Navigation – Plan du site

AccueilNuméros6VariaIs Bergson’s Theory of Memory Sim...

Varia

Is Bergson’s Theory of Memory Simulationist?

Yasushi Hirai

Résumés

Cet article présente une analyse comparative entre la Théorie de la Mémoire de Simulation (STM) de Kourken Michaelian et la Théorie de la Mémoire de Henri Bergson (BTM), mettant en lumière à la fois les convergences et les distinctions dans leurs approches de la préservation et de la construction de la mémoire. Létude examine quatre caractéristiques clés par lesquelles la STM et la BTM se croisent : le rôle constructif du souvenir, la continuité entre l’imagination et la mémoire, leur différenciation par le suivi des processus, et leur indépendance vis-à-vis de la préservation causale. Cette comparaison transfrontalière, qui diverge selon les contextes historiques, linguistiques et culturels, révèle non seulement des convergences surprenantes mais introduit également une nouvelle interprétation de la BTM. Il devient évident que l’actualisation de la mémoire pure dans la BTM n’est pas seulement une transformation ou une causalité mais une imitation basée sur la suggestion. De plus, elle clarifie la formation d’images génériques comme terrain commun entre le souvenir, l’imagination et la reconnaissance, interprétés comme des mouvements entre les plans du cône de mémoire. Inversement, cette vision restructurée de la BTM jette également un nouvel éclairage sur la STM, suscitant de nouvelles questions sur sa relation avec d’autres types de mémoire et la nécessité d’une préservation générale.

Haut de page

Texte intégral

This paper is based on the content presented at the workshop “Remembering: Analytic and Bergsonian Perspectives 2,” held at Fukuoka University in October 2022, and the workshop “Developments in the Philosophy of Memory,” held at Waseda University on June 16, 2024. I am deeply grateful to the members of Project Bergson in Japan (PBJ), the Société des amis de Bergson (SAB), and the Philosophy of Memory Network Japan (PMNJ), as well as to other participants, for their valuable feedback. Special thanks are due to the members of the Centre for Philosophy of Memory (CPM), who have courageously tackled the challenging and uncertain task of engaging with Bergson’s ideas. I am particularly indebted to Kourken Michaelian, the director of CPM, for his dedicated efforts in fostering true dialogue, without which this paper would not have been possible. This research was supported by JSPS KAKENHI Grant Numbers 19H01190 and 24K00005.

Introduction

1Recent advances in the philosophy of memory—most notably Kourken Michaelian’s simulationism—have generated fresh questions about how we remember. Although Michaelian’s view and Henri Bergson’s memory theory differ in their historical and conceptual origins (with simulationism being debated in English within the analytic tradition and Bergson’s work emerging in French within a continental context), a direct comparison can yield valuable insights for both camps. This paper aims to explore these provocative intersections.

2To guide the reader, §1 introduces Michaelian’s Simulation Theory of Memory (STM); §§2–3 develop Bergson’s Theory of Memory (BTM)—first its constructive role in recollection, then its treatment of imagination versus memory; §4 examines memory actualization in BTM; §5 investigates the origin of generic images; §6 compares STM and BTM regarding whether preservation is necessary for remembering; and §7 concludes with a summary of convergences, divergences, and open questions for future research.

The Theory of Remembering in Simulationism

3In his pivotal work, Mental Time Travel (2016), Michaelian proposes a novel memory theory termed “simulationism,” based on empirical findings from contemporary psychology and neuroscience. This theory challenges conventional distinctions between remembering and other cognitive processes, such as imagining the future or counterfactual conditions, advocating a more dynamic and integrated approach. This section introduces the following four central features of the Simulation Theory of Memory (STM).

The Constructive Role of Remembering

4Contrary to traditional causal theories that confine remembering to the retrieval of preserved content, STM emphasizes a generative process. In classical theories, remembering involves only the preserved contents, and if it contains something more than what is preserved, it is not deemed a memory. This is called preservationism. In contrast, simulationism clearly acknowledges a generative aspect. The author emphasizes that “Fundamentally, on this view, remembering is generative not preservative: it is not a matter of preserving a representation but rather of constructing a new representation of a past episode based on stored information originating from a variety of different sources as well as information available in the subject’s current environment” (Michaelian 2016, 60). This underscores memory’s active, constructive nature rather than a mere act of preservation.

Continuity between Imagination and Remembering

5STM also posits that the mechanisms underlying both imagination and memory are essentially the same. According to the author, both are different manifestations of the same “general episodic construction system” (Michaelian 2016, 13; 92; hereafter we will abbreviate it as GECS). This system enables the construction of detailed mental representations of experience, applicable to both past events (remembering) and future or hypothetical scenarios (imagining), constituting the shared foundation for these cognitive functions. This view, which does not recognize a fundamental distinction between imagination and remembering, is called continuism (Perrin 2016).

Distinguishing Imagination from Remembering through Process Monitoring

6While the underlying mechanisms of imagination and memory are closely related, distinguishing between them is crucial for cognitive reliability. “Episodic remembering would be unreliable were we unable to distinguish it from other forms of episodic imagination. But there is extensive similarity—at the neural, cognitive, and phenomenological levels—between memory and other forms of imagination, raising the question of how we manage to avoid confusing different forms of imagination” (Michaelian 2016, 125). Michaelian argues that this differentiation is achieved through “process monitoring,” a metacognitive function that assesses the cognitive process rather than its content. Certainly, when recalling an episode, whether its content stems from memory or imagination must be differentiated, a task known as “source monitoring.” However, the author suggests that before making any judgments about the source of content, it is crucial to determine the process by which the episode was obtained. The process monitoring takes this role. Specifically, “[i]f an episodic constructive process is classified as self-oriented, past-oriented, and actual rather than counterfactual, it is judged to be an instance of remembering” (Michaelian 2016, 213).

Independence from a Causal Preservation Model

7The final feature of STM to be addressed in this paper is that it does not rely on a causal model of preservation. Simulationism is not just a theory that emphasizes the constructive aspect but also states that preservation is not necessary for remembering. This distinction separates it from the Causal Constructive Theory of Remembering (CCTR), which acknowledges a constructive aspect yet considers preservation essential, unlike STM which posits that remembering can occur without preservation. It is important to note that simulationism makes a modal claim that preservation is not necessary, without denying that preservation often occurs in many everyday cases. The author states:

The simulation theory thus supports radical generationism even more emphatically than does the causal theory of constructive memory. [...] In many cases, remembering no doubt involves the sort of continuous causal connection described by the classical causal theory and at least the sort of approximate content similarity described by the causal theory of constructive memory. But it need not (Michaelian 2016, 103-104).

8This theory supports a more radical form of generationism, suggesting that remembering may involve causal connections and content similarity yet it does not necessarily depend on them.

9In the following sections, these features will be compared with Bergson’s theories, with the aim of elucidating both their divergences and surprising parallels.

The Constructive Role in Remembering (Feature 1)

10Henri Bergson’s theory of memory (BTM), as expounded in Matter and Memory, is often characterized by its singular concept of “pure memory,” which is said to preserve every experienced event in full detail. This might give the impression of a strongly preservationist view, sometimes interpreted as Platonic metaphysics (Bernet 2005). Nevertheless, recent interpreters have nuanced this view, indicating that Bergson’s idea of preservation is distinct from conventional trace theories, which he criticizes, and should not be hastily linked to causal theories (McNamara 1996; Lawlor 2003, 2006; Guerlac 2006; Ansell-Pearson 2010). Moreover, a strong preservation stance does not preclude the possibility of constructive or generative processes during retrieval (Burton 2008; Al-Saji 2004; Sinclair 2019). Indeed, Bergson acknowledges significant active intervention in memory processing, which engenders various pathological possibilities as well as with intellectual creativity (Ansell-Pearson 2010). For instance, Burton positions this within modern memory research, moving away from archival interpretations. Therefore, this paper will focus on the aspect of remembering, rather than preservation, in BTM.

11BTM does not share much of the epistemological context associated with the justification of knowledge. Instead, it is positioned within a unique scientific metaphysical context that connects empirical sciences like psychology, physiology, and biology with the metaphysics of mind and time, addressing the nature of qualia, representations, time flow, and reality of the past (Hirai 2023). Indeed, the second chapter of Matter and Memory, where the theory of memory is initially developed, starts with empirical studies of numerous recognition disorders. Not limited to pathologies in the proper sense, BTM covers a wide range of anomalies, including hallucinations, dreams, déjà vu, and the “panoramic life review”. In his view, remembering leads not only to pathology but to intellectual creativity. In his later work Intellectual Effort, it is argued that the constructive aspect inherent to remembering forms the basis for interpretation and invention (Bergson 1920, 186-230). Attention is also a significant theme (Bergson 1911, 109-117; Bergson 2018, 167-185). This co-occurrence of pathology and creativity is not a paradox but rather results from an intrinsic connection. As many scholars emphasize, BTM is characterized by its view of memory not merely as a reproduction or repetition of the past, but fundamentally as something divergent and creative (Hyppolite 1949; Ansell-Pearson 2010).

  • 1 Intellectual creativity can be interpreted as an exaptation of this utility-biased remembering.

12The idea that remembering is not merely a faithful reproduction of past facts reflects his broader view on cognition, shared between perception and memory, which posits that cognitive processes in organisms are fundamentally constrained by their current interests and needs. If a living being remembers something at all, the memory is not referenced for its own sake, but to solve problems currently faced. Unless the identical situation recurs, it is advantageous to edit and adapt the memory to suit the new contexts in which it is employed. BTM thus adopts an ecological-pragmatic principle whereby memories are selected and shaped according to present demands. “Ever bent upon action, it can only materialize those of our former perceptions which can ally themselves with the present perception to take a share in the final decision” (Bergson 1911, 188). Let us refer to this as “utility-biased remembering.”1

13This concept is illustrated using the renowned model of inverted cone of memory. It explicates that the same set of memories enters different “planes” or organizations and further undergoes “translation and rotation”—two specific operations or edits—to be transformed into representations for remembering (Bergson 1911, 220). In this framework, translation denotes the shifting of a memory across the cone’s various “planes,” thereby modulating its degree of abstraction or specificity to suit the demands of the present context. Rotation, by contrast, refers to re-orienting the memory’s internal content—shifting emphasis among its details so that those facets most pertinent to current action or recollection are foregrounded. These modifications are explicitly made in response to the “appel” [demand/appeal/call] from the current situation. Sinclair (2019, 103) notes, “pure memory can never be purely remembered. Memory retrieval is always productive or creative precisely insofar as it can actualize virtual memories only with the means available in the present, a present oriented to action now and in the future” (Sinclair 2019, 103). Consequently, in BTM, memory is susceptible to various reconstructions and reorganizations upon retrieval. A significant aspect of this theory is that the same stored memory material is not only utilized for reproducing episodic memories but also contributes to the broader capacity for constructing images, including imagination and recognition. This point will be revisited later.

The Relationship Between Remembering and Imagining (Features 2 and 3)

14As already noted, in BTM, as in STM, imagination and remembering substantially overlap (Feature 2). Both faculties utilize the same material (images) and share the capacity for constructive editing, similarly to the GECS in STM. Consequently, remembering and imagination are not independent capabilities but are intrinsically interconnected. Imagination draws upon the material from memory, and remembering employs the same mechanisms in its construction. Thus, imagination and remembering are in a continuous relationship.

15How, then, are the two differentiated (Feature 3)? In STM, this distinction is managed by a metacognitive function known as process monitoring. Bergson articulates this in Chapter Three of Matter and Memory: “To imagine is not to remember [imaginer n’est pas se souvenir],” clarifying that “the image, pure and simple, will not be referred to the past unless it was in the past that I sought it, thus following the continuous progress [progrès continu] that brought it from darkness into light” (Bergson 1911, 174).

16This indicates that imagination and remembering, when viewed solely through the images they produce, are indistinguishable. They are distinguished by “continuous progress”—the process engaged by the remembering subject. Thus, it is the process, rather than the content, that serves as the criterion for distinguishing between imagination and remembering.

What Does Actualizing Memory Entail?

  • 2 The discussion in this section is partly based on Chapter 6 of Hirai (2022).

17Let us delve deeper into the specific inner workings of the remembering process, referred to in Bergson studies as the “actualization” of memory. Why is it termed so peculiarly? It is because pure memory is preserved in a “virtual” state. At first glance, the term “actualization” may not seem constructive. This section aims to address such concerns by exploring three possible interpretations of actualization.2

18First, what does it mean for stored memory to be “virtual”? Interpretations vary, but a common understanding among scholars suggests three characteristics: non-sensory, non-representational, and non-extended (Hyppolite 1949; Al-Saji 2004; Sinclair 2019). Pure memory of individual events does not manifest as a concrete, sensible, or extended representation that the subject perceives during successful recall. Instead, actualization is a process that imbues memory with these properties. We will examine three views, ultimately advocating for the last as the most valid: 1) Metamorphosis View, 2) Causal Production View, and 3) Imitation View.

Metamorphosis View

19This view posits that memory actualization is the process by which pure memory transforms into a remembering image. Here, pure memory and the remembered image are considered numerically identical but differ merely in degree of intensity or clarity. We will refer to this as the Metamorphosis View.

20The main strength of this view lies in its alignment with naive conscious reports and its textual support. For example, in the subjective experience of remembering, the image feels as though it “comes into view like a condensing cloud” (Bergson 1911, 170), and Bergson describes this as “the progress of a memory precisely consists [...] in its becoming materialized” (Bergson 1911, 174).

21However, as any reader of Bergson is aware, his terminology is highly dependent on context and register. He often employs expressions that may not technically be acceptable to provide some overall perspective, suggesting that frequent usage does not necessarily confirm theoretical accuracy or validity. This instance is a clear example of such a situation.

22In fact, in Chapter Three of Matter and Memory, Bergson promptly revises his initial rough description in the following passage. He asks: “The question is whether the memory of pain was really pain from the beginning” (Bergson 1911, 174). Bergson’s answer is definitive: no. He argues that pure memory is never merely a “weak sensation.” He emphasizes the “difference in nature” (Bergson 1911, 179; 1920, 160) between pure memory and the sensory-laden memory image. Consequently, the Metamorphosis View cannot be adopted literally at a theoretical level.

Causal Production View

23The second interpretation posits that pure memory causally produces the remembering image. This view does not assert the numerical identity of the two, thereby avoiding direct conflict with the previously mentioned claim of a fundamental difference. Consider the following quote:

The recollection [souvenir] of a sensation is capable of suggesting the sensation, I mean of causing it to be born again [faire renaître], feeble at first, then stronger and stronger in proportion as the attention is more fixed upon it (Bergson 1920, 161).

24This quote might inadvertently be interpreted to mean that pure memory (referred to as “recollection” in the quote) causes the remembering image to emerge. However, interpreting this as literal causal production presents challenges. Causal production implies either “creation from the material itself” or “processing/editing of the given material.” If the former is true, it conflicts directly with the aforementioned “difference in nature.” If the latter, then attributing the initiative for image construction solely to pure memory complicates the reconciliation with the concept of utility-biased remembering. Thus, caution is advised in interpreting this expression literally.

Imitation View

25Finally, we advocate for what we consider the most fitting interpretation: the Imitation View. To illustrate that there is neither numerical identity nor causal production between pure memory and the remembering image, Bergson uses an unusual metaphor: the hypnotist’s suggestion. Immediately following the quote from Chapter Three of Matter and Memory mentioned in the Metamorphosis View, Bergson continues:

Because the hypnotized subject ends by feeling hot when he is repeatedly told that he is hot, it does not follow that the words of the suggestion were themselves hot. Neither must we conclude that because the memory of a sensation prolongs itself into that very sensation, the memory was a nascent sensation: perhaps indeed this memory plays with regard to the sensation which follows it precisely the part of the hypnotizer who makes the suggestion (Bergson 1911, 174).

26In this analogy, the hypnotist’s words represent pure memory, and the sensation felt by the subject represents the remembering image. The experience of heat upon hearing the words does not mean the words themselves were “hot”—in phenomenological terms, they possess an “intentional” rather than a “reell” property. Neither did the words become heat, nor did they causally generate heat. Given the intentional nature of the relationship, it is entirely possible that the same words might not produce heat in other instances. It is precisely for this reason that his position “admits between memory (le souvenir) and the nascent perception that follows it the same relations as between the hypnotist (or suggestion) and the state suggested” (Bergson 2018, 48).

  • 3 Bergson’s insistence on “suggestion” is apparent in the remarkable variety of synonymous expression (...)

27Bergson elaborates this “suggestive” relationship as a model for memory actualization in recalling events. This is deliberate, as he employs the same metaphor in other texts. In Mind Energy, he states, “the recollection [souvenir] is distinct from the sensation it suggests” (Bergson 1920, 161), and during his Collège de France lectures, he refers to this unique relationship as a “suggestive cause” (Bergson 2018, 54).3

  • 4 In the text cited under the Causal Production View (ME 161), Bergson himself does not use the verbs (...)
  • 5 Although pure memory does not causally generate the recollective image, it is no blank slate for ar (...)

28The distinction between causation and suggestion lies in the agency of production. In a causal model, pure memory is entirely responsible for image creation. However, in a suggestive model, pure memory merely indicates the target content, with the actual creation resting primarily with the rememberer.4 According to the Imitation View, pure memory itself remains “immobile,” merely indicating the target to be aimed at. The remembering image, on the other hand, materializes it. As we will see, BTM includes a mechanism analogous to the GECS in STM, from which available images are used by the rememberer to create numerically distinct imitative images. This allows for the gaps between the remembered images and pure memory.5

  • 6 Please refer to note 61 on page 271 of the revised French edition of L’énergie spirituelle (Bergson (...)

29This Imitation View offers various advantages: it not only aligns with Bergson’s theoretical stress on suggestion rather than causation but also fits well within his broader strategy of utility-biased remembering. Additionally, it is consistent with Bergson’s explanation of Ribot’s Law regarding forgetting. According to Ribot, verbs are less likely to be forgotten than nouns and adjectives. For Bergson, forgetting is failure of remembering. Contrary to Ribot’s own justification,6 verbs are less forgettable due to their greater ease of imitation: “It is simply because verbs express actions, and actions may be mimicked.” (Bergson 1920, 67[54]; Bergson 1911, 152) Another notable advantage is that imitation is indispensable even in acts of recognition. The “motor schema” (schème moteur) that extracts an object’s salient features is, in Bergson’s terms, a “movement of imitation” (Bergson 1911, 124; 130; 137). Bergson explicitly links the two domains, implying that both memory and perception operate within a shared framework of imitative reconstruction: “The brain’s role is to mimic the life of the mind, and to mimic as well the external situations to which the mind must adapt” (Bergson 1920, 47; cf. 75).

30Let us address two potential objections to this interpretation. First, there may be a mismatch with phenomenological experience; individuals might report that a memory emerges spontaneously like “a condensing cloud.” However, BTM (and STM) concern not the level of conscious report but the underlying mechanisms—like how emotional contagion stems from neural imitation (Gallese et al. 1996; Gallese 2001), suggesting a mimetic mechanism underlies even phenomenologically “passive” experiences.

31The second objection questions the source of the material for the remembering image. Bergson addresses this through the concept of “interference,” which leads to the creation of generic images. We will explore this in further detail in the next section.

The Origin of Generic Images

32In BTM, phenomena such as remembering, imagining, recognition, dreams, and hallucinations are considered different manifestations of the same underlying memory material. The Imitation View suggests that remembering involves an imitative reconstruction using this shared capability. In BTM, while the target event in an episodic remembering is given as a pure memory, the generic images used to reconstruct this target event require further explanation since they are distinct from pure memory. For instance, when envisioning a dinner, if it is recalled as a memory, there is typically a past target event. If it is imagined as a future scenario, a different setting is required. However, in both cases, generic images such as hamburgers, wine, tables, or cutlery are needed to construct the scene. The question arises: where do these images originate?

33The answer is more explicitly stated in Histoire des théories de la mémoire than in Matter and Memory. In a section of Histoire des théories de la mémoire, Bergson differentiates between the memory of one’s own personal chair and that of the chair in general. He states that the latter type of “banalized” images “are obtained precisely by a mixture, like an interference among all these personal memories related to that object.” (Bergson 2018, 115-116) This suggests that a general image of an object (semantic or object memory) is formed through the “interference” of numerous personal episodic memories:

Different memories mingle; a multitude of similar but personal memories overlapping each other and interfering will depersonalize to become impersonal memories (Bergson 2018, 129).

34Interfering memories are assumed to reinforce similar features and neutralize dissimilar ones. Thus, a wide variety of generic images are drawn from the different interferences of various memories. To use his own example of the chair, the generic image of a “chair” arises not from a single abstract definition but from the mingling and interference of countless personal memories of individual chairs one has encountered: one’s grandmother’s armchair, a school desk chair, a sleek office chair, and so on. As these distinct recollections overlap, their shared functional and structural features—such as a flat surface for sitting, support by legs, and often a backrest—become accentuated through interference. At the same time, highly specific, personal details—an upholstery’s unique hue, a particular scratch, the precise context of each encounter—are neutralized or averaged out, allowing the image to become “banalized” or depersonalized into the common representation of “chair.” Repeated interferences across memory sets yield a versatile pool of depersonalized images. These generic images then serve as ready‐made material for constructing memories, imagining new scenarios, and interpreting present perceptions.

35The transition between different “planes” of the memory cone, as described in Matter and Memory, can be interpreted to represent the formation of these generic images. The most personal pure memories are situated at the base, labeled AB. What then are the memories at the lower planes? As previously noted, each “plane” represents a certain arrangement of memories that are not fixed but are possible in diverse forms. “Within the cone so determined the general idea oscillates continually between the summit S and the base AB.” (Bergson 1911, 210) Suppose that a specific perspective of similarity defines one plane — for Bergson, who does not recognize absolute similarity, similarity is always defined “from the particular viewpoint of the action that is being carried out” (Bergson 2018, 129). Within this defined plane, the same set of memories interfere with each other in this specific way to produce a group of generic images, whereas a different plane will lead to another pattern of interference, resulting in a distinct set of generic images. These generic images serve as building blocks in constructing episodes during the remembering process.

36If the above interpretation is correct, the remembering process according to the Imitation View would be as follows: It is a process where the content of the target event, suggested by non-interfered personal (pure) memories on the ultimate plane AB, is reconstructed using groups of generic images drawn from the lower planes where memories have interfered in various ways. These generic images provide a common material for both remembering and imagining, showing that BTM possesses an equivalent to the GECS found in STM. BTM also provides an explanation of how these generic images are formed via memory interference, corresponding to the body of the memory cone.

37BTM’s application scope extends further since it is also considered applicable to the recognition of present objects. According to Bergson’s theory of “attentive recognition,” perception “only becomes complete and acquires a distinct form through that very memory, which slips into it and supplies most of its content” (Bergson 1920, 207): in short, “there is no perception which is not full of memories.” (Bergson 1911, 24) Thus, in BTM, not only past, future, and counterfactual scenarios but also present episodes possess a constructive aspect and share the same image resources.

38As noted in the preceding section, Bergson likewise invokes imitative movement (motor scheme) as the very mechanism that enables attentive recognition. Just as pure memory supplies the suggestion in remembering, so in recognition it is the external object that performs the suggestive role. The object itself is neither transformed into an act of recognition nor does it causally produce that recognition; instead, it merely prompts the mind to reconstruct the object through its own imitative activity.

39Thus, Bergson’s “continuism” posits a shared foundational capability across remembering, imagining, and recognition. This extended continuism also suggests an intrinsic link between episodic construction and other memory forms, such as semantic or object memory (Openshaw 2022).

Is Preservation Necessary? (Feature 4)

40Finally, we address the last point. As previously noted, STM acknowledges that preservation is recognized in many cases of remembering. However, it contends that preservation is not always necessary. Thus, STM distinguishes itself not only from preservationism, which does not acknowledge construction, but also from CCTR, which, while recognizing construction, still deems preservation essential.

41What would be the scenario where remembering occurs without preservation? Michaelian cites cases where:

I might see the statue in front of the train station and for one reason or another end up not encoding even a short-term representation of the scene. Or I might encode a short-term representation but end up not consolidating a long-term representation. In neither case need the absence of stored information deriving from my experience of the episode prevent me from remembering it. If sufficient information originating in other sources—for example, stored representations of similar episodes or semantic knowledge of the layout of the relevant area—is available, my episodic construction system may produce a representation of the episode when I later attempt to remember it. When this occurs, the simulation theory implies I remember the episode just as much as I do when part or even all of the content of the retrieved representation originates in my experience of the episode (Michaelian 2016, 103).

42Michaelian believes that remembering can be considered even when the corresponding episode information is missing because memories of similar episodes or semantic memories are available. Thus, if sufficient memory resources are available, remembering can proceed without the direct preservation of the target event.

43Would the same apply to BTM? This is an intriguing question. Considering the theoretical similarities discussed, one might expect a similar stance. However, as already mentioned, since Bergson did not explicitly address this question and the context of epistemic justification is not shared, the text leaves the answer undetermined. Additionally, in BTM, which asserts the preservation of all events for independent reasons, the non-preservation of a specific event would be an ad hoc assumption.

44However, from the discussion thus far, one thing can be pointed out: when it is stated that “preservation is necessary for remembering,” this “necessity” applies in two contexts:

45(1) For an episodic remembering to occur, the individual preservation of the target event (the contention between STM and CCTR) is necessary.
(2) For image-related capabilities, including remembering, to function, general preservation sufficient for the formation of generic images is necessary.

46While BTM remains non-committal on the first point, it clearly supports the second. Conversely, STM expressly denies the necessity for the first context, but does not explicitly address the second. In the quote above, “other sources” (similar episodes or semantic memories) are employed when the target event itself is not preserved. Assuming GECS facilitates common episode construction, these other sources could still be utilized even in cases of episodic remembering where the target event is preserved. This raises three possible questions for STM:

  • 7 If STM denies even the necessity of general preservation, it could lead to a serious skepticism reg (...)

47(1) Does STM consider general preservation necessary for the functionality of GECS?7
(2) Is GECS employed for recognition purposes?
(3) How does STM view the relationship between GECS and semantic or object memory?

Conclusion and Discussion

48This paper conducted a comparative analysis of four features within STM and BTM. The analysis revealed a shared framework for the first three features. Both theories emphasize the constructive role of remembering and highlight that imagination and remembering (and also recognition in Bergson) share a fundamental process of image construction, which is differentiated by process rather than content. This convergence is surprising, given the differences in time, language, and context between the two theories, prompting further reflection on its implications.

49Feature 4 divided the question into two: individual preservation for the target event and general preservation for generic images. STM explicitly addresses the former, while BTM deals with the latter, leaving the other questions open in each theory. Therefore, this comparison illuminates the nuanced relationship between pure memory and the remembered image, facilitating a restructured interpretation of the text and offering a new perspective to reorganize BTM as an extended form of continuism. These more developmental questions—such as the relation between memory and truth, or the origin of pastness—will be pursued in future research.

Haut de page

Bibliographie

Al-Saji, Alia. 2004. The memory of another past: Bergson, Deleuze and a new theory of time. Continental Philosophy Review37(2): 203–239. 

Ansell-Pearson, Keith. 2010. “Bergson on memory.” In Memory: Histories, Theories, Debates, edited by S. Radstone, and B. Schwarz, 61-76. New York: Fordham University Press.

Bergson, Henri. 1911. Matter and Memory. Translated by N. M. Paul and W. S. Palmer. London: George Allen and Unwin.

Bergson, Henri. 1920. Mind-Energy. Translated by H. Wildon Carr. New York: Henry Holt and Co.

Bergson, Henri. 2009. L’Énergie spirituelle. Paris: Presses universitaires de France.

Bergson, Henri. 2018. Histoire des théories de la mémoire : cours au Collège de France, 1903-1904. Paris: Presses universitaires de France.

Bernet, Rudolf. 2005. “A present folded back on the past (Bergson).” Research in Phenomenology 35 (1):55-76.

Burton, James. 2008. “Bergson’s non-archival theory of memory.” Memory Studies, 1(3), 321-339.

Gallese, Vittorio, Luciano Fadiga, Leonardo Fogassi, Giacomo Rizzolatti. 1996. “Action recognition in the premotor cortex.” Brain, 119(2), 593-609.

Gallese, Vittorio. 2001. “The ‘shared manifold’ hypothesis. From mirror neurons to empathy.” Journal of Consciousness Studies, 8(5-6): 33-50.

Guerlac, Suzanne. 2006. Thinking in Time: An Introduction to Henri Bergson. Ithica: Cornell University Press.

Hirai, Yasushi. 2022. Made Out of Time: An Invitation to Bergson’s Philosophy of Time. Tokyo: Seidosha.

Hirai, Yasushi. 2023. “Introduction”. In Bergson’s Scientific Metaphysics: Matter and Memory Today edited by Y. Hirai, 1-12. New York: Bloomsbury.

Hyppolite, Jean. 1949. “Aspects divers de la mémoire chez Bergson.” Revue internationale de philosophie3(10), 373-391.

Lawlor, Leonard. 2003. “The Ontology of Memory: Bergson’s Reversal of Platonism.” Epoché (Provo, Utah)8(1): 69-102. 

Lawlor, Leonard. 2006. “Bergson Revisited.” Symposium (Canadian Society for Continental Philosophy)10(1): 35–52.

McNamara, Patrick. 1999. Mind and Variability: Mental Darwinism, Memory, and Self. New York: Praeger.

Michaelian, Kourken. 2016. Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past. Cambridge Mass: The MIT Press.

Mullarkey, John. 1999. Bergson and Philosophy. Edinburgh: Edinburgh University Press.

Openshaw, James. 2022. “Remembering objects.” Philosophers’ Imprint 22(11): 1-20.

Perrin, Denis. 2016. “Asymmetries in subjective time.” In Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel, edited by Kourken Michaelian, Stanley B. Klein, & Karl K. Szpunar, 39-61, Oxford: Oxford University Press.

Sinclair, Mark. 2019. Bergson. New York: Routledge.

Haut de page

Notes

1 Intellectual creativity can be interpreted as an exaptation of this utility-biased remembering.

2 The discussion in this section is partly based on Chapter 6 of Hirai (2022).

3 Bergson’s insistence on “suggestion” is apparent in the remarkable variety of synonymous expressions he employs—la suggestion de l’image, la cause suggestive de l’image, and la cause qui suggère l’image—to indicate that memory can serve as the suggestion of an image without being the image itself (Bergson 2018, 54). Pure memory thus functions as a souvenir suggérant (“suggesting memory”) or souvenir suggesteur (“memory-as-suggester”) (Bergson 2018, 48).

4 In the text cited under the Causal Production View (ME 161), Bergson himself does not use the verbs corresponding to 'cause' or 'produce'; instead, he uses 'faire renaître' (literally, “make it reborn”). This mediating expression can be interpreted in a more delicate and accurate way from the perspective of the Imitation View. Furthermore, unlike the Metamorphosis View, you can find the transitive use of the verb “materialize” [matérialiser], which supports our interpretation: “The memory-image in its turn partakes of the “pure memory” which it begins to materialize [matérialiser].” (Bergson 1911, 170)

5 Although pure memory does not causally generate the recollective image, it is no blank slate for arbitrary invention. Each pure memory carries a distinctive “suggestive profile”—a contour of constraints that steers the imitative act toward materializing that very past occurrence, rather than just any plausible scenario. On this account, fidelity in remembering lies not in unmediated retrieval or in coherent fabrication, but in the successful imitation of the unique suggestion offered by an “immobile” pure memory, thereby preserving its essential, if indirect, connection to the past it represents.

6 Please refer to note 61 on page 271 of the revised French edition of L’énergie spirituelle (Bergson 2009).

7 If STM denies even the necessity of general preservation, it could lead to a serious skepticism regarding the existence of the past. For Bergson, this would result in significant concerns in the philosophy of time.

Haut de page

Pour citer cet article

Référence électronique

Yasushi Hirai, « Is Bergson’s Theory of Memory Simulationist? »Bergsoniana [En ligne], 6 | 2025, mis en ligne le 01 octobre 2025, consulté le 16 juin 2026. URL : http://journals.openedition.org/bergsoniana/3909 ; DOI : https://doi.org/10.4000/154jm

Haut de page

Auteur

Yasushi Hirai

Yasushi Hirai est professeur à la Faculté des lettres de l'Université Keio, au Japon.

Yasushi Hirai is Professor in the Faculty of Letters at Keio University, Japan.

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search