I am to count on induction at all.
What seems strange
about this example is not the form of the argument, but only the assumption
that one should pull his beliefs out of thin air, and that such beliefs should
turn out to be true. As it happens (for
most of us, at least), one's belief in the existence of another mind, like most
of one's beliefs, does not just pop into his head, but rather occurs to him in
connection with his evidence for it, by way of an inference that he is at least
potentially aware of. A normal person
does not count himself as psychic, because he believes as he does only on the
basis of such explicable evidence.
But there is a point
at which the difference between a normal reasoner and a genuine psychic
disappears, namely when one considers his own inferences in an abstracted way -
not as coming from within oneself, as it were, but simply as part of one's
experience of the world. In this way, my
beliefs really do just pop into my head, although I connect them with a chain
of similar 'poppings' that I like to think justifies them. Perhaps I have no better ultimate reason to
believe this than the fact that such beliefs, occurring as conclusions of what
I have taken to be proper inferences, have been reliably reconfirmed at various
times in the past. So the mere fact that I find myself believing something does
count as a reason for me to continue to do so, given my track record as a
believer so far. I suspect that this
kind of ongoing, overall inductive self-support provides an important element
of continuity to our epistemic and psychological lives. Of course, it depends on an initial, and
recurring, set of facts about the particular phenomena that we experience. Not just any set will do; only one that hangs
together in a certain, good way.
This displays, I
think, the essentially uncircular, empirical nature of my argument for other
minds. When I conclude that Denny Stampe
is a reliable source of information, this depends on the entirely contingent
fact that the sounds he makes cause verifiable propositions to occur to
me. The objective explanation for this
fact is, of course, that Denny is a person, making statements with semantic
content which I understand. But from
within the solipsistic bubble, as it were, his testimony appears as a faculty
of mine. I 'hear voices', which
is only to say that I have certain experiences.
I do not know, initially, where they come from. But I am able to discover that they 'tell the
truth', which is just to say that I find myself believing propositions,
correlated to these sounds, that I am able to verify. It is an empirical fact that I have this
reliable faculty of other-people's-testimony, and that this faculty provides me
with information of epistemological interest.
If nobody had ever said anything to me about their minds, etc., then I
would have remained in the dark about such things. It is a matter of luck, then, for each
of us, that he should be provided with his own solution to the problems posed
by solipsism. Not the blind, magical
luck of the psychic, however - just the decent luck of living around other
people who are willing to talk.
VII.
Conclusion
I am afraid that my
arguments, even if accepted, may still look like merely technical, tricky
solutions to the problems of the external world and other minds. I believe, however, that they do represent,
if a bit artificially, the way that we really come to know these facts. For surely, children learn the basic nature
of the outside world and other people, not by way of raw, 'autistic'
experiments with undifferentiated sensations, but through interaction with
their parents, by way of learning to pick out their voices, and to depend on
what they say. Of course, this process
is unlikely to proceed along the exact, crisply rational lines of my argument,
and may well be aided by a set of biological predispositions, as Noam Chomsky
and others suggest. Still, to the extent
that reasons play a role in fundamental learning, the essential reasons are, I
claim, of just the sort that I have discussed.
In any event, almost
all of our knowledge as adults relies implicitly on second-order induction,
regardless of how we initially acquired the beliefs in question. As Augustine argued, most of our beliefs
about geography, history, medicine, etc. - even who our parents are - depend
crucially on the testimony of experts, i. e. trustworthy others, who we believe
are in positions to know.[4] No reasonable person would attempt to work
out all of physics for himself, for example, without consulting other
physicists or teachers. Why should it be
different for metaphysics? It seems
different, at the fundamental level, because it seems that our access to
testimony is blocked off by the challenge of solipsism. If I am right, and it is not, then this
common form of evidence may well suffice.[5]
[1]
Bertrand Russell denies that this argument is precisely inductive,
preferring to reserve that term for arguments which extend only to potentially
observable new instances. See his Human
Knowledge: Its Scope and Limits (New York: Simon and Schuster, 1948), p.
193.
[2] Stuart Hampshire has an expanded version
of the analogical argument, which is the closest thing I have found to my view
in print. He says that one can gather
evidence for other minds, not just from the correlations between one's own
feelings and behavior, but also from the observed correctness of 'methods of
inference' used by others to establish one's own states of mind. See Hampshire, 'The Analogy of Feeling', Mind
61, no. 241 (January 1952), 1-12.
[3] There may, of
course, be other, first-order arguments for one's belief in the physical world,
e. g. the argument that physical causes best explain the regularities among our
sense impressions.
[4] See Augustine's Confessions
6.5. This is a crucial premise of his
argument for Christianity, given the greater reliability of Christian (as
opposed to Manichean, etc.) sources with respect to verifiable claims. David Hume rebuts the use of such arguments
for supernatural conclusions in his famous chapter on miracles (Enquiry
10), but not the general idea that testimony is a main source of inductive
evidence. This topic was largely ignored
in recent decades, prior to the publication of John Hardwig's 'Epistemic
Dependence', The Journal of Philosophy 82, no. 7 (July 1985), 335-349,
and especially C. A. J. Coady's Testimony: A Philosophical Study
(Oxford: Clarendon Press, 1992). Much of
the ensuing discussion has focussed on whether we have, or need, inductive
justification for the reliability of others' testimony. See, for example, Elizabeth Fricker, 'Telling
and Trusting: Reductionism and Anti-Reductionism in the Epistemology of Testimony',
Mind 104, no. 414 (April 1995), 393-411, and Jack Lyons, 'Testimony,
Induction and Folk Psychology', Australasian Journal of Philosophy 75,
no. 2 (June 1997), 163-178. Coady
argues, following Thomas Reid, that Hume's inductive view of testimony (which I
adopt implicitly in the present paper) does not work, and that testimony must
be viewed instead as a basic, irreducible (though not incorrigible) source of
knowledge, roughly equivalent to perception or memory. On this view, one's belief in other minds
must be taken almost for granted.
[5] I would like to thank Dennis Stampe, Ellery
Eells, Berent Enc, Alan Sidelle, Richard Fumerton, an audience at SUNY-Geneseo,
and two anonymous referees for their helpful comments.