Ralph Barton Perry

Journal of Philosophy, Psychology, and Scientific Methods 7 (1910): 5-14

Transcribed into Hypertext by Andrew Chrucky, March 13, 1997.

I shall deal in the present paper with a problem that is sufficiently limited to justify the hope that it may be solved on its merits. I shall seek to discover whether a certain circumstance, which has never been disputed, does or does not constitute evidence for a theory that has been much disputed. The circumstance I shall call the ego-centric predicament, and the theory, ontological idealism. I shall not attempt to determine the truth of ontological idealism, except in so far as that theory is established by an appeal to the ego-centric predicament. Furthermore, I do not attribute either the theory or the argument, in the form in which I present them, to any individual philosopher. My statement is intended to contain propositions that approach as exactly as any proposition can to a theory and an argument that are among the commonplaces of philosophy. But since the attempt to state the theory has raised doubts in my mind as to the possibility of stating it at all, and since I have found an exact statement of the argument to be equivalent to its refutation, I cannot reasonably suppose that any one ever deliberately assented to such statements. Inexact discourse cannot be criticized until it has first been converted into definite propositions; and these can never, with any certainty, be identified with the original assertions. For this reason polemics directed against historical opinions are like to prove unconvincing and futile. I propose, then, to examine certain propositions which I have myself defined. But, at the same time, I hope that what I have to say will be recognized as having an important bearing on traditional issues.

What I mean by ontological idealism is best expressed by the proposition: Everything (T) is defined by the complex, I know T. For the purposes of this proposition the "I" is in no need of any definition beyond what it contains from its being the initial term in this complex. In order to make it plain that the term is generalized, I shall substitute ego, or E, for the pronoun. The term T is primarily distinguished from other terms only in that it has unlimited denotation; it refers to anything and everything. It is desirable that the operation or relation "know" should be freed from its narrower intellectualistic meaning; and it will, therefore, prove convenient to use the expression Rc, to mean any form of consciousness that relates to an object. Thus Rc may refer to thinking, remembering, willing, perceiving, or desiring. I am justified in denominating it as a relation, because in the theory and in the argument which I am examining it plays the part of the connecting link through which E and T form one complex. Ontological idealism is, then, a name for the proposition: (E)Rc(T) defines T.

It will be observed that the proposition asserts that the specific relation Rc obtains between E and T. Ontological idealism is not to be confused, therefore, with a theory which simply asserts that some relation to E is definitive of T. Such a theory might be offered on the following grounds. No item in the universe can escape being related to every other item in the universe. Therefore, since there is at least one E in the universe, no T can escape being related to it. But a term is defined by all of its relations, hence every T is defined by its relation to an E. But such a theory would be trivial, because it would attach no peculiar importance to the relationship singled out for special mention. On the same ground one could construct a theory to the effect that T is defined by its relation to the number 7, or to Washington's crossing the Delaware, or to the flower in the crannied wall. There is an interminable series of such ontologies, and if established on such grounds, idealism would be only one of infinitely many negligible alternatives.

Not only does ontological idealism assert the specific relation Rc, but it asserts that this relation defines T as T's other relations do not. In other words, "definition" is intended in a sense in which some, but not all, relations are definitive. Otherwise, the theory would again become trivial and negligible. No theory of relations can neglect the difference, for purposes of definition, between a relation like that of a moving body to the masses of surrounding bodies, and a relation like that between a man's fortunes and his horoscope (or, "that part of the ecliptic which is on the eastern horizon at the instant of his nativity"). If the latter type of relation were as definitive as the former, then there would be no ground for preferring astronomy to astrology, or an idealistic ontology to any one of a number of others. Thus, every T is in the same universe with the number 7. Expressing the relation "with" by the symbol Rw, we can construct an ontological proposition to the effect that (7)Rw(T) defines (T). If ontological idealism is to be distinguished from the infinitely many negligible propositions which may thus be asserted, it must be contended that Rc(E), is in some sense necessary to T, while Rw(7), Rw(8), etc., are not.

Any remaining doubt of this must be dispelled, when it is observed that the only ground on which it would be possible to assert the universal proposition, every (T)Rc(E), is the discovery of the necessity of the relationship in particular instances. For complete induction is evidently out of the question. Speaking generally, the assertion that a thing is definable by all of its relations, can never throw any light on the relations that it does in truth possess. For that purpose the thing must be regarded as defined by some relations only. Before, then, it can be shown that everything possesses the relation Rc, that relation must be regarded as peculiarly indispensable to it. Rc(E) must be shown to be necessary to T, as two dimensions are necessary to a plane, or hydrogen and oxygen to water.

I desire in the present investigation to leave out of consideration a rapidly growing doubt as to the possibility of any such branch of knowledge as ontology in the traditional sense. Thus it may well be that the failure of the materialistic ontology is due not so much to the special limitations of the concept matter, as to the impossibility of obtaining any concept that shall have the unlimited denotation and connotation attributed to being or reality. Indeed, I do not feel at all sure that the words "being" and " reality" mean anything in exact discourse. But I waive that general question for the sake of isolating a narrower issue.

Ontological idealism, then, is a theory to the effect that T necessarily stands in the relation Rc to an E, or that the relationship Rc(E) is indispensable to T. Now the attempt to prove this theory at once reveals a predicament that might otherwise escape notice. One must attempt to discover the precise nature of the modification of T by Rc(E); but one promptly encounters the fact that Rc(E) cannot be eliminated from one's field of study, because "I study," "I eliminate," "I think," "I observe," "I investigate," etc., are all cases of Rc(E). In short, Rc(E) is peculiarly ubiquitous. There can be no question concerning the fact; it owes its importance in the estimation of philosophers to its being one of the few facts to which philosophy itself originally called attention. Science has occasion to eliminate errors of judgment and relativities of sense, but has no occasion to eliminate consciousness altogether; and therefore has not discovered that it is impossible. We cannot then, disagree as to the fact, nor as to its peculiarly philosophical or epistemological significance. But we are still left in doubt as to what the fact proves with reference to the problem which revealed it. My contention is that it proves nothing; or rather that it proves only the impossibility of using a certain method to solve the problem. In other words, it is not an argument, but a methodological predicament. Let me further elaborate this predicament.

In order to discover if possible exactly how a T is modified by the relationship Rc(E), I look for instances of T out of this relationship, in order that I may compare them with instances of T in this relationship. But I can find no such instances, because "finding" is a variety of the very relationship that I am trying to eliminate. Hence I cannot make the comparison, nor get an answer to my original question by this means. But I cannot conclude that there are no such instances; indeed, I now know that I should not be able to discover them if there were.

Again, with a view to demonstrating the modification of T by Rc(E), I compare T before and after it has entered into this relationship with some E other than myself. But in making the comparison, I institute the relationship with myself, and so am unable to free T altogether from such relationships.

Again, within my own field of consciousness, I may attempt to define and subtract the cognitive relationship, in order to deal exclusively with the residuum. But after subtracting the cognitive relationship, I must still "deal with" the residuum; and "dealing with" is a variety of the very relationship which I sought to banish.

Finally, just in so far as I do actually succeed in eliminating every cognitive relationship, I am unable to observe the result. Thus if I close my eyes I cannot see what happens to the object; if I stop thinking I cannot think what happens to it; and so with every mode of knowledge. In thus eliminating all knowledge I do not experimentally eliminate the thing known, but only the possibility of knowing whether that thing is eliminated or not.

This, then, is what I mean by the ego-centric predicament. It is a predicament in which every investigator finds himself when he attempts to solve a certain problem. It proves only that it is impossible to deal with that problem in the manner that would be most simple and direct. To determine roughly whether a is a function of b, it is convenient to employ Mill's "Joint Method of Agreement and Difference," that is, to compare situations in which b is and is not present. But where b is "I know," it is evidently impossible to obtain a situation in which it is not present without destroying the conditions of observation. In other words, the problem of determining the modification of things by the knowing of them is a uniquely difficult problem. The investigator here labors under a peculiar embarrassment. But this fact affords no proper ground for any inference whatsoever concerning the true solution of the problem; hence it affords no argument for any theory in the matter, such as ontological idealism.

For the purpose of further illustration, and in order to suggest specific historical applications, let me consider several varieties of ontological idealism that gain illegitimate support from this predicament. The varieties which I propose here to examine are distinguished by the type of dependence on Rc(E) which is attributed to T. The creative theory asserts that E creates T; the formative theory asserts that E forms or organizes T; the identity theory asserts that E is T.

It is characteristic of creative idealism, the most naive variety of the theory, to dispense wholly with analysis of E, T, and Rc. In other words, the necessity of the relationship is not deduced from the nature of its elements. One is held to be justified in asserting it without any previous definition of thing or ego or consciousness. Thus one might assert that "esse est percipi," or that "die Welt ist meine Vorstellung" without express reference to the nature of "esse," "percipi," "die Welt," "meine," or "Vorstellung." There would remain as evidence for the assertion only the invariable agreement of the elements denoted by these words. One finds no "case" that is not perceived, no "Welt" that is not an ego's idea. But the method of agreement, unless tested by the method of difference, affords no test proof; especially when, as in this case, there is an accidental reason for the invariability of the agreement. To rely on or employ invariable agreement when unsupported by other evidence is to commit that elementary fallacy, of which post hoc ergo propter hoc is the most common case. It is unnecessary for me to urge that this fallacy has been not infrequently committed, and that it has served on the whole as the favorite means of beguiling innocent minds into the vestibule of subjectivistic philosophies. But the degree to which this fallacy is virtually involved in the more advanced reasoning of idealism, is not, I think, sufficiently recognized.

Let us consider, for example, what I have called the "formative" theory, reducible to the proposition, E forms T. This epistemology owes its chief claim to distinction to the fact that it starts from an analysis of T, and is therefore more rational than the creative theory. It is shown that every T involves the same group of ideas or categories, so that it is possible to define thing in general in terms of that group. More specifically, it is shown that everything involves such formal characters as shall enable it to stand in determinate relations with all other things. Since everything must belong to truth, and since truth is one and systematic, everything must possess the logical qualifications for membership in one universal system. Now it is evident that this is not as yet idealistic. And yet, for some reason, it is often regarded as equivalent to idealism, or as being only one inevitable step short of idealism. This is easily explained if we allow for the surreptitious or unconscious use of the ego-centric predicament. Thus, the categories may be introduced not as the conditions of being, but as the conditions of "experience," or consciousness of being. But this means that things are already construed as instances of the (E)Rc relationship. Doubtless whatever is necessary to things is necessary to the knowledge of them; so that one may regard ontological constraints as cognitive constants. But this proves that knowledge is a function of things, and not that things are a function of knowledge. The latter assertion, unless new evidence is introduced, is simply a petitio principii. That this fact should so easily escape notice is due, I think, to the presumption that since things are severally found in the (E)Rc relationship, that relationship must be necessary to them. It is easy to beg the question because whatever one makes the starting-point of the analysis is in fact "my object." Building the results of the analysis about this, the total system becomes a system of consciousness.

But there is another motive that contributes to a looseness of reasoning here. The categories are ideas rather than sensations; they are the fruit of analysis, detached from the empirical context by thought. They do not belong to the individuals of nature. But where, then, do they belong? Now most modern philosophers scarcely regard it as necessary to prove that categories, relations, and ideas, are essentially modes of thought; and in this they are aided and abetted by common sense. For, since the overthrow of scholasticism, philosophy and common sense alike have been habitually nominalistic. Empiricists and rationalists differ only in that with the former nominalism is given a skeptical emphasis, while with the latter it is given a constructive emphasis. But to adopt a nominalistic interpretation of theories, to regard them as acts of consciousness is to commit oneself forthwith to idealism. Are categories necessarily related to a knower, are they conditioned by the relationship Rc(E)? Here again we meet with the ego-centric predicament. It is impossible to find a relation without a comparison of terms, it is impossible to find a fundamental logical concept that is not conceived. Since I cannot find a category without knowing it in the manner required by categories, I can find no category that is not a mode of thought. But since this clearly has to do with the circumstances conditioning my investigation, it must be discounted in my conclusions concerning the thing investigated. If I allow it to create the slightest presumption one way or the other, and rely on that presumption in further inferences, my construction is vicious and ungrounded.

While I very much doubt if any idealistic theory is untainted by this error, it is possible to define a third variety of the theory in which the error is much less conspicuous. This variety I have called the "identity" theory, because it reduces to the proposition E is T. This theory also bases itself on an analysis of T, or of T so far as intelligible and true. It is held that every T is definable in terms of relations through which it is connected with every other T. So far formative and identical theories agree. But in the latter more attention is given to the implications of a relational definition. If T' is definable in terms of R(T2), then R(T2) must be internal to T', or (T')R(T2) must be identical with T'. At the same time, (T')R(T2) must be identical with T2. Furthermore, since T' and T2, are defined by (T')R(T2), and not (T')R(T2) by T' and T2, or, since (T')R(T2) is intelligible per se, while T' and T2 are intelligible only in terms of (T')R(T2), the latter must be held to be prior to the former, as their ground, source, or explanation. In other words, in order that being shall be definable, it must be construed as a whole which is both identical with its parts, and also prior to them. Now this conclusion may be regarded as equivalent to a reductio ad absurdum of the relational definition; in which case it is necessary to establish idealism on entirely different grounds.{1} But what some idealists regard as beneath reason, other idealists regard as the ideal of reason. It is conceded that the conception of a whole which is both prior to, and identical with, its parts will not hold of any whole of nature, such as mechanism or organism. Nor is it possible to define it abstractly, using symbols for terms and relations. If the attempt be made it will result only in such self-contradictions as T' is identical with (T')R(T2), or (T')R(T2) is prior to T'. Indeed, if it were possible to discover this type of whole and part relationship in nature or the realm of logic, it would be impossible to infer idealism from it.{2} But it is contended that there is a unique complex in which this relationship is directly and luminously exhibited, that complex being consciousness.{3}

The crucial and more neglected question yet remains, however. Why should it be asserted that the self, subject, or ego is both identical with each of its objects and also prior to them? Why should knowledge be construed as "self-representation," "self-externalization" or "self-positing"? Now the answer to this question lies, in part, I am convinced, in a certain readiness among philosophers to assert anything of consciousness. It would appear that there is no conception too paradoxical to be harbored there. The proposition, gold is gold, is redundant, and the proposition, blue is its own other is nonsense; but the propositions, I am I, and, the self is its own other, somehow pass for intelligible discourse. Similarly, while a planetary system which is identical with each planet and prior to them, is clearly a doubtful proposition, men nod their heads sagely when they hear of a self which can dispense with its own parts, and also be wholly present to each of them. So long as the self remains obscure and unanalyzed, loosely denoted by such terms as "I," "ego," or "subject," it will doubtless afford a refuge for logical lawlessness.

But apart from this general disposition to laxity and high-handedness, how are we to account for the assertion that the thing known and the knower, the T and the E, are identical? Unless there is ground for such an assertion the conception of a whole that is identical with each of its parts and prior to them, cannot be saved from its inherent self-contradictions. As a general conception it is not to be distinguished from the obsolescent notion of substance, or of a thing-essence which is all of its attributes, and yet none of them. That it does not share the hard fare of the latter notion is due to the supposition that it is saved by special revelation. Although in general it is absurd, we are supposed to be unable to deny it because of the discovery of an unmistakable case of it. The fact being so strange, we must overcome our prejudice against fiction. We should not be entitled to invent a universal or absolute knower, identical with its objects severally, and prior to all of them, unless we had evidence that a knower, and a knower alone, is capable of just that sort of relationship. Hence everything is staked on an examination of such instances of knowing as can be observed.

It is asserted that in any typical case of knowing (E)Rc(T), the knower (E) and the thing known (T) are identical. But if we mean by E and T the terms of this relationship, then they are clearly not identical; for their identity would destroy the relationship, and the operation would lose its complexity. If, on the other hand, I mean by E and T the complete or essential natures of the entities referred to, then they do not stand in the cognitive relation. Thus I may assert that E is really (E)Rc(T), and that T is also really (E)Rc(T), and that E and T are therefore really identical. But (E)Rc(T) does not stand in the relation Rc to (E)Rc(T). In other words, I know T does not know T. In any case, then, it is impossible to assert that the knower and the thing known are identical, where these are defined as the terms of one cognitive relationship.{4}

Now if it be so simple a matter to refute the assertion that in the complex (E)Rc(T), E is T, how are we to account for that assertion? Only, I think, through the characteristic confusion of mind created by the ego-centric predicament. The T of the complex (E)Rc(T) does, as a matter of fact, stand in the relation Rc to E. This cannot be denied, albeit it is a redundant proposition when affirmed. It is only necessary to proceed, loosely and as may suit one's convenience, to substitute (E)(Rc)(T) for T or thing qua known for thing , and one has accomplished the miracle of identifying a complex with one of its own elements. Then, the other element having been dealt with in the same manner, the two elements are made equal to an identical complex, and hence to each other. But the whole question of the extent to which (E)(Rc)T can be substituted for T, depends on a very precise knowledge of the bearing of this relationship on T. The original problem, What does (E)Rc(T) mean to T? has, in all this elaborate dialectic, only been prejudged and confused. And the solution offered is not only without a shred of evidence, but is charged with the support of a logical abortion.

I have not undertaken to do more than to isolate a species of dangerous reasoning that infests a certain region of philosophical inquiry. The question of the precise modification which a thing undergoes when it is known, is a proper problem; and the theory that that modification is profound, or even in some sense definitive, is a legitimate speculative alternative. But nothing whatsoever can be inferred from the mere ubiquity of that modification, from the mere fact that nothing can be found which is not thus modified. This self-evident fact simply defines the means that must be employed for the solution of the problem. We cannot employ a method which in other cases proved a convenient preliminary step, the empirical, denotative method of agreement and difference. There remains, however, the method which must eventually be employed in any exact investigation, the method of analysis. The mere fact that T is invariably found in a certain complex since it cannot be corrected by the method of difference, must be set aside, and not allowed to weigh in our calculations. But we may still have recourse to that analysis of all the elements of the complex, of T, E, and Rc, which would be required in any case before our conclusions could assume any high degree of exactness. Having discovered just what an ego is, just what a thing is, and just what it means for an ego to know a thing, we may hope to define precisely what transpires when a thing is known by an ego. And until these more elementary matters have been disposed of we shall do well to postpone an epistemological problem that is not only highly complicated but of crucial importance for the whole system of philosophical knowledge.

Return to Problems from Wilfrid Sellars


{1} I am thinking of Mr. Bradley in particular. For him the absolute is a means of dispensing altogether with relations and hence is not argued from the necessity of a consciousness that shall supply relations. Mr. Bradley's idealism ("We perceive, on reflection, that to be real, or even barely to exist, must be to fall within sentience." Appearance and Reality, p. 144) is, so far as I can see, either a pure assumption, or a loose and unwarranted inference from the ego-centric predicament. [Back]

{2} Thus Professor Royce's contention that the part is equal to the whole in an infinite system, would prove only that being is infinite, and not that it is in any sense conscious. His subjectivism is, so far as I can see, not proven at all. In The Conception of God and in certain more recent verbal utterances he would seem to be exploiting the ego-centric predicament. In The World and the Individual, he relies mainly on the contention that, since nothing in the universe can be strictly independent of anything else, objects cannot be independent of ideas. But, as I have endeavored to point out above, this would prove that the universe can be defined in terms of anything you choose. [Back]

{3} "The unity is at once the whole of which the individuals are parts, and also completely present in every individual." "It still remains true that it is that particular relation of which the only example known to us is consciousness." McTaggart, Studies in the Hegelian Cosmology, pp. 14, 19.

"There is need of a single term to describe a One which is not a system, and for this purpose the capitalized word Individual, as qualified by the indefinite article, answers as well as any other known to the writer, It will later appear that only a self can be, in this sense, an Individual . . . . An Individual, on the other hand, has an existence fundamental, logically prior, to that of the parts or of the members. It is not separate from them, but it is distinguishable from them. It is fundamental to the parts, whereas parts, though they are real, are not absolutely essential to it; it expresses itself in the parts, instead of being made up them." Calkins,The Persistent Problems of Philosophy, pp. 378-79.

I should regard this as a rather incautious statement of the argument; not untrue to Hegel, but so express in its recognition of the priority of the whole self over its several acts or objects as to be exposed to the charge of naive spiritualism. [Back]

{4} It is evident that such considerations as these would necessitate a revision of certain current notions of "self-consciousness." But I cannot follow up the suggestion here. [Back]

Return to Problems from Wilfrid Sellars