The Impossibility of
the "Theory of Knowledge,"
by Leonard Nelson

translated by Thomas K. Brown III

Editorial Note

Leonard Nelson's lecture, "Die Unmögliclikeit der Erkenntnistheorie," was delivered on April 11, 1911, before the Fourth International Congress for Philosophy at Bologna. It was orignally published in the Abhandungen der Fries'schen Schule, III, Göttingen, 1912, No. 4., and also in Die Reformation der Philosophhie, Leipzig, 1918. In German, it is now found in Geschichte und Kritik der Erkenntnistheorie, Volume II of the Gesammelte Schriften in Neuen Bänden of Leonard Nelson, Felix Meiner Verlag, Hamburg, 1973, pp.459-483. This English translation, by Thomas K. Brown III, was originally published in Socratic Method and Critical Philosophy, Yale University Press, 1949, copyrighted by the Leonard Nelson Foundation. The book was reprinted by Dover Publications, 1965. The Leonard Nelson Foundation ceased to exist in the 1970's, after the death of its founder, L.H. Grunebaum, and the book has been out of print since then. I have inquired more than once whether Dover has any interest in reissuing the book, and whether they now consider this material to be in the public domain, but I have received no response.

From some fairly simple considerations and arguments about what a Theory of Knowledge cannot accomplish, Nelson proceeds to the argument and exposition of basic elements of Friesian epistemology, such as the distinction between mediate and immediate knowledge, and between intuitive and non-intuitive immediate knowledge. Since recent philosophy continues to ignore these distinctions, arguing that immediate knowledge does not exist because (mediate) knowledge is fallible and corrigible, or that metaphysics does not exist because it cannot be grounded on either reflection or the (supposedly) non-existent immediate knowledge, Nelson's views continue be to as novel to philosophical debate today as they were in 1911. Indeed, the many Hegelians around then have, after some years of disfavor, returned now. Unfortunately, Nelson's own students, apparently beguiled by the confident but sterile assertions of Logical Positivism and subsequent Analytic Philosophy, came to reject Friesian epistemology themselves, as can be examined elsewhere in these pages.

On the other hand, we see some of the problems in Nelson. His reference to the regressive method serves to clarify that he is indeed talking about the Regress of Reasons which, starting with Aristotle, is still one of the fundamental considerations in modern epistemology. Nelson, however, was always confused that this regress was part of Socratic Method, when it is now clearer, after the work of Karl Popper, that Socrates was using the logic of falsification rather than verification. Nevertheless, while such problems may serve to weaken Nelson's case for the unsympathetic, it does not affect the general value of his conclusions. Modern philosophy, mired in Analytic sterility, scholasticism, or in fashionable nihilism and Marxoid "Theory," still treats Nelson in the way most effective for its own foolishness:  ignoring him.

The Impossibility of the "Theory of Knowledge"

If we review the activity of this Congress so far, we become aware of two facts -- facts of which other sources might have informed us but which it is good to find here corroborated. One of them is heartening, the other must sadden us. What is gratifying is that the mere existence of this Congress bears witness to a belief in the possibility of philosophy as science. That an international Congress for philosophy is possible presupposes the conviction that there can be common philosophical endeavor; and such endeavor is possible only if we believe in philosophy as science. But this conviction is manifested even more clearly and strikingly in the unusually close connection, at this Congress, between philosophy and the exact sciences; and the union of science and philosophy in the personality of our honored president is a special symbolic expression of our faith that such union is possible.

As you know, there are many who ridicule a congress for philosophy, who feel that attendance at such a Congress is beneath the true dignity of a philosopher. Such an opinion is necessary and natural to all those who regard philosophy as a matter of personal experience, as something that cannot be molded to precise, communicable forms -- all those, in short, to whom philosophy is not a science. Those, however, who do not share this opinion must welcome the lively interest this Congress has aroused and the special emphasis its program lays on the relations of philosophy to the exact sciences, as refreshing testimony to the belief in philosophy as science.

But even though we who have convened here believe in the possibility of a scientific philosophy, we must ask ourselves whether we really possess such a philosophical science; and we must in all honesty confess -- and this is the second, saddening fact of which this Congress makes us aware -- that at present the state of philosophy is not that of a science. We have observed, and the discussions have borne it in upon us, that there is no unanimity among those present on even the most elementary philosophical questions. The more we are concerned with achieving the goal of philosophy as a science, the more important it must be to us not to gloss over this fact, that philosophy is not now a science, but rather to bring it out as clearly as possible. Thus we shall find occasion all the sooner to scrutinize the reasons why we have not yet succeeded in making a science of philosophy and to inquire in what way we may hope to put an end to this unseemly state of affairs.

This is the problem to the solution of which my address will attempt to make a contribution.

The nature of this task obliges me to deal primarily with the formulation of our problem and with questions of method. In so doing, I am undertaking nothing new: our age is strikingly rich in such methodological research. Indeed, it has been regarded as a failing, as a sort of morbid symptom, that current philosophy is preponderantly concerned with the question of its proper method. I cannot share that point of view. Whatever the case may be in other sciences, in philosophy it is not a sign of decline but a sign of convalescence, when attention is paid above all to the correct method. In other sciences the cognitive material that is to be reduced to scientific form is accessible to us through relatively simple procedures, and we do not require a special preliminary investigation to find out how we can make it our own; but in philosophy everything is based on just such a methodological preliminary study. For the sum total of the cognitions that are to form the content of philosophical science are not available to us without special effort, and everything turns on how we go about getting hold of them. Consequently, the trustworthiness of our results depends altogether on our choice of method. Difficult though it may be to achieve unanimity regarding this method, it is futile to enter into a discussion of specific results until this has been achieved. Even though the many methodological efforts of recent times have not yet led to the desired goal, we may not on that account conclude that it would be better to abandon such methodological research in order finally to turn from the method to the real matter at hand. On the contrary, it is my opinion that if methodological work has not hitherto met with the hoped-for success, that is only because it has not attacked its task vigorously enough. If only this work is carried on with the necessary earnestness and energy, the correct way will soon be found to escape from philosophical anarchy to a harmonious and planned scientific endeavor.

It must strike a dispassionate observer of the development of philosophy in recent times as particularly remarkable that the contention and variety of philosophical opinions is greatest in just that discipline whose specific intention it has been to put an end to the fruitless quarrels of the earlier academic metaphysics, namely, in the studies on the "theory of knowledge" (Erkenntnistheorie).[note] These studies were originally undertaken with the sole purpose of making the philosophical problems, which without them appeared unsolvable, accessible to scientific treatment, either by establishing thus a scientific metaphysics or by ascertaining once and for all through a "theory of knowledge" the impossibility of such a scientific metaphysics. How is it to be explained that this apparently highly justified hope for a peaceful scientific form of philosophy has not only not been realized but on the contrary has immeasurably widened the dissension among the schools? I shall show that this curious phenomenon has a very simple cause, and that the problem of the "theory of knowledge" is similar to many allied problems in other sciences. We very often find that a problem we have disputed at great length, without getting a step nearer its solution, finally turns out to be one that simply permits of no solution or one whose solution (if we can call it that) is not the hoped-for positive finding but rather the proof of its unsolvability.

I shall, therefore, prove the impossibility of the "theory of knowledge," but in order not to conclude with a negative, unconstructive result, I shall enter into the question as to what positive consequences can perhaps be drawn from this fact -- the question whether we are thus obliged to abandon the search for philosophy as science, whether, in other words, we must relapse into dogmatic metaphysics, or whether perhaps there is a third path, which really leads to our goal. I believe I can show you that this last is the case, and I hope that my address itself will provide you with an illustration that we can follow this path.

First of all, I shall say as much regarding the method I shall employ as is necessary to an understanding of my exposition.

When we concern ourselves with philosophy as science, it is natural for us to take the example of the exact sciences as a model, though we know, or at least we should know from Kant, that we cannot blindly transfer to philosophy a method that is appropriate to the mathematical sciences. Pre-Kantian metaphysics made precisely this mistake of trying to imitate in philosophy the usual dogmatic method of mathematics. Kant proved definitively the faultiness of this undertaking. But the relation of the two sciences, philosophy and mathematics, has changed in a curious way, since Kant's time: during the last century modern mathematics has developed in what is called axiomatics a method that corresponds exactly to the one Kant demanded for philosophy. This is the regressive method, the importance of which does not lie in extending our knowledge, adding new truths to the fund of those already known, elaborating their consequences, but rather in examining the known truths with regard to their preassumptions. It serves as a means of investigating the conditions of the solvabililty of a problem before we attack the problem itself; it assures us whether or not the problem is solvable at all and what presuppositions are already implicit in the mere setting of the problem; it determines what presuppositions are necessary and sufficient for a definite solution of a problem.

I wish to apply this method to the problem of the "theory of knowledge," also called the problem of knowledge. This problem is that of the objective validity of our knowledge. It is the task of a "theory of knowledge" to test the truth or objective validity of our knowledge. I maintain that a solution of this problem is impossible, and I prove this as follows:

In order to solve this problem, we should have to have a criterion by the application of which we could decide whether or not a cognition is true; I shall call it briefly the "validity criterion." This criterion would itself either be or not be a cognition. If it be a cognition, it would fall within the area of what is problematic, the validity of which is first to be solved with the aid of our criterion. Accordingly, it cannot itself be a cognition. But if the criterion be not a cognition, it would nevertheless, in order to be applicable, have to be known, i.e., we should have to know that it is a criterion of the truth. But in order to gain this knowledge of the criterion, we should already have had to apply it. In both cases, therefore, we encounter a contradiction. A "validity criterion" is consequently impossible, and hence there can be no "theory of knowledge."

One need only take any example in order to make the content of this proof clearer. For instance, someone might assert that agreement of thinking subjects with one another is the criterion in question. To be able to apply this criterion, we should have to know that agreement of various subjects is a criterion of the truth of their knowledge. But in order to know this, we should have to apply this criterion itself to the assumption that agreement is the criterion in question. We should have to convince ourselves that all subjects agree on the assertion that agreement is a criterion of the truth of their assertions. But in order to realize from this the truth of this assumption, we should already have had to presuppose that it is correct, i.e., that agreement is a "validity criterion." Thus the possibility of achieving this knowledge would involve an inner contradiction.

Or someone might claim that obviousness is the criterion in question. In order for this criterion to be applicable, it would have to be known to us as such, i.e., we should have to know that the obvious cognitions are the true ones. But we could only know this if it were obvious that the obvious cognitions are true; however, in order to deduce the truth of this assumption from its obviousness, we should already have had to presuppose that obviousness is a criterion of truth. It is therefore impossible to achieve the knonwledge in question.

Or let us take pragmatism. If the usefulness of a notion is to be the sought-after criterion of truth, we should have to know, in order to be able to apply this criterion, that usefulness is the criterion of truth. We should therefore have to know that it is useful to think that useful thinking is the true thinking, and thus we should have to presuppose that the usefulness of thinking is a criterion of its truth. So here, too, we meet the same contradiction; and it is the same in every other case.

Now, what is the preassumption that we make in setting the problem of a "theory of knowledge" and that involves the contradiction we have observed? It is important, first of all, to realize that such a preassumption is really implicit in the problem, and that the alleged absence of preassumptions, proudly proclaimed by the "theory of knowledge," is simply a chimera. If one asks whether one possesses objectively valid cognitions at all, one thereby presupposes that the objectivity of cognition is questionable at first, and that we can assure ourselves of this objectivity only indirectly, namely, through the process of the "theory of knowledge." What can be said about this preassumption, which is indispensable to the "theory of knowledge"?

Let us begin by drawing up a clearer picture of the meaning and content of this preassumption. It seems at first to be nothing more than an application of the logical principle of sufficient reason, according to which every assertion needs a verification. And indeed, the "theory of knowledge" stands or falls with the preassumption of the necessity of verifying every cognition; for the task of this discipline is none other than to verify our cognition. Although this very preassumption seems to aim at the elimination of all prejudgments, the contradiction it leads to, demonstrated above, makes us aware that some error must lie concealed here and that consequently the preassumption is itself a prejudgment.

This contradiction really lies in the following: If every cognition needs a verification, that is equivalent to saying that it presupposes another cognition as its ground, to which it must be traced back if it is to be asserted as truth. The contradiction lies in the proposition, here implied, of the mediacy of all knowledge. For if every cognition were possible only on the ground of another, we should have to execute an infinite regression in order to reach any true cognition, and hence no verification of cognitions would be possible.

We can phrase this result in another way. If one asserts the mediacy of all knowledge in the manner just outlined, one there-with asserts that every cognition is a judgment. The word "judgment" is here used in its usual sense to mean the assertion of a thought that is in itself problematic. Every judgment presupposes a notion that is not in itself assertoric but to which the assertion is only mediately added. However, this preassumption that every cognition is a judgment involves also another, namely, that the verification of a cognition can only be a proof. A proof is the tracing back of one judgment to another that contains the logical ground of the first. But if there is no other verification of judgments except proof, no verification of judgments is possible at all; for all proof consists only in the tracing back of the judgment to be proved to unproved and unprovable judgments. Therefore, either there is another means of verifying judgments besides proof, or no verification of them is possible at all.

The above-demonstrated preassumption in the "theory of knowledge" -- and on this point I should like to lay particular stress -- involves not only the logical contradiction we have discussed; it also contradicts psycholological facts. It contains, as we have seen, the psychological conclusion that every cognition is a judgment; but this statement contradicts the facts of inner experience. To convince ourselves of the existence of cognitions that are not judgments, we need only consider any intuition at all, such as an ordinary sensory perception. For example, I have a sensory perception of the sheet of paper that lies here on the table before me. This perception is, first of all, a cognition, not merely a problematic notion. The existential assertion that is all element of this cognition is, however, not a judgment. To be sure, I can also render in a judgment the same circumstances that I here cognize through the perception; but when I judge that a piece of paper is lying before me on the table, that is an altogether different sort of cognition from the perception of this situation. I need concepts for the judgment, e.g., the concept "table," the concept "paper," etc. I connect these concepts in a certain manner and assert that objective reality pertains to this combination of concepts. Perception, on the other hand, has no need of any concepts nor of any problematic notion of its objects whatsoever; rather, it is itself an originally assertoric notion -- is, in other words, an immediate cognition.

Thus we find that problematic notions are not that which is original, to which objectivity would have to be contributed from some other source, but that it is cognition itself that is original. It is correct that judgments are possible only on the basis of concepts, which are problematic notions; nevertheless, this does not apply to cognition as such. With this ascertainment the problem of the "theory of knowledge" disappears: the possibility of cognition is not a problem but a fact.

We must now scrutinize this factual character of cognition. Once one is clear in one's mind about it, one will see a problem not in the possibility of cognition but rather in the possibility of error. For if we originally have only cognitions, the question presents itself how error can arise at all. To solve this problem, we need only inquire into the relation of judgments to immediate knowledge. In and of itself a judgment is not yet a cognition; it is such only under the condition that it reiterates an immediate cognition. Judgments are acts of reflection, and to that extent arbitrary. The combination of concepts in judgments is arbitrary and hence depends on a factor that is foreign to cognition. The truth of judgments, namely, their correspondence with immediate knowledge, is not an original fact; it is rather a task that we arbitrarily set ourselves so far as our interest in truth motivates us; and in the choice of the means for the accomplishment of this task we can be mistaken.

Before I proceed to develop the consequences of my previous observations, I should like to illuminate the impossibility of the "theory of knowledge" from another angle. The impossibility of such theory can be proved also in the following manner. Since for this discipline knowledge is not a fact but a problem, the "theory of knowledge," in order to solve this problem, cannot assume any knowledge as given; rather, it must begin solely with problematic notions, that is, mere concepts. Now, only analytic judgments can be developed from mere concepts, and they never provide a new cognition, which can only consist in synthetic judgments. Thus our task amounts to deriving synthetic from merely analytic judgments.

That this task, however, is incapable of execution can be proved as follows. If we assume that it is possible to derive synthetic from merely analytic judgments, then there must appear somewhere in the series of syllogisms a syllogism both of whose premises are still analytic while the conclusion is already synthetic. But if both the premises of this syllogism, major as well as minor, are analytic, this means that, on the one hand, the major term of the syllogism is already contained in the middle term and, on the other hand, the middle term is already contained in the minor term. But with major term thus contained in middle, as well as middle in minor term, then the major term is also already contained in the minor term, i.e., the conclusion must also be analytic, contrary to our assumption. Thus it is impossible ever to derive a synthetic judgment from merely analytic judgments, and the task of the "theory of knowledge," to show how knowledge can arise from purely problematic notions, consequently cannot be accomplished.

I presuppose for this second proof of the impossibility of the "theory of knowledge" that one concedes my distinction between analytic and synthetic judgments. I shall therefore briefly consider the principal objection that is raised against this distinction. It has been held that this distinction is variable and unprecise, so that one and the same judgment can at different times and for different people be now analytic, now synthetic; consequently, a mutation of a judgment from the one type to the other is possible. This objection vanishes when we discriminate between judgments and their linguistic expression. What varies and is unprecise is only the co-ordination between the expression and the thought it stands for. The same words can, at different times and for different people, have different meanings, and for this reason, to be sure, one and the same sentence can stand here for an analytic, there for a synthetic, judgment. Whoever concludes from this that the division ot judgments into analytic and synthetic is unprecise thereby confounds [fixed] concepts and [changing] word significations.

It is only through this same confusion that the dialectical illusion arises in attempts to solve the problem of the "theory of knowledge." All these endeavors tend toward a renewal of the old, logicizing metaphysics and accordingly can be no more than a repetition of the same old errors in new guise. The apparent success of attempts to create metaphysics from mere logic rests only on the ambiguity of words. This alone makes it possible, in "theories of knowledge," unconsciously to foist off an analytic judgment as a synthetic one by expressing them both in the same sentence.

I should like to adduce two examples of what I have just said, which at the same time will serve to clarify the importance of the basic thought I have presented and to distinguish it from other views with which it might perhaps be confused.

The answer to the question whether or not we possess valid knowledge -- no matter what this answer may be -- can only be sought in a synthetic judgment, since it is concerned with a fact. And yet it would appear as if we could prove our possession of knowledge purely logically by demonstrating a contradiction in the opposite assumption. This contradiction, with which absolute skepticism has time and again been reproached since Plato, is well known. It is said: Whoever asserts his inability to know anything is contradicting himself, for he claims to know what he is asserting, namely, his inability to know anything; and it follows from this contradiction that he knows something. This reasoning is not sound. It is, to be sure, contradictory if someone claims to know that he knows nothing; but it does not follow, by any means, from this contradiction that he knows something; it follows only that he does not know what he claims to know, namely, that he knows nothing. The contradiction lies not in the skeptical assumption that we know nothing but in the other assumption that we can know, this. It is not judgment A, "I know nothing," but judgment B, "I know that I know nothing," that leads to a logical contradiction; hence it follows only that judgment B is false, not judgment A. The refutation of skepticism by such a theory rests only on the confusion of these two judgments.

This result, viz., insight into the impossibility of a positive validation of knowledge, might induce us to make the opposite attempt to decide the problem negatively. If no verification of the validity of our knowledge is possible, it seems to follow that we can know nothing about the validity of our knowledge, that we shall therefore have to regard it skeptically. But this skeptical conclusion from the impossibility of verifying knowledge is just as erroneous. It malkes the tacit assumption that we can assert only that to be valid which can be verified; and this is precisely the same prejudice of the "theory of knowledge" on the basis of which we were, at the outset, led to the contradictory demand for a verification of knowledge.

I mention this last particularly because it might be objected that my proof of the impossibility of the "theory of knowledge" simply reiterates an old idea, often enunciated by the skeptics. It can be seen from what has been said that the skeptical arguments in question prove too little. I do not assert the impossibility of the "theory of knowledge" in order to conclude that knowledge is impossible; rather, I assert that this skeptical conclusion, that knowledge is impossible, is itself merely a consequence of the prejudice held by the proponents of a "theory of knowledge." The contradiction I have pointed out is characteristic not only of the positive solution of the problem of the "theory of knowledge" but indeed of every supposed solution, and hence also of the skeptical.

The opposite mistake from this skeptical resoning is to be found in certain other arguments to which one might wish to call my attention in order to trace back my proof of the impossibility of a "theory of knowledge" to ideas that have long been known. I am thinking of the attacks on the "theory of knowledge" initiated by Hegel and Herbart and especally championed by Lotze and Busse; they all have in common the fact that they were launched in favor of dogmatism. Where the skeptical attacks prove too little, these prove too much in that they postulate the necessity of a dogmatic metaphysics -- a consequence that cannot be derived from the proofs I have offered. Moreover, nothing can be decided by such vague arguments as, for example, the argument that one cannot swim before going into the water [Hegel's argument. -- ed.], or that cognition cannot cognize itself. In this way one could just as well prove the impossibility of philology with the assertion that one cannot use language with reference to language.

The alternatives of the "theory of knowledge" and dogmatism, i.e., of the necessity of verifying every cognition and the necessity of positing some judgments without any verification, are, to be sure, inevitable as long as one adheres to the already refuted presupposition that all cognitions are judgments. For on this presupposition one must necessarily extend the application of the principle of sufficient reason to all cognitions whatsoever and, on the other hand, confuse the obvious impossibility of verifying every cognition with the postulation of unverifiable judgments. If, however, one abandons the presupposition that every cognition is a judgment, the choice between a "theory of knowledge" and dogmatism disappears. The possibility then opens up of satisfying the postulate of the verification of all judgments without falling victim to the infinite regress of the "theory of knowledge."

The criterion of truth that we here use does not give rise to the contradiction that we have found in the concept of the "validity criterion." Indeed, the criterion of the truth of judgments cannot itself be a judgment, but it does not for that reason have to lie outside cognition; instead, it lies in immediate knowledge, which in its turn does not consist of judgments.

If philosophy wishes to be a science, it, like every science, have to confine itself to this one task, the task of verifying judgments. It will only then assert its own scientific existence or, rather, will first be able to achieve it when, instead of setting itself above the competence of science and sitting in judgment on the rights and qualifications of the various sciences, it is satisfied to take over for treatment a particular field of knowledge alongside the other individual disciplines.

That this is possible and how we shall easily see if we call to mind the purpose the misinterpretation of which originally created the problem of the "theory of knowledge." If we disregard proofs, which only serve to trace judgments back to other judgments, and consider only the basic judgments, we find that these -- unless they are analytic and have their ground in mere concepts -- are verified by being traced back to intuition in accordance with the universal procedure in all special sciences. But, as Hume first pointed out, there are judgments where this means of verification fails, judgments that do not have their ground in intuition although they are not analytic: these are all the judgments through which we conceive a necessary connection between things. One such judgment is, for instance, the principle of causality. The verification of these judgments, which Kant called the "synthetic judgments evolved through mere concepts," is, indeed, the task to which metaphysicians have always, more or less gropingly, devoted their efforts but which Kant, through his generalization of the Humean problem, was the first to formulate scientifically. It is readily understandable that once the nature of these "metaphysical" judgments was clearly recognized, once the impossibility of tracing them back to the only recognized sources of knowledge -- concept and intuition -- was grasped, it was tempting (for lack of an immediate knowledge on which to ground them) to try to verify them through comparison with the object, i.e., through the "theory of knowledge."

Our rejection of this cast to the problem as an inadmissible misconstruction exposes Hume's original problem in its true signficance. The possibility of verifying metaphysical judgments, and therewith the life or death of metaphysics as a science, depends on the solution of this problem. But it can easily be demonstrated that its solution can only be sought in psychology. We cannot develop the metaphysical judgmcents immediately from their source of knowlege in the way that geometry, for instance, can be developed from the intuition of space, for the nature and even the existence of this source of knowledge are precisely what are being questioned: this source is not readily at our disposal; rather, we must first search for it. The problem at hand, correctly understood, thus concerns the existence of a certain kind of cognition, namely, immediate metaphysical knowledge. This is first of all a question of facts, and consequently a question that can only be decided by way of experience. Secondly, the object whose factuality is in question is a cognition; and cognitions, whatever their object may be, are themselves only the object of inner experience. Therefore, the Hume-Kant problem can only be resolved through psychology, i.e., through the science of inner experience.

Now, what are the possible, that is, a priori conceivable, solutions of this problem?

Initially, the opinion is conceivable that the difficulty that gives rise to the problem only seems to exist, and that the so-called "metaphysical" judgments can actually, as the metaphysicians before Hume endeavored, be traced back to the then recognized sources of knowledge, i.e., mere reflection or intuition. Indeed, if we presuppose the completeness of the disjunction between reflection and intuition as sources of knowledge, any other way to verify the judgments in question is logically inconceivable.

If one chooses reflection as the source of knowledge of the metaphysical judgments, one arrives at metaphysical logicism; if intuition, metaphysical mysticism. If, however, one rejects both sources of knowledge for the metaphysical judgments, holding still to the exclusiveness of these two sources of knowledge, there remains only the conclusion that no source of knowledge exists as basis for the metaphysical judgments, that they are therefore altogether unverifiable, and hence are fraudulent assertions. This is the consequence of metaphysical empiricism.

These attempts at solution through metaphysical logicism, mysticism, and empiricism exhaust the logical possibilities available under the presupposition of the completeness of the disjunction between reflection and intuition as sources of knowledge. It has been generally assumed heretofore that therewith all the logically possible solutions of the problem have been exhausted. This would indeed be the case if the completeness of the disjunction between reflection and intuition as sources of knowledge were logically assured.

Now, it seems to be logically self-evident that a cognition that is not intuitive must arise from concepts and therefore from reflection, and vice versa, that a cognition we possess independently of reflection must pertain to intuition. This is true, to be sure, if we define intuition as nonreflective knowledge, but such a definition does not correspond to linguistic usage. According to general usage we understand "intuition" to mean a knowledge of which we are immediately conscious. But not every immediate cognition is necessarily a cognition of which we are immediately conscious. There is no contradiction in the assumption that a cognition that does not arise from reflection reaches our consciousness only through the mediation of reflection. Immediacy of cognition and immediacy of consciousness of the cognition are logically two different things. The illusion of the logical completeness of the disjunction between reflection and intuition as sources of knowledge arises only as a result of the confusion of these two concepts, in other words, as a result of the false conclusion of the immediacy of consciousness from the immediacy of cognition.

The demonstration of the logical incompleteness of this disjunction reveals to us a fourth possible solution of our problem. It consists in tracing back the metaphysical judgments to a cognition that belongs neither to reflection nor to intuition, that is, to a nonintuitive immediate cognition. I denote this solution, which follows from the criticism of the dogmatic disjunction of the sources of knowledge, as metaphysical criticism.

[Included at right is a diagram, such as Nelson employs elsewhere, that illustrates Nelson's argument for non-intuitive immediate knowledge. The original address does not include an axiomatic diagram. -- ed.]

The disjunction, expanded by our indication of the possibility of a nonintuitive immediate cognition, is in its turn logically secured. Since thereby the completeness of the above-considered possibilities of solution is now guaranteed, we can turn to the further question of how we are to decide between these various attempts at solution, i.e., which of the various logically possible theories is psychologically correct. With this question we leave the realm of purely logical criticism and turn to the testimony of inner experience. And here we can avail ourselves of work done long ago.

Both the positive solutions possible under the dogmatic assumption of the completeness of the disjunction -- metaphysical logicism and metaphysical mysticism -- have already been refuted by Hume.

Logicism, such as formed the basis of scholastic metaphysics and which has been revived in the "theory of knowledge," breaks down on the psychological fact of the indirectness and emptiness of reflection. Reflection can analyze and elucidate cognitions elsewhere provided but cannot of itself creatively beget new cognitions, that is to say, it is a source only of analytic, but not of synthetic, judgments.

Metaphysical mysticism, such as forms the basis for neo-Platonic mysticism in its old and new forms, breaks down on the psychological fact of the original obscurity of metaphysical knowledge. There is no immediate obviousness in metaphysical truths; we cannot derive these cognitions from an "intellectual intuition"; they reach our consciousness only through thinking (reflection), through abstracting from the intuitively given content of empirical judgments.

If, then, we are not permitted to seek the source of the metaphysical judgments either in intuition or reflection, two ways are still open to us: either we can dispute altogether the existence of metaphysical knowledge; or we can abandon the assumption of the exclusiveness of reflection and intuition as sources of knowledge and assert the existence of nonintuitive immediate cognition.

Hume sought the basic fallacy of the theories he refuted in their assumption that we possess metaphysical cognition at all; in this way he was led to his negative attempt at solution and hence to metaphysical empiricism. His task then became not that of verifying the metaphysical judgments but that of explaining psychologically the illusion that calls forth these judgments, i.e., of explaining how the claim to knowledge asserted in these judgments is possible without presupposing an actual source of knowledge, merely as a product of the blind mechanism of the association of ideas. The question now arises whether this task is capable of fulfillment.

Hume believed that he could find the basis for the judgments that were the object of his problem in the psychological principle of the expectancy of similar cases. But he was not unaware of the difficulty to be encountered in basing this principle on the laws of association. Association explains only that event A reminds me of a previous event B connected with it, but not that I expect the reoccurrence of B. The remembered thought, as such, is only problematic, whereas expectation comprises an assertion which -- whether it be one of certainty or only of probability -- cannot be explained by association alone. Hume tried to overcome this difficulty by presenting the difference between problematic and assertoric notions as merely one of degree, basing this difference on a difference in the intensity of the clarity of the notions. On this supposition our recollection would, indeed, after sufficiently frequent reproduction, be able to pass over into an expectation merely through the effectiveness of the association. But this Humean hypothesis, that the difference between problematic and assertoric notions is only one of degree, contradicts the facts of self-observation. This is generally admitted today; and so Hume's attempted solution collapses.

It is easy so to generalize this criticism of Hume's theory that through it any empirical attempt at solution, whatever its nature, is excluded. The problem lies in the actual existence of certain judgments through which we conceive a necessary connection of things. What is important here is not the assertion that the merely problematic thought of a necessary connection, which is to be found in these judgments, cannot be explained through association. To be sure, every connection of ideas must be explicable by the laws of association. What we must explain here, however, is not a connection of ideas but the idea of connection. This is in its content an entirely new conception vis-à-vis the ideas of that which, in this idea, is thought of as connected; accordingly, it can never arise from these other ideas through mere association but presupposes a source of knowledge of its own.

The critical analysis of metaphysical logicism and mysticism shows us that this source of knowledge can lie neither in reflection nor in intuition. If we now consider the above psychological analysis of empiricism together with the critical analnsis of both these other theories, we have proof of the correctness of the fourth theory, the only one that still remains, viz., criticism. The mere exclusion of the first two theories permitted us only the conclusion that, if we possess metaphysical knowledge at all, the existence of nonintuitive immediate knowledge must be assumed. But we were not yet able to claim that this condition is valid; rather, the possibility was still open to us of following Hume in the opposite direction and concluding, from the dogmatic disjunction of the sources of knowledge, that metaphysical knowledge is impossible. It is only after we have also refuted this empiristic consequence that we are able, in connection with the exclusion of metaphysical logicism and mysticism, to conclude the existence of noniiituitive immediate knowledge. We thus can, at the same time, supplement the proof of the logical incompleteness of the dogmatic disjunction with the proof of its psychological falsity.

At this juncture we perceive what we have gained from our previous logical analysis of the several possible solutions of the problem. In addition to the fact that this analysis at once precludes us from the contradictory attempt to find in the "theory of knowledge" a solution of the problem, it also prevented us from too hastily excluding from our range of choices what at first sight appeared to be a logically impossible course. The service that this analysis rendered us is all the more important since in our case the course that, by and large, has not even been considered heretofore is precisely the only one that really leads to a solution. Without this preparation through such a logical criticism one continually runs the danger of blinding oneself, through the deceptive illusion of the dogmatic disjunction, to the most obvious facts of self-observation. On the supposition of the completeness of this disjunction, as we have seen, the facts of the emptiness of reflection, the nonintuitive nature of metaphysical knowledge and the actual existence of metaphysical knowledge cannot be logically reconciled with one another because one of these facts always contradicts the consequences of the other two. And so -- as, moreover, the history of philosophy bears in on us -- without that critical analysis we are constantly driven hither and thither between these three equally necessary, but mutually contradictory, consequences. But the antinomy in which we thus become entangled is immediately resolved once we become aware of the prejudice that lies behind it, set aside all dogmatic suppositions, and look only the facts themselves in the eye.

We can accordingly epitomize the results of these critical observations as follows:

On the assumption of the exclusiveness of reflection and intuition as sources of knowledge, we have a choice only between metaphysical logicism, mysticism, and empiricism; that is to say, we have only the choice of contesting either the fact of the emptiness of reflection, or the fact of the nonintuitive nature of metaphysical knowledge, or the fact of the existence of metaphysical knowledge -- or, finally, we may conclude from these three facts that nonintuitive immediate knowledge exists.

In conclusion, let us consider what we have gained in all this compared to the "theory of knowledge."

If we must admit the possibility of metaphysics, we still need a criterion by which to distinguish between legitimate and spurious metaphysical assertions. But here we are exposed to the difficulty that this criterion must itself be metaphysical in nature since we know it can lie neither in reflection nor in intuition.

The wish to escape this difficulty is the real reason why recourse had to be taken to the "theory of knowledge." For, since metaphysics can obviously no more contain in itself the basis of the validity of its judgments than can any other discipline, this basis had to be sought in another, higher discipline, which, however, in its turn could no more derive its content from mere reflection or intuition than metaphysics itself could. It is therefore not surprising that so far no one who has gone into this higher discipline has been able to disclose its origins.

But the embarrassment of which this enigmatic science is to relieve us is merely a consequence of the confusion of knowledge with judgment. If we discriminate between judgment and immediate knowledge, the fact that the ground of the validity of metaphysical judgments must itself be metaphysical in nature will not lead us to conclude that this ground itself must lie in metaphysical judgments, but we shall seek it in immediate cognition. In this immediate cognition, not in a higher discipline, lies the basis of metaphysical judgments.

To be sure, this immediate cognition is not intuition. And precisely here we see the really decisive reason why psychological critique is so fruitful for metaphysics. For even though we find the basis of metaphysical judgments in an immediate cognition, we do not become immediately conscious of this cognition in such a way as would enable us to compare it directly with the metaphysical judgments in order to verify them. Rather, in order to execute this verification, i.e., to trace back the metaphysical judgments to the immediate knowledge that forms their ground, we must first ingeniously bring this immediate knowledge to light and therefore make it the object of psychological investigation.

Hence we need a special science to verify metaphysical judgments. But this science is no "theory of knowledge": it does not itself contain the basis of metaphysical judgments but only serves to bring it to light. For this very reason, also, the empirical and psychological character of this science is entirely compatible with the rational and metaphysical nature of the propositions it is to verify. For the basis of the metaphysical propositions does not lie in the assertions of this psychological critique but in immediate metaphysical knowledge.

Perhaps we can best clarify this relation by drawing an analogy from critical mathematics. In geometric axiomatics is to be found the proposition of the unprovability of the parallel postulate. Here, then, we must distinguish between two propositions: proposition A, the parallel postulate, and proposition B, which states that proposition A cannot be proved. Now, proposition B can be proved. There is nothing paradoxical in this, because A is a proposition from the system of geometry, whereas B belongs only to critique. B does not contain the ground of A but simply has A as its object. The situation in the critique of the metaphysical propositions is altogether analogous. Let us take, for example, the principle of causality, which we shall call C. Then psychological critique proves proposition D: There exists a nonintuitive immediate cognition which contains the ground of C. C is a proposition from the system of metaphysics and as such is rational; D is a proposition from psychological critique and as such is empirical. D does not contain the ground of C but simply has C as its object.

Of course, this positive importance of psychology for the verification of metaphysics can only be asserted from the viewpoint of criticism. A logicistic or mystic metaphysics has no need of psychology. Nevertheless -- and this should no longer be overlooked -- psychology has a negative importance for every sort of metaphysics (and anti-metaphysics!), which manifests itself in the fact that every metaphysics -- consciously or unconsciously -- comprises a psychological preassumption regarding its source of knowledge, in view of which every metaphysics must submit to criticism through comparison with the psychological facts. In this general psychological criticism we are provided with a limiting principle, by the help of which, over and beyond logical criticism, we can at least exclude all those metaphysical doctrines, consistent in themselves, which stand from the very start in contradiction to psychological facts.

Herewith, also, the dispute is transferred to a field more accessible to scientific treatment, and in which it is possible to work on common problems according to a common method. Only after we have succeeded in making truly manifest the importance of the general logical and psychological criticism that we have been considering, can we hope to launch a program of cooperative and fruitful scientific endeavor in philosophy in the place of divisive and barren dogmatic quarrels.

Leonard Nelson (1882-1927)


History of Philosophy

Home Page

Copyright (c) 2008 Kelley L. Ross, Ph.D. All Rights Reserved

The Impossibility of the 'Theory of Knowledge,'
by Leonard Nelson, Note

A comprehensive history of Erkenntnistheorie since Kant can be found in Nelson's "Über das sogenannte Erkenntnisproblem," Abhandlungen der Fries'schen Schule, II (Göttingen, 1908).

Return to text