THE QUEST FOR THE IMMACULATE PERCEPTION1
How to be Sure of Yourself by Not Saying Anything
Seminar: Perception and Verification
Prof. T. M. Reed
University of Utah, April 16, 1968
"The pursuit of the incorrigible is one of the most venerable
bugbears in the history of philosophy," writes J. L. Austin as he opens
Section X of Sense and Sensibilia.(1964,) I quite agree, but not
always for the same reasons as Austin. In this paper, I propose to
supplement his criticism of the "bugbear" of incorrigibility as it
appears in A. J. Ayer's book, The Foundations of Empirical Knowledge.
(1964) More specifically, I will challenge Ayer's view that (a)
there can be propositions based upon immediate sense date – i.e. a
"language of sense-data"; and (b) that such propositions are capable of
delivering "'incorrigible' foundations of empirical knowledge."
Austin should be credited for scoring some telling criticisms of
Ayer's position (albeit, I find a few minor points of dispute with
Austin). However, since space is limited, and since conflict is far more
stimulating than concord, I will direct my major attention to Ayer.
Let us then roll out the target and look it over. To begin, I can
scarcely improve upon Austin's summary of what I will call the
''Foundations of Knowledge Doctrine:"
In a nutshell, the doctrine is that knowledge, 'empirical'
knowledge,.,. . has foundations. It is a structure of the
upper tiers of which are reached by inferences, and the
foundations are the data on which these inferences are based.
(So of course as it appears--there just have to be sense-data).
Now the trouble with inferences is that they may be mistaken;
whenever we take a step, we may put a foot wrong. Thus – so the
doctrine runs – the way to identify the upper tiers of the
structure of knowledge is to ask whether one might be mistaken,
whether there is something that one can doubt; if the
answer is Yes, then one is not in the basement. And conversely,
it will be characteristic of the data that in their case
no doubt is possible, no mistake can be made. So to find the
data, the foundations, look for the incorrigible. (105)
My principal complaint with this account is the suggestion that any
and all "foundations theorists" are in fact stuck with "the
incorrigible," when, as a matter of fact, many might get by with the
idea that the "foundations" are just the "least corrigible." (E. g.
Ouine, Popper). However, Ayer seems quite willing to be adorned with
this albatross, and so we'll let Austin's account above stand pretty much as given.
Strange to say, Ayer does not appear to accommodate us with as
concise and clear a statement of his own doctrine. The best that I can
do is to present the best samples that I could find, and to attempt to
extract therefrom, the nerve of the issue. I trust that the reader will
forgive this laboring, for it is most important that the point at issue
In our first excerpt, Ayer begins his recitation of the doctrine with
a detached, impersonal tone. But notice how, at the close, he confesses
It is held to be characteristic of an 'incorrigible'
proposition that it is completely verified by the
existence of the sense datum which it describes; and so it is
inferred that to doubt the truth of such a proposition is not
merely irrational but meaningless; for it is only significant to
doubt where there is a logical possibility of error. The
argument is, in short, that if one uses a sentence such as "this
is green" merely to designate a present sense-datum, then no
proposition is being asserted to the truth of which any further
evidence would be relevant. . . The propositions which such
sentences were supposed to express might be said to be
indubitable on the ground that it was not significant to say
that one doubted them in any other but a purely verbal sense.
From this it may be concluded that all that is involved in
the claim that there are indubitable or incorrigible empirical
propositions is that people do sometimes use sentences in the
way that the sentence "this is green" was used in my example, or
at any rate that they could use certain sentences in this way,
if they choose. And I do not see any reason to deny that this
is so. (My italics. A. J. Ayer, Foundations of Empirical
From this we might validly extract, at least, the following:
"Sense data can on some occasions, provide the bases for incorrigible
propositions." It will be the main task of this paper to invalidate the
concept of an "incorrigible sense-datum proposition." Ayer reiterates
his position in the following:
In the case of a sentence which refers to a sense-datum, it
is possible to be mistaken about the truth of the proposition
expressed, so long as one is not actually observing the relevant
fact. But there is no such possibility of error when the
sense-datum is actually being sensed. For in that case, the use
of the sentence is prescribed by a rule of the language; so that
to make an assertion that does not correspond to the fact is
either to tell a deliberate lie or else to make a verbal
We might then safely conclude that Ayer would accept the following:
"When a sense-datum is actually being sensed, it is not possible to be
mistaken about the truth of the proposition expressing this fact." This
in turn clearly points to the central target of my analysis: the belief
that sense-data can yield incorrigible propositions.
Throughout Ayer's book, we find (in one form or another) this crucial
pronouncement:2 "In the domain of sense data, whatever
appears, is real." (123) This might be interpreted to mean that of any
present sense-datum X, "X is just what it seems to be." Now what in fact
does this deny? Ayer takes great pains to show that this "usage" denies
that (a) "more is given to us than we are actually aware of," or (b)
that "only part of what we apprehend is really given." (117) Alternately
stated this usage denies (a) that "sense-data can have properties that
they do not appear to have" or (b) that some data "can appear to have properties that they do not
really have." (117) Ayer dismisses these interpretations, so as "to make
the distinction between sense-data and material things as sharp as
Now then, if we say "incorrigibly" of a present sense-datum X, "X is
just what it seems to be," what in fact are we saying? What else
is – or isn't – X? Well, it seems that we are immediately at a stop. To
say anything else, would be to link it up with something
not-given--and we can't do that without introducing contingent,
corrigible assertions. We are, remember, seeking an incorrigible proposition.
But can't we make a statement involving "facts," yet contained
within the field of awareness? Let's try it out with Ayer's own
example: "This is Green."
Of course, we can't ask "What is it that is green?" if by that we
wish to identify "this" with grass, or a tree, or whatever, for to do so
is to introduce "material objects" and with them "corrigibility." So
let's try asking "What do you mean 'this is green'?" What does
"green" mean? "Well, like what I sensed when. . ." Now hold on!
You can't leave the present, for we must not admit the
possibility of error, and "the possibility of my being mistaken, in what
is not merely a verbal sense, depends upon the fact that my judgment
goes beyond the evidence upon which it is immediately based. It connects
an experienced sense datum with other possible sense-data which are not
simultaneously given, and in doing so it allows room for doubt and
In other words, "green" in Ayer's "incorrigible" proposition "this is
green," must be as bound to the present instant as the "this." We simply
cannot bring in "green" like (some other) X" – not even "some other
And so, having no links with other times and other sense-data (i.e.
those not now present), "greenness" simply shrinks to equivalence with
"this." We must omit reference, predication, qualification, inference,
for all these must transcend the here-and-now. But without these, in
what sense have we language or meaning? And if we have sense experience
without meaning, what have we? What is left? W. M. Urban gives a
provocative response to this:
. . .something would be left if language were gone. But the
question is, just what would be left? Could we call it knowing?
Awareness. . .would still be there – the mere whatness of sense
datum and perhaps its otherness, although the latter is perhaps
questionable. But what is it that the intellect, in Bergson's
words, "will still affirm in implicit terms?" The only thing, I
feel sure, would be something like "that is that." In order that
anything else should be affirmed, language. . . would have to
come back. . .
What would be left if language were gone? Nothing that is
affirmed – even in "implicit terms." In other words, we 'have'
something, the sensation, but we 'know' nothing. . .Knowledge
does not arise until the intellect affirms, either explicitly
or implicitly, and such affirmation is impossible without
linguistic form. (342)
Be this as it may, will we still have "incorrigibility?" Of course!
If "this is green" is to mean nothing more than "that is that," or
simply "Behold!" period! – who can possibly deny it? No one can deny
this simply because there is nothing to deny. We deny judgments,
and surely this is no judgment – not unless, contra hypothesis, the predicate is
allowed to expand beyond the immediate, and thus open itself to
Well then, just what is the way out of this? Very easy. We simply do
not allow that our perception must be "immaculate." We must acknowledge
that there must be more to our data than immediately meet our senses. We
must accept that "more is given to us than we are actually aware of."
(117) In C. S. Peirce's terms, we must acknowledge that all "givens" are
signs – with "hooks" of meaning, association and inference that
reach beyond the immediacy of our here-and-now awarenesses. The
immediate "this," if it is to be an element of a proposition – that is,
be available to a meaningful language--must be spliced to a predicate
that refers beyond the immediate. That is to say, the predicate
must not be a simple repetition of the subject "this." A sentence, we
must remember, is molecular--a hybrid. It must employ concepts
derived from and comprehended in terms of other times and places. On the
other hand, sense-data, to be "incorrigible," must be now--immediate.
But if such "givens" are to be articulated in meaningful sentences, they
must be related to classes (of things, relations, ,qualities) not immediate. And if so related, the data just might turn out not to be the
sort of things they are predicated to be. Not necessarily merely because
of "a verbal error," but because they simply might be found not
to belong to the classes that they were said to belong to. You might be
convinced that "this is a green patch." But what if you look out the
window to find the grass and trees to be a radically different
color--say, what you thought was "red?" Impossible? I'm quite inclined
to agree. But inconceivable? Logically inconsistent? Hardly. Or try
this: formulate any so-called "sense-datum statement" that would
satisfy Ayer's "incorrigibility standards." Take care then to exclude
all reference or presuppositions, in any term, "to sense
data-which are not simultaneously given." (83) What then will you have?
Not a sentence in any worthy senze. And if not a sentence--then no
language. Thus the phrase "incorrigible sense-datum language" simply makes no sense!
In summary, then to the degree that sense data are "incorrigible,"
they are wholly unqualified to stand as "foundations of knowledge." The
same quality of "immediacy" that renders sense data "incorrigible"
forbids their connection with other sense data or class concepts to form
judgments and inferences or other meaningful sentence forms. Conversely,
to the degree that "sense data" are involved in meaningful (and non-tautologous)
sentences, they relate to classes and conditions beyond their immediate
manifestations, and are thus contingent and fallible.
Ayer's search among "sense data" for incorrigible foundations of
knowledge fails simply because there are no such discrete "building
blocks" underlying human knowledge. Nothing is "given" to human
awareness that is not infused with meaning. There are no "bits" of
untouched data to sort out and pile up, as one "builds" their experience
and knowledge. Rather, as Cassirer puts it:
Man lives in a symbolic universe. Language, myth, art and
religion are parts of this universe. They are the varied threads
which weave the symbolic net, the tangled web of human
experience. All human progress in thought and experience refines
upon and strengthens this net. No longer can man confront
reality immediately; he cannot see it, as it were, face to face.
Physical reality seems to recede in proportion as man's symbolic
activity advances. Instead of dealing with the things themselves
man is in a sense constantly conversing with himself. (43)
Midas-like, man transforms what he encounters, both in terms of his
past experience and his future anticipations – and as his past and
future are codified in the "symbolic universe" that surrounds him as he
"converses with himself."
The inalienable role of meaning to perception of sense-data
might be readily demonstrated. Consider the following figures:
What do you make of them? Give up? The first is a portrait of a
"vicious circle." The second depicts a soldier and his dog turning the corner of a building. Now I
defy you to "see" these figures now, just as you did a moment before.3
Meaning, I submit, is virtually inseparable from "sense-data." Alter the
meaning of the datum, and it's "appearance" changes. (Who, among your
friends, "looks" the way he did when you first met him? Recall how now
familiar places appeared when you first saw them, and how different they
It would seem, then, that we never encounter discrete bits of sensed
information, isolated for our phenomenological convenience, from past
and future entanglements. Quite the contrary, our percepts are entangled
in Cassirer's web of symbolism. We might also (and not inconsistently)
conceive of our percepts in Peirce's terms, as interpretants of the past
and hypotheses of the future. Thus interpreted, we see that our percepts
by themselves, are inherently incomplete and have meaning and
significance only as they are integrated into the process of living.4
If this be so, the "Foundations of Knowledge" doctrine is woefully
misguided and quite inadequate to the task of revealing the source and
nature of empirical knowledge.
The following comments just didn't seem to fit smoothly into the body
of my paper, but they might be of some use in provoking discussion. This
is a list of five other fundamental difficulties raised by critics of
the "foundations of knowledge doctrine":5
a) Just how are material object statements derived from sense
data statements without employing material object stigmas
to do so. (E.g.. "There is a table in the next room?" "This
floor is supported by beams?").6
b) How can a set of hypothetical statements (which, it is alleged by
some critics, sense data statements must be) add up to categorical
statements about material objects. [Hirst]
c) How can knowledge that is intrinsically private be
translated into knowledge of public objects?
d) Isn't "private language" in fact derived from the public
e) Even more, isn’t such a thing as a "private language"
impossible and incomprehensible. (Wittgenstein and
Malcolm are the most noteworthy advocates of this view).
If, upon examination, we find that these challenges (and mine) are
substantial, we might then wonder "how did anyone ever come to believe
in 'incorrigible sense data statements’ in the first place?" It would
seem that many were drawn astray by the dread of an "infinite regress."
Hard-core "basic statements" or "basic meanings" were taken as a panacea
against this trap.
Quinton summarizes the argument:
The traditional form of the doctrine of foundations holds
that there must be some intuitive belief it any beliefs are to
be justified at all. By an intuitive belief is meant one which
does not owe its truth or credibility to some other belief or
beliefs from which it can be inferred. For a belief to be
justified it is not enough for it to be accepted, let alone
merely entertained. There must also be good reasons for
accepting it. Certainly some beliefs are justifiable by
reference to others, but only if these other beliefs are
themselves established or well confirmed. If every belief was
dependent on others for its justification, no belief would be
justified at all, for in this case to justify any belief would
require the justification of an infinite series of beliefs. So
if any belief is to be justified, there must be a class a basic,
non-inferential beliefs to bring the regress of justification to
a halt. 
NOTES AND REFERENCES\
1. This marvelous pun is not, alas, my own,
I've forgotten just who is responsible for it.
2. Or better, "linguistic convention"--as Ayer
is so anxious that we not forget.
3. While at Northland College (1993-7) I showed
my classes a satellite photo of Lake Superior. I then noted that
many see a wolf’s head in this image. (See for yourself, dear
reader). Every student "saw" this immediately. I then challenged
them to perceive the image of the lake as they did, a few seconds
earlier, before they were told of the wolf’s head. Impossible!
4. I expand upon this aspect of Peirce’s
philosophy in Section III of my seminar paper, "Metaphysician, Heal Thyself! – A Dim View of Peirce’s Realism,"
5. Most of these are reviewed by Anthony
Quinton in its paper. "The Foundations of Knowledge" reprinted in
British Analytic Philosophy, Williams and Montefiore, Eds.
6. These points are carefully argued by R. J.
Hirst in his book. The Problems of Perception, (Allen & Unwin),
Chapters 2 & 3.
Austin, J. L., Sense & Sensibilia, New York, Oxford University
Ayer, A.J., The Foundations of Empirical Knowledge, London,
Cassirer, Ernst, An Essay on Man, New York, Anchor, 1954.
Hirst, R. J., The Problems of Perception, Allen and Unwin,.
Quinton, "The Foundations of Knowledge, British Analytical
Philosophy, Williams and Montefiore, eds.
Urban, W.M., Language and Reality, New York