Professional Documents
Culture Documents
Once upon a time there was a tough- ism appeared to exhaust the alternatives.
minded philosopher who said, 'What is all Compromises were attempted ('double
this talk about "minds", "ideas", and "sen- aspect' theories), but they never won
sations"? Really-and I mean really in the many converts and practically no one
real world-there is nothing to these so- found them intelligible. Then, in the mid-
called "mental" events and entities but 1930s, a seeming third possibility was dis-
certain processes in our all-too-material covered. This third possibility has been
heads.' called logical behaviorism. To state the
And once upon a time there was a nature of this third possibility briefly, it is
philosopher who retorted, 'What a master- necessary to recall the treatment of the
piece of confusion! Even if, say, pain were natural numbers (i.e. zero, one, two,
perfectly correlated with any particular three ... ) in modern logic. Numbers are
event in my brain (which I doubt) that identified with sets, in various ways, de-
event would obviously have certain prop- pending on which authority one follows.
erties-say, a certain numerical intensity For instance, Whitehead and Russell iden-
measured in volts-which it would be tified zero with the set of all empty sets,
senseless to ascribe to the feeling of pain. one with the set of all one-membered sets,
Thus, it is two things that are correlated, two with the set of all two-membered
not one-and to call two things one thing sets, three with the set of all three-mem-
is worse than being mistaken; it is utter bered sets, and so on. (This has the ap-
contradiction.' pearance of circularity, but they were able
For a long time dualism and material- to dispel this appearance by defining 'one-
membered set', 'two-membered set', 'three-
membered set', etc., without using 'one',
From R. J. Butler, ed., Analytical Philoso- 'two', 'three', etc.) In short, numbers are
phy, vol. 2 (Oxford: Blackwell, 1965). Reprinted treated as logical constructions out of sets.
by permission of the author. Notes have been re- The number theorist is doing set theory
numbered for this edition. This paper was read as without knowing it, according to this in-
part of the program of the American Association
for the Advancement of Science, Section L (His- terpretation.
tory and Philosophy of Science), December 27, What was novel about this was the
1961. idea of getting rid of certain philosophi-
2. Brains and Behavior 25
Moreover, although this is an empirical One might well stop to ask whether
fact, it underlies the possibility of talking this statement has even a clear meaning.
about pains in the particular way in which Supposing that there are Martians: do we
we do. However, it does not rule out in have any criterion for something being an
any way the possibility of worlds in which 'unconditioned pain response' for a Mar-
(owing to a difference in the environmen- tian? Other things being equal, one avoids
tal and hereditary conditions) pains are things with which one has had painful ex-
not responsible for the usual responses, or periences: this would suggest that avoid-
even are not responsible for any responses ance behavior might be looked for as a
at all. universal unconditioned pain response.
Let us now engage in a little science However, even if this were true, it would
fiction. Let us try to describe some worlds hardly be specific enough, since avoid-
in which pains are related to responses ance can also be an unconditioned re-
(and also to causes) in quite a different sponse to many things that we do not as-
way than they are in our world. sociate with pain--to things that disgust
If we confine our attention to non- us, or frighten us, or even merely bore us.
verbal responses by full grown persons, Let us put these difficulties aside, and
for a start, then matters are easy. Imagine see if we can devise an imaginary world in
a community of 'super-spartans' or 'super- which there are not, even by lenient stan-
stoics' -a community in which the adults dards, any unconditioned pain responses.
have the ability to successfully suppress Specifically, let us take our 'super-spar-
all involuntary pain behavior. They may, tans', and let us suppose that after mil-
on occasion, admit that they feel pain, but lions of years they begin to have children
always in pleasant well-modulated voices who are born fully acculturated. They are
-even if they are undergoing the agonies born speaking the adult language, know-
of the damned. They do not wince, ing the multiplication table, having opin-
scream, flinch, sob, grit their teeth, clench ions on political issues, and inter alia shar-
their fists, exhibit beads of sweat, or oth- ing the dominant spartan beliefs about the
erwise act like people in pain or people of not evincing pain (except
suppressing the unconditioned responses by way of verbal report, and even that in
associated with pain. However, they do a tone of voice that suggests indifference).
feel pain, and they dislike it (just as we Then there would not be any 'uncondi-
do). They even admit that it takes a great tioned pain responses' in this community
effort of will to behave as they do. It is (although there might be unconditioned
only that they have what they regard as desires to make certain responses-desires
important ideological reasons for behav- which were, however, always suppressed
ing as they do, and they have, through by an effort of will). Yet there is a clear
years of training, learned to live up to absurdity to the position that one cannot
their own exacting standards. ascribe to these people a capacity for feel-
It may be contended that children ing pain.
and not fully mature members of this To make this absurdity evident, let
community will exhibit, to varying de- us imagine that we succeed in converting
grees, normal unconditioned pain behav- an adult 'super-spartan' to our ideology.
ior, and that this is all that is necessary for Let us suppose that he begins to evince
the ascription of pain. On this view, the pain in the normal way. Yet he reports
sine qua non for significant ascription of that the pains he is feeling are not more
pain to a species is that its immature mem- intense than are the ones he experienced
bers should exhibit unconditioned pain prior to conversion-indeed, he may say
responses. that giving expression to them makes
30 Hilary Putnam
them less intense. In this case, the logical extent of waiving the first difficulty that
behaviorist would have to say that, comes to mind, and let us undertake the
through the medium of this one member, task of trying to imagine a world in which
we had demonstrated the existence of un- there are not even pain reports. I will call
conditioned pain responses in the whole this world the 'X-world'. In the X-world
species, and hence that ascription of pain we have to deal with 'super-super-spar-
to the species is 'logically proper'. But this tans'. These have been super-spartans for
is to say that had this one man never lived, so long, that they have begun to suppress
and had it been possible to demonstrate even talk of pain. Of course, each individ-
only indirectly (via the use of theories) ual X-worlder may have his private way
that these beings feel pain, then pain as- of thinking about pain. He may even have
criptions would have been improper. the word 'pain' (as before, I assume that
We have so far been constructing these beings are born fully acculturated).
worlds in which the relation of pain to its He may think to himself: This pain is in-
nonverbal effects is altered. What about tolerable. If it goes on one minute longer I
the relation of pain to causes? This is even shall scream. Oh No! I mustn't do that!
more easy for the imagination to modify. That would disgrace my whole family .. .'
Can one not imagine a species who feel But X-worlders do not even admit to hav-
pain only when a magnetic field is present ing pains. They pretend not to know ei-
(although the magnetic field causes no ther the word or the phenomenon to
detectable damage to their bodies or ner- which it refers. In short, if pains are 'logi-
vous systems)? If we now let the members cal constructs out of behavior', then our
of such a species become converts to X-worlders behave so as not to have
'super-spartanism', we can depict to our- pains!-Only, of course, they do have
selves a world in which pains, in our pains, and they know perfectly well that
sense, are clearly present, but in which they have pains.
they have neither the normal causes nor If this last fantasy is not, in some dis-
the normal effects (apart from verbal re- guised way, self-contradictory, then logi-
ports). cal behaviorism is simply a mistake. Not
What about verbal reports? Some only is the second thesis of logical behav-
behaviorists have taken these as the char- iorism-the existence of a near-translation
acteristic form of pain behavior. Of of pain talk into behavior talk-false, but
course, there is a difficulty here: If 'I am in so is even the first thesis-the existence of
pain' means 'I am disposed to utter this 'analytic entailments'. Pains are responsi-
kind of verbal report' (to put matters ble for certain kinds of behavior-but
crudely), then how do we tell that any only in the context of our beliefs, desires,
particular report is 'this kind of verbal re- ideological attitudes, and so forth. From
port'? The usual answer is in terms of the the statement 'X has a pain' by itself no
unconditioned pain responses and their behavioral statement follows--not even a
assumed supplantation by the verbal re- behavioral statement with a 'normally' or
ports in question. However, we have seen a 'probably' in it.
that there are no logical reasons for the In our concluding section we shall
existence of unconditioned pain responses consider the logical behaviorist's stock of
in all species capable of feeling pain (there counter-moves to this sort of argument. If
may be logical reasons for the existence of the logical behaviorist's positive views are
avoidance desires, but avoidance desires inadequate owing to an oversimplified
are not themselves behavior any more view of the nature of cluster words-
than pains are). amounting, in some instances, to an open
Once again, let us be charitable to the denial that it is possible to have a word
2. Brains and Behavior 31
is always stated, does have testable conse- this 'argument' not by challenging the
quences (perhaps some more complex premiss, be it overt or covert, that 'untest-
hypothesis does not; but then we have to able synthetic statement' is some kind of
see this more complex hypothesis stated contradiction in terms (although I believe
before we can be expected to discuss it). In that premiss to be mistaken), but simply
the second place, the usual argument for by showing that, on any but the most
the absurdity of this hypothesis rests on a naive view of testability, our hypothesis is
simplistic theory of the meaning of 'length' testable.
-and a full discussion of that situation is Of course, I could not do this if it
hardly possible without bringing in con- were true that 'by hypothesis, we couldn't
siderations from unified field theory and distinguish X-worlders from people who
quantum mechanics (the latter comes in really didn't know what pain is'. But that
connection with the notion of a 'material isn't true-at any rate, it isn't true 'by
standard'). But, the example aside, one hypothesis'. What is true by hypothesis is
can hardly challenge the point that a su- that we couldn't distinguish X-worlders
perficially coherent story may contain a from people who really didn't know what
hidden absurdity. pain is on the basis of overt behavior
Or can one? Of course, a superficially alone. But that still leaves many other
coherent story may contain a hidden con- ways in which we might determine what
tradiction, but the whole point of the logi- is going on 'inside' the X-worlders-in
cal behaviorist's sneering reference to both the figurative and literal sense of
'surface grammar' is that linguistic coher- 'inside'. For example, we might examine
ence, meaningfulness of the individual their brains.
terms, and logical consistency, do not by It is a fact that when pain impulses
themselves guarantee freedom from an- are 'received' in the brain, suitable electri-
other kind of absurdity-there are 'depth cal detecting instruments record a charac-
absurdities' which can only be detected by teristic 'spike' pattern. Let us express this
more powerful techniques. It is fair to say briefly (and too simply) by saying that
that to-day, after thirty years of this sort 'brain spikes' are one-to-one correlated
of talk, we lack both a single convincing with experiences of pain. If our X-worlders
example of such a depth absurdity, and a belong to the human species, then we can
technique of detection (or alleged tech- verify that they do feel pains, notwith-
nique of detection) which does not reduce standing their claim that they don't have
to 'untestable, therefore nonsense'. any idea what pain is, by applying our
To come to the case at hand: the logi- electrical instruments and detecting the
cal behaviorist is likely to say that our tell-tale 'brain spikes'.
hypothesis about 'X-worlders' is untest- This reply to the logical behaviorist is
able in principle (if there were 'X-world- far too simple to be convincing. 'It is true,'
ers', by hypothesis we couldn't distinguish the logical behaviorist will object, 'that
them from people who really didn't know experiences of pain are one-to-one corre-
what pain is); and therefore meaningless lated with "brain spikes" in the cas.e of
(apart from a certain 'surface significance' normal human beings. But you don't
which is of no real interest). If the logical know that the X-worlders are normal hu-
behaviorist has learned a little from 'ordi- man beings, in this sense-in fact, you
nary language philosophy', he is likely to have every reason to suppose that they
shy away from saying 'untestable, there- are not normal human beings'. This reply
fore meaningless', but he is still likely to shows that no mere correlation, however
say or at least think: 'untestable, therefore carefully verified in the case of normal
in some sense absurd'. I shall try to meet human beings, can be used to verify
2. Brains and Behavior 33
be in pain when he was, in fact, in pain. that he feels intense pain when you touch
Now consider what the 'misleading V- him with a red hot poker). On this basis,
waves' story requires: it requires that the we were quite willing to grant that the
X-worlders produce V-waves in quite a super-spartans did, indeed, feel pain-all
different way than we do, without speci- the more readily, since the deviancy in
fying what that different way is. More- their behavior had a perfectly convincing
over, it requires that this be the case, ideological explanation. (Note again the
although the reverse hypothesis-that X- role played here by considerations of co-
worlders' brains function exactly as hu- herence and simplicity.) But the X-world-
man brains do-in fact, that they are hu- ers also 'tell' us (and, perhaps, each other),
man brains-fits all the data. Clearly, this exactly the same things, albeit unwillingly
story is in serious methodological difficul- (by the medium of the involuntarily pro-
ties, and any other 'counter-explanation' duced 'V-waves'). Thus we have to say-
that the logical behaviorist tries to invoke at least, we have to say as long as the 'V-
will be in similar difficulties. In short, the wave' theory has not broken down-that
logical behaviorist's argument reduces to the X-worlders are what they, in fact, are
this: 'You cannot verify "psycho-physical" -just 'super-super-spartans'.
correlations in the case of X-worlders (or Let us now consider a quite different
at least, you can't verify ones having to argument that a logical behaviorist might
do, directly or indirectly, with pain), be- use. 'You are assuming,' he might say, 'the
cause, by hypothesis, X-worlders won't following principle:
tell you (or indicate behaviorally) when If someone's brain is in the same state as
they are in pain'. 'Indirect verification'- that of a human being in pain (not just at
verification using theories which have the moment of the pain, but before and
been 'tested' only in the case of human after for a sufficient interval), then he is in
beings-is not verification at all, because pain. Moreover, this principle is one which
X-worlders may obey different laws than it would never be reasonable to give up (on
human beings. And it is not incumbent your conception of 'methodology'). Thus,
upon me (the logical behaviorist says) to you have turned it into a tautology. But
observe what turning this principle into a
suggest what those laws might be: it is in-
tautology involves: it involves changing
cumbent upon you to rule out all other the meaning of 'pain'. What 'pain' means
explanations. And this is a silly argument. for you is: the presence of pain, in the
The scientist does not have to rule out all colloquial sense of the term, or the presence
the ridiculous theories that someone might of a brain state identical with the brain state
suggest; he only has to show that he has of someone who feels pain. Of course, in
ruled out any reasonable alternative theo- that sense we can verify that your 'X-
ries that one might put forward on the worlders' experience 'pain' -but that is not
basis of present knowledge. the sense of 'pain' at issue.
Granting, then, that we might dis- The reply to this argument is that the
cover a technique for 'reading' the un- premiss is simply false. It is just not true
spoken thoughts of X-worlders: we would that, on my conception of verification, it
then be in the same position with respect would never be reasonable to give up the
to the X-worlders as we were with respect principle stated. To show this, I have to
to the original 'super-spartans'. The super- beg your pardons for engaging in a little
spartans were quite willing to tell us (and more science fiction. Let us suppose that
each other) about their pains; and we scientists discover yet another kind of
could see that their pain talk was linguis- waves-call them 'W-waves'. Let us sup-
tically coherent and situationally appro- pose that W-waves do not emanate from
priate (e.g. a super-spartan will tell you human brains, but that they are detected
2. Brains and Behavior 35
emanating from the brains of X-worlders. There are two things that the logical
And let us suppose that, once again, there behaviorist can do: he can claim that as-
exists a simple scheme for decoding W- cribing pains to X-worlders, or even super-
waves into coherent English (or whatever spartans, involves a 'change of meaning'/
language X-worlders speak), and that the or he can claim that ascribing pains to
'decoded' waves 'read' like this: 'Ho, ho! super-spartans, or at least to X-worlders,
are we fooling those Earthians! They think is 'untestable'. The first thing is a piece of
that the V-waves they detect represent our unreasonable linguistics; the second, a
thoughts! If they only knew that instead piece of unreasonable scientific method.
of pretending not to have pains when we The two are, not surprisingly, mutually
really have pains, we are really pretending supporting: the unreasonable scientific
to pretend not to have pains when we method makes the unreasonable linguis-
really do have pains when we really don't tics appear more reasonable. Similarly,
have pains!' Under these circumstances, the normal ways of thinking and talking
we would 'doubt' (to put it mildly) that are mutually supporting: reasonable lin-
the same psycho-physical correlations guistic field techniques are, needless to
held for normal humans and for X-world- say, in agreement with reasonable con-
ers. Further investigations might lead us ceptions of scientific method. Madmen
to quite a number of different hypotheses. sometimes have consistent delusional sys-
For example, we might decide that X- tems; so madness and sanity can both
worlders don't think with their brains at have a 'circular' aspect. I may not have
all-that the 'organ' of thought is not just succeeded, in this paper, in breaking the
the brain, in the case of X-worlders, but 'delusional system' of a committed logical
some larger structure-perhaps even a behaviorist; but I hope to have convinced
structure which is not 'physical' in the the uncommitted that that system need
sense of consisting of elementary particles. not be taken seriously. If we have to
The point is that what is necessarily true is choose between 'circles', the circle of rea-
not the principle stated in the last para- son is to be preferred to any of the many
graph, but rather the principle: circles of unreason.
If someone (some organism) is in the same
state as a human being in pain in all rele- Notes
vant respects, then he (that organism) is in 1. E.g. these two points are fairly explic-
pain. itly stated in Strawson's Individuals. Strawson
-And this principle is a tautology by has told me that he no longer subscribes to
anybody's lights! The only a priori meth- point (1), however.
odological restriction I am imposing here 2. An attempted fourth alternative-i.e.
an alternative to dualism, materialism, and
is this one:
behaviorism-is sketched in "The Mental Life
If some organism is in the same state as a of Some Machines," which appeared in the
human being in pain in all respects known Proceedings of the Wayne Symposium on the
to be relevant, and there is no reason to Philosophy of Mind. This fourth alternative is
suppose that there exist unknown relevant materialistic in the wide sense of being com-
respects, then don't postulate any. patible with the view that organisms, includ-
ing human beings, are physical systems con-
-But this principle is not a 'tautology'; in
sisting of elementary particles and obeying the
fact, it is not a statement at all, but a laws of physics, but does not require that such
methodological directive. And deciding to 'states' as pain and preference be defined in a
conform to this directive is not (as hardly way which makes reference to either overt be-
needs to be said) changing the meaning of havior or physical-chemical constitution. The
the word 'pain', or of any word. idea, briefly, is that predicates which apply to
36 Hilary Putnam
a system by virtue of its functional organiza- compatible with saying that the cluster serves
tion have just this characteristic: a given func- to fix the meaning of the word. The point is
tional organization (e.g. a given inductive that when we specify something by a cluster of
logic, a given rational preference function) indicators we assume that people will use their
may realize itself in almost any kind of overt brains. That criteria may be over-ridden when
behavior, depending upon the circumstances, good sense demands is the sort of thing we
and is capable of being 'built into' structures of may regard as a 'convention associated with
many different logically possible physical (or discourse' (Grice) rather than as something to
even metaphysical) constitutions. Thus the be stipulated in connection with the individual
statement that a creature prefers A to B does words.
not tell us whether the creature has a carbon 4. Cf. "Dreaming and 'Depth Gram-
chemistry, or a silicon chemistry, or is even a mar,' "Analytical Philosophy, vol. 1.
disembodied mind, nor does it tell us how the 5. Of course, 'the cause' is a highly am-
creature would behave under any circum- biguous phrase. Even if it is correct in certain
stances specifiable without reference to the contexts to say that certain events in the brain
creature's other preferences and beliefs, but it are 'the cause' of my pain behavior, it does not
does not thereby become something 'myste- follow (as has sometimes been suggested) that
rious'. my pain must be 'identical' with these neural
3. I mean not only that each criterion events.
can be regarded as synthetic, but also that the 6. This is suggested in Wittgenstein's
cluster is collectively synthetic, in the sense Philosophical Investigations.
that we are free in certain cases to say (for rea- 7. This popular philosophical move is
son of inductive simplicity and theoretical discussed in "Dreaming and 'Depth Gram-
economy) that the term applies although the mar,'" Analytical Philosophy, vol. 1.
whole cluster is missing. This is completely