I. The Reception Of Kant’s Account Of Memory In Current Literature

In the last two decades, few commentators have recognized the importance of memory in Kant’s thought and recent studies on the empirical and transcendental function of memory are very scarce or lack profundity¹. It is, nonetheless, worth underlining some commentators like Herbert James ^{Paton (1936)} who defended the role of memory in the A edition of the Critique of Pure Reason and, particularly, in the “synthesis of recognition”². P. F. ^{Strawson claimed later, in 1966}, that experience and memory emerge together and that memory is involved in experience, recognition, consciousness, and identity of the self. More recently, Andrew ^{Brook (1994)} suggests that the transcendental function of memory may take place in the acts of the transcendental apperception. ^{Howard Caygill (1995)} maintains that memory is implied in two of the three syntheses of the ‘transcendental faculty of imagination’, namely in the synthesis of apprehension and recognition. Finally, Angelica ^{Nuzzo (2015}) argues for the seminal role of memory in the synthesis of recognition.

Gordon Nagel observes that “one of the striking “omissions” in Kant’s theory of experience is any significant role of memory in cognition. Memory, which figures so large in Locke and Hume, does not figure at all in the Critique” (^{Nagel, 1983}, p. 215)³. To be precise, in his few explicit references to memory in the CPR, ^{Kant (1998)} claims that memory constitutes an empirical condition under which our understanding is exercised.

Few commentators have explored in detail the possible role of memory in the a priori conditions of experience. For instance, G. Nagel is reluctant to admit the transcendental functions of memory in experience. Instead, he reduces memory to an empirical function of storing or accumulation of sense materials which cannot be integrated into development of knowledge. Thus, he regards memory as unnecessary on the grounds that ‘the manifold of appearance’ occupies the same place as memory (memory taken for a reconstruction). This dismissal of memory is partly motivated by a self-sufficient signification of the concept ‘appearance’, according to which “appearances are the stable correlates of the flux of sensory input. The buildup of knowledge is not the accumulation of sensory fact upon sensory fact, but the continual addition of details to a structure that obtains a priori” (^{Nagel, 1983}, p. 215). However, I think that the replacement of ‘accumulation’ with ‘continual addition’ does not mean that memory has necessarily a superfluous character, but it may be regarded as that a priori structure that prevents the appearance from losing (forgetting) every added detail in the flux of the sensory input. Nagel, nonetheless, ascribes a purely passive role of memory to Kant, by declaring that “the manifold of appearance is an active file, rather than an attic crammed with memories. It is under constant revision” (^{Nagel, 1983}, p. 215). In contrast, I shall argue (in the sixth section) that memory plays an active role in experience, via imagination, by storing and reproducing necessary items for experience.

P. F. Strawson suggests a more positive view on the role of memory in Kant’s account of experience as he points out a mutual dependence between memory and experience: “if experience is impossible without memory, memory also is impossible without experience. From whatever obscure levels they emerge they emerge together” (^{Strawson, 1966}, p. 112). Strawson says that memory is involved in “experience, recognition, consciousness, of identity of self through diversity of experience” (^{Strawson, 1966}, p. 111). However, he does not explain how memory and these elements are connected with experience.

I think that the contribution of these commentators to the study of Kant’s view of memory is very partial, because they do not analyze in detail the role of memory in experience. Instead, they restrict their analysis to the CPR, overlooking places of the Anthropology⁴ where the relation between memory and experience is discussed. Given this restriction, my analysis of memory is much broader and richer than those undertaken hitherto.

II The Empirical Function Of Memory In Cognition

According to ^{Kant (2007a}), memory is a faculty that involves fundamentally two distinct activities, namely to store and to reproduce representations. These functions are based on experience and reveal the importance of memory in human cognition. In general terms, the capacity of people’s memory (without mental deficiencies or illnesses) to achieve these two acts varies according to their interest. For the more interesting an object is for the human being, the easier it will be to store it and reproduce it afterwards. Kant suggests that “one must occupy the memory only with those things which for us are important to remember and which have a relation to real life” (^2007b, AA 9:473). Thus we should not memorize speeches or things that we learn for a short time (like a future examination), but rather we should memorize things that lead to one’s self-improvement.

First, the storing function of memory consists in the capacity to preserve the existence of different kinds of representations over time, or, in his own words: “to grasp something methodically (memoriae mandare) is called memorizing” (^{Kant, 2007a}, AA 7:183). Particularly, Kant’s distinction between ‘to memorize’ (memorieren) and ‘to study’ (studieren) is at the basis of the difference between learning philosophy and learning to philosophize. Learning philosophy means to learn thoughts or to imitate the judgements of others by memorizing them, whereas the human being who learns to philosophize ought to be led, not to be carried. The human being should be helped to walk on its own in the future and to exercise its understanding by using its own understanding. In fact, Kant thinks of philosophy from a practical point of view not as a ‘learning’ (Gelehrsamkeit) i.e. “the sum total of the historical sciences” (¹⁹⁹⁶, AA 5:138 footnote), but rather as the science of the final ends of human reason (¹⁹⁰⁰, AA 2:306; 25:978; 24:698; 9:25; 1998, CPR A838/B866).

Of course, Kant recognizes that learning can be involved in the act of understanding something and storing it in memory as he holds that catechist and historical knowledge, most of the times, is simply ‘assimilated’ (¹⁹⁰⁰, AA 24:117, 149-50, 844). Mathematical knowledge, by contrast, involves a more active role of the understanding, in so far as it is possible “to learn according to both types of knowledge. That is to say, it is possible to impress either on the memory or on the understanding that which can be presented to us as an already complete discipline” (¹⁹⁰⁰, AA 2:307). Kant favored the second kind of knowledge, arguing that philosophy is not a complete discipline that should be memorized. Philosophy is, rather, a discipline that aims to extend the capacity of the understanding in young people, by means of the method of enquiry, to the extent that they become able to acquire a more mature ‘insight’ (Einsicht) of their own (¹⁹⁰⁰, AA 2:307; 28:531, 534; 25:1037). However, this second type of knowledge does not exclude memory as it entails a comprehension of what is memorized. Thus, Kant does not disregard the role of memory within knowledge, but rather he suggests that the act of storing representations should be accompanied by the use of the understanding (¹⁹⁰⁰, AA 25:1274, 979, 555).

Both Rousseau and Kant defend the crucial role of memory in human cognition, for the former holds that “although memory and reasoning are two essentially different faculties, nevertheless the one develops truly only with the other” (^{Rousseau, 1979},p. 107). Similarly, Kant claims that “the use of reason is very necessary. For in order to learn, one needs memory and understanding, to apply what one has learned one needs judgment” (¹⁹⁰⁰, AA 25:1481). In other words, Kant does not reject the storing of information but rather he suggests that such storing is a necessary condition for developing our human cognition properly (2007a, AA 7:184; 1900, 25:1274). This storing, nonetheless, should be accompanied by an appropriate use of the information by choosing, through the faculty of the power of judgement, the most proper knowledge according to every particular case. For pragmatic anthropology seeks to help the human being at both practical and theoretical levels, by increasing one’s capacity for memory. On this point, I am in agreement with Holly L. Wilson’s claim that “insofar as the anthropology teaches students how to increase their skills, it contributes to the useful development of the technical predisposition. Anthropology can also tell one what one ought not to do” (^{Wilson, 2011}, p. 168). Kant’s approach to different ways of memorizing seeks the improvement of the human being’s cognitive faculty, which is essential for its life in community with others.

Near the beginning of the Mrongovius anthropology lectures from 1784-1785, Kant claims that “memory is like an archivarius. A memory can be artfully organized if one places all representations in certain scientific fields where they belong; this is memoria localis” (¹⁹⁰⁰, AA 25:1273). Memory is not merely a passive receptacle, in which our representations are stored, but rather it is actively responsible for the distribution of the representations that will be retained and recalled later.

In the Logik Blomberg, Kant shows an outstanding interest for the key role of memory in cognition, as he points out that a large part of our cognitions (historical, geographical, physical, etc.) arises through belief, without which we would have no greater cognitions than those of the time in which we live, or those of the place where we live. It means that our cognition would be limited by our immediate experience. In contrast, our historical cognition is based on: a) the very experience of other people, as long as they are not in an extensive deception of the senses but they should be reliable, b) the non-defective memory, because we would not believe those whose imagination brings forth other images that do not belong to memory, c) capability to communicate their past experiences, so that these people should express themselves and communicate their experiences rightly and, finally, d) honorability and inclination toward truth (¹⁹⁰⁰, AA 24:245-6).

Kant underlines the key role of memory for the human being at both theoretical and practical levels, otherwise his analysis would not be pragmatic. H. L. Wilson has correctly suggested that “what makes anthropology pragmatic is the use that it brings with it. It is useful to have more memory (…) knowing how to use a mnemonic device will increase one’s capacity for memory” (Wilson, 2011, p. 168). Memory becomes a crucial faculty of the human being, which is enhanced by both didactic practices of education and one’s moral aspirations (^{Svare, 2006}).

III Kinds Of Memorization

Kant points out that to memorize is to ‘impress’ (einprägen) something in memory and distinguishes among three different kinds of memorizing: mechanical, ingenious and judicious (2007a, AA 7:183; 1900, 25:976). I suggest that Rudolf A. Makkreel is incorrect in his assertion that “the first method is rejected as too cumbersome and the second as unreliable. What is needed is a remedial third method of judicious memorising” (^{Makkreel, 2014}, p. 34). I think that Kant is not formulating a prescriptive judgement here, according to which the first two ways should be vanished; instead, Kant is describing different ways in which memory works. Indeed, Kant does not deny that ‘mechanical’ (mechanisch) memorizing plays a relevant role, for instance, in our ability to recite poems, neither that ingenious memorizing always makes less difficult the act of recalling, for instance when we associate someone’s name with a familiar object in order to recall it more easily.

Accordingly, the mechanical memorizing is based on the frequent ‘literal’ (buchstäblich) repetition. That is, when one performs the same action frequently, such action forms particular contents that are stored by memory in the same order in which they were experienced. This form is evident in the traditional way of learning multiplication tables and it is also relevant in sciences, principally in history of epochs (1900, AA 25:1463). To this form of memorizing belongs the recital of poems, the set of procedures demanded by a plane’s regular take-off and other actions, in which one must learn or store information with an exact order. Furthermore, the mechanical storing of contents in memory has its counterpart in the mechanical reproduction of memory contents. For it is a fact that memory has the power to reproduce or evoke contents with great accuracy, without a comprehension of what has been stored:

What he merely learns, and thus entrusts to his memory, he performs only mechanically (according to laws of reproductive imagination) and without understanding. A servant who has merely to pay a compliment according to a definite formula needs no understanding, that is, he does not need to think for himself. (^{Kant, 2007a}, AA 7:197)

Kant suggests that common people often make use of this kind of memorizing, when they entrust various things, as lined up, to their minds, so that they can remember them and carry them out in succession. Thus, the order in which representations are memorized makes easier the preservation of the order in which they are reproduced, because no reasoning interferes with it. This is also suggested by Kant in the Blomberg Logic (1770s) where he holds that memory is concerned with no rational cognition, namely representations in which reason is not applied, which is not on that account irrational either (1900, AA 24:47). Thus, scholar people let many of his tasks or domestic affairs escape through distraction, because they have not caught them with enough attention. However, if there is not a good mechanical reliable memory, the art of writing can compensate for this deficiency as the latter has the power to recover precisely and without effort everything that should have been stored in the mind (1900, AA 7:184-5)⁵.

Kant shows the positive and the negative sides of this form of memorizing for the biological development of the human being. On the one hand, this form of memorizing is not highly recommended in young people, in as much as it hinders the participation of the understanding in learning. In this case, the understanding cannot act nor ‘make concepts’ (Begriffe machen) but is merely inactive. On the other hand, this form of memorizing is not merely useful to obtain historical knowledge but it becomes indispensable for the human understanding, since memory provides the stuff in which the understanding occupies later. If memory is not well furnished, the understanding will be ‘poor’ (arm) and will not have any stuff, provided by senses nor reproduced by memory, for working (1900, AA 25:976, 92, 521, 1273). Memory of young people, by contrast, can retain more easily than old people’s one because it is more receptive (1900, AA 24:522, 816).

To sum up, this form of memorizing involves the following essential characteristics: i) the subject is fully aware of the stored (memorized) representations, ii) the stored representations are short-term memories, iii) these do not demand comprehension, iv) these contain practical information that has a relatively immediate application and v) these contain linguistic properties, in so far as most of the Kant’s examples are related to word for word repetition.

The second form of memorizing is called ‘ingenious’ (ingeniös), by means of which we store a representation, by associating it with others already stored: “ingenious memorizing is a method of impressing certain ideas on the memory by association with correlative ideas that in themselves (as far as understanding is concerned) have no relationship at all with each other” (^{Kant, 2007a}, AA 7:183). This memorizing does not consist in a mere storing of representations, as it entails also an association of representations, in which we match the newly-stored representation with other similar previously-stored representation, by means of their ‘comparison’. For instance, we memorize more easily someone’s name by associating it with the name of a quite known song or a certain familiar object (1900, AA 25:977, 1274; 29:757-8). Moreover, Kant holds that ingenious memorizing is not a ‘ruleless procedure’ (regelloses Verfahren) of imagination but it involves rather a method according to which we pair together things that are not contained by the same concept in order to catch something in memory more easily. In my opinion, this form of memorizing exhibits the following essential characteristics: i) awareness of the already stored representations and of the representations that are to be memorized, ii) a comparison of stored and storable representations and iii) linguistic properties, in as much as these representations require the association of sounds of language.

The third form of memorizing, called ‘judicious’ (judiciös), is not much concerned with the way in which certain representations are memorized but rather with the kind of information stored. According to this form, we memorize a system by enumerating and retaining its linked parts, so that if one of these parts were forgotten we could recall it by remembering the connection of that forgotten part with others still retained in memory:

Judicious memorizing is nothing other than memorizing, in thought, a table of the divisions of a system (for example, that of Linnaeus) where, if one should forget something, one can find it again through the enumeration of the parts that one has retained; or else through memorizing the sections of a whole made visible (for example, the provinces of a country on a map, which lie to the north, to the west, etc.). (2007a, AA 7:184)

This form of memorizing does not involve isolated representations but representations located in certain relations, for it consists in the storing of the specific relation among some representations⁶. For instance, we remember the place of a book more easily, when we classify all books according to a framework for universal concepts (called common places) and we put them under certain labels⁷.Kant holds that judicious memorizing requires functions of the understanding which are useful for imagination. Therefore, if some representations are stored via this form of memorizing, they will exhibit the following essential characteristics: i) they require the activity of the understanding and of imagination, ii) they are useful for an immediate-practical activity as well as a theoretical one (e.g. useful for botanic), iii) they are long-term memories and iv) they involve linguistic properties, in as much as the latter two characteristics presuppose an enduring classification through a framework for general concepts.

IV. Memory And Imagination

Kant, under the influence of Wolff, defines the “power of imagination’ (Einbildungskraft) as ‘a faculty of intuition without the presence of the object” (2007a, AA 7:167; 1998, CPR A120 footnote, B151-2)⁸.Kant distinguishes between productive and reproductive imagination; the first one is a faculty of the original presentation of the object, which not only precedes but also makes experience possible as it grounds space and time. The second one, by contrast, is concerned with a derivative presentation of empirical objects. The reproductive imagination contains both fantasy and memory, in as much as they demand the act of recalling representations. Fantasy, understood as an ‘inventive’ (dichtend) power of imagination, produces images based on ‘sense representation’ (Sinnenvorstellung) that was previously given to our faculty of sense (2007a, AA 7:167-8; 1900, 25:981). Unlike memory, imagination does not have to relate representations in a temporal order that corresponds necessarily to the temporal order in which the events were experienced (1900, AA 29:881). On this basic point, I am in agreement with R. Bader on considering that “the memory is what one is aware of and which exists NOW, whereby the memory has representational content consisting in a temporally ordered sequence of remembered mental episodes” (^{Bader, 2017}, p. 134). Indeed, Memory allows us to make causal judgements about past episodes in our life, by connecting some earlier events (cause) with other later ones (effect). For instance, I can infer via memory that the displeasure of a friend was caused by an earlier negative comment expressed by me.On Kant’s picture, remembrances and imaginary ideas (fictions) should be differentiated by the faithful or unfaithful character of their content rather than by the way in which they appear in the mind (^{Dietrich, 1991}). For in both cases some representations appear in the mind as soon as they are recalled:

The power of imagination is richer and more fruitful in representations than sense, when a passion appears on the scene the power of imagination is more enlivened through the absence of the object than by its presence. This is evident when something happens that recalls the representation of an object to the mind again, which for a while seemed to be erased through distractions. (2007a, AA 7:180)

Imagination contents are initially extracted from experience but afterwards they are associated in different ways to form ‘new’ representations (1900, AA 28:236; 29:884); for instance, we imagine a stove with ears, by associating two or more images derived from experience. Kant underlines that “no matter how great an artist, even a sorceress, the power of imagination may be, it is still not creative, but must get the material for its images from the senses” (2007a, AA 7:168-9; 1998, CPR B278). It follows that a person who was born blind cannot make any colors comprehensible, because its imagination does not have the power to produce a representation without using the material that was previously given to the faculty of sense. Of course, Kant is aware of the fact that these images of imagination do not have reference necessarily to an actual external object, but they are remembrances of empirical intuitions, which can hardly be universally communicated (2007a, AA 7:168-9).

Memory must be distinguished from imagination (understood as fantasy) by the fact that memory contents should be faithful and should reproduce intuitions (‘intuitive remembrances’) as they were arranged in the original conditions:

Memory is distinguished from the merely reproductive power of imagination in that it is able to reproduce the former representations voluntarily, so that the mind is not a mere plaything of the imagination. Fantasy, that is, creative power of imagination, must not mix in with it, because then memory would be unfaithful. (2007a, AA 7:182)

Accordingly, forgetfulness is not considered as a sort of unfaithful memory. The term unfaithful means rather to remember falsely something, that is to say, to remember the occurrence of something that never happened (1900, AA 25:1463, 980). Again, memory should preserve the same temporal order of the originally stored representations.

However, in the Mrongovius anthropology lectures, Kant describes memory as tantamount to the power of imagination that aims at present time (1900, AA 25:1277). Afterwards, in the Busolt anthropology lectures from 1788-1789, he claims that memory is the faculty of the power of imagination to reproduce representations which one already has (1900, AA 25:1462). In the CPR, Kant confesses that it is unclear whether these faculties are identical:

Initially a logical maxim bids us to reduce this apparent variety as far as possible by discovering hidden identity through comparison and seeing if imagination combined with consciousness may not be memory, wit, the power to distinguish, or perhaps even understanding and reason. (1998, CPR A649/B677)

This confusion concerning the boundaries between memory and imagination is puzzling, as Kant’s descriptions of memory and imagination share similar aspects. For instance, both are tightly connected with sensibility and they are capable of forming associations among empirical representations. This has led John Llewelyn naturally to claim that

Kant’s singling out of reproduction as a power specifically of imagination reflects the already mentioned duality in the history of philosophy, according to which memory is sometimes listed as a faculty in its own right and sometimes subsumed under imagination as one of the ways in which the latter represents something absent. (^{Llewelyn, 2000}, p. 107)

As Llewelyn notices, both imagination and memory rely on material derived from senses, although they are engaged with representations that do not demand the current presence of objects in sensibility⁹.I suggest that the compelling boundary between memory and imagination does not consist much in the act through which the mind generates memory and imagination contents, but rather in the formal (temporal order) and material (material derived from senses) character of their content. That is to say, memory contents should correspond accurately to the events informed by our experience.

Furthermore, Kant distinguishes in several places of his lectures on anthropology between memory and imagination and suggests that memory’s functions depend upon imagination (2007a, AA 7:182; 1900, 25:1289, 1464, 511, 974, 1023). Memory is described as a faculty that, only via imagination, has the power to achieve three ‘actions’ (Handlungen) in the reproduction of representations, namely to grasp something, to retain it and to remember it (1900, AA 25:89).

Memory is not a self-sufficient power of the human being but is linked to the ‘re-collective’ (zurückrufend) power of imagination which brings back to the mind representations that it had previously (2007a, AA 7:167). Kant stresses that imagination and memory are determined by ‘choice’ (Willkühr), so that when we want to remember, our choice enhances our imagination or memory to recall representations of previous states (1900, AA 25:974). As a result, memory is nothing but “the capacity to avail oneself of one’s reproductive power of imagination voluntarily” (1900, AA 25:1273). On the contrary, fantasy involuntarily brings back to the mind previous images, just like a stream of images that flow incessantly (1900, AA 25: 521,314, 87; 15:126; 28:237; 2007a, 7:174-5, 180).

Furthermore, memory leans on the reproductive imagination, as long as the latter contains certain characteristics that are relevant for memory processes (^{Stephenson, 2017}). In this vein, reproductive imagination is nothing but “a faculty of the derivative presentation of the object (i.e. exhibitio derivativa), which brings back to the mind an empirical intuition that it had previously” (2007a, AA 7:167). It means that memory via reproductive imagination recollects representations that we had previously (1900, AA 25:974, 1464, 521), although the imagination’s power to reproduce past representations does not happen by chance but is conditioned by a contingent law of association among our representations (1900, AA 28:236; 29:883; 25:1272-3; 1998, CPR B152). This law of association states that “empirical ideas that have frequently followed one another produce a habit in the mind such that when one idea is produced, the other also comes into being” (2007a, AA 7:176). For instance, when we look at the scars of one’s body, we tend to reproduce, through the reproductive activity of imagination, memories or images of the circumstances in which these scars happened (1900, AA 25:1023; 2007a, 7:176). This has led P. Kitcher to take memory for an example of “a synthetic connection between states, since the contents of the later depend upon those of the earlier state. By contrast transcendental syntheses are those syntheses governed by nonassociative rules that are necessary for knowledge” (^{Kitcher, 1990}, p. 254 footnote 14). This synthetic connection is certainly an empirical act performed by imagination which differs from the transcendental synthesis achieved by the understanding.

Kant had already recognized the reliance of memory on imagination in the Friedländer lectures from 1775/1776, where he holds that memory is a faculty of imitation (i.e. reproductive image formation) which produces images (1900, AA 25:511). In brief, remembrances are images, reproduced by imagination, of the previous state of an object. The images derived from the apprehension of the state of the object are stored in memory and then reproduced by means of imagination. However, the remembrances of that previous state do not necessarily correspond to the current state of the changing object. In other words, we form images of what (objects) strikes the senses through sensibility, and these images become current remembrances¹⁰ of past states of the object, which may no longer exist.

V. The Transcendental Function Of Memory

In the first edition of the “Transcendental Deduction” (TD), Kant explains how a completely a priori concept can be related to an object of a possible experience. He is presenting here “the transcendental constitution of the subjective sources that comprise the a priori foundations for the possibility of experience.” (1998, CPR A97). In his view, the relation between the manifold of an intuition of an object and its corresponding a priori concept becomes possible by means of a threefold synthesis, which is necessarily found in all cognition. My aim is to argue that memory plays a role in these syntheses, but it is necessary first to consider previous approaches to this problem.

In “Self-Identity” (1929), H. J. Paton suggests that experience relies on ‘one’ synthesis that demands memory functions. Concerning the threefold synthesis, he emphatically asserts that “there is really only one synthesis-which we are describing here in fragments (abgesondert und einzeln). I doubt also whether it is necessary to speak of a reproduction in imagination. All that we require is memory” (^{Paton, 1929}, p. 316). Unfortunately, Paton’s article is not focused on Kant’s account of memory but rather on other authors, such as Bertrand Russell or Charlie Dunbar Broad. Later on, in 1936, Paton defended the role of memory in the synthesis of recognition, by which the manifold of intuition is “combined” into one object (^{Paton, 1936}). He even suggests that all knowledge of objects demands not only to reproduce but also to remember a series of given appearances, which refers to the order in which these appearances were given to the mind (^{Paton, 1936}). In a similar manner, Howard Caygill endorses the relevance of memory for the syntheses of apprehension and recognition:

Memory is implied in two of the three syntheses of the ‘transcendental faculty of imagination’ presented in the deduction of CPR: in the ‘synthesis of apprehension’ where it informs the consistency of appearances, and in the ‘synthesis of recognition’ where it is implied in the continuity of the consciousness of appearances. (^{Caygill, 1995}, pp. 290-1)

However, Caygill’s study of memory is too brief and cannot explain in detail how memory and imagination are connected in experience. In this vein, Andrew Brook claims that “in TD, he [Kant] hardly mentions memory (his discussion of the synthesis of recognition is a rare exception and even there memory appears only by implication)” (^{Brook, 1994}, p. 186). Although he does admit that “many, perhaps all acts of TA [transcendental apperception] make use of memory (…) to have certain sorts of representations we must be able to retain earlier representations and/ or their objects, bring them forward, and synthesize them with current ones” (1994, p. 186). Brook also recognizes the importance of memory in the synthesis of recognition, for “in order to synthesize earlier representations with current ones, we must be able to recall the earlier representations and recognize their relation to current ones” (1994, p. 186). In Brook’s view, the act of ‘recalling’ is only possible with the aid of memory, albeit the memory of the earlier representation should belong to the same consciousness (like any current representation). Otherwise, we would not have access to each particular representation. In a similar sense, Angelica Nuzzo argues for the seminal role of memory in the third synthesis:

To this extent, memory is crucial to the application of concepts, that is, to the constitution of the cognitive synthesis. (…) The consequence is that in the first CPR, Kant transcendentally deconstructs the function of memory. Now memory (i) somehow acquires a transcendental function in connection with the imagination (already in constituting the cognitive synthesis). (^{Nuzzo, 2015}, p. 190)

I am in agreement with Nuzzo, who correctly focuses on the transcendental function of memory and on the relation between memory and imagination. Unfortunately, the latter attempts overlook Kantian anthropological description of memory, so that they cannot show how memory is involved in the possibility of experience. Again, I argue for the relevance of memory in the threefold synthesis.

In his analysis of the threefold synthesis, Kant emphasizes that all representations belong to inner sense and, therefore, these should be brought into relations of time, because time is the form of inner sense. The empirical intuition is nothing but the effect of the synthesis of apprehension, so that the manifold given in sensibility is apprehended as an ‘intuition’ only, if we distinguish the time of that manifold, by going through it and taking it together (1998, CPR A99). It is well-grounded to assert that the act of taking the manifold together would not be possible, if each element of the manifold (through which we go through) were forgotten while we are successively apprehending them. Precisely, the very term apprehension entails the mind’s power to retain these empirical data, which would not be possible without a sort of sensitive memory¹¹. Second, Kant states that all successive apprehension of the manifold given in sensibility requires that each one of the apprehended representations should be reproduced, so that the preceding representation can be associated with the following one by means of a synthesis of reproduction in imagination (1998, CPR A101-2). But, how could that representation be reproduced without presupposing a storing faculty in the human being? The answer is, in a way, sketched by S. Matherne, who notices the importance of the human being’s capacity of preserving empirical data in the synthesis of reproduction:

In order, for example, to produce a representation that reflects the different aspects of the champagne flute, if by the time I am representing its curviness I have forgotten all about its glint, then I cannot produce an image of it. On Kant’s view, the synthesis of reproduction is the process through which representations in the past are brought to bear on what we are representing right now. (^{Matherne, 2015}, p. 758)

Despite of the fact that Kant is not much concerned with the retention of those representations but merely with their reproduction, it is plain that without a capacity to store and evoke those representations, there would be no stuff to be reproduced. Tom Rockmore barely notices the relation between memory and the synthesis of reproduction, by claiming that “Kant appears here to conflate the unconscious activity through which the synthesis of reproduction occurs with the problem of conscious memory” (^{Rockmore, 2012}, p. 317). Thomas Powell offers further suggestive remarks about the function of memory in the synthesis of reproduction, saying that “experience itself would be impossible if individual representations were not such that they could be reproduced in memory, synthesized in thought under object-concepts, or synthesized in some other way or ways” (^{Powell, 1990}, p. 28). Powell correctly notices that the successive addition of parts in the apprehension depends upon the successive resonance of the reproduced part. In other words, the constitution of a series is impossible, if each added part were forgotten.

But, what is the status of a preceding representation in the successive series of apprehension? I answer that that representation cannot be given as an intuition in sensibility but rather as a particular remembrance. For the persistence of a past representation in sensibility over time would make the reproduction of that representation useless. On the contrary, it is precisely the fact that the representation (intuition) no longer exists in sensibility, which makes the reproduction in imagination necessary.

The apprehended representation in the synthesis of apprehension is reproduced as a remembrance in the synthesis of reproduction in imagination and it persists until is determined by the synthesis of recognition. As a result, considering that memory partakes in the conditions that make experience possible, it is well-grounded to say that intuition and the remembrance of an intuition should have a similar status. This status could be named “intuitive remembrances” as long as it retains an almost immediate reference to these immediate impressions. Of course, the intuitive remembrance is incompatible with a strong presence-dependence account of intuition that demands “an actually existing object which is present to one in the intuition of it” (^{McLear, 2017}, p. 89). However, these could be compatible with a moderate presence-dependence one, according to which “an intuition only depends on the actual presence of an object for the initial intuition, but not necessarily for subsequent ones” (^{McLear, 2017}, p. 90). These intuitive remembrances are precisely produced by empirical objects which are not present now.

Furthermore, A. C. ^{Ewing (1967)} suggests that memory is capable of supplementing present representation in the synthesis of recognition. Since memory is grounded on imagination and “ imagination is the faculty for representing an object even without its presence in intuition”¹² (1998, CPR B151). It follows that memory contents would become a quasi-intuition. This has led Andrew Stephenson to claim that

Imaginational episodes-as occur, for example, in memory, dreams, and hallucination-involve the intuition of objects that are not, in the relevant sense, present (…) Intuition therefore does not require the existence or presence of its objects and is in no substantial sense object-dependent. (^{Stephenson, 2017}, p. 105)

Stephenson characterizes memories as imaginational episodes, which are related to cognitive states like belief and knowledge. He nonetheless, is not committed with a study of memory as a necessary condition under which experience is possible. On my account, memory stores and reproduces the stuff that is involved in the action of the understanding via the synthesis of recognition. The power of memory to preserve the intuitive remembrances rests on imagination, which is an internal condition under which the synthesis of apprehension, reproduction, and recognition is possible. In other words, the role of memory in experience can be reconstructed from the relation of two ideas. First, sensibility has two parts: senses that generate intuitions of existing objects and imagination that generates intuitions even in the absence of an existing object (2007a, AA 7:153, 167; 1998, CPR B151; 1900, AA 18:619). Second, the functioning of memory is grounded on imagination, so that memory and imagination are tightly connected in the retention and retrieval of the stuff involved in the threefold synthesis (2007a, AA 7:182; 1900, 25:1289, 1464).

Furthermore, Kant holds in the CPR that “from the fact that the existence of outer objects is required for the possibility of a determinate consciousness of our self it does not follow that every intuitive representation of outer things includes at the same time their existence” (1998, CPR B278). Kant is asserting here that even if our cognition of the existence of outer objects involves empirical intuition, we possess intuitive representations of them, without assuming necessarily the current existence of an object in sensibility. These intuitive representations are based on the capacity of imagination to achieve a sensory reproduction (^{Hanna, 2005}). For Kant suggests that the power of imagination is the substitute of the senses, so that it is nothing but the faculty of intuitions in the absence of objects. In this vein, imagination is tantamount to memory as it has the power to reproduce the past absent objects (1900, AA 28:673-4). That is to say, that possession “is possible merely through the reproduction of previous outer perceptions, which, as has been shown, are possible only through the actuality of outer objects” (1998, CPR B278). Therefore, memory can provide intuitive representations whose existence is involved in the threefold synthesis. Since memory is involved in the synthesis of reproduction, which constitutes a transcendental action of the mind, then one might say that memory is a hidden transcendental working faculty. However, Kant is not very interested in the act of storing representations but in their reproduction, which takes place in imagination. The synthesis of reproduction in imagination is an a priori ground of a necessary synthetic unity of appearances, so that it will belong to the transcendental actions of the mind (1998, CPR A102).

However, I disagree with A. Brook’s view that the synthesis of reproduction does not require memory, as long as that synhesis is just a matter of retaining earlier intuitions in such a way that it allows a transition between the later and the earlier representation. I think that he misses the fundamental question of why an “earlier intuition” should be still regarded as an “intuition” at all. He suggests that “such transitions are simply the result of acquiring an association (which, moreover, could be entirely nonconscious). This is not memory and does not even require memory” (^{Brook, 1994}, p. 127). Unlike Brook, I believe that the association is not an association of the same elements, namely the ‘intuition A’ with the successive ‘intuition B’, but rather the ‘remembrance of the intuition A’ with the ‘intuition B’. Since the reproduction (remembrance) of A is not achieved by sensibility but by memory, then, the association must depend upon memory’s power to preserve and reproduce the remembrance of the intuition A through imagination (^{Prichard, 1909}). Therefore, I support J. Bennet’s idea that imagination can be identified with memory in the synthesis of recognition: “imagination, then, is closely connected-if not identical with intellectually disciplined memory. Kant there is expounding his view that the rational grasp of one’s present experience requires the relating of it with remembered past experience” (^{Bennet, 1966}, p. 136).

Third, the synthesis of recognition in the concept declares that all reproduction in the series of representations (synthesis of reproduction) would be futile without the ‘consciousness’ (Bewußtsein) of the fact that the reproduced representation in imagination is not different from the manifold that has been apprehended as intuition. It seems reasonable to ask whether the reproduced representation could coincide with the apprehended one without presupposing memory as a faculty that preserves the existence of the former, during the act by which the mind recognizes that these representations belong to the same object. I do not think that this is possible. On the contrary, memory is required for the synthesis of recognition and Kant is aware of that condition as he warrants that we should not forget the added units in the ‘composing’ of intuitions:

If, in counting, I forget that the units that now hover before my senses were successively added to each other by me, then I would not cognize the generation of the multitude through this successive addition of one to the other, and consequently I would not cognize the number; for this concept consists solely in the consciousness of this unity of the synthesis. (1998, CPR A103; my emphasis)

The synthesis of a manifold by a concept is tantamount to the consciousness of its unity and without such consciousness, provided by the concept, neither the apprehension, nor the reproduction could provide representations of objects for us (^{Gibbons, 1994}). To my knowledge, the consciousness of the fact that the (past) reproduced and the (present) apprehended representations are ‘identical’ relies on memory, in as much as it preserves the existence of the reproduced representations sufficiently to make the consciousness of this identity possible. For Kant declares that “without consciousness that that which we think is the very same as what we thought a moment before, all reproduction in the series of representations would be in vain. For it would be a new representation in our current state” (1998, CPR A103). Thus, the manifold of the representation only becomes a ‘whole’ (Ganzes) or a ‘unity’ (Einheit) when we reach the consciousness of this identity. In contrast, if we do not presuppose the existence of memory, all the representations, which are reproduced, would be forgotten and hence experience would be merely a set of new unrelated representations (1900, AA 25:1462). In strict sense, one must conclude that without memory there would be no intuitions in plural, as long as plurality would entail that we recognize that the past representation is different from the present one. This recognition, hence, would demand the preservation of representations whose comparison leads us to recognize the difference or identity among representations. On this basic point, I agree with Nuzzo as she claims that

The third synthesis implies first the memory of past representations, and second the recognition that earlier and later representations are related as representations of the same object; and this requires the concept as a rule, which allows memory to repeat its orderly recalling and comparing of past representations. (^{Nuzzo, 2015}, p. 190)

The synthesis of recognition involves the consciousness of both unity and identity of the manifold of our empirical representations; thus, this consciousness must always be found, even if it lacks conspicuous clarity (1998, CPR A104). Kant goes on to claim that the transcendental condition of the unity of the consciousness in the synthesis of the manifold of all our intuitions is the transcendental unity of apperception (1998, CPR A106-7). Furthermore, this unity of consciousness is grounded on the consciousness of the identity of the function by means of which the empirical manifold is synthetically connected into one cognition. This leads Kant to claim that “the original and necessary consciousness of the identity of oneself is at the same time a consciousness of an equally necessary unity of the synthesis of all appearances in accordance with concepts” (1998, CPR A108). It follows that the self-consciousness is meant to be a condition of our cognition of objects (Schulting, 2015).

Again, memory performs a transcendental function, insofar as it is necessarily involved in the transcendental syntheses of apprehension, reproduction and recognition. Thus, our knowledge of appearances in general would not be possible without memory, for it demands a synthetic unity of the apprehended manifold in intuition. However, this apprehension does not suffice to guarantee the existence of this knowledge, but it requires both the a priori reproduction of the manifold and the presence of concepts in which that manifold is unified (1998, CPR A105). Yet, despite the skeptical view of A. Brook on the role of memory in the first and second synthesis, he is optimistic about the relevance of memory in the third one, by recognizing memory and consciousness as two components of the synthesis of recognition:

Synthesis by recognition requires two things. One is memory -true memory, that is to say, recovery and recognition of past representations as past, not just associative reproduction. The other, of course, is ‘consciousness’, that is to say, recognition -something in the past representations must be recognized as related to present ones. (^{Brook, 1994}, p. 129)

An object of cognition can only be constituted as such, if our mind is capable of unifying its disperse components, whose existence is preserved by our memory, through the consciousness of the identity of those components. For “[a merely reproduced] manifold would never constitute a whole, since it would lack the unity that only consciousness can obtain for it” (1998, CPR A103). Cognition, according to Kant, demands a synthesis of representations related in time and this would not be possible, if our mind were not capable of remembering and, therefore, recognizing the connection between earlier and later representations. Brook emphasizes that when our mind recognizes that the earlier and later representations represent the same object, we use then “a concept of number, a concept of quality, a concept of modality, and, of course, the specific empirical concept for the object we are cognizing” (^{Brook, 1994}, p. 129). Finally, if we take into account that, according to Kant, memory leans on imagination and memory plays a transcendental role, then, the power of memory to reproduce representations will be a transcendental act of the mind. For the acts of storing and reproduction of representations are demanded by the threefold synthesis, which is a necessary condition of experience.

I Conclusion

Kant’s theory of memory is not systematic like other aspects of his philosophy, but it is composed by elements that belong to the anthropological sphere, or to the transcendental one. Memory turns out to be a dynamic cognitive faculty that stores, organizes and reproduces representations derived from experience. Moreover, memory depends on the productive imagination, which conditions experience, as long as it is at the basis of the formation of empirical intuition in general. Thus, it is quite reasonable to claim that experience cannot be possible without memory and his allusion to memory in the threefold synthesis is a sing of its relevance in experience.