Film-Thinking and Narrative Indeterminacy
The work of Christian Metz introduced the idea that film narrative can be analysed in terms of a diegetic telling as well as having a mimetic basis, with analogical images of reality arranged to tell a story. Metz identifies a grand imagier who is responsible for the selection and ordering of the film images, which he locates as “situated somewhere behind the film, and representing the basis that makes the film possible” (21). We have here an early conception of narrational filmic agency, positing an external agent responsible for the film. This is a function that is explored with increasing sophistication throughout subsequent film theory.
One of the most comprehensive arguments for the necessity of a filmic narrator is provided by Seymour Chatman, who elaborates the narratological ideas of Metz. He describes this figure as:
the transmitting agency, immanent to the film, which presents the images we see and the sounds we hear. It is not the filmmaker or production team but bears the same relation to those real people as does the narrator to the real author of a novel. Neither is it a voiceover that introduces the action, though that voiceover may be one of its devices. (211n)
For Chatman, a narrator is responsible for the communication of narrative in film. A parallel between Chatman’s narrator and Metz’s grand imagier can be identified here, in that both conceptions posit a narrating agent external to the diegesis who is responsible for the presentation of the film. Although Chatman claims that his narrator is immanent to the film, he does not, however, sufficiently follow through the consequences of positing an immanent agency, and how this can be reconciled with the showing of film, instead retaining the literary concept of a narrator who does this showing. As will be discussed below with reference to Daniel Frampton, the former does not necessarily entail the latter, and in fact this showing activity of an agent is incompatible with immanency (involving as it does an extradiegetic showing by an immanent agent), leading Frampton (and others) to reject filmic narrators in favour of a more appropriate model of filmic agency.
Accounts of filmic narrators are derived from the work of Gerard Genette, who not only distinguished between discourse (expression) and story (content) within narrative, but also introduced a third aspect of narrative, emphasising the importance of narrative agency.  He describes the narrator as “[s]omeone recounting something: the act of narrating in itself” (26). Equivalent agents have been noted in film, with the filmic narrator regarded as narrating the diegesis of the film, functioning, to use Genette’s terminology, as an “extradiegetic” (228). However, this introduction of a narrator—traditionally a literary role—into a theory of point of view is problematic, for “the narrator does not ‘see’ the events of the fictional world, but recounts them; he or she does not observe from a post within the fictional world, but recalls events from a position outside the fictional universe” (Stam et al 86). It is this discrepancy, between the visual point of view that film seems to show and the traditional activity of a narrator as an extradiegetic figure, telling a story from a position external to it, that occupies the narratological study of filmic point of view, leading some theorists to dispense with talk of narrators altogether, offering reconceptualisations of point of view and narrative agency in the process, as we shall see.
The central problem of narration in film is therefore to theorise a model of agency—explaining how film seems to be organised to “tell” a story—compatible with the showing of film (which seems to grant more direct access to the events that constitute the narrative). Some theorists advocate the limitation or augmentation of film-narrator models, rather than their abandonment; others, however, propose a completely different model of filmic agency and mediation. 
Non-linguistic theories of narrative agency explore alternative conceptions of the manner in which film presents itself as narrative and the agency involved in this. Genette himself argued against conceiving film in linguistic terms, going as far as to claim that “film could not, properly speaking, be a narration, because it is not a linguistic being” (paraphrased by Metz 145). David Bordwell pursues an alternative conception of film narration that does not posit an external, anthropomorphic figure responsible for the narration. Bordwell instead attributes this function to the process of narration itself, claiming that “[t]o give every film a narrator or implied author is to indulge in an anthropomorphic fiction” (62). Bordwell does recognise that occasionally the narration will signal a narrator, such that the viewer feels the influence and presence of this figure as “an overriding consciousness” (225). Even in these cases “this sort of narrator does not create the narration; the narration, appealing to historical norms of viewing, creates the narrator” (62). However, this commitment to the placing of narration before narrator leads Bordwell to grant anthropomorphic functions to narration, in which case why the distinction? As Gunning remarks, “[w]hat he has excluded in theory re-emerges in his practice” (480). A similar point is made by André Gaudreault and François Jost, who criticise Bordwell’s attribution of human characteristics to an abstract process of narration (62). Chatman also objects to this theory of agency without an agent, pointing out the semantic problems involved in attributing agency to a process, stating that “[o]bjects and processes may have qualities, but only agents can do things” (127).
However, Katherine Thomson-Jones counters this criticism, objecting to Chatman’s view that narration necessarily involves a narrating agent. She argues that it does not make sense to conceive of a filmic narrator as fictional, a supposition that leads to Chatman’s distinction between a (fictional) narrator and a (real) implied author, as this leads to the idea that this showing itself is also a fiction and thus not accepted as real by a viewer. Thomson-Jones instead follows George Wilson’s argument against fictional showing to conclude that “the actual showing of a fictional story does not automatically invoke a fictional showing” and so “the narration of a fictional story does not automatically entail a fictional narration of the story” (91). The narration can therefore be attributed to a real author, occasionally involving the mediation of a fictional narrator.
Unlike Thomson-Jones (although acknowledged within her article), Wilson explores the other side of the argument against fictional showing: a fictional agent (the narrator) presenting the fictional as actual (i.e. the fictional showing of an actual story, as opposed to the actual showing of a fictional story). What is significant about this move is that it removes the need, in the act of viewing, to locate an extradiegetic source of the on-screen images; for Wilson, a viewer’s experience of a fictional narrative as a real scenario is an imagining that this presentation does not have a determinate location or source, and therefore is “without dramatic mediation” (“Le Grand Imagier” 194). This idea of immanent narrative agency, integrated with the film world experienced by a viewer, is central to Frampton’s conception of film-thinking, which celebrates the fact that Wilson “returns the intention of film-being to the ‘film’ itself”, pointing the way towards a credible alternative to “anthropomorphic fictions” and discursive figures (38).
Frampton’s model of a “filmind” helps us to overcome some of the problems and questions that have arisen from these theories of cinematic showing and agency, with consequences for narratological issues that are not fully developed in Frampton’s account. Like Bordwell, Frampton is critical of the employment of anthropomorphic figures in theorising film narration, regarding the communication model of narrative—with its assumption of sender and receiver—as unnecessary in theorising filmic narrative agency, in that it postulates an external figure responsible for the narrative. Frampton regards this as a remnant of the linguistic heritage of narratology, proposing an altogether different form of filmic articulation in its place: one that renders narrational agency immanent within the film itself.
It is worth mentioning Edward Branigan’s contribution to these debates here, paralleling Bordwell’s in employing constructivist and cognitivist principles to examine the process of film narration. However, rather than emphasising an active perceiver of the message of film narrative but no sender, as Bordwell does, Branigan sets out a hierarchy of schematic levels that form the narrative text and that viewers use to organise perceptual data into narrative patterns, attributing shots to various focalising agents:
Such concepts as “narrator,” “character” and “implied author” (and perhaps even “camera”) are then merely convenient labels used by the spectator in marking epistemological boundaries, or disparities, within an ensemble of knowledge; or rather, the labels become convenient in responding to narrative. (85 original emphasis)
Although Branigan presents a more nuanced account of narrational agency than Bordwell, while also striving to move away from literary conceptions of narration, he still falls back on the idea that there is a narrating agent distinct from characters and focalisers within the film, and whilst now “only part of the filmgoer’s narrative schema”, it is still “a film agent that is not immanent in the film” (Frampton 34).
In place of such narrational figures, Frampton puts forward the idea of a filmind, which he describes as “the theoretical originator of the images and sounds we experience, and ‘film-thinking’ is its theory of film form, whereby an action of form is seen as the dramatic thinking of the filmind” (6). He emphasises that the filmind is not some intentional agency, external to the film: “[F]ilmosophy wishes to place the origin of film-thinking ‘in’ the film itself. There is no ‘external’ force, no mystical being or invisible other. It is the film that is steering its own (dis)course” (6). There is no separation between the filmind and the world that it thinks:
In filmosophy the film does not have experience of things, it just has film-experience, or not even that, just film-thoughts. We might say that the filmind has a “film-experience” of the objects and characters—it can never be separated from the images and sounds it shows. The filmind thinks an image which includes its attention and “objects” as one. (89)
The film-world is inseparable from the thinking of the filmind, and so the filmind exists only in its thinking of the film-world, and is immanent within this: “[I]t both intends towards and ‘is’ the film-world” (90).
Frampton proposes that not only can the film “think” subjectively, as if from the perspective of one of the characters, via a point of view shot for example, but that film can also think “from itself”. This grants an autonomy to film-being that Frampton believes has not been recognised, or at least fully developed, in the work of previous theorists: “Many authors cannot make the leap to autonomous film[;]… cannot see that film may just be thinking from itself, and, only if it feels like it, through or ‘as’ characters” (85). And more drastically than this might suggest, this film-thinking is always primary: “[T]he filmind alwaysthinks its own thoughts, whether they look like the thoughts of a character or not” (85).
This has consequences for how we are to regard the distinction between objective and subjective perspectives and shots in film, reconfiguring them both as transsubjective, a term that Frampton uses to describe how the filmind can be both objective and subjective. But properly speaking, it is neither: it does not present the film-world from the subjective perspective of one of the characters within it, with its thinking belonging to a character-subject, because “the filmind always thinks it own thoughts, whether they look like the thoughts of a character or not” (85). And nor does it present it “objectively” from outside of this world, for “[t]he filmind is not outside the film, it is the film” (86). The filmind may give the impression of subjectivity or objectivity, or even seem to combine these positions, “such as when we see a drunk character through a drunken swaying defocused haze that would be their point of view” (87). However, ultimately “all ‘subjective’ and ‘objective’ shots within the film are produced through the filmind’s ‘thinking’: the filmind is neither subjective or objective, it is both, it is transsubjective” (87). This leads to the fact that “no point of view is pure, it is always the filmind’s thinking” (88).
This transsubjective film-being provides an alternative to the problematic models of narration discussed above, because as a concept it does not posit an unnecessary determinate narratorial location outside of the diegesis. Instead, it reintegrates and reconceptualises this source of the film-world as an agent that “operates from a uniquely transsubjective non-place”, immanent within this film-world as film-specific phenomenological intentionality (Frampton 88 original emphasis). The result of this conceptualisation is that “all narrational agents are grounded in the film itself, in a singular intention that gives us scene-settings and character thoughts, objective viewpoints and character experience” (99). Frampton’s theory of a filmind has the potential to free film theory from the entanglements involved in theorising the place of subjectivity and agency in narration by offering a new concept of film-being and of film-thought, which also reveals a fundamental narrative indeterminacy within this narrative agency, not fully explored by Frampton. 
Transsubjectivity and Unreliable Narration
Flashback structures in film are often accompanied by a voiceover narration that bridges the temporal shift between the diegetic present of the framing narration and the recollection visualised on screen, and gives the impression that the image is thinking with the voiceover (as in Max Ophüls’s Letter from an Unknown Woman (1948), for example). This lends the image the assertive quality of language, such that—like the voiceover narration that accompanies it—it is regarded as making a statement that the events actually took place. This “narration” provided by the image is thus regarded as reliable in this manner, unless further information is revealed that calls into question the testimony of the voiceover narration and the images that seem to stem from it, such that an effect of unreliable narration is produced.
Chatman argues that the sounds and images outside of a flashback are presented by a “cinematic narrator” whose narration frames the unreliable narration, and that both of these narrators (cinematic and homodiegetic) are framed by the intent of an implied author figure, that he describes “as the principle within the text to which we assign the inventional tasks” (132).  According to Chatman, it is this figure that enables the two narrations to be compared and thus for one to frame the other as unreliable in accordance with the perceived intention of the text, which he identifies with an implied author: “Controlling both narrations [character and cinematic] there must be a broader textual intent—the implied author. It is the implied author who juxtaposes the two narrations of the story and ‘allows’ us to decide which is true” (132). However, Frampton’s concept of an immanent, transsubjective thinking filmind enables us to regard unreliable narration in a different way as, strictly speaking, a homodiegetic voiceover narration does not frame the images when the film is thinking with a character in this manner, and, as I have discussed, the filmind removes the need to posit any kind of cinematic narrator.
What follows from Frampton’s account is an indeterminacy of the ontological status (and therefore diegetic level) of the event in film; this status—as subjective or objective—cannot be determinately asserted by the image-thinking of the film, which presents both objective and subjective events as transsubjective. Unreliable narratives featuring lying flashbacks take advantage of this transsubjective indeterminacy of film-thinking, as I will duly discuss in more detail. However, I argue that only the voiceover, as discourse, can be assigned to any one narrational location and that the images, on the contrary, are always necessarily independent of any such point of origin. As Frampton puts it:
[A] character never completely originates a narrative, the filmind is always giving its version of what the character tells. This double authoring, where personal narratives are enclosed by the larger filmind, allows us some possible scepticism over what the character is relating, allows us the possibility of not believing the character (and removes any need to call-up an implied author). The filmind can think (imagine, create) the “contents” of its character’s minds [sic]. (85)
The issue then becomes one of determining when and who the filmind might be thinking with at any particular time, while retaining the idea that the image-thinking of the filmind is always in fact transsubjective and so is never truly subjective or objective. However, Frampton does not explore the narrative indeterminacy that results from this “pure intentionality” of the moving image of thought created by film (87).
The image-thinking of film is not making an ontological statement or truth claim through its creation of thought qua moving image, and thus the supposed unreliability of the image in films featuring unreliable homodiegetic narrators is a misnomer: it is not making a claim to a reliable assertion in the first place, and so cannot be considered unreliable when the image is revealed as thinking with a fabrication. The images of a lying flashback should therefore not be considered as unreliable but, at best, as misleading, which stems from the creative power of film-thinking, thinking subjectively and objectively with equal veracity. This indeterminate ontological status (a result of the transsubjectivity of film-thought) can be exploited to mislead a viewer. In films where the flashback of a character is revealed as false (as a fabrication) the viewer may have been misled into believing that the film is thinking objectively when in fact it is thinking subjectively, with the homodiegetic narrator. They may feel deceived at having being led to this assumption, believing that the images are making statements of fact—that certain events took place—like the verbal testimony of the homodiegetic narrator. For example, in Stage Fright (Alfred Hitchcock, 1950) the conventional device of a dissolve and overlapping dialogue, such that it moves from voice-on-screen to voiceover, is used to cue a flashback sequence and give the impression of equivalence between words and image, as if the latter were a continuation of the former.
Stage Fright opens with Eve Gill (Jane Wyman) and her friend Jonathan Cooper (Richard Todd) in a car escaping from the police to Eve’s father’s boat. When they are clear of the police Eve asks Jonathan, “Could you tell me now what happened? I’d really like to know”. Jonathan then begins to explain that their plight stemmed from Charlotte Inwood (Marlene Dietrich) being “in a jam”, and that he felt he had to help her. He continues, “I was in my kitchen, it was about 5 o’clock”, this statement accompanied by an image dissolve from the car to Jonathan in a kitchen drying the dishes. He continues, saying that “the doorbell rang and I went downstairs to see who it was”, the sound of a doorbell coinciding with the words being spoken about this event, and the last part of the statement fading in volume as we see Jonathan leave the kitchen and go downstairs. The flashback over the next few minutes shows Charlotte—apparently his lover—arriving at Jonathan’s and informing him that she has killed her husband in self-defence. She persuades Jonathan to go back to her house, where the murder took place, and get a new dress to replace her blood-stained one for a performance she has to give later. Whilst there he decides to make it look like there has been an intruder to explain the murder, but is seen by Charlotte’s maid, who informs the police. Jonathan flees and goes to his friend Eve for help, hence the pursuit that the film opens with. Eve, who is in love with Jonathan, then gets involved in an elaborate ruse to prove the guilt of Charlotte. When eventually cornered by Eve about the murder, Charlotte denies having killed her husband and attempts to explain the blood on her dress and how her husband died: “Yes, some blood did splash on my dress—I was there in the room when Jonathan killed my husband, but I had nothing to do with it”. She explains that after the murder Jonathan wanted her name kept out of it and so sent her away and made it look like the work of an intruder and that he then brought her a clean dress. It eventually transpires that Jonathan has been accused of murder before but successfully claimed self-defence: the police detective—and Eve’s new love interest—Wilfred Smith (Michael Wilding) declares that Jonathan “killed Charlotte’s husband alright, and he’s killed before”. This revelation, made by a character whose honesty and integrity has not been called into question, together with the claims of Charlotte, causes a viewer to doubt the truthfulness of Jonathan’s account and his reported flashback: we now have an alternative narrative of events (provided by Charlotte) and an undermining of the character and honesty of the initial (homodiegetic) narrator. The fact that Jonathan has fabricated his account of events is confirmed when he later confesses to Eve that he did kill Charlotte’s husband.
Hitchcock’s film takes advantage of the transsubjective indeterminacy of film-thinking in order to mislead the viewer by presenting a fabricated sequence of events as if it were a statement of fact (with the images seeming to continue the assertion of the voiceover): Jonathan’s version of events exists only in his mind, expressed through his words to Eve, and yet the film is able to create these thoughts as moving images and spoken dialogue within these. Stage Fright was novel in its breaking of the convention that the portrayal of a flashback could be relied upon as a truthful version of events that took place objectively, that had actually occurred within the diegesis, to the extent that viewers felt deceived when Hitchcock flouted this convention. However, we can see that the flashback itself should be considered as unreliable and thus deceptive only insofar as it is in a certain relationship with the verbal testimony of the deceiving narrator. 
In Stage Fright the viewer is initially led by a homodiegetic narrator to regard the events that he recounts—and the images and dialogue that this voiceover seems to frame, as if making similar assertions—as a truthful recollection/presentation of veridical events. However, in Akira Kurosawa’s Rashômon (1950) there are multiple and conflicting versions of events put forward by several homodiegetic embedded narrators, such that they each lead the viewer to make certain assumptions regarding the ontological status of the image-thinking that seems to be framed by their verbal testimony, and only one of which can be true, or none of them at all. Here transsubjective indeterminacy is foregrounded, as it is not clear which flashback is veridical, nor who frames whom, as I will now demonstrate.
Rashômon features the testimony of six different characters, each giving a different version of the events that led up to the murder of a man. Unlike the above examples, Rashômon features a doubly embedded (or metadiegetic, to use Genette’s term) narrative, although this framing is unclear, as I will demonstrate. The initial frame narrative features three men taking refuge from a storm, with one of these men, a woodcutter (Takashi Shimura), describing how he found a body in the woods: “It was three days ago, I’d gone to the hills to cut wood”. We then have the familiar flashback transition of a cut to a scene where the image corresponds to a certain extent with the dialogue spoken (although without an overlapping voiceover in this case): this next scene features a low-angle shot of a tree canopy, with the woodcutter walking through the forest, first stumbling upon some discarded belongings and then a dead body, at which point he screams and runs away. The voiceover of this man then states that “I ran to tell the police”, before introducing the next scene, which follows as a causal result of this event, as well as informing the viewer of the temporal interval between the various diegetic levels: “That was three days ago. Today I was called to give evidence”.
There then follows a wipe-transition to the trial, where a series of witnesses, including an accused bandit called Tajomaru (Toshirô Mifune), give their respective versions of events within this framed situation and thus, as apparently doubly embedded. The woodcutter describes the belongings that he found in the wood, and then the priest’s testimony begins: “I saw the murdered man when he was still alive. It was three days ago in the afternoon. It was on the road between Sekiyama and Yamashina”. We are then given the impression of a flashback again, as the scene cuts from the trial to the priest walking down a road. His voiceover describes seeing the couple (the murder victim and his wife) as they are shown on horseback. In a cut-back to the trial, the priest describes how he did not think he would see the man again but then saw his dead body.
A little later the bandit confesses: “It was me, Tajomaru, who killed him”. Tajomaru is shown asleep under a tree as the couple approach and pass on horse-backhorseback. Back at the trial he explains that he desired her and “decided to have her. I’d have her even if I had to kill the man”. He is then shown stalking the couple, and after leading her husband away ties him up and attacks the woman. She does not want to be raped and lose honour in front of two men and so promises to “belong to whoever kills the other”. They fight and Tajomaru kills her husband. The scene then switches to the framing narrative, where the priest describes how the woman was found by police and also gave evidence at the trial. The woodcutter claims that both the bandit’s confession and woman’s story are lies. Back at the embedded trial the woman’s testimony then begins. She describes, and the viewer is shown, how after Tajomaru raped her she could not bear the way her husband looked at her and encourages him to kill her. She becomes angry whilst holding the dagger and then back at the trial describes how she fainted and when she came round she found a dagger in his chest, the implication being that she killed him, thus contradicting Tajomaru’s testimony.
Returning to the framing narrative once more, the priest introduces the testimony of the dead husband speaking through a medium during the trial, whose voiceover describes how Tajomaru tried to persuade his wife to go with him after raping her. We then see her agree to this and tell Tajomaru to kill her husband, but Tajomaru turns to the husband and offers to kill her instead. She escapes and he goes after her. Tajomaru then returns and tells the husband she got away, cutting the ropes that bound him. Her husband then gets up and stabs himself. Again, this contradicts both Tajomaru’s and his wife’s version of events.
In the frame narrative the woodcutter refutes this, saying “there was no dagger. He was killed by a sword”, and then reveals that he saw more than the dead body, as he initially claimed, and is therefore accused of lying in his earlier story. He goes on to describe seeing the three of them alive, with Tajomaru “in front of the woman … begging for forgiveness”. The image then cuts to this scene, with Tajomaru having just raped her, again giving the impression of a flashback. Tajomaru tells her to come away with him and asks her to marry him, threatening her with death, but she will not make this decision, instead setting her husband free so that they can fight it out. Tajomaru then kills him and she runs away. Tajomaru retrieves the sword and then also escapes. Once more, we have a version of events that contradicts those that have gone before, the only difference here being that, unlike the previous characters, the apparent narrator of the flashback in this case (the woodcutter) does not confess that he is the culprit, but accuses someone else.
What is interesting about Rashômon is that it provides multiple versions of events leading up to a man’s murder (his death seeming to be the only fact not disputed in these various accounts), but without any one of these seeming more believable than the others, so that it remains unclear which, if any, of the testimonies and the images accompanying them is to be regarded as a faithful rendering of events. There is no authoritative narrator who seems believable above all the others, and therefore it remains indeterminate which of the collections of sounds and images that are introduced as character flashbacks are to be regarded as subjective—as the filmind thinking with (i.e. creating) this character’s fabrication—and which objective—as the filmind creating events that actually happened. Indeed, it is not only an issue of subjective versus objective—of believing one character over another—but also of the attribution of the flashback to a specific character. This is due to the way in which the film sets up a hierarchy of narrative levels, opening with a frame narrative within which a character recalls a series of events, telling a story to another character, and then giving the impression that the subsequent images and sounds are a depiction of this testimony (which of course strictly they are not, as I have described). Within this framed, embedded narrative the same process occurs, so that the impression of an embedded narrator (intradiegetic, to use Genette’s term) and a doubly embedded narrative (story-within-a-story) is created. However, if we are to regard the initial flashback as the filmind thinking with a character—that is, creating his/her testimony through images of movement and sound, such that it does not belong to that character as his/her own recalled experience—then the flashback that seems to be within this is not necessarily framed by this character either (so that it is not a flashback within a flashback) but may be the filmind thinking with this embedded character and not necessarily through the framing character. Furthermore, it may indeed be interpreted as the filmind thinking this series of events with the character through the flashback of another character. In other words, a doubly embedded, metadiegetic narrative is not guaranteed in this instance, and this is due to the transsubjective mobility of film-thinking, which is not framed by the verbal testimony of character narrators, but creates the story as its own transsubjective intention, immanent within the film. It is up to the viewer to judge the relationship between the characters and film-thinking, and to what extent the latter is thinking with the former. In Rashômon this issue is complicated by the fact that the film sets up two homodiegetic narrators within the framing narrative: the woodcutter, who seems to narrate most of the embedded narrative, and the priest, who narrates (and by this I mean that he introduces) the testimony of the dead husband during the trial, channelled through a medium.
We can therefore see that the flashbacks themselves in Rashômon, apart from the verbal testimonies of the characters, are not unreliable but indeterminate, in terms of who they are thinking with and how closely they accord with the truth. And the allocation of these flashbacks affects their veracity, depending on how much we trust the character, ensuring that the viewer brings them into a position to be judged as reliable or not by associating them with a certain character and making judgements as to the reliability of that character. For example, the woodcutter reveals that he lied about finding the dead body, when he admits to the “commoner” character in the apparent framing narrative that he saw everything, but did not tell the police because he did not want to get involved, which therefore reveals that the earlier flashback was in fact a fabrication. The apparent source of this flashback admits its falsity and then proceeds to provide the impression of recalling a new version of events to put in its place. If the testimony of the bandit is regarded as embedded and the corresponding flashback as doubly embedded within the woodcutter’s flashback, then the revelation of his dishonesty may cause the viewer to also doubt the honesty of this apparent flashback to the trial of the bandit and the flashback that takes place within this (although the priest is not shown challenging this apparent version of events, and he claims he was also at the trial). Indeed, the status of the woodcutter’s flashback is subsequently complicated even further, when the commoner accuses the woodcutter of stealing the dagger, which therefore would have motivated him to deny that the dead man was killed by a dagger, as he did earlier, and invent a story in which he was killed by a sword, as he may well have done with his flashback. Are we then to believe the commoner, who at the end steals clothes from an abandoned child? Furthermore, the woodcutter redeems his character somewhat by adopting this baby at the end.
If, however, we are to regard the trial scene that follows the woodcutter’s verbal testimony as the filmind thinking objectively—presenting actually occurring events—i.e. as not framed by the woodcutter, then the flashbacks that take place within this are perhaps more likely to be trusted, at least in the sense of the filmind thinking with these characters (as opposed to thinking with these characters through the woodcutter), for they cannot both be trusted as veridical due to the mutual exclusivity of their content. Perhaps, then, the priest should be trusted, as he was not a witness and therefore offers no direct testimony of the events leading up to the murder, but seems to frame the testimony—and therefore also the flashbacks—of the husband and wife. The contradictions between these two versions of events “within” the priest’s testimony thus means that taking sides with the priest still does not enable one to locate a truthful version of events.
The question in Rashômon,and in films featuring apparent flashbacks in general, is therefore not only who is telling the truth but, more fundamentally, to whom each telling belongs. The transsubjective mobility of film-thought leads to such indeterminacy. The determination of this may, as in Rashômon, involve making a judgement as to the veracity of the account—the supposed flashback—offered by that character, if one character is more trustworthy than another, or when the film is thinking objectively (and not belonging to any character) such that it can then be regarded as reliable or not. The important point is that in order for film-thinking to be regarded as reliable or not it must first be interpreted as making an assertion of truth, which it does not do inherently but only when tied to the testimony of a character, positioning him/her as a narrator, or when regarded as thinking objectively. Although, as I have demonstrated, film-thinking is neither properly subjective nor properly objective, but fundamentally transsubjective; it is always film-thinking first, which may then be interpreted as thinking with a character or not. 
Rashômon is an early example of a film that exploits this transsubjective mobility of the filmic image through the use of indeterminate framing to produce a fundamentally indeterminate narrative, so that many interpretations seem possible, with the overall effect being that truth seems elusive. The viewer seems to be led by various narrator figures to a certain interpretation of events, but each contradicts the other, and upon further reflection the indeterminacy of these events seems only to increase, as I have demonstrated.
Indeterminacy and Film-Thinking
It is the transsubjective status of the filmic image that enables it to be open to interpretation in the manner I have suggested, thanks to the indeterminacy of film-thinking in terms of the ontological status of the event and of the character with whom it thinks at any given moment. This becomes apparent when a viewer is misled about the status of the image—it is the transsubjective indeterminacy of the image that enables them to be misled in this manner, but who is misleading here? Clearly it is not just the words of a homodiegetic narrator, who is not in control of the images that seem to accord with these words, but, in this context, it is an effect of the relationship suggested between the words and the images, with the filmic image framing speech (despite the contrary impression that can be given when the two overlap). As Frampton proposes, it makes sense to conceptualise the agent responsible for this as a filmind rather than as any kind of narrating agent, with the inherent connotations of discourse and transcendent (as opposed to immanent) figures.
As we have seen, Chatman makes the distinction between a cinematic narrator and an implied author (a figure on which Wilson also falls back (Narration in Light 123)) in order to explain the effect of unreliable narration that he locates in “lying flashback” films such as Stage Fright: “If the sole source of the ostensible story is a narrator, and if we come to believe that the ‘facts’ are not as the narrator presents them, there can only be some other and overriding source of the story, the source we call the implied author” (Chatman 131). However, we can now see, since I have argued that it only makes sense to ascribe unreliability to the verbal testimony of a homodiegetic narrator, that this distinction is not required if we are to reconceptualise this unreliable cinematic (as opposed to homodiegetic) narration as indeterminate film-thinking instead.
To return to the example of Stage Fright, Chatman believes that, during the lying flashback, “everything that we see and hear follows Johnny’s scenario. Thus, even when his voice falls silent, he remains the controlling, if unreliable, narrator of the flashback”; and that this is distinct from the cinematic narrator who controls everything else we see and hear (132). Chatman claims that it “is the implied author who juxtaposes the two narrations of the story and ‘allows’ us to decide which is true” (132). I argue instead, influenced by Frampton, that this distinction between narrators is not necessary here, and that the agent responsible for each of these views is better regarded as a transsubjective filmind. However, in an extension of Frampton’s theory to cover narratological concerns with truth, authority and narrational agency, I propose that the nature of transsubjective film-thinking is inherently indeterminate; it does not present facts that are asserted as reliable in the sense that a verbal narrator does, but is indeterminate in terms of the status of the action that it presents and can mislead through the actions of its film-thought. It does not make sense to talk of film-thinking itself as unreliable and therefore an implied author is not necessary to explain the misleading effect of films like Stage Fright. The image-thinking of narrative film is unique in that, in terms of the events presented through the image, it is not a question of disbelieving a narrator, or believing one narrator over another, as if facts were being asserted through the image only to be challenged by other images, but of the viewer determining the ontological status and related diegetic level of the events being shown, which remains indeterminate within film-thinking.
1. Tom Gunning, for example, describes how a filmic narrator organises the different mimetic aspects of narrative discourse in order to tell a story, “creating a hierarchy of narratively important elements within a mass of contingent details …, thus carving a story out of photographed reality” (474). Gunning’s analysis of this narrating agency is systematic in terms of its treatment of the specificity of film, identifying the different levels on which the filmic narrator can exert an influence in order to communicate narrative to the viewer, including the selection of objects to be filmed, the arrangement of these objects within the frame, and the ordering of these framed images through the process of editing: “These three aspects of filmic discourse—the pro-filmic, the enframed image, and editing—almost always work in concert and represent the medication [sic] between story and spectator in film. They are how films ‘tell’ stories. Taken together, they constitute the filmic narrator” (477).
2. For example, in order to account for the fact that film seems to directly show events without the mediation of a narrator, André Gaudreault identifies various levels of narrativity, making a distinction between “monstration” and narration, with monstration referring to the mimetic capacity of film to depict events as if they are happening in the present (citation?). According to Gaudreault, the narrator is manifest in the editing together of these images, which brings about a temporal manipulation of the natural “presentness” of the image, such that the narrator creates a narrative in this way: “Only the narrator (= the editor) can inscribe between two shots (by means of cuts and articulation) the mark of its viewpoint, can introduce a guided reading and thereby transcend the temporal oneness which unavoidably constrains the discourse of narration” (Gaudreault 33). Both of these activities are regulated by a “meganarrator” which, as a grand imagier, is “responsible for the meganarrative—the film itself” (Gaudreault and Jost 58). Gaudreault’s theory allows for the diegetic activity of a narrator and the mimetic expression of the film image, distinguishing between the communicative mode of each (in a way which Gunning does not). However, this account retains the figure of an external narrating agent, derived from linguistic conceptions of narrative hierarchies.
3. There is indeed a legacy of film-as-thought throughout film theory, and, whilst a proper consideration of this is beyond the scope of this article, Frampton does himself engage with this (15–26). Of particular note here is the work of Bruce Kawin and his idea of “mindscreen narration” as a kind of visual first-person narration through images of thought, distinct from what a character says in first-person, through voiceover, and what they see, through subject camera shots (10). This is an earlier attempt at considering the specific visual ontology of film in relation to narratological issues such as point of view and levels of narration. However, as with many of the figures who may be considered the forebears of his filmind, Frampton does not think that Kawin goes far enough in shedding literary models, being too concerned with finding equivalents to first-person narration, however novel the terms employed. Frampton’s notion of transsubjectivity allows him to avoid problems in attaching film-thought to determinate points of origin, writing of Kawin: “This clinging to the first-person does not reflect the powerful free-play of film, and disallows the possibility of ‘false’ character thoughts or memories” (Frampton 21). However, while supplying an alternative model of film-thought and narrational agency, Frampton does not fully explore the narratological consequences of this film-being and its ramifications for conceptions of indeterminacy and unreliability, as I do here. Indeed, Frampton’s neglect of the narratological implications of his own work runs the risk of undermining his dismissal of prior theorisations of narrational agency in film, which is a situation I wish to address.
4. Genette uses this term to describe a narrator who is “present as a character in the story he tells”, with a heterodiegetic narrator “absent from the story he tells” (244–5).
5. Frampton, while recognising the transsubjective nature of film-thought, does not follow through the implications of this for a conception of unreliable narration, i.e. the unsuitability of this label as applied to the image, and how this leads to an inherent narrative indeterminacy of film-thought, as I explore here.
6. There is a legacy of theories of unreliable narration in film theory, including those of Sarah Kozloff (112–17), David Bordwell (60–1), George Wilson (Narration in Light 39–44), Greg Currie (202–7) and Seymour Chatman (131–7). However, these notions of unreliability tend to depend on the figure of an implied author or some other external agent in order for the narration to be regarded as such, or cannot account for this agency coherently enough, as in Bordwell’s case. I argue here that the transsubjectivity of the filmic image means that we should conceive of the image-thinking of film in different terms, as will become clearer in my discussion of Chatman’s position.
1. Bordwell, David. Narration and the Fiction Film. London: Routledge, 1985. Print.
2. Branigan, Edward. Narrative Comprehension and Film. London and New York: Routledge, 1992. Print.
3. Chatman, Seymour. Coming to Terms: The Rhetoric of Narrative in Fiction and Film, Ithaca and London: Cornell University Press, 1990. Print.
4. Currie, Greg. “Unreliability Refigured: Narrative in Literature and Film”. Philosophy of Film and Motion Pictures. Ed. Noel Carroll and Jinhee Choi. Malden/Oxford/Victoria: Blackwell Publishing Ltd., 2005. 200–10. Print.
5. Frampton, Daniel. Filmosophy.London and New York: Wallflower Press, 2006. Print. Gaudreault, André. “Narration and Monstration in the Cinema”. Journal of Film and Video 39 (1987): 29–36. Print.
6. Gaudreault, André and François Jost. “Enunciation and Narration”. A Companion to Film Theory. Ed. Toby Miller and Robert Stam. Malden/Oxford/Victoria: Blackwell Publishing, 2004. 45–63. Print.
7. Genette, Gerard. Narrative Discourse. Oxford: Basil Blackwell, 1980. Print.
8. Gunning, Tom. “Narrative Discourse and the Narrator System”. Film Theory and Criticism: Introductory Readings. Ed. Leo Braudy and Marshall Cohen. New York and Oxford: Oxford University Press, 2004. 470–81. Print.
9. Kawin, Bruce. Mindscreen: Bergman, Godard, and First-Person Film. Princeton, New Jersey: Princeton University Press, 1978. Print.
10. Kozloff, Sarah. Invisible Storytellers: Voice-Over Narration in American Fiction Film. Berkeley/Los Angeles/London: University of California Press, 1988. Print.
11. Letter from an Unknown Woman. Dir. Max Ophüls. British Film Institute, 1948. Film.
12. Memento. Dir. Christopher Nolan. Fox Pathé, 2000. Film.
13. Metz, Christian. Film Language: A Semiotics of the Cinema. New York: Oxford University Press, 1974. Print.
14. Rashômon. Dir. Akira Kurosawa. RKO Radio Pictures, 1950. Film.
15. Stam, Robert, Robert Burgoyne and Sandy Flitterman-Lewis. New Vocabularies in Film Semiotics: Structuralism, Post-Structuralism and Beyond, London and New York: Routledge, 1992. Print.
16. Stage Fright. Dir. Alfred Hitchcock. Warner Bros., 1950. Film.
17. The Usual Suspects. Dir. Brian Singer.MGM, 1995. Film.
18. Thomson-Jones, Katherine. “The Literary Origins of the Cinematic Narrator”. British Journal of Aesthetics 47.1 (2007): 76–94. Print.
19. Wilson, George M. Narration in Light: Studies in Cinematic Point of View. Maryland: John Hopkins University Press, 1986. Print.
20. ---. “Le Grand Imagier Steps Out: The Primitive Basis of Film Narration”. Philosophy of Film and Motion Pictures. Ed. Noel Carroll and Jinhee Choi. Malden/Oxford/Victoria: Blackwell Publishing Ltd., 2005. 185–99. Print.
Billingham, J. (2012) 'Film-thinking and narrative indeterminacy', Alphaville: Journal of Film and Screen Media, 4, pp.68–82. https://doi.org/10.33178/alpha.4.05.
Jimmy Billingham was awarded his AHRC-funded PhD from the University of Sussex in August 2011 for a thesis entitled “The Act of Viewing: Indeterminacy and Interpretation in Narrative Film”. He now lives in London and works as a freelance editor. Jimmy is currently writing an article on drone music and altered states, and is himself an ambient-drone musician, with releases on various notable labels, including his own imprint.