- 1 There are, broadly, two ways in which memory can constitutively figure in experience: (i) remembere (...)
1When we consider our experience from a broad perspective — taking into account our cognitive, emotional, and evaluative attitudes — it is evident that in many cases memory does not possess a merely causal role in relation to experiences that have temporal contents. At least some of these aspects of experience are constitutively related to memory. That is, they would not be the kind of experiences they in fact are if memory did not figure in them.1 Our experiences of nostalgia, of grief, our reflections on the way time has been passing faster as we have been growing older, our appreciation of the progress we’ve been making — all of these cases arguably include the experienced recollection of past events and are thus not only causally related to past experiences. A constitutive relation to memory also seems to obtain in the case of our more banal reflections on temporal order, for example when I try to remember whether I watered my plants yesterday, or the day before.
- 2 The term “non-inferential experience,” frequently qualified as “immediate experience,” is meant to (...)
- 3 Bergson’s views on the faculty of memory and its role in experience broadly construed have been, by (...)
2My focus here, however, is on a much more tightly circumscribed topic — the relation between memory and our non-inferential experience of change and motion.2 And my central claim will be that Bergson’s views on this relation, which have not been discussed in significant detail in the literature, can shed new light upon recent debate on the topic.3 I approach this in three steps. First, I examine the “memory theory” of non-inferential temporal experience in its two main forms — a “traditional” version and a “modified account” — situating Bergson’s views vis-à-vis this distinction. Second, I explore a contemporary defence of the memory theory (Phillips 2010; 2018) and underline a limitation in this account. Finally, I focus on a feature of Bergson’s reflections on memory and temporal experience that can potentially address this limitation.
3Various thinkers have defended a memory theory of immediate temporal experience, on which memory is taken to be not only causally related to but also partially constitutive of our experience of change. An influential source for the position is Reid’s view. According to Reid, sense perception is limited to the present moment in a “strict” sense:
[I]f we speak strictly and philosophically, no kind of succession can be an object either of the senses, or of consciousness; because the operations of both are confined to the present point of time, and there can be no succession in a point of time (Reid 1785, 325-326).
- 4 Augustine, unlike Reid, does not consider sense perception to be confined to the present, and does (...)
4Reid is working with an Augustinian idea of the present as having no temporal extension whatsoever, an “interval with no duration” (Augustine 2008, 232) or an “indivisible point of time, which divides the future from the past” (Reid 1785, 326). And since sense perception is limited to the present, we cannot be said to perceive succession by sense. Now, and here Reid also follows Augustine, we do often speak as if the present has temporal extension. We say that we see motion (Reid 1785, 326), and we consider the present as varying according to the particular context of discussion — we speak of the present hour, year, or century (Reid 1785, 326; Augustine 2008, 231-232).4 This kind of inaccuracy, Reid notes, is perfectly legitimate in “common life” (Reid 1785, 326). Philosophers, however, should insist on a “strict” meaning of the expression “the present moment.” This first point leads to the positive aspect of Reid’s argument concerning the origin of our idea of succession:
[I]t is only by the aid of memory that we discern motion, or any succession whatsoever: we see the present place of the body; we remember the successive advance it made to that place: the first can then only give us a conception of motion, when joined to the last (Reid 1785, 327).
- 5 It might be suggested that Reid’s account is of the way we acquire a “conception” (Reid 1785, 327), (...)
5We do in fact experience temporal features, despite the fact that this awareness is not strictly perceptual. Reid is here offering a bare-bones version of the traditional memory theory. On the view, our awareness of change essentially depends on memory, it is memorial awareness.5
6The Reidian account faces numerous challenges. Here’s a first. Take the opening notes of Bach’s Cello Suite #1 in G Major: G-D-B
Figure 1.
Opening notes of Bach’s Cello Suite #1
7The form of the view as presented by Reid above needs to be refined in order to seem at least initially plausible. We would need, for example, to be able to account for a structure in our awareness of change (in memory) that maps onto the progression of the notes. When we experience the first B, it wouldn’t be enough to say that we remember G and D together with the sense perception of B, which would be compatible with D having been heard before the G, and with the G having been heard two months ago (see Dainton 2006/2000, 124). We would want to say that we remember having just heard D with a short-term memory of G and having previously heard G when the D was played. Without temporal distance and order as part of the traditional memory theory, it is wildly implausible.
- 6 Broad frames this as a “notorious fact” (Broad 1923, 351) rather than a particularly original insig (...)
- 7 Another prima facie powerful argument against the memory theory, which can also be levelled against (...)
8It could perhaps be claimed that Reid’s view already includes something like the idea of temporal distance and order, when he writes (in relation to the spatial case) of remembering the “successive advance [the object] made to [its present] place.” Even with the idea of distance and order in place, however, the view seems to face an insurmountable challenge. On the traditional memory theory, remembering what occurred a year ago and what happened two seconds ago involves experiences of the same kind. Likewise, noticing that something changed in relation to the way it was a minute ago, and being aware of a note played half a second ago on the cello involves the same type of awareness. This kind of identity lies in conflict with what is widely taken to be a striking experiential difference between perceiving change and noticing that something has changed, a difference influentially noted by Broad (1923, 351) and Russell (Russell 1927, 280-281).6 Watching a shooting star, for example, is different than noticing that the moon has changed its position. This kind of difference does not seem to amount to one between being aware that change has taken place over one minute, say, rather than one second, a type of difference that can be accommodated within the traditional proposal. The suggestion is rather that are aware of change in the two cases in a different way.7
- 8 Another way for a theory of temporal experience to accommodate this difference is to understand it (...)
- 9 The type of memory described here has also been called “retention,” following Husserl (1991), altho (...)
9A way to respond to this challenge is available from conceptual resources supplied by a non-traditional form of the memory theory.8 On such a view, the perception of change is explained by the contribution of a particular type of memory, distinguished from recollection or episodic memory. The locus classicus of the position is James’ chapter on memory from The Principles of Psychology. The type of memory associated with the immediate perception of change is termed, in James’ account, “primary” or “elementary”:9
[W]hat elementary memory makes us aware of is the just past. The objects we feel in this directly intuited past differ from properly recollected objects. An object which is recollected, in the proper sense of that term, is one which has been absent from consciousness altogether, and now revives anew. It is brought back, recalled, fished up, so to speak, from a reservoir in which, with countless other objects, it lay buried and lost from view. But an object of primary memory is not thus brought back; it never was lost; its date was never cut off in consciousness from that of the immediately present moment. In fact, it comes to us as belonging to the rearward portion of the present space of time, and not to the genuine past (James 1890, 646-647).
- 10 The modified memory theory is defended, for example, in Stout (1930), Husserl (1991), O’Shaughnessy (...)
10James is here articulating the idea setting apart what can be called the modified memory theory (following Phillips 2010; 2018 and Hoerl 2009) from the traditional account. Our perception of change, on this version of the view, is accounted for by the constitutive contribution of memory, in that temporal perception would not be what it is were it not for the involvement of primary memory. This involvement, however, is not susceptible to the challenge expressed by Broad and Russell. The modified view allows us to bring out a difference in kind between perceiving change (e.g., watching a shooting star) and noticing that change happened (e.g., noticing the moon has moved). Direct, or immediate, perception of change is due to the preservation of the perceived which is characteristic of memory, but this is a preservation of an object which “was never cut off in consciousness from […] the immediately present moment.”10
- 11 Though see a number of remarks in Sinclair’s Bergson (2019, 47-48, 93), and a brief related note in (...)
11The relation of Bergson’s thought to the development of the memory theory has not been examined in significant detail.11 In Time and Free Will, Bergson hints that memory plays an essential role in our perception of change, when he explains the experience of temporal properties as due to the “synthesis carried out by our consciousness between the actual positions and what our memory calls the former positions [la synthèse opérée par notre conscience entre la position actuelle et ce que notre mémoire appelle les positions antérieures]” (Bergson 1910, 124). But this is not decisive. The work does not explore the relations between perception and memory, so it does not seem to be possible to ascertain whether the view is a “traditional” or “modified” version of the account. From 1896’s Matter and Memory, however, we can see Bergson presenting an account of change perception which appeals to memory, and which involves distinguishing this kind of memory from recollective or “secondary” memory:
However brief we suppose our perception to be, it always occupies a certain duration, and involves, consequently, an effort of memory which prolongs, one into another, a plurality of moments […] Memory in […] two forms, covering as it does with a cloak of recollections a core of immediate perception, and also contracting a number of external moments into a single internal moment, constitutes the principal share of individual consciousness in perception, the subjective side of the knowledge of things (Bergson 1988, 34).
- 12 This distinction is noted by Sinclair (2019, 93) and Perri (2017, 510-511).
12The relevant point in this passage is the distinction Bergson makes between (i) recollection and (ii) a memory which preserves phases forming the continuity characteristic of durée in a “single internal moment,” and in that sense seems to be part and parcel of immediate perception itself.12 Memory is constitutive of the “concrete present” — the present as experienced — quite a different thing than what one might suppose when taking up a “strict” definition of the present as an “indivisible limit which divides the past from the future” (Bergson 1988, 150).
13The relation of durée to memory, and the difference between the two types of memory is made further explicit in a passage from Duration and Simultaneity’s third chapter (“Concerning the Nature of Time”), in which Bergson characterises the continuity of immediate experience:
What is this continuity? That of a flow or passage […] and this transition, all that is naturally experienced, is duration itself. It is memory, but not personal memory, external to what it retains, distinct from a past whose preservation it assures; it is a memory within change itself, a memory that prolongs the before into the after (Bergson 1965, 44).
- 13 By the claim that durée “is” memory, I take Bergson to be arguing that memory plays an essential ro (...)
- 14 Bergson’s recently published Collège de France course on memory, delivered between the years 1903-1 (...)
14Two things can be underlined here. First, Bergson explicitly identifies durée with memory, which rules out taking memory to possess only a causal role with regards to the experience of change.13 Secondly, Bergson further distinguishes between the type of memory implied in the continuity of experience, a “memory within change itself” (James’ primary memory, which is not “cut off in consciousness [from] the immediately present moment”), from “personal memory”14 which is “external to what it retains” (James’ secondary memory, which “has been absent from consciousness […] and now revives anew”).
- 15 One can also find frequent discussion of the relation between memory and temporal experience in the (...)
15We can, then, secure an understanding of Bergson’s position as taking up the idea central to the modified memory theory: memory of a specific kind, a memory “within change itself,” plays a constitutive role in immediate temporal experience.15
16Contemporary work on temporal experience does not frequently appeal to the memory theory. Accounts of the non-inferential experience of time typically fall under one of three models: cinematic, retentional, or extensional. The cinematic view, on which perception is limited to instantaneous “snapshots,” relies rather straightforwardly on the memory theory in its traditional form. A cinematic theorist seems to need recourse to memory in order to explain our experiential grasp of temporal properties. But the cinematic model has been largely rejected during the past century (though see Chuard 2011; 2017 for a recent defence of the view).
- 16 See Tye (2003), Grush (2005; 2007; 2016), and Lee (2014) for recent versions of the retentional mod (...)
- 17 See Dainton (2006/2000; 2008; 2014; 2016), Phillips (2010; 2018), Hoerl (2009; 2013; 2017), and Sot (...)
- 18 Lee’s (2014) view is a version on the retentional model, in that on his account our experience of c (...)
17Both the retentional and extensional models allow for the immediate perception of temporal properties, while differing with regards to the relation they describe as obtaining between temporal experience and its content. On the retentional view, temporal properties (succession, order, duration) are considered as contents of instantaneous, or quasi-instantaneous experiences. The model is sometimes understood as involving the application of a general lesson concerning representation to the case of temporal experience: just as we do not take our experience of a red flower, for example, to be itself red, our experience of time need not be taken to possess (significant) temporal structure. The representational vehicle (in this case, the experience representing temporal properties) need not be understood as possessing the structure of its contents (the temporal properties).16 The extensional model, by contrast, involves identifying an explanatory connection between the experience of time and its contents. On this view, our experience of temporal properties itself unfolds, or progresses in time. Our experience of temporal properties is explained by the temporal extension of experience.17 Versions of both the retentional and the extensional theories, the two competing models of temporal experience in contemporary literature, do not typically employ the memory theory. One can find, for example, explicit criticism of the memory theory in Lee’s (2014) retentional model,18 as well as in Dainton (2006/2000) and Hoerl’s extensional accounts of temporal experience (Hoerl 2009).
18In recent work, Phillips has defended a form of the memory theory which he takes to apply to both the retentional and extensional models (Phillips 2010; 2018). A broad aim of his paper involves suggesting that the distinction between the models of temporal experience does not hinge on whether they appeal to the memory theory. Seeing that all the models might well involve attributing to memory a constitutive role in change perception would allow us, according to Phillips, to focus on the central issue with regards to which the views diverge: the relation between temporal experience and the temporal profile of experience itself. More specifically, on the question of whether temporal experience itself unfolds in time (Phillips 2018, 294-295). I will return to this suggestion in the conclusion of this paper.
- 19 Phillips here draws upon Martin’s (2001) account of episodic memory.
19Phillips pursues this goal by specifying the claims made within the memory theory in its modified form and showing that the theory can be understood as applicable to an understanding of temporal experience on the retentional view — contra Lee’s rejection of the memory theory (Lee 2014) — as well as the extensional view — as defended by Dainton (2006/2000). Phillips’ argument starts from appreciating that the distinguishing feature of all forms of memory is a preservation of “cognitive contact” with the past, or the continuation of past psychological success.19 Episodic memory, James’ “secondary” memory, involves preservation of past acquaintance with things or events; semantic memory involves the preservation of past knowledge (Phillips 2010, 194; 2018, 291). Primary, or elementary, memory can be seen as involving the continuation of past psychological success by allowing us to have experiences of things succeeding one another. If we are considering the succession of notes do to re, for example:
[H]earing re involves primary memory insofar as it involves hearing re in a particular way, namely as succeeding from do. This modification of one’s manner of awareness plausibly counts as a form in which psychological success (namely awareness of do) can be preserved (Phillips 2018, 291).
20Crucially, the idea of preservation of past psychological success in primary memory does not depend on the presentation of preceding parts or phases of a succession as past — the modified form of the memory theory does not involve tensed experiences of the phases in succession (Phillips 2018, 290-291). Successive phases can be considered as parts of one unified, albeit complex, perceptual experience of change. In that way, the modified memory theory can hold on to the distinction between noticing that change happened — which would include tensed experiences — and the immediate perception of change. The latter involves the preservation of past success in a memory “internal to change,” experiences which are not retrieved since they have never been absent.
21This construal of the memory theory in its modified form, in line with the ideas from James and Bergson presented above, allows Phillips to argue that both the retentional and extensional accounts can be understood as appealing to memory (Phillips 2018, 289-294). The idea of the preservation of past psychological success, according to Phillips, is implied in the very idea of an experience of succession, regardless of the relation a theory of temporal experience takes to obtain between the content of temporal experience and its structure. That is, regardless of whether experiences of succession are themselves successive. Both the retentional and extensional approaches assert that we immediately experience succession, and experiencing succession involves the preservation of past psychological success — and hence memory.
22One may wonder, however, whether the turn to memory has been decisively motivated in the context of Phillips’ argument. The preservation of past psychological success is meant to explain our experience of temporal properties, but it has seemed to a number of theorists that the appeal to memory in this context is unnecessary. Hoerl, for example, sketches out an alternative on which succession simply falls under the scope of a temporally extended experience (Hoerl 2009, 8), which need not involve any appeal to memory. The modified memory theory, according to Hoerl, does not seem to describe a real contact with the past, but only a “proxy for that contact” (Hoerl 2009, 4). As I understand Hoerl’s point, it is that when we look at what the memory theory is meant to account for — the experience of succession — perception itself already possess all the necessary resources.
23Dainton also expresses a similar thought, recently noting with regards to the involvement of memory in Bergson’s durée that “a case can be made […] for holding that in these contexts [i.e., the contexts in which Bergson examines the relation of durée to memory] Bergson’s ‘memory’ is simply the unifying relation which connects the earlier and later phases into a single episode of durée” (Dainton 2017, 104). On Dainton’s view of temporal experience, the experiential unifying relation which binds together phases of change into a single experience — “co-consciousness” — does not involve memory of any kind (Dainton 2006/2000, 126-127; 2010/2001, 105-106). An analogous line of thought comes up in Lee (2014), who takes it that a single state, devoid of any relation to memory, can satisfactorily be considered as having succession as its content. There is no work to be done by memory in the immediate apprehension of succession.
24It is difficult to make out who has the upper hand in this debate. Given that what we are accounting for is the experience of succession, there does not seem to be a decisive reason either in favour, or against, the memory theory of change in its modified form. Phillips’ version of the view, in line with the aforementioned points mentioned in relation to James and Bergson, is coherent, and successful in escaping the challenge articulated in Broad’s work. There is a measure of intuitive appeal to the idea of a preservation of past psychological success as required for an account of change perception, but it seems that the modification of awareness involved in experiencing succession can just as well be thought (as suggested by Hoerl, Dainton, and Lee) to be inherent to temporal experience itself, as distinct from memory.
25Bergson, as noted above, presents a view in accordance with the idea at the heart of the modified memory theory. As in James’ and Phillips’ accounts, on Bergson’s view a type of memory distinct from recollection is implied in our non-inferential experience of change. Memory prolongs the “before into the after” (Bergson 1965, 44) and is thus essential to our experience of succession. As I noted with regards to Phillips’ proposal, this idea in itself does not seem to decisively carve out an explanatory niche for the memory theory. What I would like to suggest, at this stage, is that we can find in Bergson an additional claim that could make the appeal to memory more persuasive. The claim is that our awareness of change does not only involve the preservation of past psychological success; it involves the preservation of that which is “receding” within the present of our perception, the prolongation of that which is no longer in what is.
26According to Bergson, durée is the “continuous life of a memory which prolongs the past into the present” (Bergson 1946, 211). In this sense, the role of (primary) memory in immediate temporal experience involves the preservation of the past, the “building up” of states in the course of changes and movements. My experience of the succession of the notes G-D-B in the opening of Bach’s Cello Suite #1, for example, is in this sense due to memory. So far, we are in the realm of the modified memory theory as presented by James (1890), and as recently defended in Phillips (2018). But the awareness of the change also involves memory because these notes recede into the past as they are perceived; or rather, as Bergson might have preferred, a new present is arising at each stage of the succession, constantly replacing the past moments. This is particularly evident in a passage from Duration and Simultaneity:
- 20 There is a certain looseness to the expression “impossible to distinguish between” here (as also no (...)
[I]t is impossible to distinguish between the duration, however short it may be, that separates two instances and a memory that connects them,20 because duration is essentially a continuation of what no longer exists into what does exist. This is real time, perceived and lived (Bergson 1965, 49; italics mine).
27The idea, I think, is experientially compelling, even if we seldom pause to appreciate it or its significance. When we hear the note B in the phrase G-D-B of the cello suite phrase, for example, there is a sense in which the G is no longer, though it is still present, or preserved, in our consciousness. Jumping into a cool lake on a summer’s day, when we hit the water our experience of swiftly moving through the air is preserved in our awareness but is “no longer.” What is recognised by Bergson’s suggestion is quite simple, notwithstanding the apparently paradoxical formulation of a “continuation of what no longer exists into what does exist.”
- 21 Dainton’s point concerning the relation of memory to durée noted above — that it is “simply the uni (...)
28The idea of a continuation of what no longer is into what exists is arguably absent from the modified form of the memory theory as discussed above, on which what memory explains is only the preservation of past moments. And it seems to allow decisively singling out a role for memory in change perception. Beyond preservation of successive terms, the experientially appealing idea of a perception of what is “no longer,” in the present, is the key.21
29Now, the immediate response to this idea could well be that it seems to bring us within the remit of the critique applied to the traditional version of the memory theory, based on the idea captured in Broad’s (1923) and Russell’s (1927) remarks. The criticism would apply if what we would be describing was an experience that is of the past, in the same sense that episodic memories are. This would mean that when we experience the note B in the succession of notes G-D-B within the specious present, for example, the G would be “no longer” in the same sense that hearing the G 30 seconds or 5 hours ago would be “no longer.” Watching a shooting star would involve phases which are “no longer” in the same sense that the moon has moved involves past experiences which are “no longer.” We would not then be able to account for a distinction between noticing that change happened and perceiving change immediately, which is the important insight expressed by Broad. This would indeed be a bad result.
30This, however, is not what the version of the memory theory which involves recognising “the continuation of what no longer exists into what does exist” implies. The key, I think, is that the “continuation” we’re invited to notice is not of an object or content as past (e.g., in the past experiences of the moon’s location), but rather what might be called the becoming present of the past. To put it in terms of tense, we’re invited to consider not a tensed experience, but an experience which can be put in terms of “dynamic tensing” — of the moments becoming present, becoming past, continuously. This experience of “tensing” in the scope of the present, absent from the account presented by James and Phillips, indicates the manner in which a primary form of memory (distinct from episodic, or “personal,” memory) is constitutive of the non-inferential experience of change. The experience is tensed when it is no longer within the scope of the specious present, but while it is, it is dynamically experienced as being preserved while becoming past.
- 22 The dynamic memory theory I’m describing has interesting similarities with Husserl’s “modal retenti (...)
31Bergson’s ideas on memory and temporal perception allow us, then, to articulate what might be called a “dynamic memory theory.” The view can be considered as a version of the modified memory theory, insofar as the role attributed to memory as partially constitutive of change perception, prolonging the “before into the after” (Bergson 1965, 44), is essentially the same as that described by Phillips. The dynamic view, as with Phillips’ construal of the modified account, avoids the objection which was seen to be deadly for the traditional memory theory. The modified theory as such, however, is meant to be compatible with views (e.g., Lee 2014) that deny any role for tense in non-inferential temporal experience. As I’ve suggested above, Phillips’ articulation of the theory does not in itself provide a conclusive motivation for the appeal to memory. The motivation, I’ve claimed in this section, can be provided in the context of the dynamic view, differentiated from other possible versions of the modified theory by the way it accommodates the phenomenon of “dynamic tensing” within the scope of non-inferential change perception.22
32I have argued that Bergson’s view is in agreement with insights expressed as early as William James’ work and more recently in publications by Phillips, concerning the involvement of a form of primary memory in temporal experience. I have also suggested that Bergson’s construal of the memory theory might allow to address a challenge arising with regards to a recent formulation of the account.
- 23 See Dainton (2017). In recent and forthcoming work (Wolf 2020), I discuss Bergson’s relation to the (...)
33There might be reason to worry, however, that the memory theory of temporal experience lies in tension with an additional theoretical commitment taken up in Bergson’s work. As I noted above, Phillips’ argument in favour of the memory theory is meant to apply to both the retentional and extensional models. And Bergson’s position on the experience of time is arguably a version of the latter — on his proposal, temporal experience itself unfolds in time just like its content.23 Now, the problem here concerns the fact that the contribution of memory in (immediate) experience seems to rely on a notion of simultaneous presence with a precept or sensation. Taking memory of phases, a change (e.g., the opening notes G-D of Bach’s suite) to be partially constitutive of the experience of change appears to amount to the view that the remembered contents are simultaneous with the successive phases (e.g., the B note which succeeds them). The result would be pulling apart the content (the succession G-D-B) from the time it takes the experience to occur, since the temporal content allegedly results from the simultaneous conjunction of a precept (e.g., the note B) and the memory of preceding phases (e.g., notes G-D). The conjunction might itself have some temporal extension (e.g., of experiencing the B note), but this extension does not conform to the extension of the experienced content (e.g., of the phrase G-D-B).
34This issue merits further discussion, but one way to approach this concern is to consider the memory constitutive of the immediate experience of change apart from the notion of simultaneous presence. Primary memory, a memory “within change itself,” might not rely on the kind of simultaneous conjunction with present precepts which characterises the availability of secondary, or episodic, memory. It can be taken to be implicated in the “distention,” or “stretching out” of the perceived present, characteristic of experience itself and not merely its objects.