WE HAVE NOW defined the idea of knowledge. In the act of cognition this idea is directly given in human consciousness. Both outer and inner perceptions, as well as its own presence are given directly to the “I,” which is the center of consciousness. (It is hardly necessary to say that here “center” is not meant to denote a particular theory of consciousness, but is used merely for the sake of brevity in order to designate consciousness as a whole.) The I feels a need to discover more in the given than is directly contained in it. In contrast to the given world, a second world — the world of thinking — rises up to meet the I and the I unites the two through its own free decision, producing what we have defined as the idea of knowledge. Here we see the fundamental difference between the way the concept and the directly given are united within human consciousness to form full reality, and the way they are found united in the remainder of the world-content. In the entire remainder of the world picture we must conceive an original union which is an inherent necessity; an artificial separation occurs only in relation to knowledge at the point where cognition begins; cognition then cancels out this separation once more, in accordance with the original nature of the objective world. But in human consciousness the situation is different. Here the union of the two factors of reality depends upon the activity of consciousness In all other objects, the separation has no significance for the objects themselves, but only for knowledge. Their union is original and their separation is derived from the union. Cognition separates them only because its nature is such that it cannot grasp their union without having first separated them. But the concept and the given reality of consciousness are originally separated, and their union is derived from their original separation; this is why cognition has the character described here. Just because, in consciousness, idea and given are necessarily separated, for consciousness the whole of reality divides into these two factors; and again, just because consciousness can unite them only by its own activity, it can arrive at full reality only by performing the act of cognition. All other categories (ideas), whether or not they are grasped in cognition, are necessarily united with their corresponding forms of the given. But the idea of knowledge can be united with its corresponding given only by the activity of consciousness. Consciousness as a reality exists only if it produces itself. I believe that I have now cleared the ground sufficiently to enable us to understand Fichte's Science of Knowledge through recognition of the fundamental mistake contained in it. Of all Kant's successors, Fichte is the one who felt most keenly that only a theory of consciousness could provide the foundation for knowledge in any form, yet he never came to recognize why this is so. He felt that what I have called the second step in the theory of knowledge, and which I formulated as a postulate, must be actively performed by the I. This can be seen, for example, from these words:
“The science of knowledge, insofar as it is to be a systematic science, is built up in the same manner in which all possible sciences, insofar as they are systematic, are built up, that is, through a determination of freedom; which freedom, in the science of knowledge, is particularly determined: to become conscious of the general manner of acting of the intelligence. ... By means of this free act, something which is in itself already form, namely, the necessary act of the intelligence, is taken up as content and put into a new form, that is, the form of knowledge or of consciousness. ...” 
What does Fichte here mean by the “acting of intelligence” if we express in clear concepts what he dimly felt? Nothing other than the production of the idea of knowledge, taking place in consciousness. Had Fichte become clear about this, then he would have formulated the above principle as follows: A science of knowledge has the task of bringing to consciousness the act of cognition, insofar as it is still an unconscious activity of the I; it must show that to objectify the idea of knowledge is a necessary deed of the I.
In his attempt to define the activity of the I, Fichte comes to the conclusion: “The I as absolute subject is something, the being (essence) of which consists merely in postulating its own existence.”  For Fichte, this postulation of the I is the primal unconditioned deed, “it is the basis of all consciousness.”  Therefore, in Fichte's sense too, the I can begin to be active only through an absolute original decision. But for Fichte it is impossible to find the actual content for this original activity postulated by the I. He had nothing toward which this activity could be directed or by which it could be determined. The I is to do something, but what is it to do? Fichte did not formulate the concept of knowledge which the I must produce, and in consequence he strove in vain to define any further activity of the I beyond its original deed. In fact, he finally stated that to investigate any such further activity does not lie within the scope of theory. In his deduction of representation, he does not begin from any absolute activity of the I or of the not-I, but he starts from a state of determination which, at the same time, itself determines, because in his view nothing else is, or can be contained directly in consciousness. What in turn determines the state of determination is left completely undecided in his theory; and because of this uncertainty, one is forced beyond theory into practical application of the science of knowledge.  However, through this statement Fichte completely abolishes all cognition. For the practical activity of the I belongs to a different sphere altogether. The postulate which I put forward above can clearly be produced by the I only in an act which is free, which is not first determined; but when the I cognizes, the important point is that the decision to do so is directed toward producing the idea of cognition. No doubt the I can do much else through free decision. But if epistemology is to be the foundation of all knowledge, the decisive point is not to have a definition of an I that is “free,” but of an I that “cognizes.” Fichte has allowed himself to be too much influenced by his subjective inclinations to present the freedom of the human personality in the clearest possible light. Harms, in his address, On the Philosophy of Fichte, (p. 15) rightly says: “His world-view is predominantly and exclusively ethical, and his theory of knowledge has no other feature.” Cognition would have no task to fulfill whatever if all spheres of reality were given in their totality. But the I, so long as it has not been inserted by thinking into the systematic whole of the world-picture, also exists as something merely directly given, so that it does not suffice to point to its activity. Yet Fichte is of the opinion that where the I is concerned, all that is necessary is to seek and find it. “We have to search for the absolute, first, and unconditioned fundamental principle of human knowledge. It cannot be proven nor determined if it is to be absolute first principle.”  We have seen that the only instance where proof and definitions are not required is in regard to the content of pure logic. The I, however, belongs to reality, where it is necessary to establish the presence of this or that category within the given. This Fichte does not do. And this is why he gave his science of knowledge a mistaken form. Zeller  remarks that the logical formulas by which Fichte attempts to arrive at the concept of the I only lightly hide his predetermined purpose to reach his goal at any cost, so that the I could become his starting point. These words refer to the first form in which Fichte presented his science of knowledge in 1794. When it is realized that, owing to the whole trend of his philosophy, Fichte could not be content with any starting point for knowledge other than an absolute decree, it becomes clear that he has only two possibilities for making this beginning appear intelligible. One possibility is to focus the attention on one or another of the empirical activities of consciousness, and then crystallize out the pure concept of the I by gradually stripping away everything that did not originally belong to consciousness. The other possibility is to start directly with the original activity of the I, and then to bring its nature to light through self-contemplation and self-observation. Fichte chose the first possibility at the beginning of his philosophical path, but gradually went over to the second.
On the basis of Kant's synthesis of “transcendental apperception” [The perception of an object involving the consciousness of the pure self as subject. (Translator)] Fichte came to the conclusion that the activity of the I consists entirely in combining the material of experience into the form of judgment. To judge means to combine predicate with subject. This is stated purely formally in the expression: a == a. This proposition could not be made if the unknown factor x which unites the two a's did not rest on an absolute ability of the I, to postulate. For the proposition does not mean a exists, but rather: if a exists, then so does a. In other words there is no question of postulating a absolutely. In order, therefore, to arrive at something which is valid in a quite straightforward way, the only possibility is to declare the act of postulating as such to be absolute. Therefore, while a is conditional the postulation of a is itself unconditional. This postulation, however, is a deed of the I. To the I is ascribed the absolute and unconditional ability to postulate. In the proposition a == a, one a is postulated only because the other a is already postulated, and indeed is postulated by the I. “If a is postulated in the I, then it is postulated, or then it is.”  This connection is possible only on condition that there exists in the I something which is always constant, something that leads over from one a to the other. The above mentioned x is based on this constant element. The I which postulates the one a is the same as the I which postulates the other a. This means that I == I. This proposition expressed in the form of a judgment: If the I exists, then the I exists, is meaningless. The I is not postulated by presupposing another I; it presupposes itself. This means: the I simply is, absolutely and unconditionally. The hypothetical form of a judgment, which is the form of all judgments, when an absolute I is not presupposed, here is transformed into a principle of absolute existence: I simply am. Fichte also expresses this as follows: “The I originally and absolutely postulates its own being.”  This whole deduction of Fichte's is clearly nothing but a kind of pedagogical discussion, the aim of which is to guide his reader to the point where knowledge of the unconditional activity of the I dawns in him. His aim is to bring the activity of the I emphatically home to the reader, for without this activity there is no I.
Let us now survey Fichte's line of thought once more. On closer inspection one sees that there is a break in its sequence; a break, indeed, of a kind that casts doubt upon the correctness of his view of the original deed of the I. What is essentially absolute when the I postulates? The judgment is made: If a exists, then so does a. The a is postulated by the I. There can, therefore, be no doubt about the postulation as such. But even if the I is unconditioned insofar as its own activity is concerned, nevertheless the I cannot but postulate something. It cannot postulate the “activity, as such, by itself,” but only a definite activity. In short: the postulation must have a content. However, the I cannot derive this content from itself, for by itself it can do no more than eternally postulate its own postulation. Therefore there must be something which is produced by this postulation, by this absolute activity of the I. Unless the I sets to work on something given which it postulates, it can do “nothing” and hence cannot postulate either. Fichte's own principle actually shows this: The I postulates its existence. This existence is a category. This means we have arrived at our principle: The activity of the I is to postulate, as a free decision, the concepts and ideas of the given. Fichte arrives at his conclusion only because he unconsciously sets out to prove that the I “exists.” Had he worked out the concept of cognition, he would then have arrived at the true starting point of a theory of knowledge, namely: The I postulates cognition. Because Fichte is not clear as to what it is that determines the activity of the I, he simply characterizes this activity as the postulation of being, of existence. In doing so, he also limits the absolute activity of the I. If the I is only unconditioned in its “postulation of existence.” everything else the I does must be conditioned. But then, all possible ways to pass from what is unconditioned to the conditioned are blocked. If the I is unconditioned only in the one direction described, it immediately ceases to be possible for the I to postulate, through an absolute act, anything but its own being. This makes it necessary to indicate the basis on which all the other activities of the I depend. Fichte sought for this in vain, as we have already seen.
This is why he turned to the other of the two possibilities indicated for deducing the I. As early as 1797, in his First Introduction to the Science of Knowledge, he recommends self-observation as the right method for attaining knowledge of the essential being of the I:
“Be aware of yourself, withdraw your attention from all that surrounds you and turn it toward your inner being — this is the first demand that philosophy makes on the pupil. What is essential is not outside of you, but solely within yourself. 
To introduce the science of knowledge in this way is indeed a great advance on his earlier introduction. In self-observation, the activity of the I is actually seen, not one-sidedly turned in a particular direction, not as merely postulating existence, but revealing many aspects of itself as it strives to grasp the directly given world-content in thinking. Self-observation reveals the I engaged in the activity of building up the world-picture by combining the given with concepts. However, someone who has not elaborated the above considerations for himself — and who therefore does not know that the I only arrives at the full content of reality when it approaches the given with its thought-forms — for him, the process of knowledge appears to consist in spinning the world out of the I itself. This is why Fichte sees the world-picture more and more as a construction of the I. He emphasizes ever more strongly that for the science of knowledge it is essential to awaken the faculty for watching the I while it constructs the world. He who is able to do this appears to Fichte to be at a higher stage of knowledge than someone who is able to see only the construction, the finished product. He who considers only the world of objects does not recognize that they have first been created by the I. He who observes the I while it constructs, sees the foundation of the finished world-picture; he knows the means by which it has come into being, and it appears to him as the result of presuppositions which for him are given. Ordinary consciousness sees only what is postulated, what is in some way or other determined; it does not provide insight into the premises, into the reasons why something is postulated in just the way it is, and not otherwise. For Fichte it is the task of a completely new sense organ to mediate knowledge of these premises. This he expresses most clearly in his Introductory Lecture to the Science of Knowledge, delivered at Berlin University in the autumn of 1813:
“This science presupposes a completely new inner sense organ, through which a new world is revealed which does not exist for the ordinary man at all.” “The world revealed by this new sense, and therefore also the sense itself, is so far clearly defined: it consists in seeing the premises on which is based the judgment that ‘something is’; that is, seeing the foundation of existence which, just because it is the foundation, is in itself nothing else and cannot be defined.” 
Here too, Fichte lacks clear insight into the content of the activity carried out by the I. And he never attained this insight. That is why his science of knowledge could never become what he intended it to be: a philosophical foundation for science in general in the form of a theory of knowledge. Had he once recognized that the activity of the I can only be postulated by the I itself, this insight would also have led him to see that the activity must likewise be determined by the I itself. This, however, can occur only by a content being given to the otherwise purely formal activity of the I. As this content must be introduced by the I itself into its otherwise quite undetermined activity, the activity as such must also be determined by the I itself in accordance with the I's own nature. Otherwise its activity could not be postulated by the I, but at most by a “thing-in-itself” within the I, whose instrument the I would be. Had Fichte attempted to discover how the I determines its own activity, he would have arrived at the concept of knowledge which is to be produced by the I. Fichte's science of knowledge proves that even the acutest thinker cannot successfully contribute to any field of knowledge if he is unable to come to the right thought-form (category, idea) which, when supplemented by the given, constitutes reality. Such a thinker is like a person to whom wonderful melodies are played, but he does not hear them because he lacks an ear for music. Consciousness, as given, can be described only by someone who knows how to take possession of the “idea of consciousness.”
Fichte once came very near the truth. In his Introduction to the Science of Knowledge (1797), he says that there are two theoretical systems: dogmatism — in which the I is determined by the objects; and idealism — in which the objects are determined by the I. In his opinion both are possible world-views. Both are capable of being built up into a consistent system. But the adherents of dogmatism must renounce the independence of the I and make it dependent on the “thing-in-itself.” For the adherents of idealism, the opposite is the case. Which of the two systems a philosopher is to choose, Fichte leaves completely to the preference of the individual. But if one wishes the I to retain its independence, then one will cease to believe in external things and devote oneself to idealism.
This line of thought fails to consider one thing, namely that the I cannot reach any choice or decision which has some real foundation if it does not presuppose something which enables it to do so. Everything determined by the I remains empty and without content if the I does not find something that is full of content and determined through and through, which then makes it possible for the I to determine the given and, in doing so, also enables it to choose between idealism and dogmatism. This something which is permeated with content through and through is, however, the world of thinking. And to determine the given by means of thinking is to cognize. No matter from what aspect Fichte is considered, we shall find that his line of thought gains power and life when we think of the activity of the I, which he presents as grey and empty of content, as filled and organized by what we have called the process of cognition.
The I is freely able to become active of itself, and therefore it can also produce the category of cognition through self-determination; in the rest of the world, by objective necessity the categories are connected with the given corresponding to them. It must be the task of ethics and metaphysics to investigate the nature of this free self-determination, on the basis of our theory of knowledge. These sciences will also have to discuss whether the I is able to objectify ideas other than those of cognition. The present discussion shows that the I is free when it cognizes, when it objectifies the ideas of cognition. For when the directly given and the thought-form belonging to it are united by the I in the process of cognition, then the union of these two elements of reality — which otherwise would forever remain separated in consciousness — can only take place through a free act.
Our discussion sheds a completely new light on critical idealism. Anyone who has acquainted himself intimately with Fichte's system will know that it was a point of vital importance for this philosopher to uphold the principle that nothing from the external world can enter the I, that nothing takes place in the I which is not originally postulated by the I itself. Yet it is beyond all doubt that no idealism can derive from the I that form of the world-content which is here described as the directly given. This form of the world-content can only be given; it can never be constructed out of thinking. One need only consider that if all the colors were given us with the exception of one single shade, even then we could not begin to provide that shade out of the I alone. We can form a picture of distant regions that we have never seen, provided we have once personally experienced, as given, the various elements needed to form the picture. Then, out of the single facts given us, we combine the picture according to given information. We should strive in vain to invent for ourselves even a single perceptual element that has never appeared within our sphere of the given. It is, however, one thing merely to be aware of the given world: it is quite another to recognize its essential nature. This latter, though intimately connected with the world-content, does not become clear to us unless we ourselves build up reality out of the given and the activity of thinking. The essential What of the given is postulated for the I only through the I itself. Yet the I would have no occasion to postulate within itself the nature of something given if it did not first find itself confronted by a completely undetermined given. Therefore, what is postulated by the I as the nature and being of the world is not postulated without the I, but through it.
The true shape is not the first in which reality comes before the I, but the shape the I gives it. That first shape, in fact, has no significance for the objective world; it is significant only as a basis for the process of cognition. Thus it is not that shape which the theory of knowledge gives to the world which is subjective; the subjective shape is that in which the I at first encounters it. If, like Volkelt and others, one wishes to call this given world “experience,” then one will have to say: The world-picture which, owing to the constitution of our consciousness, appears to us in a subjective form as experience, is completed through knowledge to become what it really is.
Our theory of knowledge supplies the foundation for true idealism in the real sense of the word. It establishes the conviction that in thinking the essence of the world is mediated. Through thinking alone the relationship between the details of the world-content become manifest, be it the relation of the sun to the stone it warms, or the relation of the I to the external world. In thinking alone the element is given which determines all things in their relations to one another.
An objection which Kantianism could still bring forward would be that the definition of the given described above holds good in the end only for the I. To this I must reply that according to the view of the world outlined here, the division between I and external world, like all other divisions, is valid only within the given and from this it follows that the term “for the I” has no significance when things have been understood by thinking, because thinking unites all opposites. The I ceases to be seen as something separated from the external world when the world is permeated by thinking; it therefore no longer makes sense to speak of definitions as being valid for the I only.