Can Science Help Film Theory?*

Malcolm Turvey

“Science has never shaken off the impress of its origin in the historical revolt of the later Renaissance. It has remained predominantly an anti-rationalistic movement, based upon a naïve faith.... Science repudiates philosophy. In other words, it has never cared to justify its faith or explain its meanings.”

A. N. Whitehead

Science and the Modern World

What is film theory?

At its most general, to borrow David Bordwell's words, "'film theory' refers to any reflection on the nature and functions of cinema."[1] But what sort of reflection is this? Of what type of theoretical inquiry does it consist?

NoÁl Carroll has recently reminded us that "the philosophy of science provides us with some of our best models for understanding theoretical inquiry."[2] According to one of these models, a theory stands a good chance of approximating the truth about the phenomenon it is investigating if it attempts to improve upon competing theories. As Carroll puts it:

Theory-building builds on previous histories of theorizing as well as upon data (which may be theory-laden). Present theories are formulated in the context of past theories. Appraised of the shortcomings in past theories, through processes of continued scrutiny and criticism, present theories try to find more satisfactory answers to the questions that drive theoretical activity. (p. 57)

If formulated in robust, dialectical competition with other theories, using the best standards of evidence and reasoning available, a theory has a good chance of taking a step toward the truth about the phenomenon under investigation. But this also means that a theory is fallible. It is proposed in the full knowledge that it, too, will probably be revised and even rejected in the future, that it is merely an approximation of the truth, an hypothesis that will be improved upon in the light of new evidence and/or more stringent reasoning.

This post-positivist model of theory-building now more or less dominates mainstream philosophy of science. It admits the historicity of natural scientific theories, but resists the "framework relativism" that a few philosophers have argued is the consequence of this historicity. For as Carroll points out:

The fact that theorizing has a history does not compromise the possibility of discovering what is the case, since that history may involve, among other things, the successive elimination of error. Furthermore, the fact that we are constantly revising our theories in the light of continued criticism and new evidence does not preclude the possibility that our theories are getting closer and closer to the truth. (p. 58)

Certainly, this post-positivist model seems to be a good explanation for why natural scientific theorizing has been so successful in explaining and predicting the natural world. But is it also an appropriate model for film theorizing?

Carroll seems to think it should be. Although he is careful to state that "We should not attempt to slavishly imitate any of the natural sciences" (p. 59) and that "We need to be alert to the special features of our own field of inquiry" (ibid.), he expresses the hope that "what can be claimed for science may be claimed eventually for film theory" (ibid.). What I take him to mean here is something like the following: while film theory is certainly not a natural science, it has a good chance of achieving the explanatory successes of the natural sciences if it is modeled on their method of theory-building—if, in other words, film theories are conceived of as fallible hypotheses that are formulated through dialectical criticism of other theories according to the best available standards of evidence and reasoning.

This seems to me to be a noble goal, and Carroll proposes it with the best of intentions. For it is undoubtedly the case that rigorous dialectical criticism is essential for progress in any pursuit of knowledge. Furthermore, it is perfectly reasonable to turn to the natural sciences to find a model for film theorizing, considering their enormous explanatory success and prestige. However, the sciences do not have a monopoly on dialectical criticism—it can be a feature of any rational form of inquiry, whether scientific or not, if participants are willing. And while the post-positivist conception of theory-building recommended by Carroll undoubtedly provides a good model for some film theorizing, I think we need to examine carefully whether that model, or, indeed, any conception of theory-building in the natural sciences, provides an appropriate model for all film theorizing. As Carroll himself points out, the subject matter of humanistic disciplines such as cinema studies possesses features of its own that set it apart from the phenomena studied by the natural sciences. But what Carroll does not take into consideration is the possibility that, because of these features, the methods and forms of explanation employed by the natural sciences, including the method of theory-building he recommends, may be logically inappropriate for investigating humanistic subject matter.

My suggestion is that before we adopt Carroll's proposal wholesale, we need something like the philosophy of the humanities called for by philosopher G. H. von Wright, a propaedeutic philosophy that will clarify precisely which methods and forms of explanation are logically appropriate for investigating humanistic subject matter, and which are not. [3] Once we have such a philosophy, we will understand better what type of questions the method of natural scientific theory-building recommended by Carroll can help us to answer, and what type this method may be of no help in answering.

Obviously, such a philosophy is beyond the scope of this paper. Here, I will simply point to one feature of the cinema that is a good candidate for distinguishing it from the phenomena studied by the natural sciences, a feature it shares with humanistic phenomena in general and that any, more general, philosophy of the humanities would have to take into consideration. I will then argue that this feature rules out, on logical grounds, the method of theory-building recommended by Carroll for investigating much humanistic subject matter.

Carroll's recommendation that we adopt the post-positivist conception of natural scientific theory-building as a model for film theorizing rests on the assumption that the cinema is like the phenomena studied by the natural sciences in at least one way: in order to find out about it, we need to propose fallible hypotheses in the form of theories that we continually attempt to improve upon using the best available standards of reasoning and evidence.

It is easy to understand why the natural sciences attempt to find out about natural phenomena in this way. For the natural sciences aim to discover the causal or probabilistic laws that govern the nature and behavior of natural phenomena, laws of which human beings have no prior knowledge. Because we have no prior knowledge of these laws, because they are contingent and empirical, there are only three ways—at least according to the standard picture of scientific discovery—for us to gain knowledge of them: either scientists begin empirically by making observations and conducting experiments on natural phenomena, and then gradually work up to hypotheses and theories about these phenomena through induction; or they begin theoretically with hypotheses about the natural universe, deduce some observable consequences from these hypotheses, and conduct empirical experiments on natural phenomena to test them; or they combine the two approaches (which is what they usually do in practice).

Carroll seems to be claiming that what film theorists aim to discover about the cinema—its nature and functions—resemble the laws governing natural phenomena in at least one way: we have to find out about them through induction and/or deduction, in the same way that scientists find out about natural laws. In other words, the nature and functions of cinema are not known to us in advance. Nor can we ever know them with absolute certainty. Rather, like natural laws, they are contingent and empirical, and the only way we can find out about them is through proposing fallible hypotheses arrived at inductively and/or deductively, hypotheses that, at best, approximate the truth, and that will be improved upon using the best standards of reasoning and evidence available.

But if we look at the history of film theory, one of the most obvious things we notice is that film theorists rarely if ever treat their propositional claims about the cinema as fallible hypotheses about contingent, empirical phenomena of which we have no prior knowledge. For example, they rarely systematically test their claims against hard empirical evidence. Although they sometimes draw on empirical research from the human and natural sciences, they do not themselves collect and analyze data in support of their claims. When Eisenstein, for instance, says "Cinema is montage," or Metz says "Cinema is the Imaginary," neither of them offer much if any empirical evidence to support these assertions.

Carroll would probably argue that this is because these particular film theorists are insufficiently scientific. But if we look at more recent film theories putatively formulated in a more scientific spirit, we find a similar failure. Consider David Bordwell's constructivist cognitivist theory of film comprehension proposed in Narration in the Fiction Film. In this theory, Bordwell makes some major claims about the cinema and human beings. For example, he argues that "the organism [by which he means a human being's hard-wired mental faculties] enjoys creating unity" in order to support his claim that the spectator of a fiction film attempts to mentally construct a unified "fabula" out of the perceptual data the film provides.[4] However, Bordwell does not offer any empirical evidence in support of this claim, such as data about what real spectators actually enjoy doing when they view a film. [5] Similarly, in his book Engaging Characters, the film theorist Murray Smith proposes a theory of our engagement with characters in cinematic fictions by developing the concepts of "recognition, alignment, and allegiance."[6] However, Smith does not attempt to test his theory through controlled experimentation on real spectators and the way they actually feel about fictional characters. Although Bordwell and Smith both draw on research from the human sciences in order to formulate their theories, just as Eisenstein and Metz did, neither of them seem to think it is necessary to test their own hypotheses against data.

Why is this? Why is there a lack of basic empirical research in film theory if the nature and functions of cinema are like the laws governing natural phenomena? Why does such research, somehow, seem unnecessary to film theorists? And how is it that film theories ever convince anyone that they are plausible in the absence of such sustained research? Considering the influence of Metz's and Bordwell's theories, why is it that so many of us so readily assent to them, and so quickly?

One reason, I would suggest, is because unlike theories of the natural world, film theories concern what human beings already know and do, and this points to a major difference between natural phenomena and the cinema. Human beings have no prior knowledge about the laws that govern natural phenomena and must find out about those laws through scientific investigation. In the case of the cinema, however, human beings already know a good deal prior to any scientific investigation. This is because the cinema is a human creation, and it is embedded in human practices and institutions that most of us engage with on a daily basis. For example, most of us know what a comedy film is, and how to respond to one appropriately. Although the average person may not be able to state in a propositional form what comedy films are (he or she may not be able to formulate a theory of cinematic comedy) most of us nevertheless know comedies when we see them and have no problem understanding them.

This is one reason, I suggest, why film theories command assent in the absence of sustained empirical research and experimentation. [7] We already have a good idea of what they are talking about, and the most powerful and plausible ones seem, at least in part, to clarify what we already know and do. When someone like Metz, for example, compares the cinema to phenomena such as dreams, daydreams, and fantasies, some of us are tempted to agree with him because we think, "Ah yes, he's right. Some films are like dreams, the experience of cinema is sometimes fantasy-like," and so on. Or when Bordwell claims that we try and construct a unified fabula when watching a fiction film, some of us might concur because we think, "Yes, that's true. When I watch a film, I do make inferences, try and fill in missing information, anticipate and form hypotheses."

When we look at theoretical explanations of the nature and behavior of natural phenomena, however, we immediately see a difference. For with such explanations, the question of whether or not they clarify something we already know and do does not play a role in whether we accept them or not. For example, the claim that the "ventromedial prefrontal region" of the "frontal lobe" of the human brain controls normal decision-making, and that damage to this region (as in the famous case of Phineas Gage) impairs our ability to plan for the future, is not something we accept because it clarifies something we already know.[8] Rather, we accept it because the empirical data shows that it is the most plausible hypothesis. When this area of the brain is damaged either by accident or in controlled experiments, we find that it repeatedly causes an impairment of normal decision-making. For all we knew prior to such empirical research, it could have been the area of the brain known as Broca's Area that controls normal decision-making. This is simply another way of saying that nature is a "blind watchmaker," to use Richard Dawkins's felicitous analogy, that its design is meaningless and its laws are blind.

Is this an important difference, and does it tell us anything about the logical appropriateness of the natural scientific method of theory-building for investigating humanistic subject matter? Carroll might agree that we know more about a humanistic phenomenon such as the cinema than we do about natural phenomena, and that this knowledge plays a role in our acceptance or rejection of film theories that it does not play in regard to theories of the natural world. But he might characterize this knowledge as empirical data, and argue that we simply have more empirical data about the cinema because of our everyday engagement with it. Because it is a humanistic phenomenon, he might stay, it stands to reason that we know more about it pre-theoretically than we do about natural phenomena because we invented it. But this does not mean, he might continue, that we need to stop proposing fallible hypotheses about cinema's nature and functions, thereby improving upon our pre-theoretical knowledge of the cinema.

However, I think the fact that we know more about the cinema shows that it is a fundamentally different phenomenon than natural phenomena. This is because we need to know more about the cinema—or at least certain aspects of it—in order to be able to engage with it, while the same is not true of natural phenomena. For example, when watching a fiction film at the local multiplex, we can use our eyes perfectly well without any knowledge of their design, of the laws that govern their nature and behavior. We simply rely on the fact that they work. But we cannot understand the same film if we do not already have a concept of what fiction is, what the difference between fiction and non-fiction is, what a cut as opposed to a dissolve means, what the difference between horror and comedy is; if we have not been taught the appropriate ways to respond intellectually and emotionally to cinematic fictions and the role they play in our culture; if we do not know the customs, conventions, and a whole host of other norms that govern films and film viewing; as well as much else that film theorists are interested in. If these things really were equivalent to the empirical, contingent laws governing natural phenomena, then knowledge of them would be as unnecessary to us in order to be able to engage with cinema as is knowledge of the laws that govern the nature and behavior of our eyes.

Philosophers such as von Wright and Charles Taylor tend to draw this distinction in the following way: while natural phenomena are what they are independently of human knowledge of the laws that govern them, a humanistic phenomenon such as the cinema is constituted, at least in part, by human norms and human beings who know and use these norms. The cinema would not be what it is if there were not human beings who knew these norms, while natural phenomena are what they are independently of whether human beings know the laws that govern them and understand their design.[9] Wittgenstein clarifies this difference by using games as a paradigm for humanistic phenomena. When we learn to play tennis, for example, we master various concepts, customs, and conventions, and the patterns of behavior they inform.[10] We learn what a serve is, what it means in terms of winning or losing a game, what an acceptable return of serve is, what a lob is, what "out" means, that we should not shout at our opponent during play, what emotions we should feel and when, and so on and so forth. In order to play tennis, we have to master these various types of norms. Tennis, however, is also shaped by the natural world. If gravity was six times weaker than it actually is, for example, tennis would be a very different game. However, we do not need to know anything about the laws governing gravity to be able to play tennis, and gravity is the way that it is independently of whether we know anything about these laws. Knowledge of such laws is external to playing tennis. These laws do not constitute it, although they shape it. When we play tennis, we simply rely on the fact that we do not float away from the earth. Knowledge of the variety of norms of tennis, however, is internal—they constitute it. Without knowledge of them, we would not be able to play tennis. Indeed, there would be no such thing as tennis.

Similarly, our ability to understand and respond appropriately to a film is dependent, at least in part, on our knowledge of a whole host of norms that are internal to our engagement with cinema. Certainly, like tennis, the cinema is shaped by the natural world. Think, for example, how different cinema would be if our eyes were able to perceive each individual frame of a film, or if our vision was triocular instead of binocular, or if we could only see in black and white, or if, to use one of Murray Smith's examples, we could not reidentify characters from scene to scene. But knowledge of the design of our eyes, of the empirical, contingent laws governing their nature and behavior, is external to our practice of engaging with cinema. We do not need to possess this knowledge in order to understand and respond appropriately to a film.

III

If this distinction between external and internal is valid, then it means that at least some of the theoretical questions we ask about the cinema are likely to be about the internal norms that constitute it, such as the distinction between fiction and non-fiction, or the concept of comedy. And if this is the case, then natural scientific theory-building is not a logically appropriate method for answering such questions. For this method is designed to make discoveries about empirical, contingent phenomena, such as natural laws, of which we have no prior knowledge. But the various norms that constitute cinema are ones we already know and use. Hence, we do not need to propose fallible hypotheses in order to find out about them because we already know and use them. Instead, in order to answer questions about them, we need to clarify what we already know and do, the concepts, customs, and conventions we already have and use, rather than finding out new information that we have no prior knowledge of through induction and/or deduction. Answering questions of this kind is akin to clarifying the variety of grammatical rules that govern our everyday use of language, for we follow grammatical rules all the time when we speak or write, such as the rules that govern the position of adjectives before nouns. But we are typically not able to state in a propositional form what these rules are until they are clarified for us by grammarians.

It is also akin, as the philosopher Oswald Hanfling has pointed out, to the type of knowledge aesthetics seek. Not all forms of rational inquiry, Hanfling has reminded us, involve finding out new information, aesthetics being the classic example.[11] Typically, gaining aesthetic knowledge about a work of art consists of having our attention drawn to features of that art work that lie in clear view before our eyes but which we have not noticed or paid attention to…or paid the right kind of attention to. Think, for example, of the way in which David Bordwell has illuminated the playful patterning of space in Ozu's films, something that lies in clear view before any viewer of an Ozu film, but that is not necessarily noticeable to that viewer until pointed out to him or her. Similarly, I am suggesting, answers to theoretical questions about the internal norms that constitute cinema do not consist of new information, but of drawing our attention to the various concepts, customs, and conventions we already know and use when engaging with cinema, norms that—like grammatical rules—in one sense lie in plain view before our eyes because we use them everyday, but which we are often not able to couch in a propositional form until they have been clarified for us.

Of course, this does not mean that the natural scientific method of theory-building recommended by Carroll is of no help at all to film theory, for there are undoubtedly other questions asked by film theorists that are about empirical features of cinema, and that do require new information in order to be answered. Knowledge of the human perceptual mechanism, for example, may explain a good deal about aspects of film style, such as why the center of the frame is privileged in most filmmaking practices. The fact that it is privileged may have a lot to do with the design of our eyes, and questions about the design of our eyes, like the design of nature in general, are empirical questions. Thus, theories that attempt to answer such questions would be well advised to employ the best method available for finding out about the design of nature, which for most is the method of theory-building recommended by Carroll.

Nor does what I have said rule out using scientific methods to answer questions about empirical features of cinema that are posed in other areas of cinema studies, such as film history. A good example of just such a method is statistical analysis.[12] Just as statistical analysis is used in order to find out, for instance, about certain empirical features of tennis—the percentage of first serves in, speed of service, number of unforced errors, and so on—so too it has been used by Barry Salt and others, with impressive results, to find out about empirical features of the cinema of interest to film historians, such as the average length of shots during certain periods in certain group styles.

Nevertheless, to return to film theory, the fact that some theoretical questions about cinema are empirical and require theory-building, while others are about the norms that constitute cinema and require clarification of those norms, means that when we pose a theoretical question about cinema, we need to be sure about the type of question we are asking in order to know how to answer it. If we are not, we risk serious confusion. We may, for example, attempt to explain film comedy by formulating a theory that draws on contemporary neuropsychology and argues that laughter is an affect caused by a certain region of the brain, when what we really should be doing is clarifying our concept of comedy. Or alternatively, we may try to explain why it is we see continuous movement when a film is projected above 16 frames per second by trying to clarify our concept of seeing, instead of investigating the design of our perceptual mechanism.

This distinction between the methods of clarification and theory-building also means that, in principle at least, our answers to theoretical questions about cinema's internal norms are not fallible and do not need improving upon in the future. They are instead measured by how well they make sense of the norms they seek to clarify. For unlike the laws governing natural phenomena, the norms that govern the cinema or a game such as tennis are public and visible. If they were not, we would not be able to learn about them or use them. If, for example, the distinction between fiction and non-fiction—like a natural law—was unknown to us and was therefore something we needed to make fallible hypotheses about in order to find out about, we would not be able to use it in the first place. We would not currently know how to distinguish between fiction and non-fiction in practice. Potentially, therefore, we can clarify once and for all the distinction that we currently operate with between fiction and non-fiction, because it will consist of a clarification of something we already know and use.

Of course, the distinction between fiction and non-fiction may change over time in subtle and not so subtle ways, just like any conceptual distinction, thereby mandating that we clarify it anew. But new clarifications are not better than their predecessors. Rather, they are clarifications of different conceptual distinctions, of practices in which the distinction between fiction and non-fiction is drawn differently. Similarly, the game of tennis may have new rules added to it, and any overview of the rules of tennis would have to be revised to take these new rules into consideration. But such an overview would not be better than its predecessors. Rather, it would be an overview of a different game—a new, modified tennis.

None of what I have said here is enough, probably, to change the mind of a committed proponent of the view that all film theorizing should be modeled on the method of theory-building employed by the natural sciences. To possibly do that, a fully fledged philosophy of the humanities is necessary, one that will flesh out in more detail the interrelated distinctions between internal and external and the methods of clarification and theory-building which I have labored to illuminate here and which will point to other features of humanistic subject matter requiring methods of inquiry different from the natural sciences.[13] However, even if these distinctions are, in the end, shown to be spurious, it is, I think, crucial that we have a debate first about whether there are such distinctions to be made. For without such a debate, we risk scientism, the dogmatic insistence that the sciences provide a model for all rationalistic inquiry, the illicit extension of the methods and forms of explanation of the sciences into domains of knowledge where they have no application. In his classic work Science and the Modern World, the philosopher and mathematician A. N. Whitehead argued that one of the pre-conditions for the scientific revolution of the sixteenth century was the absence of just such a debate.[14] While things have certainly improved in the natural sciences since Whitehead made this claim, I do not think they have in humanistic studies. Certainly, such a debate has never taken place in cinema studies, even though what Roland Barthes once referred to as the "euphoric dream of scientificity" has remained a perennial attraction to film theorists, whether in the form of specific scientific theories and paradigms—structuralism, Althusserian Marxism, psychoanalysis, and now cognitive psychology—or the methods of science, such as the model of theory-building recommended by Carroll. If we engage in such a debate, we may find that, at least in answering some theoretical questions, the sciences are of no help to us at all.