A       Cognitive theories of
narration (Lost Highway)
Introduction
Martha Nochimson warns film critics and theorists that their linguistic and rational tools will miss or at least distort the specificity of David Lynch's films, which represent 'a level of non-rational energy on which all kinds of meaningful activity takes place' (Nochimson 1997: 6). Rather than suggest that Lynch has abandoned language and rationality, Nochimson argues that he places language and reason in the context of the subconscious, which has the effect of relativizing their 'imperialism'. Moreover, Lynch celebrates rather than distrusts the non-rational energy of the subconscious. He invites the spectator to 'suspend the desire for [rational] control by engaging in an empathetic relationship with a protagonist who, as a matter of survival, must learn to permit a channel to the subconscious in order to open the self to the universe' (Nochimson 1997: 11). For Lynch, the rationalist illusion of control and mastery creates a barrier to the real. In his films, Lynch unshackles himself from Western society's ultra-rational way of thinking, realizes that there is meaning beyond rationality, and tries to convey that our fear of letting go of rationality is simply a fear created by our prison-house of language.
Lynch uses the cinema to express non-rational energy in tangible form (visually and aurally). This energy is familiar to us all, but has been repressed in us by language, rationality, and education. This is one reason why Lynch's films seem to be nonsensical, but nonetheless evoke powerful feelings. It is easy to make nonsensical films that don't evoke any feelings at all, because they don't engage with the non-rational energy that Lynch evokes.
We can use the term 'uncanny' to describe the powerful feelings the non-rational energy evokes in us. This term designates a mood more than anything else - a mood created by uncertainty and confusion as the familiar world starts to become strange, and we begin to lose our bearings. Seemingly ordinary events - everything from a blazing fire in the fire place, traffic lights changing from red to green, the sound of the wind, to a drive along the road
Cognitive theories of narration     169
are rendered strange in Lynch's films. But Lynch does not simply present strange events; he also creates an uncanny feeling in the spectator. Spectators quite literally come out of his 1997 film Lost Highway feeling disoriented, since the film has challenged a number of the certainties we hold within the boundaries of our linguistic and rational world. Nonetheless, the strange world seems familiar at the same time - it is that realm of experience that language and rationality have made us repress. In this sense, Lynch's films are receptive to the subconscious, since they evoke a 'return from the repressed'. Although Lynch's films are receptive to the non-rational energy of the subconscious, most spectators do not switch off their rational faculty when watching a film like Losf Highway, although this faculty does become challenged and has to work harder. Our aim in this chapter is to analyse Lost Highway in terms of its 'comprehensibility'. We do not aim to locate mysterious hidden meanings in the film, but want to understand how the film can be comprehended on the basic level of its story structure, and in terms of how the film's narration conveys that story to spectators. We shall attempt to demonstrate that the cognitive theory of narration David Bordwell pioneered in his book Narration in the Fiction Film (1985), and which Edward Branigan developed further in Narrative Comprehension and Film (1992), are ideally suited to analysing complex films such as Lost Highway. This is because cognitive theory constructs a model of the norms, principles, and conventions that explain how spectators routinely comprehend films. Cognitive theory can therefore highlight and account for the moments in a film where comprehension breaks down. Or, more positively, it can focus on the moments where the film goes beyond the spectator's routine and rational ('common sense') way of comprehending a film, and begin to determine how non-rational energy has influenced the film's structure and meaning.
6.1. Theory: Narration in the Fiction Film
In Narration in the Fiction Film David Bordwell develops a cognitive theory of film comprehension, which he explicitly opposes to a psychoanalytic theory of film. Psychoanalytic film theorists (whom we discuss in Chapters 8 and 9) define the experience of reality as not being delimited by the horizon of consciousness (or 'common sense'), but argue that it includes myth, ideology, and unconscious desires and fantasies. According to psychoanalysts, our consciousness is merely the tip or peak of our identity, most of which remains hidden and repressed. But for cognitive scientists, consciousness is not a mere superstructure, but the base, or basis, of identity. Following the cognitive scientists, Bordwell argues that film theorists should begin with cognitive explanations of filmic phenomena, and should move on to psychoanalytic
170    Studying Contemporary American Film
explanations only if a cognitive account is found wanting: 'The theory 1 advance attends to the perceptual and cognitive aspects of film viewing. While 1 do not deny the usefulness of psychoanalytic approaches to the spectator, 1 see no reason to claim for the unconscious any activities which can be explained on other grounds' (Bordwell 1985: 30).'
The basic premise of Bordwell's theory is that narration is the central process that influences the way spectators understand a narrative film. Moreover, he argues that spectators do not simply absorb a finalized, preexisting narrative, but must actively construct its meaning. Bordwell develops his theory within what is called the 'constructivist school' of cognitive psychology, which studies how perceivers 'make sense' of the world from inherently fragmentary and incomplete data and experiences. For example, we can only directly see three sides of a six-sided solid cube. But from this incomplete experience, we complete the cube by 'appending' the other three sides. Bordwell and other cognitive film theorists argue that film is like a six-sided cube in which spectators see at most only three sides on screen. The spectator has to complete the film by appending the other three sides, so to speak.
In Chapters 3 and 4 of Narration in the Fiction Film Bordwell outlines a cognitive theory of film that tries to explain how spectators complete a film's narrative, rendering it coherent. Spectators are not free rational agents who can simply 'fill in the gaps' in a film in any way they wish. Instead, intersubjective norms, principles, and conventions guide them. When watching a narrative film, spectators do not simply 'absorb' the data, because it is not complete in itself. Instead, they have to process this inherently incomplete data. And they process it using what cognitive psychologists call schemata - norms and principles in the mind that organize the incomplete data into coherent mental representations. Schemata are activated by 'cues' in the data. Bordwell notes that gaps in the data are the most evident cues, for they are simply the missing data that spectators need to fill in. For example, a cube 'suggests' its three hidden sides (the missing data) by a variety of cues, including the way the three visible sides are projected in space, the way the visible sides form edges, and so on. More accurately, the cube cues us to fill in the three hidden sides. This process of filling-in is called hypothesis or inference generation.
Narrative films cue spectators to generate inferences or hypotheses - but not just any inferences. When comprehending a narrative film, one schema in particular guides our hypotheses - the one that represents the canonical story format:
Nearly all story-comprehension researchers agree that the most common template   structure   can   be  articulated   as   a   'canonical'   story   format,
Cognitive theories of narration     171
something like this: introduction of setting and characters - explanation of a state ot alfairs - complicating action - ensuing events - outcome - ending.
(Bordwell 1985: 35)
Moreover, comprehension of a narrative is made easier if it is organized around a goal-oriented protagonist - a character who drives the narrative forward towards his or her predefined goal.
Spectators do not, therefore, enter the cinema with a blank mind and passively absorb the film's narrative. Just as each language-learner internalizes the rules of his/her native language, so each film spectator internalizes a schema, a template or set of norms and principles with which to comprehend narrative films. In Western societies, spectators internalize a schema called the canonical story format.
But exactly how does the narrative schema work? Bordwell notes that, when spectators are presented with two events in a film, they employ the narrative schema to attempt to link the events together - either spatially, temporally, and/or causally. As the film progresses, spectators rearrange events, disambiguate their relations and order, and in doing so, gradually construct a story. Following the Russian formalists, Bordwell calls the resulting story the fabula: 'the fabula embodies the action as a chronological, cause-and-effect chain of events occurring within a given duration and spatial field' (Bordwell 1985: 49). Bordwell calls the actual order in which the fragments of the fabula events are presented the plot, or syuzhet: 'The syuzhet (usually translated as "plot") is the actual arrangement and presentation of the fabula in the film. It is not the text in toto. It is a more abstract construct, the patterning of a story as a blow-by-blow recounting of the film could render it' (p. 49).
The third element (after the fabula and syuzhet) that influences film comprehension is style, which Bordwell simply defines as a film's 'systematic use of cinematic devices' (p. 50). Bordwell defines narration as a combination of syuzhet and style, which interact with the spectator's narrative schema in constructing the fabula.
Bordwell emphasizes that the fabula is a mental representation, and that spectators construct the fabula on the basis of cues in the syuzhet (and style) interacting with the narrative schema. Moreover, he argues that the key to comprehending a particular film is determined largely by the relation between the fabula and the syuzhet - or, more specifically, by whether the syuzhet facilitates or blocks the spectator's construction of the fabula.
Because the film's fabula is a mental representation the spectator constructs during her ongoing experience of the film's syuzhet, the tabula is in a constant state of change, due to the spectator's ongoing generation of new hypotheses, strengthening of existing hypotheses, and abandonment of existing hypotheses. Spectators may need to abandon hypotheses because they only
172    Studying Contemporary American Film
have a probable reality. Or they may generate several conflicting hypotheses on the basis of a few cues, and then reduce them as the syuzhet presents additional cues. Moreover, a film may deliberately lead spectators to generate incorrect hypotheses (the phenomenon of unreliable narration'), or the film may deliberately challenge the canonical story format: 'If the film does not correspond to the canonic story, the spectator must adjust his or her expectations and posit, however tentatively, new explanations for what is presented' (Bordwell 1985: 36). This is the case with Lost Highway.
6.2. Method
The most analytically important variable is the set of formal correspondences between fabula and syuzhet. That is, to what extent does the unfolding syuzhet correspond to the logical, temporal, and spatial nature of the fabula we construct?
(Bordwell 1985: 54)
The film analyst can extract a fairly rich methodology from Bordwell's cognitive theory of narration. Bordwell introduces a series of concepts to explain how narration works - that is, how syuzhet and style facilitate and hinder the spectator's construction of the fabula. These concepts include: various hypotheses the syuzhet encourages the spectator to generate; various gaps constructed by the syuzhet constructs; types of exposition employed by the syuzhet; tactics for delaying the release of fabula information; redundancy (the way some information is conveyed several times); and broad narrational strategies, such as knowledgeability, self-consciousness, and communicativeness.
A general principle is to look for cues in the film, including patterns, gaps, and the way the syuzhet is organized. More specifically, look for the way the canonical story format is cued or thwarted. Bordwell writes that: 'the schemata [particularly the narrative schema] need a firm foothold somewhere. The sequential nature of narrative makes the initial portions of the text crucial for the establishment of hypotheses' (Bordwell 1985: 38). Analysing the opening scenes of a film is therefore crucial to a cognitive analysis of narration.
Not all cues and hypotheses are the same. Bordwell identifies several types of hypothesis. You need to ask yourself: Is the film encouraging me to generate a curiosity hypothesis (that is, an hypothesis about past events) or a suspense hypothesis (is the film asking me to anticipate forthcoming events)? Is the hypothesis probable or improbable? Is it exclusive or non-exclusive? And at what level is the process of hypothesis generation taking place - a micro-level (moment by moment) or macro-level (large scale)?
Cognitive theories of narration    173
Bordwell also identifies several types of gap (the most recognizable cue in the text). When analysing gaps, we need to ask: Are they temporary or permanent? Most are temporary - that is, resolved by the end of the film. Second, are they flaunted or suppressed? A gap is flaunted when the spectator is made aware that there is some information they need to know about the fabula, whereas a suppressed gap does not call attention to itself. Finally, are the gaps diffused or focused? A diffused gap is open ended, leading the spectator to generate a series of non-exclusive hypotheses, whereas a focused gap is clearly defined and leads the spectator to generate an exclusive hypothesis. A diffuse gap introduced at the beginning of a film can be gradually brought into focus as the film progresses.
The expositional moments in a film introduce pertinent background information about the settings, characters, and states of affairs. Exposition can be concentrated into a few scenes or, more rarely, diffused throughout the whole film. If concentrated, it may be preliminary (appearing at the beginning of the film) or delayed until the end (as in detective films). The syuzhet can also set up false leads, complications in the action, and subplots to delay fabula information. Or it may convey some information on several occasions (redundancy), to reinforce the importance ofthat information and ensure its effective communication. (This is why redundancy is a standard principle of classical narration.) We shall see that Lost Highway employs very little exposition. As the film unfolds, spectators gradually expect a delayed scene of concentrated exposition at the end. However, this explanation never arrives, which is one reason why the film is disorienting.
More generally, a film's syuzhet is constructed using broad narrational strategies, including knowledgeability, self-consciousness, and communicativeness. Under knowledgeability, Bordwell includes a syuzhet's range of knowledge and its depth. Is the knowledge limited to what one character knows about fabula events (restricted narration), or does it go beyond what any character knows (omniscient narration)? And is that knowledge deep (does it delve into the character's mental life)? Or does it remain on the surface (simply showing the characters' behaviour)? Second, a self-conscious narration displays a recognition that it is addressing an audience. Cues of self-consciousness include: characters looking into the camera, voiceovers addressing the spectator, and frontality of figure position. Third, the narration may withhold from the spectator some of the available information. For example, if the narration shows us the fabula through a character's eyes, is it willing to show us all the relevant information that character sees (in which case it is being highly communicative)? Or does the narration suddenly change perspective at a crucial moment, thus denying the spectator some important information (in which case it is less communicative)?
From these various principles we can begin to identify the film's internal
174    Studying Contemporary American Film
norms - its specific syuzhet structure. We shall see that our experience of Lost Highway is strongly determined by a syuzhet consisting of flaunted focused gaps, as well as a radical challenge to the canonical story format.
6.3. Analysis
Armed with the notion of different narrative principles and the concept of the syuzhet's distortion of fabula information, we can begin to account for the concrete narrational work of any film.
(Bordwell 1985:51)
The credit sequence of Lost Highway consists of a shot of a camera attached to the front of a car travelling very fast along a highway at night. The car's headlights illuminate the road. The credits appear from the middle of the screen, travel rapidly towards the camera, and pause momentarily (the letters appear to stick to the film screen) before disappearing 'behind' the camera and spectator.
In the opening scene we are introduced to Fred Madison (Bill Pullman) at home. Because of his frequent appearance in the opening scenes, we assume that Fred is the film's main protagonist. The scene opens with Fred sitting in the dark on the edge of his bed. He is smoking a cigarette and looking at himself in the mirror. The front-door intercom buzzes, and he hears the message 'Dick Laurent is dead'. He goes over to the window in another part of the house to look out, but he sees no one. As he heads towards the window we hear the off-screen sound of tyres screeching and a police siren.
We have seen Bordwell argue that a film's beginning is crucial because the spectator's hypotheses need to establish a foothold in the film early on. The intercom message leads the spectator to generate at least two hypotheses, focused around the questions: Who rang the bell? And, who is Dick Laurent? These two hypotheses are generated in response to the gaps in the fabula that the syuzhet has constructed. Firstly, knowledge about the fabula is severely limited. But this limitation is motivated because the narration is linked to Fred's level of awareness and experience of tabula events: the spectator sees and hears what Fred sees and hears. The knowledge is therefore deep and restricted, and the syuzhet is being communicative, because it gives the spectator access to this knowledge. (We shall see later in this chapter that Fdward Branigan discusses character awareness and experience in terms of the concept of focalization.) The gaps in the fabula are, first, spatial. The restricted narration does not show us the identity of the person outside and does not show us the source of the off-screen sounds. This spatial gap in the fabula is evident to the spectator, and is therefore a flaunted (rather than a suppressed) gap. It is also a clearly delineated gap, and is therefore specific
Cognitive theories of narration     175
(rather than diffuse). Finally, it is temporary (rather than permanent) because it is eventually filled in at the end of the film. The hypothesis we generate about this spatial gap is a suspense, non-exclusive hypothesis operating at the film's macro-level. It is a suspense hypothesis because we assume the gap will be resolved in the future (so we anticipate the filling in of this gap at a later time in the film's unfolding); it is non-exclusive because it could have been anyone (we cannot generate a hypothesis suspecting a particular person); and it operates on the macro-level because it spans the entire film. The scale of probability/improbability usually refers to the hypotheses we generate. But in this case, the way the syuzhet fills in this gap at the end of the film is highly improbable. Although our hypotheses were non-exclusive, it is highly unlikely that any spectator would generate the hypothesis that Fred is also outside the house pressing his own doorbell!
The lack of information on Dick Laurent's identity is a temporary, flaunted, focused gap that leads the spectator to generate an exclusive curiosity hypothesis that operates on the macro-level (for his identity is not immediately resolved). In a more conventional film (one that follows the conventions of the canonical story format), the spectator's narrative schema would condition her to expect the next scene to contain exposition explaining who Dick Laurent is.
The screeching tyres and the police siren are not coded as prominent cues, and many spectators may not perceive them as cues, but as part of the film's 'reality effect' - that is, backgroud noise that one may expect to hear, rather than a significant narrative event. In summary, the opening scene enables the narrative schema to gain a foothold in the film, since the spectator generates hypotheses in response to the gaps the syuzhet has constructed, and is anticipating events in future scenes.
The first scene ends on an establishing shot, a very long shot of the front of Fred's house in the early morning light. After a fade, the second scene begins by repeating this exterior establishing shot, except that this time it is night. Inside the house, we see Fred packing a saxophone into its case, and talking to Renee (Patricia Arquette), who wants to stay home and read rather than go to the club with him. This seemingly simple scene nonetheless keeps the spectator busy. It appears to follow the canonical story format by continuing to introduce the setting and characters, and by explaining a state of affairs. On the basis of the two exterior establishing shots (shown back to back), we generate the hypothesis that the film has now progressed from morning to evening of the same day. In other words, using our narrative schema, we establish a linear temporal relation between the two scenes. Second, information about Fred is conveyed indirectly: we assume he is a musician, and we find out from his talk with Renee that the two of them are married (Renee: T like to laugh, Fred.' Fred: 'That's why I married you.') The deadpan
176    Studying Contemporary American Film
way the two characters interact, plus the sparse dialogue, may suggest that the marriage is at a stalemate, to the point where Fred's sax-playing bores Renee, and she invents improbable reasons for wanting to stay home (she does not appear to be the type of person who will spend her evenings at home reading).
In contrast to the end of scene 1, scene 2 ends abruptly, as we cut from the quiet interior of the Madisons' house to an image of the exterior of the Luna lounge and the very loud sound of sax music. This sudden break from scene 2 jolts the spectator, not only because of the contrast in sound and image, but also because there is no reference to the two gaps in scene 1. Fred does not mention the message he received on the intercom, and therefore we are no closer to rinding out who the messenger was, or who Dick Laurent is. Reference to these cues would have strengthened the causal relationship between scenes 1 and 2. As it is, the two scenes are linked more tenuously -visually (the visual repetition of the establishing shot), and linearly (the progression from day to night), rather than causally. The syuzhet is marked by a lack of redundancy between scenes.
In scene 3 Fred is shown playing his sax and phoning his wife during an intermission. But no one answers the phone at home; the house appears empty. In scene 4 Fred returns home to find his wife asleep in bed. These two scenes introduce a discrepancy between Renee's words and actions. In combination with the way Fred and Renee interact in scene 2, the discrepancy enables the spectator to group these actions together and call them a complicating action, the next stage of the canonical story format. The complicating action can be called 'unhappy marriage', with the probability of infidelity on Renee's part. The spectator's hypothesis of infidelity is a near-exclusive hypothesis - one with only a few alternatives - and is generated on the basis of Renee's absence from the house in scene 3 (a flaunted, focused gap in the film's fabula). The infidelity hypothesis is the most probable, but because the narration is restricted to Fred's perspective, the spectator does not gain any more information than Fred possesses in order to confirm or disconfirm this hypothesis.
Scene 5, the next morning. Renee picks up the newspaper outside, and discovers a videotape on the steps, with no addresser, addressee, or message. After watching the tape, which shows the outside of their house plus a closer shot of the front door, the Madisons arc understandably perplexed, and Renee generates the weak, improbable hypothesis that an estate agent may have made the tape. This scene again presents another flaunted, focused gap in the fabula (which can be formulated into the following question: Who made the tape?), and its only link to the previous scenes is a continuity of characters and settings. There is no narrative continuity between this scene and the film's previous scenes. But the narration does seem to establish an internal norm,
Cognitive theories of narration     177
whereby it selects very specific portions of the fabula to show - namely, actions and events performed early in the morning or late at night.
Scene 6. Renee and Fred in bed at night (the film therefore continues to follow the internal norm of only showing actions performed in the morning or at night). We see several of Fred's memory images - of him at the Luna lounge playing his sax, and seeing Renee at the club leave with another man (we later find out that he is Andy). Fred and Renee then make love, and afterwards Fred recounts a dream he had the previous night. We then cut to several of the dream images: Fred looking around the house at night, hearing Renee call out to him, and a shot of Renee in bed being frightened by a rapidly approaching, off-screen agent. The spectator shares this agent's vision as he approaches Renee, but we do not see who it is. Renee looks into the camera and screams (her look into the camera makes the narration mildly self-conscious). Fred is then shown waking up, and looking at Renee. But another face is superimposed upon her (we later find out that it belongs to the 'mystery man').
In this scene the narration continues to be restricted and deep, and seems communicative, as we gain access to Fred's memory and dream images. But the status of the memory images is ambiguous. To make sense of these images, we can generate the probable hypothesis that they refer to a fabula event that took place before the film begins, and that Fred is generating these images to fill in the gap in scene 3. We can paraphrase these images in the following way: 'Last night Renee was probably with the guy who accompanied her to the Luna lounge on a previous occasion.'
The dream images are also ambiguous. Above we noted that the narration in this scene is restricted and deep, and appears to be communicative. But a close analysis of later scenes in the film makes the film analyst realize that the narration in this scene is in fact being very uncommunicative, although its uncommunicative status is disguised. At this stage in the film, we are used to a communicative narration, with flaunted gaps. But in the dream sequence, we (or, at least, the film analyst) retrospectively realize that the narration contains suppressed gaps and is uncommunicative. This is why later scenes in the film jolt us.
Furthermore, Fred does not question Renee about her whereabouts the previous evening (although in the script he does'). He only mentions the dream he had the previous night. Therefore, the link between this scene and previous scenes is only temporal, rather than causal (although the memory images of the club do at least refer us back to the location of scene 3). Due to this lack of causal cues, the spectator tries to generate weak hypotheses to connect the mystery narrative (who is Dick Laurent?) to the romance narrative (perhaps the guy in the club is Dick Laurent, etc.).
Other events in this scene are even more vague. As Renee and Fred make love,
178    Studying Contemporary American Film
the screen suddenly turns white for a moment (a white fade) and the action is slowed down. Although these make the narration self-conscious (both devices challenge the conventions of standard speed and black fades), it is unclear what these devices are meant to cue, other than the director's intervention (thereby defining the mise en scene as mannerist; see Chapter 3). Furthermore, the moment in which Fred wakes up and sees the mystery man's face superimposed upon Renee's face still constitute part of Fred's dream he had the previous evening. In other words, the scene ends inside Fred's dream; the narration does not return to the image of Fred in bed narrating the dream. But one thing at least is clear from this scene: Renee and Fred make love dispassionately, which strengthens the hypothesis that their marriage is in crisis.
Scene 7: the following morning. Renee finds another videotape on the steps. But this one shows more than the previous tape by filming inside the house, and ending on Fred and Renee asleep in bed. Renee then calls the police. The second videotape adds to the 'complicating action' chunk of the by now strained canonical story format.
In scene 8, two detectives watch the second videotape, and look around the house for possible signs of a break-in. Information is planted in the dialogue to facilitate the spectator's hypothesis-generating process. Fred tells the detectives he hates video cameras, because he likes to remember things his own way, not necessarily the way they happen. Many critics who reviewed Lost Highway saw this line as a nodal point on which to focus the previous scenes, as a key to the film's meaning - namely, many of the narration's twists can be motivated psychologically, as Fred's distorted view of events. The spectator also has the opportunity to test out this reading of the film in the next scene.
Scene 9: party at Andy's house, late at night. (Andy was shown previously in Fred's memory images - in scene 6 - of Renee and Andy leaving the Luna lounge while Fred plays his sax.) The canonical story format is challenged, and begins to break down in this scene. In the previous scene, Fred and Renee did not talk about going to a party at Andy's house. If they had done so, it would have strengthened the causal relation between the scenes. More radically, Fred is shown drinking two whiskies and then talking to 'the mystery man' (Robert Blake), whose face we have already seen superimposed over Renee's race in Fred's dream (recounted in scene 6). The mystery man says to Fred that they have met before (although Fred does not remember), and then defies Newtonian space-time physics by suggesting that he is in Fred's house at that very moment (that is, is in two places at once). The mystery man 'confirms' this by persuading Fred to phone his home, where indeed the mystery man also answers! In terms of their meeting before, we tend to side with the mystery man; first because of the comment Fred made in scene 8 (he likes to remember things his own way, not necessarily the way they happened); and
Cognitive theories of narration     179
second, because the mystery man appeared in Fred's recounted dream in scene 6 (his face is superimposed over Renee's face).
Challenging one of the fundamental background assumptions of the canonical story format (a character cannot be in two places at once) begins to open up inferential possibilities, as we have to try to explain or motivate this fundamental discrepancy in the film's tabula. Is Fred simply delusional, and are we sharing his delusions? This assumes that the suyzhet continues to be highly communicative by conveying Fred's deep experiences.
When Fred asks Andy who the mystery man is, he responds that he is a friend of Dick Laurent, to which Fred replies: 'Dick Laurent is dead, isn't he?' Troubled by this news, Andy protests that Dick Laurent can't be dead. (He also says to Fred, T didn't think you knew Dick', which is an odd thing to say immediately after telling Fred that the mystery man is a friend of Dick Laurent.) The mention of Dick Laurent's name finally brings into sharper focus one of the gaps opened up in scene 1. Through its linear progression from scene to scene, the syuzhet is now finally beginning to refer back to gaps in previous scenes.
Scene 10: Fred and Renee in the car on the way back from Andy's party. Renee describes how she met Andy (T met him at this place called Moke's ... We became friends ... He told me about a job ...'.) This is one of the few fragments of exposition the spectator has received from the syuzhet, which means that it is impossible to slot it into a context and make more sense of it. Instead, it remains a fragment of exposition. But from Fred's reaction we understand that he dislikes Andy, possibly because he thinks Renee is having an affair with him.
Scene 11. At home, Fred checks around the house. He switches off the alarm, hears the telephone ring (but doesn't answer it), and he and Renee then prepare for bed. The events that follow refer back to Fred's recounted dream in scene 6. What is extraordinary is that parts of Fred's dream are now repeated and 'played out' as non-dream events. As Fred wanders around the house, we hear Renee call out Fred's name in exactly the same way she did in the dream. In a few images we also see Fred moving around the house in the same way he did in his dream. In the bathroom, he looks intensely at himself in the mirror. He then walks towards the camera, blocking our view entirely (a self-conscious moment of narration). In the next shot, the camera is positioned outside a door.
This scene challenges the canonical story format further, since it distorts the notion of linear progression (unless we read the dream as a premonition). At the end of the scene, the syuzhet is also uncommunicative, because it creates a gap in the fabula as the camera is positioned outside a closed door. But on the basis of a shot in Fred's dream - of an unseen agent attacking Renee in bed, together with the hypothesis that Renee is having an
180    Studying Contemporary American Film
affair - we can generate a near-exclusive, probable hypothesis that Fred is attacking Renee.
This hypothesis is confirmed in scene 12. Fred finds a third videotape on the doorstep, plays it, and sees, in addition to the initial footage on the second tape, a series of shots depicting him murdering and dismembering Renee. He acknowledges the video camera filming him, by looking directly into it (making the narration self-conscious). But for Fred watching the tape, the images are horrifying, and in desperation he calls out to Renee. He is suddenly punched in the face by one of the detectives who visited the house in scene 8. There is a flaunted ellipsis in the fabula at this moment in the film, as Fred is now being questioned about Renee's murder (scene 13). The syuzhet is both communicative and uncommunicative, since it shows us (via the video camera images) Fred murdering Renee, but it is uncommunicative in supplying information about who recorded the videotapes, who Dick Laurent is, the mystery man's ability to be in two places at once, and the identity between Fred's recounted dream and Renee's murder the following evening. More generally, the film is marked by a lack of synchronization between its fabula and unfolding syuzhet.
Retrospectively, we can now re-evaluate the film so far as a detective film, which Bordwell (1985: 64) defines as having the following characteristics: a crime (cause of crime, commission of crime, concealment of crime, discovery of crime) and investigation (beginning of investigation, phases of investigation, elucidation of crime). We can characterize the film as enacting a crime, with emphasis on its concealment and discovery, with a very condensed investigation (at this stage consisting of identification of criminal and consequences of identification). We hypothesize that Fred is the causal agent, motivated by jealousy, who carried out the murder soon after Andy's party.
The policeman throws his punch directly at the camera, suggesting the syuzhet's continued alignment with Fred. It also makes the narration self-conscious, not only because the action is directed at the camera, but also because it reminds a cine-literate spectator of similar moments in Hitchcock's films - most notably, Strangers on a Train and North by Northwest, where punches are similarly directed at the camera. In another Hitchcockian moment, Fred's trial is not shown, but is reduced to the voiceover of the judge pronouncing sentence, as Fred is led to his cell. This goes beyond Hitchcock's rapid depiction of Margot Wendice's trial in Dial M for Murder. (In Lost Highway, a scene taking place in the courtroom is in the script, but has been omitted from the final cut.)
There follows a quick series of scenes (sometimes consisting of three or four shots) as Fred is taken to his cell (scene 14), which is intercut with video images of Renee's murder (coded as Fred's memory images). Scene 15 continues with this theme, as Fred tries to figure out what is happening to him. In scene 16 he
Cognitive theories of narration    181
collapses in the prison courtyard, complaining of a headache. In scene 17 the prison doctor forces him to swallow some sleeping pills, and in scene 18, he has a vision of an exploding cabin in the desert, although the explosion is shown backwards (making the narration self-conscious). The mystery man then comes out of and goes back into the cabin. The iconography reminds the cine-literate spectator of the exploding beach house at the end of Kiss Me Deadly (Robert Aldrich, 1955) - a film that also begins with a shot from a camera attached to the front of a car travelling very fast along a highway at night. During these scenes in the prison, it becomes evident that Fred is not a rational, goal-driven agent who causally motivates narrative events, since he is unable to remember or explain his actions. But his state of mind motivates the lack of synchronization between the fabula and syuzhet.
The events in the second half of scene 18 and in the following scenes completely defy and undermine the canonical story format. Scene 18 ends on the following shots:
•     Fred's cell; there is a sudden flash of bright light, and light bulb in his cell goes dim (perhaps representing the effects of an electrocution on the rest of the prison);
•     the highway at night, repeating the image of the credit sequence; but this time, the car stops in front of a young man (whom we later find out is Fete Dayton);
•     cut in closer to Pete, with a superimposed shot of his girlfriend, Sheila, and Pete's parents; Sheila is screaming Pete's name;
•     big close-up of Pete's eyes, superimposed over an image of the light in Fred's prison cell;
•     Fred frantically rocking from side to side in his prison cell, screaming and covered in blood (this image seems to be strongly influenced by Francis Bacon's portraits, to the extent that it can be read as a filmic equivalent to Bacon's still images);
•     shot of the prison ceiling; the camera pans down to Fred;
•     cut to an image of what looks like an open wound, and the camera moves towards it.
Working along the lines of a surreal logic, the syuzhet presents a series of fragmented fabula events, which retrospectively we infer signifies Fred's transformation into Pete. (The version of this event in the script is more explicit about the transformation.)
Scenes 19-25 depict a prison guard discovering Pete in Fred's cell, Pete's identification, his release, and his home and work life. It is as if the film has 'started again' or, more accurately, we seem to be watching the second half ot another film, because the syuzhet has identified Pete as the film's main protagonist, and has introduced a new setting and additional characters. This
182    Studying Contemporary American Film
sudden jolt in the film's fabula is caused by the fact that the previous protagonist, to whom we were given privileged access, and from whom the camera rarely departed, has suddenly and inextricably disappeared from the fabula.
This jolt is far more radical than superficially similar scenes in other films -such as the murder in Psycho of Marion Crane, the film's primary protagonist up to that point. In Psycho, the transfer from iMarion to Norman takes place within a stable fabula. In Lost Highway, the fabula has been severely disrupted, creating a flaunted but diffuse permanent gap that is never filled in. But to attempt to fill in this gap, the spectator needs to generate the two mutually exclusive hypotheses: Is Pete the same person as Fred? That is, are two actors playing the same character? Or are the two actors playing two different characters? However, the syuzhet does not contain sufficient cues to enable us to choose one hypothesis over the other.
After the spectacular transfer of agency from Fred to Pete, we start to question the communicative status of the syuzhet. It seems to hide more than it shows. As we continue to watch the film unfold, the unresolved issues remain, because the film does not address or even acknowledge them - that is, until scene 24, when Pete's girlfriend, Sheila, mentions to Pete that he has been acting strangely since 'the events' of the previous evening. Also, the syuzhet continues to follow the internal norm establish at the beginning of the film - to depict events taking place early in the morning or late at night. In scene 25 Pete returns to work (as a mechanic), and in scene 26 he repairs the car of a gangster, Mr Eddy. The scene ends with Mr Eddy driving away from the garage where Pete works, and two cops who are following Pete identify Mr Eddy as Dick Laurent. One of the gaps presented in the fabula at the beginning of the film (who is Dick Laurent?) is now brought into clearer focus, although it raises another question: Why is Mr Eddy also called Dick Laurent? From this moment onwards, the suyzhet makes additional and more frequent references to the first part of the film, enabling the spectator to focus other gaps and refine hypotheses.
At the beginning of scene 27, Pete looks at himself in the mirror in the same way as Fred looked at himself just before he murdered Renee. But in this part of the film, Pete takes Sheila out on a date.
Scene 28. At work the following morning, Pete hears sax music on the radio - identical to the music Fred played at the Luna lounge. The music distresses Pete, and he switches it off. A few moments later, he meets Mr Eddy's girlfriend, Alice, played by Patricia Arquette, who also played Fred's wife, Renee. But, as Renee, Arquette looked vampish; as Alice, she conforms to the stereotype of the blonde femme fatale. Hypotheses we generated when Fred transformed into Pete recur here, but inverted. Now we need to ask: Are Renee and Alice the same character in disguise (because they are played by the same actress)? Or is Patricia Arquette playing two characters?
Cognitive theories of narration    183
Scenes 29-32 depict the affair that Pete develops with Alice. In scene 33, after Alice has had to break off a date with Pete, we see Pete in his room, experiencing hallucinations and hearing strange sounds. The syuzhet is restricted, deep, and communicative because the spectator directly shares these experiences (the camera goes out of focus, we see Pete's hallucinations of Alice, and so on). In scene 34 Pete decides to go and see Sheila, and in scene 35, Pete's parents talk to him. They know what happened to him, but refuse to tell him everything. They tell him that he came home with Sheila and 'a man', but say no more. This scene is interrupted by a montage sequence repeating the shot, in scene 18, of Pete's parents and Sheila screaming, although this time it is not superimposed over an image of Pete. This is followed by a shot of the open wound, and a video image of Renee's mutilated body. These shots are coded as Pete's memory images, whereas previously they were coded as Fred's.
In following scenes, Mr Eddy threatens Pete, and Alice devises a plan whereby she and Pete will rob Andy and run away together. In scene 39 Alice tells Pete how she met Mr Eddy. She uses the same line that Renee used to describe to Fred (scene 10) how she met Andy ('I met him at this place called Moke's ... We became friends ... He told me about a job ...'.)
Scene 40. Pete breaks up with Sheila, and then takes a phone call from Mr Eddy-Dick Laurent and the mystery man (confirming Andy's comment in scene 9 that the mystery man is a friend of Dick Laurent). When the mystery man talks to Pete, he uses the same phrases as he did in scene 9, when speaking to Fred at Andy's party ('We've met before, haven't we?' etc.). He then indirectly threatens Pete.
In scene 41, Pete carries out Alice's plan to rob Andy, but in the process Andy is killed. Pete finds at Andy's house a photo of Mr Eddy, Renee, Alice, and Andy standing together. Pete asks Alice 'Are both of them you?', echoing a hypothesis the spectator generates when first seeing Alice in scene 29. Pete then goes upstairs to clean up, but the corridor in Andy's house looks like a hotel corridor. Furthermore, we sec flashes of light in the corridor, as we did in Fred's prison cell.
In scene 42 Pete and Alice drive to the desert to sell Andy's valuables to the mystery man at his cabin. Shots of the highway at night, and the shot of the burning cabin are repeated. In scene 43, Alice and Pete make love in the desert. Alice then goes inside the cabin, and Pete gets up. However, he has now inexplicably transformed back into Fred. But before we have time to adjust to Fred's sudden return to the fabula, the syuzhet presents a series of unusual shots and scenes. Fred looks into his car and sees the mystery man inside, staring back. We then hear the mystery man's voice off-screen, and he suddenly appears in the entrance of his cabin. He then goes inside the cabin in the same way he did in Fred's memory image in scene 18. Fred enters the
184    Studying Contemporary American Film
cabin where he sees the mystery man, but no Alice. Fred asks him where Alice is, and he replies that her name is Renee. Despite the photograph in Andy's apartment, Renee and Alice may be the same person (although this still does not explain her disappearance after entering the cabin). The mystery man then confronts Fred about his name, and begins to film him using a video camera. As Fred runs out of the cabin into his car, several of the shots are from the mystery man's perspective filmed through the video camera, and the resulting images are the same as the three tapes sent to the Madison's home. We can now fill in one of the gaps generated in the first part of the film, for we have conclusive evidence that the mystery man made the videotapes. Moreover, the hypotheses about the relation between Alice and Renee, and Fred and Pete, are brought into focus, but they are not resolved, for the 'solutions' the syuzhet present are highly improbable.
Scene 44. Fred drives along the highway, and shots of the highway at night are repeated. Scene 45 takes place at the Lost Highway motel. Fred wanders along the corridors in the same way Fete did in Andy's house. Fred enters an empty room, while Renee and Dick Laurent make love in another room. Another gap in the fabula is filled in, as we realize that Renee is having an affair with Dick Laurent, not Andy. (Andy therefore presents the spectator with a false lead.) In scene 46, Renee leaves the hotel and Fred knocks out Dick Laurent, watched by the mystery man. In scene 47 Fred takes Laurent to the desert where he kills him, with the help of the mystery man. In fact, the mystery man suddenly appears just at the right moment, to hand Fred a knife. The mystery man then shoots Laurent, but a few moments later he suddenly disappears, leaving Fred with the gun in his hand.
In an attempt to make sense of what is happening in these scenes, we can return to the opening scenes, when we generated the hypothesis that Renee may be having an affair. We now see that Fred has followed her to the Lost Highway motel, and discovers that she is having an affair with Dick Laurent, whom Fred subsequently kills. The mystery man and Alice now seem to be figments of Fred's imagination. However, if we accept this, then it generates more questions and additional gaps in the syuzhet, such as: Who made the three video tapes?
Scene 48. The police are at Andy's house. The two detectives who questioned Fred look at the photo - but she is missing, strengthening the hypothesis that Alice is a figment of Fred's imagination. The two detectives realize that Pete's prints are all over the place, so they head towards Fred's house. For this scene, then, the film has 'flipped back' to the Pete side of the fabula, but has erased Alice from it. The police have generated the hypothesis that Pete/Fred has murdered Andy, possibly because Fred thought that Andy was having an affair with Renee. In terms of the film's fabula, perhaps Fred followed Renee to Andy's place first, killed Andy, and then followed her to the
Cognitive theories of narration     185
Lost Highway motel, where he subsequently kills Dick Laurent. At some point in the fabula, he also kills Renee. (Obviously, in this reconstruction of fabula events, he does not kill her after going to Andy's party, because Andy is already dead.)
Scene 49. Outside Fred's house. Fred has just returned from the desert, and presses the intercom and leaves the message 'Dick Laurent is dead.' The two cops then turn up, and give chase. Scene 50: the film ends with the police chasing Fred as he continues to drive along the highway at night. He appears to undergo another transformation, but we are left with an image of the highway at night.
The gap opened up in scene 1 is now filled - it is Fred who rings his own doorbell and who leaves the message that Laurent is dead! The off-screen sounds of screeching tyres and the police siren are similarly repeated, but now as on-screen sounds. The final scenes fill in most of the gaps the syuzhet has generated, but they do not lead to a resolution, because the 'answers' they present pose additional questions since they are improbable answers.
From this cognitive reading of Lost Highway, we can discern several irresolvable ambiguities and inconsistencies. First, concerning character stability: in scene 9, at Andy's party, the mystery man is in two places at once. Fred is also in two places at once: in scene 1, he is inside his own house receiving the message on the intercom that 'Dick Laurent is dead' and in scene 49, which returns to scene 1, he is outside his house delivering the message. Other instabilities of character include Fred's transformation into Pete at the end of scene 18, and his transformation back again in scene 43; in scene 12 it appears that Renee is murdered, but in scene 28 she seems to return to the film's fabula. The spectator needs to ask whether Fred and Pete are the same character played by two different actors, or whether they are different characters. Are Renee and Alice two different characters played by the same actress, or the same character in disguise? And why is Mr Fddy also called Dick Laurent? Other irresolvable ambiguities include: in scene 6, in Fred's dream images, he sees the mystery man's face superimposed over Renee's face; but in scene 9, when the mystery man introduces himself, Fred cannot remember meeting him before. Scene 35 repeats the video images of Renee's murder, and a shot of the open wound; when they first appeared in scenes 15 and 18 (respectively) , they were coded as Fred's memories, but now they are coded as Pete's memories. Furthermore, the photo in Andy's apartment is shown in scene 41 and repeated in scene 48, but Alice is missing when the photo is shown again. Finally, there are ambiguities concerning the linear, temporal ordering of events: the events in scene 6 (Fred's recounted dream) are repeated (as non-dream events) in scene 11; and Fred's visions in scene 18, of an the exploding cabin from which the mystery man appears and disappears, are repeated in scenes 42 and 43, although they are not coded as Fred's visions.
186     Studying Contemporary American Film
But how can Fred's dreams and visions so accurately predict forthcoming events - unless those events have already happened? This suggests that the narrative of Lost Highway is organized like a loop - or better, a Möbius strip -rather than linearly. If this is the case, then scene 18, in Fred's prison cell, represents the twist in the Möbius strip, the twist where the topside is transferred to the underside. Scenes 1 and 49 are the moments where the two edges of the strip are connected together, with Fred represented outside his house on one side, and inside the house on the other. Moreover, to travel around the entire length of the strip, one needs to go around it twice - first on one side (from the intercom message to Fred's transformation in his cell), then on the other side (from Pete being released from prison to his transformation back into Fred), before we are returned to the moment where the two sides are joined (Fred conveying the intercom message to himself). The metaphor of the Möbius strip appears to accurately represent the structure of Lost Highway.
It is important to remember when discussing such ambiguous moments that our aim is not to disambiguate them, for this is a reductionist move that attempts to explain them away. Instead, we should attempt to explain how the ambiguities are produced, and what effects they achieve. These scenes contain either too few cues, or too many cues that contradict one another; or there are too many flaunted and suppressed gaps; or maybe the cue is a permanent gap. All these cues may lead the spectator to generate non-exclusive, diffuse hypotheses that are not brought into focus, or are 'resolved' in an improbable manner. Lynch's films are open to analysis as long as we do not try to reduce these ambiguous moments to a rational logic, but recognize that a non-rational but meaningful energy governs them. Lost Highway also prevents spectators from automatically applying schemata to it, since it goes beyond the common-sense, rational logic embedded in these schemata; instead, spectators become aware of the schemata's conventions, and work hard to apply them in new and unforeseen ways (spectators unwilling to do this stop watching the film).
6.4. Theory: Narrative Comprehension and Film
David Bordwell's Narration in the Fiction Film pioneered the cognitive theory of film, which flourished in the 1990s with books such as Joseph Anderson's The Reality of Illusion (1996), Edward Branigan's Narrative Comprehension and Film (1992), Gregory Currie's Image and Mind (1995), Torben Grodal's Moving Pictures (1997), Carl Plantinga and Greg Smith's (eds) Passionate Views (1999), Murray Smith's Engaging Characters (1995) and Ed Tan's Emotion and the Structure of Narrative Film (1996). These authors acknowledge the originality of Bordwell's book, and then proceed to refine the
Cognitive theories of narration     187
ideas articulated there, either by exploring underdeveloped areas (the role of emotions in cognition, how genres determine comprehension, the different levels on which spectators engage characters, the role of imagination and intentionality in comprehending fiction films), or re-establish cognitive film theory on a deeper foundation (such as ecology, biology, or neuroscience).
One frequent criticism emerges from these authors: Bordwell is an 'atheistic' narratologist because he does not recognize the role of an external 'master of ceremonies' controlling the events in the fabula. In other words, he does not posit the existence of external narrative agents (external to the fabula). He asks: 'must we go beyond the process of narration to locate an entity which is its source?' (Bordwell 1985: 61-2) and answers: 'To give every film a narrator or implied author is to indulge in an anthropomorphic fiction. ... [This strategy takes] the process of narration to be grounded in the classic communication diagram: a message is passed from sender to receiver' (p. 62). In place of this communication model, Bordwell argues that narration 'presupposes a perceiver, but not any sender, of a message' (p. 62). Branigan's cognitive model of narration presupposes both a sender and receiver of a film - in fact several senders and receivers.
Branigan draws upon concepts from cognitive science, narratology, and linguistics to develop his theory of film narrative and narration - more specifically, a theory of a story world's space, time, causality, of point of view, levels of narration, the relation between subjective and objective narration, and the relation between fiction and narrative. We shall not give a complete overview of Branigan's theory, but will instead focus on its most fundamental concepts and unique methodology. Like Bordwell, Branigan employs the concept of schema to explain the role ol narrative in organizing the spectator's experience of a film. Moreover, Branigan does not represent the narrative schema as a linear list, as Bordwell does when writing about the canonical story format. Instead, Branigan develops a more open and dynamic model, one organized as a hexagon with the main narrative actions (exposition, complicating action, and so on) represented at the points of the hexagon, and linked together by connecting lines (Branigan 1992: 17), This model captures the complexity of narrative more than a linear model because it describes the recursive nature of narrative: 'Narrative is a recursive organization of data; that is, its components may be embedded successively at various micro- and macro-levels of action' (p. 18). The narration conveys these narrative events to spectators, and the uniqueness of Branigan's theory and methodology lies in the complex model of narration he develops in Chapters 3 and 4 of Narrative Comprehension and Film.
While chapter 3 outlines disparities and hierarchies of knowledge conveyed by film narration (concepts that are similar to Bordwell's concepts of the range, depth, and communicativeness of the narration), it is in chapter 4 that
188    Studying Contemporary American Film
he develops a systematic theory and methodology of film narration. This theory is based on eight levels of narration, with a 'sender' and 'receiver' on each level (see diagram on p. 87 of Narrative Comprehension and Film). Branigan remains neutral on the controversial issue of whether we can describe narration as a form of communication (p. 107-10), but it is clear that he goes beyond Bordwell by theorizing the role of narrators in films.
Branigan defines narration as 'the overall regulation and distribution of knowledge which determines how and when the spectator acquires knowledge [of narrative events]' (Branigan 1992: 76). Whereas to study narrative is to find out what happens in a film, to study narration is to find out how spectators acquire knowledge of the narrative. The film agent is a crucial component in this process of knowledge acquisition.
For Branigan, a theory of film agents requires a fundamental distinction between historical authors, implied authors, narrators, characters, and focalizers. For the purposes of this section, we shall only focus on the latter three, since they are the most relevant in terms of methodology and textual analysis. Spectators comprehend characters as agents who exist on the level of narrative; the character is therefore an agent who directly experiences narrative events and who acts and is acted upon in the narrative world. A character whose experiences of the narrative world are then conveyed to spectators become focalizers. Narrators, on the other hand, do not exist in the narrative; they exist outside it on the level of narration. This means they have the ability to influence the shape and direction of the narrative.
One of the most important contributions Branigan makes to the study of film narration is his rigorous theory of focalization in film:
Focalization (reflection) involves a character neither speaking (narrating, reporting, communicating) nor acting (focusing, focused by), but rather actually experiencing something through seeing or hearing it. Focalization also extends to more complex experiencing of objects: thinking, remembering, interpreting, wondering, fearing, believing, desiring, understanding, feeling guilt.
(p. 101)
Branigan therefore distinguishes two types of focalization, each representing a different level of a character's experiences: external focalization, which represents a character's visual and aural awareness of narrative events (the spectator sees what the character sees, but not from the character's position in the narrative; the spectator shares the character's attention, rather than their experience); and internal focalization, which represents a character's private and subjective experiences, ranging from simple perception (optical vantage point) to deeper thoughts (dreams, hallucinations, memories).
The narrator is the third agent in film.  For Branigan, a narrator by
Cognitive theories of narration     189
definition does not exist in the narrative world, but on the level of narration. The narrator is an omniscient 'master of ceremonies' who does not see anything from a perspective within the narrative. Although the narrator is absent from the narrative, its presence is felt on the level of narration. For example, elements of the film that spectators cannot attribute to characters attest to the narrator's existence, including unmotivated camera movements (not motivated by the movement of characters or objects), inter-titles, and foreshadowing effects. (In classical mise en scene, shot changes are usually motivated by character movement, character glances off-screen, or by offscreen sounds and voices.) If a character in the narrative does not motivate a technique, then the spectator attributes it to the external narrator. Classical narration is defined by its attempt to conceal the narrator's presence from the spectator, whereas modernist narration continually reveals the narrator's presence (by means of unmotivated cuts, camera movements, and so on).
To avoid confusion, we should note that a character can become a narrator in the narrative world, where we see the character narrating the events in the form of flashbacks (as in films such as Double Indemnity or Sunset Boulevard). But these character-narrators are still characters, and a narrator external to the narrative still narrates the film.
Finally, Branigan emphasizes that these film agents and the levels at which they operate are not immanent in the film, but constitute part of the spectator's narrative schema: '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' (Branigan 1992: 85). We can go so far as to say that what exists on the movie screen is simply changing patterns of light and shade, from which the spectator then generates hypotheses to construct the film's fabula, including characters. It may sound strange to say that a character is simply a hypothesis generated by the spectator from a series of cues in the film, because characters seem so permanent. But as we have already seen, in Lost Highway the characters are not permanent, which prevents spectators from automatically applying their 'character' schema, making them aware of the schema's conventions.
From this brief outline, it should be evident that Branigan's theory of narration is more subtle than Bordwell's because Branigan makes more and finer distinctions. Moreover, Branigan does not use the same terminology as Bordwell. Branigan talks about 'diegesis' and 'narrative', rather than 'fabula' (although the terms are not equivalent); and 'levels of narration' rather than 'syuzheť and 'cues', a difference that marks a fundamental philosophical difference between Bordwell and Branigan. 'The notion of levels of narration' Branigan writes, 'provides a way of escaping a simple structuralism as well as
190    Studying Contemporary American Film
a strict empiricism, because comprehension is not made to depend upon a few basic surface units, or "cues," which may be endlessly combined in strings through addition and subtraction' (p. 112). For Branigan, schemata do not simply prime spectators to spot cues on the surface of a text. More fundamentally, they act as a cognitive frame or guiding procedure that enables the spectator to transform data and create perceptual boundaries. The basic information is present in the data, but spectators need to shape, transform, and segment it. A spectator can completely transform a 'text' by perceiving it on a different level. Moreover, the levels of narration that Branigan outlines are not mutually exclusive; spectators can interpret the data in the film on several levels simultaneously; in terms of agents of narration, the data, when interpreted as a shot, can be attributed to several agents at once. This philosophical principle should become clearer in the following two sections.
6.5. Method
From Branigan's theory of agents and levels of narration in chapter 4 of Narrative Comprehension and Film we can construct a typology of four types of shot:
1.    objective shots (not focused around the consciousness of any character within the film's diegesis [the film's narrative world]; instead, it is a shot motivated by an agent outside the film's diegesis - the narrator);
2.    externally focalized shots (shots focused, or focalized around a character's awareness of diegetic events, such as over the shoulder shots; they do not represent the character's experience, but their awareness);
3.    internally focalized shots (surface) - represent a character's visual experience of diegetic events, as in point of view (POV) shots (that is, optical POV shots; when we call a shot a POV shot in the following analysis, we mean an internally focalized shot (surface));
4.    internally focalized shots (depth) - represent a character's internal events, such as dreams and hallucinations.
From this typolog)', the analyst is able to label and identify any shot in a narrative film in terms of the agents who control it and the level(s) on which it operates.
6.6. Analysis
For the purposes of this section I shall limit myself to the first half of the film, concentrating on the distinctive characteristic of Branigan's theory and methodology - the various agents and levels of narration he outlines. The
Cognitive theories of narration     191
focus of the following analysis is therefore: What narrative agent (if any) motivates the selected images? And: What level of narration can they best be described as operating upon?
After the Madison's view the first video that has been sent to them (scene 5), the film cuts to the hallway leading up to the Madisons' bedroom. This hallway is draped by a distinctive red curtain (a characteristic feature of Lynch's films). What is the status of this shot? Is it simply a transitional shot between scenes? It seems to be a non-focalized shot - that is, a shot not controlled by any narrative agent in the film's diegesis, but controlled by an agent outside the diegesis - the narrator.
In scene 6 Fred's recounted dream consists of the following shots, which also raise intriguing questions in terms of agency and levels:
•     We see Fred walking around the house and hear Renee calling out to him; we also hear Fred's voice-over recounting the dream. All of the recounted dream shots are therefore internally focalized (depth) shots (type 4).
•     Image of fire (with exaggerated sound, rendering the fire uncanny). (Because this shot is part of the dream, it belongs to type 4; but within the dream, it is non-focalized (type 1).)
•     Fred and voice of Renee.
•     A puff of smoke rises from the stairway (as with the red curtain, smoke is another characteristic symbol in Lynch's films). This shot is coded as Fred's POV shot. In other words, within his recounted dream (level 4) we have a POV shot (level 3).
•     Fred in hallway (there is an ellipsis, since he have moved location between cuts).
•     Hallway. Again, this is a POV shot.
•     Fred.
•     Hallway and red curtain, and bedroom (coded as Fred's POV).
Here we have a repetition of the red curtain, but this time it is coded as Fred's POV. Whereas previously the shot could be read as a transitional shot, which means that it is non-focalized (objective, or belonging to the narrator), here Fred has now appropriated this image, as it is focalized around his vision and is part of his recounted dream.
With Fred still recounting or narrating the dream in voice over, the camera quickly moves towards Renee, and she screams. Fred then 'wakes up' - but this seems to be part of the dream. (This is the conclusion we reached in the 'Bordwellian' analysis of this scene.) In Branigan's terms, is this image of Fred waking up an internally focalized (depth) image (i.e. part of the recounted dream), or has Fred stopped recounting the dream? There are insufficient (or conflicting) data in the image to enable us to decide one way or the other. The voice-over has ended, and the film has returned to Fred and Renee in bed, the
192    Studying Contemporary American Film
place where Fred began narrating the dream. This suggests that Fred has stopped narrating the dream. However, there is no continuity between this shot of Renee and Fred in bed with the shot of Fred beginning to narrate the dream. This is discontinuous because both of them are now asleep, and Fred is waking from the dream. He then sees the mystery man's face imposed over Renee's face. This means that he is not only narrating the dream to Renee but is also telling her that he woke up and did not recognize her. The film is inherently ambiguous about which description is correct. Furthermore, there are no other cues in the film indicating that Fred stops narrating the dream. If the second description is correct, it means that the dream remains open-ended - we don't know when and where it ends.
The second video (first shown in scene 7) also deserves closer scrutiny. As Renee and Fred watch the second video, Renee turns to Fred and calls his name. Fred looks at the TV screen. Cut to a shot of the hallway. This is a complex shot to describe. First, it is Fred's POV shot as represented in his dream. But another narrative agent has also appropriated it - this time the agent who has made the video (the mystery man). But as Fred watches this shot on screen, it becomes his POV shot again. He is therefore watching his POV shot within his dream, now being manifest in reality via the video. There are multiple layers of agency attached to this shot (as there are with many shots; however, here the presence of the various agents becomes apparent). Perhaps part of Fred's fear is that he feels someone has got inside his head and is now reproducing his dream on video. Cut to a close-up of Fred's eyes, and then cut back to the video image, now showing the red curtain, a shot used twice before, but this time it is attributable to the mystery man (or the mystery man appropriates it from Fred, who appropriated it from the narrator). The video then shows Renee and Fred in bed.
When Fred comments in the next scene that he does not own a video camera because he prefers to remember events his own way, not necessarily the way they happened, this comment seems (as we saw in the Bordwellian analysis of this scene) to be a nodal point on which to focus the previous scenes. Yet by looking at the previous scenes more closely, through the lens of Branigan's theory, we come to realize that the video images appropriate Fred's POV shots. In other words, there is no conflict between what Fred sees and remembers and what we see on video; yet Fred's comment serves to distinguish video images from his experience.
Later we can attribute the video images to the mystery man. The fluctuating attribution of agency to these shots gives us textual evidence to link the mystery man to Fred. Furthermore, we argued that the shot ofthe red curtain, when it is first seen, is a non-focalized shot (i.e. is attributable to the narrator); on its second appearance, Fred has appropriated it; on the third occasion, the mystery man appropriates it. We can use this description to link narrator to
Cognitive theories of narration    193
Fred and the mystery man, and perhaps go even further and link the narrator with the historical director, Lynch. It is easy to make wild assertions (or hypotheses) about the relation between Fred, the mystery man, the narrator, and Lynch; what we need is textual evidence to support these hypotheses, so that we can attach or ground these assertions in the film itself.
In scene 11, Fred in the house after Andy's party, the shot ofthe red curtain is repeated for the fourth time, which could be the POV shot of an unseen agent, who then seems to confront Fred. Could it be the mystery man with his video camera recording what's going to happen in the house? (If so, then do we identify the narrator's-Lynch's camera with the mystery man's video camera?) Otherwise, it could be a non-focalized shot. Preparations for bed. Fred then looks at himself in the mirror in the same way he looked at the camera a moment ago. Renee then calls out, in the same way she did in Fred's dream. We then have a shot of the living room with two shadows (are they replacing the image of the blazing fire in Fred's dream?). Is this the mystery man following Fred with his video camera, ready to record what's going to happen next? Instead of finding out what happens next, the spectator is positioned outside a door. We then cut to the next morning, where Fred picks up the third videotape, which fills in the ellipsis of the previous scene. The dream can also be added to fill in this ellipsis, since we sec Fred approaching Renee as she sleeps in bed.
This reanalysis of key scenes and shots from the first half of Lost Highway only begins to demonstrate Branigan's theory of agents and levels of film narration. But from this short analysis, its ability to make more and finer distinctions than Bordwell's theory make it a powerful tool, particularly in analysing moments of ambiguity, in more detail and with more subtlety. The spectator's hypotheses can be formulated more clearly, and exclusive hypotheses can be related to one another more precisely (by linking each to a particular agent and level of narration). Whereas Bordwell's theory offers a methodology that reads a film as a linear or horizontal string of cues that spectators try to identify, Branigan develops a methodology that reads a film both horizontally and vertically, which enables the analyst to recognize the complexity of a individual shot or scene. A notable example from Lost Highway is the shot of the hallway in the second video, which Fred is watching. Branigan's method of analysis not only revealed the complexity of this shot, but also supplied the tools to analyse it in detail.
Endnotes
1 Noel Carroll similarly argues that: 'where we have a convincing cognitivist account, there is no point whatsoever in looking any further for a psychoanalytic account. ... For a psychoanalytic theory to reenter the debate, it must be
194    Studying Contemporary American Film
demonstrated that there is something about the data of which cognitivist (or organic) explanations can give no adequate account' (Carroll 1996b: 65). In this polemical statement, Carroll shifts the burden of proof to the psychoanalytic film theorists: they must explain why they use psychoanalysis as the starting point to theorize filmic phenomena.
2      See Bordwell (1985: 60-61) and Buckland (1995).
3      See Lynch and Cifford (1997: 11-12). At breakfast, Fred questions Renee (who is reading a book), after they have looked at the first videotape. This scene has been deleted and replaced with the single shot of the red curtain.
Realism in the photographic and digital image [Jurassic Pork and The Lost World)
Introduction
We currently see before our very eyes in contemporary Hollywood cinema a composite - a composite of the optical (or photographic) and the digital (or post-photographic) image. What is the ontological status (or 'mode of being') of this composite, that is, what is the status of its 'realism'? We shall attempt to untangle this complex issue in this chapter by reviewing the work of André Bazin, because he successfully identified the complex facets, aesthetics, and styles of realism in the cinema. We shall then apply Bazin to Spielberg's two films Jurassic Park (1993) and The Lost World (1997), focusing on their conformity to the aesthetics and style of realism as defined by Bazin. However, Bazin's theory only relates to the photographically produced image - that is, images based on optics, mechanics, and photochemistry. In the second half of this chapter we shall therefore update Bazin with the philosophy of modal logic - particularly its theory of possibility and possible worlds - to analyse the digital images in Spielberg's two films. We shall focus on the ontological status of the digitally produced dinosaurs - for they are not, of course, produced by means of optics, mechanics, and photochemistry, and so they cannot easily be analysed within the framework of Bazin's theory of realism without creating a paradox. By employing the philosophy of possible worlds we hope to overcome this paradox.
7.1. Theory: André Bazin on realism
In this section we shall consider Bazin's overlapping contributions to film history and film theory: specifically, his dialectical history of film (in which he
7