Politicizing Eye tracking Studies of Film – William Brown


This essay puts eye tracking studies of cinema into contact with film theory, or what I term film-philosophy, so as to distinguish film theory from specifically cognitive film theory. Looking at the concept of attention, the essay explains how winning and keeping viewers’ attention in a synchronous fashion is understood by eye tracking studies of cinema as key to success in filmmaking, while film-philosophy considers the winning and keeping of attention by cinema to be a political issue driven by economics and underscored by issues of control. As such, film-philosophy understands cinema as political, even if eye tracking studies of film tend to avoid engagement in political debate. Nonetheless, the essay identifies political dimensions in eye tracking film studies: the legitimization of the approach, its emphasis on mainstream cinema as an object of study and its emphasis on statistical significance all potentially have political connotations/ramifications. Invoking the concept of cinephilia, the essay then suggests that idiosyncratic viewer responses, as well as films that do not synchronously capture attention, might yield important results/play an important role in life in an attention-driven society.

In this essay, I wish to put eye tracking studies of film into dialogue with a more political approach to film, drawn from film theory, or what, for the benefit of distinguishing film theory from cognitive film theory, I shall term film-philosophy. In doing so, I shall draw out what for film-philosophy are some of the limitations of eye tracking, including its emphasis on statistical significance, or what most viewers look at when they watch films. I shall argue that we might learn as much, if not more, about cinema by paying attention not only to statistically significant and shared responses to films (what most viewers look at), but also to those viewers whose responses to a film do not form part of the statistically significant group, and/or to films that may not induce in viewers statistically significant and shared responses. In effect, we may find that there are insights to be derived from those who look at the margins of the cinematic image, rather than at the centre, even if those viewers are themselves ‘marginal’ in the sense that they are pushed to the margins of most/all eye tracking studies of film viewers. There is perhaps also value to be found in looking at ‘marginal’ films. In this way, we might find that idiosyncratic responses to a film or films is as important as the shared response. I shall also argue that there is a politics to the idiosyncratic response, especially when it is put into dialogue with film theoretical/film-philosophical work on cinephilia, and that as a result there is also a politics to eye tracking and its emphasis on statistical significance. I shall start, however, by looking at the state of eye tracking film research today.

On 29 and 30 July 2014, the Academy of Motion Picture Arts and Sciences (AMPAS) – the same American academy that distributes the so-called Oscars – held two events under the combined title of ‘Movies in Your Brain: The Science of Cinematic Perception’. The events included contributions from neuroscientists Uri Hasson, Talma Hendler and Jeffrey M. Zacks, psychologist James E. Cutting, directors Darren Aronofsky and Jon Favreau, editor Walter Murch and writer-producer Ari Handel. The host of the first evening was psychologist Tim J. Smith, whose eye tracking studies of cinema have arguably become the best known and most influential over recent years (see, inter alia, Smith 2012a; Smith 2013; Smith 2014). Through these events, as well as through coverage of these events in fashionable journals like Wired (Miller 2014a; Miller 2014b), we can see how eye tracking – together with the study of film using brain scanning technologies such as functional Magnetic Resonance Imaging (fMRI) – is clearly becoming important for our understanding of how films work. This in turn means that such studies are surely important to film studies.

For a detailed history and overview of eye tracking, explaining how it works and what it tells us about film, I cannot do better than to guide readers to the afore-mentioned work by Smith. Smith has soundly demonstrated, and with great clarity, how the human eye moves via small movements called saccades, and that in between saccades the human eye fixates. It is during fixations that humans take in visual information, with fixations being linked therefore to attention and to working memory; we tend to remember objects from our visual field upon which we have fixated, or to which we have paid attention. Clearly this is important to the study of film, since viewers typically attend only to parts of the movie screen at any given time, and not necessarily to others or to the whole of the screen (and the surrounding auditorium). Can/do filmmakers exert influence over where we look, for how long, and thus what we remember about a film – with those memories themselves lasting for greater or lesser periods of time? And if filmmakers do influence such things, how much influence do they exert and through which techniques? These are the questions that eye tracking technology can help to answer – and scholars like Smith do so with great skill and eloquence.

My aim, however, is not simply to reproduce findings by Smith and others who have used eye tracking devices to study film. In order to construct a theoretical argument concerning the importance of the idiosyncratic, or ‘cinephilic’, response to a film or films in general, as well as the importance of a filmmaker not necessarily ‘controlling’ where a viewer looks, but instead allowing/encouraging viewers precisely to look idiosyncratically, cinephilically, or where they wish, I need instead to bring the scientific and ‘apolitical’ use of eye tracking devices into a political discourse concerning the nature of cinema, power, hegemony and the issue of cinematic homogeneity and/or heterogeneity. This is a controversial maneuver – in that it will bring together two areas of film studies that often seem to stand in ‘opposition’ to each other, namely cognitive film theory and a film theory that still plies its trade using Continental philosophy, or what for the sake of simplicity I shall term film-philosophy. My desire is not simply to be controversial, however. Rather it is to engage with what eye tracking means to film studies, both currently and potentially in the future.

To begin to bring eye tracking studies of film into the ‘political discourse’ mentioned above, I shall relate an anecdote. A semi-regular response from colleagues in film studies, when I tell them about eye tracking studies of film viewing, is that eye tracking doesn’t tell us anything about films that we didn’t already know. Is it a surprise that we tend to look more often at the center of the screen? Is it a surprise that we typically attend more to brightly illuminated parts of the screen than to dimly lit ones? Is it a surprise that we tend to direct our attention towards human faces when watching a film that features human characters? Anyone who has consciously thought about what they do while watching a film will be able to tell from memory alone that these things are all true. As a result, eye tracking studies of film can sometimes be filled with what, at least to the film student/scholar, are truisms. By way of an example, Paul Marchant and colleagues say that ‘these strategies and techniques… [capture] the audience’s visual attention: focus, camera movement, eye line match, color and contrast, motion of elements within the shot, graphic matching’ (Marchant et al. 2009, 158). On my print-out of Marchant et al.’s essay, my own apostil next to this assertion reads as follows: ‘Do we not know this already (otherwise cinema would not have developed these techniques)?’ Many, if not all, film viewers will know simply from experience that these techniques help to guide their attention, even if they are blissfully unaware of the relationship between eye fixations, attention and memory. Of course, it is pleasing to have our introspective responses to/our intuitive knowledge about cinema ‘scientifically’ confirmed (to a large extent, but not entirely – about which, more later); but essentially, so my colleagues’ argument goes, eye tracking studies tell us what we already know.

Now, even if I myself find some eye tracking studies of film to be ‘truistic’, I nonetheless believe that eye tracking studies of film are of great importance. However, their importance is perhaps in playing a role that is different from the one that eye tracking studies of film seem to give to themselves, which is as a key component of cognitive film theory. Instead, I think that eye tracking studies of film are important for film theory, or what today is termed film-philosophy. I shall explain the distinction between cognitive film theory and film-philosophy presently.

Little in this world is uniform, and so by definition I generalize when I say that the basic tenet of cognitive film theory is, with David Bordwell and Noël Carroll’s Post Theory: Reconstructing Film Studies (1996) serving as its figurehead, for film studies to move towards a theory of cinema based on the analysis of films themselves, and away from a film theory that uses cinema as a means of confirming or denying a Lacanian understanding of the human and/or an Althusserian/Marxist conception of contemporary capital. In spite of cognitive film theory’s lack of uniformity, eye tracking studies of film are nonetheless part of cognitive film theory’s project to help us to look at cinema ‘as it is’, and not to use cinema as a political football. Conversely, film-philosophy is in general informed by the kinds of Continental philosophers, often though not limited to Gilles Deleuze, that cognitive film theorists reject, and it engages not just with films ‘as they are’, but with the politics of films.

Now, to claim that we can isolate films and film viewing from a human world that is perhaps always political, and to claim that we can then analyse films ‘as they are’, is perhaps absurd: films ‘as they are’ are part of a political world, and cognitive film theorists are not unaware of this, just as film-philosophers are not incapable of scientific analysis. However, how much politics is allowed into the analysis of films perhaps informs the broad distinction between cognitive film theory and film-philosophy, as I hope to clarify by looking briefly at the role of attention in the work of two scholars, Tim J. Smith and Jonathan Beller. In his ‘Attentional Theory of Cinematic Continuity’ (AToCC), Smith (2012a) uses eye tracking studies to demonstrate how filmmakers capture and maintain viewers’ attention, with certain techniques, mainly those associated with continuity editing, being more successful than others. Meanwhile, in his Cinematic Mode of Production: Attention Economy and the Society of the Spectacle, Beller (2006) suggests that capturing attention is not necessarily an aesthetic, but rather a political project: the more attention a film garners, the more success one will have in monetizing that film, with the making of money becoming the bottom line of cinema. Beller does not appeal to some early cinema that did not attempt to elicit viewers’ attention and thus make money; such an early cinema did not necessarily exist. Rather, Beller argues that cinema has always been part of an economy that is based on attention; indeed, cinema plays a key role in naturalizing this attention economy, meaning that cinema has not always been necessarily capitalist, but that the capitalist world endeavors as much as possible to become cinematic, to capture our attention as much as possible in order to ‘win’ the economic race, since capturing eyeballs means making money. Smith explains how attention is captured; Beller offers an explanation as to why. Even though filmmakers rely on natural processes in order to capture attention (Smith), the process of consistently trying to capture our attention (‘cinema’) is not natural, but political and economic (Beller).

James E. Cutting, in commenting on an earlier draft of this paper, says that the results of eye tracking studies of film, which reveal how filmmakers capture attention, are

big news… because almost nothing else does this – not static pictures (photographs, artworks), not class room behavior by teachers, not leaders of business meetings, and often not even spectacles of various kinds (sporting events, rock concerts, etc); even TV is typically not as good as the average narrative, popular movie. (Cutting, signed peer review 2014)

If cinema is indeed better at capturing our attention than these other media, and if in some senses it is better at capturing our attention than those parts of the world that do not feature such media – i.e. if cinema is better at capturing our attention than reality – then cinema, and the making-cinematic of reality in a bid to capture attention, to make money and/or to influence people (Cutting compares cinema in particular to teachers and to business leaders) is profoundly political. It is profoundly political because learning about how to capture attention – learning about how cinema works – is tied to the shaping of our material reality (putting screens everywhere) and to controlling attention (encouraging us to look at those screens, and not at the rest of reality). Cognitive film theory is apolitical; film-philosophy, meanwhile, engages in the very political dimensions of cinema. Eye tracking studies of film tend to position themselves as part of the former; my aim here is to bring them into dialogue with the latter.

If eye tracking studies of film tend to position themselves as part of a would-be apolitical approach to cinema, then in their investigation into cinema, they are nonetheless conducting an investigation into politics, as per Beller’s equation of cinema with politics highlighted above. However, while eye tracking studies of film position themselves as apolitical, politics do creep into eye tracking studies, especially through what I shall call their absences. What is more, these politics do relate to film-philosophy’s ‘political’ approach to film. In order to demonstrate this, I shall begin by analyzing how eye tracking studies of film have sought historically to legitimate themselves.

Early in an essay that gives an overview of eye tracking studies of film, Smith asserts, without naming any, that the hypotheses of film theory ‘generally remain untested’ (Smith 2013, 165). In this almost throwaway comment, we perhaps find important information. For in asserting that eye tracking is what can help us to ‘test’ out some theories of film, as Smith goes on to do in relation to Sergei M. Eisenstein’s writing about his own film, Alexander Nevsky (USSR, 1938), he perhaps overlooks how film theorists often (but perhaps not always) try (though not always with success) to construct their theories based on the films that they have seen, studied and perhaps made, and not the other way around. That is, Smith seems not to consider that watching films is itself a means of testing our theories about films – without the need for eye tracking devices. On a related note, while he does consider filmmakers like Eisenstein, D.W. Griffith, Edward Dmytryk and others as ‘experimentalists’ of sorts (who have tested their own theories), Smith also does not fully acknowledge that the history of cinema can itself be seen as a prolonged ‘test’ in what ‘works’ or ‘does not work’ with audiences – with that which ‘works’ being regularly adopted as either a short- or a long-term strategy by the film industry, be that in terms of re-using storylines, adopting a specific cinematic style, employing bankable film stars, using topical settings, engaging with zeitgeist themes and so on. Instead, it is Smith’s intervention that will validate or otherwise that history of theory and practice, and which will confirm what filmmakers, and perhaps also many audience members, have probably known for a long time, even if putting their knowledge into practice sometimes proves harder than we might imagine (because otherwise films would presumably not have ‘mistakes’ in them).

Now, it’s natural that a (relatively) new approach to studying film would need to legitimize itself in order to gain credibility and following – and Smith clearly charts the c30 year trajectory of eye tracking in film studies since the 1980s onwards (Smith 2014: 90). Nonetheless, if the history of cinema is not ‘test’ enough for Smith, then implicitly a claim is being made here about what constitutes a ‘real’ test, and, by extension, what sort of person can carry out a ‘real’ test. In other words, eye tracking, and the cognitive framework more generally, here legitimizes itself as being a tool for verifying (scientifically) what previously were ‘mere’ and speculative theories (these are my terms) – with the people qualified to carry out these tests being neither filmmakers nor audience members, but psychologists. By justifying eye tracking in this way, Smith is not just making a statement of fact (eye tracking demonstrates that viewers look at the same things at the same time during films made using the continuity editing style), but he is also – I assume unintentionally – making an implicit value judgment that carries political assumptions regarding what constitutes a/the most legitimate framework for learning and knowing about film. If, as per my anecdote above, I can and do know the same things via introspection that eye tracking tells me, then why is introspection not equally legitimate as a framework, even if the former involves less visible labor, and certainly less sexy imagery, and thus does not seem to involve any real ‘testing’?

Eye tracking thus seeks ‘politically’ to legitimate itself as a tool for film analysis. To be clear: eye tracking is legitimate, but it is also always already making claims about what constitutes knowledge: introspection is not knowledge, while science is – even if both can lead to the same understanding. Importantly, in producing visible evidence (the afore-mentioned ‘sexy imagery’ of colored clusters of eye-gaze on scenes from films), then eye tracking studies are also always already cinematic, by which I mean to say that they affirm a system whereby the visual/the cinematic (here are pictures of attention being captured) are validated above invisible (here, introspective) approaches to the same knowledge. This in turn always already affirms the process of cinema and attention-grabbing as being the (political) system that is most powerful.

If eye tracking affirms a politically cinematic world, in that cinematic forms of knowledge are more valid than invisible, i.e. uncinematic, ones, then within that cinematic world eye tracking might also, and in some respects implicitly does, legitimate some forms of cinema over others. This is suggested by the way in which eye tracking studies look predominantly at Hollywood/mainstream cinema in their analyses of film. For example, in his AToCC, Smith (2012a) cites a diverse range of movies, including L’escamotage d’une dame au théâtre Robert Houdin/The Vanishing Lady (Georges Méliès, France, 1896) and L’année dernière à Marienbad/Last Year at Marienbad (Alain Resnais, France/Italy, 1961), but eye tracking data are given mainly for contemporary Hollywood films, including Blade Runner (Ridley Scott, USA/Hong Kong/UK, 1982), Requiem for a Dream (Darren Aronofsky, USA, 2000) and There Will Be Blood (Paul Thomas Anderson, USA, 2007), with Smith suggesting that continuity editing is the form of cinema best suited to capturing attention.1

The absence of eye tracking data on those other, non-Hollywood films is perhaps telling, as suggested by two respondents to Smith’s essay, who query how his theories would apply to different cinemas, including the avant garde (Freeland 2012, 40-41) and, at least by implication, Japanese cinema (Rogers 2012, 47-48). Eye tracking would of course yield important insights into avant-garde and other forms of cinema, but that information is not offered here.

Furthermore, Smith’s suggestion that continuity editing is the form best suited to capturing attention, also prompts Paul Messaris and Greg M. Smith to argue that continuity editing violations, in particular jump cuts, are quite regular and not particularly detrimental to the continuity of the film viewing experience (Messaris 2012, 28-29; Greg M. Smith 2012, 57). Malcolm Turvey, meanwhile, argues that the film viewing experience is always continuous, meaning that the ‘continuity’ of continuity editing ‘is not continuity of viewer attention per se… but rather the manner in which films engage and manage that attention’ (Turvey 2012, 52-53; for Smith’s riposte to these responses and more, see Smith 2012b).

These responses highlight how filmmaking ‘perfection’ (an absence of continuity errors) need not be fetishized too much; audiences are quite happy to watch films with continuity errors (many of which they will not notice). Furthermore, many audiences love what Jeffrey Sconce (1995) might term ‘paracinema’ – i.e. ‘trash’ cinema and ‘bad’ movies – be they intentionally ‘bad’ or otherwise. In other words, it would seem that as long as audiences are primed regarding how they should receive a film (or, in Turvey’s language, as long as their attention is managed and then engaged in the right way), then you don’t need to care about and can even love the stylised acting, the ropey mise-en-scène, the unmotivated camera movements, the strange edits and the story loopholes of, say, The Room (Tommy Wiseau, USA, 2003), supposedly the worst film in history. Under the right circumstances (with the right management/ preparation), it would seem that audiences can like pretty much anything, including a 485-minute film of the Empire State Building (Empire, Andy Warhol, USA, 1964). In other words, while in his AToCC Smith mentions Méliès and Resnais, and while he engages with Eisenstein and other filmmakers elsewhere, the AToCC puts an emphasis on mainstream Hollywood cinema and its predominant system of continuity editing, since this cinema elicits a synchronicity of response, or control over attention, in that viewers attend to the same parts of the screen at the same time – while also often failing to detect edits done in the continuity editing style (see Smith and Henderson 2008). There is a seeming bias here towards mainstream, narrative filmmaking, the engrossing nature of which is lauded at the expense of other cinemas.

Let us move away from Smith in order to demonstrate how this bias is not his alone. Jennifer Treuting suggests that ‘[t]he use of eye tracking… can help filmmakers and other visual artists refine their craft’ (Treuting 2006: 31). In some respects, this is an innocent comment; I have no doubt that eye tracking can help filmmakers and other visual artists to refine their craft. But suggested in this ‘refinement’ is also the move towards validating the mainstream/continuity style at the expense of its alternatives. A combined eye tracking and fMRI study carried out by Uri Hasson and colleagues also makes this clear: much fuss is made over how work by Alfred Hitchcock elicits greater synchrony (‘inter-subject correlation’) in viewers than does an ‘unstructured’ shot of a concert in Washington Square Park, a film that is simply a ‘point of reference’ and which ‘fails to direct viewers’ gaze’ (Hasson et al. 2008, 13-14; emphasis added). My reference above to Warhol’s Empire here becomes apposite: what Hasson and colleagues dismiss as a ‘point of reference’ and as a ‘failure’ in various respects defines one of the great experimental films. Perhaps ‘marginal’ films like Empire should also be considered successful – but at achieving something different to the work of Hitchcock, and perhaps Hasson’s film is not a ‘point of reference’, but an experimental work that equally inhabits the totality of films in the world that we shall call cinema.

If Hitchcock ‘succeeds’ in controlling viewers’ attention, while Warhol by implication ‘fails’, then eye tracking becomes implicitly/inevitably embroiled in not just what film is, but in what film could or should be – as Treuting’s suggestion that eye tracking might feed back into filmmaking also makes clear. This suggests that there is a politics to eye tracking film studies, particularly in the UK where universities are increasingly relying on ‘impact’, particularly on the economy, in order to survive: they don’t just observe films, but feed back into how films are, or should be, made, by exploring what is ‘successful’ in terms of eliciting attention, getting bums on seats and thus making money. In some respects, eye tracking in particular and cognitive film theory in general are now dragged back towards the Marxist approach to cinema that cognitive film theory initially sought to reject: it, too, shapes/seeks to shape cinema just as Marxist film theory in effect lobbied for alternatives to the mainstream. However, where Marxist film theory lobbied for a rejection of mainstream cinematic techniques, eye tracking studies seem to validate them – and to suggest that filmmakers might ‘refine their craft’ by adopting/intensifying them. Saving the thorny issue of ‘control’ and ‘influence’ for later, there is still a political dimension to this potential validation of mainstream cinema techniques, because it reaffirms the economic hegemony of one style over others and it also validates in some degree a homogeneity of product (and of audience?) – all within a ‘cinematic’ economic system that is itself predicated upon gaining attention. Cinema is both business and art, but if art is one thing it is unique/different, and so a move towards homogeneity is a move towards the reduction of art in favor of business. If it requires an artist rather than an academic to make this clear, then Darren Aronofsky’s apprehensive response to Hasson’s work at the AMPAS events hopefully serves this purpose: ‘“It’s a scary tool for the studios to have,” Aronofsky said. “Soon they’ll do test screenings with people in MRIs.” The audience laughed, but it didn’t seem like he was joking, at least not entirely’ (Miller 2014b).

I have so far argued that cinema is political, that eye tracking studies have required some political maneuvering in order to legitimate themselves, and that the focus on continuity editing/mainstream cinema by eye tracking studies may also have a political dimension. However, are eye tracking studies themselves without methodological politics, in that they simply report findings? I wish presently to suggest that eye tracking research does have methodological limitations – which is why I asserted above that eye tracking film studies are only to a large extent and not entirely reliable – and that these limitations also have a political dimension. The methodological limitations are not simply a case of potential inaccuracies regarding the type of eye-tracker used, determining how long the eye needs to be still for a fixation to take place, what algorithm is used to measure this, or how accurate is the eye-tracker in determining where exactly the eye is looking – all ongoing issues with eye tracking technologies (see, inter alia, Wass et al. 2013; Saez de Urabain et al. 2014). It is also a case of issues of statistical significance and the politics thereof, particularly what I shall call the temporal politics, and to a lesser extent the social politics, of eye tracking. In relation to the latter, many eye tracking studies involve students in order to carry out their research (e.g. Tatler et al. 2010; Võ et al. 2012). As a result, the findings might pertain not universally, but to population members who are of a certain age and, if we can say that university students tend to be from more affluent backgrounds, a certain socioeconomic status. In relation to statistical significance, meanwhile, all studies tend to discount those viewers who do not look where the researchers want them to look; for example, in a study of where people look when viewing moving faces, only 87 per cent of fixations targeted the face region when shown a moving face with sound, with that figure dropping to 82 per cent when shown a moving face without sound (Võ et al. 2012, 7). Of course, when what one is investigating is where people look when they look at faces, it is correct to discount those 13-18 per cent of fixations that were not directed at the face. But the point is that similar discounts happen all the time, not least in the process of averaging that we see in various experiments, including those mentioned by Marchant et al., Hasson et al., and Smith. And yet, where neuroscience is based in large part upon the study of anomalous brains – from autists to damage sufferers to perceived geniuses – psychologists engaged in eye tracking tend to go with force majeure and report the average, or what most people do. There may however be in human populations a ‘long tail’ (to use the terminology of Chris Anderson, 2006) that may not in any one experiment be statistically significant, but which over a number of experiments might begin to show patterns that could help us to understand vision and attention in a more ‘holistic’ fashion.

To continue by way of another anecdote: a film scholar took part in an eye tracking film study at a leading European university. Upon completion, the colleague conducting the study told the scholar that they looked in completely different places – generally at the margins of the screen – to where most of the other participants looked, and that their participation was therefore useless to the study. If we can say that the film scholar looked (perhaps deliberately) where others do not look, then to what degree is film viewing a matter of, to use Turvey’s language, management and engagement? That is, do film scholars look differently at films, perhaps even at the world? And if so, what can we make of this?

The Russian ‘godfather’ of eye tracking studies, Alfred Yarbus. famously published in the 1960s that setting viewers different tasks will completely modify where they look at an image (Yarbus 1967; see also Tatler et al. 2010). There is much to extrapolate from this. For while eye tracking studies will use terms like ‘naïve’ to define how participants are unaware of the aims of the study, when it comes to film viewing, humans are rarely naïve at all. Advertising, reviews and other publicity materials are always – at least on an implicit level – telling us how and where to look at films, just as the media and our conspecifics are telling us how and where to look in the real world. Now, it may well be that humans who have never before seen a movie have little trouble understanding Hollywood cinema, as affirmed, inter alia, by both Messaris (2012, 31-33) and Smith (2012b, 74). Nonetheless, our attention is not just managed and engaged in the cinema, but it is also managed and engaged for the cinema, and I have not read any studies where psychologists showed a non-Hollywood film to first-time audiences and in which those audiences had trouble understanding the film; that is, these studies affirm nothing about the comprehension of continuity editing per se, although they might affirm that humans can understand cinema without training – as is presumably affirmed worldwide everyday as the first film shown to children is not a Hollywood film but a Bollywood, Nollywood, Filipino, Chinese or other movie; what is more, the studies perhaps only affirm the cultural hegemony enjoyed by Hollywood, in that psychologists present a Hollywood and not another film to those first-time viewers – and then use that research to affirm Hollywood’s economic primacy as being a result of its filmmaking style and not also as a result of historical and other factors. As Cynthia Freeland reminds us in her response to Smith’s AToCC, James Peterson in Post Theory argued that

a common feature of avant-garde film viewing – one that usually passes without comment: viewers initially have difficulty comprehending avant-garde films, but they learn to make sense of them. Students who take my course in the avant-garde cinema are at first completely confused by the films I show; by the end of term, they can speak intelligently about the films they see. (Peterson 1996, 110; quoted in Freeland 2012, 41)

In other words, as per my assertions re: The Room above, it is quite possible that humans would quite easily watch – and enjoy – all manner of different films, but that they do not because their attention is not ‘managed and engaged’. Again, this is a political issue, because if it is true, then it is about who can afford to use the mass media to manage and engage the attention of the most people in the quest for profit – meaning that alternative approaches to filmmaking are forced either to adopt the same system of filmmaking to compete, or they are pushed to the margins where the struggle to find audiences – because people are not prepped to watch them. The scholar at the European university has had a long education in film, and this potentially helps to manage and engage differently how they attend to them; their ‘statistically insignificant’ response might well be important in helping to demonstrate how we can not just view different/marginal films, but also view mainstream films differently.

Cutting and colleagues suggest that film editing correlates with a 1/f pattern, with 1/f (one over frequency) referring to the ‘natural’ amount of time that humans attend to objects in the real world (Cutting et al. 2010). In other words, the suggestion is that Hollywood editing rhythms reflect human attention spans – ‘evolving toward 1/f spectra… [meaning that] the mind can be “lost”… most easily in a temporal art form with that structure’ (Cutting et al. 2010, 7). Now, since David L Gilden only came up with the 1/f structure in 1995 (Gilden et al. 1995), it remains untested, and untestable without a time machine, as to whether the human attention span itself changes over time, or according to culture. That said, if cinema has always been going at about the pace that human attention was working, and if cinema cutting rates have accelerated since the 1930s and through to the present era, then attention spans may well interact with culture, and even be shaped by our media.

I often ask my students how long they should look at a painting for. It’s a trick question, because of course there is no right or wrong answer. It is my (untested) hypothesis, however, that the amount of time humans look at paintings has been shaped by the media, including films; that is, in galleries, I see people look at paintings for about the average duration of a film shot (four to five seconds) – although recently they have begun to look at a painting for about the amount of time that it takes them to take a photo of that painting with their mobile handheld device.2 Smith, citing Cutting’s work, suggests that

[i]n an average movie theatre with a 40-foot screen viewed at a distance of 35 feet, this region at the centre of our gaze will only cover about 0.19 per cent of the total screen area. Given that the average shot length of most films produced today is less than 4 seconds… viewers will only be able to make at most 20 fixations covering only 3.8 per cent of the screen area. (Smith 2013: 168)

Given that paintings vary in size, one cannot rightly say how long it would take to see a ‘whole’ painting. But if one looks at a cinema-screen sized painting for 4 seconds, then one would, after Smith, fixate on about 4 per cent of that painting. In order to see the whole painting, then more time is needed, just as more time is needed to take in our natural, rather than cinematic, environment, since we also only ever see a small proportion of that at any one time.

Relating to film the foregoing foray into painting, we might add that, given that we do not take in visual information while saccading, and given that saccades have a duration of 20-50 miliseconds (Smith 2013, 168), this means that we do not take in visual information for 0.7 seconds during every four-second shot. At 90 minutes in length, there are on average 1,350 shots per film, meaning that we do not take in visual information for 15 minutes and 45 seconds per film – blinks and turning away from the screen for snogging and toilet breaks not included. If spatially we only see 3.8 per cent of the screen during a shot, and if we only see 82.5 per cent of a film’s duration, this means that we see around 3.14 per cent of the average (Hollywood) film (no spooky π references intended).3 To be clear, these statistics apply not just to Hollywood: I would only see 3.14 per cent of Empire if I were to watch it at the cinema, too. But since it is a film comprised of a single-seeming shot and a static frame, Empire clearly encourages viewers to look for longer at the space within the frame, while Hollywood arguably does not give viewers the time to do so, since the content and duration of images is concerned uniquely with story-telling, and not with anything else. This in turn affects for how long we think that we are supposed to look at objects in our everyday lives, if for the sake of argument my gallery hypothesis be allowed to stand. Neither paintings, nor Empire, nor the world itself is organized to be seen ‘cinematically’, even if Empire is undoubtedly a work of cinema. That is, they all invite contemplation, but what they often receive is a shot-length of attention before they become boring (Empire perhaps deliberately so). Neither paintings, nor Empire, nor large swathes of the world itself controls our attention in the way cinema does; there would be much more idiosyncrasy and less synchrony of attention when looking at Empire than at a mainstream film. If the proliferation of screens featuring cinematic techniques is the making-cinematic of reality in the services of capital, then the refusal to attend to paintings, Empire and the world itself suggests not just that our attention is controlled while watching a film, but that our attention is working at a ‘cinematic’ rhythm – a rhythm that Empire uses the very apparatus of cinema in order to try to break.

The ‘temporal politics’ that I mentioned above, then, is to do with the management and engagement of attention rhythms/patterns not just in cinema, and not just for cinema (we are prepped to be movie viewers), but also by cinema for the world (people pay attention to paintings in galleries about as long as they would attend to a film shot/as long as a film shot would allow them to attend to it, before ‘cutting’, or turning away, likely getting out one’s phone, the screen of which one can also cut across with the swipe of a thumb). Politics rear their head again as homogeneity of attention span, perhaps even of life rhythm, jump into bed with the political and economic concerns that govern the structures of our society. Almost certainly in an unwitting fashion (this is not a conspiracy), validating certain cutting rates and attentions spans over others becomes an issue linked to social control, and the economic bottom line of both cinema and perhaps society as a whole. Eye tracking studies of film are part of this political ecology.

A final throw of the dice. Those of us engaged in education are of course part of a system that prepares our students for the real world. But I am personally also committed to encouraging my students sociably and communicatively to develop their individuality, to become ‘idiosyncratic’, to look at the world differently and various other notions that have long since been corporatized disingenuously as advertising slogans. Being a film teacher, I do this through encouraging my students to look differently at films. Hollywood films employ techniques that do not encourage us to look differently at movies; instead, our attention (and our brain activity) are synchronized. What is more, the idiosyncratic viewers that do look at films differently (the European film scholar) are discounted from eye tracking studies for not conforming to the norm (for not confirming to us what we already know, even if not through a scientific framework). Not only might we encourage our students to look at the world differently (to become the idiosyncratic, perhaps ‘educated’ viewer), but we might also encourage our students to make films differently, since films can also play a role in encouraging us to see the world differently, to become ‘idiosyncratic’ individuals (Hasson’s research involved the production of an interesting avant garde work, regardless of his own thoughts on the matter). Perhaps eye tracking (and fMRI) studies can help in this by turning their attention not to the majority, but to the minority, to the marginal people who look, both figuratively and literally, at the margins of the screen, and at marginal films. And this perhaps involves slowing attention down, and making it (willfully?) deeper rather than rapid and superficial. I know that the longer I look at a painting, the more the power of its creation comes to my mind, the more I marvel at it and also at the world that sustains it. In other words, it brings me joy. As I repeat often to those students who do not seem committed to participating in my classes: the more you put in, the more you get out.

Would to educate (to manage and engage attention) both in the classroom and through making and showing different sorts of (slower?) films not simply replace one trend with another, and itself be prey to political issues regarding what type of ‘idiosyncrasy’ is best? Of course, such questions are going to be of ongoing importance and would need constant attention. In relation to eye tracking film studies, though, the introduction of a ‘temporal’ dimension might help enrich our understanding of idiosyncrasy. The spatial information that idiosyncratic eye-tracks give to us is chaotic and without pattern – and thus of not much use to the psychologist; however, there may well be temporal patterns that emerge when we consider ‘idiosyncrasy’ as a shared process (to be encouraged?), rather than as a reified thing to be commoditized.

Paul Willemen has written about cinephilia as being the search for/paying attention to otherwise overlooked details in movies (Willemen 1994, 223-57). Meanwhile, Laura Mulvey has argued that DVD technology allows the film viewer to develop a deeper, cinephilic relationship with movies, since she can now pause and really analyse a film – by ‘delaying’ it/slowing it down (Mulvey 2006, 144-60). To look idiosyncratically at a movie is thus to look ‘cinephilically’; it is to look at cinema with love, perhaps to look with love tout court – but in this instance at cinema. My argument comes full circle, then, as we bring cognitive film theory, via eye tracking film studies, into contact with film theory/film-philosophy, exemplified here by Mulvey as a major figure from the Screen movement/moment. There is no I in eye tracking – but if we can accept that eye tracking studies of cinema are embroiled in a political discourse (and a political reality) concerning which films are validated as better than others and why, then perhaps by putting an ‘I’ into eye tracking, by looking at the idiosyncratic in addition to the statistically significant, then we may be able to bring about different ways of seeing and making films.


  1. The exception is Dancer in the Dark (Lars von Trier, Spain/Argentina/ Denmark/Germany/Netherlands/Italy/USA/UK/France/Sweden/Finland/ Iceland/Norway, 2000).
  2. One of my peer reviewers took issue with the speculative nature of this suggestion. The other agreed with it.
  3. Note that I insist on the term ‘visual information’ – since film does not just engage us visually, but also aurally and via other senses (as Freeland, 2012, also reminds Smith in her response to his AToCC essay).



Anderson, Chris. 2006. The Long Tail: Why the Future of Business is Selling Less of More. New York: Hyperion.

Beller, Jonathan. 2006. The Cinematic Mode of Production: Attention Economy and the Society of the Spectacle. Lebanon, N.H.: Dartmouth College Press.

Bordwell, David. 2010. “Now you see it, now you can’t.” Observations on Film Art: Kristin Thompson and David Bordwell, June 21. http://www.davidbordwell.net/blog/?p=8509.

Bordwell, David, and Noël Carroll. 1996. Post-Theory: Reconstructing Film Studies. Madison: University of Wisconsin Press.

Cutting, James E.. 2014. Peer Reviewer’s Comments. Received October 1.

Cutting, James E., Jordan E. DeLong and Christine E. Nothelfer. 2010. “Attention and the Evolution of Hollywood Film.” Psychological Science 20:10, 1-8.

Freeland, Cynthia. 2012. “Continuity, Narrative, and Cross-Modal Cuing of Attention.” Projections: The Journal for Movies and Mind 6:1, 34-42.

Gilden, D.L., T. Thornton, and M.W. Mallon. 1995. “1/f Noise in Human Cognition.” Science 267:1837-39.

Hasson, Uri, Ohad Landesman, Barbara Knappmeyer, Ignacio Vallines, Nava Rubin and David J. Heeger. 2008. “Neurocinematics: The Neuroscience of Film.” Projections: The Journal for Movies and Mind 2:1, 1-26.

Marchant, Paul, David Raybould, Tony Renshaw and Richard Stevens. 2009. “Are you seeing what I’m seeing? An eye tracking evaluation of dynamic scenes.” Digital Creativity 20:3, 153-163.

Messaris, Paul. 2012. “Continuity and Its Discontents.” Projections: The Journal for Movies and Mind 6:1, 28-33.

Miller, Greg. 2014a. “How Movies Manipulate Your Brain to Keep You Entertained.Wired, August 26.

Miller, Greg. 2014b. “How Movies Synchronize the Brains of an Audience.” Wired, August 28.

Mulvey, Laura. 2006. Death 24x a Second: Stillness and the Moving Image. London: Reaktion.

Peterson, James. 1996. “Is a Cognitive Approach to the Avant-Garde Cinema Perverse?” In Post Theory: Reconstructing Film Studies, edited by David Bordwell and Noël Carroll, 108-129. Madison: University of Wisconsin Press.

Rogers, Sheena. 2012. “Auteur of Attention: The Filmmaker as a Cognitive Scientist.” Projections: The Journal for Movies and Mind 6:1, 42-49.

Saez de Urabaín, Irati R., Mark H. Johnson and Tim J. Smith. 2014. “GraFIX: A semiautomatic approach for parsing low- and high-quality eye tracking data.” Behavior Research Methods, March 27, pp. 1-20.

Sconce, Jeffrey. 1995. “‘Trashing’ the academy: taste, excess, and an emerging politics of cinematic style.” Screen 36:4, 371-393.

Smith, Greg M. 2012. “Continuity Is Not Continuous.” Projections: The Journal for Movies and Mind 6:1, 56-61.

Smith, Tim J. 2012a. “The Attentional Theory of Continuity Editing.” Projections: The Journal for Movies and Mind 6:1, 1-27.

Smith, Tim J. 2012b. “Extending AToCC: A Reply.” Projections: The Journal for Movies and Mind 6:1, 71-78.

Smith, Tim J. 2013. “Watching You Watch Movies: Using Eye Tracking to Inform Cognitive Film Theory.” In Psychocinematics: Exploring Cognition at the Movies, edited by Art P. Shimamura, 165-191. New York: Oxford University Press.

Smith, Tim J. 2014. “Audiovisual Correspondences in Sergei Eisenstein’s Alexander Nevsky: A Case Study in Viewer Attention.” In Cognitive Media Theory, edited by Ted Nannicelli and Paul Taberham, 85-105. London: Routledge/American Film Institute.

Smith, Tim J, and John M. Henderson. 2008. “Edit Blindness: The relationship between attention and global change blindness in dynamic scenes.” Journal of Eye Movement Research 2(2):6, 1-17.

Tatler, Benjamin W., Nicholas J. Wade, Hoi Kwan, John M. Findlay and Boris M. Velichkovsky. 2010. “Yarbus, eye movements, and vision.” i-Perception 1:7-27.

Treuting, Jennifer. 2006. “Eye Tracking and the Cinema: A Study of Film Theory and Visual Perception.” SMPTE Motion Imaging Journal 115:1, 31-40.

Turvey, Malcolm. 2012. “The Continuity of Narrative Comprehension.” Projections: The Journal for Movies and Mind 6:1, 49-56.

Võ, Melissa L.-H., Tim J. Smith, Parag K. Mital and John M. Henderson. 2012. “Do the eyes really have it? Dynamic allocation og attention when viewing moving faces.” Journal of Vision 12(13):3, 1-14.

Wass, Sam V., Tim J. Smith and Mark H. Johnson. 2013. “Parsing eye tracking data of variable quality to provide accurate fixation duration estimates in infants and adults.” Behavior Research Methods 45:1, 229-250.

Willemen, Paul. 1994. Looks and Frictions: Essays in Cultural Studies and Film Theory. Bloomington: Indiana University Press.

Yarbus, Alfred L. 1967. Eye Movements and Vision. Translated by Basil Haigh. New York: Plenum Press.


William Brown is a Senior Lecturer in Film at the University of Roehampton, London. He is the author of Supercinema: Film-Philosophy for the Digital Age (Berghahn, 2013) and, with Dina Iordanova and Leshu Torchin, of Moving People, Moving Images: Cinema and Trafficking in the New Europe (St Andrews Film Studies, 2010). He is the co-editor, with David Martin-Jones, of Deleuze and Film (Edinburgh University Press, 2012). He is also a filmmaker.