it is usual to designate kant as the founder of the theory of knowledge in the modern sense. against this view it might plausibly be argued that the history of philosophy records prior to kant numerous investigations which deserve to be regarded as something more than mere beginnings of such a science. thus volkelt, in his fundamental work on the theory of knowledge,1 remarks that the critical treatment of this discipline took its origin already with locke. but in the writings of even older philosophers, yes, even in the philosophy of ancient greece, discussions are to be found which at the present day are usually undertaken under the heading of theory of knowledge. however, kant has revolutionised all problems under this head from their very depths up, and, following him, numerous thinkers have worked them through so thoroughly that all the older attempts at solutions may be found over again either in kant himself or else in his successors. hence, [281]for the purposes of a purely systematic, as distinct from a historical, study of the theory of knowledge, there is not much danger of omitting any important phenomenon by taking account only of the period since kant burst upon the world with his critique of pure reason. all previous epistemological achievements are recapitulated during this period.

the fundamental question of kant’s theory of knowledge is, how are synthetic judgments a priori possible? let us consider this question for a moment in respect of its freedom from presuppositions. kant asks the question precisely because he believes that we can attain unconditionally certain knowledge only if we are able to prove the validity of synthetic judgments a priori. he says: “should this question be answered in a satisfactory way, we shall at the same time learn what part reason plays in the foundation and completion of those sciences which contain a theoretical a priori knowledge of objects;”2 and, further, “metaphysics stands and falls with the solution of this problem, on which, therefore, the very existence of metaphysics absolutely depends.”3

are there any presuppositions in this question, as formulated by kant? yes, there are. for the possibility of a system of absolutely [282]certain knowledge is made dependent on its being built up exclusively out of judgments which are synthetic and acquired independently of all experience. “synthetic” is kant’s term for judgments in which the concept of the predicate adds to the concept of the subject something which lies wholly outside the subject, “although it stands in some connection with the subject,”4 whereas in “analytic” judgments the predicate affirms only what is already (implicitly) contained in the subject. this is not the place for considering the acute objections which johannes rehmke5 brings forward against this classification of judgments. for our present purpose, it is enough to understand that we can attain to genuine knowledge only through judgments which add to one concept another the content of which was not, for us at least, contained in that of the former. if we choose to call this class of judgments, with kant, “synthetic,” we may agree that knowledge in judgment form is obtainable only where the connection of predicate and subject is of this synthetic sort. but, the case is very different with the second half of kant’s question, which demands that these judgments are to be formed a priori, i.e., independently of all experience. for one thing, it is altogether possible6 that such judgments do not occur at [283]all. at the start of the theory of knowledge we must hold entirely open the question, whether we arrive at any judgments otherwise than by experience, or only by experience. indeed, to unprejudiced reflection the alleged independence of experience seems from the first to be impossible. for, let the object of our knowledge be what it may—it must, surely, always present itself to us at some time in an immediate and unique way; in short, it must become for us an experience. mathematical judgments, too, are known by us in no other way than by our experiencing them in particular concrete cases. even if, with otto liebmann,7 for example, we treat them as founded upon a certain organisation of our consciousness, this empirical character is none the less manifest. we shall then say that this or that proposition is necessarily valid, because the denial of its truth would imply the denial of our consciousness, but the content of a proposition can enter our knowledge only by its becoming an experience for us in exactly the same way in which a process in the outer world of nature does so. let the content of such a proposition include factors which guarantee its absolute validity, or let its validity be based on other grounds—in either case, i can possess myself of it only in one way and in no other: it must be presented to me in experience. this is the first objection to kant’s view. [284]

the other objection lies in this, that we have no right, at the outset of our epistemological investigations, to affirm that no absolutely certain knowledge can have its source in experience. without doubt, it is easily conceivable that experience itself might contain a criterion guaranteeing the certainty of all knowledge which has an empirical source.

thus, kant’s formulation of the problem implies two presuppositions. the first is that we need, over and above experience, another source of cognitions. the second is that all knowledge from experience has only conditional validity. kant entirely fails to realise that these two propositions are open to doubt, that they stand in need of critical examination. he takes them over as unquestioned assumptions from the dogmatic philosophy of his predecessors and makes them the basis of his own critical inquiries. the dogmatic thinkers assume the validity of these two propositions and simply apply them in order to get from each the kind of knowledge which it guarantees. kant assumed their validity and only asks, what are the conditions of their validity? but, what if they are not valid at all? in that case, the edifice of kantian doctrine lacks all foundation whatever.

the whole argumentation of the five sections which precede kant’s formulation of the problem, amounts to an attempt to prove that the propositions of mathematics are [285]synthetic.8 but, precisely the two presuppositions which we have pointed out are retained as mere assumptions in his discussions. in the introduction to the second edition of the critique of pure reason we read, “experience can tell us that a thing is so and so, but not that it cannot be otherwise,” and, “experience never bestows on its judgments true or strict universality, but only the assumed and relative universality of induction.”9 in prologomena,10 we find it said, “first, as regards the sources of metaphysics, the very concept of metaphysics implies that they cannot be empirical. the principles of metaphysics (where the term ‘principles’ includes, not merely its fundamental propositions, but also its fundamental concepts), can never be gained from experience, for the knowledge of the metaphysician has precisely to be, not physical, but ‘metaphysical,’ i.e., lying beyond the reach of experience.” lastly kant says in the critique of pure reason: “the first thing to notice is, that no truly mathematical judgments are empirical, but always a priori. they carry necessity on their very face, and therefore cannot be derived from experience. should anyone demur to this, i am willing to limit my assertion to the propositions of pure [286]mathematics, which, as everybody will admit, are not empirical judgments, but perfectly pure a priori knowledge.”11

we may open the critique of pure reason wherever we please, we shall always find that in all its discussions these two dogmatic propositions are taken for granted. cohen12 and stadler13 attempt to prove that kant has established the a priori character of the propositions of mathematics and pure natural science. but all that kant tries to do in the critique may be summed up as follows. the fact that mathematics and pure natural science are a priori sciences implies that the “form” of all experience has its ground in the subject. hence, all that is given by experience is the “matter” of sensations. this matter is synthesised by the forms, inherent in the mind, into the system of empirical science. it is only as principles of order for the matter of sense that the formal principles of the a priori theories have function and significance. they make empirical science possible, but they cannot transcend it. these formal principles are nothing but the synthetic judgments a priori, which therefore extend, as conditions of all possible empirical knowledge, as far as that knowledge but no further. thus, the critique of pure reason, so far from proving the a priori character of mathematics and [287]pure natural science, does but delimit the sphere of their applicability on the assumption that their principles must become known independently of experience. indeed, kant is so far from attempting a proof of the a priori character of these principles, that he simply excludes that part of mathematics (see the quotation above) in which, even according to his view, that character might be called in question, and confines himself to the part in which he thinks he can infer the a priori character from the bare concepts involved. johannes volkelt, too, comes to the conclusion that “kant starts from the explicit presupposition” that “there actually does exist knowledge which is universal and necessary.” he goes on to remark, “this presupposition which kant has never explicitly questioned, is so profoundly contradictory to the character of a truly critical theory of knowledge, that the question must be seriously put whether the critique is to be accepted as critical theory of knowledge at all.” volkelt does, indeed, decide that there are good grounds for answering this question in the affirmative, but still, as he says, “this dogmatic assumption does disturb the critical attitude of kant’s epistemology in the most far-reaching way.”14 in short, volkelt, too, finds that the critique of pure reason is not a theory of knowledge free from all assumptions.

in substantial agreement with our view are [288]also the views of o. liebmann,15 holder,16 windelband,17 ueberweg,18 eduard von hartmann,19 and kuno fischer,20 all of whom acknowledge that kant makes the a priori character of pure mathematics and physics the basis of his whole argumentation.

the propositions that we really have knowledge which is independent of all experience, and that experience can furnish knowledge of only relative universality, could be accepted by us as valid only if they were conclusions deduced from other propositions. it would be absolutely necessary for these propositions to be preceded by an inquiry into the essential nature of experience, as well as by another inquiry into the essential nature of knowing. the former might justify the first, the latter the second, of the above two propositions. [289]

it would be possible to reply to the objections which we have urged against the critique of pure reason, as follows. it might be said that every theory of knowledge must first lead the reader to the place where the starting-point, free from all presuppositions, is to be found. for, the knowledge which we have at any given moment of our lives is far removed from this starting-point, so that we must first be artificially led back to it. now, it is true that some such mutual understanding between author and reader concerning the starting-point of the science is necessary in all theory of knowledge. but such an understanding ought on no account to go beyond showing how far the alleged starting-point of knowing is truly such. it ought to consist of purely self-evident, analytic propositions. it ought not to lay down any positive, substantial affirmations which influence, as in kant, the content of the subsequent argumentation. moreover, it is the duty of the epistemologist to show that the starting-point which he alleges is really free from all presuppositions. but all this has nothing to do with the essential nature of that starting-point. it lies wholly outside the starting-point and makes no affirmations about it. at the beginning of mathematical instruction, too, the teacher must exert himself to convince the pupil of the axiomatic character of certain principles. but no one will maintain that the content of the axioms is in any way made dependent on these [290]prior discussions of their axiomatic character.21 in exactly the same way, the epistemologist, in his introductory remarks, ought to show the method by which we can reach a starting-point free from all presuppositions. but the real content of the starting-point ought to be independent of the reflections by which it is discovered. there is, most certainly, a wide difference between such an introduction to the theory of knowledge and kant’s way of beginning with affirmations of quite definite, dogmatic character.