The Positive and the Logical
Arthur F. Bentley
ONE is tempted to look upon the positive and the
logical somewhat as one looks upon the quick and the dead. Yet the issue is hardly that sharp. Viability has strange possibilities and varied forms, and must often be appraised with an eye directed as much towards the environment as towards the claimant organism. Stretching the application of the word `viable' to complexes of behavior such as the philosophies and theories of knowledge, we may ask: Is the combination of the positive and the logical in logical positivism viable? And is this logical positivism itself as it stands either positive or logical?
These are interesting questions and no doubt endlessly debatable, though I have no thought whatever of entering the lists of the disputants. I find, however, that a certain amount of light can be thrown upon them by a method as elementary and as humble as that which a high-school student is taught to employ in his first contacts with microscope and bug. We shall try such a method, jot down certain observations in our note-book, and study our reports as they come, with no pretense to deep or soulful understanding.
Logical positivism is an activity or behavior of certain men who call themselves logical positivists; it is their highly characteristic
(473) behavior in the region of their most active interest. The logical positivists are alive, and the viability of their theory is dependent upon their continuing life and activity and upon that of such successors as they may have. We may ask whether men of their manner of thought and action are on the way to enjoy successful and widening participation in the great human activities of knowing and of interpreting the known. We may ask whether their behaviors are of a type fitted to maintain themselves in the kinds of environments we may anticipate for their immediate future.
The material that I shall here examine is limited to five contributions to Philosophy of Science by Carnap, Feigl, and Blumberg. The inquiry itself is one phase of a wider inquiry undertaken to appraise the present status of the interpretation of science, for which the file of this magazine offered an appropriate assemblage of materials. Various strange situations were found, involving often astonishing abuses of words, even in papers purporting to deal with symbolic logic. Among these the procedures of Carnap and his associates present but a single illustration.
Taking our materials, i.e., the papers above described, just as they come, I apply two tests which I make as simple and as little exacting as possible in the hope that they will not exceed what any reasonable man will think may be required in any serious discussion of science and of knowledge.
Logical. If a procedure is to be regarded as logical, its advocate may not within the limits of any one essay declare a proposition to be fundamental, use it as such, and then later on casually introduce a protecting clause to the effect that the proposition itself may of course be defective or perhaps even false. Nor, when he makes an assertion, may he follow this later within the same essay by its contradictory, without explaining the transition.
Positive. If a procedure is to be regarded as positive, it must employ materials of investigation and adopt attitudes of inquiry comparable to those used by the successful sciences of our era —
(474) the ones we most generally recognize and describe as positive with respect to the form of the knowledge they offer. It may not merely steal the use of a popular name. It must, in particular, exclude the use of all fairies, spooks, and spirits, whether fair or foul.
Lethal. Where a construction shows itself defective by both tests, and where, moreover, the defects of the first type can be directly traced to improper procedures of the second type, we may fairly regard the whole construction as an inadequate form of human behavior, lethal in its type, and doomed to disappear.
It may as well be stated at once that the evil spirit we shall find at work in the selected papers before us is one that manifests itself in the use of the word `concept.' The `concept' as our logical positivists employ it proclaims a reserved right on their part to "really mean" something different from what they say. Their `concepts' may thus be taken as marking gaps, or holes, in their language—places where words fail and where the expounder waves a magic wand to assure us nevertheless that all is well.
As to the status more generally of this word `concept,' we may remind ourselves that it derives from the old language of `consciousness' and `mind,' and that all formal definitions of it are in terms of some `mental' that is a `not-physical.' Its peculiarity is that it is in common daily use by men who have long since abandoned all belief in any such `mental'; they take it for granted, apparently without once asking themselves what expressly it is that they are talking about. They face some kind of a `factual' situation; they find the word `concept' not only convenient but eminently respectable; they proceed to use it. The result is that `concept' becomes a weasel word in philosophy's service far more efficient in deception than any word the amateurs of the politicial party platforms have evolved. One can find it with as many as half a dozen distinguishable applications in a single paper. While indicating sometimes a "bit of consciousness," or "a thought," or an assumption of "precise meaning," at other times it can be directly replaced in the text by the word `word' itself. It may offer any desired combination of, or stage of transition between, `mind' and `word.' It may cover sentence, proposition, theory,
(475) or doctrine. The part it plays for logical positivists is therefore not unique, but merely symptomatic of a disease of language, wide-spread in our era.
The purpose of Philosophy of Science as announced by the editor in his introductory statement (I, i) was to aid in the clarification of philosophy and science, and perhaps in their unification through such typical projects as "the examination of fundamental concepts and presuppositions in the light of the positive results of science, systematic doubt of the positive results, and a thoroughgoing analysis and critique of logic and of language." Upon all of these questions Carnap and his associates have had much to say, and Carnap's paper "On the Character of Philosophic Problems" was well indicated for the leading place in the first issue. It was Carnap's emphatic assertion, his main thesis, that by "purely formal procedure" the logic of science can "finally arrive at the answering" of all connotative questions (I, 9). This "connotative" of Carnap's (in the German, inhaltliche) covers all issues of what we most generally describe as `meanings,' since he places the inhaltliche in sharp opposition to the formale, and expressly defines the latter as covering the study of symbols and propositions "without reference to the meaning" (I, 9). In plain everyday language Carnap asserts that all questions of meaning can be answered without reference to meaning; or, in more erudite language, that a symbolic procedure stripped of all implications of meaning will be humanity's final resort for solving all problems of the import of knowledge.
Remember now that we are to examine only the materials immediately on the table before us. If we do this in a highly "formal" way, Carnap certainly should not be the man to object. It does not matter to us what Carnap writes elsewhere nor how he may modify his construction from year to year; our only concern is the integrity of the particular specimen of work before us. It is as if the boy in the laboratory studying the bug should dis-
(476) -regard all speculation as to its place in Nature's Mighty Scheme, and confine himself to such simple observations as how well its wings are fastened to its body, and whether its alimentary canal is open at both ends.
We begin by applying our first simple test which we labelled with the word `logical.' And here we can follow a lead already taken in the pages of the magazine by John Dewey. Dewey read Carnap's paper and at once asked him certain pointed questions which, to anyone interested in the simpler "facts of life," have devastating import (I, 237). Carnap gave no sign of attention either to facts or import, but instead spun the web of his reply (I, 359) in a manner much nearer the medieval than the modern. Since questions and answers with accompanying page citations are directly available to all readers who have this magazine in hand, I may be permitted to paraphrase and condense.
Dewey first cited Carnap's assertion that by "purely formal procedure" the logic of science can "finally arrive" at the answering of all inhaltliche questions. He then set side by side with it three of Carnap's further statements as follows: mathematical meanings derive from empirical considerations; statements of fact are synthetic (that is, neither analytic nor contradictory, but involving definite truth or falsity); the very selection of a linguistic form on which to rely (such as formale or inhaltliche itself) requires an empirical decision.
Dewey was very gentle and considerate in his comment. He merely said that Carnap's position "is not clear to me beyond a certain point." But he proceeded to show that among Carnap's tabulated illustrations, seventeen in number, the formale expressions were apt to be the ones with precision of `meaning,' while their inhaltliche alternatives were too often either meaningless or just vaguely ambitious. And beyond that he was surprised to find Carnap at the end of his paper, after having built entirely on a severance of formale from inhaltliche, calmly remarking that
(477) there was nothing fundamental at all about his distinction between the two approaches.
Carnap in reply admitted a lack of clarity in his fifteen-page essay. He said this was because "concepts" were necessary which were far too technical for use in any such place. Even in personally addressing Dewey he was compelled to fall back upon descriptions which he admitted to be "somewhat inexact." He "distinguished" three sorts of sentences—"real object-sentences," "pseudo-object-sentences," and "syntactical sentences"; the first of which deal with "extra-lingual objects" (elsewhere called "empirical matter"), the second "deal seemingly" with them (although actually with "the lingual signs of those objects"), while the third "deal with lingual expressions." He assured Dewey that the whole distinction between formale and inhaltliche concerned only those sentences "which deal with language," as though that were extenuation enough for factual falsity.
We have here an authoritative sample of Carnap's "logical analysis. The question is whether it is entitled to be called 'positive'? More particularly we may ask: Does the array of sorts of sentences which Carnap "distinguishes" offer us a positive classification of linguistic materials in any sound sense of the word `positive'? How far can we outline `operations' or establish `facts' to correspond? How far can we so much as attempt this?
The first difficulty we find lies in the criteria of classification, one sort of sentence being identified with respect to language, another with respect to what is not-language, and a third with respect to what seems to be not-language. We have here 'sentences,' 'objects,' and 'seemings' used for tests in classifying sentences themselves, and we may expect much confusion in the search for applicable `operations.'
Next to attract attention is the peculiarity of the hyphenation. Order the groups as follows: syntactical sentences, real object-sentences, pseudo-object-sentences. Observe this closely. You
(478) will find that it would take several pages to bring out all the issues the hyphens evade. The "syntactical" is not balanced against the "object" in a first stage of classification, nor the "real" against the "pseudo" in a second stage; it is not even clear which stage has precedence; no `genus' can be identified for the three `species.' The hyphens are disingenuous; they are Ethiopians—'conceptual' Ethiopians, so to speak—in the linguistic wood-pile. They mark, not classification, but casuistry.
So much for the "syntax" of Carnap's procedure. Inspect it now rather for its "content." Не fails to show that his test, i.e., what the sentences are `about,' is dominant for his problem; other important possibilities for ordering are available, but we may pass this over. Most troublesome is the fact that he posits his "real object-sentences" and his "syntactical sentences" in such conceptual purity that it would be difficult for him (personally I would say impossible) ever to exhibit a specimen of either. All the interest of inquiry lies in the intermediate territory, and here the procedure is deliberately that of facing both ways. The reader will note (a) that instead of clearing the presentations from the difficulties of the `seeming' and the `actual,' this complexity is expressly maintained; (b) that the phrase "material mode of speech" is here applied because of its verbal suggestion of empirical matter on the one side and of formal mode on the other; and (c) that the parallelism of "being a thing" with "being a thing-sign" is another bit of prestidigitation between matter and mode. In effect Carnap arranges a logical key-board to be played by both hands in the pious wish that in time the right hand may come to `know' what the left hand is doing. Right hand plays language-language with a permitted shift from a formal to a material mode. Left hand plays thing-language with a set of pseudo-thing variations. The organ stop marked "parallel syntactical quality," when pulled out at the right moment by the right man, is supposed to make "logical analysis" become "positive."
One further conflict in Carnap's statements may be noted,
(479) though its nature is much like what has preceded. In connection with his assertion that all science is language in the sense of a system of propositions (I, 8, 14), he maintains that language is a "calculus" and that its formal rules of procedure are "mechanical operations" (I, 10). Then later he remarks that "mere calculating about" is not enough (I, 19). Without the first attitude he could not set up his main thesis, nor undertake his proposed sentence-classification. Without the second he could get nowhere after setting his thesis up. The status is this: A certain "parallelism" of ancient lineage is disclaimed and expelled. Later, however, hobbled and veiled, it is admitted through the back door. We are told that this does not matter, for now it is fully under linguistic control. "The distinguishing of the formal and the material modes of speech concerns only those sentences which deal with language" (I, 360). There it is justified, there it rules, there it characterizes "logical analysis"; otherwise, incidentally, it is false.
Carnap, of course, himself sees unity and coherence in all this; and we must seek the locus of his confidence. But first observe how powerful that confidence is. Dewey had asked him a further question, one concerning the distinction he had set up between "assertions" and "proposals." He could not answer this question, but he assured Dewey nevertheless that the distinction was sound, and that his criterion could be applied, "true or false," to any sentence, "provided of course the sentence Si as well as the whole connection to which it belongs is given in a perfectly correct and complete form" (I, 3ßo). Here is a requirement which a disciple may no doubt accept, but which to the ordinary inquirer midway in the passage of our lives may perhaps seem an ideal just a shade too bright and good. Nevertheless it exhibits Carnap's reliance, his typical reliance, in all his work.
Proceed, Carnap tells us in effect, with an infinite series of regressive "meanings," being very careful at every stage to strip yourself of factual or experiential debasement; make every "meaning" a "symbol," and purify the symbol so thoroughly of
(480) all meaning that you become sure it has "no meaning at all"; then in the end you will secure final and complete control of all "meanings" and of all knowledge. He casts this "proposal," this device, up against the logical sky in the form of a huge calculating machine. He does not, however, assert it to be independent and absolute perpetual motion. It has to be produced in some way, and it evidently has some kind of an operator. What operator? His operator is in plain view, even though its head is buried in the verbal sands. It is what in mathematical analogy we may call the "point-soul," or in physical analogy the "particle-mind." It is that from which everything starts and to which everything comes back, the limit of all possible statement. From it arise the verbal flights through which knowledge is to be acquired and appraised. In it lie the attributes of truth on which Carnap relies even when all his words and sentences prove false. It alone is the authority whereby the logician may dare to soar above all those situations which currently we regard as experientially factual or as approximations to fact.
This operator of Carnap's possesses a tool—a magic tool—the `concept.' The 'concept' does everything the operator wants it to do. When he wishes, the concept garbs itself as a `word'; then the system deals with `language.' When he wishes otherwise, then the concept transforms itself into `meaning,' and the system runs far away from the `language' of its primary confession of faith. By means of the `concept,' itself a word, every word is empowered to take shifting values—concept, meaning, symbol, proposition, language, science, even "no meaning at all." The intention of the `concept' is pure, and everything must come home in the end to roost on the wrist of its master, the point-soul; then why bother with minor temporary inconsistencies, or tests, or practical `operations' en route?
(481) Carnap and his followers take advantage, nevertheless, of whatever opportunity opens to associate their procedure with Bridgman's operations; at times they draw support from him, at others they allot him approval as an earnest seeker, though of the lower caste of science. Bridgman makes 'concept' his key-word and brings in a credo of the solipsistic self like a refrain to his writing. The difference between Bridgman and the logical positivists is, nevertheless, almost antipodal. What Bridgman does with concept and self is to bring them under harsh control by physical operations of the most advanced laboratory type. What the logical positivist does with them is to run to them for help whenever he gets into trouble.
Point-souls of the type Carnap uses are largely discredited in more recent psychological and sociological inquiry, and they are wholly discredited by the phenomenon of Gestalt; but they are so deeply embedded in the German language with its terminologies of Geist and Seele, Vernunft and Verstand, that linguistically they dominate there beyond all suggestion of observational search or test. Even the experimental work of Gestalt is most often construed in one or other of their forms; even the Marxian is still bound. Carnap is of their inner faith. The last stronghold of such point-souls is in logic, and there the issue will be fought out. What concerns us of course is no question of `reality,' whether such souls actually `exist' or not being of no direct importance. We are concerned strictly with the phenomena before us, namely, those linguistic behaviors in which point-souls whether by direct assertion or implicit use are prominent factors.
We have one more astonishing observation to make with regard
(482) to this point-soul of Carnap's as it actually performs. As itself concept and bearer of concepts, and as sole point of ultimate reference, it refuses to remain a point. It splits. It becomes a two-faculty soul of a type so ancient and elsewhere so discredited that it is astounding to find it surviving and strongly entrenched in a recent logic. The proposition about the "two languages" which the logical positivists make and all their distinctions of data and constructs arise in this way. Our own discussion has covered this situation in a different linguistic form, and it will be enough here to give footnote reference to other comments upon it in the pages of Philosophy of Science.
If we take Carnap's formale and inhaltliche out of the rigid 'conceptual' forms in which he propounds and contrasts them, and if we examine them in everyday life where although Carnap recognizes their union he has no `logical' use for it, we find our separations of them all temporary and provisional, and the return always necessary to their combined behavioral process. We forever strive to formalize our expressions as fully as we can, and to succeed we must limit. But also we strive forever to give our expression as rich, as complete, as factual a reference as we can. And here each success means a new destruction of the old `conceptual' identifications; it means the widening and organizing and systematizing of our knowledge. Carnap has every right to make his split, and to make it as sharp and rigid as he can. But this is the right to postulate and to test. He cannot expect to convince the world by fiat or by spiritist appeal or by glowing picture of ulti-
(483) -mate ideal success. He must work within the stream of knowledge and within that portion of the stream within which we find ourselves today. It is this which can and will guarantee factual orientation on the one hand and heightened consistency of expression on the other—the two hand in hand.
The papers of Feigl and Blumberg are strikingly similar to Carnap in the characteristics we have been considering. Feigl's logical positivism results in verbal anarchy. Blumberg surpasses him by expressly recognizing that such is the outcome, while at the same time holding as firmly as ever to the basic principles of the creed.
Feigl proposes to exhibit "the strict identity of the `mental' life with certain processes in the `physical' world" and to show that this is "not a matter of belief. . . but a truth capable of logical demonstration" (I, 420). His confidence is so great that he declares (I, 444) that if his analysis leaves any riddle unsolved "then it is an insoluble one that cannot be expressed as a significant question in any logically legitimate language." His procedure, he says, will be by way of "a metaphysically neutral logical analysis of fundamental concepts" (I, 420). He uses an "epistemological subject"; he declares this to be a necessity for logic since it is the only source of answer to the question: "How do we justify our factual assertions?" (I, 428); nevertheless he protests vehemently all the time that it is a fiction. He disclaims all dogma that makes such a subject "individual," but he says that in use it inevitably becomes "individual," and that in no other form is it possible. If this is "solipsistic," says Feigl, it is "only in the methodological sense, which is metaphysically strictly neutral" (I, 429). In other words Feigl exorcises metaphysics, but goes right ahead with "fundamental concepts," "epistemological subjects," isolated "individuals," and whatever else he wishes, all under the name of "logic" and of "method." He gets rid of some of the spoiled chicks of metaphysics, and makes a great virtue of it,
(484) but he builds a gilded temple over the metaphysical incubator, and proclaims a priestly rule in the name of the logical and of the positive.
In various other ways the dangers of the `conceptual' procedure are demonstrated in Feigl's pages. He begins, for example, by declaring "perceptions, thoughts, desires, dreams" to be "empirical existences"—data that all trained people can agree upon under the very same tests that are applied to "tables, trees, people, stars...." etc (I, 423). He winds up by turning the "physical" and the "mental" into two separate "symbolic systems," "universes of discourse, characterized by their logical syntax" (I, 442). In the course of his development for these "two languages," both under the aegis of a "language about languages," he arrives at the following sentences which we may cite without further comment:
"The physical language is retranslatable into the language of data" (I, 430).
"The physical language is not only translatable into the language of data but also vice versa" (I, 435).
"Mutual translatability means nothing but identity of structure" (I, 436).
"The two languages have incompatible syntaxes. . . they are structurally completely disparate" (I, 439).
Blumberg uses the title "Philosophic Analysis" for his form of logical positivism. I will set down its characteristics as literally as I can in the immediate terms of its presentation. He tells us that we must say "verify" (but if we do not like to say "verify," then we may say "confirm"; II, 4). If we "confirm," however, we must be prepared to admit all factual statements as "at least theoretically capable of verification," which means that such statements must permit themselves to be "completely translated" into "atomic propositions" (but if we do not believe in "atomic propositions," then we may "take the terms `simple,' `ultimate,' `basic,' and `atomic' in a relative and not an absolute sense"; II,
(485) 5). Proceeding thus he quickly finds "true definition" and "correct usage" in close quarters, but it is wholly wrong to attempt to equate them; instead we may carefully say: "the analysis which yields the definition will be correct if the definition does actually render the usage examined" (II, 7). All this, however, is merely "description" and philosophers must rise to "prescription," though when they get there they have nothing to rely on except "considerations of utility, convenience, and aesthetic value" (II, 7); nevertheless they may in the end come to "speak of the meaning of a sign, the correct analysis of an assertion." While all this is "difficult in practice" it is "simple in theory" and will finally result in a substitution of the "concrete, practical" for the "vague, exalted, mystical" (II, 8).
The above somewhat astonishing outcome of "philosophic analysis" flows professedly from "the nature of language" which involves a "fundamental distinction" (II, 3) between statements of syntax and statements of fact. The parentage of this linguistic base itself is in point-soul and concept. Blumberg does not use the word `concept' in his paper, and abstention here, unless it is a minor treachery, is always a favorable sign. His stress is heavy on the behaviorally phrased inquiry: "In what ways do people use a certain sign ?" His outcome is close to immolating point-soul and concept alike, and by that test we may say that it offers something that a normal scientific investigator would regard as a fair, even though still crude, starting point, for his inquiry. One may even anticipate on his own showing that he, or some other inquirer in similar state, will soon strip off the apparatus of logical positivism in order to undertake direct and positive investigation.
It is hardly necessary for me to say that with the great objectives of the logical positivists—the expulsion of metaphysics, and the development of a linguistic frame for appraisal and organization—I am in the fullest sympathy. What is criticized in the present paper is solely some of the detail of workmanship.