Curt Ducasse, Philosophy as a Science, 1941

CHAPTER FOURTEEN

The Method of Knowledge in Philosophy




IN THIS chapter, I propose to summarize and to supplement at a number of points the conclusions that have gradually been reached in the preceding discussion concerning philosophy and its method.

Our initial postulate, it will be recalled, was that philosophy resembles the natural and other sciences in being essentially a knowledge-seeking enterprise. This entails that its method must be "scientific," that is, knowledge-yielding; and scientific method in general -- as distinguished from the particular procedures dictated by application of it to particular sorts of subject matter -- may be characterized broadly as involving scrutiny of data (so far as possible under experimental conditions), empirical generalization, explanatory theorizing, empirical resting of the theories constructed, and organization of the validated theories belonging to the given field of inquiry into a logically complete whole.{1} As a product of the construction of valid theory additional laws of the given field, too exact or too recondite to have been discovered by direct generalization of observations, are usually obtained. That these are the essential features of knowledge-yielding method would probably be granted so far as the natural sciences are concerned; and something has been said to show that they may be traced also in the procedures of the formal sciences, and what forms they assume in those sciences.

To vindicate the contention that these features also characterize knowledge-yielding method in philosophy, what was needed was similarly to make clear what specific forms they assume where the subject matter about which knowledge is sought is philosophical.

1. The Facts of Observation in Philosophy. -- For this it was first of all necessary to determine what exactly is the subject matter distinctive of philosophy. To do this we were forced to distinguish between the primitive and the derivative facts of any knowledge-seeking enterprise, and we found that the subject matter distinctive of any given such enterprise is ultimately to be defined in terms of the sort of facts the enterprise takes as primitive. The facts primitive for philosophy, it was asserted, consist of spontaneous particular appraisals in so far as made available for public study through veracious formulation of them. That they consist of appraisals was evidenced on the one hand by pointing out that philosophy is generally thought of as being the search for wisdom, and that the maxims or aphorisms in which wisdom is typically formulated state, either directly or by implication, appraisals of various kinds of things, and on the other hand, by pointing out that theorizing of the kind generally called philosophical is what automatically occurs when one attempts to settle rationally a doubt or dispute of a spontaneous appraisal.

Formulated spontaneous appraisals of particular entities (which entities may be acts, objects, situations, thought processes, feelings, or anything else) are thus the philosophical analogues of the publicly perceived particular facts that are primitive for natural science, and are accordingly what primitive philosophical experience, or primitive facts of observation in philosophy, consist in. But in philosophy as in natural science, there are facts of observation which, although still particular, are derivative instead of primitive. This is the case whenever a character observed in a given particular entity is of a kind implicit in a certain kind of primitive character present in this entity -- that is, derivative in the sense that the former kind of character was originally discerned through interpretive reflection upon primitives of the latter kind.

This reflection, however, may have been performed not by us but by someone else, and its results then directly or indirectly imparted to us by him. This means that a character familiar to us may be derivative and yet not have been derived by us. As Whewell emphasizes, something which yesterday would have been called "interpretation of a perceived fact" is today called "perception of a fact" if in the meantime the interpretation has been tested and validated and has become habitual. The physicist, for instance, will say not that he perceived a streak in the vapor of a cloud-chamber, but that he perceived the emission of an electron. Or, to take a less esoteric example, we say that we perceive the sun being eclipsed by the moon, instead of saying that we perceive the shape of the sun gradually changing from a circle to a crescent that diminishes to nothing.

Similarly in philosophy, we must distinguish between primitive characters, which are the various value-characters, and derivative philosophical characters, which are the ones implicit in the value-characters. For example, "being erroneous" is a primitive philosophical character. But "being erroneous" implies "being an opinion." Therefore "being an opinion" is a derivative philosophical character. And if further analysis reveals that an opinion is, let us say, a proposition plus some degree of inclination to believe or to disbelieve it, then "being a proposition," "being disbelief," etc., would thereby be revealed as derivative philosophical characters. Accordingly, the judgment expressed by the statement, "That these words are printed in blue ink is an opinion," would be a derivative particular philosophical fact; whereas the judgment expressed by the statement, "That these words are printed in blue ink is erroneous," would be a primitive particular philosophical fact.{2}

These are the sorts of particular facts with which philosophical theories are concerned, and by reference to which they are tested. These facts are different from the particular facts relevant to the theories of natural science, but they are facts equally, and experiential facts; and the testing of philosophical theories by reference to them is therefore as genuinely empirical testing. A philosophical theory, to the effect, for instance, that, as suggested above, an opinion is a proposition plus one or another degree of inclination to believe or to disbelieve it, must, in order to be acceptable, fit among others the particular philosophical fact expressed by the statement, "That these words are printed in blue ink is an opinion," and also the particular philosophical fact expressed by the statement, " 'Are these words printed in blue ink?' is not an opinion."

2. The Empirical Generalization of Philosophy. -- By empirical generalization of the primitive philosophical facts, empirical norms -- empirical axiological laws -- are obtained. They are the philosophical analogues of the empirical laws which, in the field of natural science, describe the commonly recognized (nontechnical) properties of various kinds of things -- for instance, the property or law of steel, that it rusts in moist air;{3} the property or law of wax, that it is fusible by heat, etc. An example of a philosophical law obtained by the empirical generalizing of derivative instead of primitive particular facts would be the one expressed by the statement, "All entities expressible by sentences of the form 'S is P' or 'S is not P' are opinions."

3. Primitive Philosophical Facts, and Therefore Also Empirical Norms, Always "Functions" of Specific Person. -- All facts primitive for philosophy are appraisals, and an appraisal essentially involves three things: something appraised, some appraiser of it, and the sort of appraisal made. Therefore, no statement of an appraisal is complete unless it specifies all three of these constituents. That is, every statement of an appraisal, and therefore every statement of a philosophical primitive and likewise of a norm, whether empirical or not, must, in order to be unambiguous, have the form "the appraisal made of entity (or entities) E by person (or persons) P is of kind A." Accordingly, a peculiarity of primitive philosophical facts (and of value-facts in general) is that such facts are all essentially relative to a given person or group, in the sense that some particular person or group -- and, moreover, an appraising one -- is an intrinsic constituent of all of them. Specification of the person or group concerned may, as we have seen, be made either denotatively, or else descriptively, in terms of its "position," i.e., of its relation, to the entity appraised.

The primitive facts of natural science, on the other hand, are not in this sense essentially relative to -- are not "functions" of -- some person. They are indeed likewise objects of judgment by persons, but in them the object of judgment does not itself contain a person as one of its intrinsic constituents. Accordingly, whereas mention of a person or persons is indispensable to the adequate description of a primitive philosophical fact or of an axiological law, mention of a person in the description of a primitive fact (or law) of natural science is irrelevant in all cases except those where the fact happens to be specifically an anthropological one; and in these "person" means only a human body making perceptually public responses to perceptually public stimuli.

4. The Task, the Nature, and the Validity of Philosophical Theories -- The sort of situation which calls for construction of a philosophical theory is this: A given philosophical predicate is spontaneously applied by given persons to a variety of entities which, so far as observation directly reveals, seem so have in common no character invariably lacking in the variety of other entities to which the opposite of that predicate is also spontaneously applied by these persons. The task of theory is to explain this situation conceptually, and this is done by constructing conceptually the unperceived character satisfying that requirement. This character is ex hypothesi a hidden one, since otherwise merely abstractive observation would be capable of exhibiting it, and theory construction, i. e., invention of it, would not be needed. That, in spite of its hiddenness, its presence or absence in a given entity is nevertheless empirically ascertainable is due to the fact that its presence or absence is not ascertained by direct observation, but only through the indirect test specified in the "method" which, as we have seen, always has to accompany the "definition" proposed by the theory for the given predicate.

The validity of the theory -- the "reality" of the character it invents -- is a matter, first, of whether it fits the facts which actually gave rise to the need for a theory, and, second, of whether it successfully predicts facts other than the ones it was specifically intended to fit. More specifically, the two tests are as follows:

The facts which actually gave rise to the need for a theory consisted in the spontaneous application by given persons of the given predicate and its opposite, respectively, to two sets of given entities; and the first test is whether applicability of the predicate and its opposite respectively to these two sets of entities is deducible from the theory.

The second test is whether some other sorts of entities, which were not taken into consideration in the devising of the theory but to which (according to the theory) the given predicate, or its opposite, should be applicable, turn out to be entities to which respectively the given predicate or its opposite are in fact spontaneously applied by the same persons.

A theory which satisfies the first test, and for the validity of which there is also considerable evidence of the kind defined by the second test, provides a basis for rational settlement of cases of doubt or conflict among the applications of the given predicate or of its opposite by the given persons; for to settle such cases rationally consists in deciding them in a manner whose theory is identical with a theory already acknowledged by these persons as adequate to explain the majority or the most confident of their own applications of that predicate and its opposite. A theory whose validity has been confirmed in this manner moreover makes possible the discovery -- deductively from it -- of new (i.e., nonspontaneous) applications (affirmative or negative) of the given predicate which are valid, in the sense of coherent in principle with those already made by the persons concerned.

Before passing to concrete illustrations of the manner in which the empirical tests of validity of philosophical theories are applied in practice, we must reply briefly to an objection which might suggest itself at this point.

5. Two Ways of "Knowing" the Meaning of a Philosophical Predicate. -- As already pointed out, a philosophical theory reduces in essentials to a definition of a philosophical predicate, together with a method for the empirical identification of cases of the definiens. But the definition cannot be an arbitrary one, since there are tests it must meet. A question therefore suggests itself as to how it can be needed at all. For either one knows what a term means or one does not and, since most philosophical predicates are terms in common use, one must already know what they mean. Otherwise one would not be able to use them.

The answer obviously is that there are ways and ways of "knowing what a term means." More particularly, there are two. One of them may be described as intuitive, superficial, and good only for everyday practical purposes. To "know the meaning" of a word, in this sense, consists in being able to act appropriately in response to familiar sentences in which the word is used, and in being able to use the word in sentences ourselves. This is the way in which every English-speaking person knows, for instance, what is meant by the word "time": he is able to act appropriately in response to such sentences as, "You have time for lunch before the train starts"; "What time is it?"; "How did you spend your time?" But only very few persons have, in addition to this, also the analytical and discursive knowledge of the meaning of "time" needed for theoretical purposes -- the sort of knowledge of its meaning, namely, that would enable them to answer adequately the question "What is time?" or questions that presuppose knowledge of the answer to it. The knowledge which philosophical theory seeks of the meaning of the predicates with which it is concerned is knowledge of this second kind. If we distinguish between (1) a term T, (2) that which it means, M, i.e., the meant, and (3) the conceptual analysis of M, we may say that when M is a philosophical character (and T therefore a philosophical predicate), conceptual analysis of M (as distinguished from directly observational analysis of it) is what philosophical theorizing undertakes.

6. Two Kinds of "Meaning" of a Term. -- Terms, and among them those whose meaning philosophy investigates, obviously do not have a meaning in themselves but only as used by a given person or persons. When, however, we speak of the use made of a given predicate, we may be referring either to the applications made of it -- i.e., to the variety of subjects of which we predicate it -- or to the implications of it, i.e., to the set of attributes we assert of anything by means of the given predicate. In the first case, "to mean" is essentially to orient us to -- to call our attention to, to indicate to us -- certain things (as to the attributes of which we may know very little); whereas in the second case "to mean" is essentially to prepare us for or prepare us to -- to prepare us for some experience, or prepare us to act in some particular way. The first may be called the indicative or monstrative use of a term, and the second its quiddative use. The distinction is ultimately that between where and what, between the place of something and the properties of it. It is of first importance for the theory of knowledge and should not be blurred by, for instance, using the word "referent" indiscriminately for both. The distinction is customarily described as one between the denotation and the connotation, or the extension and the intension of a term. These two pairs of terms often are used synonymously, although both pairs, I believe, are needed to mark a further important distinction. This matter, however, which it seems to me even the best writers have not satisfactorily discussed, would demand a chapter in itself and need not be further pursued for our present purpose. I need say only that I shall use the terms "extension" or "extensional meaning" and "intension" or "intensional meaning," or the terms "application" and "implication," as J. N. Keynes does -- to refer to the general distinction made irrespective of further refinements.

The matter is brought up here only because the application and the implication of a term are indissolubly connected and because of the bearings of this fact upon the method of empirical investigation of the meaning of given terms. Although the fact of cardinal importance here is familiar enough, its methodological bearings are only too often ignored. That fact is that the moment either the application or the implication of a term is given, the other is thereby implicitly determined, and therefore cannot be independently given but must be (or must have been) discovered empirically. Conversely, there is no possibility of empirical inquiry into either the application or the implication of a term unless the other is given to begin with. That is, as cannot be repeated too often, there is no such thing as a genuine problem without data; and in particular there is no genuine problem as to the implication of a term except as relative to given applications of it.

The indissoluble connection between application and implication is evident enough if we turn to examples. If, for instance, it is given that the term "tree" is applied to this, that, and that object, then it cannot also be given that that term implies certain characters, such as having feathers, etc., which any of these objects happens not to possess. For the term "tree" can then imply none but characters possessed by all of them; and which characters these are can ultimately be decided only by observation of these objects. What we can stipulate in such a case is only that the term shall be used to imply only certain ones of these characters, e.g., those which together differentiate the objects to which "tree" is applied from any objects that resemble these in some stated respect but to which the name tree is in fact denied.

On the other hand, if it is given to begin with that the term tree implies possession at least of a certain character C, then it cannot be also given that it applies to a certain object, for this object may happen not to possess it. The term can be applied to -- predicated of -- none but objects that have character C, and only empirical investigation, not stipulation, can tell us which ones have it. What we can stipulate, however, is that the term shall be applied only to certain ones of these objects, e.g., only to such of them as have in addition certain other characters.

In practice, when we are called upon to investigate the meaning of a term in common use, we do not start with a tabula rasa on which arbitrary stipulation would have to write the data of the problem. Rather, we already know (whether intuitively or discursively) some characters which the term implies and some entities to which it applies -- all these characters being in fact possessed by all these entities. Out problem is then either that of discovering which other entities also possess all of these characters (the latter than functioning as data of our problem), or that of discovering what other characters are all possessed by all the given entities (the latter then functioning as data of our problem).

In the light of these remarks and of the considerations summarized in the preceding sections of this chapter, we shall now attempt to state in the form of a number of precepts the sort of method which can yield genuine knowledge in theoretical philosophy.

7. Precepts of the Method of Knowledge in Theoretical Philosophy. -- Let us recall to begin with the illustration from the field of ethics previously employed, namely, the problem of the nature of right and wrong. To be even theoretically -- to say nothing of practically -- soluble the problem to which these words only allude must be stated in a manner specifying its data, for, once more, a problem without data is a question about nothing. As that problem presents itself on a given occasion, an adequate statement of it might be as follows: What theory of wrongness and rightness explains conceptually the application of the predicate "wrong" to, for instance, killing, stealing, and adultery, and of the predicate "right" to, for instance, kindness, honesty, and veracity, by persons P, Q, R? In so stating the problem, we are obeying the first of the methodological precepts that follow from all our preceding discussion. This precept is: Specify the data of the problem.

There is nothing revolutionary about this precept, which only articulates the procedure ordinary common sense would dictate. Yet in philosophy it is only too often unheeded, the data of the problems attacked being assumed to be already understood, or being only alluded to or hinted at instead of explicitly stated. This first precept, moreover, loses some of its triteness if we particularize it specifically to fit problems such as the one here exemplified, where what we seek is the intensional meaning -- the "implication" -- of a given term; for it then becomes: List the application of the given term, and the applications of the contrary term from which we seek to distinguish it, that are to be explained. To comply with this precept as to give such lists as: "killing is wrong"; "adultery is wrong"; "stealing is wrong"; "kindness is right"; "honesty is right"; "veracity is right."

When this is done, the second precept readily suggests itself, namely: In the listed application substitute for the term the proposed definition of it. If this substitution leaves unchanged the truth-value of the respective applications, the proposed definition "fits" the term as applied to the entities mentioned in them. That is, it explains these applications of the term in the sense that they can be deduced from the proposed definition of it.

It may be asked, however, how we can tell whether or not the truth-values remain unchanged -- more specifically, how we can tell whether the truth-value of, for instance, "stealing is a case of D" (where D is the definiens of a proposed definition of "wrong"), is the same as the truth-value of "stealing is wrong." The answer is as follows: That the truth-value of the latter is "truth" is known to us to begin with ex hypothesis, for as emphasized earlier the statement "stealing is wrong" formulates simply the spontaneous disapproval of stealing by the given persons P, Q, R. We assume here only that when they assert their disapproval of stealing they are not dissembling their actual attitude. As regards the truth-value of "stealing is a case of D," we are enabled to ascertain whether it is "truth" or "falsity" by the very "method for identification of cases of the definiens" which as we have seen must accompany a proposed definition whenever that definition is a theoretical one instead of one obtained directly by abstractive analysis of observed facts.{4}

It may be asked further what precept can be given for discovery of a theoretical definition which, in the sense stated, "fits" the specified applications. The answer is that no precept for this can be given other than: Try, and if you don't succeed, try again and again. Thinking of a definition which will fit is a matter of the fertility and sagacity of the investigator's mind. The method of trial and error, if it can be described as a method at all rather than as absence of method, is the only one available here; and we can give a precept only for determining whether a definition we somehow came to think of does or does not fit.

The second precept above formulates the method for determining whether or not a definition proposed for a given term fits the applications of that term which it was specifically devised to fit. As we have seen, however, a definition is not authoritatively explanatory unless it has in addition the capacity to predict successfully spontaneous applications of the term that were not considered at all in the devising of the definition; that is, spontaneous applications of it that were no part of the specific facts that gave rise to the problem and in terms of which it was formulated. The precept that governs the testing of a proposed definition in respect to that capacity is as follows: Consider kinds of entities other than those figuring in the data of the problem and determine, by the method which accompanies the proposed definition, whether, according to that definition of the given term the term, the term is, or on the contrary is not applicable to entities of their other kinds. If its applicability and inapplicability to such entities (according to the definition) coincides with the application and nonapplication spontaneously made of it to those very entities by the same persons P, Q, R, then the definition is thereby shown to posses to this extent the required sort of predictive capacity.

These are the chief precepts by compliance with which the status of knowledge -- probable knowledge -- can be obtained for the definitions which theoretical philosophy is so largely occupied in seeking. The procedure these precepts dictate perhaps seems artificial and remote from anything ever actually occurring in philosophical investigations. But any appearance of artificiality is, I believe, only a case of the sort of result which formal description of the familiar almost always has. The formal description of the process of hitting a golf ball, for instance, which could be formulated through analysis of it by slow-motion pictures, would similarly seem artificial. Again, if a person, even one whom we have known all our lives, were minutely dissected and the relations of the pieces described, we should doubtless say that what we are offered is nothing much like him. What analysis exhibits is always more or less unfamiliar even when the analysis is correct, and this just because analysis does reveal within the familiar certain features that had not hitherto been individually discriminated. Moreover, examples are not lacking in which the procedure described by these precepts has been actually used by philosophers, even if less systematically than would be desirable or than would have been the case if their use of it, instead of being instinctive, had been fully conscious and deliberate. The last of the precepts given, for instance, is the very one Thomas Reid was following when, to demonstrate the invalidity of Hume's definition of the "cause" of a phenomenon as the antecedent which the phenomenon invariably has followed in our past experience, he pointed out that according to this definition the term "causation" would be applicable to the relation of day to night and of night to day, whereas in fact our spontaneous judgment refuses to apply that term to this relation.

It may be added that what is essential, in order that our method of philosophical investigation be knowledge-yielding, is in the first place that the nature of the data relevant to problems of definition of philosophical terms in general be clearly realized, and in the second that the data of the specific problem of definition to which we address ourselves at a given time be clearly present to mind. Formal listing of them is a device for making sure that they are present to mind, and are the same in the minds of all parties to the given investigation. If in a given case we are already sure of this, the formal listing of them would only be pedantic.

I shall now give illustrations, in terms of some familiar philosophical problems, of the procedure which the precepts stated would dictate.

8. Illustrations of Procedure. -- Let us consider first the problem of the nature of ready. The first question which the precepts require us to ask is, "What are the data of the problem?"; and the attempt to answer it immediately forces us to decide which one of two different but very similar-sounding questions is the one we seek to answer. They are "What being is real?" and "What is being real?" The second calls for an account of the character constituting the criterion of realness as distinguished from unrealness, whereas the first calls for mention of the kind or kinds of being which possess that character. The second is a question as to the intension of the adjective real, and if it is a question susceptible of a nonarbitrary answer, its datum can consist only of an extension spontaneously given to the adjective. The first on the contrary is a question as to the extension of that adjective, and a nonarbitrary answer to it can be reached only if we have as datum of the problem an intension spontaneously given to the adjective real.

Problems of philosophical theory are for the most part concerned with the intension which theoretical purposes demand for a given philosophical term; but occasionally a question arises as to extension instead. An example would be the controversy between Bradley and McTaggart as to whether the individual or only the whole universe is real. The intensional assumption they both took as datum might be stated in some such way as this: "To be real is to be susceptible of being taken without contradiction as ontologically ultimate." This, however, can hardly be claimed to state the intension spontaneously ascribed to the term real, even by them. It states much rather a theory they both accept as to the intension of the term, although I can hardly imagine what spontaneous extensional data it would fit. But even if that definition of "to be real" were wholly arbitrary, it would, under a certain proviso, enable those who do take it as datum to settle a controversy among them as to whether individuals, or the universe as a whole, or both, or neither, or mind, or matter, etc., are real. The proviso is that that definition of real should be accompanied by the stipulation of a method for deciding whether any given thing is or is not "susceptible of being taken without contradiction as ontologically ultimate." I do not believe that this proviso was met even informally, and if it was not, the controversy, because pervaded by ambiguity, remains insoluble irrespective of whether the definition of "to be real" used in it as datum is or is not arbitrary.

The problem of "the nature of reality" has in most cases been discussed without preliminary discrimination between the two questions "What being is real?" and "What is being real?"; without realization that both these questions cannot be raised at once, since each requires as datum what in the other is the quaesitum; without realization that the only nonarbitrary data for either consist of uses spontaneously made of the adjective real; and without initial statement of such data. It is therefore little wonder that no authoritative answer has yet been found to the question as to "the nature of reality." If, however, we ask, "What definition of real would explain our spontaneous assertion that Spain is a real country but Utopia is not ?" we have a genuine problem, which is susceptible of solution because we are furnished its data, viz., with an extension of "real" which any intension proposed for this adjective must fit. It may be said, perhaps, that although this is a genuine and soluble problem, it is not the one ontologists have been attempting to answer. This may well be true, for "real" possibly has other kinds of use than that made of it in the statement that Spain is a real country. I insist only that if the ontologists' problem is to be a genuine and soluble one, its data have to be specified to begin with, and that, if they are not to be arbitrary, they can consist only of common spontaneous predications of the adjective real -- if not about Spain, then about some explicitly mentioned other sort of entity. If the ontologist is unable to mention them then neither we nor he himself know what his problem is.

As another example, we may take the question "What is it to exist?" which, although it is often regarded as synonymous with "What is it to be real?" is as I believe nevertheless distinct from it. Suppose that instead of asking 'What is it to exist?" we ask "What definition of 'to exist' would explain our spontaneous assertions that, for instance, black swans exist but green swans do not, and that a square root of 9 exists but a square root of 3 does not?" We have then mentioned the data of the problem -- the facts which any proposed definition of "to exist" is to fit -- and we are therefore in position to ascertain empirically, in accordance with the second and third precepts above, whether a given definition of it does or does not fit the facts.{5}

Again, the problem of the relation between mind and nature presupposes an answer to the questions "What is mind?" and "What is nature?" So long as they are phrased in this way, the correctness or incorrectness of any answer proposed to them remains impossible to check. But we may rephrase them, asking, "What definitions of 'mental' and of 'physical' would account for the spontaneous common assertions that wondering, remembering, imagining, are mental events, whereas breathing, walking, eating, are physical events?" Then, because we have in this way been furnished with the facts which the definitions sought are desired to fit, it becomes possible for us to show whether any proposed definitions do or do not fit them.

These illustrations will make clear how a realization of the nature of the data of the philosophical problems we raise, and an explicit statement of these data at the outset, open the way to genuinely knowledge-yielding procedure -- as described in the precepts given -- in the field of philosophical inquiry. The knowledge obtained in this manner will, like scientific knowledge in other fields, be for the most part probable knowledge only. That is, what we shall know will usually be only that a certain philosophical assertion is, on specified evidence we have at the time, more probably true than a certain other. This knowledge, although at first fragmentary and not organized into a system, will, because it is based on facts of philosophical experience, have the verifiability and the capacity for genuine growth which has been so conspicuously lacking in philosophy as compared with the other sciences, and which a philosophical system constructed a priori cannot possess, however formally perfect it may be.

The systematic organizing of philosophical knowledge -- the ascertaining of the mutual relations of the philosophical facts already discovered at a given time, and the looking specifically for the other facts needed to fill gaps still existing -- is a proper and necessary task in the field of philosophy as in any other. Its method is the same in philosophy as in any other field, and it can be begun as soon as any knowledge, as material to be organized, has been won. The great obstacles to real progress in philosophy have been, it seems to me, the lack of clear ideas as to the subject matter distinctive of philosophy, as to the general nature and function of theory construction, as to the sort of data by reference to which the validity of philosophical theories can be empirically tested, and as to the method by which this testing must proceed. It is on these obscure matters that I have specifically attempted to shed such light as I believe I see. I have therefore said little or nothing as to the criticism of theories, philosophical or other, in respect to their internal consistency or formal rigor. This is a matter commonly much better understood, and I have therefore concentrated instead on the problem of method for criticism of philosophical theories in respect to their fitting or failing to fit the facts they concern, and on the problem of the general nature of these facts.

9. Philosophy, Nature, Mind and Language. -- In earlier chapters the philosophical sciences were distinguished from the natural, the mental, and the formal sciences on the basis of the subject matter peculiar to each. The subject matter distinctive of philosophy, it was contended, is appraisals -- spontaneous particular appraisals -- and whatever turns out to be implicit in them. That of the natural sciences consists of facts perceptually public and whatever proves to be implicit in them. That of the mental sciences is facts introspectively ascertainable plus whatever they imply. And the subject matter of the formal sciences consists of facts created by stipulations as to relations of words or other discursive entities to one another, plus such other facts as are implicit in these stipulations.

The question might still be asked, however, how these four sorts of subject matter are related to one another. Do some of them perhaps in part overlap? Or are one or more of them perhaps species of the rest? Or are they wholly independent? More specifically, is mind perhaps only a part of nature? Are pure logic and mathematics parts of the science of mind, or perhaps of the science of nature? And are appraisals facts of mind or facts of nature? Little or no attention has been given to these questions in the present discussion, and the effort has been made not to base anything said concerning philosophy and its method upon any assumed particular answer to them. Accordingly, if the account that has been given of philosophy as normative science or science of appraisals holds at all, it holds no matter what may be the right answers to these questions. This, of course, is as it should be, since our inquiry belongs to the field of philosophical metatheory rather than directly to that of philosophical theory. Because of this, any discussion of these important and interesting questions is best left to some future occasion.
[Table of Contents]

Notes

{1} Experimental scrutiny is esentially observation of the differential outcome of an operation, i.e., of the outcome of the operation alone. The operation may be performed by ourselves (as in the laboratory use of the "method of difference" or, in mathematics, of the operation which, performed upon a number, reveals whether or not it is divisible by 7); or the operation may be performed by nature (as in instance in an eclipse).[Back]

{2} What is a philosophical fact in each use is the judgment mentioned, i.e., the fact is that somebody makes these judgments. What the person who makes them asserts, however, may or may not also be a fact, i.e., be true. [Back]

The judgment expressed by the statement, "That these words are printed in blue ink is erroneous," may be otherwise expressed: "No, these words are not printed in blue ink." That is, it essentially consists in disagreement with the opinion that the words are printed in blue ink. For disagreement or agreement with an opinion one understands, no training is needed. It is a primitive capacity. Even the savage and the young child have it, although, of course, the opinions they are able to understand are relatively few. On the other hand, the judgment expressed by the statement, "That these words are printed in blue ink is an opinion" is a judgment which only a highly sophisticated person is capable of making; for it is not an act of agreement but an act of classification, for which the concept of "opinion," is indispensable. And this concept (which is more abstract than the relatively familiar one of "mere opinion," i. e., belief as contrasted with knowledge) is possessed only by persons epistemologically sensitized enough to discriminate at least in practice between opinion in general and either erroneous or sound opinion, and also between sound opinion and knowledge. This sensitiveness is derived (either at first hand or more often at second hand) from reflection upon given cases of the application of such value predicates as "erroneous," "sound," "known," etc. [Back]

{3} This illustration is borrowed from Campbell's What Is Science? p. 40. [Back]

{4} A definition proposed for a term no matter whether obtained by would-be abstractive analysis of observations or by speculation is not acceptable unless its definiens can he substituted for the term in any statement whatever where the term occurs, without thereby altering the truth-value of the proposition formulated by the statement. The precept given in the text therefore applies no matter how the proposed definition has been obtained; but it has been obtained by abstractive analysis of observed facts, the question whether is fits or doe not fit these facts can be answered by more careful direct scrutiny of them. If on the other hand, the definition was obtained by invention, whether or not it fits can be ascertained only through the indirect test specified in the "method" with which any speculative definition has to be accompanied. [Back]

{5} In a paper entitled "On Our Knowledge of Existence," I have offered a definition of to exist which seems to me to fit these facts. See Proceedings, 7th International Congress of Philosophy, pp. 163-7. [Back]


[Table of Contents]