Notes on the Bergsonian bits of Deleuze, G. (1992) Cinema 1.The Movement  Image

Dave Harris

Chapter one: Theses on movement.  First commentary on Bergson

Bergson has three theses on movement, with the first really acting only as an introduction.  The first thesis says that movement is distinct from the space covered, 'space covered is past, movement is present, the act of covering'(1).  However, movement is an integral part of the whole: we cannot split it up into static spaces or instance in time, 'immobile sections', since by themselves, they would never move.  There has to be some movement added, but the usual notion implies some 'abstract idea of the succession, of a time which is mechanical, homogeneous, universal and copied from space, identical for all movements'.  However, there is another movement which explains this succession, something which occurs in the interval between points allocated by mechanical time, 'a concrete duration', something qualitative. 

At first, Bergson saw the cinema as offering an illusory notion of time, with static images joined by some abstract [in the mechanist sense] time provided by the apparatus itself.  This will be false movement.  He saw cinema simply reproducing the natural perception of movement, something abstract and mechanical, something 'constant and universal'(2).  However, cinema offers something else, not just immobile sections, 24 images per second, but some 'intermediate image', something which gives us movement immediately.  Unlike natural perceptions, there is nothing to correct the movement of images [nothing which stabilizes or selects among them as a result of human intentionality].  The cinematic image appears 'without conditions', some immediate 'movement - image', something which is itself mobile.  Bergson had already suggested this possibility, in Matter and Memory, in the first chapter [I rendered it as:  The notion of the image works both ways, to deny both idealism and simple materialism. We already have the notion of a relation of movement replacing static binary divisions.].  Bergson probably 'forgot' this statement, because the new always takes time to emerge, as in his view that 'when it began life was forced to imitate matter' (3).  And we have to remember that in the early days, cinema was a matter of fixed shots which were immobile, and the technology did indeed combine cameras with projectors 'endowed with a uniform abstract time'.  The real breakthrough was in the development of 'montage, the mobile camera and the emancipation of the viewpoint, which became separate from projection'.  The shot became a temporal not a spatial one, and the section itself mobile.  This is what was to support Bergson's original idea in Matter and Memory [confirmed in Canales' book].

The second thesis is found in Creative Evolution, where the illusions of natural perception are further analyzed.  The classical conception sees movement as a transition from one ideal form to another, 'an order of poses or privileged instants, as in a [classical eg Greek] dance' (4).  However, modern thinking involves the notion of 'any - instant - whatever', a series of normal material elements which were immanent rather than transcendental.  Deleuze gives a number of examples from Kepler, Galileo, Descartes on movements on a straight line, and finally Newton and Leibniz on calculus and 'sections which could be brought infinitely closer together'.  This also had the effect of seeing time as an independent variable [that is, a separate quantifiable variable, not linked to some transcendental motion of necessary or ideal movement?].  The film camera has a role in this transformation, expressing movement between ordinary instants [and Deleuze says we see this in the films of Wenders].  What the film camera does, in Bergson's first thoughts, is to take snapshots rather than posed photos, and to suggest an equal distance between them, made possible by a perforated film and a mechanism to move images at standard rates.  Classical drawing might still rely on poses, but drawing for animated cartoons shows figures already forming and being dissolved 'through the movement of lines and points taken at any-instant-whatevers of their course' [the mundane and tedious 'tweens' that luckily computers now do for you]. 

You can still find 'privileged instants' in cinema, such as Eisenstein's films which work with 'certain moments of crisis', 'peaks and shouts' which come into collision, and which he called 'the"pathetic"' (5).  By analogy, we know that the classic films of galloping horses also show particular interesting instants, where only one hoof is on the ground, for example, but these are best seen as 'remarkable occasions'[as in Leibniz where remarkable occasions are those points on a curve where something unusual happens, amid all the unremarkable points].  These are also called 'singular points', but they do not arise because some transcendental principle is being actualized in a privileged pose, they are still any-instant-whatevers.  The interesting question is how singular points get produced by movements among ordinary ones, how 'the qualitative leap is achieved by the accumulation of banalities (quantitative process)'(6).  In Eisenstein's terms, the pathetic presupposes the organic, which organizes the set of any-instant-whatevers.

So cinema shows movements between any-instant-whatevers, and this seemed to contradict the analytic interest in science.  And art seem to be stuck with the notion of some higher synthesis of movements between privileged poses or forms.  In this sense, cinema was always ambiguous, neither science nor art.  However, some arts were also getting interested in any-instant-whatevers, especially dance, and mime.  One cinematic example is Fred Astaire's '"action dance"' associated with musical comedy, which takes place in the street or along the pavement, in other words in 'any-location-whatever'(7).  Another example would be Chaplin developing the 'action - mime'.

Bergson was aware of these developments, but showed two contradictory ways to think about them.  In the first, we can still see movement as being something which links static images, whether these are eternal poses or merely [temporarily]  immobile sections.  This is still flawed, because the elements are thought of as some abstract self contained whole, in which everything is already given.  However, he wanted to argue that real movement is something outside these given wholes, not something that takes place within a closed system [where possible movements are already circumscribed].  In the second thoughts, we insist that even these predictable movements allude to 'the production of the new, that is, of the remarkable and the singular at any one of these moments'.  This is what Bergson wanted to do, including his project to give even [Einstein's] science a metaphysics.  It follows that a philosophy might also be developed from art and cinema, that these media have a role to play in developing this new way of thinking.  This is the second thesis, which sees cinema as'the organ for perfecting the new reality' (8).

The third thesis is found in Creative Evolution.  We can describe the instant as 'an immobile section of movement', but go on to see another kind o fmovement  as 'a mobile section of duration'.  It thus becomes a section of the whole, something capable of bringing change: duration 'changes and does not stop changing', while matter merely moves 'but does not change'[qualitatively, that is].  Ordinary movements occur as 'a translation in [ordinary] space', but those movements also produce qualitative changes in the whole.  In Matter and Memory, for example we can explain the movement of an animal between two points abstractly [as distance covered in a particular time], but there's also a qualitative dimension, as when that movement is produced by a search for food: when the movement is completed, 'what has changed is not only my state, but the state of the whole which encompassed [the points] and all that was between them'.  Similarly, the fall of a body presupposes another one, and results in a change in the whole 'which encompasses them both'.  Bergson has discovered not just the translation of a fixed state, but other kinds of movements 'vibration, radiation'.  The qualities can be seen as 'pure vibrations which change at the same time as the alleged [self contained] elements move'(9).  There is another famous example in Creative Evolution which involves putting sugar in a glass of water and having to wait until it dissolves [and thus experience duration].  The point is that a qualitative change is going on from water to sugared water, a change in the whole [which was formally made up of separated sugar and water, and might also include a glass and a spoon].

Bergson explains the difference between the old conception of change as a combination of an immobile section and a movement, while the new conception has a mobile section related to qualitative change.  That is the proper representation of reality.  The example of the glass of water certainly adequately represents subjective reality, but it is also a change in a objective reality.  The mistake that positivist science makes is to attempt to fully define the whole [close the system].  The recurrent failures to do that adequately  led to skepticism [or operationalism, so there is a good enough closure of the system], but what it really tells us is that the whole is 'the Open', and that the Open endures—'its nature is to change constantly, or to give rise to something new, in short to endure'.  When we experience duration we can conclude that 'there exists somewhere a whole which is changing and which is open somewhere'.  It is not surprising that Bergson explored duration first of all via consciousness, but later he concluded that consciousness helps us experience duration only because it opens itself to the whole, discovers the opening up of that whole.  He goes on to say that living beings change when they do the same thing, open themselves to the whole [expose themselves to something outside a closed system—this is what drives evolution]. 

We see the importance of relations, ones which are external to their terms [that is which are not predicted in terms of pre specified possibilities within a closed system].  They are always linked to the open.  They do not belong to objects or closed sets of objects.  They bring about transformative or qualitative change.  Duration or [philosophical] time reveals a whole set of relations [or all the relations in the Open.  This is another way of thinking about the multiplicity, of course, which contains all the possible relations between virtual and actuals themselves, and all the ones within each category].

The whole cannot be considered in terms of sets, which are always closed.  The whole does have parts, but only 'in a very special sense, since it cannot be divided without changing qualitatively at each stage of the division' (10) [just like the definition of the rhizome or the multiplicity again].  On the contrary, the whole shows that sets can never be closed, must always be kept open, even in the most tenuous way, always connected to the rest of the universe.  The whole creates itself, 'like the pure ceaseless becoming which passes through [specific, actual] states'.  This is how it appears as something spiritual or mental, something graspable by consciousness, which can see the inadequacy of abstractions and closed systems.

Closed sets or systems can be well founded.  The link to the open can be reduced, and this might even be necessary given particular organizations of matter and space [and the need to work with them].  But it is still a mistake to confuse a set with the whole and to ignore qualitative change, and we still need to bear in mind that there are two conceptions of the relation between movements and time.  In fact, we have three 'levels': sets or closed systems which can be defined in terms of their objects or distinct parts; movements as translation between these objects which modify their position; the whole which constantly changes 'according to its own relations' (11).  Thus we have two kinds of movement—translation and duration.  Overall, movement relates fixed elements and systems to duration and vice versa [duration has the effect of opening up the systems.  It's starting to remind me of Kuhn on the progress of science, with normal science being interrupted now and then by something new from the outside producing a scientific revolution, although not without a lot of resistance].  To summarize, in terms that will guide the rest of the book, Bergson suggests that: there may be 'instantaneous images...  immobile sections of movement', but also images which show immobile sections of duration [the movement-image proper], and also images of duration and qualitative change, beyo
nd simple movement, and these are 'time images'.

Chapter four: The movement - image and its three varieties.  Second commentary on Bergson

Earlier conceptions of movements thought images were held in consciousness, and movements in space.  The images in consciousness could only be 'qualitative and without extension' (56), while movements were only 'extended and quantitative'.  The problem was to move from one to the other, to explain that movements can suddenly produce an image as in perception and conversely that image can produce a movement as in voluntary action.  One answer was to look at the 'miraculous power' of the brain, but there was still problems because movement could clearly be seen already as 'at least a virtual image', while the image could be seen as possible movement.  The problem really lay in an old confrontations between materialism and idealism, with the first seeing consciousness as a matter of material movements, and the other seeing reality as produced by pure images in consciousness.  There were also certain new developments external to philosophy, including the clear role of images in the material world, and the cinema was a key development.  This duality had to be overcome, and both Bergson and Husserl attempted to do this.  The alternatives were that consciousness just is something, or that all consciousness is consciousness of something, respectively.

Neither mentioned the cinema, or at least Bergson did only negatively at first.  Sartre ignored the cinema.  Merleau-Ponty attempted a phenomenology, but like all phenomenology, it took natural perception as the norm, with images as  'existential coordinates' which anchor the knowing subject in the world [intentionality is the concept I thought of to explain the phrase that all consciousness is consciousness of something].  Movement becomes a sensible form, a gestalt, which organizes the field in which intention and perception can operate.  This connects with a classical motion of movement as a matter of poses, although existential rather than essential ones.  The cinema was not seen as very helpful here, because it can challenge both the existential anchor of the subject and the horizon of perception, offering 'an implicit knowledge and a second intentionality' (57), unlike the other arts which can produce something unreal from the real world: with cinema 'it is the world which becomes its own image'.  This is why the cinema was initially seen as offering false perceptions, but also as extending them. 

For Bergson, the cinema misrepresents movement, but this is just the same as natural perception: both take snapshots of reality.  Bergson's model is not that of natural perception, but of things that constantly change, 'a flowing-matter in which no points of anchorage nor center of reference would be assignable'.  Instead, centres have to be formed and imposed to produce fixed views, an operation in both natural and cinematic perception that requires investigation.  However, cinema did offer one possibility not open to natural perception in this respect: the latter moves from the 'acentred state of things to centred perception', but cinema can do the reverse and show the acentred state of things itself.  Bergson saw at least saw this possibility [apparently in the first chapter of Matter and Memory—see my notes]. 

We can develop this by thinking of a world [of cinema] where 'IMAGE=MOVEMENT' (58).  We can initially see the image as the 'set of what appears', with no interaction with other images, no mobile section, nothing moving away from the initial perceived moment.  Instead, the image is 'indistinguishable from its actions and reactions' [ie a closed set].  However, [and this is the crucial step] we can also see it as 'universal variation', via Bergson, acting as a conduit for all the modifications at work in the entire universe.  Although images might seem to be immobile, they are therefore acting and reacting at this universal level, with all their facets and elements reacting.  Every atom is an image like this, with actions and reactions extended almost infinitely. If we think of our brains or bodies as images like this that means they are also open to external images and movements from the universe.  The brain is not some universal arbiter above images in its consciousness.  Instead 'me, my body, are rather a set of molecules and atoms which are constantly renewed'.  Even atoms should really be understood as a set of 'interatomic influences' [real posthumanism here!].  In fact, 'matter [is] too hot for one to be able to distinguish solid bodies in it', and we see only universal variation, 'universal rippling' with no centres or other fixed points or dimensions.

'The infinite set of all images constitutes a kind of plane of immanence' (58-9).  When images exist in themselves on this plane, they constitute matter, an 'absolute identity of the image and movement'(59).  This identity of image and movement means that 'the movement image and matter are identical...  The movement-image and flowing-matter are strictly the same thing'.  The operations are not mechanical, since mechanism implies closed systems and snapshots ['immobile instantaneous sections'].  These do exist, but they are 'cut from this universe or on this plane', something made exterior.  By contrast, 'the plane of immanence is the movement (the facet of movement) which is established between the parts of each system and between one system and another, which crosses them all, stirs them all up together, and subjects them all to the condition which prevents them from being absolutely closed'.  It is itself a section [of pure chaos], but 'a mobile section, a temporal section or perspective...  A bloc of space time'.  The time of movement within it is also a section of 'all time' [presumably Aion].  There is an infinite series of these mobile sections, 'so many presentations of the plane, corresponding to the succession of movements in the universe'.  This is machinism as opposed to mere mechanism, and the material universe, [and note that the material includes the plane of immanence] is 'the machine assemblage of movement images'.  This is citing Bergson, and one implication is the famous one that 'the universe [acts] as cinema in itself, a metacinema'.  Thus have we developed a theory of cinema from Bergson, although it is completely different to his own account.

What we're left with is 'images in themselves which are not for anyone and are not addressed to anyone'.  These are not the same as bodies, however [too actualised].  Bodies are a result of our perception and our language which can assign qualities and actions to them.  We customarily work with a system of nouns, adjectives and verbs.  But action in this case is not [pure] movement, but rather a more limited notion involving something which is directed at some intended result.  The same goes for quality which appears to offer a limited movement between a state which persists 'whilst waiting for another to replace it'.  Body confines movements to 'the idea of the subject...  Or of an object which would submit to it' (60).  Again we do find such images, appearing as action images, affection images and perception images, but they are different from movements 'which are called images in order to distinguish them [for the sake of convenience] from everything that they have not yet become'

More positively, we can see that the plane of immanence 'is entirely made up of Light', and this light diffuses without resistance or loss [citing Bergson again -- light for him meant electromagnetic radiation of all kinds, or the basis forces of the universe?] ].  Light is identical to matter as well, and this provides the identity between image and [real] movement. Apparently, Bergson was to apply this notion of movement as light to relativity theory.  This shows his intention to produce a philosophy of modern science, not an epistemology, but involving the 'invention of autonomous concepts' which would fit with the new terms of science.  Bergson wanted to start with a diffusion of light on the plane of immanence, initially producing movement images which were only lines or figures of light.  These become apparent only when light is interrupted, reflected or stopped, say by a human eye.  However, this implies that the eye 'is [itself] in things, in luminous images in themselves' [not something separated from the universe, but a potential so to speak?], and thus human perception is both interior to the universe and capable of acting anywhere in it [maybe -- this is difficult].  There is no opposition of consciousness and things, as in phenomenology with its notion of consciousness as some kind of light ray aimed at aspects of the exterior [this is why we had to talk about light from the beginning, I assume --everyone used optical metaphors].  For Bergson, things are already luminous, consciousness is a part of things which share the image of light.  Consciousness is not grounded in a particular source of light giving insight, but is potentially at least everywhere, itself located on the plane of immanence and with a particular capacity to stop or reflect light [I heard echoes of Hegel here and the role of Spirit, because somehow this is important for the plane of immanence as well—incidentally, I think there's a typo on 61 which talks about the capacity to stop light as something 'which the plate {sic} lacked], anywhere on the plane. [Deleuze wants to also describe this capacity as providing a black screen, to link with that strange stuff about screens, actually white ones, in ATP].  When humans actually perceive something, they are picking up one of these opacities, intercepting light that is being propagated universally. 

After this difficult section, Deleuze summarizes: 'the plane of immanence or the plane of matter is: a set of movement images; a collection of lines or figures of light; a series of blocs of space time'(61).

We can only explain actual happenings using the characteristics or factors already identified.  The first possibility is that an interval can appear on the plane, 'a gap between the action and reaction'.  An interval implies a dimension of time.  This interval gives living things or matters their distinctive characteristics as images.  All other images act and react using all their facets and parts, but living images receive actions on one facet or in one part, and, produce reactions on other parts—they are '"quartered"'.  This is a way of explaining their receptive or sensorial capacities.  What they do is isolate certain images from all those that might be available in the universe, and organize these in a more closed system as perceptions.  This is the same operation as cinematic framing.  Similarly, reactions select elements after an interval and organize these as new movements.  Again the delay introduces an element of selection and unpredictability and this is '"action" strictly speaking'(62).  These processes of selection mean that living images are '"centers of indetermination"'

When light encounters living images, it becomes opaque.  The reflected image is a perception.  This introduces another system, a center.  Images can be referred in two ways: to their connections with all the other images reacting across all  facets and parts, and secondly to their relation to a single image [the living image] which receives other images and then reacts to them.  It is not the human brain that is the center of images, because it is itself a special image.  We have to start with the whole plane of images interacting among themselves.  For Bergson, the brain image implies 'a highly complex and organized state of life' (63) [in Matter and Memory, apparently, before we went on to consider the elan vital].  All living beings are capable of inserting intervals of the kind discussed above, and we can suggest that these intervals even occurred in the primeval soup at the start of life, when forms of matter emerged.  This only occurs once the earth, the earthly plane of immanence has cooled down sufficiently [this is the basis for DeLanda's account].  The evolution of solid matter is accompanied by the 'organization of images in more and more elaborate perception', corresponding to the development of more elaborate solids.

The thing and the perception of the thing are united in the image.  These images are related to two systems of reference: as a thing, there is a relation to all the other images as above; as a perception, there is a relation to the special living image and its selective activity.  Human perceptions select from things, subtract the aspects which are not relevant to needs, those needs themselves provided for by the receptive facets and actions retained from our own relation to other things.  Human subjectivity is at first subtractive.  This is only possible once the thing itself is presented as an initially diffuse perception, as image.  Non living things do not subtract like this, hence the famous statement that 'the atom perceives infinitely more than we do' (64), [which might excite fans of Barad].  We can use the term prehension [to refer to this diffuse perception before human intentionality provides apprehension, if I remember my phenomenology].  It follows that there are objective prehensions that things have of other things, and subjective prehensions that living beings have of things.

The interesting thing about cinema is that it does not just rely on the subjective perceptions.  It can deal with mobile centres and variable framings, and thus exceed natural perception, displaying the first system of relations as above, 'universal variation, total, objective and diffuse perception' (64).  It can actually change between the two, showing both the 'total objective perception which is indistinguishable from the thing' and subjective perception which features elimination or subtraction.  When we do this subjective perception, we have the perception image as the 'first avatar of the movement-image'[curious terminology], which begins to organize by awakening the centre of indetermination.  Once we have a center, we can organize the universe around it, 'incurve' it, treat it as a horizon.  This also awakens the possibility for action, which arises after a delay, as we saw, while things are organized.  The whole thing combines to provide the sensory motor schema, the connections between perception and action: perception reveals certain unstable facets of things and this makes action possible to use them.

It is possible to experience and understand this combination that links periphery and center as both '"virtual"' and '"possible"' action  (65). What actually happens is that the distance between core and periphery is altered, a spatial dimension that corresponds to the delay.  The more distant, the more that virtual action is extracted [virtual here meaning potential,but also emanating from the unperceived real?].  This produces the second avatar, the action image which follows imperceptibly from perception, once the universe has been incurved.  This incurving allows for possible actions on our behalf and also 'the virtual action of things on us'.  This provides another classic material [objective]  aspect of subjectivity.  Linguistic activity thinks of actions as something to do with acts or verbs aimed at some assumed end.

The interval is provided because there are two 'limit-facets', perceptive and active, but there's also an 'in between', occupied by affection that connects perceptions and hesitant actions.  It offers a 'coincidence of subject and object', involving what experience feels like [when impacted by objects?], as an aspect of subjectivity.  It lends a quality to movement, as expressed in adjectives.  It is the returning of aspects of the external that had been selected out by perception, still not actually perceived subjectively but representing some pure relation between object and subject.  It is necessary to the whole system of perception and action, overcoming the apparent immobility [passivity?] of perception compared to the necessary activation of action.  Affect 'replaces the action which is become momentarily or locally impossible'(66) [which is what gets the politics of affect advocates so interested - affect produces something immediate, before cognition catches up and has the chance to censor reaction].  This is why Bergson defines affection as a motor tendency acting on some receptive plate which has been [temporarily] immobilized [Matter and Memory].  Thus affection reestablishes the relation between perception and action, making action a form of expression, not just a mere translation, adding a quality.  Micromovements in the face show this well.  Thus movement-image is divided into perception images, action images and affection images, and human subjects are 'nothing but an assemblage [agencement] of three images', a consolidation of them. 

We can consider this in reverse, seeing how we can work back from human consolidation of these images to get to the pure movement-image, before it had a center.  This is what Beckett attempts with Film, [see the discussion here]  trying to show how humans are perceived and how they think this demonstrates their essence, their persisting notion of self.  [The film is then described...  We see the character first acting, moving along walls and up staircases, but this action image is constrained by certain conventions of the camera, only showing the character within particular angles {classic subjective pov shots?}.  The camera takes different positions to offer both subjective and objective perceptions, perception images that combine the double system described above.  The character eliminates all subjective perceptions by taking away the animals and mirrors.  Here, the restriction of camera angles supports objective perceptions.  In the last sections, when the character himself sits before the camera, the total elimination of subjective perception threatens, amplified by the removal of restriction on how the camera can move and show the character.  However, each time it shoots the character from a new angle, he can wake up and force the camera back to the pov, until his exhaustion permits the camera to show him in close up, and the camera itself becomes identified with the subject, even though the subject displays anguish.  This is 'the domain of the perception of affection'(67), showing that when all else is destroyed, only affection remains, the only thing preventing subjective death.  Yet even subjective death is not the end point, because Beckett wants to show 'the world before man...  where movement was...  under the regime of universal variation'(68).  He tries to end with the 'luminous plane of immanence', showing only matter and movement images, tracing the human avatars back to the movement-image.  Deleuze says this is a common ambition in experimental cinema, often using complex technology, but Beckett simply uses conventions to eliminate human centered images.]

However, mostly analysis goes for movement-image to the varieties or avatars. Although identifying three varieties, other kinds of images are possible, because the plane of movement images is only one section of 'a Whole which changes, that is of a duration or of a "universal becoming"'.  Movement-images offer a temporal perspective which also opens to some deeper 'real Time' beyond the specific plane of movement-images and beyond specific movements.  This means there must be time images to depict real time specifically.  There can be indirect images of time produced by relating movement-images together or combining their three avatars, and we see this in montage, but this is still not a 'time-image for itself' (68-9).  There can be a relationship with this whole or duration, but Bergson himself suggests direct time images, in his case 'the "memory image"' (69).

Peirce can be used to attempt to classify these different images, although we will have to work on the relation between this concept of the image and his notion of signs.  Deleuze suggests that we can see signs as particular images, representing a type of image, sometimes related to the composition of the image, sometimes to its formation or extinction.  Until this is clarified, Deleuze proposes to still use Peirce.

Back to illustrations, The Man I Killed offers 'an exemplary perception image' when a cripple perceives a parade through the gap in the crowd.  Lang offers an example of the action image in Dr. Mabuse the Gambler, where an organized action corresponds with synchronized watches, and this produced lots of classic action images in film noir, with the depiction of 'detailed segmentarised action'.  Westerns present action images and also perception images, where 'the hero only acts because he is the first to see, and only triumphs because he imposes on action the interval or the second's delay which enables him to see everything' (70), with the example of Winchester '73.  Dreyer on Joan of Arc offers the classic affection image.

The three images are combined in montage, which assembles movement-images with their corresponding avatars.  However, usually films have one dominant type of image which affects montage [action films or affection films -- Dreyer offers almost exclusively affect, Vertov perception, Griffith action].  Each has a characteristic shot as well, long shots with perception, medium shots for action, closeups for affection.  However, for directors like Eisenstein, each of these movements is supposed to offer a pov on the whole of the film, a way of grasping it, so that it is really the whole that becomes affective in closeups etc.

 Notes on the Bergsonian bits of Deleuze, G (1989) Cinema 2 -- the time-image, London The Athlone Press.
(I know the date of the second volume is before the date of the first one -- shows the effects of different publications in English)

Chapter three.  From Recollection to Dreams: third commentary on Bergson

Bergson says that recognition can be automatic/habitual, a form of sensori-motor recognition that takes place through movements reacting to  and extending perceptions so that we pass from one object to another, but always on the same plane.  Attentive recognition is not just the extension of perception through movement, however, but a 'return to the object' (44) through attempts to analyze and add characteristics to it: the object goes through 'different planes'.  If the first is a sensori-motor image, the second image is 'a pure optical (and sound) image of the thing'.  It looks like the first one is richer, while the second one involves description which replaces the concrete object [and apparently Robbe Grillet uses these Bergsonian terms to describe the cinematographic image].  However, the first description is also organic [which includes being connected to our own life and its adaptation] while the other is 'physical - geometrical, inorganic'.  We start to see these in shots like Rossellini's depictions of landscape as almost abstract,  while Godard [Les Carabiniers] offers just a series of descriptions in each shot, 'pure descriptions which are unmade at the same time as they are outlined'.  So the main characteristic of the new cinema at first is that it operates with a different theory of description.

It is possible to see the sensori-motor image as the impoverished one, retaining only what interests us, including provoking reactions from a character.  The sensori-motor plane provides interests in general.  The pure optical image on the other hand points to 'an essential singularity' which also endlessly refers to other descriptions at the same time.  The first set of images look more useful, while the use of the optical image remains unclear.  Bergson suggest its importance lies in its ability to call up a '"recollection image"' (46), but is also possible to see suggestions of the relations between 'the real and the imaginary, the physical and the mental, the objective and the subjective, description and narration, the actual and virtual', the relations between terms that are different in nature, although they refer to each other.  It no longer matters which one comes first.  These images invoke 'a plane or circuit' suggesting infinite characteristics of the thing which has its own layers or aspects.  In this way, it is possible for characters in films to read images quite differently [as reported in dialogue], almost as hallucinations.

In practice we are talking about a circuit of 'obliteration and creation'[descriptions supplant then add meanings],with different planes intersecting and contradicting each other to show different layers but via 'forking' [not smooth succession?]. But this depicts reality as layered as well as memory [referring to the quote in Matter and Memory about layers in the cone also representing strata of reality] [the example is Rossellini and Stromboli with the different perspectives offered by the character climbing higher and higher on the island].  Sensori-motor images and their extensions are replaced by 'much more complex circular links between pure optical and sound images on the one hand, and on the other hand images from time and thought' (47).  All coexist 'by right', representing, in this case, 'the soul and body of the island'. 

So pure optical and sound signs link with a virtual image not with movement.  So what is a virtual image?  Bergson talks of recollection images, although these are already implicated in automatic recognition.  However, their proper role is to link with attentive recognition, which positively requires them.  A new model of subjectivity is involved, not just a gap between received and executed movement.  The recollection image fully fills the gap [affection only describes the process that goes on internally in the gap].  The recollection image 'leads us back individually to perception', involving a subjectivity that is 'temporal and spiritual' actually adding something to matter. 

One obvious way to see this is in the flashback, which goes from present to past and then loops back to the present.  Some filmmakers [Carné in Daybreak] actually offer us 'a multiplicity of circuits', operating in a whole 'zone of recollections and returns' (48).  However, a flashback is still a 'conventional extrinsic device', offering a psychological causality, which is 'still analogous to a sensori-motor determinism' and does not threaten linear narration.  Flashbacks arise because it is necessary to go back to the past, to either 'mark or authenticate the recollection image', to explain destiny, for example.  However, we might still be hinting at 'a pure power of time which overflows all memory'[more description of Carné and how he uses expressionist figures].

Mankiewicz [Joseph that is] offers the best examples of flashbacks, but does this not to explain or suggest linearity, but to offer 'an inexplicable secret, a fragmentation of all linearity, perpetual forks like so many breaks in causality' (49).  It is time that forks, and this can be unacknowledged by the characters.  The flashback involves several people indicating the multiplicity of circuits, and even the circuits fork within themselves [sometimes the characters can pursue these forks to find out what is happening].  Nor is there some easy alignment, some common destiny, but only new meanders or breaks in causality, 'a collection of nonlinear relations'[examples from the films follow—as I say in the other notes the only one I recall is All About Eve, which shows nonlinear progress as the heroine schemes to take the place of the actress, but as a deviation which makes a circuit and offers a secret].  Sometimes we see that flashbacks do not even explain the present, and sometimes the secrets remain forever.  Sometimes the forks seem to constantly complicate so that everything has to start again and become even more enigmatic.  Overall, it is the forks that have to be explained by the flashbacks, and the recollection images are used to point to the forks.  Originally, they are often pretty well imperceptible and need to be recalled by an attentive memory [there are links here with the structure of the novella in ATP.  Mankiewicz is seen as combining the novella with the more theatrical element].  The memory is still 'story behaviour', represented by voice and the voice-off.  Sometimes this is further in bodied as a phantom or ghost, or some absent character barely glimpsed, but theatricality is added because we see the characters and listen to their dialogue [implying that they act out the story, sometimes in a way which is emphasized by camera movements or sound qualities which vary in volume]. The fork, also understood as 'a skidding, a detour', is usually revealed after the event through flashback, but one character always foresees it [examples follow, including the way in which an assistant overhears Eve, off-frame, producing a false story].  Here, we are seeing 'the birth of memory' which will reappear in the future.  [This links with Bergson on time, showing that the past had already been constituted in the present, as an aim to come]. Mankiewicz uses the cinema particularly to show us this happening, the present recurring in the future as past.  In particular, he uses the notion of out-of-field [the eavesdropping or overlooking]. 

Sometimes this is depicted directly, without flashback [still with Mankiewicz] as in films like Julius Caesar or Cleopatra.  The characters are remembering history, sometimes with visual clues, like the frescoes in Cleopatra, but there are still forks.  The characters are depicted differently in psychological terms, the linear Brutus, for example, although even here he lets Mark Anthony speak in his absence and is easily outmaneuvered.  Mark Anthony is 'a supremely forked being', with his many faces, complete with plebeian origins, acting as a simple soldier and so on.  Cleopatra is also  'the eternal forking woman, devious, capricious'(53), and  Anthony hides behind a pillar to witness one of these duplicitous scenes with Octavian.

The flashback is still limited by requiring a justification, although in Manciewicz there is forking time at least.  There is also a similar problem with the lack of autonomy of the recollection image: it exists because it needs to find a pure recollection, discover pure virtuality..  Recollection images only really emerge as actualizations of pure recollection [when I read Matter and Memory, I wasn't sure if this was just a way for the philosopher to understand them—Deleuze seems to think pure memories really do actualize on their own, which is also what Bergson actually says, to be fair].  So there is another element to subjectivity.  The recollection memory can only represent 'the former present that the past "was"'(54) in the process of it being actualized.  Again this is being hinted at with the 'deeper time images' seen in later discussion.  Attentive recognition has to go through recollection images, but it assists sensation and movement.  It informs us, but often more when it fails them when it succeeds, when we cannot remember, and have to suspend action, or where there's a disconnect between recollection image and the actual optical image.  Recollection images point to the virtual directly in those cases, appearing as 'feelings of déjà vu or past "in general"'(55) [as these examples suggest, Bergson really sees such disconnection or suspension as psychological pathology, and discusses it as such, while Deleuze wants to turn it into philosophical insight]. 

There are dream images, or fantasies.  Avoiding links with action [the most common way to do so says Bergson], these provide 'the recollection image or attentive recognition [with] the proper equivalent of the optical-sound image'.  We soon find mental phenomena like amnesia, hallucination, dream depicted in film, especially in Europe.  This was partly due to a reaction against American films with action images, and partly with stressing subjectivity against American objectivity [note that it is Freud and surrealism that is an influence as much as Bergson, only briefly acknowledged by Deleuze].  The motor extension is lost, sensations and perceptions no longer link with recognition in a predictable way.  A whole set of memories, images of the past in general appear as something free floating.  Cinema uses dissolves and super impositions.  Expressionism tries to restore a panoramic vision, even where people are being assaulted or are near drowning.  We also see it in Fellini and 8 1/2.  Bergson still sees a connection with sensation, but no longer a tight one, with dreams as the most extended of the circuits' connecting with action.  Any sensory image whatever can be connected with the dream image.  In normal life, pure recollections become recollection images, 'summoned by the perception image'(56) but in dreams, a virtual image actualizes in the third [dream] image, which can itself be actualized in a whole range of the other images, this is why dreams are not metaphors but 'anamorphoses' in a circuit [so a dig at Lacan here—but there is no real discussion of how these transformations or condensations occur, and metaphor must be invovled too, surely -- the thin cloud and the razor?].  Thus may a green surface broken by white patches become a meadow or a billiard table [Bergson's actual example], linked by 'a becoming which can by right continue to infinity' (57) ]Entr'acte {watch it here} and Un Chien Andalou {here} are both given as examples]. 

American cinema grasp the implications in 'Buster Keaton's burlesque', through surrealism and dadaism [examples from Keaton's dream sequence in Sherlock Junior, {here}  where the hero ,who is a cinema projectionist, dreams and this shown as a film in the cinema where he is working: balancing chairs become somersaults and then edges of precipices and then steps among lions, all just leaping from one scene to the other matched on action].  Dream sequences are not the only way to represent dreams [in Spellbound, it is not just the Dali sequence but things like the impressions of the fork on a sheet turning into stripes on pajamas].  Each image in the circuit acts as the virtual which becomes actual in the next, and sometimes they all return together [in the final shot of rosebud the sledge].  Dreams in films are depicted in various ways.  One overloads the visuals with dissolves de-framings, special effects and so on, while the other heads for abstractions offering 'a perpetual unhinging' between concrete objects.  These are actually both like linguistic operations.  [Examples of the first are Murnau {Nosferatu} and The Last Man {aka The Last Laugh}, and Keaton as above for the second, or between Entr'acte and Un Chien...  The latter apparently retains the circular shape in common between the different objects. 

This circuit of dream images often ends with a depiction of real objects, or shows us an actual dreamer, or depicts the dream in a separate distinctive visual way [so it is not a very acute critique of the real].  A certain Devillers introduced ambiguity in terms of the '"implied dream"' (57), with no link between optical or sound image and motor extension, and no attempt to compensate by becoming a dream image.  Instead, there is another connection with movement, but this time the world itself moves, in a depersonalized form.  [Some gripping examples, with the slippery road itself sliding, all the world running away from a frightened child as if on a conveyor belt, the camera causing the path on which characters and to move.  We see these features also in dreams, but their effect is then mitigated by acknowledging that they are dreams].  The example is Epstein and The Fall of the House of Usher, where landscapes or houses extend, Hathaway [Peter Ibbetson] or Laughton's Night of the Hunter, where nature apparently intervenes to save the children [the boat is motionless, but the landscape slips by].  Malle also uses the technique [how about David Lynch when the fairy descends to save Sailor in Wild at Heart, or the miniature characters come to life in Mulholland Drive?].  Different dream worlds can be entered, sometimes indicated by peculiar animals, or other inversions.  This is the whole 'cinema of enchantment' (60).  We can also see this at work in the depiction of fairgrounds in Fellini which apparently transport viewers to alternative space times.

Even musical comedy 'outlines a dreamlike world' as bodies and movements are depersonalized [Berkeley and the girls' legs as a kaleidoscope]. There are still identifiable individuals, but they can still display 'a supra-personal element... a movement of world that the dance will outline' (61) [examples where walks turn into dances in Astaire or Kelly— critics do a lot of work here, and there's also an attempt to link both dancers into what Kleist says about grace [see Bogue's commentary and my additions]—Astaire dances quite unconsciously, while Kelly is simply able to impose his consciousness through athleticism].  The idea is to get us to dream, and we enter the dream state as soon as the dancing starts,  a dream-like metamorphosis, an implied dream again.

A definite ambiguity remains, as normal sensory motor movements turn into dance, and then return.  We can speculate about whether the normal action was itself just a pure optical-sound situation, pure description, not real action.  In Donen's films [which include Singing in the Rain], the ambiguity is deliberate, with real action already represented by postcards or snapshots, using the same colors as on the film set, so that dance adds proper depth to the world [also think implied here is that a movement of dance is clearly not normally connected to the sensori-motor adaptations of normal life, but is itself an excess].  Minnelli was able to depict plural worlds through dance, finally questioning dominant reality: each set becomes a world, offering pure description, although we have 'to let ourselves become absorbed'(63) [the examples here are Yolande {aka Yolanda and the Thief}  and The Pirate].  Other devices are also used, sometimes which depict reality as 'the heart of a nightmare', or trapped in the dream of the other.  Minnelli show best how musical comedy can offer a point of indiscernibility between the real and the imaginary.  Sometimes in these pieces, another dreamer is featured!

Burlesque offers similar possibilities [examples here include Jerry Lewis]. Modern burlesque celebrates sensori-motor situations and links them in various kinds of series [including the Laurel and Hardy 'dismantled series' (64).  But there's also a strong 'emotive affective' element [specially in silent film] which can appear in both large and small forms of action.  The characters are strange, lunatic, or dumb [and lunatic] as in Harpo.  There are still links with the sensory motor schema at first, sometimes connected by absurd yet logical relations, Groucho or Fields.  This will lead to another stage, Deleuze predicts, which breaks with th the sensori-motor altogether and becomes pure description.  Apparently we can see this in Jerry Lewis, which are sometimes openly theatrical or which resemble musical comedy [with the dance in The Patsy].  The sets are unrealistically intense.  The character sets off movements of world which often become catastrophic.  Lewis is a loser, but he makes the world move, often oft are extraordinary bodily movements of his own, and can triumph.  We can also see a new notion of the machine, the remote controlled electronic machine, which goes out of control and 'ravages the set' (66).  This evokes an 'energy band', not centered on the character alone, but appearing as a movement of the world. 

We have to go beyond Bergson on the comic here, because this is not just something added to something living, but something which carries away the living.  Pure optical and sound situations are connected to this waiver of energy or movement of world, in a special kind of fantasy or implied dream [lots of Jerry Lewis examples].  Tati offered something similar (66).  Sound also ceases to be integrated into sensory motor schemes visual.  Again Tati rides a wave, although these are not traced to personal fantasizing, instead, he offers 'sound and visual configurations capable of making up a new op'art, a new son'art'(67), and others and named in developing situations as 'pure set valid for itself', or optical and sound situations turning into song rather than conventional action [like the extraordinary Japanese film Zaitoichi (2003) about the one armed samurai which ends in song and dance].  [Demy {The Umbrellas of Cherbourg} is given as the best example, where 'set...replaces situation, and the to-and-fro...replaces action'.

Chapter five.  Peaks of present and sheets of past: fourth commentary on Bergson

The crystal [discussed in chapter four, see notes] reveals time without having any movement, showing how time itself splits into present and past.  This raises the possibility of two time images in fact.  We can see the first in the inverse cone model, where the pure past is not managed by recollections
tied to the present but exist as a virtual element.  These can exist autonomously, and we have to access them with a seed.  This shows that 'memory is not in us; it is we who move in a Being-memory, a world memory' (98), it seems something already there or preexisting, towering over the point of the cone which is the present.  Fellini says something similar. Within the cone are many 'regions, strata, and sheets' (99), some extended, some contracted, each with different tones, themes and shining points.  This is quite different from linear clock time.  They're only layered in terms of the perspective offered by the actual present.  When we search for recollections, we place ourselves in these regions through a 'jump', and if we do not find a recollection image there, we have to return to the present and jump again.  It is the same with sense and language, ideas and thought, which are similarly organized in circles or sheets and into which we jump. 

We can get a sense of the present by thinking about something which is replaced by something else, something in the past or future, implying a form of succession.  However, we can also grasp a single event as something 'purely optical, vertical, or, rather, one in depth' (100) [this is the crystal, surely?].  If we pursue this, though we can detach events from the space in which they are located, and from the present.  This leaves the time of the event as including past, present and future, 'a time is revealed inside the event', and the elements of the past, present, and future of the event can be seen as 'deactualized peaks of present [sic]'.  This appears paradoxical, but we can see the events, or life itself as containing a different number of presents—of the past and of the future as well.  The paradox can lead to disruptions in interpretation or in social relationships.  The appropriate chronosign is the accent or peak of view ['pointes de vue'], as opposed to the signs of aspects, regions or layers as in the first example.

Robbe-Grillet shows the peaks best, including in Last Year...—'the three implicated presents are constantly revived, contradicted, obliterated, substituted, recreated, fork and return' (101).  This as an effect on narration which now becomes abstracted from action, distributing 'different presents to different characters' which are plausible enough until they are combined.  In the end, we realize that the man 'lives in a present of past', while the woman 'lives in a present of future', and neither of them live in the present of present where the husband is.  [Other examples are given 102 F, and there's even a short meditation on relativity and chance {via a critic}].  We see depictions of unconventional time not regulated by movement in Bunuel, through things like repetition [The Exterminating Angel] or cyclical time, or alternative endings, including the substitution of different actors for the same character in That Obscure Object of Desire.

Back to Last Year...  The two main contributors Resnais and Robbe-Grillet, had quite different understandings of the film and how it should be constructed.  Both were aware of the crisis of the action image, in the wandering of characters.  R maintained that something real always did exist, even in Marienbad, while R- G saw everything taking place in the head of the characters or even the viewer.  However, for Deleuze, the image remains whatever the viewer thinks, and that something can be distinct and yet indiscernible [I think because the whole distinction between real and imaginary is reversible].  The two might also disagree about whether the film is set in a perpetual present stripped of time, or whether it shows an architecture of time, points of present or sheets of time.  Again the issue is that the time-image can do both.  The film also raises issues of what is true and what is false, especially if the film is to be read as offering not only an' indiscernible becoming of distinct image' or 'undecidable alternatives' (105).  We can see time as depicted as simultaneously both sheets of the virtual past, and deactualized peaks of present.  In either case, the film shows that cinema is not always 'by nature' rooted in the present.

The direct time image, this time of sheets of the past, is found in Citizen Kane.  Witnesses are found to offer recollection images of the man in a series of subjective flashbacks, which also offer 'a slice of Kane's life, a circle or sheet of virtual past'. The whole point is to try to discover what 'Rosebud' means.  No sheet is prioritized, each has shining points, so the overall picture of Kane is nebulous.  The recollection image can never offer unity, however.  The first kind of recollection image shows a  'motor series of former presents' or events (106), such as those developing the desire to make Susan a singer.  The culmination of this sequence is the shot showing depth of field, with Kane in the background and Susan about to commit suicide in the foreground: here images are arranged in depth rather than shown as a series of presents, to represent regions of the past. 

This restored depth to one of its earlier functions in cinema, before montage or camera mobility.  However, it was not depth of field as such, but rather parallel planes which harmonize.  The breakthrough comes, in painting first, when the planes refer to each other, say when characters from different planes address each other.  We can see the same history in the cinema, with Welles inventing a depth of field, and having the characters in the background make direct contact with those in the foreground.  In other examples, characters communicate by watching each other along a diagonal.  In Welles, there is visual distortion too, and the combination of light and shadow can rightly be termed baroque or neo expressionist.  The deep shot shows temporality, or 'continuity of duration' (108), a depth of time not space, indicating regions of time, linked to other regions by optical aspects or borrowed elements.

Many functions are offered by this depth of field, all of them producing 'a direct time image' (109).  Is not just depth, however which offers direct time images, but this one corresponds to definitions of memory and shows us processes of remembering.  This is the usual location for depths shots which appear necessary.  Flashbacks can show 'psychological memory' (109) as a series of recollection images, and there are a number of occasions when the present passes according to chronological time [clocks or calendars].  But the depths shot can show how someone evokes the past in an actual present as a preliminary to gathering recollection images, or where a sheet of the past is explored before being consolidated into recollection images, both the contraction and the expansion in Bergson's terms.  So pure and psychological memory are connected again as in Bergson.

We see contraction in Kane, for example the high angle shot picking out the alcoholic Susan which then can 'force her to evoke'(110).  Other examples are given in other Welles films, 110f.  We see expansion as the witnesses go back to their own sheet of the past, searching for the meaning of 'rosebud'.  In all these examples, high and low angle shots offer visual contractions, while oblique and lateral tracking shots show sheets.  Depths of field in each type of shot shows evocation and exploration taking place.  Depth serves as 'a function of remembering...a figure of temporalization'.  It can show adventures of the memory, including pathologies of disconnection where evocations for example can no longer be applied to perceptions. There are also montage techniques in their own rights, relating the sheets of past with the shorter shots of passing presents [that initiate them]; relating sheets together; relating the sheets to the contracted present. 

Each witness evokes but Kane is actually dead, his death is the fixed point [not movement is the point, I think, so movements cannot be used to allude to time].  The same goes with the end of the film when rosebud is revealed, but there is no one there to record it as it burns, 'it is totally pointless and of interest to no one'(111).  This raises questions about the relevance of the sheets of past, but again maybe this is the point, since once a person is dead, there is no point.  A similar structure is apparent in Ambersons, where sheets of the past are evoked, but no longer fit the transformed present where the key figures are dead.  In The Lady, sheets of the past can no longer be evoked or even distinguished, so recollections seem like hallucinations [more discussion 113 F].  In Arkadin, the character erases his own past, having feigned amnesia, in order to construct a 'grandiose, paranoid unity'.  All the films can be seen as offering a progression towards the impossibility of evocation and recollection.  In The Trial, the hero cannot recall anything about the offense and experiences and number of puzzling regions of the past which offer more hallucinations rather than precise recollections so that 'nothing is decidable anymore'.  Behind all the episodes, is not some notion of superior Justice, but rather a machine [to quote D's and G's book on Kafka].

So for Welles, the sheets of past exist but may not be usable. Death is also 'a permanent present, the most contracted region', or the sheets themselves have become twisted or scattered.  We see time as a perpetual crisis or even 'at a deeper level, time as primary matter, immense and terrifying, like universal becoming' (115): either the strata are unending, or there is some substance without stratification at the bottom.  This is not transcendental but immanent, some immanent justice or earth.  We contact it directly, it is 'autochthonous' [no origin or history, found in its original place].  This is best seen in Welles's Macbeth with its stone age setting.

At least Welles offers some fixed point, some present offered to our view, which is the starting out for the exploration of the past.  But Resnais films have no center, and 'the present begins to float' (116) [apparently, in Je T'Aime Je T'Aime, the time machine returns to a different present and past each time].  As a results, there is no way to decide between sheets of the past, seem to each has their own present [the example is Hiroshima..., where even apparently agreed characters in their pasts turn out to be quite different, in a movement of 'generalized relativity'].  There are 'undecidable alternatives' between sheets of the past.  [Then Deleuze attempts to show continuities between the films as opposed to the normal way of dealing with them critically which is to see a number of 'oppositions': JeT'Aime over the memory of one character, Last Year as complications because there are two, Hiroshima shows completely different memories between the two characters with nothing in common ...].

The sheets of the past can belong to a single memory, a whole memory world, or some reconstructed notion of the ages of the world itself. Tracking shots are used to construct circuits and continuums are particularly pliable [apparently offering an example of the 'Boulanger transformation'. Known in English as the baker's transformation, this involves mixing elements in such a way, liking cutting and kneading dough,  that apparently unconnected points become adjacent to each other.  The point seems to me that characters in each of the sheets are separated but can also be related in some topology, demonstrating 'each of the characters', or the 'various periods of Van Gough'.  In Last Year... there is one sheet where the characters are actually quite close to each other, although they also have a separate existence on other sheets.  The issue is whether each of these sheets can be seen as a regular transformation of the other or of some underlying single one, and this is how you would unify Resnais films for Deleuze.  More discussion 120 F, including Resnais' rejection of the flashback in favor of transformations, sometimes shown with fragmented objects which are redistributed or when ages are superimposed.]

What these films show is Bergsonian pure recollection as virtual, and the recollection image as a way of making it the actual in relation to a present.  Bergson insists that the two are separate, and that pure recollection is found in a sheet preserved in time, or an age.  When we occupy it in memory, we can either discover specific points which will be actualized as a recollection image, but we're also aware that the past itself is not always accessible, that it belongs to a different age.  However, as Resnais shows, there is another possibility of a sheet of transformation, where fragments from different ages are brought together.  This can happen in dreams where one recollection image turns into a series of others referring to different points.  Spectating ,reading or writing can involve something analogous, where we construct a sheet of communication to weave a network of detached relations between them [transversality?].  'In this way we extract nonchronological time'[a mixture of timescales, as in the event?].  Sometimes this will fail and end in incoherence or bland generalities.  Sometimes this can be shown in discussions of false recollections. 

However, art can construct such streets which are both a past and something to come. This is one way to read Last Year, where both a man and the woman are seen as characters occupying two sheets from which the third character has constructed a transverse link.  We see hints of this in other Resnais pieces too, especially Providence.  Here 'the work of art crosses coexistent ages' (124) producing something like a new montage [perhaps in a Van Gogh painting]. Resnais attempts to stop pure memory collapsing into mere recollection, by preserving all the other mental operations as well, including a 'forgetting, false recollection, imagination, planning, judgement'.  It is a structure of feeling that underpins all these functions, not the characters themselves.  Feelings unite past and present.  Music is important in conveying them.  Feeling circulates from one sheet to another and they can be set free on a sheet of their own.  In that event, they can make us aware of consciousness or thought as well, 'a becoming conscious' (125), as a kind of hypnosis, bringing new awareness of things that appear to be mere shadows.  Thought is the transversal operating in non chronological time across the ages picked out by the feedings.  Apparently, this happens in the brain, in cerebral mechanisms, which offer a continuity between feelings.  The operation can be compared precisely to the unrolling of the film. Resnais intended that such thoughts all to operate around and behind the image.

This happens in the time image, where the world has become memory, but the brain itself recreates matter through transversals.  The brain itself can be seen as a screen where confrontations take place between past and future, inside and outside, in dependent of any fixed point.  'The image no longer has space and movement as its primary characteristics but topology and time' (125).