But there is a complicated and diverse history to mental images in cinema - and possibilities within that history that still beckon to artists and theorists alike. Let's take two moments from two ends of cinema history. The first is from the recently revived silent classic Our Hospitality (1923) by Buster Keaton. An inter-title tells us about the house that Buster has heard he is going to inherit. We go to a knee-to-head frame of Buster, standing still, looking blank, open in his famous way. This look of Keaton's can mean many different things in many different situations but, since there is no gag in sight right now, we suspect that it means he's thinking, or dreaming, or remembering; at any rate, his mind is working. Then the film takes us, through a slow super-imposed lap-dissolve, to the facade of a grand mansion. And then back, after a while, via another lap-dissolve, to the first shot of Buster.

What is so incredible about this sequence of shots, and so many like it in the annals of silent cinema, is how unforced, how open it is. We take it, pretty naturally, to signify Buster's mental fantasy about the house he has inherited (of course, the comic pay-off to this comes later, when we see that in reality the mansion is a run-down dump). But the pieces, the elements of this sequence seem floating, disconnected. The shot of the house looks for all the world like a strictly objective image, a documentary image. The shot of Buster has no conventional acting or mise en sc²ne indications to tell you that a man is thinking, and that we, and the film, are about to enter into his head. If you shot and cut this kind of scene exactly this way today, it would look surreal, avant garde; it would confuse some people - why are we getting a cut-away to some house between two shots of an inexpressive actor? I suspect this kind of 'primitive' scene must have looked strange and wonderful to some sophisticated folk even back then in the 1920s, since Bunuel and Dali's classic Un Chien Andalou (1928) is full of such odd connections and disconnections achieved via lap-dissolves, as if in parody of a typical melodrama of the period.

So how would such a sequence of narrative actions be staged and shot and cut today? Basically, in the most conventional cases, you would have a profusion of indicators that would leave you in no doubt whatsoever that a mental image - a flashback or fantasy imagining or hallucination - was about to occur. Going into and out of such mental images is an elaborate, rather over-coded business in popular cinema - over-coded precisely where the Keaton example seems under-coded. The actor would furrow his brow, or give a dreamy look. The camera would zoom or dolly into his face. Music would begin, or the soundscape would somehow thin out and become unrealistic, abstract. If there was a prop in this scene, some object that triggers the mental-image, the film would insist on the cause-and-effect relation between this object and the thinking man, by cutting back and forth between them, and matching the zoom-in movement on both. There might well be some optical distortion that separates the mental image from its bordering realist images: the image of the object goes out of focus, or the screen is washed with colour, or slow-motion enters. And then after the mental image, all of these indicators would change back: the normal sound ambience would slowly return, the camera would pull back again, the image would resume its normal look and texture and speed, and the actor would re-compose himself, ready for the next action.

In the annals of 1970s film theory, one read a lot about what were called 'shifters'. Shifters were verbal, literary indicators of narrative placement - things like pronouns and tenses. In the late '70s and early '80s, there was a small international movement in world cinema dubbed the 'new talkies'. It was a movement exploring experimental narrative - not straight narrative, and not straight avant garde abstraction, either. The stars of this movement included filmmakers such as Yvonne Rainer, Mark Rappaport, Marguerite Duras and Alexander Kluge. Their films were called the 'new talkies', and not even sarcastically, because they were indeed very wordy films - full, particularly, of voice-over texts. But these wall-to-wall texts were wildly shifty - the speaking voice goes from I to you to he to she to they; the events described by this voice swing madly in their locution and location from past to present to future.

And the exact identity of this speaker was always in doubt, always vacillating, the exact status of this speaking personage quite undecidable: at one moment an omniscient narrator, at another a character in the fiction, at yet another a mere quotation or citation from some unnamed archive of social-cultural texts. It was fashionable and thrilling for a while, in those days, to pluck out examples of this shifty work in independent film and ask pointedly, in the classroom or at the seminar podium: who is speaking here? Or even more tersely: who speaks? And the correct answer was usually: everyone and no one, all at once.

Strangely, under the verbal and intellectual spell of the new talkie, no one much thought to ask a logical supplementary question: who sees? From which head has this image been projected, whose eyes claim it within the purview of their vision? I believe this to be a fundamental question in cinema. Vision is as ambiguous in its shifts and nuances as anything spoken or heard on a soundtrack, and these images enter into some extraordinarily devious and dancing relations with the natural and contrived ambiguities of sound.

Consider this case study in the fundamental narrational ambiguity of cinema. We hear a lot these days about the mythic and therapeutic power of storytelling - telling the story of your life to find your self, to find your way. We have seen a fad, in popular-classy literature, for novels that are mosaics of many personal life-stories: The Joy Luck Club (1994) and How to Make an American Quilt (1995), for instance, where you get ritual ceremonies of storytelling, around a mahjong game or a quilting bee, if not exactly a scouting campfire. In literature, on the page, this mosaic is easily achieved. It's simple and direct for a novel to be a tapestry of first-person stories: you can have both the tale and the teller, the narrative and the narration, the information and the inflected quality of a personal voice, all equally evident at once. It's easy to maintain the illusion (and sometimes it is not even an illusion) that each story belongs to its narrator.

But in a movie, to whom does a story belong? Who, which character in the fiction, can ever lay claim over his or her story? Images can contradict voice-over, and gallop away into their own fuzzy space between subjective and objective realms. The Australian film What I Have Written (1996), adapted by director John Hughes from the novel by John Scott, prompts me to ponder these problems, these conditions. Even the title, What I Have Written, does not really work well as the title of a film. In the novel, the literary conceit is perfect. The entire story revolves around a manuscript, a duplicitous manuscript, left by a character who is lying in hospital dying. This manuscript begins the book; it is there for you to read, as if it were the actual material object left behind by the character. It sits there as evidence, as fact, as a text. What he has written is what you read. Now, Hughes' film does not simply film the pages of this manuscript and the words on the pages: that would have been far too avant garde. Hughes keeps the words of the manuscript as a voice-over, and he adds to them pictures, images - still, sometimes blurred images that eventually, across the film, nudge their way into movement.

Where do these images come from? Hughes could have invented a filmic conceit to match Scott's literary conceit: these could have been, at a stretch, images somehow left behind by the character, polaroid snapshots or some kind of video document, even a reel of film. There are of course quite a few films constructed around audio-visual documents that are left behind, or placed in hiding, by fictional characters, documents that are unveiled at a high moment of drama or revelation, as in Crystal Gazing (1982), Benny's Video (1993), Friendship's Death (1990) Ulysses' Gaze (1995) or Irma Vep (1996). Sometimes these films only go to prove that it is better, finally, to keep this magical audio-visual document veiled rather than reveal it, because often the moment of revelation is a let-down, not even half of what you imagined it would be: that's why Jacques Rivette keeps a certain trembling, troubling painting completely hidden from our gaze for the entire four hours of his film about art and the artistic process, La Belle Noiseuse (1991).

But Hughes does not go down this road. He presents the manuscript as a voice and a set of images: words and images from nowhere, authorised or claimed by no one. It's not neat, but it's interesting and intriguing, because it makes you think of how completely unmoored, how completely unhinged at a fundamental level, every illustrated voice-over narration in a movie really is. The screen narrative becomes a paradoxical thing: its images belong neither to the subjective regime of the character telling his or her story, nor the objective regime of the filmmaker telling that story from a more distant perspective, as is the classical manner. So, in a way, we are in both spaces at once, the subjective and the objective.

There are films which exaggerate this blurry double-effect, this strange superimposition, this fuzzy border of subjective and objective regimes or levels. The films of Alain Robbe-Grillet, of Marguerite Duras, of Raul Ruiz. I will never forget the shock of first seeing Duras' film India Song (1975). Very early on, there are objective, wide angle views of rooms, objects, decadent people lounging about in a group - and we approach this interior space, in the conventional manner, through a a successive focalising, a passage from establishing exterior views, to the interior ones. And there is a first-person narration, a voice, a text that insists heavily on its 'I', that seems to be taking us, guiding us into this space. But suddenly there is a shot, excessively close, almost barbaric, quite rupturing in its effect, of a woman's bared breast, its nipple pointed to the ceiling. Whose eye is seeing this shock image, which consciousness is giving us this shock cut? From that moment on, India Song is a veritable carnival of shifters: different voices and points-of-view and cues of narrative direction and narrative sense rise and die in a constant, impossible warring, and you never know where you are with the film.

The films of Ruiz, like Pasolini's Arabian Nights (1974), are the modernist version of the mosaic-quilt-mahjong-campfire multi-narratives so popular today. They do not resist narrative, or drain it with minimalism, or detonate it or destabilise it in quite that Durasian, deconstructive way. Ruiz multiplies the stories, stories within stories, until you no longer know who is telling, who is remembering, who is speaking. The fictional voices, the claimants, roam all over the narrative map, grabbing whatever piece of the territory they can wield or manage.

A corresponding set of narrative and stylistic perturbations happened, once upon a time, in mainstream films. Cinema, at many of its levels, went crazy in the 1960s. Suddenly, in art house and popular movies alike, the conventions governing the presentation of mental images were stripped bare. No more wavery lines and soppy music cues and neat book-ended shots of a thinking or gazing human face; not for a while, anyway. English-language films from Accident (1966) and Two For the Road (1967) to Bob Fosse's All That Jazz (1979) used every kind of short-cut and short-circuit and daring leap imaginable in the depiction of memories and dreams. Images flashed up in all their ambiguity - some of that ambiguity cleared up later in the plot, retrospectively, the way Atom Egoyan does it in Exotica (1994); some of the ambiguity remaining for good. The difference in status between a fantasy and a memory was blurred; so too the difference between a flashback and that pretty rare commodity, a true '60s invention, the flash-forward. All bases ensuring immediate comprehension, immediate legibility, were scrambled.

The inauguration of this history in the '60s can be fairly safely attributed to Resnais' and Robbe-Grillet's Last Year in Marienbad (1961), a film whose narrative, in its successive variations, is purely speculative, a set of experimental possibilities that are like the splintered shards of some lost, coherent plot. Robbe-Grillet took this game even further in his own subsequent films, such as L'Immortelle (1963) and The Man Who Lies (1968), the latter being a movie that generates devious dream-sequences, devious flashbacks, devious waking fantasies, the whole devious works; and Resnais returned to this turf in his great film Providence (1977).

In a prescient article written on the release of Last Year in Marienbad and published in Sight and Sound, the merry surrealist Jacques Brunius wrote that it was a film for which he and his comrades had been waiting all their lives: a film that presented the perfect picture of inner mental processes. How we think, how we remember, how we conjure, how we fantasize - the real-time flow of all that in our heads, complete with all the lived diversions and interruptions and super-impositions that go on - for Brunius, Last Year in Marienbad provided the analogue of all that.

There was something new and jolly and progressive in what Brunius said and celebrated - and also something instantly, deadeningly conservative, tame and predictable. The notion that these crazy films of the 60s showed what went on inside a person's head was, finally, a way of coding the action of these films, of constraining their dynamism and ambiguity, a way of anchoring and regulating their unnerving mysteries and sweet confusions. What Brunius said in 1961 is still being recycled 36 years later with boring regularity, by the likes of Philip Adams when he trots out his eulogy for Dennis Potter, and especially the '80s TV series The Singing Detective. How do we get outside these all-encompassing heads that contain all thoughts, these handy, fictitious eyes that see and guide and channel every apparition?

The fact is that the loony, skipping-around films of the '60s were not always so constrained by the laws and logics - however expanded these laws and logics - of an individual, subjective consciousness. They were sometimes, often, about characters who remembered and dreamt and lived particularly vivid inner, imaginative lives; but they were also great, hulking, formal contraptions, veritable time-machines, launched off the springboard, the pretext, the artifice of a central, ruminating character in crisis. From the trippy, psychedelic films of that bygone era to Sergio Leone's Once Upon a Time in America (1984), cinema shuttled back and forth on these corroded rails of subjectivity. Dream-like films certainly, but not simply dream-films, because their journeys are tense and fraught, not just dreamy: the anguish they can sometimes generate comes from this tension between levels, this friction of gears between a scary, fragmented world, and the fragmented consciousness that tries in vain to apprehend and compose it.

Today, we too can use these giddily disrupted and disrupting popular movies of the '60s as our own time machine. Using them, staying open to their ambiguities and multiple paths and wilfully false or confusing trails, we can travel back to those primal moments of under-coded mayhem in the silent days of Keaton's Our Hospitality - and we can recapture the shock, the birth of possibilities, the poetic delirium, that are present and flowering in such moments. And from there, immersed in that primal experience, we can see and grasp almost any passage of mental imagery in cinema - any flashback, projection or memory, however conventionally presented - as tattered and broken, heterogeneous scraps fighting towards some pitiful, homogenised sense.

Cinema liberates itself - and our idea of it - when it opens itself to such ruination of sense and linkage. At such moments, cinema refinds its truly poetic vocation. Think of the wonderful and haunting beginning of Clint Eastwood's A Perfect World (1994): a series of shots, free and completely disconnected from the story yet to come, of objects floating in the sky, drifting to the ground; and a man (Kevin Costner) in some secret, mysterious ceremony of voluptuous surrender and falling, his body and identity subject to the grace of a strange and unfathomable oblivion. What a day for a daydream... but a daydream like those 'eternal moments' that flash up inexplicably in cinema, the kind of surreal dream-images eulogised once upon a time by Jean Epstein, pregnant with a hundred stories like a good Cindy Sherman 'movie-still' (stories of guilt, death, punishment, desire, regression, redemption), suggesting them all at once, presenting them fan-like, and yet compelled to follow-through on none of them in particular. This is an associative cinema, animated and driven forth by the merest flicker or hint or echo of a connection (logical or lyrical or phantasmatic) between discrete units or elements as they flare up or swim through the time and space of the frame.

But it is not just a matter, ultimately, of isolated moments of rupture or confusion or poesis, 'special effects' of editing or optical treatments; as Bunuel's films (such as Tristana, 1970) show us, what begins as a point of aesthetic 'transgression' (like a shock cut or insert) can slowly expand and extend like a contagion, infecting and destablising whole plateaux of narrative, character and meaning. Back, today, in the realm of popular cinema (or its margins), we find this kind of Bunuelian contagion at work in the films of David Cronenberg (Crash, 1996) or David Lynch (Lost Highway, 1997). What begins as the pristine interiority of a seeming mental image works its way up and out and across, transforming the depicted material world, with its objects and bodies, into a kind of responsive, sensual sponge, expanding and contracting under the influence of primal, unconscious forces and fancies.

Jane Campion is another mistress of this cinematic contagion. In The Piano (1993), Campion stages an absolutely stunning moment - reminiscent of a key image in Hitchcock's profoundly dream-infected Vertigo (1958) - where the camera approaches Holly Hunter from behind, as she dangles her hand in and odd manner, to close in on the tight bun of her hair; this then cuts to a wild, unruly thicket of trees. It is as if this charged, fetishistic 'nodal point' offers a gateway to the unconscious itself, into which the film eagerly plunges. When a film emerges from such dives - if it knows, even if only unconsciously or intuitively, how to stay on the waves and off the rails - the world it creates and the world it offers us to experience will be utterly touched, transformed and enriched. And the condensing, pin-point moments of the mental image or flash will be (in the words of the Surrealist poet P³tr Kral) "simple preludes to these magnificent orgies".

END NOTE
This is the first part of an on-going project devoted to mental imagery in cinema. Part Two, "The Ever Tested Limit", is forthcoming in a publication sponsored by the Centre for Contemporary Photography.