P. E.

Meehl and Wilfrid Sellars

being absorbed into the physical system. But apparently they want their - ----WILFRID SEll.ARS - - - - -
laws to be both effectual and at the same time no part of the physical
system ... (pp. 243--44)
First a terminological point. Among the various meanings of the word
physical let us distinguish the following for present purposes: Empiricism and the Philosophy of Mind
Physicali: an event or entity is physicali if it belongs in the space-
time network.
Physical2: an event or entity is· pJ1ysical2 if it is definable in terms
of theoretical primitives adequate to describe completely the actual
states. though not necessarily the potentialities of the universe pefore I. An Ambiguity in Sense-Datum Theories
the appearance of life. I PRESUME that no philosopher who has attacked the philosophical idea
Now, an emergentist account (of the kind we have been construct· of givenness or, to use the Hegelian term, immediacy, has intended to
ing) of raw feels denies that the latter are physica12 • But this in no wa;y deny that there is a difference ~etween interring t1Jat something is the
involves the denial that they are physicalt. And indeed this emergentist case and, for example, seeing it to be the case. If the term "given"
account definitely gives ,them a physicali .status. And if the equation~ referred merely to what is observed as being observed, or, perhaps,
a= g(q,r) to a proper subset of the things we are said to detem1ine by observa-
=
b h(s,t) tion, the existence of "data" would be as noncontroversial as the
permit the elimination of a and b from the descriptive function relat- existence of philosophical perplexities. But, of course, this just isn't
ing the physical2 variables q,r,s, and t, this fact, as we have just seen, so. The phrase "the given" as a piece of professional-epistemological-
by no means involves that the emergent entities with which the shoptalk carries a substantial theoretical commitment, and one can
vaiables a and b are associated roust also be phy_sica12 • deny that ther~ are "data" or that anything is, in this sense, "given"
Whether or not there are any emergents in the sense we have sought without flying in the face of reason.
to clarify is an empirical question. Our only aim has been to ~how Many things have been said to be "given": sense contents, material
that Pepper's "formal" demonstration of the impossibility of non- objects, universals, propositions, real connections, first principles, even
epiphenomenal emergents is invalid. givenness itself. And there is, indeed, a certain way of construing the
situations which philosophers analyze in these terms which can be
REFERENCE said to be the framework of givenness. This framework has been a
l. Pepper, Stephen C. "Emergence," Journal ot Pl1ilosopJ2y, 23:241-45 (1926). common feature of most of the major systems of philosophy, includ-
ing, to use a Kantian turn of phrase, both "dogmatic rationalism" and
"skeptical empiricism." It has, indeed, been so pervasive that few, if
any, philosophers have been altogether free of it; certainly not Kant,
and, I would argue, not even Hegel, that great foe of "immediacy."
Often what is attacked under its name are only specific varieties of
"given." Intuited first principles and synthetic necessary connections
No·~·v.:This paper w;1s first pr~scntcd as the University of London Special Lectures on
l'lulosophy for 1955- 56, delivered on March 1, 8, and 15, 1956, under lhe til'le
"The Myth of the Given: Three Ledurcs on E111pirids111 and the l'hilosophy of
Mind."

253
252
 Wilfrid Sellars RLVIPIRICISM AND THE PHILOSOPHY OF MIND

were the first to come under attack. And many who today attack "the 3. Now if we hear in mind that the point of the epistemological
whole idea of givenness"-and they are an increasing number-are category of the given is, presumably, to explicate the idea that empiri-
really only attacking sense data. For they transfer to other items, say cal knowledge rests on a 'foundation' of non-inferential knowledge of
physical objects or relations of appearing, the characteristic features of matter of fact, we may well experience a feeling of surprise on noting
the "given." If, however, I begin my argument with an attack on sense that according to sense-datum theorists, it is particulars that are sensed.
datum theories, it is only as a first step in a general critique of the For what is known, even in non-inferential knowledge, is facts rather
entire framework of givenness. than particulars, items of the form something's being tlws-and-so or
2. Sense-datum theories characteristically distinguish between an act somct:hing's standing in a certain relation to something else. It would
of awareness and, for example, the color patch which is its object. seem, then, t~t the sensing of sense contents cazmot constih1te knowl-
The act is usually called sensing. Classical exponents of the theory edge, inferential or non-inferential; and if so, we may well ask, what
have often characterized t hese acts as "phenomenologically simple" light does the concept of a sense datum throw on the 'foundations of
and "not further analyzable." But other sense-datum theorists-some empirical knowledge?' The sense-datum theorist, it would seem, must
of them with an equal claim to be considered "classical exponents"- choose between saying:
have held that sensing is analyzable. And if some philosophers seem (a) It is particulars which are sensed. Sensing is not knowing. The
to have thought that if sensing is analyzable, then it can't be an act, existence of sense-data does not logically imply the existence of
this has by no means been the general opinion. There are, indeed, knowledge.
deeper roots for the doubt that sensing (if there is such a thing) is or
an act, roots which can be traced to one of two Jines of thought tangled (b) Sensing is a form of knowing. It is facts rather than particulars
together in classical sense-datum theory. For the moment, however, which are sensed.
I shall simply assume that however complex (or simple) the fact that On alternative (a) the fact that a sense content was sensed would be
x is sensed may be, it has the form, whatever exactly it may be, by a non-epistemic fact about the sense content. Yet it would be hasty
virtue of which for x to be sensed is for it to be the object of an act. to conclude that this alternative precludes any logical connection be-
Being a sense datum, or sensum, is a relational property of the hveen the sensing of sense contents and the possession of non-inferential
item that is sensed. To refer to an item which is sensed in a way knowledge. For even if the sensing of sense contents did not logically
which does not entail that it is sensed, it is necessary to use some imply the existence of non-inferential knowledge, the converse might
other locution. Sensibile has the disadvantage that it implies that well be true. Thus, the non-inferential knowledge of particular matter
sensed items could exist without being sensed, and this is a matter of of fact might logically imply the existence of sense data (for example,
controversy among sense-datum theorists. Sense content is, perhaps, as seeing that a certain physical object is red might logically imply sens-
neutral a term as any. ing a red sense content) even though the sensing of a red sense con-
T h ere appear to be varieties of sensing, referred to by some as visual tent were not itself a cognitive fact and did not imply the possession
sensing, tactual sensing, etc., and by others as directly seeing, directly of non-inferential knowledge.
hearing, etc. But it is not clear whether these are species of sensing On the second alternative, (b), the sensing of sense contents would
in any full-blooded sense, or whether "x is visually sensed" amounts logically imply the existence of non-inferential knowledge for the
to no more than "x is a color patch which is sensed," "x is directly simple reason thnt it would he this knowledge. But, once again, it
heard" than "x is a sound which is sensed" and so on. In the latter would be facts rather t:hnn part·iculars wliicli arc sensed.
case, being a visual sensing or a direct hearing would be a relational 4. Now it might· seem that when <·011fro11t cil hy this choice, the
property of :m act of sensing, just as being a sense datum is a relational sense-datum theorist seeks lo linvc his ('11ke i111d l'nt it. For lie cltarnc-
property of a sense content. lcrist·ically insists hollt tlrnt· sc11.~i 11 g is 11 k11owi111-: 11m/ that it is particn-

254
 Wilfrid Sellars EMPIRICISl\·f AND THE P HILOSOPHY OF MIND

lars which are sensed. Yet his position is by no means as hopeless as work, and thus ·sever the logical connection between sense data and
this formulation suggests. For the 'having' and the 'eating' can be non-inferential knowledge to which the classical form of the theory is
combined without logical nonsense provided that he uses the word committed. This brings us face to face with the fact that in spite of
know and, correspondingly, the word given in two senses. He must the above considerations, many if not most sense-datum theorists have
say something like t11e following: thought of the givenness of sense contents as the basic notion of the
sense-datum framework. What, then, of the logical connection in the
The non-inferential knowing on which our world picture rests is the
knowing that certain items, e.g. red sense contents, arc of a certain direction seming sense contents~ l1aving non-inferential knowledge?
character, e.g. red. Wben such a fact is non-inferentially known about Clearly it is severed by those wbo think of sensing as a unique and
a sense content, I will say that the sense content is sensed as being, e.g., unanalyzable act. Those, on the other hand, who conceive of sensing
red. I will then say that a sense content is sensed (full stop) if it is as an analyzable fact, while they have prima facie severed this connec-
sensed as being of a certain character, e.g. red. Finally, I will say of tion (by taking the sensing of sense contents to be the basic concept
a sense content that it is known if it is sensed (full stop), to empha-
size that sensing is a cognitive or epistemic fact . of the sense-datum framework) will nevertheless, in a sense, have main-
tained it, if the result they get by analyzing x is a red sense datum
Notice that, given these stipulations, it is logically necessary that turns out to be the same as the result they get when they analyze
if a sense content be sensed, it be sensed as being of a certain character, x is non-inferentiaJJy known to be red. The entailment which was
and that if it be sensed as being of a certain character, the fact that thrown out the front door would have sneaked in by the back.
it is of this cliaracter be non-inferentially known. Notice also that the It is interesting to note, in this connection, that those who, in the
being sensed of a sense content would be knowledge only in a stipu- classical period of sense-datum theories, say from Moore's "Refutation
lated sense of know. To say of a sense content-a color patch, for of Idealism" until about 1938, analyzed or sketched an analysis of
ex.ample-that it was 'known' would be to say that some fact about it sensing, did so in non-epistemic terms. T ypica1Iy it was held that for
was non-inferentially known, e.g. that it was red. This stipulated use a sense content to be sensed is for it to be an element in a certain
of know would, h owever, receive aid and comfort from the fact that kind of relational array of sense contents, where the relations which
t here is, in ordinary usage, a sense of know in which it is followed constitute the array are such relations as spatiotemporal juxtaposition
by a noun or descriptive phrase which refers to a particular, thus (or overlapping), constant conjunction, mnemic causation-even real
Do you know John? connection and belonging to a self. There is, however, one class of
Do you know the President? terms which is conspicuous by its absence, namely cognitive terms.
Because these questions are equivalent to "Are you acquainted with For these, like the 'sensing' which was under analysis, were taken to
John?'' and "Are you acquainted with the President?" the pl\rase belong to a higher level of complexity.
"knowledge by acquaintance" recommends itself as a useful metaphor Now the idea that epistemic facts can be analyzed without re-
for this stipulated sense of know and, like other useful metaphors, has mainder-even "in principle"-into non-epistemic facts, whether phe-
congealed into a technical term. nomenal or behavioral, public or private, with no mat ter how lavish
5. We have seen that the fact that a sense content is a datum (if, a sprinkling of subjunctives and hypotheticals is, I believe, a radical
indeed, there are such facts) will logically imply that someone has non- mistake-a mistake of a piece with the so-called "naturalistic foll::icy"
inferential knowledge only if to say that a sense content is given is in ethics. I shall not, however, press this point for the moment» though
contextually defined in terms of non-inferential knowledge of a fact it will be a centra l theme in a later stage of my argument. What· I do
about this sense content. If this is not clearly realized or held in mind, want to stress is that whether classical sense-datum philosophers have
sense-datum theorists may come to think of the givenness of sense conceived of the givenness of sense contents as :111:1lyzal>lc i11 11011-
co11lcnl·s as the basic or primitive concept of the scnsc-dattm1 frame- cpistemic terms, or as conshtut·cd by acts which arc so111clinw both

256 257
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

irreducible and knowings, they have without exception taken them 1) He can abandon A, in which case the sensing of sense contents
to be fundamental in another sense. becomes a noncognitive fact-a noncognitive fact, to be sure which
6. For they have taken givenness to be a fact which presupposes no may be a necessary condition, even a logically necessary condition, of
learning, no forming of associations, no setting up of stimulus-response non-inferential knowledge, but a fact, nevertheless, which cannot
connections. In short, they have tended to equate sensing sense contents constitute this knowledge.
with being conscious, as a person who has been hit on the head is not 2) He can abandon B, in which case he must pay the price of
conscious whereas a new born babe, alive and kicking, is conscious. cutting off the concept of a sense datum from its connection with
They would admit, of course, that the ability to know that a person, our ordinary talk about sensations, feelings, afterimages, tickles and
namely oneself, is now, at a certain time, feeling a pain, is acquired itches, etc., which are usually thought by sense-datum theorists to
and does presuppose a (complicated) process of concept formation. be its common sense counterparts.
But, they would insist, to suppose that the simple ability to feel a pain 3) But to abandon C is to do violence to th e predominantly
or see a color, in short, to sense sense contents, is acquired and involves nominalistic proclivities of the empiricist tradition.
a process of concept formation, would be very odd indeed. 7. It certainly begins to look as though the classical concept of a sense
But if a sense-datum philosopher takes the ability to sense sense datum were a mongrel resulting from a crossbreeding of two ideas:
contents to be unacquired, he is clearly precluded from offering an (1) TI1e idea that there are cert.ain inner episodes-e.g. sensations
analysis of x senses a sense content which presupposes acquired abilities. of red or of C# which can occur to human beings (and brutes) with-
It follows that he could analyze x senses red sense content s as x non- out any prior process of learning or concept formation; and without
inferentia11y knows tJ1at s is red only if 11e is prepared to admit that which it would in some sense be impossible to sec, for example, that
the ability to have such non-inferential knowledge as that, for example, the facing surface of a physical object is red and triangular, or hear
a red sense content is red, is itself unacquired. And this brings us face that a cert.ain physical sound is C#.
to face with the fact that most empirically minded philosophers are (2) The idea that there are certain inner episodes which are the
strongly inclined to think t hat all classificatory consciousness, all knowl- non-inferential knowings that certain items arc, for example, red
edge that somet11ing is thus-and-so, or, in logicians' jargon, all subsum p- or Cl; and that these episodes are the necessary conditions of em-
tion of particulars under universals, involves learning, concept forma- pirical knowledge as providing the evidence for all oth er empirical
tion, even the use of symbols. It is clear from the above analysis, there- propositions.
fore, that classical sense-datum theories-I emphasize the adjective, for And I think that once we are on the lookout for them, it is quite
there are other, 'heterodox,' sense-dat~m theories to be taken into ac- easy to sec how these two ideas came to be blended together in
count- are confronted by an inconsistent triad made up of the follow- traditional epistemology. TI1e first idea clearly arises in the attempt
ing three propositions: ' to explain the facts of sense perception in scientific style. How does
A. X senses red sense contents entails x non-inferentially knows that it happen that people can have the experience which th ey describe
sis red. by saying "It is as though I were seeing a red and triangular physical
B. 1be ability to sense sense contents is unacquired. object" when either there is no physical object there at all, or, if there
C. The ability to know facts of the form x is </> is acquired. is, it is neither red nor triangular? The explanation, roughly, posits that
A and B together entail not-C; B and C entail not-A; A and C entail in every case in which a person has an experience of this kind, whdher
not-B. veridical or nol:, he has what is called a 'seusat·iou' or 'impression' 'of
Once the c1assical sense-datum theorist faces up to the fact that a red triangle.' The core idea is 1'11at· t·he proximat e < "llll.~C of such a
A, 13, and C do form an inconsistent triad, which of them will he sensation is 011/y for tl1c: 111osl part l>ronght 11 1>0111 by I he pre.~c11ce in
choose to abandon? the neighborhood of the perceiver of 11 red 1111d hi11 11g11lar physical

258 259
 Wilfrid Sellazs EMPIRICISM AND THE PHILOSOPHY OF MIND

object; and that while a baby, say, can have the 'sensation of a red which are veridical, and from which the non-veridical members cannot
triangle' without either seeing or seeming to see that the facing side be weeded out by 'inspection,' this foundation cannot consist of such
items as seeing that the facing surface of a physical object is red and
of a physical ob;ect is red and triangular, there usually looks, to adults, triangular.
to be a physical object with a red and triangular facing surface, when Thus baldly put, scarcely anyone would accept this conclusion. Rather
they are caused to have a 'sensation of a red triangle'; while without they would take the contrapositive of the argument, and reason that
such a sensation, no such experience can be had. since the foundation of empirical knowledge is the non-inferential
I shall have a great deal more to say about this kind of 'explanation' knowledge of such facts, it does consist of members of a class which
of perceptual situations in the course of my argument. What I want to contains non-veridical members. But before it is thus baldly put, it
emphasize for the moment, however, is that, as far as tlte above formu- gets tangled up with the first line of thought. The idea springs to
lation goes, there is no reason to suppose that having the sensation of mind that sensations ot red triangles have exactly the virtues which
a red triangle is a cognitive or epistemic fact. There is, of course, a ostensible seeings ot red triangular physical surfaces lack. To begin with,
temptation to assimilate "having a sensation of a red triangle" to the grammatical similarity of 'sensation of a red triangle' to "thought
" thinking of a celest.ial city" and to attribute to the former the of a celestial city" is interpreted to mean, or, better, gives rise to the
epistemic character, the 'intentionality' of the latter. But this tempta· presupposition, that sensations belong in the same general pigeonhole
tion could be resisted, and it could be held that having a sensation as t110ughts-in short, are cognitive facts. Then, it is noticed that
of a red triangle is a fact sui generis, neither epistemic nor physical, sensations are ex hypothesi far more intimately related to mental proc-
having its own logical grammar. Unfortunately, the idea that there esses than external physical o.bjects. It would seem easier to "get at"
are such things as sensations of red triangles-in itself, as we shall see, a red triangle of which we are having a sensation, than to "get at" a
quite legitimate, though not without its pllzzles-seems to fit the red and triangular physical surface. But, above all, it is the fact that
requirements of another, and less fortunate, line of thought so well it doesn't make sense to speak of unveridical sensations .which strikes
that it has almost invariably been distorted 'to give the latter a rein- these philosophers, though for it to strike them as it does, they must
forcement without which it would long ago have collapsed. This un- overlook the fact that if it makes sense to speak of an experience as
fortunate, but familiar, line of thought runs as follows: veridical it must correspondingly make sense to speak of it as unveridi-
T he seeing that the facing surface of a physical object is red and ~- Let me emphasize that not all sense-datum theorists-even of the
triangular is a veridical member of a class of e>--periences-let us call classical type-have been guilty of all these confusions; nor are these all
them 'ostensible seeings' - some of the members of which are non- the confusions of which sense-datum theorists have been guilty. I shall
veridical; and there is no inspectible hallmark which guarantees that
have more to say on this topic later. But the confusions I have men-
any such experience is veridical. To suppose that the non-inferential
knowledge on which our world picture rests consists of such ostensible tioned are central to the tradition, and will serve my present purpose.
seeings, hearings, etc., as happen to be veridical is to place empirical For the upshot of blending all these ingredients together is the idea
knowledge on too precarious a footing-indeed, to open the door to that a sensation of a red triangle is the very paradigm of empirical
skepticism by making a mockery of the word knowledge in the phrase knowledge. And I think that it can readily be seen that this idea leads
"empirical knowledge."
straight to the orthodox type of sense-datum theory and accounts for
Now it is, of course, possible to delimit subclasses of ostensible see-
ings, hearings, etc., which are progressively less precarious, i.e. more the perplexities which arise when one tries to think it through.
reliable, by specifying the circumstances in which they occur, and the
vigilance of the perceiver. But the possibility that any given ostensible II. Another Language?
seeing, hearing, etc., is non-veridical can never be entirely eliminated. 8. I shall now examine briefly a heterodox suggestion hy, for example,
T herefore, given that the foundation of empirical knowledge cannot
c:ousist of the veridical members of a class not all the memhcrs of Ayer ( l) (2) to the effect that discourse about sense data is, so lo speak,

260 261
 Wilfrid Sellars EMPIRICISM AND TlIE P HILOSOPHY OF MIND

another language, a language contrived by the epistemologist, for situa- of which they are parts. For example, the code symbol for "Someone
tions which the plain man describes by means of such locutions as "Now on board is sick" might contain the letter S to remind us of the word
the book looks green to me" and "There seems to be a red and "sick," and, perhaps, the reversed letter E to remind those of us who
triangular object over there." 111e core of this suggestion is the idea have a background in logic of the word "someone." Thus, the flag for
that the vocabulary of sense data embodies no increase in the content "Someone on board is sick" might be '3S.' Now the suggestion at
of descriptive discourse, as over and against the plain man's language which I am obviously driving is that someone might introduce so-called
of physical objects in Space and Time, and the properties they have and sense-datum sentences as code symbols or ''flags," and introduce th e
appear to have. For it holds that sentences of the form vocables and printables they contain to serve the role of reminding
X presents S with a cf> sense datum us of certain features of the sentences in ordinary perceptual discourse
are simply stipulated to have the same force as sentences of the form which the flags as wholes represent. In particular, the role of the
X looks cf> to S. vocable or printable "sense datum" would be that of indicating that
Thus "TI1e tomato presents S with a bulgy red sense-datwn" ·would be the symbolized sentence contains the context " . . . looks . ..," the
the contrived counterpart of "The tomato looks red and bulgy to S" vocable or printable "red" that the correlated sentence contains the
and would mean exactly what the latter means for the simple reason context " . . . looks red . . ." and so on.
that it was stipulated to do so. 9. Now to take this conception of sense datum 'sentences' seriously
As an aid to explicating this suggestion, I am going to make use of is, of course, to take seriously the idea that there are no independent
a certain picture. I am going to start with the idea of a code, and I logical relations between sense-datum 'sentences.' It looks as though
am going to enrich this notion until the codes I am talking about are there were such independent logical relations, for these 'sentences' look
no longer mere codes. Whether one wants to call these "enriched codes" like sentences, and they have as proper parts vocables or printables which
codes at all is a matter which I sha11 not attempt to decide. function in ordinary usage as logical words. Certainly if sense-datum
Now a code, in the sense in which I shall use the term, is a system talk is a code, it is a code which is easily mistaken for a language proper.
of symbols each of which represents a complete sentence. Thus, as Let me illustrate. At first sight it certainly seems that
we initially view the situation, there are two characteristic features of A. The tomato presents S with a red sense datum
a code: (1) Each code symbol is a unit; the parts of a code symbol entails both
are not themselves code symbols. (2) Such logical relations as obtain B. There are red sense data
among code symbols are completely parasitical; they derive entirely and
from logical relations among t11e sentences they represent. Indeed, to C. The tomato presents S with a sense datum which has some specific
speak about logical relations among code symbols is a way of tal}dng shade of red.
which is introduced in terms of the logical relations among t he sen- This, however, on the kind of view I am considering, would be a
tences they represent. Thus, if "O" stands for "Everybody on board mistake. (B) would follow-even in the inverted commas sense of
i.<> sick" and "6" for "Somebody on board is sick," then "6" would 'follows' appropriate to code symbols-from (A) only because (B) is
follow from "O" in the sense that the sentence represented by "6" the flag for (P ), "Someth ing looks red to somebody," which doc.~ follow
follows from the sentence represented by "O". from {a), "111c tomato looks red to Jones" which is rcprcscntccl in
Let me begin to modify this austere conception of a code. 111ere is the code by (A). And (C) wonlcl 'follow' from (A), i11 spilc of appear-
no reason why a code symbol might not have parts which, without ances, only if {C) were tl1c flag for :1 sc111·c11cc which follows from (").
becoming full-fledged symbols on their own, do play a role in the I shall l1ave more to say aho11t lliis cx:1111plc i11 a 1110m t·11 I. 'l'lic point
system . Thus they might play the role of mnemonic devices serving to to he stressed now is that to rnrry 0111 !h is vit·w t'o11~i.~ l c11 lly ouc must
pul 11s in mind of features of the sentences represented hy the symbols clcny l'o such vocables 1111<1 pri 11lnlil c.~ us "tpmlit y," "is," " red," "color,''
262 261
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

"crimson," "determinable," "determinate," "all," "some," "exists," etc., that is, to treat sense-datum flags as though they were sentences in a
etc., as they occur in sense.<fatum talk, the full.blooded status of their theory, and sense.datum talk as a language which gets its use by coordi-
counterparts in ordinary usage. They are rather clues which serve to nating sense-datum sentences with sentences in ordinary perception talk,
remind us which sense-datum 'flag' it would be proper to By along as molecule talk gets its use by coordinating sentences about popula-
with which other sense.<fatum 'flags.' T hus, the vocables which make tions of molecules with talk about the pressure of gases on tI1e walls
up the two 'flags' of their containers. After all,
(D) All sense·data are red x looks red to S - == · there is a class of red sense data which
and belong to x, and are sensed by S
(E) Some sense data are not red has at least a superficial resemblance to
remind us of the genuine logical incompatibility between, for example, g exerts pressure on w • == • there is a class of molecules which
(F) All elephants are grey make up g, and which are bouncing
and off w,
(G) Some elephants are not grey, a resemblance which becomes even more striking once it is granted
and serve, therefore, as a clue to the impropriety of flying these two that the former is not an analysis of x looks red to S in terms of sense
'flags' together. For the sentences they symbolize are, presumably, data.
{S) Everything looks red to everybody There is, therefore, .reason to believe that it is the fact that both
and codes and theories are contrived systems which are under the control
(t) There is a cofor other than red which something looks to some- of 'the language with which they are coordinated, which has given aid
body to have, and comfort to the idea that sense-datum talk is "another language"
and these are incompatible. for ordinary discourse about perception. Yet although the logical re·
But one would have to be cautious in using these clues. Thus, from lations between sentences in a theoretical language are, in an important
the fact tl1at it is proper to infer sense, under the control of logical relations between sentences in the
(H) Some elephants have a determinate shade of pink observation language, nevertheless, within the framework of this con-
from trol, the theoretical language has an autonomy which contradicts the
(I) Some elephants are pink very idea of a code. If this essential difference between theories and
it would clearly be a mistake to infer that the right to fly codes is overlooked, one may be tempted to try to eat his cake and
(K) Some sense data are pink have it. By tllinking of sense-datum talk as merely anot.h er language, one
carries with it the right to fly draws on the fa ct that codes have no surplus value. By thinking of
(L) Some sense data bave a determinate shade of pink. sense·datum talk as illuminating the "language of appear'ing," one draws
9. But if sense-datum sentences are really sense-datum 'sentences'-i.e. on the fact that theoretical languages, though contrived, and depend-
code flags-it follows, of course, that sense-datum talk neither ciaxifies ing for their meaningfulness on a coordination with the language of
nor explains facts of the form x looks ef> to S or x is <f>. That it \Vould observation, have an explanatory function. Unfortunately, these two
appear to do so would be because it would take an almost superhuman characteristics are incompatible; for it is just because theories have
effort to keep from taking the vocables and · printables which occur in "surplus value" t11at they can provide explanations.
the code (and let me now add to our earlier list the vocable "directly No one, of course, who thinks-as, for example, docs Ayer- of the
known") to be words which, if homonyms of words in ordinary usage, existence of sense clata as entailing the existence of "clired knowledge,''
have their ordinary sense, and which, if invented, have a meaning speci- would wish to say lha t sense dat;r arc theoretica l enti ties. H could
fied hy their relation to the others. One would be constantly tempted, scarcely he a theoretical fact tha t I nm directly k11owi11g lhnt a certain
264 26S
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

sense content is red. On the other hand, the idea that sense contents culties which become apparent once the role of "looks" or "appears" is
are theoretical entities is not obviously absurd-so absurd as to preclude understood. And it is to an examination of this role that I now turn.
the above interpretation of the plausibility of the "another-language"
approach. For even those who introduce the expression "sense con- III. The Logic of 'Looks'
tent" by means of the context " . . . is directly known to be ..." may l 0. Before turning aside to examine the suggestion that the language
fail to keep this fact in mind when putting this expression to use-for of sense data is "another language" for the situations described by the
example, by developing the idea that physical objects and persons alike so-called '1anguage of appearing," I had concluded that classical sense.
are patterns of sense contents. In such a specific context, it is possible datum theories, when pressed, reveal themselves to be the result of
to forget that sense contents, thus introduced, are essentially sense data a mismating of two ideas: (I ) TI1e idea that there arc certain "inner
and not merely items which exemplify sense qualities. Indeed, one may episodes," e.g. the sensation of a red triangle or of a C# sound, which
even lapse into thinking of the sensing of sense contents, the givenness occur to human beings and brutes without any prior process of learn-
of sense data, as non-epistemic facts. ing or concept formation, and without which it would-in some scnsc-
I think it fair to say that those who offer the "another-language" in- be impossible to see, for example, that the facing surface of a physical
terpretation of sense data find the illumination it provides to consist object is red and triangular, or hear that a certain physical sound is C#;
primarily in the fact that in the language of sense data, physical objects (2) The idea that there are certain "inner episodes" which arc the
are patterns of sense contents, so that, viewed in this framework, there non-inferential knowings that, for example, a certain item is red and
is no "iron curtain" between the knowing mind and the physical world. triangular, or, in tlie case of sounds, c,, which inner episodes are the
It is to elaborating plausible (if schematic) translations of physical- necessary conditions of empirical knowledge as providing the evidence
object statements into statements about sense contents, rather than to for all other empirical propositions. If this diagnosis is correct, a reason-
spelling out the force of such sentences as "Sense content s is directly able next step would be to examine these two ideas and determine how
known to be red," that tbe greater part of their philosophical ingenuity that which survives criticism in each is properly to be combined with
has been directed. the other. Clearly we would have to come to grips with the idea of
However this may be, one thing can be said with confidence. If the inner episodes, for this is common to both.
language of sense data were merely a code, a notational device, then Many who attack the idea of the given seem to have thought that
the cash value of any philosophical clarification it might provide must the central mistake embedded in this idea is exactly the idea that there
lie in its ability to illuminate logical relations within ordinary discourse are inner episodes, whether thoughts or so-called "immediate experi-
about physical objects and our perception of them. Thus, the fact (if ences," to which each of us has privileged access. I shall argue that this
it were a fact) that a code can be constructed for ordinary perce.Ption is just not so, and that the Myth of the Given can be dispelled with·
talk which 'speaks' of a "relation of identity" between the components out resorting to the crude vcrificationisms or operationalisms characteris-
("sense data") of "minds" and of " things," would presumably have as tic of the more dogmatic forms of recent empiricism. TI1en there are
its cash value the insight that ordinary discourse about physical objects those who, while they do not reject the idea of inner episodes, find the
and perceivers coul<l (in principle) be constructed from sentences of Myth of the Given to consist in the idea that knowledge of these epi-
the form "There looks to be a physical object with a red and triangular sodes furnishes premises on which empirical knowledge rests as on a
facing surface over there" (the counterpart in ordinary language of the foundation. But while this idea has, indeed, been the most wiclesprcacl
basic expressions of the code). In more traditional terms, the clarifica· form of the Myth, it is far from constituting its csscuc:e. t•:veryt·hing
tion would consist in making manifest the fact that persons and things hinges on wJ1y these philosophers reject it. If, for cxnmple, ii is 011 the
arc alike logical constructions out of lookings or appcarings (not a ppcar- ground that the lcamiug of a language is :i puhlic prnC'c:ss whi('h proceeds
ancc.~I). Bnt :my claim to this effect soon rnns into insuperable diffi· in a domain of p11hlic: objects and is governed hy p11hlic s:111t l ions, so that
266 267
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

private episodes-with the exception of a mysterious nod in their direc- 12. I shall begin my examination of "X looks red to S at t" with
tion-must needs escape the net of rational discourse, then, while these the simple but fundamental point that the sense of "red" in which
philosophers are immune to the fom1 of the myth which has Bowered things look red is, on the face of it, the same as that in which things
in sense-datum theories, they have no defense against the myth in the arc red. \ Vhen one glimpses an object and decides that it looks red (to
form of the givenness of such facts as that physical object x looks red me, now, from here) and wonders whether it really is red, one is surely
to person S at time t, or that there looks to person S at time t to be a wondering whether the color-red-which it looks to have is the one
red physical object over there. It will be useful to pursue the Myth in it really does have. This point can be obscured by such verbal manipu-
this direction for a while before more general issues are raised. lations as hyphenating the words "looks" and "red" and claiming that
11. Philosophers have found it easy to suppose that such a sentence it is the insoluble unity ''looks-red" and not just "looks" which is the
::is ."The tomato looks red to Jones" says that a certain triadic relation, relation. Insofar as this dodge is based on insight, it is insight into the
looking or appearing, obtains among a physical object, a person, aµd a fact that looks is not a relation between a person, a thing, and a
quality.• "A looks q, to S" is assimilated to "x gives y to z"-or, better, quality. Unfortunately, as we shall see, the reason for this fact is one
since giving is, strictly speaking, an actjon rather than a relation-to "x which gives no comfort at all to the idea that it is looks-red rather than
is between y and z," and taken to be a case of the general form looks which is the relation.
"R (x,y,z) ." Having supposed this, they tum without further ado to I have, in effect, been claiming that being red is logically prior, is a
the question, "Is this relation analyzable?" Sense-datum theorists have., logically simpler notion, than looking red; the function "x is red" to
on the whole, answered "Yes," and claimed that facts of the form "x looks red to y." In short, that it just won't do to say that x is red
x looks red to X are to be analyzed in terms of sense data. Some of is analyzable in terms of x looks red to y. But what, then, are we to
them, without necessarily rejecting this claim, have argued that facts make of the necessary truth-and it is, of course, a necessary truth-that
of this kind are, at the very least, to be explained in terms of sense x is red . = · x wo.uld look red to standard observers in standard
data. Thus, when Broad ( 4) writes "If, in fact, nothing elliptical is conditions?
before. my mind, it is very hard to understand why the penny sh ould 111ere is certainly some sense to the idea that this is at least the schema
seem elliptical rather than of any other shape (p. 240)," he is appeal- for a definition of physical redness in terms of looking red. One begins
ing to sense-data as a means of explaining facts of this form. The dif- to see the plausibility of the gambit that looking-red is an insoluble
ference, of course, is that whereas if x looks </> to S is correctly analyzed unity, for the minute one gives "red" (on the right-hand side) an inde-
in terms of sense data, then no one could believe that x looks 4> to S pendent status, it becomes what it obviously is, namely "red" as a
witho11t believing that S has sense data, the same need not be true if predicate of physical objects, and the supposed definition becomes an
x looks q, to S is explained in terms of sense data, for, in the case of obvious circle.
some types of explanation, at least, one can believe a fact without be- 13. The \vay out of this troubling situation has two parts. The
lieving its explanation. second is to show how "x is red" can be necessarily equivalent to "x
On the other hand, those philosophers who reject sense-datum theo· would look red to standard observers in standard situations" without
ries in favor of so-called theories of appearing have characteristically held this being a definition of "x is red" in terms of "x looks red." But the
that facts of the form x looks q, to S are ultimate and irreducible, and lirst, and logically prior, step is to show that "x looks red to S" does
that sense data are needed neither for their analysis nor for their not assert either an unanalyzable triadic relation to obtain between x,
explanation. If asked, "Doesn't the statement 'x looks red to S' have as red, and S, or an unanalyzable dyadic relation to obtain between x
part of its meaning the idea that s stands in some relation to something and S. Not, however, because it asserts an analyzable relation to obtain,
that is red?" their answer is in the negative, and, I believe, rightly so. but because looks is not a relation at all. Or, to put the matter in a
• A useful discussion of views of this type is to be found in (9) nnd ( I 3). familiar way, one can say that looks is a relation if he likes, for the

268 269
 Wilfrid Sellars EMPIRICISM AND TUE PHILOSOPHY OF MIND

sentences in which this word appears show some grammatical analogies "I don't know what to say. If I didn't know that the tie is blue-
to sentences built around words which we should not hesitate to classify and the alternative to granting this is odd indeed-I would swear that
as relation words; but once one has become aware of certain other I was seeing a green tie and seeing that it is green. It is as though I
features which make them very unlike ordinary relation sentences, he were seeing the necktie to be green."
will be less inclined to view his task as that of finding tl1e answer to If we bear in mind that such sente~ces as "This is green" have both
t he question "Is looks a relation?" a fact-stating and a reporting use, we can put the point I have just
14. To bring out the essential features of the use of "looks," I shall been making by saying that once John learns to stifle the report "lbis
engage in a little historical fiction. A young man, whom I shall call necktie is green" when looking at it in the shop, there is no oth er report
John, works in a necktie shop. He has learned the use of color words about color and the necktie which h e knows how to make. To be sure,
in the usual way, with this exception. I shall suppose that he has never he now says "This necktie is blue." But he is not making a reporting
looked at an object in other than standard conditions. As he examines use of this sentence. He uses it as the conclusion of an inference.
his stock every evening before closing up shop, he says "This is red," 15. We return to the shop after an interval, and we find that when
"That is green," "This is purple," etc., and such of his linguistic peers John is asked "What is the color of this necktie?" he makes such state-
as happen to be present nod their heads approvingly. · ments as "It looks green, but take it outside and see." It occurs to us
Let us suppose, now, that at this point in the story, electric lighting that perhaps in learning to say "This tie looks green" when in the shop,
is invented. His friends and neighbors rapidly adopt this new means he has learned to make a new kind of report. Thus, it might seem as
of illumination, and wrestle with the problems it presents. John, how· though his linguistic peers have helped him to notice a new kind of
ever, is the last to succumb. Just after it has been installed in his shop, objective fact, one which, though a relational fact involving a per-
one of his neighbors, Jim, comes in to buy a necktie. ceiver, is as logically independent of the beliefs, the conceptual frame-
"Here is a handsome green one," says John. work of the perceiver, as the fact that the necktie is blue; but a minimal
" But it isn't green," says Jim, and takes John outside. fact, one which it is safer to report because one is less likely to be
" Well," says John, "it was green in there, but now it is blue." mistaken. Such a minimal fact would be the fact that the necktie looks
"No," says Jim, "you know that neckties don' t change their color green to John on a certaiµ occasion, and it would be properly reported
merely as a result of being taken from place to place." by using the sentence ''This necktie looks green." It is this type of
"But perhaps electricity changes t heir color and they change back account, of course, which I have already rejected.
again in daylight?" But what is the alternative? If, that is, we are not going to adopt the
"TI1at would be a queer kind of change, wouldn't it?" says Jim. sense-datum analysis. Let me begin by noting that there certainly seems
"I suppose so," says bewildered John. "But we saw that it was green to be something to the idea that the sentence "This looks green to me
in there." now" has a reporting role. Indeed, it would seem to be essentially a
"No, we didn't see that it was green in tbere, because it wasn't green, report. But if so, what does it report, if not a minimal objective fact,
and you can't see what isn't so!" and if what it reports is not to be analyzed in terms of sense data?
" W ell, this is a pretty pickle," says John. "I just don' t know what 16. Let me next calJ attention to the fact that the experience of
to say." having something look green to one at a certain time is, insofar as it
The next time John picks up this tie in his shop and someone asks is an experience, obviously very much like that of seeing something to
what color it is, his first impulse is to say "It is green." He suppresses he green, insofar as the latter is an experience. But the latter, of conrsc,
this impulse, and, remembering what happened before, comes out with is not just :in experience. And this is the heart of the mnttcr. For to
" It is blue." fie doesn't see that it is blue, nor would he say that he say that a certain experience is a seeing tl1:1t someth ing is Ilic t :isc, is
secs it to he blue. What docs he sec? Let us ask him. to clo more th:m describe Ilic experience. I t is to cliaraclcrizc it as, so
270 27 1
 Wilfrid Sellars EMPIRICISM AND THE PmLOSOPHY OF M IND

to speak, making an assertion or claim, and-which is the point I wish the other merely as a case of something's looking green. Of course, if
to stress-to endorse that claim. As a matter of fact, as we shall see, I say "X merely looks green to S" I am not only failing to endorse
it is much more easy to see that the statement "Jones sees that the the claim, I am rejecting it.
tree is green" ascribes a propositional claim to Jones' experience and en- Thus, when I say "X looks green to me now" I am reporting the fact
dorses it, than to specify how the statement describes Jones' experience. that my experience is, so to speak, intrinsically, as an experience, in-
I realize that by speaking of experiences as containing propositional distinguishable from a veridical one of seeing that x is green. Involved
cJaims, I may seem to be knocking at closed doors. I ask the reader in the report is the ascription to my experience of the claim 'x is green';
to bear with me, however, as the justification of this way of talking is and the fact that I make this report rather than the simple report "X is
one of my major aims. If I am permitted to issue this verbal currency green" indicates that certain considerations have operated to raise, so
now, I hope to put it on the gold standard before concJuding the to speak in a higher court, the question 'to endorse or not to endorse.'
argument. I may have reason to think that x may not after an be green.
16. It is dear that the experience of seeing that something is green If I make at one time the report "X looks to be green"-which is
is not merely the occurrence of the propositional claim 'this is green'- not only a report, but the withholding of an endorsement-I may later,
not even if we add, as we must, that this claim is, so to speak, evoked when the original reasons for withholding endorsement have been
or wrung from the perceiver by the object perceived. Here Nature- rebutted, endorse the original claim by saying "I saw that it was green,
to turn Kant's simile (which he uses in another context) on its though at the time I was only sure that it looked green." Notice that
head-puts us to the question. The something more is clearly what I will only say "I see that x is green" (as opposed to "X is green")
philosophers have in mind when they speak of "visual impressions' ~ or when the question "to endorse or not to endorse" has come up. "I see
"immediate visual experiences." What exactly is the logical status of that x is green" belongs, so to speak, on the same level as "X looks
these "impressions" or "immediate experiences" is a problem which wiJJ green" and "X merely looks green."
be with us for the remainder of this argument. For the moment it is 17. There are many interesting and subtle questions about the dialec·
the propositional claim which concerns us. tics of "looks talk," into which I do not have the space to enter.
I pointed out above that when we use the word "see" as in "S sees Fortunately, the above distinctions suffice for our present purposes. Let
that the tree is green" we are not only ascribing a claim to the experi- us suppose, then, that to say that " X looks green to S at t" is, in
ence, but endorsing it. It is this endorsement which Ryle has in mind effect, to say that S has that kind of experience which, if one were
when he refers to seeing that something is thus and so as an achieve- prepared to endorse the propositional claim it involves, one would char-
ment, and to "sees" as an achievement word. I prefer to call it a "so acterize as seeing x to be green at t. Thus, when our friend John learns
it is" or "just so" word, for the root idea is that of truth. To char,,ac- to use the senten ce "This necktie looks green to me" he learns a way
terize S's experience as a seeing is, in a suitably broad sense-which I of reporting an experience of the kind which, as far as any categories
shall be concerned to explicate-to apply the semantical concept of I have yet permitted him to have are concerned, he can only charac-
truth to that experience. terize by saying that as an experience it does not differ from seeing
Now the suggestion I wish to make is, in its simplest terms, that something to be green, and that evidence for the proposition 'This
the statement "X looks green to Jones" differs from "Jones sees that necktie is green' is ipso facto evidence for the proposition that the
x is green" in that whereas the latter both ascribes a propositional claim experience in question is seeing tliat the necktie is grccu.
to Jones' e>..1JCrience and endorses it, the former ascribes the claim but Now one of the chief merits of this account is that it permits a parallel
does not endorse it. This is the essential difference between the two, treatment of 'c1ualitative' and 'existential' seeming or looking. Thus,
for it is clear that two experiences may be identical as experiences, and when J say "1be tree looks hent" I am cnclorsing that part of the claim
yet one be properly referred to as a seeing tJ1at something is green, and involved in my experience which concerns the cxistc1H·e of the I rec, but
272 271
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF M IND
withholding endorsement from the rest. On the other hand, wh en I is a necessary truth not because the right-hand side is the definition
say "Th ere looks to be a bent tree over there" I am refusing to endorse of "x is red," but because "standard conditions" means conditions in
any but the most general aspect of the claim, namely, that there is an which things look what they are. And, of course, which conditions are
'over there' as opposed to a 'here.' Another merit of the account is that standard for a given mode of perception is, at the common-sense level,
it explains how a necktie, for example, can look red to S at t, without specified by a list of conditions which exhibit the vagueness and open
looking scarlet or crimson or any other determinate shade of red. In texture characteristic of ordinary discol!rse.
short it explains how things can have a merely generic look, a fact which 19. I have arrived at a stage in my argument which is, at least prima
would be puzzling indeed if looking red were a natural as opposed to facie, out of step with the basic presuppositions of logical atomism.
epistemic fact about objects. The core of the explanation, of course, is 111Us, as long as Iookirig green is taken to be the notion to which
that the propositional claim involved in such an experience may be, being green is reducible, it could be claimed with considerable plausi-
for example, either the more determinable claim 'This is red' or the bility that fundamental concepts pertaining to observable fact have
more determinate claim 'This is crimson.' TI1e complete story is more that logical independence of one another which is characteristic of the
complicated, and requires some account of the role in these experiences empiricist tradition . Indeed, at first sight the situation is quite disquiet-
of the 'impressions' or 'immediate experiences' the logical status of ing, for if the ability to recognize that x looks green presupposes the
which remains to be determined. But even in the absence of these addi- concept of l'eing green, and if this in turn involves knowing in what
tional details, we can note the resemblance between the fact that x can circumstances to view an object to ascertain its color, then, since one
look red to S, without it being true of some specific shade of red that can scarcely determine what the circumstances are without noticing that
x looks to S to be of that shade, and the fact that S can believe that certain objects have certain perceptible characteristics-including col-
Cleopatra's Needle is tall, without its being true of some determinate ors-it would seem that one couldn't form the concept of being green,
number of feet that S b elieves it to be that number of feet tall. and, by parity of reasoning, of the other colors, unless he already had
18. The point I wish to stress at this time, however, is that the con- them.
cept of looking green, the ability to recognize that something looks Now, it just won't do to reply that to have the concept of green, to
green, presupposes the concept of being green, and that the latter con- know what it is for something to be green, it is sufficient to respond,
cept involves the ability to tell what colors objects have by looking at when one is in point ot fact in standard conditions, to green objects
them-which, in turn, involves knowing in what circumstances to place with the vocable "T his is green." Not only must the conditions be of
an object if one wishes to ascertain its color by looking at it. Let me a sort that is appropriate for determining th e color of an object by
develop this latter point. As our friend John becomes more and more looking, the subject must know that conditions of this sort are appro-
sophisticated about his own and other people's visual experiences, he priate. And while this does not imply that one must have concepts
learns under what conditions it is as though one were seeing a necktie before one has them, it does imply that one can have the concept of
to be of one color when in fact it is of another. Suppose someone asks green only by having a whole battery of concepts of which it is one
him "Why does this tie look green to me?" John may very well reply element. It implies that while the process of acquiring the concept of
"Because it is blue, and blue objects look green in this kind of light." green may- indeed does-involve a long history of acquiring piecemeal
And if someone asks this question when looking at the necktie in plain habits of response to various objects in various circumstances, there is
daylight, John may very well reply "Because the tie is green"-to which an important sense in which one has no concept pertaining to the
he may add "We are in plain daylight, and in daylight things look wlwt observable properties of physical objects in Space and Time unless one
tl1ey are." We thus sec that has them :ill- and, inclcccl, as we shall see, a gre.a.t deal more besides.
x is red . = . x looks red to standard observers in slaud:ml con- 20. Now, 1 l'hink it· is dear what a logical atomist, supposing that
ditions he found any merit al :ill in the :ihovc argument, would say. He would
274 27S
 Wilfrid Sellars E MPIRICISM AND THE PHILOSOPHY OF MIND

say that I am overlooking the fact that the logical space of physical ]east as far as I have pushed it to date, revealed any such items as sense-
objects in Space and Time rests on the logical space of sense contents, contents. And it may be relevant to suggest that once we see clearly
and he would argue that it is concepts pertaining to sense contents that physical redness is not to be given a dispositional analysis in terms
which have the logical independence of one another which is character- of looking red, the idea that it is to be given any kind of dispositional
istic of traditional empiricism. "After all," he would point out, "con- analysis loses a large measure of its plausibility. In any event, the next
cepts pertaining to theoretical entities-molecules, for exa mpl~have move must be to press further the above account of qualitative and
the mutual dependence you have, perhaps rightly, ascribed to concepts existential looking.
pertaining to physical fact. But," he would continue, "theoretical con-
IV. Explaining Looks
cepts have empirical content because they rest on-are coordinated
21. I have already noted that sense-datum theorists are impressed
with-a more fundamental logical space. Until you have disposed, there-
by the question "How can a physical object look red to S, unless some-
fore, of the idea that there is a more fundamental logical space than
thing in that situation is red and -S is taking account of it? If S isn't
that of physical objects in Space and T ime, or shown that it too is
experiencing something red, how does it happen that the physical
fraught with coherence, your incipient Meditations Hegeliennes are
premature." object looks red, rather than green or streaky?" There is, I propose to
show, something to this line of thought, though the story turns out to
And we can imagine a sense-datum theorist to interject the follow-
be a complicated one. And if, in the course of telling the story, I shall
ing complaint: "You have begun to write as though you had shown
be led to make statements which resemble some of the things sense-
not only that physical redness is not to be analyzed in terms of looking
datum theorists have said, this story will amount to a sense-datum
red-which I will grant-but also that physical redness is not to be
theory only in a sense which robs this phrase of an entire dimension
analyzed at all, and, in particular, not to be analyzed in terms of the
of its traditional epistemological force, a dimension which is character-
redness of red sense contents. Again, you have begun to write as
istic of even such heterodox forms of sense-datum theory as the "an-
though you had shown not only that observing that x looks red is not
other language" approach.
more basic than observing that x is red, but also that there is no form
Let me begin by formulating the question: " Is the fact that an
of visual noticing more basic than seeing that x is red, such as the
object looks to S to be red and triangular, or that there looks to S to
sensing of a red sense content. I grant," he continues, "that the tend-
be a red and triangular object over there, to be explained in terms of
ency of sense·datum theorists has been to claim that the redness of
the idea that Jones has a sensation-or impression, or immediate experi-
physical objects is to be analyzed in terms of looking red, and then to
claim that looking red is itseJf to be analyzed in tem1s of red sense ence-of a red triangle? One point can be made right away, namely
contents, and that you may have undercut this line of analysis. But that if these expressions are so understood that, say, the immediate
what is to prevent the sense-datum theorist from taking the line that experience of a red triangle implies the existence of something-not a
the properties of physical objects are directly analyzable into the quali- physical object-which is red and triangular, and if the redness which
ties and phenomenal relations of sense contents?" this item has is the same as the redness which the physical object looks
Very well. But once again we must ask, How does the sense-datum to have, then the suggestion runs up against the objection that the red-
theorist come by the framework of sense contents? and How is be ness physical objects look to have is the same as the redness physical
going to convince us that there are such things? For even if looking objects actually do have, so that items which ex hypotl1esi are not
red doesn't enter into the analysis of physical redness, it is by asking physical objects, and which radically, even categorially, differ from
us to reflect on the experience of having something look red to us that physical objects, would have the same redness as physical objects. And
he hopes to make this framework convincing. And it therefore becomes while this is, perhaps, not entirely out of the question, it certainly pro-
relevant to note that my analysis of x looks red to S at t has 11ot, at vides food for thought. Yct when it is claimed that "obviously" physical
276 277
 Wilfrid Sellars E MPIRICISM A:'iD THE PfilLOSOPHY OF M IND

objects can' t look red to one unless one is experiencing something that experiences' thought of the latter as the most untheoretical of entities,
is red, is it not presumed that the redness which the something has is indeed, as the observables par excellence.
the redness which the physical object looks to I1ave? Let us therefore tum to a second way in which, at least prima facie,
Now there are those who would say that the question "Is the fact there might be an additional, but equally legitimate explanation of
that an object looks red and triangular to S to be explained-as opposed existential and qualitative lookings. According to this second account,
to notationally reformulated-in terms of th e idea that S has an impres- when we consider items of this kind, we find that they contain as com·
sion of a red triangle?" simply doesn't arise, on the ground that there ponents items which are properly referred to as, for example, 'the
are perfectly sound explanations of qualitative and existential lookings immediate experience of a red triangle.' Let us begin our exploration
which make no reference to 'immediate experiences' or other dubious of this suggestion by taking another look at our account of existential
entities. Thus, it is pointed out, it is perfectly proper to answer the and qualitative lookings. It will be remembered that our account of
question "Why does this object look red?" by saying "Because it is qualitative looking ran, in rough and ready terms, as follows:
an orange object looked at in such and such circumstances." The ex- 'x looks red to S' has the sense of 'S has an experience which involves
planation is, in principle, a good one, and is typical of the answers in a unique way the idea that x is red and involves it in such a way
we make to such questions in everyday life. But because these explana· that if this idea were true, the experience would correctly be charac-
tions arc good, it by no means follows that explanations of other kinds terized as a seeing that x is red.'
might not be equally good, and, perhaps, more searching. '11ms, our account implies that the three situations
22. On the face of it there are at least two ways in which additional, (a) Seeing that x, over there, is red
but equally legitimate explanations might b e forthcoming for such a (b) Its looking to one that x, over there, is red
fact as that x looks red. 'The first of these is suggested by a simple (c) Its looking to one as though there were a red object over there
analogy. ~·fight it not be the case that just as there are two kinds of differ primarily in that (a ) is so formulated as to involve an endorse-
good explanation of the fact that this balloon has expanded, (a) in ment of the idea that x, over there, is red, whereas in (b) this idea
terms of the Boyle-Charles laws which relate the empirical concepts is only partially endorsed, and in ( c) not at all. Let us refer to the
of volume, pressure, and temperature pertaining to gases, and (b) in idea that x, over there, is red as the common propositional content of
terms of the kinetic theory of gases; so there are two ways of explaining these three situations. (This is, of course, not strictly correct, since
the fact that this ob ject looks red to S: (a ) in terms of empirical gen- the propositional content of ( c ) is existential, rather than about a pre-
eralizations relating the colors of objects, th e circumstances in which supposedly designated object x, but it will serve my purpose. Further-
they are seen, and the colors they look to have, and (b) in terms of more, the common propositional content of these three experiences
a theory of perception in which 'immediate experiences' play a role is much more complex and determinate than is indicated by the sen-
analogous to that of the molecules of tl1e kinetic theory. tence we use to describe our experience to otbers, and which I am
Now there is such an air of paradox to the idea that 'immediate ex- using to represent it. Nevertheless it is clear that, subject to the first
periences' are mere theoretical entities- entities, that is, wh ich are postu- of these qualifications, the propositional content of these three experi-
lated, along with certain fundamental principles concerning them, to ences could be identical. )
explain uniformities pertaining to sense perception, as molecules, along 'The propositional content of these three experiences is, of course,
with the principles of molecular motion, are postulated to explain the but a part of that to which we are logically committed by characl'crizing
experimentally determined regularities pertaining to gases-that I am them as situations of these three kinds. Of the remainder, as we have
going to lay it aside until a more propitious context of thought may seen, part is a matter of l'hc extent to which this propositional coutcnt
make it seem relevan t. Certainly, th ose who have though t f'liat q11alita- is endorsed. It is the rcsicl11c with \.vhich we :ire uow couc:cmccl . Let
l ivc and existential lookings are to be explained in terms of ' immediate us call this residue the descriptive corrtcrrt . I c 111 lhc11 point out that
278 279
 Wilfrid Sc11ars l<:MPIIUCISM AND THI': PlllLOSOPllY OF MINO

it is impliccl hy my account that not only the propositional content, this entails seeing that its facing surface is red. A red surface is a two-
but nlso Ilic descriptive content of these three experiences may be dimensional red expanse-two-dimensional in that though it may be
i<lc11tical. I shall suppose this to be the case, though that there must bulgy, and in tfos sense three-dimensional, it has no thickness. As far
as .the ~nalysis of perceptual consciousness is concerned, a red physical
be some factual difference in the total situations is obvious. obiect 1s one that has a red expanse as its surface.
Now, and this is the decisive point, in characterizing these three Now a red expanse is not a physical object, nor does the existence
experiences as, respectively, a seeing that x, over there, is red, its look- of a red expanse entail the existence of ~ physical object to which it
ing to one as though x, over there, were red, and its looking to one as belongs. .( Indee_d, there ar~ "wil~". expanses which do not belong to
though there were a red object over there, we do not specify this com- ~ny physical obiect.) The descnpttve content"-as you put it-which
mon descriptive content save indirectly, by implying that if the common is ~ommon to the three experiences (a), (b) and ( c) above, is exactly
this sort of thing, a bulgy red expanse.
propositional co11tent were true, then all these three situations would
be cases of seeing that x, over there, is red. Both existential and qualita- Spelled out thus baldly, the fallacy is, or should be, obvious; it is
tive lookings are experiences that would be seeings if their propositional a simple equivocation on the phrase "having a red surface." We start
contents were true. out by thinking of the familiar fact that a physical object may be of
Thus, the very nature of "looks talk" is such as to raise questions one color "on the surface" and of another color "inside." We may
to which it gives no answer: What is the intrinsic character of the express this by saying that, for example, the 'surface' of the object is
common descriptive content of these three experiences? and How are red, but its 'inside' green. But in saying this we are not saying that
they able to have it in spite of the fact that whereas in the case there is a 'surface' in the sense of a bulgy two-dimensional particular,
of (a) the perceiver must be in the presence of a red object over a red 'expanse' which is a component particular in a complex particular
there, in (b) the object over there need not be red, while in (c) there which also includes green particulars. The notion of two-dimensional
need be no object over there at all? bulgy (or flat) particulars is a product of philosophical (and mathe-
23. Now it is clear that if we were required to give a more direct matical) sophistication which can be related to our ordinary conceptual
framework, but does not belong in an analysis of it. I think that in its
characterization of the common descriptive content of these experi-
place it has an important contribution to make. (See below, Section
ences, we would begin by trying to do so in terms of the quality red.
61; (5), pp. 325-26.) But this place is in the logical space of an ideal
Yet, as I have already pointed out, we can scarcely say that this descrip-
scientific picture of the world and not in the logical space of ordinary
tive content is itself something red unless we can pry the term "red"
discourse. It bas nothing to do with the logical grammar of our ordinary
loose from its priroa-facie tie with the category of physical objects. And color words. It is just a mistake to suppose that as the word "red" is
there is a line of thought which has been one of the standard gambits actually used, it is ever surfaces in the sense of two-dimensional par-
of perceptual epistemology and which seems to promise exactly this. ticulars which are red. The only particular involved when a physical
If successful, it would convince us that redness-in the most basic sense object is "red on the outside, but green inside" is the physical object
of this term-is a characteristic of items of the sort we have been calling itself, located in a certain region of Space and enduring over a stretch
sense contents. It runs as follows: of Time. The fundamental grammar of the attribute red is physical
While it would, indeed, be a howler to say that we don't see chairs, ob;ect x is red at place p and at time t. Certainly, when we say of
tables, etc., but only their facing surfaces, nevertheless, although we an object that it is red, we commit ourselves to no more than th.at
see a table, say, and although the table has a back as well as a front, it is red "at the surface." And sometimes it is red at the surface by
we do not see the back of the table as we see its front. Again, although
we see the table, and although the table has an 'inside,' we do not see having what we would not hesitate to call a "part" which is red
the inside of the table as we see its facing outside. Seeing an object through and through-thus, a red table which is red by virtue of a
entails seeing its racing surface. If we are seeing that an object is red, layer of red paint. But the red paint is not itself red by virtue of a
280 281.
 Wilfrid Sellars El'vlPIRICISM AND THE PHILOSOPHY OF MIND

component-a 'surface' or 'expanse'; a particular with no thickness- the experience, but we should, of course, be only verbally better off
which is red. There may, let me repeat, turn out to be some place in than if we could only refer to this kind of experience as the kind which
the total philosophical picture for the statement that there " really could be the common descriptive component of a seeing and a quali-
are" such particulars, and that they are elements in perceptual experi- tative or existential looking. And this makes it clear that one way of
ence. But this place is not to be found by an analysis of ordinary putting what we are after is by saying that we want to have a name
perceptual discourse, any more than Minkowski four-dimensional Space- for this kind of experience which is truly a name, and not just short-
Time worms are an analysis of what we mean when we speak of physi- hand for a definite description. D oes ordinary usage have a name for
cal objects in Space and Time. this kind of C.'C perience?
I shall return to this quest in a moment. In the meantime it is
V. Impressions and Ideas: a Logical Point important to clear the way of a traditional obstacle to understanding
24. Let me return to beating the neighboring bushes. Notice that the status of such things as sensations of red triangles. Thus, suppose
the common descriptive companent of the three experiences I am I were to say that while the experience I am examining is not a red
considering is itself often referred to (by philosophers, at least ) as an experience, it is an experience of red. I could expect the immediate
experience-as, for example, an immediate experience. Here caution is challenge: "Is 'sensation of a red triangle' any better off than 'red and
necessary. The notorious "ing-ed" ambiguity of "experience" must be triangular experience'? Does not the existence of a sensation of a red
kept in mind. For although seeing that x, over there, is red is an experi- triangle entail t11e existence of a red and triangular item, and h ence,
encing-indeed, a paradigm case of experiencing-it does not follow always on the assumption t11at red is a property of physical ob;ects, of
that the descriptive content of this experiencing is itself an experiencing. a red and triangular physical object? Must you not, therefore abandon
Furthermore, because the fact that x, over t11ere, looks to Jones to be this assumption, and return to the framework of sense contents which
red would be a seeing, on Jones' part, th at x, over t11ere, is red, if its you have so far refused to do?"
propositional content were true, and because if it were a seeing, it One way out of dilemma would be to assimilate ';Jones has a sensa-
would he an experiencing, we must beware of concluding that the fact tion of a red triangle" to "Jones believes in a divine Huntress." For
that x, over there, looks red to fones is itself an experiencing. Certainly, the truth of the latter does not, of course, entail the existence of a
the fact that something looks red to me can itself be experienced. But divine Huntress. Now, I think that most contemporary philosophers
it is not itself an experiencing. are clear that it is possible to attribute to the context
All this is not to say that the common descriptive core may not . . . sensation of . . .
turn out to be an experiencing, though the chances that this is so the logical property of being such that "There is a sensation of a red
appear less with each step in my argument. On the other hand, I can triangle" does not entail "There is a red triangle" without assimilating
say that it is a component in states of affairs which are experienced, the context " . . . sensation of . .." to the context " . . . believes
and it does not seem unreasonable to say that it is itself experienced. in . . ." in any closer way. For while mentalistic verbs characteristically
But what kind of experience (in the sense of experienced ) is it? If provide nonextcnsional contexts (when they are not "achievement" or
my argum ent to date is sound, I cannot say that it is a red experience, "endorsing" words), not all nonextensional contexts are mentalistic.
that is, a red experienced item. I could, of course, introduce a new Thus, as far as the purely logical point is concerned, there is no reason
use of "red" according to which to say of an 'immediate experience' why "Jones has a sensation of a red triangle" should be assimilated 1:0
that it was red, would be the stipulated equivalent of characterizing it "Jones believes in a divine Huntress" rather than to " lt is possihlc H1at
as that which could be the common descriptive component of a seeing the moon is made of green cheese" or to any of the other no11cxt cmio11al
th:it something is red, and the corresponding qualitative and cxistcnti:1l contc:xts familiar t·o logicians. l mlcccl there is 110 rcasou why it .~ lt ould
1ooki11gs. 111is would give us a predicate hy which to describe :tncl rcpmt he :issimila tcd to any of these. " ... sc11satio11 of ... " or " ... im-
282 283
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

pression of . . ." could be a context which, though sharing with these "sensation of a red triangle" had the sense of "episode of the kind
others the logical property of nonextensionality, was otherwise in a which is the common descriptive component of those experiences which
class by itself. would be cases of seeing that the facing surface of a physical object is
25. Yet there is no doubt but that historically the contexts " . . . red and triangular if an object were presenting a red and triangular
sensation of . . ." and " ... impression of . . ." were assimilated to facing surface" then it would have the nonextensionality the noticing
such mentalistic contexts as " ... believes .. .," " . . . desires ...," of which led to this mistaken assimilation. But while we have indeed
" . . . chooses ...," in short to contexts which are either themselves escaped from this blind alley, it is small consolation. For we are no
'propositional attitudes' or involve propositional attitudes in their further along in the search for a 'direct' or 'intrinsic' characterization
analysis. This assimilation took the form of classifying sensations with of 'immediate experience.'
ideas or thoughts. Thus Descartes uses the word "thought" to cover
not only judgments, inferences, desires, volitions, and (occurrent) ideas VI. Impressions and Ideas: an Historical Point
of abstract qualities, but also sensations, feelings, and images. Locke, 26. There are those who will say that although I have spoken of
in the same spirit, uses the term "idea" with similar scope. TI1e appa- exploring blind alleys, it is really I who am blind. For, they will say,
ratus of Conceptualisrn, which had its genesis in the controversy over if that which we wish to characterize intrinsically is an experience, then
universals, was given a correspondingly wide application. Just as objects there can be no puzzle about knowing what kind of experience it is,
and situations were said to have 'objective being' in our t11oughts, when though there may be a problem about how this knowledge is to be
we think of them, or judge them to obtain-as contrasted with the communicated to others. And, indeed, it is tempting to suppose that
'subjective' or 'formal being' which they have in the world-so, when if we should happen, at a certain stage of our intellectual development,
we have a sensation of a red triangle, the red triangle was supposed to be able to classify an experience only as of the kind which could be
to have 'objective being' in our sensation. common to a seeing and corresponding qualitative and existential look-
In elaborating, for a moment, this conceptualistic interpretation of ings, all we would have to do to acquire a 'direct designation' for this
sensation, let me refer to that which has 'objective being' in a thought kind of experience would be to pitch in, 'examine' it, locate the kind
or idea as its content or immanent object. Then I can say that the which it exemplifies and which satisfies the above description, name
fundamental difference between occurrent abstract ideas and sensations, it-say ".p" -and, in full possession of the concept of ef>, classify such
for both Locke and Descartes, Jay in the specificity and, above all, the experiences, from now on, as cf> experiences.
complexity of the content of the latter. (Indeed, both Descartes and At this point, it is clear, the concept-or, as I have put it, the myth-
Locke assimilated the contrast between the simple and the complex of the given is being invoked to explain the possibility of a direct
in ideas to that between the generic and the specific.) Desantes thinks account of immediate ex-perience. The myth insists that what I have
of sensations as confused thoughts of their external cause; Spinoza of been treating as one problem really subdivides into two, one of which
sensations and images as confused thoughts of bodily states, and still is really no problem at all, while the other may have no solution. These
more confused thoughts of the external causes of these bodily states. problems are, respectively
And it is interesting to note that the conceptualistic thesis that abstract
(1) How do we become aware of an immediate experience as of one
entities have only esse intentionaie (their esse is ooncipi) is extended by sort, and of a simultaneous immediate experience as of another sort?
Descartes and, with less awareness of what he is doing, Locke, to incJudc (2) :How can I know that the labels I attach to the sorts to which
the thesis that colors, sounds, etc., exist "only in the mind" (their cssc my immediate experiences belong, are attached by you t·o 1'11c same
is percipi) and by Berkeley to cover all perceptible qualities. sorts? May not the sort I call "red" he 1'11e sort· yon cull "green"-and
Now, I think we would all agree, today, that this assimilation of so on systematically throughout the spectrum?
sensations to thoughts is a mistake. Jt is sufficient to note that if We shaJl find that the second q11cslio11, to he u philosophicu] per-
284 285
 Wilfrid Sellars EMPIRICISI\I AND TIIE PHILOSOPHY OF MIND

plexity, presupposes a certain answer to the first question-indeed the notorious that this won't account for the relation of the idea of being
answer given by the myth. And it is to this first question that I now red to the idea of being crimson. By thinking of con;unction as the
turn. Actually there are various forms taken by the myth of the given fundamental logical relation involved in building up complex ideas
in this connection, depending on other philosophical commitments. from simple ones, and as the principle of the difference between
But they all have in common the idea that the awareness of certain determinable and determinate ideas, Locke precluded himself from
sorts- and by "sorts" I have in mind, in the first instance, dctcm1inate giving even a plausible account of the relation between ideas of de-
sense repeatables-is a primordial, non-problematic feature of 'imme· tenninab]es and ideas of dctenninates. It. is interesting to speculate
diate experience.' In the context of conceptualism, as we have seen, what turn his thought might have taken had he admitted disjunctive
this idea took the form of treating sensations as though they were as well as conjunctive complex ideas, the idea of being A or B along·
absolutely specific, and infinitely complicated, tltouglJts. And it is side the idea of being A and B.
essential to an understanding of the empiricist tradition to realize that 27. But my purpose here is not to develop a commentary on the
whereas the contemporary problem of universals primarily concerns the shortcomings of Locke's treatment of abstract ideas, but to emphasize
status of repeatable determinate features of particular situations, and that something which is a problem for us was not a problem for him.
the contemporary problem of abstract ideas is at least as much the And it i~ therefore important to note that the same is true of Berkeley.
problem of what it is to be aware of determinate repcatables as of His problem was not, as it is often construed, "How do we go from
what it is to be aware of determinable repeatables, Locke, Berkeley and, the awareness of particulars to ideas of repeatables?" but rather "Granted
for that matter, Hume saw the problem of abstract ideas as the prob· that in immediate experience we are aware of absolutely specific sense
lem of what it is to be aware of determinable repeatables. * T hus, an qualities, how do we come to be conscious of genera pertaining to
examination of Locke's Essay makes it clear that he is thinking of a them, and in what does this consciousness consist?" (This is not the
sensation of white as the sort of thing that can become an abstract only dimension of "abstraction" that concerned him, but it is the one
idea ( occurrent) of Wbite-a thought of 'W hite "in the Understand· that is central to our purpose.) And, contrary to the usual interpreta·
ing"-merely by virtue of being separated from the context of other tion, the essential difference between his account and Locke's consists
sensations (and images) which accompany it on a particular occasion. in the fact that whereas Locke was on the whole# committed to the
In other words, for Locke an abstract (occurrent) idea of the detenni· view that there can be an idea which is of the genus without being of
nate repeatable Whiteness is nothing more than an isolated image of
• I say that Locke was "on the whole" committed to tlie view that there can be
wl1itc, which, in turn, differs from a sensation of wllite only (to use a an idea which is of the genus without being of any of its species, because while he
modern turn of phrase) by being "centrally aroused.'' saw that it couldn't be of any one of the species to the exclusion of the others, and
saw no way of avoiding this except by making it of none of the species, he was
In short, for Locke, the problem of how we come to be aware of greatly puzzled by this, for he saw that in some sense the idea of the genus must be
determinate sense repeatablcs is no problem at all. Merely by virtue of all the species. We have already noted that if he had admitted disjunction as a
of having sensations and images we have this awareness. His problem principle of compounding ideas, he could have said that the idea of the genus is the
idea of the disjunction of all its species, that th e idea of being triangular is the idea
of abstract ideas is the problem of how we come to be able to think of being scalene or isosceles. As it was, b e thought that to be of all tl1e species it
of generic properties. And, as is clear from the Essay, he approaches would have to be the idea of beiug scalene and isosceles, whicl1 is, of course, the
idea of an impossibility.
this problem in terms of what might be called an "adjunctive theory of It is interesting to note that if Berkeley had faced up to the implic.1tions of the
specification," that is, the view that (if we represent the idea of :1 criterion we shall find him to have adopted, this disj unctive conception of 1'11c generic
idea is the one he wo111cl have hccn led to adopt. For since being G-whcrc '(;' slands
determinable as the idea of being A) the idea of a determinate for111 for a generic charactcr- ('lltails beings. ors. ors• . . . . . or s.,-whcrc ·s,. stands
of A can be represented as tl1e idea of being A and B. It is, of course, for a specific charadcr falling under C-Bcrkclcy should have taken as th(' 1111it of
ideas concerning l ri:111i;lc~. lhc idc:i of th e gc1111s Triangle as dirTcrl'llli:1IC'cl into the
• l•'or a systematic clahomtion :md dcfcnt·e of the following i11terprcl11tin11 ol sci of specific for111s of lria11g11l:1ritr. But . 11 ccdlcs~ In say, if llcrkt·lcy Imel t:1kc11 this
I .ode, llcrkclcy, and I lnmc, the reader should <·onsnlt ( 11 ). skp, he could 1wl hnvc lhm1ulit o 11 Sl'll~aliu11 ul ('ri111~011 a~ a tlt·l<'n11i1111tc 1/11111g/1t.

286 2R7
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

any of its species, Berkeley insists that we can have an idea of a genus 29. Now, it takes but a small twist of Hume's position to get a radi-
only by having an idea of the genus as, to borrow a useful Scotist term, cally different view. For suppose that instead of characterizing the initial
'contracted' into one of its species. elements of experience as impressions of, e.g. red, Hume had charac-
Roughly, Berkeley's contention is that if being A entails being B, terized them as red particulars (and I would be the last to deny that
then there can be no such thing as an idea which is of A without being not only Hume, but perhaps Berkeley and Locke as well, often treat
of B. He infers that since being triangular entails having some determi- impressions or ideas of red as though they were red particulars) then
nately triangular shape, there cannot be an idea which is of triangle Hume's view, expanded to take into account determinates as well as
without being of some determinately triangular shape. We can be aware determinables, would become the view that all consciousness of sorts
of generic triangularity only by having an idea which is of triangularity or repeatables rests on an association of words (e.g. "red") with classes
as 'contracted' into one of the specific forms of triangularity. Any of of resembling particu1ars.
the latter will do; they are all "of the same sort.'' It clearly makes all the difference in the world how this association
· 28. Now, a careful study of the Treatise makes it clear that Hume is conceived. For if the formation of the association involves not only
is in the same boat as Berkeley and Locke, sharing with tbem the pre- the occurrence of resembling particulars, but also the occurrence of
supposition that we have an unacquired ability to be aware of determi- the awareness that they are resembling particulars, then the givenness
nate repeatables. It is often said that whereas he begins the Treatise of determinate kinds or repeatables, say crimson, is merely being re-
by characterizing 'ideas' in terms which do not distinguish between placed by the givenness of facts of the form x resembles y, and we
images and thoughts, he corrects this deficiency in Book I, Part I, Sec- are back with an unaequired ability to be aware of repeatables, in this
tion vii. What these students of Hurne tend to overlook is that what case the repeatable resemblance. Even more obviously, if the forma-
Hume does in this later section is give an account not of what it is to tion of the association involves not only the occurrence of red particu-
think of repeatables whether determinable or determinate, but of what lars, but the awareness that they are red, then the conceptualistic form
it is to think of determinables, thus of color as contrasted with par- of the myth has merely been replaced by a realistic version, as in the
ticular shades of color. And his account of the consciousness of deter- classical sense-datum theory.
minables takes for granted that we have a primordial ability to take If, however, the association is not mediated by the awareness of facts
account of determinate repeatables. T h us, his later account is simply either of the form x resembles y, or of the form x is </i, then we have
built on, and in no sense a revision of, the account of ideas with which a view of the general type which I will call psychological nominaiism,
he opens the Treatise. according to which all awareness of sorts, resemblances, facts, etc., in
How, then, does he differ from Berkeley and Locke? The latter two short, all awareness of abstract entities-indeed, all awareness even of
had ·supposed that there must be such a thing as an occurrent thought particulars-is a linguistic affair. According to it, not even the aware-
of a determinable, however much they differed in their account of such ness of such sorts, resemblances, and facts as pertain to so-called im-
thoughts. Hume, on the other hand, assuming that there are occurrent m ediate experience is presupposed by the process of acquiring the use
thoughts of determinate repeatables, denies that there are occurrent of a language.
thoughts of determinab1es. I shall spare the reader the familiar details Two remarks are immediately relevant: ( l) Alth ough the form of
of Hume's attempt to give a constructive account of our consciousnes.q psychological n ominalism which one gets by modifying Hume's view
of determinables, nor shall I criticize it. For my point is that however along the above lines has the essential merit that it avoids the mistake
much Locke, Berkeley, and Hume differ on the problem of abstract of supposing that there arc pure episodes of being aware of sensory
ideas, they all take for granted that the human mind h as an innate rcpeatables or sensory facts, and is committed to the view lfo1t any
ability to be aware of certain determinate sorts-indeed, that we arc event which cnn he referred to in these terms must he, to use Rylc's
aware of them simply by virtue of having sensations and imugcs. expression, a mongrel cntcgorical-hypolhctical, i11 partirnl11r, n vcrlml
288 2fi9
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

episode as being tlle manifestation of associative connections of the the logical space of which the child is supposed to have this undiscrimi-
word-object and word-word types, it nevertheless is impossibly crude nating awareness is conceived by us to be that of physical objects or of
and inadequate as an account of the simplest concept. (2) Once sensa- private sense contents.
tions and images have been purged of epistemic aboutness, the primary The real test of a theory of language lies not in its account of what
reason for supposing that the fundamental associative tie between lan- h as been called (by H . H. Price) "th inking in absence," but in its
guage and the world must be between words and 'immediate experi- account of " thinking in presence"-that is to say, its account of those
ences' has disappeared, and the way is clear to recognizing that basic occasions ou which the fundamental connection of language with non-
word-world associations hold, for example, between "red" and red linguistic fact is exhibited. And many theories which look like psycho-
physical objects, rath er than between "red" and a supposed class of logical nominalism when one views their account of thinking in absence,
private red particulars. turn out to be quite "Augustinian" when the scalpel is turned to their
The second remark, it should be emphasized, does not imply that account of th inking in presence.
private sensations or impressions may not be essential to the formation 31. Now, the friendly use I have been making of the phrase "psycho-
of these associative connections. For one can certainly admit that the logical nominalism" may suggest that I am about to equate concepts
tie between "red" and red physical objects-which tic makes it possible with words, an d thinking, in so far as it is episodic, with verbal episodes.
for "red" to mean the quality red-is causally mediated by sensations I must now hasten to say that I shall do nothing of the sort, or, at least,
of red without being committed to the mistaken idea that it is "really" that if I do do sornet11ing of the sort, the view I shall shortly be de-
sensations of red, rather than red physical objects, which are the primary veloping is only in a relatively Pickwickian sense an equation of tltinki11g
denotation of the word "red." with th e use of language. I wish to emphasize, therefore, that as I :1111
using the term, the primary connotation of "psychological nomi11alis111"
VII. The Logic of 'Means' is the denial that there is any awareness of logical space prior h>, or
30. There is a source of the Myth of the Given to which even phi- independent of, the acquisition of a language.
losophers who are suspicious of the whole idea of inner episodes can However, although I shaJJ later be distinguishing between lhougltl·s
fall prey. This is the fad that when we picture a child-or a carrier of and their verbal expression, there is a point of funda mental i111porl :111ce
slabs-learning his first language, we, of course, locate the language which is best made before more subtle distinctions arc draw11 . 'l'o
learner in a structured logical space in which we are at home. Thus, we begin with, it is perfectly clear that the word "red" woulcl not he a
conceive of him as a person (or, at least, a potential person) in a world predicate if it didn't have the logical syntax characteristic of predirnlcs.
of physical objects, colored, producing sounds, existing in Space and Nor would it be the predicate it is, unless, in certain frames of 111i11d,
Time. But though it is we who are familiar with this logical space, we at least, we tended to respond to red objects in standard circ11111sl1111ccs
run the danger, if we are not careful, of picturing the language learner with something having the force of "This is red." Ancl once we lt:ivc
as having ab initio some degree of awareness-"pre-analytic," limited abandoned the idea that learning to use the word "red" involves 1111l c·
and fragmentary though it may be-of this same logical space. \:Ve cedent episodes of the awareness of redness-not to he c1111 f11sed, of
picture his state as though it were rather like our own when placed course, with sensations of red-there is a tcmpt·at·ion lo s11ppwa· tli111·
in a strange forest on a dark night. In other words, unless we are care- the word "red" means the quality reel hy virtue of t lt e.~e I wo foci s:
ful, we can easily take for granted that the process of teaching a child briefly, the foct that it has the syntax of a predica te, 11 11<1 tlte fud llt111·
to use a language is that of teaching it to discriminate elements with in it is a response ( in certain circumstan ce.~) lo n.:d oh jcrl:I.
a logical space of particulars, universals, facts, etc., of which it is already But this acco1111t of the 111c:111i11gful11css of "rnl." whiclt 1'1 i<'e h:1s
uncliscriminatingly aware, and to associate these discriminated clements correctly stigmatizccl as lhc "l hcnuo111cler view," would l111vc lil tlc
wi~h verbal symbols. And this mistake is in principle the same whether plausibility if it were 11ot rci11forccd hy 111101her li11c of I h 1111~;lil whi<'h
290 291
 Wilfrid Sellars EMPIRICISM AND THE PHil.OSOPHY OF MIND
takes its point of departure from the superficial resemblance of These considerations make it clear that nothing whatever can be
(In German) "rot" means red inferred about the complexity of . the role played by the word "red"
to such relational statements as or about the exact way in which the word "red" is related to red
Cowley adjoins Oxford. things, from the truth of the semantical statement "'red' means the
F or once one assimilates the form quality red." And no consideration arising from the 'Fido'-Fido aspect
". . ." means - - - of the grammar of "means" precludes one from claiming that the role
to the form xRy of the word "red" by virtue of which it can correctly be said to have
and thus takes it for granted that meaning is a relation between a the meaning it does is a complicated one indeed, and that one cannot
word and a nonverbal entity, it is tempting to suppose that the understand the meaning of the word "red"-"know what redness is"-
relation in question is that of association. unless one has a great deal of knowledge which classical empiricism
The truth of the matter, of course, is . that statements of the form would have held to have a purely contingent relationship with the
"'.. .' means - - -" are not relational statements, and that while it possession of fundamental empirical concepts.
is indeed the case that the word "rot" could not mean the quality red
unless it were associated with red things, it would be misleading to VIII. Does Empirical Knowledge Have a Foundation?
say that the semantical statement "'Rot' means red" says of "rot" that 32. One of the forms taken by the Myth of the Given is the idea
it associated with red things. For this would suggest that the semantical that there is, indeed must be, a structure of particular matter of fact
statement is, so to speak, definitional shorthand for a longer statement such that (a) each fact can not only be noninferentially known to
about the associative connections of "rot," which is not the case. The be the case, but presupposes no other knowledge either of particular
rubric " '. . .' means - - -" is a linguistic device for conveying the matter of fact, or of general truths; and (b) such that the noninferential
information that a mentioned word, in this case "rot," plays the same knowledge of facts belonging to this structure constitutes the ultimate
role in a certain linguistic economy, in this case the linguistic economy court of appeals for all factual claims-particular and general- about the
of German-speaking peoples, as does the word "red," which is not world. It is important to note that I characterized the knowledge of
mentioned but used-used in a unique way; exhibited, so to speak-and fact belonging to this stratum as not only noninferential, but as pre-
which occurs "on the right-hand side" of the semantical statement. supposing no knowledge of other matter of fact, whether particular or
We see, therefore, how the two statements general. It might be thought that th is is a redundancy, that knowledge
" Vnd" means and (not belief or conviction, but knowledge) which logically presupposes
and knowledge of other facts must be inferential. This, however, as I hope
"Rot'' means red to show, is itself an episode in the Myth.
can tell us quite different things about "und" and "rot," for the first Now, the idea of such a privileged stratum of fact is a familiar one,
conveys the information that "und'' pJays the purely formal role of a though not without its difficulties. Knowledge pertaining to this level
certain logical connective, the second that "rot" plays in German the is noninferential, yet it is, after all, knowledge. It is ultimate, yet it has
role of the observation word "red"-in spite of the fact that means has authority. The attempt to make a consistent picture of these two re-
the same sense in each statement, and without having to say that the quirements has traditionally taken the following fonn:
first says of "und" that it stands in "the meaning relation" to Conjunc· Statements pertaining to this level, in order to 'express knowledge'
tion, or the second that "rot" stands in "the meaning relation" to must not only be made, but, so to speak, must be worthy of hcing made,
Redness.* credible, that is, in the sense of worthy of credence. Furt·hcnnorc, and
• ro'or an analysis of the problem of abstract entities built on this intcrprch1tion of
this is a crucial point, they must be made in a way which i11volvcs this
scnumtical statements, sec (20). · credibility. F or where l'licre is no connection hclwecn the mnking of
292 293
 Wilfrid Sellars EM PIRICISM AND THE PHlLOSOPHY OF MIND

a statement and its authority, tl1e assertion may express conviction, but type or token. This is, or seems to be, the case with certain sentences
it can scarceJy be said to express knowledge. · used to make analytic statements. The credibility of some sentence
The authority-the credibility-of statements pertaining to this level types accrues to them by virtue of their logical relations to other sen-
cannot exhaustively consist in the fact that they are supported by other tence types, thus by virtue of the fact that they are logic-al consequences
statements, for in that case all knowledge pertaining to this level would of more basic sentences. It would seem obvious, however, that the
have to be inferential, which not only contradicts the hypothesis, but credibility of empirical sentence types cannot be traced without re-
flies in the face of good sense. The conclusion seems inevitable that if mainder to the credibility of other sentence types. And since no empiri-
some statements pertaining to this level are to express noninferentia1 cal sentence type appears to have intrinsic credibility, this means that
knowledge, rhey m ust have a credibility which is not a matter of being credibility must accrue to some empirical sentence types by virtue of
supported by other statements. Now there does seem to be a class of their logical relations to certain sentence tokens, and, indeed, to sen-
statements which fill at least part of this bill, namely such statements tence tokens the authority of which is not derived, in its tum, from
as would be said to report observations, thus, "This is red." These state- the authority of sentence types.
ment~, candidly made, have authority. Yet they are not expressions of 111e picture we get is that of their being two ultimate modes of
inference. How, then, is this authority to be understood? credibility: (I) The intrinsic credibility of analytic sentences, which
Clearly, the argument continues, it springs from the fact that they accrues to tokens as being tokens of such a type; (2) the credibility
are made in just the circumstances in which they are made, as is of such tokens as "express observations," a credibility wh ich flows from
indicated by the fact that they characteristically, th ough not necessarily tokens to types.
or without exception, involve those so-called token-reflexive expressions
which, in addition to the tenses of verbs, serve to connect the circum- 33. Let us explore this picture, which is common to all traditional
stances in which a statement is made with its sense. (At this point it empiricisms, a bit further. How is the authority of such sentence tokens
will be helpful to begin putting the line of thought I am developing as "express observational knowledge" to be understood? It has been
in terms of the fact-stating and observation-reporting roles of certain tempting to suppose that in spite of the obvious differences which
sentences.) Roughly, two verbal performances which are tokens of a
exist between "observation reports" and "analytic statements," there
non-token-reflexive sentence can occur in widely different circumstances
and yet make the same statement; whereas two tokens of a token- is an essential similarity between the ways in which they come by
rcf:lexive sentence can make the same statement only if they are uttered their authority. T hus, it has been claimed, not without plausibility,
in the same circumstances (according to a relevant criterion of same- that whereas ordinary empirical statements can be correctly made with-
ness). And two tokens of a sentence, whether it contains a token- out being true, observation reports resemble analytic statements in that
reflexive expression-over and above a tensed verb-or not, can make
the same report only if, made in all candor, they express the presence- being correctly made is a sufficient as well as necessary condition of
in some sense of "prescnce"-of the state of affairs that is being re- their truth. And it has been inferred from this-somewhat hastily, I
ported; if, that is, they stand in that relation to the state of affairs, believe-that "correctly making" the report "111is is green" is a matter
whatever the relation may be, by virtue of which they can be said to of "following the rules for the use of 'this,' 'is' and 'green.'"
formulate observations of it. Three comments are immediately necessary:
It would appear, then, that there arc two ways in which a sentence
(1) First a brief remark about the tenn "report." In ordinary usage
token can have cred ibility: ( 1) The authority may accrue to it, so to
speak, from above, that is, as being a token of a sentence type all the a report is a report made by someone to someone. To make a report
tokens of wl}.ich, in a certain use, have credibility, e.g. "2 + 2 = 4." J11 is to do something. In the literature of epistemology, however, the word
this case, let us say that token credibility is inherited from type authority. "report" or "Konst:iticrung" has acquired a technical use in wh ich a
(2) The credibility may accrue to it from the fact that it came to exist sentence token c.1n pby a reporting role (a) without hci11g a11 overt
in a certain way in a certain set of circumstances, e.g. "This is red." verbal performance, :incl (h) without having I he diaraclcr of hci11g "hy
Ilcre token credibility is not derived from type credibility.
Now, the credibility of some sentence types appears to be i11lrinsic- someone to somco11e"- cvc11 onesel f. There is, of rnursc, s11d1 :1 tlt i11g
at least in tl1c limited sense tl1at it is not derived from other sc11l c11ccs, as "talking to 0t1<.;self"-i11 foro iul cr110- b11t, as I sltall lie cn1ph:isizi11g

294 295
 Wilfrid Sellars EMPIRICISM .AND THE PHILOSOPHY OF MIND

in the closing stages of my argument, it is important not to suppose in the presence of a green item is a Konstatierung and expresses observa·
that all "covert" verbal episodes are of this kind. tional knowledge if and only if it is a manifestation of a tendency to
(2) My second comment is that while we shall not assume that be· produce overt or covert tokens of "This is green"-given a certain set-
cause ' reports' in the ordinary sense are actions, 'reports' in the sense if and only if a green object is being looked at in standard conditions.
of Konstatierungen are also actions, the line of thought we are consider· Clearly on this interpretation the occurrence of such tokens of "TI1is
ing treats them as such. In other words, it interprets the correctness of is green" would be "following a rule" only in the sense that they are
Konstatierungen as analogous to the rightness of actions. Let me empha· instances of a uniformity, a uniformity differing from the lightning·
size, however, that not all ought is ought to do, nor all correctness the thunder case in that it is an acquired causal characteristic of the Ian·
correctness of actions. guage user. Clearly the above suggestion, which corresponds to the
(3) My third com.m ent is that if the expression "following a rule" is "thermometer view" criticized by Professor Price, and which we have
taken seriously, and is not weakened beyond all recognition into the already rejected, won' t do as it stands. Let us see, however, if it can't
bare notion of exhibiting a uniformity-in which case the lightning, be revised to fit the criteria I have been using for "expressing observa-
thunder sequence would "follow a rule"-then it is the knowledge or tional knowledge."
belief that the circumstances are of a certain kind, and not the mere T he first hurdle to be jumped concerns the authority which, as I have
fact that they are of this kind, which contributes to bringing about th e emphasized, a sentence token must have in order that it may be said
action. to express knowledge. Clearly, on this account the only thing that can
34. In the light of these remarks it is clear that if observation reports remotely be supposed to constitute such authority is the fact that one
are construed as actions, if their correctness is interpreted as the cor· can infer the presence of a green object from the fact that someone
redness of an action, and if th e authority of an observation report is makes this report. As we have already noticed, the correctness of a
construed as the fact that making it is "following a rule" in the proper report does not have to be construed as the rightness of an action.
sense of this plirase, then we are face to face with givenness in its most A report can be correct as being an instance of a general mode of
straightforward form. For these stipulations commit one to the idea behavior which, in a given linguistic community, it is reasonable to
that the authority of Konstatierungen rests on nonverbal episodes of sanction and support.
awareness-awareness that something is the case, e.g. that thi.s is green- T he second hurdle is, however, th e decisive one. For we have seen
which nonverbal episodes have an intrinsic authority ( they are, so to that to be the expression of knowledge, a report must not only have
speak 'self-authenticating') which the verbal performances (the Konsta· authority, this authority must in some sense be recognized by the person
tierungen) properly performed "e.'<press." One is committed to a whose report it is. And this is a steep hurdle indeed. For if the authority
stratum of authoritative nonverbal episodes ("awarenesses") the a uthority of the report "This is green" lies in the fact that the existence of green
of wh ich accrues to a superstructure of verbaJ actions, provided that items appropriately related to the perceiver can be inferred from the
the expressions occuning in these actions are properly used. These self· occurrence of such reports, it follows that onJy a person who is able
authenticating episodes would constitute the tortoise on which stands to draw this inference, and therefore who has not only the concept
the elephant on which rests the edifice of empirical knowledge. The green, but also the concept of uttering "This is green"-indeed, the
essence of the view is the same whether these intrinsically authoritative concept of certain conditions of perception, those which would cor·
episodes are such items as the awareness that a certain sense content rectly be called 'standard conditions'-could be in a position to token
is green or such items as the awareness that a certain physical object "This is green" in recognition of its authority. In other words, for a
looks to someone to be green. Konstatierung "This is green" to "express observational knowledge," not
35. But what is the alternative? We might begin by trying some- only must it he a .~yrnptom or sign of the presence of 11 grcc11 object
thing like the following: An overt or covert token of "This is ·green" in standard conditions, hut the perceiver must know thnt tokens of
296 2Q7
 Wilfrid Sellars EMPIRICISM AND -THE PHILOSOPHY OF MIND

"11lis is green" are symptoms of the presence of green objects in con- it in the logical space of reasons, of justifying and being able to justify
ditions which are standard for visual pcrc"Cption. what one says.
36. Now it might be thought that there is something obviously absurd 37. Thus, all that the view I am defending requires is that no token-
in the ide-a that before a token uttered by, say, Jones could be the ing by S now of "This is green" is to count as "expressing observational
expression of observational knowledge, Jones would have to know that knowledge" unless it is also correct to say of S that he now knows the
overt verbal episodes of this kind are reliable indicators of the existence, appropriate fact of the form X is a reliable symptom of Y, namely that
suitably related to the speaker, of green objects. I do not think that (and again I oversimplify) utterances of "This is green" arc reliable
it is. Indeed, I think that something very like it is true. The point I indicators of the presence of green objects in standard conditions of
wish to make now, however, is that if it is true, then it follows, as a perception. And while the correctness of this statement about Jones
matter of simple logic, that one couldn't have observational knowledge requires that Jones could now cite prior particular facts as evidence for
of any fact unless one knew many· other things as well. And let me the idea that these utterances are reliable indicators, it requires only
emphasize that the point is not taken care of by distinguishing between that it is correct to say that Jones now knows, thus remembers, that
knowing how and knowing that, and admitting that observational knowl- these particular facts did obtain. It does not require that it be correct
edge requires a lot of "know how." For the point is specifically that to say that at the time these facts did obtain he t11en knew them to
observational knowledge of any particular fact, e.g. that this is green, obtain. And the regress disappears.
presupposes that one knows general facts of the form X is a reliable Thus, while Jones' ability to give inductive reasons today is built on
symptom of Y. And to admit this requires an abandonment of the a long history of acquiring and manifesting verbal habits in perceptual
traditional empiricist idea that observational knowledge "stands on situations, and, in particular, the occurrence of verbal episodes, e.g.
its own feet." Indeed, the suggestion would be anathema to traditional "This is green," which is superficially like those which are later properly
empiricists for the obvious reason that by making observational knowl- said to express observational knowledge, it does not require that any
edge presuppose knowledge of general facts of the form X is a reliable episode in this prior time be characterizeable as expressing knowledge.
symptom of Y, it runs counter to the idea that we come to know (At this point, the reader shouJd reread Section 19 above.)
general facts of thls form only after we have come to know by observa- 38. The idea that observation "strictly and properly so-called" is con-
tion a number of particular facts which support the hypothesis that stituted by certain seJf-authenticating nonverbal episodes, the authority
X is a symptom of Y. of which is transmitted to verbal and quasi-verbal performances when
And it might be thought that there is an obvious regress in the view these performances are made "in conformity with the semantical rules
we are examining. Does it not tell us that observational knowledge at of the language," is, of course, the heart of the Myth of the Given. For
time t presupposes knowledge of the form X is a reliable symptom of the given, in epistemological tradition, is what is taken by these self-
Y, which presupposes prior observational knowledge, which presupposes authenticating episodes. These 'takings' are, so to speak, the unmoved
otller knowledge of the form X is a reliable symptom of Y, which pre- movers of empirical knowledge, the 'knowings in presence' which are
supposes still other, and prior, observational knowledge, and so on? presupposed by all other knowledge, both the knowledge of general
This charge, however, rests on too simple, indeed a radically mistaken, truths and the knowledge 'in absence' of other particular matters of
conception of what one is saying of Jones when one says that he knows fact. Such is the framework in which traditional empiricism makes its
that p. It is not just that the objection supposes that knowing is an characteristic claim that the perceptually given is the foundation of
episode; for clearly there are episodes which we can correctly character- empirical knowledge.
ize as knowings, in particular, ol>scrvings. The essential point is that Let me make it clear, however, that if I reject this framework, ii· is
in eharnctcrizing nn episode or a state ns that of knowing, wc arc not not because I should deny th:it oh.5crvings arc i1111cr episodes, 11or that
giving an empirical description of that episode or state; wc :arc placing strictly spcaki11g they arc 11011vcrlml episodes. It· will be my c:onl cnl"ion.
298 299
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

however, that the sense in which they are nonverbal-which is also the the Philosophy of Science. Nor shall I attempt to locate this new
sense in which thought episodes are nonverbal-is one which gives no specialty in a classificatory system . The point I wish to make, however,
aid or comfort to epistemological givenness. In the concluding sections can be introduced by calling to mind the fact that classificatory schemes,
of this paper, I shall attempt to explicate the logic of inner episodes, however theoretical their purp~se, have practical consequences: nominal
and show that we can distinguish between observations and thoughts, causes, so to speak, have real effects. As Jong as there was no such
on the one hand, and their verbal expression on the other, without subject as 'philosophy of science,' all students of philosophy felt obli-
making the mistakes of traditional dualism. I shall also attempt to gated to keep at least one eye part of the time on both the methodo-
explicate the logical status of impressions or immediate experiences, and logical and the substantive aspects of the scientific enterprise. And if
thus bring to a successful conclusion the quest with which my argument the result was often a confusion of the task of philosophy with the
began. task of science, and almost equally often a projection of the framework
One final remark before I begin this task. If I reject the framework of the latest scientific speculations into the common-sense picture of
of traditional empiricism, it is not because I want to say that empirical the world (witness the almost unquestioned assumption, today, that
knowledge has 110 foundation. For to put it this way is to suggest that the common-sense world of physical objects in Space and Time must
it is really "empirical knowledge so-called," and to put it in a box be analyzable into spatially and temporally, or even spatiotemporally,
with rumors and hoaxes. There is dearly some point to the picture of related events), at least it had the merit of ensuring that refl ection on
human knowledge as resting on a level of propositions-observation re- the nature and implications of scientific discourse was an integral and
ports-which do not rest on other propositions in the same way as vital part of philosophical thinking generally. But now that philosophy
other propositions rest on them. On the other hand, I do wish to of science has nominal as well as real existence, there has arisen the
insist that the metaphor of "foundation" is misleading in that it keeps temptation to leave it to the specialists, and to confuse the sound idea
us from seeing tbat if there is a logical dimension in which other that ph ilosoph y is not science with the mistaken idea that philosophy
empirical propositions rest on observation reports, there is another logi· is independent of science.
cal dimension in which the latter rest on the former. 40. As long as discourse was viewed as a map, subdivided into a side-
Above all, the picture is misleading because of its static character. by-side of sub-maps, each representing a sub-region in a side-by·side of
One seems forced to choose between the picture of an elephant which regions making up the total subject matter of discourse, and as long
rests on a tortoise (What supports the tortoise?) and the picture of as the task of the philosopher was conceived to be the piecemeal one
a great H egelian serpent of knowledge with its tail in its mouth ('Where of analysis in the sense of definition-the task, so to speak, of "making
does it begin?). Neither will do. For empirical knowledge, like its little ones out of big ones"-one could view with equanimity the exist-
sophisticated extension, science, is rational, not because it has a founda· ence of philosophical specialists- specialists in formal and mathematical
tion but because it is a self-correcting enterprise which can put any logic, in perception, in moral philosophy, etc. For if discourse were as
claim in jeopardy, though not all at once. represented above, where would be the harm of each man fencing him-
self off in his own garden? In spite, however, of the persistence of the
IX. Science and Ordinary Usage slogan "philosophy is analysis," we now realize that the atomistic con·
39. There are many strange and exotic specimens in the gardens of ception of philosophy is a snare and a delusion. For "analysis" no longer
philosopby: Epistemology, Ontology, Cosmology, to name but a few. connotes the definition of terms, but rather the clarification of the
And clearly there is much good sense-not only rhyme hut reason-to logical structure- in the broadest sense-of discourse, and discourse no
these labels. It is not my purpose, however, to animadvert on the botnn· longer appears as one plane parallel to anot11cr, but as a hmgle of in-
izing of philosophies and things philosophical, other than to call uttcn· tersecting dimensions whose relations with one another 1111<1 with extra-
tion to a recent addition to the list of philosophical flora and founu , linguistic fa ct conform to no single or simple pattern. No longer c:an
300 301
 Wilfrid Sellars EMPIRICISM AND THE PilILOSOPHY O F MIND

the philosopher interested in perception say "let him who is interested colored. Later it was pointed out that if this is interpreted as the
in prescriptive discourse analyze its concepts and leave me in peace." claim that the sentence "Physical objects have colors" expresses an
Most if not all philosophically interesting concepts are caught up in empirical proposition which, though widely believed by common sense,
more than one dimension of discourse, and while the atomism of early has been shown by science to be false, then, of course, this c1aim is
analysis has a h ealthy successor in the contemporary stress on journey- absurd. T he idea that physical objects aren't colored can make sense
man tactics, the grand strategy of the philosophical enterprise is once only as the (misleading) expression of one aspect of a philosophical
again directed toward that articulated and integrated vision of man- critique of the very framework of physical objects located in Space and
in-the-universe-or, shall I say discourse"'<tbout-man-in-all-discourse- enduring through T ime. In short, "Physical objects aren't really colored"
which bas traditionally been its goal. makes sense only as a clumsy expression of the idea that there are no
But the moral I wish sp ecifically to draw is that no longer can one such things as the colored physical objects of the common-sense world,
smugly say "Let the person who is interested in scientific discourse where this is interpreted, not as an empirical proposition-like "There
analyze scientific discourse and let the person who is interested in are no nonhuman featherless bipeds"-witl1in the common-sense frame,
ordinary discourse analyze ordinary discourse." Let me not be misunder- but as the expression of a rejection (in some sense) of this very frame-
stood. I am not saying that in order to discern the logic-the polydimen- work itself, in favor of another built around different, if not unrelated,
sional logic-of ordinary discourse, it is necessary to make use of the categories. This rejection need not, of course, be a practical rejection.
results or the methods of the sciences. Nor even that, within limits, It need not, that is, carry with it a proposal to brain-wash existing popu-
such a division of labor is not a sound corollary of the journeyman's lations and train them to speak differently. And, of course, as Jong as
approach. M y point is rather that what we call the scientific enterprise the existing framework is used, it will be incorrect to say-otherwise
is the flowering of a dimension of discourse which already exists in th.an to make a philosophical point about the framework-that no object
what historians caU the "prescienti:fic stage," and that failure to under- is rea1ly colored, or is located in Space, or endures through Time. But,
stand this type of discourse "writ large"-in science-may lead, indeed, speaking as a p11iiosopl1er, I am quite prepared to say that the common-
has often led to a failure to appreciate its role in "ordinary usage," and, sense world of physical objects in Space and Time is unreal- that is,
as a result, to a failure to understand the full logic of even the most that there arc uo such things. Or, to put it less paracloxicalJy, that in
fundamental, the "simplest" empirical terms. the dimension of describing and explaining the world, science is the
41. Another point of equal importance. The procedures of philosophi- measure of a11 things, of what is tb at it is, and of what is not that
cal analysis as sucb may make no use of the methods or results of the it is not.
sciences. But familiarity with the trend of scientific thongh t is essential 43. There is a widespread impression that reflection on how we learn
to the appraisal of the framework categories of the common-sense pic- the language in which, in everyday life, we describe the world, leads to
ture of the world. For if the line of thought embodied in the preceding the conclusion that the categories of the common-sense picture of the
paragraphs is sound, if, that is to say, scientific discourse is but a con- world have, so to speak, an uncba11engeable authenticity. There are, of
tinuation of a dimension of discourse which has been present in human course, different conceptions of just what this fundamental categorial
discomse from the very beginning, then one would expect there to be framework is. For some it is sense contents and phenomenal relations
a sense in which the scientific picture of the world replaces t he common- between them; for others physical objects, persons, and processes in
sense picture; a sense in which the scientific account of "what there is" Space and Time. But whatever their points of difference, the philoso-
supersedes the descriptive ontology of everyday life. phers I have in mind arc united in the conviction that what is called
H erc one must be cautious. For there is a right way and a wrong the "ostcnsivc tic" hclwccn our fundamental clcscripl ivc vot'a lmbry ancl
way to make this point. Many years ago it used to be conficlcnlly said the WOrld rules Olli of court :IS utterly absurd :111 y 11ol io11 that lltcrc
that science 11as shown, for example, that physical objects arc11:1 reall y arc 110 such things as this rra111cwork talks ahoul.
302
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

An integral part of this conviction is what I shall call (in an extended have the background and motivation to keep track of it, but an em·
sense) the positivistic conception of science, the idea that the frame- ployment which is fundamentally a hobby divorced from the perplexi-
work of theoretical objects (molecules, electromagnetic fields, etc. ) and ties of the mainland. But, of course, this summing up won't quite do.
their relationships is, so to speak, an auxiliary framework. In its most For all philosophers would agree that no philosophy would be complete
explicit form, it is the idea that theoretical objects and propositions unless it resolved the perplexities which arise when one attempts to
concerning them are "calculational devices," the value and status of think through the relationship of the framework of modem science to
which consist in their systematizing and heuristic role with respect to ordinary discourse. My point, however, is not that any one would reject
confirmable generalizations formulated in tl1e framework of terms which the idea that this is a proper task for philosophy, but that, by approach-
enjoy a direct ostensive link with the world. One is tempted to put this ing the language in which the plain man describes and explains empiri-
by saying that according to these philosophers, the objects of ostensively cal fact with the presuppositions of givenness, they are led to a ''reso·
linked discourse behave as if and only as if they were bound up with lution" of these perplexities along the lines of what I have called the
or consisted of scientific entities. But, of course, these philosophers positivistic or peninsular conception of scientific discourse-a "resolu·
would hasten to point out (and rightly so) that tion" which, I believe, is not only superficial, but positively mistaken.
X behaves as if it consisted of Y's
makes sense only by contrast with X. Private Episodes: the Problem
X behaves as it does because it does consist of Y's 45. Let us now return, after a long absence, to the problem of how
whereas their contention is exactly that where the Y's are scientific the similarity among the experiences of seeing tliat an object over there
objects, no such contrast makes sense. is red, its looking to one that an object over tl1ere is red (when in
The point I am making is that as long as one thinks that there is a point of fact it is not red) and its looking to one as though there were
framework, whether of physical objects or of sense contents, the abso· a red object over there (when in fact there is 110thing over there at all)
lute authenticity of which is guaranteed by the fact that the learning is to be understood. Part of this similarity, we saw, consists in the
of this framework involves an "ostensive step," so long one will be fact that they all involve the idea-the proposition, if you please-that
tempted to think of the authority of theoretical discourse as entirely the object over there is red. But over and above this there is, of course,
derivative, that of a calculational auxiliary, an effective heuristic device. the aspect which many philosophers have attempted to clarify by the
It is one of my prime purposes, in the following sections, to convince notion of impressions or immediate experience.
the reader that this interpretation of the status of the scientific picture It was pointed out in Sections 21 ff. above that there are prima facie
of the world rests on two mistakes: ( 1) a misunderstanding (which I two ways in which facts of the form x merely looks red might be ex-
have already exposed ) of the ostensive element in the learning and plained, in addition to the kind of explanation which is based on
use of a language-the Myth of the Given; (2) a reification of the empirical generalizations relating the color of objects, the circumstances
metllodological distinction between theoretical and non-theoretical dis· in which they are seen, and the colors they look to have. These two
course into a substantive distinction between theoretical and non· ways are (a) the introduction of impressions or immediate experiences
theoretical existence. as theoretical entities; and (b} the discovery, on scrutinizing these situa-
44. One way of summing up what I have been saying above is by tions, that they contain impressions or immediate experiences as com-
saying tl1at there is a widespread impression abroad, aided and abetted ponents. I called attention to the paradoxical character of the first of
by a naive interpretation of concept formation, that philosophers of these alternatives, and refused, at that time, to take it seriously. But in
science deal with a mode of discourse which is, so to speak, a pcnins11lnr the meantime the second alemative, involying as it does the Myth of
offshoot from the mainland of ordinary discourse. TI1c study of scicn· the Given, has turned out to be no more satisfactory.
tific clisconrsc is conceived to be a worthy employment for those wl10 for, in the first place, how are these impressions to be described,
104 305
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

if not by using such words as "red" and "triangular.'' Yet, if my argu- sentence (e.g. "S has a toothache") of which it is logically true that
ment, to date, is sound, physical objects alone can be literally red and whereas anybody can use it to state a fa ct, only one person, namely S
triangular. Thus, in the cases I am considering, there is nothing to himself, can use it to make a report. But while this is a useful formula-
be red and triangular. It would seem to follow that "impression of a tion, it does not do justice to the supposedly episodic character of the
red triangle" could mean nothing more than "impression of the sort items in question. And _that this is the heart of the puzzle is shown
wl1icll is common to those experiences in which we either see that by the fact that many philosophers who would not deny that there are
something is red and triangular, or something merely looks red and short-tem1 hypothetical and mongrel hypothetical-categorical facts about
triangular or there merely looks to be a red and triangular object over behavior which others can ascribe to us on behavioral evidence, but
there." And if we can never characterize " impressions" intrinsically, but which only we can report, have found it to be logical nonsense to speak
only by what is logically a definite description, i.e., as the kind of entity of non-behavioral episodes of which this is true. Tims, it has been
wbicll is common to such situations, then we would scarcely seem to claimed by Ryle (17) that the very idea that there are such episodes
be any better off than if we maintained that talk about "impressions" is a category mistake, while others have argued that though there are
is a notational convenience, a code, for the language in which we speak such episodes, they cannot be characterized in intersubjective discourse,
of how things look and what there looks to be. learned as it is in a context of public objects and in the 'academy' of
And this line of thought is reinforced by the consideration that once one's linguistic peers; It is my purpose to argue that both these con-
we give up the idea that we begin our sojourn in this world with any- tentions are quite mistaken, and that not only are inner episodes not
even a vague, fragmentary, and undiscriminating-awareness of the category mistakes, they are quite "effable" in intersubjective discourse.
logical space of particulars, kinds, facts, and resemblances, and recognize And it is my purpose to show, positively, how this can be the case.
that even such "simple" concepts as those of colors are the fruit of a I am particularly concerned to make this point in connection with such
long process of publicly reinforced responses to public objects ( includ- inner episodes as sensations and feelings, in short, with what has-un-
ing verbal performances ) in public situations, we may well be puzzled · fortunately, I th ink-been called "immediate experience." For such an
as to how, even if there are such things as impressions or sensat ions, .account is necessary to round off this examination of the Myth of t he
we could come to know that there arc, and to know what sort of thing Given. But before I can come to grips with these topics, the way must
they are. For we now recognize tliat instead of coming to have a COJJ · be prepared by a discussion of inner episodes of quite another kind,
cept of sometlling because we have noticed t]1at sort of t11ing, to have namely thoughts.
the al)ility to notice a sort of t11ing is already to have the concept of XI. Thoughts: the Classical View
that sort of thing, and cannot account for it.
46. Recent empiricism has been of two minds about the .status of
Indeed, once we think t his line of reasoning through, we are struck
thoughts. On the one hand, it has resonated to the idea that insofar as
by the fact that if it is sound, we are faced not only with t he question
there are episodes wh ich are thoughts, they are verbal or linguistic epi-
" How could we come to have the idea of an 'impression' or 'sensation?' "
sodes. C learly, however, even if candid overt verbal behaviors by people
but by the question "How could we come to have the idea of somc-
who had learned a language were th ough ts, there are not nearly enough
thing's looking red to us, or," to get to the crux of the matter, "of
of them to account for all the cases in which it would be argued that
seeing that something is red?" In short, we are brought face to face
with the general problem of understanding how there can he i1111cr a person was thinking. Nor can we plausibly suppose that the remainder
episodes- episodes, that is, which somel10w combine privacy, in tl1:1t is accounted for by those inner episodes which are often very clumsily
each of us has privileged access to his own, with intcrs11bicc:tivity. i11 lumped together under the heading "verbal imagery."
that· cach of us can, in principle, know about the other's. W e 111igl11· try On the other hand, they have been tempted to suppose that the
to put th is more JinguistiraHy as the problem of how lla(;rc <::111 he 11 episodes which are referred to by verbs pertaining to thinking include

106 307
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND
all forms of "intelligent behavior," verbal as wen as nonverbal, and that so I shall develop distinctions which will later contribute to a resolu-
the "thought episodes" which are supposed to be manifested by these tion, in principle, of the puzzle of immediate experience. But before
behaviors are not rea1ly episodes at all, but rather hypothetical and T continue, let me hasten to add that it will turn out that the view
mongrel hypothetical-categorical facts about these and still other be- I am about to expound could, with equal appropriateness, be repre-
haviors. This, however, runs into the difficulty that whenever we try sented as a modified form of the view that thoughts are linguistic
to explain what we mean by calling a piece of nonhabitual behavior episodes.
intelligent, we seem to find it necessary to do so in terms of thinking.
The uncomfortable feeling will not be downed that the dispositional XII. Our Rylean Ancestors
account of thoughts in terms of intelligent behavior is covertly circular. 48. But, the reader may wen ask, in what sense can these episodes
47. Now the classical tradition claimed that there is a family of epi- be "inner" if they are not immediate experiences? and in what sense
sodes, neither overt verbal behavior nor verbal imagery, which are am they be '1inguistic" if they are neither overt linguistic performances,
thougl1ts, and that both overt verbal behavior and verbal imagery owe nor verbal imagery "in faro interno"? I am going to answer these and
their meaningfulness to the fact that they stand to these thoughts in the other questions I have been raising by making a myth of my own,
the unique relation of "expressing" them. These episodes are intro- or, to give it an air of u~to-date respectability, by writing a piece of
spectable. Indeed, it was usually believed that they could not occur science fiction-anthropological science fiction. Imagine a stage in pre-
without being known to occur. But this can be traced to a number history in which humans are limited to what I shall call a Rylean
of confusions, perhaps the most important of which was the idea that language, a language of which the fundamental descriptive vocabulary
thoughts belong in the same general category as sensations, images, speaks of public properties of public objects located in Space and
tickles, itches, etc. This mis-assimilation of th oughts to sensations and enduring through Time. Let me hasten to add that it is also R ylean
feelings was equally, as we saw in Sections 26 ff. above, a mis-assimilation in that although its basic resources are limited (how limited I shall be
of sensations and feelings to thoughts, and a falsification of both . The discussing in a moment ), its total expressive power is very great. For
assumption that if there are thought episodes, they must be immediate it makes subtle use not only of the elementary logical operations of
experiences is common both to those who propounded the classical conjunction, disjunction, negation, and quantification, but especially
view and to those who reject it, saying tbat they " find no such experi- of the subjunctive conditional. Furthermore, I shal1 suppose it to be
ences." If we purge the classical tradition of these confusions, it becomes characterized by the presence of the looser logical rc1ations typical of
the idea that to each of us belongs a stream of episodes, not themselves ordinary discourse which are referred to by philosophers under the head-
immediate experiences, to which we have privileged, but by no means ings "vagueness" and "open texture."
either invariable or infallible, access. These episodes can occur without I am beginning my myth in medias res with h umans who have already
being '!expressed" by overt verbal behavior, though verbal behavior is- mastered a Rylcan language, because the philosophical situation it is
in an important sense-their natural fruition. Again, we can "hear our· designed to clarify is one in which we are not puzzled by how people
selves think," but the verbal imagery which enables us to do this is acquire a language for referring to public properties of public objects,
no more the thinking itself than is the overt verbal behavior by which but are very puzzled indeed about how we learn to speak of inner
it is expressed and communicated to others. It is a mistake to suppose episodes and immediate experiences.
that we must be having verbal imagery- indeed, any imagery-when we 111ere are, I suppose, still some philosophers who are inclined to
"know what we are thinking" -in short, to suppose that "privileged l'hink that by allowing these mythical ancestors of ours the nsc ad
access" must be construed on a perceptual or quasi-perceptual model. 1il>itum of subjuncl'ive conditionals, we have, in effect, enabled them
Now, it is my purpose to defend such a revised classical analysis of lo say anything lhal we (~lll s:1y when wc speak of tlio11gl1ts, cxpcric11ccs
our common-sense conception of thoughts, and in the course of doing (seci ng, hearing, ctr.), n11<l i111111cdi:1tc cxpcric11c:cs. I douht r ha!' there
308 309
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

are many. In any case, the story I am telling is designed to show exactly 50. With the resources of semantical discourse, the language of our
how the idea that an intersubjective language must be Rylean rests on fictional ancestors has acquired a dimension which gives considerably
too simple a picture of the relation of intersubjective discourse to more plausibility to the claim that they are in a position to talk about
public objects. thoughts just as we are. For characteristic of thoughts is their inten-
49. The questions I am, in effect, raising are "What resources would tionality, reference, or .aboutness, and it is clear that semantical talk
have to be added to the Rylean language of these talking animals in about the meaning or reference of verbal expressions has the same
order that they might come to recognize each other and themselves as structure as mentalistic discourse concerning what thoughts are about.
animals that tl1ink, observe, and have feelings and sensations, as we It is therefore all the more tempting to suppose tl1at the intentionality
use these terms?" and "How could the addition of these resources be of thoug11ts can be tra-c ed to the application of semantical categories to
construed as reasonable?" In the first place, the language would have overt verbal performances, and to suggest a modified Ryle.an account
to be enriched with the fundamental resource;S of semantical discourse- according to which talk about so-called "thoughts" is shorthand for
that is to say, the resources necessary for making such characteristically h ypothetical and mongrel categorical-hypothetical statements about
semantical statements as "'Rot' means red," and "'Der Mond ist rund' overt verbal and nonverbal behavior, and that talk about the intention-
is true if and only if the moon is round." It is sometimes said, e.g., by ality of these "episodes" is correspondingly reducible to semantical talk
Carnap (6), that these resources can be constructed out of the vocabu· about the verbal components.
lary of formal logic, and that they would therefore already be contained, What is the alternative? Classically it has been the idea that not
in principle, in our Rylean language. I have criticized this idea in only are there overt verbal episodes which can be characterized in
another place (20) and shall not discuss it here. In any event, a semantical terms, but, over and above these, there are certain inner
decision on this point is not essential to the argument. episodes which are properly characterized by the traditional vocabulary
Let it be granted, then, that t hese mythical ancestors of ours are of intentionality. And, of course, the classical scheme includes the idea
able to characterize each other's verbal behavior in semantical terms; that semantical discourse about overt verbal performances is to be
that, in other words, they not only can talk about each other's predic· analyzed in terms of talk about the intentionality of the mental epi-
tions as causes and effects, and as indicators (with greater or less relia-
sodes which are "expressed" by these overt performances. M y immediate
bility) of other verbal and nonverbal states of affairs, but can also say
problem is to see if I can reconcile the classical idea of thoughts as
of these verbal productions that they mean thus and so, that they say
inner episodes which are neither overt behavior nor verbal imagery and
that such and such, that they are true, false, etc. And let me emphasize,
which are properly referred to in terms of the vocabulary of intention-
as was pointed out in Section 31 above, that to make a semantical state·
ality, with the idea that the categories of intentionality arc, at bottom,
ment about a verbal event is not a shorthand way of talking about its
semantical categories pertaining to overt verbal performances.*
causes and effects, although there is a sense of "imply" in which
semantical statements about verbal productions do imply information
XIII. Theories and Models
about the causes and effects of these productions. Thus, when I say
••'Es regnet' means it is raining," my statement "implies'" that the 51. But what might these episodes be? And, .iJ.1 terms of our science
causes and effects of utterances of "Es regnet" beyond the Rhine fiction, how might our ancestors have come to recognize their existence?
parallel the causes and effects of utterances of "It is raining" by myself The answer to these questions is surprisingly straightforward, once the
and other members of the English-speaking community. And if it didn't logical space of our discussion is enlarged to include a distinction, central
imply this, it couldn't perform its role. But this is not to say thnt to the philosophy of science, between the language of theory nnd the
semantical statements are definitional shorthand for statements ahout language of observation. Although this distinction is a familiar one, l
the causes and effects of verbal performances. • An earlier nl.tcmpt nlong lhc.~c lines i~ to he found iu ( 18) nn<l ( 19).

310 311
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

shall take a few paragraphs to highlight those aspects of the distinction sophisticated scientists today might and perhaps, on occasion, do pro-
which are of greatest relevance to our problem. ceed in true logistical style. I do, however, wish to emphasize two points:
Informally, to construct a theory is, in its most developed or sophisti- {I) The first is that the fundamental assumptions of a theory are
cated form, to postulate a domain of entities which behave in certain usually developed not by constructing uninterpreted calculi which
ways set down by the fundamental principles of the theory, and to might correlate in the desired manner with observational discourse, but
correlate-perhaps, in a certain sense to identify-complexes of these rather by attempting to fin<l a model, i.e. to describe a domain of
theoretical entities with certain non-theoretical objects or situations; familiar objects behaving in familiar ways such that we can see how
that is to say, with objects or situations which are either matters of the phenomena to be explained would arise if they consisted of this
observable fact or, in principle at least, describable in observational sort of thing. The essential thing about a model is that it is accom-
terms. This "correlation" or "identification" of theoretical with observa- panied, so. to speak, by a commentary which qualifies or limits-but not
tional states of affairs is a tentative one "until further notice," and precisely nor in all respects-the analogy between the familiar objects
amounts, so to speak, to erecting temporary bridges which permit the and the entities which are being introduced by the theory. It is the
passage from sentences in observational discourse to sentences in the descriptions of the fundamental ways in which the objects in the
theory, and vice versa. Thus, for example, in the kinetic theory of gases, model domain, thus qualified, behave, which, transferred to the theo-
empirical statements of the form "Gas g at such and such a place and retical entities, correspond to the postulates of the logistical picture
time has such and such a volume, pressure, and temperature" are cor- of theory construction.
related with theoretical statements specifying certain statistical measures (2) But even more important for our purposes is the fact that the
of populations of molecules. These temporary bridges are so set up that logistical picture of theory construction obscures the most important
inductively established laws pertaining to gases, formulated in the thing of all, namely that the process of devising "theoretical" explana-
language of observable fact, are correlated with derived propositions tions of observable phenomena did not spring full-blown from the
or theorems in the language of the theory, and that no proposition in head of modern science. In particular, it obscures the fact that not all
the theory is correlated with a f:alsified empirical generalization. Thus, common-sense inductive inferences are of the form
a good theory (at least of the type we are considering) "explains" estab- All observed A's have been B, therefore (probably) all A's are B.
lished empirical laws by deriving theoretical counterparts of these laws
from a small set of postulates relating to unobserved entities. or its statistical counterparts, and leads one mistakenly to suppose that
These remarks, of course, barely scratch the surface of the problem so-called "hypothetic-deductive" explanation is limited to the sophisti-
of the status of theories in scientific discourse. And no sooner have cated stages of science. 111e truth of the matter, as I shall shortly be
I made them, than I must hasten to qualify them-almost beyond illustrating, is that science is continuous with common sense, and the
recognition. For while this by now classical account .of the nature of ways in which the scientist seeks to explain empirical phenomena are
theories (one of the earlier formulations of which is due to Norman refinements of the ways in which plain men, however crudely and
Campbell (5), and which is to be bound more recently in the writings schematically, have attempted to understand their environment and
of Carnap (8), Reichenbach (15, 16), Hempel (10), and Braithwaite their fellow men since the dawn of intelligence. It is this point which
(3)) does throw light on the logical status of theories, it emphasizes I wish to stress at the present time, for I am going to argue that the
certain features at the expense of others. By speaking of the construc- distinction between theoretical and observational discourse is involved
tion of a theory as the elaboration of a postulate system which is tenta- in the logic of concepts pertaining to inner episodes. I say "involved
tively correlated with observational discourse, it gives a highly artificial in" for it would be paradoxical and, indeed, incorrect, to say that these
and unrealistic picture of what scientists have actually done in the concepts arc theoretical concepts.
process of constructing theories. J don't wish to deny that logically 52. Now I think it fair t·o say that some light l1as already hccn thrown
312 313
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND
on the expression "inner episodes"; for while it would indeed be a concerns the construction of new concepts. As a methodological thesis,
category mistake to suppose that the inflammability of a piece of wood it involves no commitment whatever concerning the logical analysis
is, so to speak, a hidden burning which becomes overt or manifest when of common-sense mentalistic discourse, nor does it involve a denial that
the wood is placed on the fire, not all the unobservable episodes we each of us has a privileged access to our state of mind, nor that these
suppose to go on in the world are the offspring of category mistakes. states of mind can properly be described in terms of such common-sense
Clearly it is by no means an illegitimate use of "in" -though it is a concepts as believing, wondering, doubting, intending, wishing, infer-
use which has its own logical grammar-to say, for example, that "in" ring, etc. If we permit ourselves to speak of this privileged access to
the air around us there are innumerable molecules which, in spite of our states of mind as "introspection," avoiding the implication that
the observable stodginess of the air, are participating in a veritable there is a "means" whereby we "see" what is going on "inside," as
turmoil of episodes. Clearly, the sense in which these episodes are "in" we see external circumstances by the eye, then we can say that Behavior·
the air is to be explicated in terms of the sense in which the air "is" a ism, as I shall use the term, does not deny that there is such a thing
population of molecules, and this, in turn, in terms of the logic of as introspection, nor that it is, on some topics, at least, quite reliable.
the relation between theoretical and observational discourse. The essential point about 'introspection' from the standpoint of Be-
I shall l1ave more to say on this topic in a moment. In the meantime, haviorism is that we introspect in terms of common sense mentalistic
let us return to our mythical ancestors. It will not surprise my readers concepts. And while the Behaviorist admits, as anyone must, that much
to learn that the second stage in the enrichment of their Rylean knowledge is embodied in common-sense mentalistic discourse, and
language is the addition of theoretical discourse. Thus we may suppose that still more can be gained in the future by formulating and testing
these language-using animals to elaborate, without methodological hypotheses in terms of them, and while he admits that it is perfectly
sophistication, crude, sketchy, and vague theories to explain why things legitimate to call such a psychology "scientific," he proposes, for his
which are similar in their observable properties differ in their causal own part, to make no more than a heuristic use of mentalistic discourse,
properties, and things which are similar in their causal properties differ and to construct his concepts "from scratch" in the course of develop-
in their observable properties. ing his own scientific account 0£ the observable behavior of human
organisms.
XIV. Methodological versus Philosophical Behaviorism 54. But while it is quite clear that scientific Behaviorism is not the
53. But we are approaching the time for the central episode .in our thesis that common-sense psychological concepts are analyzable into
myth. I want you to suppose that in this Neo-Rylean culture there concepts pertaining to overt behavior-a thesis which has been main-
now appears a genius-let us call him Jones-who is an unsung fore· tained by some philosophers and which may be called 'analytical' or
runner of the movement in psychology, once revolutionary, now com· 'philosophical' Behaviorism-it is often thought that Behaviorism is
monplace, known as Behaviorism. Let me emphasize that what I have committed to the idea that the concepts of a behavioristic psychology
in mind is Behaviorism as a methodological thesis, which I shall bo must be so analyzable, or, to put things right side up, that properly
concerned to formulate. For the central and guiding theme in the historl· introduced behavioristic concepts must be built by explicit definition-
cal complex known by this term has been a certain conception, or family in the broadest sense-from a basic vocabulary pertaining to overt be-
of conceptions, of how to go about building a science of psychology. havior. The Behaviorist would thus be saying "Whether or not the
Philosophers have sometimes supposed that Behaviorists are, as such, mentalistic concepts of everyday life are definable in terms of overt
committed to the idea that our ordinary menta)istic concepts are 11nnly:11• behavior, I shall ensure that this is true of the concepts that I shall
able in terms of overt behavior. But although behaviorism hns ofl'cm employ." And it must he confessed that many hchavioristically oriented
been characterized by a certain metaphysical bias, it is not n thoslN psychologists have believed themselves committ:ccl l«> lhis auslcrc pro-
about the analysis of existing psychological concepts, but one which gram of concept formation.
314 11~
 W iHrid Seliars EMPIRlCISM AND THE PHILOSOPHY OF MIND
Now I think it reasonable to say that, thus conceived, the behavioristic explaining the selected phenomena of which the theory is the theory;
program would be unduly restrictive. Certainly, nothing in the nature (b) their role as candidates for integration in what we have called the
of sound scientific procedure requires this self-denial. Physics, the "total picture." These roles are equally part of the logic, and h ence
methodological sopl1istication of which has so impressed-indeed, overly the "meaning," of theoretical terms. Thus, at any one time the terms
impressed-the other sciences, does not lay down a corresponding re- in a theory will carry with them as part of their logical force that which
striction on its concepts, nor has chemistry been built in terms o_f it is reasonable to envisage-whether schematically or determinately-
concepts explicitly definable in terms of the observable properties and as the manner of their integration. However, for the purposes of my
behavior of chemical substances. The point I am making should now argument, it will be useful to refer to these two roles as though it
be clear. The behavioristic requirement that all concepts should be were a matter of a distinction between what I shall call pure theoretical
introduced in terms of a basic vocabulary pertaining to overt behavior concepts, and hypotheses concerning the relation of these concepts to
is compatible with the idea that some behavioristic concepts are to concepts in other specialties. \\That we can say is that the less a scien-
be introduced as tlleoretical concepts. tist is in a position to conjecture about the way in which a certain
55. It is essential to note that the theoretical terms of a behavioristic theory can be e~"Pected to integrate with oth er specialities, the more
psychology are not only not defined in terms of overt behavior, they
the concepts of his theory approximate to the status of pure theoretical
are also not defined in terms of nerves, synapses, neural impulses, etc.,
concepts. To illustrate: We can imagine that Chemistry developed a
etc. A behavioristic theory of behavior is not, as such, a physiological
sophisticated and successful theory to explain chemical phenomena
explanation of behavior. The ability of a framework of theoretical con-
before either electrical or magnetic phenomena were noticed; and that
cepts and propositions successfully to explain behavioral phenomena
chemists developed as pure theoretical concepts, certain concepts which
is logically independent of the identification of these theoretical con·
cepts with concepts of neurophysiology. What is true-and this is a it later became reasonable to identify with concepts belonging to the
logical point-is that each special science dealing with some aspect of framework of electromagnetic theory.
the human organism operates within the frame of a certain regulative
ideal, the ideal of a coherent system in which the achievements of each XV. The Logic of Private Episodes: Thoughts
have an intelligible place. Thus, it is part of the Behaviorist's business 56. With these all too sketchy remarks on Methodological Behavior-
to keep an eye on the total picture of the human organism which is ism under our belts, let us return once again to our fictional ancestors.
beginning to emerge. And if the tendency to premature identification We are now in a position to characterize the original Rylean language
is held in check, there may be considerable heuristic value in specula· in which they described themselves and their fellows as not only a
tive attempts at integration; though, until recently, at least, neuro· behavioristic language, but a behavioristic language which is restricted
physiological speculations in behavior theory 11ave not been particularly to the non-theoretical vocabulary of a behavioristic psychology. Suppose,
fruitful. And wbile it is, I suppose, noncontroversial that when tho now, that in the attempt to account for the fact that his fellow mc11
total scientific picture of man and his behavior is in, it will involvo behave intelligently not only when their conduct is threaded on a siring
some identification of concepts in behavior theory with concepts per· of overt verbal episodes-that is to say, as we would put it, when I hey
taining to the functioning of anatomical structures, it shonld not be "think out loud"-but also when no detectable verbal output is present',
assumed that behavior theory is committed ab initio to a physiological Jones develops a theory according to which overt utterances arc hnt. the
identification of a11 its concepts,-that its concepts arc, so to speak, culmination of a process which begins with certain inner epi.5otks. J\ml
physiological from the start. let us suppose tlwt l1is model for these episodes which initiate the events
We have, in effect, hccn distinguishing between two dimensions of wh ich culminate in overt verbal behavior is tlwt of overt verl>al l>clwvior
the logic (or 'mcthodologic') of thcorcticnl terms: (a ) their role in itself. Tn other words, 11si11g tl1c /:111g11:1gc of tl1c moclcl, Ilic theory is lo
316 117
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

the effect that overt verbal behavior is the culmination of a process has introduced is a framework of "unobserved," "nonempirical" "inner"
which begins with "inner speech." episodes. For we can point out immediately that in these respects they
It is essential to bear in mind that what Jones means by "inner are no worse off than the particles and episodes of p hysical theory. For
speech" is not to be confused with verbal imagery. As a matter of fact, these episodes are " in" language-using animals as molecular impacts
Jones, like his fellows, does not as yet even have the concept of an are "in" gases, not as "ghosts" are in "mach in es." T h ey are "nonem-
image. pirical" in the sim ple sense that they are theoretical-not definable in
It is easy to see the general lines a Jonesean theory will take. Accord- observational terms. Nor does the fact that they are, as introduced, un-
ing to it the true cause of intelligent nonhabitual behavior is "inner observed entities imply that Jones could not have good reason for sup-
speech." Thus, even when a hungry person overtly says "Here is an posing them to exist. Their "purity" is not a metaphysical purity, but,
edible object" and proceeds to eat it, the true- theoretical-cause of so to speak, a methodological purity. As we have seen, the fact that
his eating, given his hunger, is not the overt utterance, but the "inner they are not introduced as physiological entities does not preclude the
utterance of this sentence." possibility that at a later methodological stage, they may, so to speak,
57. The first thing to note about the Jonesean theory is that, as built "turn out" to be such. 11ms, there are many who would say that it
on the model of speech episodes, it carries over to these inner episodes is already reasonable to suppose that these t11oughts are to be "identi-
the applicability of semantical categories. TilUs, just as Jones has, like fied" with complex events in the cerebral cortex functioning along the
his fellows, been speaking of overt utterances as meaning this or that, Jines of a calculating machine. Jones, of course, has no such idea.
or being about this or that, so he now speaks of these inner episodes (3) Although the theory postulates that overt discourse is the culmina-
as meaning this or that, or being about this or that. tion of a process which begins with "inner discourse," this should not
The second point to remember is that although Jones' theory in- be taken to mean that overt discourse stands to "inner discourse" as
volves a model, it is not identical with it. Like all theories formulated voluntary movements stand to intentions and motives. True, overt lin-
in terms of a model, it also includes a commentary on the model; a guistic events can be produced as means to ends. But serious errors
commentary which places more or less sharply drawn restrictions on creep into the interpretation of both language and thought if one
the analogy between the theoretical entities and the entities of the interprets the idea that overt linguistic episodes express t houghts, on
model. Thus, while bis theory talks of "inner speech," th e commentary the model of the use of an instrument. Thus, it should be noted t hat
hastens to add that, of course, the episodes in question are not the Jones' theory, as I have sketched it, is perfectly compatible with the
wagging of a hidden tongue, nor are any sounds produced by this "inner idea that the ability to have thoughts is acquired in the process of
speech." acquiring overt speech and that only after overt speech is well estab-
58. The general drift of my story should now be clear. I shall there- lished, can "inner speech" occur without its overt culmination.
fore proceed to make the essential points quite briefly: (4) Although the occurrence of overt speech episodes wh ich are
(1) What we must suppose Jones to have developed is the germ of characterizable in semantical terms is explained by the theory in terms
a theory which permits many different developments. We must not of thoughts which are also characterized in semantical tem1s, this docs
pin it down to any of the more sophisticated forms it takes in the not mean that the idea that overt speech "has meaning" is being
hands of cJassical philosophers. Thus, the theory need not be given a analyzed in terms of the intentionality of thoughts. It must not be
Socratic or Cartesian form, according to which this "inner speech" is a forgotten that the semantical diaractcrization of overt vcrha1 episodes
function of a separate substance; though primitive peoples may have had is tl1e primary use of semantical terms, and tlrnt overt li11g11is/'ic cvc111's
good reason to suppose that humans consist of two separate things. :1s semantically cliaractcrizcd arc the 1110Jc1 for f·hc i1111cr cpi.mc/cs i11I ro-
(2) Let us suppose Jones to have called these discursive cnt·il'ics duced by the theory.
thoughts. We can admit at once that the framework of thonglit·s he (5) One fina l point before we come t·o I he dc11011c111c11I of t lic lirst
118 11 ')
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

episode in the saga of Jones. It cannot be emphasized too much that reporting role of these concepts-the fact that each of us has a privileged
although these theoretical discursive episodes or thoughts are intro- access to his thoughts-constitutes a dimension of the use of these con-
duced as inner episodes-which is merely to repeat that they are intro- cepts which is built on and presupposes this intersubjective status. My
duced as theoretical episodes-they are not introduced as immediate myth has shown that the fact that language is essentially an intersubjec-
experiences. Let me remind the reader that Jones, like. his Neo-Rylean tive achievement, and is learned in intersubjective contexts-a fact
contemporaries, does not as yet have this concept. And even when he, rightly stressed in modern psychologies of language, thus by B. F. Skinner
and they, acquire it, by a process which will be the second episode in (21), and by certain philosophers, e.g. Carnap (7), Wittgenstein (22) -
my myth, it will only be the philosophers among them who will sup- is compatible with the "privacy" of "inner episodes." It also makes clear
pose that the inner episodes introduced for one theoretical purpose- that this privacy is not an "absolute privacy." For if it recognizes that
thoughts-must be a subset of immediate experiences, inner episodes these concepts have a reporting use in which one is not drawing in-
introduced for another theoretical purpose. ferences from behavioral evidence, it nevertheless insists that the fact
59. Here, then, is the denouement. I have suggested a number of times that overt behavior is evidence for these episodes is built into the very
that although it would be most misleading to say that concepts per- logic of these concepts, just as the fact that the observable behavior of
taining to thinking are theoretical concepts, yet their status might be gases is evidence for molecular episodes is built into the very logic of
illuminated by means of the contrast between theoretical and non- molecule talk.
theoretical discourse. We are now in a position to see exactly why
this is so. For once our fictitious ancestor, Jones, has developed the XVI. The Logic of Private Episodes: Impressions
theory that overt verbal behavior is the expression of thoughts, and 60. We are now ready for tl~e problem of the status of concepts per-
taught his compatriots to make use of the theory in interpreting each taining to immediate experience. The first step is to remind ourselves
other's behavior, it is but a short step to the use of this language in that among the inner episodes which belong to the framework of
self-description. 11rns, when Tom, watching Dick, has behavioral evi- tI10ugI1ts will be perceptions, that is to say, seeing that the table is
dence which warrants the use of the sentence (in the language of the brown, hearing that the piano is out of tune, etc. Until Jones intro-
theory) "Dick is thinking 'p' " (or "Dick is thinking that p" ), Dick, duced this framework, the only concepts our fictitious ancestors had
using the same behavioral evidence, can say1 in the language of the of perceptual episodes were those of overt verbal reports, made, for
theory, "I am thinking 'p'" (or "I am thinking that p.") And it now example, in the context of looking at an object in standard conditions.
turns out-need it have?-that Dick can be trained to give reasonably Seeing that something is the case is an inner episode in the J onesean
reliable self-descriptions, using the language of the theory, without theory which has as its model repo1ting on looking that something is
having to observe his overt behavior. Jones brings this about, roughly, tl1e case. It will be remembered from an earlier section that just as
by applauding utterances by Dick of "I am thinking that p" when the when I say that Dick reported that the table is green, I com!llit myself
behavioral evidence strongly supports the theoretical statement "Dick to the truth of what he reported, so to say of Dick that he saw that
is thinking that p"; and by frown ing on utterances of " I am thinking the table is green is, in part, to ascribe to Dick the idea 'this table is
that p," when the evidence does not support this theoretical statement. green' and to endorse this idea. The reader might refer back to Sections
Our ancestors begin to speak of the privileged access each of us hns 16 ff. for an elaboration of this point.
to his own thoughts. What began as a language with a purely t11coret-icnl With the enrichment of the originally Rylean framework to include
use I1as gai11ed a reporting role. inner perceptual episodes, J have established contact with my original
As I see it, this story helps us understand that concepts pc1taini11g formulation of the problem of inner experience (Sections 22 ff.). For
to snch inner episodes as thoughts arc prim:irily and cssc11tially _i11l cr· I can readily reconstruct in this fram ework my earlier account of the
snhjcc:tivc, :is intcrsubjcctivc ;1s the concept of n positron, :111d th:it the l:111g11:igc of ;1ppc::iri11g, hoth qualit;itive and cxistc11tinl. Let· us turn,
~20 32 1
 Wilfrid Se1Ia1s EMPIRICISM AND THE PHILOSOPHY OF MIND

therefore to the final chapter of our historical novel. By now our an- 61. A number of points can be made right away:
cestors speak a quite un-Rylean language. But it still contains no ref· (1) The entities introduced by the theory arc states of the perceiving
erence to such things as impressions, sensations, or feelings-in short, subject, not a class of particulars. It cannot he emphasized too strongly
to the items which philosophers lump t ogether under the heading "im· that the particulars of the common-sense world arc such things as books,
mediate experiences." It will be remembered that we had reached a pages, turnips, dogs, persons, noises, Bashes, etc., and the Space and
point at which, as far as we could see, the phrase "impression of Time-Kant's Undinge-in which they come to be. W hat is likely to
a red triangle" could only mean something like ''that state of a per- make us suppose that impressions are introduced as particulars is that,
ceiver-over and above the idea that there is a red and triangular as in the case of thoughts, this ur-theory is formulated in terms of a
physical object over there-which is common to those situations in model. This time the model is the idea of a domain of "inner replicas"
which which, when brought about in standard conditions, share the perceptible
characteristics of their physical source. It is important to see tl1at the
(a) he sees that the object over there is red and triangular;
model is the occurrence "in" perceivers of replicas, not of perccivings
(b) the object over there looks to him to be red and triangular;
of replicas. Thus, the model for an impression of a red triangle is a
(c) there looks to him to be a red and triangular physical object over red and triangular replica, not a seeing ot a red and triangular replica.
there."
The latter alternative would have the merit of recognizing that impres·
Our problem was that, on the one hand, it seemed absurd to say that sions are not particulars. But, by misunderstanding the role of models
impressions, for example, are theoretical entities, while, on the other, in the formulation of a theory, it mistakenly assumes that if the enti·
the interpretation of impressions as theoretical entities seemed to pro- ties of the model are particulars, the theoretical entities which are
vide the only hope of accounting for the positive content and explana- introduced by means of the model must themselves he particulars-
tory power that the idea that there are such entities appears to have, thus overlooking the role of the commentary. And by taking the model
and of enabling us to understand how we could have arrived at this to be seeing a red and triangular replica, it smuggles into the language
idea. The account I have just been giving of tlwughts suggests how of impressions the logic of the language of thoughts. For seeing is a
this apparent dilemma can be resolved.
cognitive episode which involves the framework of thoughts, and to
take it as the model is to give aid and comfort to the assimilation of
For we continue the myth by supposing that Jones develops, in crude
impressions to thoughts, and thoughts to impressions which, as I have
and sketchy form, of course, a th eory of sense perception. Jones' theory
already pointed out, is responsible for many of the confusions of the
does not have to be either well-articulated or precise in order to be
classical account of both thoughts and impressions.
the first effective step in the development of a mode of discourse which
(2) T he fact that impressions are theoretical entities enables us to
today, in the case of some sense-modalities at least, is extraordinarily
understand how they can be intrinsically characterized-that is to say,
subtle and complex. W e need, therefore, attribute to this mythical characterized by something more than a definite description, such as
theory only those minimal features which enable it to throw light on "entity of the kind which has as its standard cause looking at a red
the logic of our ordinary language about immediate experiences. From and triangular physical object in such and such circumstances" or
this standpoint it is sufficient to suppose that the hero of my myth "entity of t11e kind w11ich is common to the situations in which there
postulates a class of inner-theoretical-episodes which lie calls, say, looks to be a red and triangular physical object." For although the
impressions, and which arc the end results of the impingement of predicates of a theory owe their meaningfulness to the fact tha t they
physical objects and processes on various parts of the body, and, in are logically related to predicates which apply to the observable phc-
particular, t o follow up the specific form in wl1ich I have posed our nomemt which the theory explains,, f'hc predicates of a theory arc not
problem, the eye. shorthand for definite descriptions of properties in t·cn11s of Ihcsc
322 323
 \Viltrid Sellars EMPIRICISM AND THE PIIILOSOPHY OF MIND

observation predicates. When the kinetic theory of gases speaks of (4) It might be concluded from this last point that the concept of
molecules as having mass, the term "mass" is not the abbreviation of the impression of a red triangle is a "purely formal" concept, the con-
a delinite description of the form "the property which ..." Thus, cept of a "logical form" which can acquire a "content" only by means
"impression of a red triangle" does not simply mean "impression such of "ostensive definition." One can see why a philosopher might want
as is caused by red and triangular physical objects in standard condi- to say this, and why be might conclude that in so far as concepts per-
tions," though it is true-logically true-of impressions of red triangles taining to immediate experiences are intersubjcctive, they are "purely
that they are of that sort which is caused by red and triangular objects structural," the "content" of immediate experience being incommuni-
in standard conditions. cable. Yet this line of thought is but another expression of the Myth
(3) If the theory of impressions were developed in true logistical style, of the Given. For the theoretical concept of the impression of a red
we could say that the i11trinsic properties of impressions are "implicitly triangle would be no more and no less "without content" than any
defined" by the postulates of the theory, as we can say that the in- theoretical concept. And while, like these, it must belong to a frame-
trinsic properties of subatomic particles are "implicitly defined" by the work which is logically connected with the language of observable fact,
fundamental principles of subatomic theory. For this would be just the logical relation between a theoretical language and the language of
another way of saying that one knows t]1e meaning of a theoretical term observable fact has nothing to do with the epistemological fiction of
when one knows (a) how it is related to other theoretical terms, and an "ostensive definition."
(b) how the theoretical system as a whole is tied to the observation (5) The impressions of Jones' theory are, as was pointed out above,
language. But, as I have pointed out, our ur-behaviorist does not formu- states of the perceiver, rather than particulars. If we remind ourselves
Jate his theory in textbook style. He formulates it in terms of a model. that these states are not introduced as physiological states (see Section
Now the model entities are entities which do have intrinsic proper- 55), a number of interesting questions arise which tie in with the
ties. They are, for example, red and triangular wafers. It might there· reflections on the status of the scientific picture of the world (Sections
fore seem that the theory specifies the intrinsic characteristics of im- 39~4 above) but which, unfortunately, there is space only to adum-
pressions to be the familiar perceptible qualities of physical objects and brate. Thus, some philosoph ers have th ought it obvious that we can
processes. If this were so, of course, the theory would be ultimately expect that in the development of science it will become reasonable
incoherent, for it would attribute to impressions-which are clearly not to identify all the concepts of behavior theory with definable terms in
physical objects-characteristics which, if our argument to <late is sound, neurophysiological theory, and these, in turn, with definable terms in
only physical objects can have. Fortunately, this line of thought over- theoretical physics. It is important to realize that the second step of
looks what we have called the commentary on the model, wh ich quali- this prediction, at least, is either a truism or a mistake. It is a truism
fies, restricts and interprets the analogy between the familiar entities if it involves a tacit redefinition of "physical theory" to mean "theory
of the model and the theoretical entities which are being introduced. adequate to account for the observable behavior of any object (includ-
Thus, it would be a mistake to suppose that since the modcJ for the ing animals and persons) which has physical properties." While if
impression of a red triangle is a red and triangular wafer, the impression "physical theory" is taken in its ordinary sense of "theory adequate to
itself is a red and triangular wafer. 'Vhat can be said is that the im- explain the observable behavior of physical objects," it is, I believe,
pression of a red triangle is analogous, to an extent which is by 110 mistaken.
means neatly and tidily specified, to a red and triangular wafer. The To ask how impressions fit together with electromagnetic fields, for
essential feature of the analogy is that visual impressions stand to ouc example, is to ask a mistaken question . It is to mix the framework of
another in a system of ways of resembling and differing which is molar behavior theory with the framework of the micro-theory of
structurally similar to the ways in which the colors and shapc.11 of visible physical objects. The proper question is, rather, "What wonld cor·
objects rcscmhlc and differ. respond in a micro-theory of sentien t organisms to mofor concepts per-
324 325
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND

taining to impressions?" And it is, I believe, in answer to this question will include such introspectible inner episodes as its looking to one
that one would come upon the particulars which sense-datum theorists as tbougJ1 there were a red and triangular physical object over t11ere, as
profess to find (by analysis) in the common-sense universe of discourse well as overt behavior.) F inally he succeeds in training them to make
(cf. Section 23). F urthermore, I believe that in characterizing these a,. reporting use of this language. He trains them, that is, to say "I have
particulars, the micro-behaviorist would be led to say something like the impression of a red triangle" wh en, and only when, according to
the foJJowing: "It is such particulars wh ich (from the standpoint of the theory, they are indeed having the impression of a red triangle.
the theory) are being responded to by the organism wh en it looks to Once again the myth helps us to understand that concepts pertain-
a person as though there were a red and triangular physical object over ing to certain inner episodes-in this case impressions-can be primarily
there." It would, of course, be incorrect to say that, in the ordinary and essentially intersubjective, without being resolvable into overt be-
sense,. such a particular is red or triangular. \Vhat could b e said,* how- havioral symptoms, and that the reporting role of these concepts, their
ever, is that whereas in the common-sense picture physical objects are role in introspection, the fact that each of us has a privileged access
red and triangular but the impression "of" a red triangle is neither to his impressions, constitutes a dimension of these concepts which is
red nor triangular, in the framework of this micro-theory, the theo- built on and presupposes their role in intersubjective discourse. It also
retical counterparts of sentient organisms are Space-Time worms char- makes clear why the "privacy" of these episodes is not the "absolute
acterized by two kinds of variables: {a) variables which also characterize privacy" of the traditional puzzles. For, as in the case of thoughts, the
the theoretical counterparts of merely material objects; (b) variables fact that overt behavior is evidence for these. episodes is built into the
peculiar to sentient things; and that these latter variables are the very logic of these concepts as the fact that the observable behavior
counterparts in this new framework of the perceptible qualities of the of gases is evidence for molecular episodes is built into the very logic
physical objects of the common-sense framework. It is statements such of n1olecule talk.
as these which would be tl1e cash value of th_e idea that "physical objects Notice that what our "ancestors" have acquired under the guidance
aren' t really colored; colors exist only in the perceiver," and that "to of Jones is not "just another language" -a "notational convenience" or
see that the facing surface of a physical object is red and triangular is "code"-which merely enab1es them to say what they can already say in
to mistake a red and triangular sense content for a physical object with the language of qualitative and existential looking. 'I 'hey have acquired
a red and triangular facin~ side." Both these ideas clearly treat what another language, indeed, but it is one which, though it rests on a
is really a speculative philosophical critique (see Section 41 ) of the framework of discOlUSC about public objects in Space and Time, has an
common-sense framework of physical objects and the perception of autonomous logical structure, and contains an explanation of, not just a
physical objects in the light of an envisaged ideal scientific framework, code for, such facts as that there looks to me to be a red and triangular
as though it were a matter of distinctions w~ich can be drawn within physical object over there. And notice that while our "ancestors" came
the common-sense framework itself. to notice impressions, and the language of impressions embodies a "dis-
62. This brings me to the final chapter of my story. Let us suppose covery" that there are such things, the language of impressions was no
more tailored to fit antecedent noticings of these entities than the lan-
that as his final service to mankind before he vanishes without a t race,
guage of molecules was tailored to fit antecedent noticings of molecules.
Jones teaches his theory of perception to his fellows. As before in the
And the spirit of Jones is not yet dead . F or it is the particulars of
case of tJ10ugl1ts, they begin by using the language of impressions to
the micro-theory discussed in Section 61 ( 5) wh ich arc the solid core
draw theoretical conclusions from appropriate premises. (Notice that·
of the sense contents and sense fields of the scnsc-cfat11m 1·1tcorisl . 1•:11-
the evidence for theoretical statements in the language of impressions
visaging the general lines of that framework, even skcl'd1i11g some of
" For a discussion of some logical points r crtaining to lhis framework, Ilic render it·s regions, he l 1as tm1ght himself t·o play wit·h it· ( i11 his st 11dy) as :1
sl10111cl consult the essay, "The Concept o Emergent'<!," hy P:111l K Meehl 11rnl
Wilfri~I Scllari;, 0 11 pp. 239- 52 of this vol ume. report language. Uuforhmatcly, he mislornlt:s I he Irull1 of t ltcs.c con-
326 127
 Wilfrid Sellars EMPIRICISM AND THE PHILOSOPHY OF MIND
ceptions, and, with a modesty forgivable in any but a philosopher, 19. Sellars, Wilfrid. "A Semantical Solution of the Mind-Body Problem," Methodos,
confuses his own creative enrichment of the framework of empirical 5:45-84 (1953).
20. Sellars, Wilfrid. "Empiricism and Abstract Entities," in Paul A. Schlipp (ed.),
knowledge, with an analysis of knowledge as it was. He construes as Tl1e Philosopliy of Rudolf Carnap. Evanston (Ill.) : Library of Living Philoso-
data the particulars and arrays of particulars which he has come to be phers (forthcoming}. (Available in mimeograph form from the author. )
21. Skinner, B. F. "The Operational Analysis of Psychological Terms," Psycl1ological
able to observe, and believes them to be antecedent objects of knowl- Review, 52:270-77 (1945). Reprinted in H. Feig! and M. Br~dbeck (eds.),
edge which have somehow been in the framework from the beginning. Readings in the Philosophy of Science, pp. 585-94. New York: Applcton-
Century-Crofts, 1953.
It is in the very act of taking that he speaks of the given. 22. Wittgenstein, Ludwig. PI:Ulosophical Investigations. London: Macmillan, 1953.
63. I have used a myth to kill a myth-the Myth of the Given. But
is my myth really a myth? Or does the reader not recognize Jones as
Man himself in the middle of his journey from the grunts and groans
of the cave to the subtle and polydimensional discourse of the drawing
room, the laboratory, and the study, the language of Henry and William
James, of Einstein and of the philosophers who, in their efforts to break
out of discourse to an arclle beyond discourse, have provided the most
curious dimension of all.
RF.FERENCES
l. Ayer, A. J. Foundations of Empirical Knowlcd~e. London: Macmillan, 1940.
2. Ayer, A. J. "The Terminology of Sense Data, ' in Philosophical Essays, pp. 66-
104. London: Macmillan, 1954. Also in Mind, 54, 1945, pp. 289-312.
3. Ilraitliwaitc, R. B. Scientific Explanation. Cambridge: Cambridge Univ. Pr., 1953.
4. Ilroad, C. D. Scientific Thought. London: Kegan Paul, 1923.
5. Campbell, Norman. l'hysics: Tlie Elements. Cambridge: Cambridge Univ. Pr.,
1920.
6. Camap, Rudolf. Introduction to Semantics. Chicago: Univ. of Chicago Pr., l 942.
7. Camap, Rudolf. "Psycl10Jogie in Physikalischer Spracl1c,'' Erkenntnis, 3:107-42
(1933).
8. Carnap, Rudolf. "The In terpretation of Physics," in H. Feigl and M. Brodbeck
(eds.), Readings in the Philosopliy of Science, pp. 309-18. New York: Appleton-
Centuiy-Crofts, 1953. Tiiis selection consists of pp. 59-69 of his Foundations
of Logic and Matltematics. Chicago: Univ. of Cl1icago Pr., 1939.
9. CJ1isholm, Roderick. "11le Theory of Appearing," in Max Dlack (eel.), Pliilo·
sophical Analysis, pp. 102-18. Ithaca: Cornell Univ. Pr., 1950.
IO. Hempel, C. G. fundamentals of Concept Formation in Empirical Science. Chi·
cago: Univ. of Chicago Pr., 1952.
11. Linnell, John. "Berkeley's Critique of Abstract Ideas." A Ph.D. thesis submitted
to tl1e Graduate Faculty of the University of Minnesota, June 1954.
12. Paul, G. A. "Is there a Problem about Sense Data?" in Supplementary Volume
XV of the Aristotelian Society Proceedings. Also in A. C . N. Flew (ed.), Logic
and Language. New York: Philosophical Lib., 1951.
13. Price, H. H. Perception. London: Methuen, 1932.
14. Price, H. H. Tl1inking and Experience. London: Hutchinson's Univ. Lib., 1953.
1S. Reichenbach, H. Pliilosophic dcr Raum-Zeit-Lehre. Berlin: de Gmytcr, 1928.
16. Reichenbach, H. Experience and Prediction. Chicago: Univ. of Chicago l'r.,
1938.
17. Ryle, Gilbert. The Concept of Mimi. London: Hutchinson's Univ. I.ih., 1949.
18. Sellars, \Vilfrid. "Mind, Mca11ing :mcl Behavior," J>l1ilosopl1ic:rl Studies, 3:83-91
(l 952).
328 329