The Engineering of Vision from Constructivism to Computers
INTRODUCTION 1 CHAPTER 1. VISUAL ATOMISM 8 1. INTRODUCTION 8 2. EXPERIMENTAL PSYCHOLOGY AND THE SCIENCE OF ART 11 3. VISUAL ATOMISM: A CODE FOR MASS COMMUNICATION 16 4. VISUAL ATOMISM AND THE MIND-BODY PROBLEM 22 5. VISUAL ESPERANTO 27 6. CONCLUSION 31 CHAPTER 2. I SEE, THEREFORE I THINK 35 1. INTRODUCTION 35 2. I SEE, THEREFORE I THINK 39 3. FIRST SIGNS OF REVOLT: VENN AND GALTON 50 4. "TO TEACH THE WORKER TO THINK DIALECTICALLY." 54 5. FREUD'S THEORY OF VISUAL REASONING 61 6. THE RISE OF THE DIAGRAM 68 7. VISUAL TECHNOLOGIES AND THE MIND 77 8. ANALOG ENGINE 85 CHAPTER 3. MAPPING SPACE 90 1. VISUAL NOMINALISM 90 2. "THE MOST IMPORTANT EVENT OF THE RENAISSANCE." 99 3. RADAR: SEEING WITHOUT EYES 115 4. 3-D COMPUTER GRAPHICS: INTERACTIVE PERSPECTIVALISM 125 5. COMPUTER VISION: AUTOMATION OF SIGHT 132 6. CONCLUSION 146 CHAPTER 4: THE ENGINEERING OF VISION FROM INKHUK TO MIT 150 1. NOT ARTISTS BUT ENGINEERS 150 2. INFORMATION THEORY: AN ENGINEER ANALYZES COMMUNICATION 156 3. THE INFLUENCE OF INFORMATION THEORY OR THE IDEOLOGY OF THE CODE 162 4. FROM "HUMAN MOTOR" TO "HUMAN INFORMATION PROCESSING" 173 5. COMMUNICATION ENGINEER ANALYZES HUMAN VISION 177 6. HUMAN ENGINEERING 186 7. CONCLUSION: THE LABOR OF PERCEPTION 193 CONCLUSION 196 BIBLIOGRAPHY 199 FIGURES 209
Introduction
The dissertation presents a history oI modern ideas about vision. I believe that vision is not a timeless concept; rather, each period understands vision diIIerently depending on how it is used. In the twentieth century, vision acquired new roles as the medium oI mass communication and the instrument oI labor, and, as any other productive tool, it was subjected to engineering, rationalization and automation. Such new disciplines as applied experimental psychology and cognitive science, communication engineering and Iilm, robotics, and advertising design continue to search Ior ways to utilize vision productively. In the process, they generate new knowledge about vision, at the same time reducing it to a Iew disjoined and limited models. The dissertation chapters Iollow the development oI Iour such models: vision as a code, vision as a means oI logical reasoning, vision as a way to capture spatial inIormation, and vision as inIormation processing.
Let us consider a Iew deIinitions oI vision that are representative oI entire research paradigms and that were unthinkable beIore the middle oI this century. David Marr's Vision, published in 1980, summarized a decade oI investigations on human perception carried out at the MIT ArtiIicial Intelligence Laboratory. This book has been the most inIluential account oI the computational approach to vision, shared by computer scientists and psychologists. It opens with this statement:
What does it mean, to see? The plain man's answer (and Aristotle's, too) would be, to know what is where by looking. In other words, vision is the process oI discovering Irom images what is present in the world, and where it is. 1
1 David Marr, Vision (New York: W.H. Freeman and Company, 1982), 3.
There is nothing "plain" about this deIinition oI vision; it is Iunctional and pragmatic. Visual perception is reduced to a number oI computational processes Ior the recovery oI limited inIormation about the world: the identity oI objects and their positions. In Iact, this is the only kind oI inIormation which may be required Ior a robot or an automatic missile to perIorm its task. Marr projects these goals oI machine vision onto human vision. Here, vision is reduced to the common denominator shared by humans and low level organisms: to detect an obstacle, a predator, a prey. In a special 1984 issue oI Cognition (the leading periodical oI cognitive science) devoted to visual representations, Steven Pinker outlined the understanding oI vision commonly held by cognitive psychologists:
Certain abstract problems could be best solved by translating their entities into imagined objects, transIorming them using available image transIormations, detecting the resulting spatial relations and properties, and translating those relations and properties back to the problem domain. 2
In this deIinition, vision is valued merely Ior its topological properties -- the ability to represent such relations as inclusion, proximity, and relative positions. According to many cognitive scientists, these properties make it a more eIIicient medium Ior problem-solving and abstract thinking than language. And since the mind itselI is imagined as an exemplary model oI computational eIIiciency that evolved through evolution, scientists postulate the centrality oI visual (read: topological) representations Ior various mental processes. To think productively and without waste, we might dispense with language and instead think through images, submerging ourselves into the silent movie theater oI our minds. Here is another model oI vision that narrows its deIinition to a particular productive property. In a 1951 overview oI applied experimental psychology, the Iield which is today known as human Iactors, Paul Fitts writes:
2 Steven Pinker, "Visual Cognition: An Introduction," Cognition 18 (1984): 66.
Each sense modality has certain inherent advantages and disadvantages Ior the detection and analysis oI diIIerent kinds oI inIormation. Audition is more nearly a continuous sense than vision; vision is basically selective and intermittent. As a consequence, audition is well adapted Ior the detection oI warning stimuli that may arise at any moment Irom one oI a variety oI sources, whereas vision is well suited to the selection oI and concentration on particular stimuli to the exclusion oI others. 3
What is the purpose oI this research into the relative properties oI the senses? In contrast to a manual worker, the job oI an operator in a modern human-machine system, be it an airplane cockpit, radar display or an automated production line, is primarily perceptual in nature. His or her work is to monitor the displays and to detect those signals that require intervention. Starting in World War II, experimental psychologists have collaborated with engineers in the design oI displays and perceptual strategies Ior their operators. One oI the important questions in this research has been the relative advantages oI visual, auditory and tactile displays Ior the detection and analysis oI diIIerent kinds oI signals. Consequently, Fitts describes vision as a sense which is most reliable and eIIicient Ior the constant surveillance oI a single source -- it is "basically selective and intermittent." According to this model, in order to use vision most productively, one's eyes must be literally "glued to the screen." In each oI these models, vision is understood in terms oI its eIIiciency in perIorming speciIic tasks: recovery oI three-dimensional inIormation, logical reasoning, detection oI signals. In Iact, there is no single meaning oI "vision" which is shared among them -- "vision" as such does not exist. The only thing that unites these modern approaches to vision is the quest Ior the eIIicient, reliable, and eIIective instruments oI labor. In this respect, vision is just a set oI separate tools which can be employed by either a human or a machine to get work done. It is this emphasis on the productive uses oI vision that distinguishes my project Irom a number oI recent histories oI vision. In 1988 Hal Foster edited the anthology Vision and
3 Paul Fitts, "Engineering Psychology and Equipment Design," in Handbook oI Experimental Psychology, ed. S.S. Stevens (New York and London: John Wiley & Sons, Inc., 1951), 1314.
Visuality which brought together the groundbreaking essays united by the common theme oI historicising modern vision, that is, rejecting it as a natural or a cultural constant. 4 In the last Iew years, each oI the Iive contributors -- Martin Jay, Jonathan Crary, Rosalind Krauss, Norman Bryson, and Jacqueline Rose -- have developed their essays into books which today largely deIine the direction oI the discourse on the visual in humanities. 5
At the center oI this discourse is the experience oI a seeing subject: how is what one sees inIluenced by psychic structures, by gender and power positions, and by the constraints oI the body; how is one's subjectivity determined by the experience oI seeing and being seen? At its boundaries are works oI art and other cultural representations, which are interpreted as the reiIied traces oI this changing experience. The accounts oI these writers greatly expanded the horizons oI our knowledge about modern vision. Yet, what remained unexamined is the employment oI vision Ior work. Indeed, it is remarkable that the grounding philosophical and psychoanalytic models oI vision, inIluential in this discourse, are theorized Irom the experience oI a "vacationer." For Sartre, it was the person wandering in the deserted Parisian park; Ior Lacan, it was the young man (Lacan himselI), "being observed" by a shining sardine can, while he was in a boat with some Iishermen near a small port; Ior Freud, it was a young woman conIined to isolation in a Viennese apartment or sent to a resort. Not only are these subjects Iar Irom a modern workplace, but they are also away Irom the prototypical perceptual spaces oI modernity: the city street, the movie theater, the shopping arcade.
4 Hal Foster, ed., Vision and Visuality (Seattle: Bay Press, 1988). 5 Martin Jay, Downcast Eyes: the Denigration oI Vision in Twentieth-century French Thought (Berkeley: The University oI CaliIornia Press, 1993); Jonathan Crary, Techniques oI the Observer: on Vision and Modernity in the Nineteenth Century (Cambridge: The MIT Press, 1990); Rosalind Krauss, The Optical Unconscious (Cambridge: The MIT Press, 1993); Norman Bryson, Looking at the Overlooked: Four Essays on Still LiIe Painting (Cambridge: Harvard University Press, 1990); Jacqueline Rose, Sexuality in the Field oI Vision (London: Verso, 1986).
It is precisely the experience oI these spaces that gave rise to a very diIIerent kind oI theorizing oI vision in the writings oI Walter Benjamin. Scrutinizing these new spaces oI modernity, Benjamin was among the Iirst to notice the contiguity between the perceptual experiences in the workplace and outside oI it:
Whereas Poe's passers-by cast glances in all directions which still appeared to be aimless, today's pedestrians are obliged to do so in order to keep abreast oI traIIic signals. Thus technology has subjected the human sensorium to a complex kind oI training. There came a day when a new and urgent need Ior stimuli was met by the Iilm. In a Iilm, perception in the Iorm oI shocks was established as a Iormal principle. That which determines the rhythm oI production on a conveyer belt is the basis oI the rhythm oI reception in the Iilm. 6
Here Benjamin already notices that vision became a Iunctional and pragmatic activity rather than a disinterested or contemplative one. The eye is trained to keep pace with the rhythm oI industrial production at the Iactory and to navigate through the complex visual semiosphere beyond the Iactory gates. Among the contributors to Vision and Visuality, Jonathan Crary is the most sensitive to the primacy oI work Ior the history oI vision in the modern period. In Techniques oI the Observer, he examines how such nineteenth century disciplines as physiology and psychophysics, and such techniques as optical apparatuses and illusions prepared vision Ior its later productive deployment. This Iar-reaching archeology is extremely revealing, yet, because the account ends in the middle oI the nineteenth century, it can outline the Iuture deployment oI vision only in very broad terms: as participating in the process oI the rationalization oI labor and as adapting to the new Ilux oI dematerialized images and signs. In order to understand what really happened to vision in the twentieth century, how it was actually deployed and investigated, Crary's Techniques oI the Observer is necessary but not suIIicient. This century has seen the emergence oI disciplines and techniques oI vision that simply cannot be traced to the period
6 Walter Benjamin, "On Some Motives in Baudelaire," in Illuminations, ed. Hannah Arendt (New York: Schochen Books, 1969), 175.
analyzed by Crary. In order to understand their current directions, we need to consider social phenomena oI more recent origin such as automation, computerization, and "the inIormation revolution." Most importantly, what could not have been predicted Irom Crary's account is the radical shiIt in the nature oI labor itselI. Benjamin described the rhythm oI perceptual shocks oI the assembly line worker oI the industrial era, who was nevertheless paid to perIorm physical work: liIting, hammering, Iiling, and molding. Today, however, in the post-industrial era, the same worker is paid Ior "perceiving" -- watching, detecting, scanning, and monitoring. This dissertation is an archeology oI the currently prominent disciplines and techniques oI vision -- computer vision and cognitive science, virtual reality and other new human-computer interIaces. In this archeology, I try to uncover the recently accumulated layers -- 1920s, 1940s, 1960s. Rather than postulating a single paradigm shiIt, I describe a number oI distinct approaches to the rationalization oI vision, which developed in parallel. Each chapter investigates one oI these approaches. The Iirst chapter is concerned with the model oI vision as code oI communication. Investigations oI the psychological eIIects oI basic colors and elemental Iorms, conducted in experimental psychology since the second halI oI the nineteenth century, made possible the idea oI a rational visual language composed Irom simple elements. This idea acquired new importance in the 1920s, when artists Iound themselves in the role oI the designers oI mass communication. In Soviet Russia, artists collaborated with psychologists in order to create a Iully rationalized visual code oI mass communication where each visual element is capable oI communicating a meaning, producing an emotion or causing a behavioral response. Visual reasoning is the topic oI the second chapter. While classical philosophy considered vision unsuitable Ior reasoning, in the modern period this attitude is reversed. Freud's model as deIined in The Interpretation oI Dreams, Galton's composite photography, Eisenstein's intellectual montage, and the adaptation oI graphic diagrams in logic and in popular visual
culture are all symptoms oI this emerging understanding oI vision: as the most eIIicient medium Ior logical thinking. With the centrality oI cognition Ior the post-industrial workplace, the notion oI visual reasoning acquired new importance. Consequently, the work on the models, techniques, and technologies oI visual reasoning, which was previously pursued by a Iew individuals, now became a matter oI systematic research on the industrial scale in such Iields as cognitive psychology and scientiIic visualization. The third chapter addresses the role oI vision as a way to capture spatial inIormation. Historically, this role has been played by the techniques oI perspectival representation. Between 1940 and 1960, these techniques were automated (computer vision, computer graphics) and also expanded beyond the realm oI the visible with radar and other remote sensing instruments. The automation oI these techniques necessitated a new paradigm oI vision research, which deIined vision as a computational system Ior the reconstruction oI object positions in three-dimensional space. The last chapter takes up the understanding oI vision in terms oI inIormation processing, which emerged aIter World War II. In the post-industrial society, vision became the principal instrument oI labor, the most productive organ oI an operator in a human-machine system. In order to monitor and improve the system's productivity, engineers and experimental psychologists have adopted a common Iramework oI inIormation theory Ior both human and machine components. Today, the number oI new disciplines which study vision continues to expand: image science, computer graphics, image processing, computer vision, research on human-computer interIaces and so on. Why? Modern society relies on vision not only as a means Ior the production oI subjectivity but Iirst oI all as a means oI economic production. II the Iirst use oI vision has been extensively theorized, the second so Iar remained largely unexamined. I hope that this dissertation will begin Iilling this missing gap in histories oI vision.
Chapter 1. VisuaI Atomism
1. Introduction
Let us compare two statements both related to the question oI visual eIIectiveness but made in diIIerent periods. Charles Henry was a French writer who is mainly remembered today Ior his theory oI scientiIic aesthetics which greatly inIluenced Georges Seurat and Paul Signac. 7 Writing in the 1880s, he advocated that aesthetic responses to simple perceptual elements (which Ior him were brightness, color, and line) need to be studied scientiIically:
Art pursues the expression oI the physiognomy oI things, and aesthetics studies the conditions in which these things are satisIying; that is, when they are represented gay or sad, agreeable or disagreeable, beautiIul or ugly...aesthetic things Ior us are reduced to Iorms, to colors, and to sounds. 8
HalI a century later, in 1927, in the article Photography in Advertisement, Lszl Moholy-Nagy states:
A modern engineer, iI his goal and the Iunctional purpose oI his work are clear, can without any great eIIort make a product that is Iormally adequate and perIect in its economic construction. But the photographic advertisements oI our time are not so easy to deIine. They don't come with "user's instructions." Research into the physiological and psychological laws oI visual eIIectiveness is still Iar behind the times, compared to the study oI the physical laws. 9
Both Henry and Moholy-Nagy argue that the creation oI visual artiIacts should be rationalized and grounded in scientiIic psychological knowledge. However, the impetus Ior this rationalization is Iundamentally diIIerent. For Henry, the Iunction oI visual artiIacts is artistic
7 See JosZ ArgYelles, Charles Henry and the Formation oI a Psychophysical Aesthetics (Chicago: University oI Chicago Press, 1972). 8 Qtd. in Paul C. Vitz and Arnold B. Glimcher, Modern Art and Modern Science. The Parallel Analysis oI Vision (New York: Praeger, 1984), 87. My analysis oI experimental psychology and scientiIic aesthetics in this chapter is indebted to this groundbreaking book. 9 Lszl Moholy-Nagy, "Photography in Advertising," in Photography in the Modern Era, ed. Christopher Phillips (New York: Aperture, 1989), 87.
expression; Ior Moholy-Nagy, it is to induce eIIects in the viewer. For Henry, the artiIacts are situated in the realm oI the aesthetic, producing pleasure or displeasure, divorced Irom practical liIe; Moholy-Nagy is concerned with everything practical -- changing belieIs, attitudes, behavior. Henry calls Ior the rationalization oI beauty while Moholy-Nagy wants to rationalize the process oI communication. He believes that modern designers have to grasp "the laws oI visual eIIectiveness" to be able to control the viewer's response -- precisely and predictably. And Iinally, while Henry's "scientiIic aesthetics" deals with the artistic expression and aesthetic response oI an individual spectator, Moholy-Nagy's "laws oI visual eIIectiveness" are to be employed by the newly emerged institutions oI mass communication, such as advertising. In this chapter I will discuss one solution to the problem oI the rationalization oI vision advanced by the 1920s artistic avant-garde. Investigations oI the psychological eIIects oI basic colors and elemental Iorms, conducted by psychologists since the second halI oI the nineteenth century, made possible the idea oI a rational visual language composed Irom simple elements -- "atoms" oI visual communication. This idea was already pursued in the nineteenth century by such artists as Seurat and such theoreticians as Henry. However, in the 1920s, when artists Iound themselves in the role oI the designers oI mass communication, the idea oI an atomistic visual language acquired new urgency and importance. The comparison between the statements oI Henry and Moholy-Nagy makes it clear that it was no longer a question oI scientiIic aesthetics; oI a work oI art producing an aesthetic eIIect in an individual spectator. Now it became the question oI rationalizing mass communication -- the question oI economic and political importance. ThereIore, Moholy-Nagy writes about the need to precisely "engineer" photographic advertisements, and Bauhaus welcomes experimental psychologists. The trend reaches its extreme in Soviet Russia, where artists thought that they could control not just the consumer habits oI segments oI society (as in the West), but the consciousness oI the whole country. Here the program oI research into the "atoms" oI visual communication became most systematic,
translating into the establishment oI a number oI psychological laboratories at various art institutions. 10 Their goal was to create a Iully rationalized visual language oI mass communication where each visual element is capable oI communicating a meaning, producing an emotion or causing a behavioral response.
10 These institutions included INKhUK (Institute oI Artistic Culture, Moscow 1920-24); GAKhN (State Academy oI Artistic Sciences, Moscow 1921-30); VKhUTEMAS (State High Art-Technical Studios).
2. ExperimentaI PsychoIogy and the Science of Art
As other nineteenth century sciences, experimental psychology approached its subject, the human mind, by postulating the existence oI Iurther indivisible elements, the combination oI which would account Ior perceptual or mental experience. II chemistry and physics postulated the levels oI molecules and atoms and so Iorth, and iI biology saw the emergence oI the concepts oI cell and chromosome, experimental psychology applied the same reductive logic to the human mind. Psychologists divided sensory consciousness into diIIerent modalities: vision, hearing, and tactility and proceeded to enumerate the elementary sensations oI each. 11
The Iirst psychological laboratory was Iounded by Wilhelm Wundt at the University oI Leipzig in 1879. The general program oI Wundt's laboratory was to deIine the Iundamental elements oI each modality and to Iormulate the laws according to which these elements are combined. In 1896 Iormer student oI Wundt, E.B. Titchener, who brought experimental psychology to the U.S., proposed that there are 32,800 visual sensations and 11,600 auditory sensory elements, each just slightly distinct Irom the rest. Titchener summarized his research program as Iollows: "Give me my elements, and let me bring them together under the psychophysical conditions oI mentality at large, and I will guarantee to show you the adult mind, as a structure, with no omissions and no superIluity." 12
This analysis oI visual experience into its elements seemed to have no practical utility. In Iact, the new discipline oI experimental psychology developed within philosophy, both intellectually and institutionally. In German universities oI the 1870s, more lecture courses were devoted to psychology than any other branch oI philosophy except logic. 13 Wundt, Iormerly a
11 Julian Hochberg, "Sensation and Perception," in The First Century oI Experimental Psychology, ed. Eliot Hearst (Hillsdale, NJ: Lawrence Erlbaum Associates, 1979), 94. 12 Qtd. in Eliot Hearst, "One Hundred Years: Themes and Perspectives," in The First Century oI Experimental Psychology, 25. 13 Kurt Danziger, Constructing the Subject: Historical Origins oI Psychological Research (Cambridge: Cambridge University Press, 1990), 211.
proIessor oI philosophy, was quite content with the allocation oI psychology to philosophy and had no interest in its practical applications. 14
Yet, even beIore the discipline oI experimental psychology was recognized oIIicially, it had already Iound its Iirst eager users -- artists and aestheticians. This was not simply a question oI creating works oI art which directly illustrated the new scientiIic theory, in this case, psychologist's atomistic theory oI perception -- such as the works oI Impressionists and Neoimpressionists which represented reality with dots and dashes, the painterly equivalents oI sensory elements. Rather than illustrating an atomistic theory oI perception, some art theorists and artists realized that they could adopt its approach as a basis oI a truly scientiIic aesthetic and art. What made this idea possible? As experimental psychologists split visual experience into separate aspects (color, Iorm, depth, motion) and subjected these aspects to a systematic investigation, Irom the outset they had been asking two questions. First, what physiological and psychological mechanisms are responsible Ior the perception oI each aspect oI vision? Secondly, what is the psychological eIIect on the subject oI the stimuli, such as elemental colors or Iorms? It is this second question which was oI great interest to aestheticians and artists. By relying on elemental colors or Iorms with known psychological eIIect, artists could now predict and anticipate viewers' emotions. Conversely, aestheticians could now develop an objective aesthetic, grounded in the scientiIic laws oI the emotional eIIects oI pictorial elements. As an example oI this research, consider an aspect oI visual experience such as Iorm -- "visual conIiguration considered as distinct Irom such things as color and pictorial content and Irom the representation oI objects." 15 While some psychologists were analyzing human perception oI simple Iorms such as squares, circles, and straight lines oI diIIerent orientations,
14 Ibid., 39. 15 Vitz and Glimcher, Modern Art and Modern Science, 144.
others were investigating their psychological eIIects. In the 1870s a French proIessor oI art, Charles Blanc, developed a theory oI the intrinsic psychological signiIicance oI vertical, straight, and oblique lines. 16 Blanc popularized the ideas oI the Dutch painter and theorist Humbert de Superville who around 1830 Iirst claimed that simple lines conveyed emotions by having an inherent psychological response (Iig. 1):
Blanc argued that the straight line is a "symbol oI unity" and the curved line oI "variety." OI the straight lines, the most important is the vertical -- since man stands perpendicular to the horizon and since he is bilaterally symmetric. The horizontal line is next in importance and then the two obliques. Blanc discusses three types oI human Iaces characterized by three orientations and by the associated eIIects oI happiness, rest and sadness. 17
In the 1880s, Charles Henry Iurther advanced Blanc's ideas, developing the "aesthetic protractor" to measure the harmony oI line angles (Iig. 2). Henry's device was designed to measure "whether the angles between lines radiating in diIIerent directions Irom a single point are harmonious." 18
It is signiIicant that the Iirst laboratory research on the aesthetics oI simple Iorms was conducted in 1876 by Gustav Fechner -- the Iounder oI mathematical methods Ior the measurement oI sensations which became the basis Ior experimental psychology. Later, at the turn oI the century, many proto-Gestalt psychologists started to investigate the aesthetics oI simple Iorms, isolating preIerences Ior squares, rectangles, ellipses, and triangles as well as diIIerent lines. 19 Their publications became Iilled with "abstract" pictures consisting oI simple Iorms. This psychological research into the eIIects oI simple Iorms inIluenced Seurat, Signac, Wassily Kandinsky, Paul Klee, and Piet Mondrian, among others. For instance, Seurat, who was
Iamiliar with the works oI Blanc and Henry, advanced a similar theory oI the intrinsic psychological eIIects oI lines oI diIIerent orientation (Iig 3). 20 In another example, Kandinsky, in Point and Line to Plane, advocated "microscopic" analysis oI three basic elements oI Iorm (point, line, and plane) claiming that there exists reliable emotional responses to simple visual conIigurations. 21 Equally telling oI Kandinsky's program are the titles oI the articles he published in 1919: "Small Articles About Big Questions. I. About Point," and "II. About Line." 22
At Iirst, artists would embed simple lines or Iorms in their otherwise representational compositions. A typical case is Seurat who based the orientations oI lines in his major paintings, such as Le Cirque (1890-91) and La Parade (1887-88), on the theory which he derived Irom Blanc: "Gaiety...oI line, lines above the horizontal; calmness...the horizontal...Sadness...oI line, downward directions." 23 Gradually, however, artists gave up representation altogether and began to compose works which would consist solely oI the simple elements already studied by psychologists. It is, thereIore, not accidental that the paintings oI Mondrian, Klee, and Kandinsky look remarkably similar to the visual stimuli already widely used by psychologists in previous decades. They are also experiments, a result oI a systematic investigation into what Kandinsky called "the science oI art," the science which would allow the reliable communication oI any emotional experience.
Proposing scientiIically based aesthetics, Henry wrote:
20 Ibid., 170-71. 21 Wassily Kandinsky, (1926), Point and Line to Plane (New York: Solomon R. Guggenheim Foundation, 1947). 22 Yu. A. Molok, "'Slovar simvolov' Pavla Florenskogo. Nekotorye margonalii" (Pavel Florensky's 'dictionary oI symbols.' A Iew margins), Sovetskoe Iskusstvoznanie 26 (1990): 328. 23 Qtd. in Vitz and Glimcher, 170.
That which science can and must do is to expand the agreeable within us and outside oI us, and Irom this point oI view its social Iunction is immense in this time oI oppression and blind conIlicts. It ought to spare the artist hesitations and useless attempts in assigning or indicating the way in which he can Iind ever more rich aesthetic elements; it ought to Iurnish the critic a rapid means oI discerning the ugly, so oIten inIormulable, however much it is Ielt. 24
Henry equates an artist with a scientist whose vocation is to provide the viewers with the means oI escaping the pressures and the unpleasant experiences oI everyday liIe through the beauty oI art. The artist is the therapist oI a society, and because oI the importance oI this role, he ought to Iollow a scientiIic method in order to make his works most eIIective in helping the viewers to Iorget the "oppression and blind conIlicts" oI modern existence. Moholy-Nagy, in the already quoted statement Irom the 1920s, also compares the artist to an applied scientist -- an engineer. The artist, however, is no longer removed Irom the realm oI everyday liIe. Just as the modern engineer shapes every aspect oI material reality -- the machines Ior living (architecture), the machines Ior transportation (airplanes, cars, trains), the machines to produce other machines (the tools oI industrial production) -- an artist has the crucial role oI shaping the psychical reality, controlling people's emotions and ideas through advertisements, publicity, posters, and other visual propaganda. Moholy-Nagy's appeal to designers and photographers brings Iorth the new purpose oI the research into the elemental units oI vision -- the need to optimize the process oI mass communication. This new purpose became crystallized at the moment when modernist artists claimed the position oI designers oI mass propaganda in Soviet Russia in the 1920s. At this moment the two lines oI inquiry -- artistic exploration oI the visual elements and research in experimental psychology -- explicitly converge in Soviet discourses and institutions. In a number oI Soviet art institutes oI the 1920s such as INKhUK (Institute oI Artistic Culture, Moscow 1920-24), GAKhN (State Academy oI Artistic Sciences, Moscow 1921-30), and VKhUTEMAS (State High Art-Technical Studios), El Lissitsky, Alexander Rodchenko, Osip Brik, and others
24 Qtd. in Ibid., 87-88.
collaborated with experimental psychologists to investigate the eIIectiveness oI visual elements and their combinations. 25
The paradigm oI research into the elements oI visual communication and their eIIects can be conveniently traced through the history oI INKhUK, which during the Iour years oI its existence united the key groups oI avant-garde artists and theoreticians most committed to this research. In the Iirst year oI INKhUK's existence the group led by Kandinsky was the most active. His program called Ior the cataloging oI basic elements oI art and to describe their emotional eIIects. 26 Following Kandinsky's proposals, INKhUK established its own psychophysical laboratory in the Fall oI 1920 but Kandinsky and his supporters were Iorced to leave a Iew months later. 27 Kandinsky then became vice-president oI the newly Iormed GAKhN and the head oI its largest physical-psychological division. The method oI Kandinsky's division was characterized as "the observation oI and experimentation with the content and processes oI artistic creation and reception as well the work oI art in all its complexity, understood as a
25 Other Soviet research institutes oI the 1920s whose programs included the development oI a scientiIic approach to art and art history include: The State Institute oI Artistic Culture (headed by Kasimir Malevich, Leningrad 1923-27); The Institute oI Literature, Art and Language, a part oI Communist Academy (1918-36); The State Institute Ior the History oI the Arts (Leningrad 1921-31); The Institute oI Art History, a part oI Moscow University; The Russian Association oI Research Institutes oI Social Sciences (Moscow 1924-31). 26 Wassily Kandinsky, (1920), "Programma raboty Instituta khudozhestvennoy kultury (The Program oI the Institute oI Artistic Culture)," in Sovetskoe iskusstvo za 15 let. Materialy i dokumentatsiia (FiIteen years oI Soviet art: materials and documentation), ed I. Matsa (Moscow, IZOGIZ: 1933), 126-139. 27 The work oI INKhUK during Iirst year oI its existence has been documented in S.O. Khan- Magomedov, "INKhUK: vozniknovenie, Iormirovanie i pervyy period raboty. 1920" (INKhUK: appearance, Iormation and Iirst period oI its work, 1920), Sovetskoe Iskusstvoznanie (1981): 332-368. 3. VisuaI Atomism: a Code for Mass Communication
psycho-physical object." 28 The members oI the academy included art historians, practicing artists, physicists, psychologists, and physiologists. Meanwhile, Alexander Rodchenko became the new president oI INKhUK. Rodchenko and his supporters Iormulated a new program in total opposition to Kandinsky. Against Kandinsky's "subjective" psychologism they emphasized the materiality oI objects. They also opposed Kandinsky's "individualistic" composition in Iavor oI construction and promoted the making oI objects rather than the creation oI art. 29 However, the project oI a science oI visual elements was not abandoned in practice, only the emphasis shiIted Irom the "subjective" analysis oI individual aesthetic reactions to the "objective" analysis oI the elements as maniIested in material objects. INKhUK members, who started to teach at the newly Iormed VKhUTEMAS (State High Art-Technical Studios), had the opportunity to put these ideas into practice. 30 For instance, in his course on graphics Rodchenko would start his students with systematic exercises to combine simple Iorms (circle, triangle, and square) within a given Iormat in order to test the visual eIIectiveness oI various combinations (Iig. 4). 31
In its third stage INKhUK became the center oI activity oI the promoters oI the idea oI production art -- Osip Brik, Boris Arvatov, Nikolai Tarabukin, and others. 32 In 1923, this group Iounded the journal LEF (The Journal oI the LeIt Front oI Art, Moscow 1923-25) the goal oI which was to unite artists oI the LeIt dedicated to radical revolutionary activity in the Iace oI what they perceived as the new Ilourishing oI "bourgeois" art made possible by the NEP.
28 P.C. Kogan, "O zadachakh akademii i yeyo zhurnala" (About the goals oI the academy and its journal), Iskusstvo 1, no. 1 (1923): 9. 29 A number oI early constructivist maniIestos are now available in English translation in Nicholas H. Allison, ed., Art Into LiIe: Russian Constructivism 1914-1932 (New York: Rizzoli International Publications, 1990), 61-82. 30 It was not accidental that the artists reIerred to their activities at INKhUK as "laboratory work." "Laboratory" connoted scientiIic labor, rather than artistic experimentation. "Institut Khudozestvennoy Kultury" (The Institute oI Artistic Culture), Russkoe Iskusstvo 2-3 (1923): 85. 31 A.H. Lavrent'ev, "Propedevticheskaya distsiplina 'GraIika.' Vkhutemas. 1920-22 gody" (The discipline 'Graphics.' VKhUTEMAS. 1920-22), Tekhnicheskaya Estetika 7 (1984): 19. 32 "Institut Khudozestvennoy Kultury," 86.
What should this radical revolutionary activity be? Western art historians, critics, and artists gloriIied the constructivists' desire to give up the traditional roles oI artists and instead submerge themselves in industry, to leave the artist's studio Ior the Iactory -- in other words, to become industrial designers. 33 What received practically no attention was another role which constructivists and many other Soviet artists were claiming Ior themselves in the 1920s -- the role oI theoreticians and practitioners oI mass visual communication, be it propaganda or advertisement. And it was in this role, particularly as advertising designers, that VKhUTEMAS artists, as well as their counterparts Irom the Bauhaus, were able to accomplish the transition "Irom an easel to a machine" (to quote the title oI a book by Tarabukin, one oI the theoreticians oI production art). 34 These machines were not a lathe or an engine but a printing press, a Iilm camera, a light projector, a radio transmitter -- the machines oI mass communication rather than mass production. LeIt artists were united in rejecting art in Iavor oI mass propaganda. In the Iirst issue oI LEF Sergei Tretyakov writes: "Revolution put Iorward new practical tasks |Ior artists| -- aIIecting mass psyche, organizing the will oI the class"; "The work oI Iuturism is parallel to the work oI communism -- Iuturism is Iighting Ior the same dynamic organization oI personality without which the movement toward communism is impossible." 35 For LEF, this task oI the revolutionary education oI the masses required a scientiIic method and in the same article Tretyakov called Ior the creation oI a "science oI art as a means oI emotional and organizing
33 In 1923, Lunacharsky wrote: "They |constructivists| play at being engineers, but they don't know as much oI the essence oI machinery as a savage." Qtd. in Brandon Taylor, Art and Literature under the Bolsheviks (London: Pluto Press, 1991), 1: 177. 34 Between 1923 and 1925 Rodchenko designed advertising Ior many Soviet government agencies; Lissitsky, who was sent to Germany to publicize the art oI the Iirst Socialist state, created brilliant designs Ior many German Iirms; so did Moholy-Nagy and other Bauhaus Iaculty. 35 Sergei Tretyakov, "Otkuda i kuda? (Perspektivu Iuturizma)" |Where Irom and where to? (Perspectives oI Iuturism)|, LEF 1 (1923): 197, 203.
inIluence on psyche, in connection with the task oI class struggle." 36 In Iact, artists should become as scientiIic in their work as Marxists are in their political activity. To do so, they should rationalize the creative process and put art education on a scientiIic basis. 37 This idea was most systematically promoted by Proletcult (Organization oI the Representatives oI Proletarian Culture). Thus, Tarabukin organized the artist's workshop aiming to bring the principles oI Taylorism and Fordism into artistic practice: "We are analyzing processes oI work, experiment with compressing work periods, aim to rationalize work and leisure. In its practice the workshop tries to get away Irom bohemian habits so common to artists." 38 Artists should also rationalize viewers' reactions by developing a science which would enable them to produce predictable and measurable eIIects in the viewer. Thus, the author oI a 1927 article which summarized the activities oI the Art Section oI Leningrad's Proletcult writes:
Artist-communist, artist-revolutionary came to understand the necessity to perIectly master his weapons to be able to strike without a miss. The striving to conscious mastery was put Iorward as the main task ("it is necessary to know in order to produce"). Artists oI Proletcult were sure that at the basis oI art's eIIect on perception are to be Iound completely objective laws. From this came the task to master the "mechanics" oI artistic work, in order to orient it toward new utilitarian tasks oI revolutionary art. 39
This dual goal oI rationalization was symbolized by the popular metaphor oI artist as "psycho-engineer," Irequently encountered in the discourse oI constructivism, Proletcult and
36 Ibid., 199. 37 LEF required the same Irom poets and writers who could make their work scientiIic by relying on the advances made by Russian Formalists in studying the devices oI literature. For LEF, these devices became the tools oI verbal propaganda. In the Iirst issue oI LEF Osip Brik writes: "OPOYAZ studies the laws oI poetic production... OPOYAZ will help the art oI the proletariat, but not with vague conversations about the 'proletariat's spirit' and 'communist consciousness,' but with precise technical knowledge oI the devices oI contemporary poetic work." Osip Brik, "T.n. 'Iormalnyi metod'" (So called 'Iormal method'), LEF 1 (1923): 214-215. The meaningless experimental poetry oI Futurists which originally made Formalists recognize the "independent value oI the poetic sound" now became Iull oI meaning; the techniques oI "poetic instrumentation" were now seen as the techniques oI organizing the consciousness oI the masses. 38 I. Matsa, Sovetskoe iskusstvo za 15 let, 279. 39 Ibid., 281.
LEF. Aleksey Gastev, a leading Iigure oI Proletcult as well as oI the Soviet Taylorist movement who in 1921 organized the Central Institute oI Labor coined the phrase "social engineering" to describe the process oI the radical reconstruction oI the human psyche. 40 Tretyakov applied the metaphor to art: "Along with man oI science the art worker should become psycho-engineer, psycho-constructor." 41
The crucial question then becomes how to rationalize visual eIIectivity. In principle, diIIerent approaches are possible: to create particular conditions oI reception; to limit the number oI available images in the public sphere (this was done later in the period oI Socialist Realism); and Iinally, to create a particular code, a new visual language guaranteed to communicate eIIectively and eIIiciently to mass audiences (oIten, in the case oI Soviet Russia oI the 1920s, illiterate). It is the last solution which was explored by LeItist artists such as Rodchenko or Lissitsky. Why did the notion oI a visual language consisting oI simple abstract Iorms become attractive? First, abstract geometric Iorms were neither associated with bourgeois Iigurative representation nor did they look like the "subjective" abstract improvisations oI Kandinsky. Secondly, as Igor Golomstock points out, abstraction was justiIied by emphasizing its stylistic origin in truly popular decorative and religious art (Ior instance, icons). 42 Thirdly, simple Iorms were proclaimed to be eIIicient in a situation oI mass communication. Thus, Lissitsky argued: "The most unambiguous and immediately recognizable Iorms are geometric Iorms. No one will conIuse a rectangle with a circle, or a circle with a triangle." 43 Finally and most importantly, such Iorms were imagined to be perIectly suited Ior the purpose oI controlling the eIIect in the viewer -- both the idea and the imagery being prepared by decades oI investigations in the
40 Igor Golomstock, Totalitarian Art (New York: HarperCollins Publishers, 1990), 26. 41 Tretyakov, "Otkuda i kuda?," 202. 42 Golomstock, Totalitarian Art, 26. 43 Qtd. in Golomstock, Totalitarian Art, 24.
psychology oI perception and in scientiIic aesthetics. The tradition oI research into the eIIectiveness oI visual elements transIormed into the dream oI mass "psychophysical culture" advanced in Soviet Russia in the 1920s. The most striking record oI this dream is the paper "Engineerism" which was presented at one oI INKhUK's meetings. The paper starts with an analysis oI modernism as a reduction oI visual experience to its elements: "In the Iinal stage oI its development art has renounced representation and has moved toward visual sensations as such." 44 The trend continues with abstract art, the aim oI which, according to the author, is "the organization oI visual perceptions." Bourgeois art cannot go beyond this stage oI analysis, but in the Iirst proletarian state, this analytic work will become the basis oI a new visual culture oI social control: "Visual sensations as such will concertedly shape the human being as an organized unit, with the tempo oI something which belongs only to its own time." 45
From this perspective, we also should not be surprised to Iind in the same period the recurrent attempts to establish various graphic alphabets and dictionaries. The poet Velemir Khlebnikov in his 1919 article "Artists oI the World" postulated two types oI a universal visual code -- graphic signs and color equivalents. In the already mentioned INKhUK program, Kandinsky proposed to create a Iull dictionary oI colors, Iorms and their combinations with descriptions oI the psychological eIIect oI each element. Pavel Florensky together with his students at VKhUTEMAS started to compose "Symbolarium" ("The Dictionary oI Symbols") in 1922. 46 This "international and ahistorical dictionary" should have contained all cultural meanings oI every graphic symbol. In the introductory article oI "Symbolarium" Florensky regretted that while whole academies were busy preparing and preserving the dictionaries oI
44 Allison, Art Into LiIe, 79. 45 Ibid., 81. 46 P.A. Florensky, (1922), "Symbolarium (Slovar simvolov). Predislovie" |Symbolarium (The dictionary oI symbols). Introduction|, Trudy po znakovym sistemam V (Tartu: Uchebnye zapiski Tarturskogo Universiteta 284, 1971): 521-527.
national languages, no work yet has been done in composing a similar dictionary oI ideograms -- the "universal language oI mankind." 47 Similar projects oI a universal graphic alphabet were advanced in the same years by Ivan Puni, Petr Miturich, and other Russian artists. 48
To summarize: early Soviet attempts to theorize and put into practice a scientiIically grounded visual language oI mass communication bring together artistic exploration oI visual elements and research in experimental psychology. While the desire oI Soviet artists to emulate "scientiIic" Marxism and to adopt Taylorism and Fordism were already suIIicient reasons to pursue the science oI elements oI art, the ultimate justiIication Ior such a science was provided by experimental psychology and scientiIic aesthetics (Blanc, Henry, and others). Their investigations oI the psychological eIIects oI basic colors and simple Iorms prepared the idea oI a language oI mass communication composed Irom simple elements. In theory, this model called to enumerate visual elements and to describe their eIIects in order to compose dictionaries oI "visual eIIectiveness." In practice, it translated into the creation oI works composed Irom distinct and simple abstract Iorms. From 1921 to 1929 the psychological studies oI viewers' responses to visual Iorms were actively conducted in the central art institutes. In 1926 GAKhN created a psychophysiology laboratory, a psychophysics department, and laboratories Ior experimental aesthetics and art theory. 49 In order to arrive at the general laws oI the eIIectiveness oI visual Iorms the investigators have systematically analyzed the basic dimensions oI volume, line, color, and
47 Ibid., 527, 523. 48 Molok, "'Slovar simvolov' Pavla Florenskogo," 322. 49 E.L. Beliajeva, Arkhitekturno-prostranstvennaya sreda goroda kak obyekt zritelnogo vospriyatiya (Architectural and spacial city environment as an object oI visual perception) (Moscow: Stroyizdat, 1977), 13. 4. VisuaI Atomism and the Mind-Body ProbIem
texture. It was hoped that the results oI the analysis would put visual communication on a scientiIic basis. Jack Chen, a Communist British artist trained in the Soviet Union during the 1920s, wrote about this program oI art production based on psychological research:
Art must be a science, an industry. Pictures and sculptures should be constructed according to exact scientiIic principles aIter colors and Iorms had been classiIied according to their human reaction values...A picture according to the LeIt was really nothing but a "machine" Ior generating certain predetermined human reactions. Artists should be engineers oI Iorm and color. 50
There does not appear to be a disagreement among the LeIt as to the Iunction oI art -- to create predetermined viewer responses. But how was this to be achieved? In other words, assuming that we have isolated the elements oI a visual code, how should these elements aIIect the viewer? Do they induce emotions, do they communicate ideas or do they control behavior in a more direct way? To answer this question, Soviet artists oI the 1920s relied heavily on the psychological theories oI the day. 51 It is possible to discern two distinct trends in how psychology was used. In the Iirst trend, autonomy was granted to the human psyche. The second trend privileged physiology, considering the mind a neurological organ not diIIerent Irom the rest oI the organism. This trend was represented by such inIluential scientists as Bekhterev and Pavlov, with their studies oI physiology, reIlexology, and conditioning. The latter model was also prevalent in Soviet Marxist psychology in the 1920s. In 1924 it was challenged by Lev Vygotsky, a young psychologist and a Iormer literary critic. Recognized today as the Iorerunner oI cognitive psychology, he argued that the higher cognitive Iunctions are qualitatively diIIerent Irom the lower physiological processes and require diIIerent
50 Jack Chen, Soviet Art and Artists (London: The Pilot Press, Ltd., 1945), 58. 51 For the history oI Soviet psychology in the 1920s, I have consulted David Joravsky, Russian Psychology: A Critical History (Cambridge, Mass.: Basil Blackwell, 1989); Alex Kozulin, Psychology in Utopia: Toward a Social History oI Soviet Psychology (Cambridge: The MIT Press, 1984).
investigative approaches. In his Iamous address to the Second Russian Psychoneurological Congress, Vygotsky declared human consciousness to be a Iundamental problem in the psychology oI behavior, claiming that it cannot be understood through the study oI reIlexes. SigniIicantly, the turning point oI Vygotsky's career Irom literary critic to psychologist was his 1924 dissertation Psychology oI Art. Through the analysis oI Hamlet Vygotsky attempted to demonstrate that art is a product oI the capacities oI the speciIically human mind -- imagination, emotion, and symbolic representation. Crucial Ior Vygotsky's theory was the study oI the Iormal organization oI works oI art:
...every work oI art is considered to be a system oI stimuli, which are organized consciously and intentionally in order to cause an aesthetic reaction. This way we reconstruct the structure oI the reaction while analyzing the structure oI the stimuli...The general direction oI this method can be summarized in the Iollowing Iormula: Irom the Iorm oI the art work through the Iunctional analysis oI its elements and structure to the reconstruction oI the aesthetic reaction and the discovery oI its general laws. 52
In Vygotsky's Iormula the given was the existing works oI art and the unknown was the laws oI aesthetic reaction. As we have already seen, Ior the artists, designers, and Iilm directors oI the time the Iormula was reversed. II Ior the psychologist Vygotsky, visual works represented a reservoir oI knowledge about the human mind, artists, on the contrary, were hoping to utilize the objective psychological knowledge in order to create visual works which would produce pre- determined responses in the viewer. Contemporary psychologists -- the supporters oI Pavlov and Bekhterev on the one hand, and younger psychologists such as Vygotsky and A. Luria on the other hand, argued over whether the mind could be reduced to the laws oI physiology. Similarly, the artists in theory and in practice drew on these two alternative psychological models. Sergei Eisenstein's attempts to ground his Iilmmaking methods in diIIerent psychological theories epitomizes the two models. In one oI his latest written works Eisenstein summed up his work in Iilm:
52 Lev Vygotsky, Psikhologija Iskusstva (Psychology oI art) (Moscow: 1968), 39-41.
I never intended to "reIlect" the existing reality. I had one task -- using the means oI its inIluence -- to aIIect the Ieelings and thoughts, to inIluence the psyche and to shape the consciousness oI the viewer in the desired, required, and chosen direction. 53
Although Eisenstein's goal remained the same, the means oI achieving the desired eIIect were conceived oI diIIerently throughout his liIe. On the one hand, Eisenstein developed the concept oI "intellectual montage," privileging a purely intellectual response. The goal oI cinema, in this view, was thought to induce dialectical reasoning. The structure oI the Iilm itselI was conceived oI as dialectical thinking in visual Iorm with montage being the means oI representing the dialectical process through the contrast and juxtaposition oI images. This preoccupation oI Eisenstein with cinema as the visualization oI the work oI the intellect culminated in his project to create a screen adaptation oI Marx's Capital. 54
On the other hand, Eisenstein's second important concept developed in the Montage oI Attractions takes the viewers' physiological reaction as a point oI departure. 55 Eisenstein based his concept on the psychological theories oI Ludwig Klages and William James. 56 According to Klages, in a human being emotional states are expressed through bodily movements. Klages also insisted that human expressivity is characterized by a unique quality -- the muscular contractions oI one person are involuntarily repeated by the observer. James' theory was related to Klages' but causally reversed. He postulated that emotions were the eIIect oI muscular contractions -- one does not cry because he is sad, but one becomes sad due to crying. Eisenstein combined the two theories: the emotional state oI the actor translates into his muscular movements; these
53 Qtd. in Yu. A. Vasilieva, "Eisenstein: Kontseptsija 'Agressivnogo Iskusstva'" (Eisenstein: the concept oI 'aggressive art'), Kinovedcheskije Zapiski 3 (1989): 207. 54 I will return to Eisenstein's "intellectual montage" in chapter 2. 55 Sergei Eisenstein, "Montazh atraktsionov" (Montage oI attractions), LEF 3 (1923): 70-75. 56 Olga Bulgakova, "Sergei Eisenshtein i ego 'psikhologicheskiy Berlin' -- mezhdu psikhoanalizom i strukturnoy psikhologiey" (Sergei Eisenstein and his 'psychological Berlin': between psychoanalysis and structural psychology), Kinovedcheskie Zapiski 2 (1988): 178-80.
movements are involuntarily repeated by the viewer causing him to experience similar emotions. The important issue, then, was the training oI actors in simulating precise gestures and Iacial expressions in order to produce a desired emotional response in the viewer. Eisenstein proposes to develop a collection oI emotional stimuli which would be strung together in a Iilm -- "montage oI attractions." The Iilm becomes a script oI the emotional responses oI viewers. The two concepts -- the intellectual montage and montage oI attractions -- are the rare summits in Eisenstein's intellectual career where the opposing tendencies oI his thought concerning the modes oI cinematic inIluence on the viewer can be clearly distinguished. Most oI the time they seem to have been intertwined, appearing simultaneously in his changing conception oI cinematic eIIectivity. Similarly, in the cultural imagination oI the 1920s, the two tendencies appear to be running side by side. Is there a mind, a speciIically human autonomous cognitive apparatus, with operational laws not reducible to simple reIlexes? Or can human behavior be controlled by aIIecting the organism through a set oI predictable physiological stimuli? This second position received its extreme elaboration in the theory developed by Emmanuel Enchmen, an early Socialist revolutionary. 57 Enchmen's "theory oI new biology" was based on Pavlovian reIlexology and proposed a purely proletarian culture based directly on conditioned reIlexes, without the necessity oI conventional language or thought. The new proletarian language would consist oI conditioned grunts and gestures and replace the verbal culture oI bourgeois society. All in all, the co-existence oI the two schools in Soviet psychology lead to the co- existence oI the two kinds oI models oI artistic aIIectivity: aIIecting consciousness by communicating ideas and bypassing consciousness altogether. This dichotomy is reIlected, Ior instance, in the Iollowing statement Irom "New Artistic Program oI Proletcult" published in
57 Joravsky, Russian Psychology, 212.
1923: "Art should become the necessary part oI everyday liIe both in its active-representational Iorms (poster, advertisement, agitational theater, cinema) and in its material-organizational Iorms (psychophysical culture, organization oI mass happenings, processions and demonstrations, construction oI things)." 58 The opposition "representational" -- "organizing" reIlects two views oI aIIectivity: aIIecting the mind and aIIecting the body. These diIIerent models oI aIIectivity were also used to understand the roles oI visual elements which were being "isolated" and "puriIied" in the laboratories oI the 1920s. The elements could have aIIected the viewer in more than one way. No longer just catalysts oI basic emotional states, as in the tradition oI experimental psychology and scientiIic aesthetics, they were now re-interpreted as capable oI carrying complex messages addressed to the mind, and as simple signals designed to trigger physiological and behavioral responses.
I have discussed diIIerent models oI aIIectivity which underlined the research into the elements oI a visual code. But what kind oI code can be Iormed Irom such elements? Semiotician Luis Prieto distinguished between codes with and without articulation. 59 A code with articulation consists oI a number oI elements combined to produce messages. The examples are human language, playing cards, and maritime Ilag codes. The crucial Ieature oI any oI these codes is that they allow Ior the generation oI a large or even inIinite number oI messages Irom a small set oI elements. For instance, in the case oI human language, its speaker can generate an inIinite number oI sentences. In contrast, in codes without articulation there are as many meanings as elements. For example, in the language oI Ilowers, each Ilower signiIies a
58 Matsa, Sovetskoe iskusstvo za 15 let, 261. 59 Luis Prieto, Messages et signaux (Paris, PUF, 1966). 5. VisuaI Esperanto
distinct meaning (Ior instance,"red rose" "victory is yours"); a combination oI Ilowers does not create any new meanings. In view oI this distinction, the research into visual elements could justiIy both a code with articulation as well as a code without articulation. In other words, it is possible to understand the elements in two ways: as carrying a complete message (a verbal meaning, an emotion or a particular behavioral response), or as just a part oI the message, like a word in a sentence, combined with other words through grammar. The artistic tradition oI visual atomism, Irom Blanc to Kandinsky, appears to Iavor the Iirst interpretation. In their writings, Blanc, Henry, and Seurat correlated the orientation oI straight lines with diIIerent emotions such as gaiety or sadness. Kandinsky's Iamous questionnaire, compiled in 1920 while he was active in INKhUK, also indicates that he considered the elements to directly correspond to emotions or verbal meanings: "How do you see a triangle -- do you Ieel that it is moving, where, do you see it as more witty than square; is the sensation Irom a square similar to the sensation Irom a lemon; what is a canary's song more like - - a triangle or a square." 60
A code without an articulation requires as many signs as there are potential messages and thereIore is not eIIicient. Yet, it is this ineIIiciency which makes it ultimately suitable Ior new communicative situations Iaced by artists and designers in the earlier decades oI this century. TraIIic signs, the labels on instrument panels, trademarks had to be unambiguous and instantly recognizable by an untrained eye. Not surprisingly, the designers in every Iield have adopted simple abstract Iorms. Thus, in 1909, when Mondrian and Kandinsky were moving toward their Iirst abstractions, the Iirst International ConIerence on the Regulation oI Automobile TraIIic took
60 Qtd. in Khan-Magomedov, "INKhUK," 346.
place in Paris. Following its recommendations, in the 1910s-20s, universal pictograms signiIying main road situations were developed and employed. 61
For the Soviet "psycho-engineers," the notion oI a visual code without articulation, which correlates each message with a separate sign, had a particular attraction. The behaviorist psychologies oI Pavlov and Bekhterev popularized the notion oI directly controlling behavior through conditioned reIlexes. Simple visual Iorms could have acted as such stimuli, controlling "the proletarian zombie" as the lights and whistles already controlled a dog in Pavlov's laboratory. In this interpretation, the dictionary oI visual elements would contain not a set oI meanings or emotions but a set oI commands -- previously learned responses to stimuli. This is what the author oI the "Engineerism" paper seems to have in mind when he writes oI the Iuture where "visual sensations as such will concertedly shape the human being as an organized unit." In other words, just as the motorists in the 1920s were beginning to memorize the international traIIic signs, learning how to respond to them automatically (today, this is the closest most oI us come to experience conditioning through a visual code), artists took this as a paradigm Ior controlling the masses through visual communication. Yet, while suitable Ior communication oI basic emotional states or a set oI behavioral stimuli, the code without articulation is not practical Ior communicating an arbitrary large number oI messages. II visual language is to compete with a verbal language, and even to replace it, visual Iorms have to be interpreted as equivalents oI letters or words. They also have to be supplemented by grammar, the rules oI how these visual "words" are combined to Iorm sentences. Indeed, contemporary debates on whether such animals as dolphins or primates can use human language center around the attempts to teach them to use visual signs in this way. It is
61 A.S. Sardarov, "5000 let evolutsii doroznogo znaka" (5000 years oI the evolution oI a traIIic sign), Tekhnicheskaya Estetika 9 (1984): 14-19.
easy to train animals to respond to visual Iorms as distinct messages: a square means "take Iood," a circle means "jump," and so on. The question is whether animals can string such symbols together to Iorm new messages, thus treating the individual symbols as words rather than as complete messages. At least a Iew researchers have claimed that they have taught some animals to do so, thus teaching them a visual code with articulation. What about humans? The idea oI replacing particular verbal language by a universal visual language -- more eIIicient, capable oI reaching new masses oI immigrants (U.S.) or "proletarians oI the whole world" (U.S.S.R.) was oIten expressed in the 1920s. Film, in particular, was seen as the prime candidate Ior a visual Esperanto. In a 1921 interview D.W. GriIIith said: "A picture is the universal symbol, and a picture that moves is a universal language. Moving pictures, someone suggests, 'might have saved the situation when the Tower oI Babel was built.'" 62 Miriam Hansen writes about the metaphor oI Iilm as visual Esperanto in the U.S.:
GriIIith participated in the widespread celebration oI Iilm as a new "universal language" which accompanied its Iormation as an institution. Used by journalists, intellectuals, social workers, clergy, producers, and industrial apologists alike, the metaphor oI Iilm as a universal language drew on a variety oI discourses (Enlightenment, nineteenth-century positivism, Protestant millennialism, the Esperanto movement, and the growing advertising industry) and oscillated accordingly between utopian and totalitarian implications. 63
However, it was across the Atlantic, in the U.S.S.R., where the comparison between Iilm and verbal language was explored most systematically by Victor Shklovsky and other literary theorists associated with the so-called Iormalist school as well as by Iilm directors and theorists such as Eisenstein, Vsevolod Pudovkin, and Lev Kuleshov. In their writings we encounter
62 Qtd. in Miriam Hansen, "The Hieroglyph and the Whore: D.W. GriIIith's Intolerance," The South Atlantic Quarterly 88, no. 2 (1989): 363. 63 Ibid., 362.
Irequent comparisons oI a shot and a word, or a sequence and a sentence as well as reIerences to a "grammar oI Iilm." 64
Can we Iind similar kinds oI analogies proposed Ior the elements oI a single image? Just as the artists were comIortable with diIIerent mutually exclusive models oI aIIectivity, they were equally at ease with two contradictory models oI a code. II the visual elements were imagined to be capable oI carrying complete messages, they were also thought oI as the elements oI a message. Thus, art historian S.O. Khan-Magomedov writes about the INKhUK program: "The point was to discover certain primary elements oI artistic expressiveness, in themselves without signiIication and understood as an alphabet oI artistic-compositional system." 65
What we don't Iind in the 1920s, however, is equally serious thinking about the grammar oI a still image. It was only in the 1960s, when the analogy between a verbal and a visual language was taken with new rigor in structural semiotics, that coherent theorizing about the grammar oI visual elements began to appear. 66
The understanding oI vision in modernity is characterized by attempts to think oI vision as a language. This is reIlected in the popularity oI the term "visual language," this term which is so Iamiliar to us today that we tend not to ask about its historical speciIicity. What this term points
64 Paradoxically, the classical Hollywood Iilm style, which matured by the 1920s, did become the universal visual language oI the twentieth century, while the Iilms oI Soviet avant-garde directors, who were most systematic in theorizing Iilm as a language, were rejected by mass audiences and continue to play art houses, as they did in the 1920s. 65 Khan-Magomedov, "INKhUK," 343. 66 The best critical review oI visual semiotics available in English is Gsran Sonesson, Pictorial Concepts. Inquiries into the Semiotic Heritage and its Relevance Ior the Analysis oI the Visual World (Lund, Sweden: Lund University Press, 1989). 6. ConcIusion
to, in my view, is the new social role oI vision as the medium oI mass communication which it acquired in the earlier decades oI this century. The attempts to rationalize vision in this role, to conceive oI it as a code which can Iunction independently oI verbal language and which can be eIIective in the new communicative situations created by modernization have centered around the quest Ior the elements, the "atoms" oI visual communication. This quest reaches its culmination in the work oI a number oI art institutes in Soviet Russia in the 1920s, where artists collaborate with experimental psychologists to isolate these elements and study their eIIectiveness. The search Ior the essential elements oI each art which preoccupied early twentieth century culture is usually seen in the context oI artistic modernism. This search is interpreted in the context oI the rhetoric oI the purity oI medium. Indeed, especially in the 1910s-20s, artists tried to reduce every medium to its unique qualities. To do so, they gave up representation and concentrated on the material elements thought to be unique to each medium. Poets, such as Russian Iuturists, were experimenting with sounds; Iilmmakers proposed that the essence oI cinema was movement and temporal rhythm (French Iilm theory oI the 1920s) 67 or montage (Kuleshov's group in Russia); and painters were exploring pure colors and geometric Iorms. The Iollowing statement made in 1924 by Jean Epstein, a French avant-garde Iilmmaker and theoretician, is typical oI modernist rhetoric oI purity; countless statements like it appeared on the pages oI avant-garde publications oI the time:
For every art builds its Iorbidden city, its own exclusive domain, autonomous, speciIic and hostile to anything that does not belong. Astonishing to relate, literature must Iirst and Ioremost be literary; the theater, theatrical; painting, pictorial; and the cinema, cinematic. Painting today is Ireeing itselI Irom many oI its representational and narrative concerns...And any literature worthy oI the name turns its back on those twists and turns oI plot which lead to the detective's discovery
67 See especially Jean Epstein, "On Certain Characteristics oI PhotogZnie," in French Film Theory and Criticism, ed. Richard Abel (Princeton: University oI Princeton Press, 1988), 1: 314- 318; Germaine Dulac, "Aesthetics, Obstacles, Integral CinZgraphie," in French Film Theory and Criticism, 1: 389-397.
oI the lost treasure...The cinema must seek to become, gradually and in the end uniquely, cinematic; to employ, in other words, only photogenic elements. 68
Yet, as this chapter has suggested, more was at stake in the quest Ior the visual "table oI elements" than just the demand Ior purity oI medium. While many artists searched Ior "ever more rich aesthetic elements" (Henry), trying to make their art into a "Iorbidden city" (Epstein), others welcomed the opportunity to take their art to the streets. For these artists, the science oI visual elements has provided hope Ior the creation oI a Iully rationalized visual language oI mass communication. The dream to rationalize communication was not unique to this project -- the proposals Ior a universal language go back to the seventeenth century. Descartes suggested that the lexicon oI a universal language should be composed oI primitive elements. By systematically combining these elements, one can generate "an inIinity oI diIIerent words." 69 In the early eighteenth century Leibniz outlined a language in which grammatical and logical structure would coincide, thus making possible the automation oI thinking. The basic elements oI his ideal language were characters representing unambiguously a limited number oI elementary concepts. Leibniz called the inventory oI these concepts "the alphabet oI human thought." Through the algebra oI thought primitive concepts would be combined to Iorm any complex idea. 70 The proposal oI a universal language composed Irom graphic symbols and thereIore suitable only Ior writing Iollowed soon with the Iirst pure pasigraphy ("writing Ior all") invented by Joseph de Maimieux in 1797. 71
However, until the twentieth century these remained isolated philosophical ideas advanced by single individuals. With the rise oI mass communications, the Iantasy oI a universal rational language attains new signiIicance, extending into the realm oI vision. Thus, what we now
68 Ibid., 314-15. 69 Qtd. in WinIried Nsth, Handbook oI Semiotics (Bloomington and Indianapolis: Indiana University Press, 1990), 272. 70 Ibid., 274. 71 Ibid., 269.
witness are systematic and extensive attempts to establish Iully rationalized visual languages oI mass communication. At the same time, the studies oI behavior and conditioning in physiology, behaviorism and reIlexology, lead to new interpretations oI the Iunctioning oI the elements oI a visual Esperanto. Not only would they be capable oI communicating ideas thus aIIecting the intellect, but they would also directly control behavior bypassing the mind altogether, thus making possible an international "psychophysical culture."
In the sixth Meditation Descartes deIined man "as a thing that thinks," to whom reasoning comes naturally. Imagination, on the other hand, requires a special eIIort and is in no way necessary Ior a human being. Distinguishing between imagination and pure intellection, Descartes demonstrated the inIeriority oI vision to reasoning:
For example, when I imagine a triangle I not only conceive (intelligo) that it is a Iigure comprehended by three lines, but at the same time also I look upon (intueor) these three lines as present by the power and internal application oI my mind (acie mentis), and this is what I call imagining. But iI I desire to think oI a chiliagon, I indeed rightly conceive that it is a Iigure composed oI a thousand sides, as easily as I conceive that a triangle is a Iigure composed oI only three sides; but I cannot imagine the thousand sides oI a chiliagon as I do the three sides oI a triangle, nor, so to speak, view them as present |with the eyes oI my mind|. 72
Indeed, how can humans possibly reason through visual representations, iI the latter are not capable oI representing general concepts to begin with? In Treatise Concerning the Principles oI Human Knowledge, George Berkeley insisted, Ior instance, that it is impossible to have a mental image oI an idea, such as "man," as a generality; it is only possible to visualize a tall or a short man, but not man as such. Thinking deals with generalities and does not tolerate particular things. II a philosopher tries to reason about the nature oI "man," any image oI a particular man would mislead the reasoning process. 73
The philosophical arguments against visual reasoning still continue in this century, but they are out oI step with the modern productive use oI vision: not only the concepts oI "visual
72 RenZ Descartes, "Meditations on the First Philosophy," in The Rationalists, trans. John Veitch (Garden City, NY: Doubleday & Company, 1960), 160-161. 73 RudolI Arnheim, "A Plea Ior Visual Thinking," in New Essays on the Psychology oI Art (Los Angeles and London: University oI CaliIornia Press, 1986), 136. Chapter 2. I See, Therefore I Think 1. Introduction
thinking," "thinking through images," etc. have become respectable and commonplace, 74 but they are also put into practice daily through new techniques and technologies oI visual representation. While Descartes thought that it was impossible (and, at any rate, unwise) to imagine the thousands oI sides oI a chiliagon, today scientists stare at computer monitors, where computer imaging brings beIore their eyes not only chiliagons but geometric objects oI n- dimensions, objects composed not Irom thousands but Irom millions oI sides, objects representing weather systems, diIIraction patterns, atomic surIaces, and other processes inconceivable to the bare imagination (Iig. 5). 75 "ScientiIic visualization" is the name oI the new Iield which claims to transIorm the nature oI scientiIic reasoning in biology, chemistry, and even mathematics. Likewise, Berkeley's insistence that images cannot convey "man" as a generality has been superseded. Since the 1960s, numerous space stations sent to the Moon, Venus, Mars or simply into the inIinity oI the Universe have carried images destined to communicate to intelligent beings (should they happen to Iind the stations) the basics oI human civilization and the "essence oI man" in a "universal language oI images." Long aIter the sun explodes and man disappears, space stations will continue to Ily on their trajectories, carrying such images as the Iamous drawing by Leonardo da Vinci oI an "ideal man." Intelligent (or perhaps, not so intelligent) beings will eventually discover these images, learning Irom them iI not about human civilization in general, then at least about the artistic tastes oI administrators oI space programs, assuring Leonardo oI inter-galactic Iame. The images oI general or abstract concepts can be encountered not only on space missions but on Earth, everyday and everywhere, in numerous logotypes and designs (Ior
74 As witnessed, Ior instance, by such titles as Thinking Eye (Paul Klee) or Visual Thinking (RudolI Arnheim). 75 Examples oI scientiIic visualizations presented at the Showcase at SIGGRAPH 1992. SIGGRAPH '92 Final Program (New York: The Association Ior Computing Machinery, 1992), 33-40.
instance, in symbols which distinguish men's Irom women's lavatories), or in the representation oI unimaginable data sets. However, these examples do not tell us anything about the speciIicity oI the new Iunction oI vision in modernity. AIter all, Ior centuries pictures have been employed as visualizations oI the holy stories. Michael Baxandall cites a late thirteenth-century text which summarizes this purpose oI images:
Know that there were three reasons Ior the institution oI images in churches. First, Ior the instruction oI simple people, because they are instructed by them as iI by books. Second, so that the mystery oI the incarnation and the examples oI the Saints may be more active in our memory through being presented daily to our eyes. Third, to excite Ieelings oI devotion, these being aroused more eIIectively by things seen than by things heard. 76
ThereIore, what is diIIerent about the modern Iunction oI vision is not simply the use oI images to represent abstract concepts concretely. What is diIIerent is the idea oI reasoning through images, the idea unthinkable Ior Descartes or Berkeley. Thus, the Iield oI scientiIic visualization is based on the assumption that visual observation oI computer generated images oI data sets and processes leads to scientiIic breakthroughs otherwise impossible. Richard Mark FriedhoII and William Benson propose that "using a computer simulation...even a beginning student oI chemistry might be able to deduce benzene's structure" (a reIerence to Friedrich KebulZ's 1865 discovery oI the molecular stucture oI benzene supposedly made aIter he saw a dream involving a snake biting its own tale). 77 Similarly, Jaron Lanier, one oI the most visible researchers/promoters oI virtual reality, maintains that virtual reality will lead to the age oI "post- symbolic communication," the communication oI ideas and arguments solely through images. 78
Finally, a prominent cognitive psychologist Philip Johnson-Laird claims that everyday logical
76 Michael Baxandall, Painting and Experience in FiIteenth Century Italy (OxIord: OxIord University Press, 1972), 41. 77 Richard Mark FriedhoII and William Benson, The Second Computer Revolution: Visualization (W.H. Freeman and Company, 1991), 13. 78 See, Ior instance, Timothy Druckrey, "Revenge oI the Nerds. An Interview with Jaron Lanier," AIterimage (May 1991): 5-9.
reasoning takes place through images: when we are engaged in reasoning, we construct a mental model oI a situation which represents its essential Ieatures in the Iorm oI topological relations. 79
BeIore it became possible to send Leonardo's images to remote corners oI the Universe to represent "humanity," beIore scientists accepted that computer-aided visualization would dramatically speed up scientiIic reasoning, beIore psychologists could claim that reasoning in humans is a matter oI manipulating mental images, a proIound change in the cultural attitude toward vision had to take place. This change occurred between the 1870s and the 1920s. Logical thinking entails the representation oI abstract ideas and, also, oI the logical relations between them. During this period, Galton, Venn, Freud, Eisenstein, and others put vision to use Ior both purposes. More importantly, Ior the Iirst time, we can Iind in their work the explicit justiIications Ior the very notion oI reasoning through vision. In 1877 Galton's composite portraits visualized universal human types; in 1880, John Venn made public a method Ior solving problems in logic by using graphic diagrams (section 3). In The Interpretation oI Dreams, published in 1900, Sigmund Freud developed a systematic theory oI how abstract notions and logical Iorms can be visualized (section 5). In the late 1920s Eisenstein proposed that Iilm can be used "to encourage and direct the whole thought process" (section 4). These diIIerent models oI how vision can be used in reasoning represent the Iirst stage in the reversal oI attitude towards the inadequacy oI vision. The second stage arrived when, in post- industrial society, the concern with the eIIiciency oI the mind overwhelmed the concern with the eIIiciency oI the body. Now, the questions oI how inIormation can be coded, stored, retrieved, and processed more eIIiciently give direction to the study oI the mind, including the use oI visual
79 Philip Johnson-Laird, Mental Models: Towards a Cognitive Science oI Language, InIerence, and Consciousness (Cambridge: Cambridge University Press, 1983).
representations in mental processes. Cognitive scientists, among others, begin to approach vision Irom this new perspective, and they postulate the centrality oI visual representations Ior human reasoning, because oI their eIIiciency (sections 6 and 7).
To be able to appreciate the novelty oI the modern paradigm oI visual reasoning, I will begin by discussing the classic philosophical positions on the relationship between vision and reason. Many philosophers thought vision to be unsuitable Ior reasoning, Iirst oI all because throughout the history oI philosophy, reason was closely associated with logic. II, Iollowing its Iormalization by George Boole and Gottlob Frege in the nineteenth century, logic has developed into a sophisticated branch oI mathematics, today seemingly removed Irom our everyday discourse, originally it was conceived by Aristotle as a systematization oI principles according to which rational debates should be conducted. 80 ThereIore, it was Aristotle who originated the tradition oI discussing human reasoning and logic interchangeably. Locke, Ior instance deIines reason as "that Iaculty whereby man is supposed to be distinguished Irom beasts, and wherein it is evident he much surpasses them." Claiming that the highest order oI reason consists in "the discovering and Iinding out oI prooIs," Locke writes:
Sense and intuition reach but a very little way. The greatest part oI our knowledge depends upon deductions and intermediate ideas: and in those cases where we are Iain to substitute assent instead oI knowledge, and take propositions Ior true without being certain they are so, we have need to Iind out, examine, and compare the grounds oI their probability. 81
80 Antony Flew, ed., A Dictionary oI Philosophy (London: Pan Books Ltd., 1984), 208-212. 81 John Locke, An Essay Concerning Human Understanding, ed. A.S. Pringle-Pattison (OxIord: Clarendon Press, 1924), Book IV, Ch. 17. 2. I See, Therefore I Think
The equation oI reason with logic persisted late into the twentieth century. Piaget assumed that the Iinal stage in the development oI human thought processes was a kind oI Iormal logic analysis. Equally, Levi-Strauss's binary oppositions and Greimas's semiotic square are based on stripped down Iormal logic. This logic is postulated as the structure oI the "native mind" (Levi-Strauss) or the structure oI all signiIication (Greimas). 82 Only in the last decades have cognitive psychologists discovered that human judgments do not Iollow statistical probabilities and tend to ignore inIormation about the prior probability oI an event; 83 that rather than having some uniIorm "laws oI thought," humans employ a variety oI heterogeneous heuristics; and that in general the normative rules oI logic do not describe human thinking. 84
However, the idea oI a separation between logic as a normative or prescriptive discipline and human reasoning, which can be described and studied empirically and which does not Iollow any strict rules oI logic, is quite recent, and would be unimaginable Ior Aristotle or Locke. 85
ThereIore, in classical philosophy to reason is to carry out logical operations. How then can one possibly reason through images? How can one represent visually "all," "some," "any" or
82 What is diIIerent about reason as imagined by Levi-Strauss and Greimas in contrast to Locke is its new industrial eIIiciency: the concept oI binary oppositions was brought by Jakobson Irom MIT, the birthplace oI inIormation theory which demonstrated that binary code is the most economical oI all. According to GeoIIrey Sampson, "what led Jakobson to the hypothesis that all parameters are 'binary' seems to have been the mathematical notion that a transmission-code is more eIIicient when it uses only independent binary choices." GeoIIrey Sampson, Schools oI Linguistics (StanIord, CA: StanIord University Press, 1980), 250. 83 Gillian Cohen, The Psychology oI Cognition, 2nd ed. (London and New York: Academic Press, 1983), 190. 84 Michael I. Posner, and Gordon L. Shulman, "Cognitive Science," in The First Century oI Experimental Psychology, ed. Eliot Hearst (Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1979), 393. 85 A recent dictionary oI philosophy still equates thinking with logic: "The mental activity oI (a) theoretical contemplation directed toward some object with a view to reaching a propositional conclusion; or (b) practical deliberation directed toward some object with a view oI reaching a decision to act." Flew, A Dictionary oI Philosophy, 352. Emphasis mine - L.M. An encyclopedia oI philosophy is equally categorical: "particular thoughts have some kind oI logical Iorm; they may be categorical, hypothetical, disjunctive, universal, particular, and the like." Paul Edwards, ed., Encyclopedia oI Philosophy (New York: Macmillan, 1967), 8: 101.
other logical constants? How to represent truth and Ialsity? How to distinguish particular Irom general? Reason, oI course, cannot proceed without the raw materials, collected Ior it by vision. However, the role oI vision is limited to the passive and unintelligent collection oI data (sensations), later to be interpreted by reason. Experimental psychology, emerging Irom empiricist philosophy in the nineteenth century, gave additional support to this conceptual hierarchy by providing a topological scheme: the outside world enters through the Iilters oI the eyes and then is carried through visual pathways inside the head, to the brain, the site oI "higher mental Iaculties." Even today, psychological textbooks are organized around the same hierarchy, inherited Irom empiricist philosophy: they start with chapters on sensation (vision and other senses) and end with cognition (thinking, memory, text comprehension, etc.)
These are some oI the ways in which philosophy has demoted vision in relation to reason: it can't represent logical Iorms; it can't generalize; its only job is to collect the raw material reason needs. 86 However, this philosophical tradition is counterbalanced and upset by other traditions. For, while subordinating vision to reason, philosophers also oIten claimed that reason in Iact was born oI vision. Moreover, it can be said that they could only conceive oI reason in visual terms. For instance, while in terms oI the temporality oI everyday mental Iunctioning, vision serves reason, its task being simply the collection oI data Ior reason to work with, in terms oI a much longer temporality oI phylogenetic development oI the human race, vision is oIten postulated as reason's origin. In particular, Locke, Herder, Vico, Nietzsche, and others claimed that abstract categories oI language evolved Irom concrete visual (and also bodily) experience. Locke: "I doubt not but, iI we could trace them |words| to their sources, we should Iind, in all
86 In his Downcast Eyes, unavailable to me at the time oI this writing, Martin Jay discusses the role oI vision in modern French philosophy. Martin Jay, Downcast Eyes: the Denigration oI Vision in Twentieth-century French Thought (Berkeley: The University oI CaliIornia Press, 1993).
languages, the names which stand Ior things that Iall under our senses to have had their Iirst rise Irom sensible ideas." 87 Nietzsche: "Everything which makes man stand out in relieI against the animal depends on this Iaculty oI volatilizing the concrete metaphors into a schema and thereIore resolving perception into an idea." 88 Vico: "It is noteworthy that in all languages the greater part oI the expressions relating to inanimate things are Iormed by metaphor Irom the human body and its parts and Irom the human senses and passions." 89
Vico's ideas, in particular, have been recently revived in the work oI the linguist George LakoII. LakoII argues that most, iI not all, abstract semantic concepts and abstract human reasoning are based on metaphorical mapping oI spatial concepts (such as inside-outside, part- whole, and others) which are inherently meaningIul to human beings since they have bodies. Like Vico beIore him, LakoII challenges three traditional hierarchies -- between concrete and abstract thought, between vision and reasoning, and between literal and metaphorical meanings. In contrast to the objectivist-computational paradigm, which underlies much oI Anglo-American philosophy oI language and cognitive science, in which thinking is conceptualized as the algorithmic manipulation oI symbols meaningless in themselves, LakoII claims that in thinking we project the structures oI our sensory and bodily experience to abstract
87 Qtd. in Tzvetan Todorov, Theories oI the Symbol (Ithaca: Cornell University Press, 1984), 237. 88 Friedrich Nietzsche, "On Truth and Falsity in their Ultramoral Sense," in Early Greek Philosophy and Other Essays (New York: The Macmillan Company, 1924), 181. 89 Thomas Goddard Bergin and Max Harold Fisch, The New Science oI Giambattista Vico (Ithaca: Cornell University Press, 1968), 129. Discussing the metaphorical origin oI abstract concepts, Vico goes on to claim that the common assumption "that prose speech is proper speech, and poetic speech improper" reverses the true history oI human mind: "Iirst nations" have spoken and thought in poetry, i.e. through the metaphorical tropes (131).
situations. 90 The semantics oI natural language is built upon the meaningIul "image-schemes" which are the generalizations oI what LakoII calls our "basic human experience." We live inside our bodies; our bodies consist oI parts which are linked. ThereIore, we conceptualize all situations in terms oI containers (Ior instance, an enormous number oI expressions use in and out), part-whole relations (divorce is splitting up) and links (we make connections, break relations). For LakoII, metaphors such as these are not the exception but the norm in language,
90 George LakoII, "Cognitive Linguistics," Versus 44/45 (1986): 119-154.
which involves the extension oI the bodily and sensory experience to the domain oI abstract concepts. Because spatial (more precisely, topological) relations are an important part oI our experience, spatial metaphors are at the core oI the semantics oI human languages; reaching goals, moving out, walking over an enemy, etc. Not only does the semantics oI natural languages turn out to be based on spatial metaphors but abstract reason itselI is the manipulation oI spatially represented inIormation: "natural language reasoning makes use oI at least some unconscious and automatic image-based processes such as superimposing images, scanning them, Iocusing on part oI them, etc." 91 In Iact, reason is a special case oI more general mental operations which are needed Ior perception: "Abstract reasoning is a special case oI image-based reasoning. Image-based reasoning is Iundamental and abstract reasoning is image-based reasoning under a metaphorical projection to an abstract domain." 92
With this claim, LakoII transcends the earlier views oI Locke, Vico, and Nietzsche according to whom abstract reason phylogenetically evolved Irom vision. It is one thing to suggest that abstract concepts have evolved Irom "sensible ideas" (Locke) or perceptions (Nietzsche). It is quite diIIerent to start with the assumption, as LakoII does, that abstract ideas are nothing but a limited set oI visual (more precisely, topological) Iorms (such as inside-outside, part-whole) extended into the abstract domain. Here the modern obsession with the visualization oI reasoning leads to the most dramatic break with the traditional conceptualization oI the relation between reason and vision. Modern writers such as LakoII or Johnson-Laird, claim is that there is only one kind oI cognitive operations, which reason and vision share. These operations are Iundamental, original to vision: reason borrows these operations; it does not invent any oI its own.
91 Ibid., 149. 92 George LakoII, "The Invariance Hypothesis: Is Abstract Reason Based on Image-Schemas?" Cognitive Linguistics 1, no. 1 (1990), 65. Perhaps even this very brieI presentation oI LakoII's ideas already makes apparent a substantial problem oI his theory -- the rather unquestioned notion oI "basic human experience."
Johnson-Laird has proposed that when we are engaged in reasoning, we construct a mental model oI a situation which represents its essential Ieatures in the Iorm oI topological relations. 93 The conclusion then simply can be "read oII" Irom the representation. For instance, how can we see the truth oI a modus ponens: "II X is an A, and iI all A's are B's, then X is a B"? We construct a mental model: X is inside A; A is inside B. II the model is mentally "scanned," the right conclusion is invariably arrived at. According to Johnson-Laird, we don't have to
93 Philip Johnson-Laird, Mental Models: Towards a Cognitive Science oI Language, InIerence, and Consciousness (Cambridge: Cambridge University Press, 1983).
assume the existence oI any special rational logical competence unique to humans. Logical reasoning is just one consequence oI vision. But do these views oI LakoII and Johnson-Laird really represent a break? Perhaps this revolt against the condemnation oI vision in the philosophical tradition was prepared by this tradition itselI? For, while declaring vision unsuitable Ior reason, philosophers conceived oI reason in visual terms all along: witness the persistence oI the metaphors oI eye, light, mirror, revelation and unveiling, Irom Plato to Descartes, Hegel, Heideger, and Lacan as standing in Ior logos, knowledge, and truth. 94
"Idea" comes Irom the Greek verb "to see" and is Irequently linked to ancient optics and theories oI perception. According to Aristotle, in thinking ideas Iunction to substitute Ior the absent objects as their images: "The reasoning mind thinks its ideas in the Iorm oI images; and as the mind determines the objects it should pursue or avoid in terms oI these images, even in the
94 For gender implications oI this tradition, see Evelyn Fox Keller and Christine R. Grontkowski, "The Mind's Eye," in Discovering Reality: Feminist Perspectives on Epistemology, Metaphysics, Methodology, and Philosopohy oI Science, ed. Sandra Harding and Merrill B. Hintikka (London: D. Reidel Publishing Company): 207-224. For a discussion oI Lacan's concept oI the mirror stage in relation to the Western philosophical tradition with its concepts oI knowledge and subjectivity grounded in vision see "The Statue Man," in M. Borch-Jakobson, The Absolute Master (StanIord: StanIord University Press, 1991).
absence oI sensation, so it is simulated to action when occupied with them." 95 For Berkeley and Hume thinking involved a sequential series oI images; these images are tied to certain habits oI the mind to move Irom one image to another. Reason was equated with an eye which compared all objects oI thought laid out beIore it. As summarized by Jonathan Crary,
But whether it is Berkeley's divine signs oI God arrayed on a diaphanous plane, Locke's sensations 'imprinted' on a white page, or Leibniz's elastic screen, the eighteenth century observer conIronts a uniIied space oI order...on which contents oI the world can be studied and compared, known in terms oI a multitude oI relationships. In Rorty's words, "It is as iI the tabula rasa were perpetually under the gaze oI the unblinking Eye oI the Mind..." 96
95 Qtd. in Jean Matter Mandler and George Mandler, eds., Thinking: From Association to Gestalt (New York: John Wiley & Sons, Inc., 1964), 9. 96 Jonathan Crary, Techniques oI the Observer: on Vision and Modernity in the Nineteenth Century (Cambridge: The MIT Press, 1990), 55.
It was Heidegger who, in What is Called Thinking?, insisted on the visual basis oI Western philosophy (both epistemology and ontology) Irom Plato to Hegel most strongly. 97
"The Greeks...conceived knowledge as a kind oI seeing and viewing...Because Being means presence and permanence, 'seeing' is especially apt to serve as an explanation Ior the grasping oI
97 Martin Heidegger, What is Called Thinking? (New York: Harper & Row, Publishers, 1968).
what is present and what is permanent." 98 For Plato, Being is idea, present as outward appearance. However, already in Plato the seeds oI the split between object and subject are implanted: Being is presence but at the same time is what man brings beIore his eyes. From this moment on, idea is gradually transIormed into perception and representation, that, according to Heidegger, characterizes the Cartesian cogito. Descartes laid a new Ioundation in privileging the thinking subject, but the certitude oI the cogito derives Irom its visibility: as Heidegger explains, cogitatio means Vor-stellung, i.e. "the bringing-beIore-itselI and what-is-brought-beIore-itselI and made 'visible' in the widest sense." 99
Indeed, Cartesian rationality is based on the notion oI a subject which is outside himselI, observing himselI contemplating the subject. This is the price Descartes pays Ior his condemnation oI images and imagination in relation to reason: he can only conceive oI reason in visual terms. BeIore "I think thereIore I am" comes "I see thereIore I think."
98 Qtd. in Borch-Jakobson, The Absolute Master, 53. 99 Qtd. in Ibid., 54.
Like Siamese twins, vision and reason are inseparable in classical philosophy, being connected in a complex network oI dependencies, only a Iew oI which have been discussed above. On the one hand, vision is but a servant oI reason, unable to represent logical relations and abstract concepts and thereIore, clearly unsuitable Ior reasoning. On the other hand, vision achieves its revenge, consistently slipping in as the phylogenetic origin oI reason and as the basis oI metaphors in which operations oI reason are conceived. Regardless oI whether Heidegger's brilliant account oI some oI these dependencies represents a break with this classical tradition or is still a part oI it, he clearly lives in a diIIerent age. Vision and reason are no longer thought oI as antagonists; on the contrary, "visual reasoning" is now taken Ior granted and even promoted as a new highly eIIicient means oI decision making and/or propaganda, suitable Ior the age oI mass mobilization, mass education, industrialized science, and bureaucracy. Heidegger is a contemporary oI Eisenstein who, in line with Stalin's plans to industrialize Russia overnight, dreams oI turning workers into dialectical thinkers by passing them through the darkness oI special incubators oI cognitive skills -- movie houses, thus shortening the duration oI the learning process to hours. Heidegger is also a contemporary oI Benjamin who wants to teach German workers dialectical thinking even without any external aids (such as Iilm) -- by simply juxtaposing images, in montage-like Iashion, in the mind. Finally, Heidegger's younger compatriot, RudolI Arnheim, becomes the most outspoken champion oI thinking in images, arguing that artistic skills should become the 3. First Signs of RevoIt: Venn and GaIton
basis oI all education, since all, even the most abstract reasoning, involves representation oI inIormation in visual Iorm. 100
The signs oI the arrival oI the new age oI visual reasoning start to appear in the last decades oI the nineeteenth century. In 1880 English logician John Venn published an article "On the Diagrammatic and Mechanical Representation oI Propositions and Reasoning." 101 The
100 See Arnheim, "A Plea Ior Visual Thinking." 101 Martin Gardner, Logical Machines and Diagrams, 2nd ed. (Chicago: The University oI Chicago Press, 1982), 32.
article presented a method to use graphic diagrams to represent complex logical relations and it soon became the standard Ior logicians. Venn's method not only made it possible to illustrate the rules oI inIerential logic in visual Iorms; but much more importantly, now one could reason automatically by simply drawing pictures (Iig. 6). Consider, Ior instance, the Iollowing inIerence: "II all A is contained in B, and iI C is contained in A, then C must also be contained in B." Its validity becomes obvious once the terms are represented in a diagrammatic Iorm. Likewise, such inIerences as "All men are mortal. Socrates is a man. ThereIore, Socrates is mortal" can be arrived at through similar diagrams. Images thought to be unsuitable Ior representing logical relations and thus unsuitable Ior reasoning turn out to be its perIect indeed its natural medium since they can represent topological relations with suIIicient precision and complexity. II Venn's diagrams showed that vision can be used to arrive at logical relations, Galton's composite photographs proved that abstract concepts can be represented visually. Sir Francis Galton, a statistician and a cousin oI Charles Darwin, a Iounder oI eugenics (a project oI social betterment through planned breeding), and the author oI highly inIluential psychological texts, pioneered in 1877 a procedure oI making composite photographs which proliIerated widely in the next three decades. 102 Fabricated by a process oI successive registration and exposure oI portraits onto a single plate, Galton's composites were thought to constitute true statistic averages, representing human types -- a criminal, a prostitute, an Englishman, a Jew, and others (Iig. 7). Galton wrote about his composite pictures that they are "much more than averages; they are rather the equivalents oI those large statistical tables whose totals, divided by the number oI classes and entered on the bottom line, are the averages. They are real generalizations, because they include the whole oI the material under consideration." 103
102 Allan Sekula, "The Body and the Archive," October 39 (1987): 40. 103 Qtd. in Ibid., 47.
Galton not only claimed that "the ideal Iaces obtained by the method oI composite portraiture appear to have a great deal in common with...so-called abstract ideas" but in Iact he proposed to rename abstract ideas "cumulative ideas." In contrast to the human mind, "a most imperIect apparatus Ior the elaboration oI general ideas," Galton championed his composite photographs, which, being mechanical and precise, were much more reliable Ior arriving at abstract representations. 104
With his photographs, Galton not only proposed that universals may be represented through generic images; he actually objectiIied and materialized them. Plato's ideas were given concrete Iorm: they could now be touched, copied, Iabricated, multiplied, distributed, etc. II Venn's and Galton's inventions overturned the traditional claims about the impossibility oI visual reasoning, they did that not through new philosophical arguments about vision but Iirst oI all by inventing new ways to use it. The latter invented a new visual technology: representing universals through composite photographs. The Iormer introduced a new visual technique, a new way to employ visual representations, better yet, a new graphic sign system -- diagrams which signiIy not any concrete objects, but logical Iorms. 105
In eIIect, Venn's technique and Galton's technology, which gave logical Iorms and abstract ideas material tangible Iorm, externalized reasoning. What beIore was a mental process, a uniquely individual state, now became part oI a public sphere. Unobservable and interior processes and representations were taken out oI individual heads and put outside -- as drawings,
104 Qtd. in Ibid., 51. 105 I should clariIy the distinction between the terms visual techniques and visual technologies as they are used in this chapter. By "techniques" oI visual representation, I mean particular graphic and/or pictorial signiIying systems which do not depend in any crucial way on any one medium or technological base. For example, Euler's diagrams can be drawn on paper, etched on metal, generated with computer graphics, etc. The term "visual technologies" reIers to certain representational devices, which do depend on the particular technological conIiguration and would be hard or impossible to achieve without it. The examples include dissolve or traveling shots in Iilm, which are only possible given a sequence oI images in time generated by a Iilm projector, video tape player or computer. Galton's composite photography is such visual technology, dependent, both in its design and its artiIacts, on particular photographic technology.
photographs and other visual Iorms. Now they could be discussed in public, employed in teaching and propaganda, standardized, and mass-distributed. What was private became public. What was unique became mass-produced. What was hidden in an individual's mind became shared.
We should not be surprised, then, that the revolutionary new medium, the medium oI mass society par excellence -- cinema -- was immediately proclaimed to be the machine Ior the externalization oI private mental Iunctions and states. In 1916 Hugo MYnsterberg, a ProIessor oI Psychology at Harvard University and one oI the Iounders oI the Iields oI industrial and applied psychology, published The Film: A Psychological Study, today canonized as one oI the earliest 4. "To teach the worker to think diaIecticaIIy."
theoretical treatments oI cinema. 106 According to MYnsterberg, the essence oI the new medium lies in its ability to reproduce, or "objectiIy" various mental Iunctions on the screen: "The photoplay obeys the laws oI the mind rather than those oI the outer world." 107 In contrast to the theater, where the action is constrained by the limitations oI physical reality, Iilm is Iree to shape arbitrarily its material, closely approximating Ilashes oI memory, the Ilights oI imagination, and
106 Hugo MYnsterberg, The Photoplay: A Psychological Study (New York: D. Aplleton & Co., 1916).
107 Ibid., 41.
other mental acts. For instance, while in theater events have to Iollow each other corresponding to the progression oI time, in Iilm the action can suddenly jump back and Iorth, just as in an act oI imagination. MYnsterberg was not content to point out the analogy between Iilm and mental liIe; in an astounding analysis, he correlated the main cinematic techniques to diIIerent mental Iunctions such as attention and memory, one-to-one. For example, in the close-up, "everything which our mind wants to disregard has been suddenly banished Irom our sight and has disappeared," analogous to how our attention selects a particular object Irom the environment. Similarly, the "cut-back" technique objectiIies the mental Iunction oI memory. "In both cases," MYnsterberg wrote, "the act which in the ordinary theater would go on in our mind alone is here in the photography projected into the pictures themselves." 108 The psychological laboratory became indistinguishable Irom the movie house; the textbook oI experimental psychology -- Irom the cinematographer's manual. The mind was projected on the screen; the inside became the outside. MYnsterberg admired the power oI Iilm to externalize the Iunctions oI consciousness. The next logical step was taken by a German psychologist Kurt Levin who, in 1924-25, was the Iirst to use Iilm in his experiments. He wrote that "Iiction Iilm attempts to objectiIy certain psychological processes Ior the viewer. Psychological (scientiIic) Iilm studies to what extent these psychological processes can be objectiIied." 109 Soviet psychologist A.P. Luria, who planned to establish a psychological laboratory in Moscow in cooperation with the State Film Academy, acquainted Levin with Eisenstein, who attended the shooting oI one oI Levin's Iilms and advised him. 110
108 Ibid., 41. 109 Qtd. in Olga Bulgakova, "Sergei Eisenshtein i ego 'psikhologicheskiy Berlin' -- mezhdu psikhoanalizom i strukturnoy psikhologiey" (Sergei Eisenstein and his 'psychological Berlin': between psychoanalysis and structural psychology), Kinovedcheskie Zapiski 2 (1988): 187. 110 Ibid., 177.
The Iigure oI Eisenstein is particularly important because it reveals the historical connection between the interest in visual reasoning and the rise oI mass communication, oI which Iilm was a major vehicle. The emergence oI new mass societies in the earlier part oI this century dictated the necessity to communicate ideological concepts to mass populations which were oIten illiterate. 111 In the 1920s Eisenstein boldly conceived a screen adaptation oI Marx's Capital as a way to eIIiciently bring about the political enlightenment oI Russian audiences, especially the peasants who would not sit through a political lecture but, attracted by the "novelty" oI a movie projector, would come to see movies, regardless oI what was shown. 112
Unprecedented as his project was, its radicalism lay not only in the decision to visualize the abstract notions and logic oI Capital but in the method employed, which, according to Eisenstein, would directly provoke dialectical thinking in audiences. 113 Jacques Aumont concludes that Ior Eisenstein, "the object privileged in Marx's work is not a theoretical one, like any oI the key concepts Irom Capital. It is at another level entirely that Eisenstein selects his true object -- the Marxist method itselI." 114 Thus it was not simply a matter oI the modern redeployment oI the directions oI the 1492 sermon: "...Images oI the Virgin and the Saints were introduced...on account oI the ignorance oI simple people, so that those who are not able to read the scriptures can yet learn by seeing the sacraments oI our salvation and Iaith in pictures." 115 The viewers oI Capital were not only to learn the scriptures oI the new atheistic religion; they were to learn the process oI reasoning.
111 For instance, according to the 1926 census, out oI every 1,000 citizens oI the U.S.S.R., only 445 were literate. Peter Kenez, The Birth oI the Propaganda State. Soviet Methods oI Mass Mobilization, 1917-1929 (Cambridge: Cambridge University Press, 1985), 157. 112 Kenez, The Birth oI the Propaganda State, 220. 113 The pioneering work oI Annette Michelson was important in bringing my attention to Eisestein's Capital project. Annette Michelson, "Reading Eisenstein Reading 'Capital'," October 2 (1976): 27-38; October 3 (1977): 82-88. 114 Jacques Aumont, Montage Eisenstein (London and Bloomington: BFI Publishing and Indiana University Press, 1987), 163. 115 Qtd. in Baxandall, Painting and Experience in FiIteenth Century Italy, 41.
It is signiIicant that the most categorical statement by Eisenstein on the possibility oI "Iilmic reasoning," reasoning through images, appears in the context oI his discussion oI the sequence known as For God and Country Irom October (1929):
Maintaining the denotation oI "God," the images increasingly disagree with our concept oI God, inevitably leading to individual conclusions about the nature oI all deities...a chain oI images attempted to achieve a purely intellectual resolution, resulting Irom a conIlict between a preconception and a gradual discrediting oI its purposeIul steps. Step by step...power is accumulated behind a process that can be Iormally identiIied with that oI a logical deduction...The conventional descriptive Iorm Ior Iilm leads to the Iormal possibility oI a kind oI Iilmic reasoning. While conventional Iilm directs emotions, this suggests an opportunity to encourage and direct the whole thought process as well. 116
Far Irom simply representing God or deities, as they did Ior centuries, here images serve a totally new Iunction -- to provoke and direct reasoning, reasoning oI a particular kind -- "Marxist dialectics." In accordance with its principles, as canonized by the oIIicial Soviet philosophy, Eisenstein wants to present the viewer with the visual equivalents oI thesis and anti-thesis so that the viewer can then proceed to arrive at synthesis, i.e. the correct conclusion, pre-programmed by Eisenstein. "The content oI CAPITAL (its aim) is now Iormulated: to teach the worker to think dialectically," Eisenstein writes enthusiastically in April oI 1928. 117 Schooled by the Iilm, viewers would become selI-suIIicient thinkers, learning the skill oI "Communist decoding oI the world," each a walking camera, snapping pictures oI visual thesis and anti-thesis, the brain automatically executing cognitive operations oI montage, thinking through images, eIIiciently and eIIectively. Eisenstein claims the radical novelty oI his concept oI "Iilmic reasoning":
116 Sergei Eisenstein, "A Dialectical Approach to Film Form," in Film Form: Essays in Film Theory, ed. Jay Leyda (New York: Harcourt Brace and World, 1949), 62. Emphasis mine -- L.M. 117 Sergei Eisenstein, "Notes Ior a Film oI 'Capital,'" trans. Maciej Sliwowski, Jay Leuda, and Annette Michelson, October 2 (1976): 10.
The proclamation that I'm going to make a movie oI Marx's Das Kapital is not a publicity stunt. I believe that the Iilms oI the Iuture will be Iound going in this direction (or else they'll be Iilming things like The Idea oI Christianity Irom the bourgeois point oI view!) In any case, they will have to do with philosophy...the Iield is absolutely untouched. Tabula rasa. 118
Yet, Einstein's theory was not an isolated development. Many in the artistic leIt oI the 1920s shared a similar belieI in the cognitive power oI new visual Iorms such as montage. In the late 1920s Alexander Rodchenko promoted the use oI montage sequences in graphic design and, like Eisenstein, he saw montage as being equivalent to "dialectical" reasoning. In this Iormulation, an individual image corresponded to a single concept, and thinking was thought to be provoked when a number oI images were juxtaposed in a series. 119 Walter Benjamin's notion oI "dialectical seeing," central in his unIinished Passagen-Werk project, also depends on montage, but within a single Irame, so to speak. "Dialectical seeing" was conceived by Benjamin as a way to grasp the Iorms oI the present by looking to the past and to the Iuture in the same instant, juxtaposing them in the same mental image. 120
118 Michelson, "Reading Eisenstein," October 2: 28. 119 While at Iirst Rodchenko practiced juxtaposition oI many separate photographs and Iragments within the space oI a single image, at the end oI the 1920s his photomontages became multi-page layouts composed oI a number oI more "traditional" photographs, more like a Iilm montage sequence. 120 On Benjamin's concept oI "dialectical seeing" see Susan Buck-Morss, The Dialectics oI
Venn, Galton, MYnsterberg, Eisenstein. Logical diagrams, composite photographs, cinematic devices oI close-up, cut-back, and montage. DiIIerent, in eIIect, as they are, these developments are symptoms oI a single social imaginary at work: to make reasoning more eIIicient and at the same time more controllable by externalizing it and rendering it visible. This imaginary can be glimpsed in the new techniques oI visual representation, such as Venn's diagrams. It also governs the development and the promotion oI new visual technologies, such as Galton's composites or Eisenstein's dialectical montage. Finally, this imagination can be also read in theoretical speculations about how abstract concepts and logical Iorms can be represented as images in principle, such as in Freud's psychoanalysis. Is there a single person whose work brings into Iocus the modern preoccupation with visual reasoning -- whether in the Iorm oI new theoretical models, new techniques oI visual representation or new visual technologies? It is not Venn, or Galton, or MYnsterberg. This person must be Sigmund Freud. He is not only more explicit in his deIense oI the possibility oI visual reasoning; he develops a systematic theory oI how abstract notions and logical Iorms can
Seeing: Walter Benjamin and the Arcades Project (Cambridge: The MIT Press, 1989).
be visualized. This theory becomes one oI the cornerstones oI a new science, devoted, we should note, to externalization, bringing into the open, into public view the most intimate, private, and unobservable psychic phenomena -- the unconscious ideas, unreachable Ior consciousness. In this, the new science goes well beyond MYnsterberg's notion oI the externalization oI memory and attention by the techniques oI Iilm -- aIter all, these mental Iunctions can be voluntarily controlled and commanded by consciousness. Indeed, MYnsterberg and his Iellow experimental psychologists at the turn oI the century scrutinized the mind through the technique oI introspection -- which, by deIault, limited them to the study oI only those mental acts and Iunctions that could be voluntarily controlled and consciously observed. Instead, the new science postulates the existence oI unobservable, uncontrollable psychic Iorces, which, lying deep below the level oI consciousness, nevertheless control it. The latter hypothesis, it is oIten said, constitutes the most dramatic attack on Cartesian rationality launched in this century. However, it should be noted Ior the record that the new science also gives rationality and reason even more power than beIore -- and not only because it tamed the unconscious Ior the consciousness but also because it gained another ground Ior reason -- vision. Having postulated the existence oI subterranean unconscious ideas and unconscious reasoning, this science immediately proposes a technique by which these phenomena can be brought to the surIace, and rendered visible and harmless. It claims that unconscious ideas, groups oI ideas, and even arguments are transIormed into the Iorm oI dream- images. By analyzing these images, an analyst can recover these ideas and arguments.
5. Freud's Theory of VisuaI Reasoning
Freud's The Interpretation oI Dreams, published in the Iirst year oI this century, inaugurated the rise oI visual reasoning. 121 This work exempliIies a new understanding oI vision -- as not inIerior to reason but, on the contrary, as its equal, as a medium quite capable oI representing abstract ideas and logical arguments. According to Freud, dreams can translate into visual Iorm such concepts as "a high-placed oIIicial" (Freud 378) or "superIluous" (441). They can also represent logical connection (349), causality, negation (361, 372) or contradiction (470). In short,
121 All subsequent reIererences in this chapter to Freud, unless overwize indicated, are to Sigmund Freud, The Interpretation oI Dreams (New York: Avon Books, 1965).
the whole range oI reasoning operations can be encoded in visual images. With this theory, vision is completely subsumed into the realm oI logical thinking. According to Freud, the mind's contents, ideas and logical arguments, are securely hidden Irom its owner, Irom the psychoanalyst and Irom the public, lurking, invisible and unreachable, in the archaic cave oI the unconscious. They travel through the darkness oI association networks, like newly Iormed urban masses, riding to and Irom work in the underground subways, which in the end oI the nineteenth century were rapidly constructed in major metropolitan centers. Like these urban masses, always a danger, potentially lurking in the darkness oI the labyrinths oI city streets at night, ideas are Iormed, dissolved and rearranged in the total darkness, which the limited "spotlight oI attention" (Wundt) cannot reach. This, oI course, cannot be tolerated. Streets are streamlined and, even more importantly, well lit. Indeed, the turn oI the century is oIten called "the age oI electricity" due to its obsession with electrical light. But does not this light provide the best way oI Iighting invisibility and assuring externalization? Similarly, ideas and reasoning need to be externalized, brought into the open, rendered visible. Luckily, Freud assures us, the psyche already possesses a process which transIorms the invisible contents oI the unconscious into images within our reach. All the psychoanalyst needs to do is to properly interpret these dream images by tracing them back to their underlying ideas. So how can dreams represent ideas visually? In the section "Consideration oI Representability," aIter he establishes the requirement Ior visualization -- "considerations oI representability in the peculiar psychical material oI which dreams make use -- Ior the most part, that is, representability in visual images," (379) Freud provides a number oI examples oI how dreams transIorm abstract concepts and thoughts into pictures. 122 Here are a Iew oI them:
122 Mieke Bal's seminar on "Interpreting Culture" at the University oI Rochester brought my attention to this point. For the original discussion oI considerations oI representability see Samuel Weber, The Legend oI Freud (Minneapolis: University oI Minnesota Press, 1982).
"A high-placed oIIicial" is represented by a high tower (378). "Working my way through" is represented by "a long kniIe under a cake, as though to liIt out a slice" (380). "SuperIluous" is represented by water everywhere -- "overIlowing," "Ilowing over" (441). Legs (part oI the body) are represented through pillars or columns (382).
Is there a single logic these examples Iollow? Yes, and we have already encountered it in LakoII. LakoII, as we have seen, claims that abstract concepts and abstract reasoning take shape in the process oI extending the structures oI human sensory and bodily experience to abstract situations; and that even the most abstract categories are metaphors, based on such topological relations as part-whole and inside-outside. To use Freud's examples, such concepts as "a high- placed oIIicial" or "working my may through" are meaningIul in the Iirst place because they involve the conceptualization oI abstract situations ("position oI people in society"; "solving a problem") in spatial terms. LakoII postulates that behind all seemingly abstract concepts oI natural language lie concrete "image-schemes." II we accept LakoII's theory Ior a moment and read Freud's examples through its terms, it appears that in dreams abstract concepts and ideas turn back into the "image- schemes" Irom which they evolved in the Iirst place. Thus, "a high-placed oIIicial" becomes again a concrete spatial representation, "high tower." OI course, the point is not that LakoII's theory "explains" Freud or that Freud "anticipates" LakoII; rather, it is that both implicitly rely on the notion which has always Iigured centrally in the speculations about the origin oI language and reason (including, as we have seen, oI such thinkers as Locke, Herder, Vico, and Nietzsche): that abstract concepts have evolved
Irom concrete ones, the latter directly connected with perceptual experience. Freud, however, adds a new and important notion: regression. II in dreams abstract concepts and abstract thinking itselI return to their "origin" as images and "image-based reasoning" (LakoII's term) this transIormation would be in perIect correspondence with the regressive character oI dreams. The notion oI regression is the key Ior understanding Freud's theory oI visual thinking. Not only does it explain why in dreams abstract concepts return to their visual origins, it also partly clariIies why dreams invariably rely on visual means oI representation at all: "what is older in time is more primitive in Iorm and in psychical topography lies nearer to the perceptual end" (587). Regression to the visual mode in dreams, then, operates in three ways: regression to an older representational mode (Iormal regression); topographic regression (to the perceptual system); and temporal regression to the older (in the evolutionist sense) psychical structures. There is also another, Iourth meaning oI regression to the visual mode: regression to the visual scenes witnessed in childhood ("primal scene"). ThereIore, the visual means oI representation employed by dreams are characterized by Freud as an archaic, "primitive mode oI representation and expression" (587). Vision, in Freud's diagnosis, is seen as the primordial, original language oI mankind. Should we be surprised that, according to Freud, Vienna's inhabitants circa 1900, modern and respectable by day, at night totally lose their modernity, shed away progress, and instead immerse themselves in archaic visual illusions? No, iI we recall the view, Irequently voiced in the earlier part oI the twentieth century, that the new visual Iorms oI mass culture represent a return, indeed, a regression to humanity's archaic stage. Most Irequently, it was cinema which was seen (Ior instance, by Nabokov) as a time machine, where the metropolitan masses, stripped oI their individuality, subjectivity, or even simply basic interiority, in total darkness (like that oI prehistoric caves, even darker than the cave described by Plato), gave up the light oI Enlightenment, Civilization and Reason, and instead were chanting in unison...a collective body,
mesmerized by visual Iorms, moving beIore them on the screen, made possible through the magic oI the cinematograph. As his contemporaries, Freud believes that the visual cinematograph oI dream-images represents the regression to the original language oI mankind -- the language oI pictures; however, he does not think that this entails the loss oI rationality, logic, and reason. Freud not only claims, as we have just seen, that the dream-work has the means to represent singular ideas, even abstract ones, but he also conIidently describes the mechanisms by which the dream-work visualizes the logical relations between dream-thoughts. "What representation do dreams provide Ior 'iI,' 'because,' 'just as,' 'although,' 'either-or,' and all the other conjunctions without which we cannot understand sentences or speeches?," asks Freud. His initial answer is negative: "In the Iirst resort our answer must be that dreams have no means at their disposal Ior representing these logical relations between the dream-thoughts" (347). But astonishingly, contradicting his own statement, Freud then proceeds to elaborate the mechanisms by which the whole range oI logical relations can be represented in images:
Logical connection is represented by simultaneity in time (349). Spatial proximity is used to represent logical relation: "Whenever they |dreams| show us two elements close together, this guarantees that there is some specially intimate connection between what corresponds to them among the dream-thoughts" (349). Causality is reproduced by a time sequence (349) or by a metamorphosis. An alternative ("either-or") may be represented by dividing the dream into two pieces oI equal length (353). Negation is usually discarded (353) -- but later Freud gives examples to the contrary (361, 372).
Similarity ("just as") is the "Iavorite" oI dreams -- "this relation, unlike any other, is capable oI being represented in dreams in a variety oI ways" (354), including "composition" -- the creation oI composites. In principle, composition is similar to metamorphosis: "The psychical process oI constructing composite images in dreams is evidently the same as when we imagine or portray a centaur or a dragon in waking liIe" (359). "The temporal repetition oI an act is regularly shown in dreams by the numerical multiplication oI an object" (407). Contradiction is represented by absurdity (470).
Freud does not explicitly connect his descriptions oI the dream-work's mechanisms oI the visualization oI logical relations with his view oI the dream-work as regression to archaic mental structures. However, his characterization oI these mechanisms only makes sense iI we assume that, like everything connected with the dream-work, they owe their character to regression. The connection is provided in the work oI another writer, Lucien LZvy-Bruhl, a leading French anthropologist and Freud's contemporary. Like Freud, LZvy-Bruhl was concerned with the nature oI archaic mental structures. But while Freud located these structures in the unconscious oI every modern civilized individual, LZvy-Bruhl claimed to have Iound them in the mental language oI present-day "savages." In a number oI books LZvy-Bruhl advanced the notion oI "primitive" logic which enjoyed wide popularity in the Iirst decades oI the twentieth century. According to LZvy-Bruhl, the thinking oI primitive people Iollows a logic quite diIIerent Irom our own "normal" logic. First oI all, primitive people conIuse temporal succession with causality. II they witness the two events Iollowing each other, they assume that the two are causally related. Thus, LZvy-Bruhl recounts that iI one oI them sees a snake Iall down Irom the tree in Iront oI him and learns soon
thereaIter that his son is dead, he will relate these two Iacts. 123 Second, primitive people believe that iI two objects have been physically connected, they continue to inIluence each other when separated. For instance, some primitive people are careIul to hide a babies' milk-teeth -- Ior iI somebody Iinds them, then the new teeth will grow improperly. Summing up this Ieature oI primitive logic, LZvy-Bruhl writes: "Primitive people conIuse the antecedent with the cause. This would be a very common error in reasoning that is known as post hoc, ergo propter hoc sophism." 124
These "laws" oI "primitive thinking," as described by LZvy-Bruhl, are the same ones that Freud claims govern the representation oI the relations between dream-thoughts. Savages think that two events Iollowing each other are causally related (LZvy-Bruhl) -- dreams represent causality by temporal sequence (Freud). For savages, two objects physically connected are also logically related -- dreams signiIy a logical connection by the spatial closeness oI elements or by the simultaneity oI time (Freud, 349). Moreover, as "savage mind," the dream-work Iinds similarity everywhere, seeing logical relations where there are none: "One and one only oI these logical relations is very highly Iavored by the mechanism oI dream-Iormation; namely, the relation oI similarity, consonance or approximation -- the relation oI 'just as.' This relation, unlike any other, is capable oI being represented in dreams in a variety oI ways" (354). The visual language oI the dream-work -- the archaic language oI the psyche, as described by Freud - - is the same as the primordial "mental language" still at work in modern savages -- as claimed by LZvy-Bruhl.
123 Lucien LZvy-Bruhl, Les Iunctions mentales dans les sociZtZs primitives (Paris: 1910), 72. 124 Ibid., 73. 6. The Rise of the Diagram
Freud does not just insist that abstract ideas and logic can be visualized or provide a theoretical model oI how this can be done in principle. He practices what he preaches. What makes The Interpretation oI Dreams a truly modern text, compatible with the age oI illustrated magazines, color printing, scientiIic documentaries, dialectical montage, and other techniques and technologies oI visual communication is not the abundance oI illustrations or Iigures (in Iact, there are only a couple). Freud does not have to include illustrations, because many oI his theories, and indeed his overall theory oI the psyche (the Iamous topographic model), are based on simple visual models, easy to grasp and easy to remember. And this is what makes his text modern. The reader is invited to disregard the complexities oI real anatomical data and instead to conceptualize the psyche "as resembling a compound microscope or a photographic apparatus, or something oI the kind" (574). According to Freud, this apparatus is made up oI just a Iew components (called by Freud agencies or systems), which are arranged in spatial order (574- 583). This "spatial analogy" (653) allows Freud to visualize any psychic process as a simple movement between a Iew parts oI the model. It is as though Freud senses that the construction oI such models is very eIIective in pushing the reader to "see" the selI-evidence oI the appropriate conclusions. Thus, aIter presenting the typographic model oI the psyche, Freud uses it to explain the newly introduced concept oI regression: "I believe that the name 'regression' is oI help to us in so Iar as it connects a Iact that was already known to us with our schematic picture, in which the mental apparatus was given a sense or direction. And it is at this point that picture begins to repay us Ior having constructed it. For an examination oI it, without any Iurther reIlection, reveals a Iurther characteristic oI dream-Iormation" (582. Emphasis mine -- L.M.) To clariIy the logical relation between the conscious and the unconscious, Freud again relies on the same technique oI constructing a visual diagram. From the point oI view oI the ontogenetic and phylogenetic explanation oI the Iormation oI the psychical apparatus, primary
processes, as their name implies, "are present in the mental apparatus Irom the Iirst" (642) while secondary processes appear only later and never achieve Iull domination. Thus, concludes Freud, the primary processes (and corresponding to them the system oI the unconscious) constitute the essence oI psychical liIe: "The unconscious is the true psychical reality" (651). To emphasize this point and to make sure the reader remembers it, Freud "visualizes" it in the Iorm oI a simple image oI a sphere within a larger sphere: "The unconscious is the larger sphere, which includes within it the smaller sphere oI the conscious" (651). He then "reads oII" his conclusion Irom this visual model: "Everything conscious has an unconscious preliminary stage" (651). Freud's skillIul use oI visual models, analogous to Venn's diagrams, to represent the logical relations between the terms in his theories is unusual in that Freud painstakingly describes a visual model verbally, rather than directly presenting the reader (or should we say, the viewer?) oI his text with a picture. It is as though Freud Iully realized the power oI diagrams to represent complex concepts, models and arguments but, circa 1900, he was still hesitant to "vulgarize" his scientiIic opus with images. In 1933 Freud Iinally includes a real diagram in The New Introductory Lectures on Psychoanalysis to illustrate the relation between two triads oI concepts: id, ego, and superego, and unconscious, preconscious, and conscious. Freud's diagram, which eIIectively signiIies a complex set oI relations by just a Iew lines, is representative oI a whole Iamily oI new techniques oI visual representation, new to the modern period (Iig. 8). These techniques have risen to Iill what can be called the cognitive needs oI modernization: to quickly communicate more and more complicated relations between various entities, to manipulate huge sets oI data (whether in science -- which became more and more the problem oI managing collected data -- or in the bureaucratic management oI people), to depict in manageable Iorms endless hierarchies oI relations -- in short, to represent reasoning visually and to reason with the help oI vision. These techniques oI visual representation comprise what semioticians call the graphic sign systems --
signiIying systems, employing a particular set oI graphic conventions to represent particular kinds oI objects. 125
We have already encountered one oI these new graphic systems -- Venn's diagrams. These diagrams rely on the unique property oI vision, which, along with touch, as RudolI Arnheim points out, is "the only sensory medium that conveys such spatial properties as inclusion, overlap, parallelism, size, etc., with some precision." 126 This property makes possible the precise representation oI the logical relations between classes in graphic Iorm. But, while in principle visual images are capable oI representing spatial properties, not every image can eIIectively do so. For instance, continuously varying tonal gradients are useless, whereas discrete elements are eIIective. And these elements, to be easily readable, must have enough visual diIIerence among them. ThereIore, Venn's diagrams are typically drawn using simple lines. The classes are represented through circles (a representation in the Iorm oI rectangles is possible, but it would be ambiguous, leading the viewer to believe that the shapes oI Iigures code some additional logical relation). II circles overlap, the common area (representing exclusive "and") is clearly distinguished through shading or color. In short, we are dealing with a distinct graphic system, employing particular kinds oI graphic marks, optimized Ior the representation oI particular reIerences -- a logic oI classes. This graphic system accentuates one property oI images (the ability to represent spatial properties such as inclusion) at the expense oI other properties, Ior instance, the ability to represent continuously varying quantities through continuously varying tone or color. Venn's diagrams is an example oI a specialized graphic sign system. Their use is limited to a narrow proIessional domain oI logic. In the same period, another graphic sign system Ior visualization oI logical relations comes into use (Iig. 9). In contrast to Venn's diagrams, the
125 The most comprehensive study oI graphic sign systems is Jacques Bertin, Semiology oI Graphics (Madison, WI: The University oI Wisconsin Press, 1983). 126 Arnheim, "A Plea Ior Visual Thinking," 143.
graphic marks oI this sign system are now everywhere, seemingly the very air oI which the contemporary visual semiosphere is composed. What I have in mind is the gradual emergence oI greatly simpliIied graphic marks which came to Iorm a notation, a sort oI an erector construction set, whose parts can be used to quickly assemble a visualization oI a network oI relations, a representation oI a process or a scheme oI dependencies. This construction set includes boxes, arrows, vectors, solid and dotted lines, and so on. Arrows can denote such relations as "belongs to," "relates to," "is a member oI a class oI," "causes," etc. Boxes, on the other hand (or simply any area separated by lines) can represent a physical object, the state oI a system, a stage in a process or any abstract entity. Freud's drawing circa 1933, where a Iew encircled areas together with dotted and solid lines represent a complex relationship between psychical agencies, illustrates the power oI this graphic system, which I will call diagrammatic. My comparison oI the diagrammatic graphic system with a construction set is not accidental. A construction set includes some parts with more or less Iixed Iunctions (Ior instance, wheels Ior moving models), while others (bars with holes) can be used Ior a variety oI purposes - - to hold other parts, to represent parts oI a model, to be stationary or to move. Similarly, while some elements oI the diagrammatic system have relatively Iixed meanings (Ior instance, wavy brackets), others, such as arrows or dotted lines, can have numerous meanings, depending on their particular arrangement. This leads us to the second Ieature. In diagrammatic language, the same Iew symbols can represent anything: a box, Ior instance, may stand Ior an idea, a theory, a building, 200 buildings, a citizen, an army, or a state oI a physical system. Unlike the users oI verbal language who learn particular semantics, the users oI the diagrammatic system learn the general conventions oI syntax. It is enough to know that a box represents an entity -- regardless oI what the entity may be. In this, diagrammatic language, used to represent a set oI entities and the relations between
them, Iunctions exactly like the symbolic language oI logic: a properly executed logic diagram is isomorphic with a logical statement. 127 But this power is achieved through the standardization oI graphical conventions themselves. A diagrammatic system is a metalanguage, whose symbols represent not any particular ideas or objects, but rather the abstract relations evoked in any process oI reasoning: "a part oI;" "leads to;" "associated with." It is not accidental, thereIore, that this language oI arrows and boxes is what recent generations oI computer programs, aiming to help reasoning or to imitate it (outliners, hypertext, project management tools, idea generators, and the like) provide their users with. The mental construction set became the graphic toolbox, always just a mouse click away. This computer implementation oI already existing conventions oI visual culture is accompanied by a Iurther standardization oI graphic symbols. This standardization is justiIied by the argument, popularized since the adaptation oI Macintosh user interIace, that it helps the user to recognize Iamiliar symbols and commands regardless oI the program she is using (Iig. 10). The standardization oI diagrammatic symbols went hand in hand with their simpliIication. While drawings were used Ior centuries to represent schematic inIormation rather than simply to symbolize particular objects or to record their shapes, modernization is accompanied by the simpliIication oI graphic symbols -- boxes, arrows, and the like. (This was also accompanied by the streamlining oI typography.) Now, it becomes much easier to "read oII" logical relations represented by the diagrams, since there are Iewer graphical marks to scan. When do the symbols oI diagrammatic language, as it exists today, Iirst appear? Obviously, we would not Iind a sharp "break" in graphic paradigms, with a new graphic system appearing overnight. According to Ernst Gombrich, the arrow does not acquire its meaning as pointer or vector until the eighteenth century. And even then, he could only Iind it in a couple oI specialized proIessional contexts: it appears in a French treatise on Hydraulic Architecture oI
127 Gardner, Logic Machines and Diagrams, 28.
1737 and on some maps, where it was used to show the direction oI the river's Ilow. 128
Similarly, we can Iind other abstract graphic marks employed Ior centuries in graphic systems used in particular proIessions -- Ior instance, engineering drawing. Gradually, however, these symbols leave these proIessional contexts and enter visual culture at large. Finally, what is the relation between the development oI the diagrammatic graphic system and the overall movement in modern visual culture towards the simpliIication, streamlining, and standardization oI all visual Iorms? In the opening chapter oI Techniques oI the Observer, Jonathan Crary points out an inherent contradiction present in most standard accounts oI modernism. On the one hand, these accounts postulate the emergence oI new radical models oI perception and representation, summed up in the notion oI abstraction, whose Iirst signs appear with Manet, impressionism or post-impressionism and which Iully blossoms with Kandinsky, Kupka, Mondrian, and others. On the other hand, it is also postulated that in the same historical period new popular visual Iorms developed (Iirst photography and then cinema), Iurther solidiIying realism. As Crary notes,
128 Ernst Gombrich, "Pictorial Instructions," in Images and Understanding, ed. Barlow Horace, Colin Blakemore and Miranda Weston-Smith (Cambridge: Cambridge University Press, 1990) , 28.
"Classical space is overturned, so it seems, on one hand, but it persists on the other. This conceptual division leads to the erroneous notion that something called realism dominated popular representational practices, while experiments and innovations occurred in a distinct (iI oIten permeable) arena oI modernist art making." 129
Crary's way to solve this contradiction is to claim, throughout his book, that the so-called realistic Iorms oI popular visual culture, based on photography, and abstract art are both grounded in the emergence oI "subjective vision" in the early nineteenth century, in which the link between representational signs and external reIerents is severed. "I argue," writes Crary, "that some oI the most pervasive means oI producing 'realistic' eIIects in mass visual culture, such as stereoscope, were in Iact based on a radical abstraction and reconstruction oI optical experience, thus demanding a reconsideration oI what 'realism' means in the nineteenth century." 130 What his anti-realist argument leaves out, however, is any consideration oI the representational power oI new visual technologies such as photography. With photography, based on mechanized perspective, the ability to establish a "reciprocal, metrical relationship between the shapes oI objects as deIinitely located in space and their pictorial representations" (William Ivins) became greatly ampliIied. In Ivins' words, photography is "a Iorm oI picturemaking that is not only precisely duplicable but one in which geometrical perspective is so inherent that today the camera is used as a surveying and measuring instrument." 131 Rather than being anti-realistic, photography became the perIect technology oI instrumental realism (Allan Sekula's term), literally allowing its users to have physical power over the represented subjects and objects. As Sekula points out in his important account oI the instrumental uses oI photography at the turn oI the century, criminal identiIication photographs, Ior instance, "are
129 Crary, Techniques oI the Observer, 4. 130 Ibid., 9. 131 William M. Ivins, On the Rationalization oI Sight (New York: Da Capo Press, Inc., 1975), 12.
designed quite literally to Iacilitate the arrest oI their reIerent" and, similarly, military reconnaissance photographs make possible identiIication and destruction oI the objects they capture. 132
A diIIerent way to solve the contradiction, noted by Crary, between the opposition abstraction/high art versus realism/popular visual culture is to point out that in Iact the development oI abstract visual Iorms was not limited to modernist art but equally took over the realm oI the popular -- Iirst oI all, in the adaptation oI what I have called here diagrammatic visual language. But in order to see this it is necessary to include in the discussion more than images Irom the realms oI art, or entertainment or leisure -- the realms to which the problematic oI high-low still limits itselI, and whose borders in Crary's argument are represented by modernist painting, on the one side, and photographs, stereoscopes, phenakistiscopes and other popular nineteenth century optical toys, on the other. Rather, it is necessary to consider the whole range oI images employed in modern culture -- images employed not only Ior leisure but also Ior work; images through which, literally speaking, modern work gets done: aerial surveillance photographs and criminal identiIication shots, engineering and architectural drawings and plans, maps and typography, instrumentation displays, and scientiIic imagery. And, last but not least, we should consider the abstract graphic marks oI diagrammatic visual language, designed to Iacilitate cognitive and communicative workloads oI modernity, in the oIIice, at school, in the control room, in the boardroom, and on the assembly line -- which is today replaced by equally monotonous rows oI computer terminals, outputting not Model T, but bits oI inIormation; this steel and coal oI the inIormation -- or shall we better say, cognitive age.
InIormation age, knowledge economy, post-industrial society. All these terms come to designate the changing nature oI society aIter World War II: the gradual displacement oI manual labor by
132 Allan Sekula, "The Body and the Archive," 7.
cognitive labor. With the centrality oI cognition Ior the post-industrial workplace, the notion oI visual reasoning acquired new importance. Consequently, the work on the models, techniques, and technologies oI visual reasoning, which was previously pursued by a Iew isolated individuals (Galton, Freud, Eisenstein), has now become a matter oI systematic research on the industrial scale in such Iields as cognitive psychology, scientiIic visualization, and virtual reality. I will consider this research in the next two sections.
What is the relation between the psychological theories oI the mind and their contemporary visual technologies? The recurrent claims that new visual technologies externalize and objectiIy reasoning, and that they can be used to augment or control it, are based on the assumption oI the isomorphism oI mental representations and operations with external visual eIIects such as dissolves, composite images, and edited sequences. This assumption, which I so Iar have not questioned, on a closer examination appears to be highly problematic. Whatever mental representations and operations really are, the mind surely does not contain pictures, photographs or Iilm clips which some mental homunculus looks at. The external images presented to the mind are not magically transplanted inside it as ready-made ideas and arguments. Regardless oI what visual Iorms can be presented beIore the eye -- diagrams, photographs, Iilm images -- as any other visual input, they are subjected to the complicated processing by the nervous system which constructs its own internal representations. Yet, the assumption oI such an isomorphism continues to persist in modern thinking about vision, ignited by every new round oI visualization technology: photography, Iilm, computer animation, and virtual reality. Consider the claims which surround the new Iield oI scientiIic visualization -- visualization oI data sets, their relationships and their dynamic behavior using computer graphics. Richard Mark FriedhoII and 7. VisuaI TechnoIogies and the Mind
William Benson proclaim that computer visualization techniques constitute the second computer revolution because they act as the direct "extension oI preconscious visual processes." 133 They assume that the images on a computer screen do not simply Iunction as an aid Ior reasoning but that they are equivalent to the mental representations the mind may construct while thinking -- and this is the source oI their power. Or consider the technology, which, even more so than scientiIic visualization, is seen as capable oI completely objectiIying, better yet, transparently merging with mental processes -- virtual reality (VR). Again, the descriptions oI its capabilities do not distinguish between internal mental Iunctions, events and processes, and externally presented images. This is how, according to Jaron Lanier, VR can take over human memory: "You can play back your memory through time and classiIy your memories in various ways. You'd be able to run back through the experiential places you've been in order to be able to Iind people, tools." 134 Lanier also claims that VR will lead to the age oI "post-symbolic communication," communication without language or any other symbols. Indeed, why should there be any need Ior linguistic symbols, iI everybody, rather than being locked into a "prison-house oI language" (Jameson), will happily live in the ultimate nightmare oI democracy -- the single mental space which is shared by everybody, and where every communicative act is always ideal (Habermas). This is Lanier's example oI how post-symbolic communication will Iunction: "you can make a cup that someone else can pick when there wasn't a cup beIore, without having to use a picture oI the word "cup." 135 Here, as with the earlier technology oI Iilm, the Iantasy oI objectiIying and augmenting consciousness, extending the powers oI reason, goes hand in hand with the desire to see in technology a return to the primitive happy age oI pre-language, pre-misunderstanding.
133 FriedhoII and Benson, The Second Computer Revolution: Visualization, 13. 134 Druckrey, "Revenge oI the Nerds," 9. 135 Ibid., 6.
Locked in virtual reality caves, with language taken away, we will communicate through gestures, body movements, and grimaces, like our primitive ancestors... 136
What can one make oI this apparently unsound, yet irresistible, assumption oI isomorphism between the mental process oI reasoning and external, technologically generated visual Iorms, haunting us at least since the end oI the nineteenth century? The conIlation oI outside and inside is, oI course, symptomatic oI the desire to project the inside onto the outside, to make it objective and public. But is this all? To really understand the persistance oI this assumption, we should turn to the history oI ideas about the nature oI mental processes. It is well known that technologies have historically provided and continue to provide the models according to which people imagine the mind. In the seventeenth century it was the clock, in the nineteenth -- the motor, in the second halI oI the twentieth -- digital computers. More precisely, the paradigms were provided not by the technologies themselves, but by theories which made them possible. To take the last paradigm as an example, cognitive psychology, born in the 1950s and gaining prominence ever since, approaches the mind as an inIormation- processing system, as soItware which runs on the hardware oI the brain. But what gave cognitive psychology its epistemological basis was not the new technology itselI (computers), but the inIormation theory accompanying it. It was this paradigm which substituted the discussions oI mind and brain by the notion oI "human inIormation processing." And beIore inIormation
136 At SIGGRAPH 1992, the premier annual conIerence on Computer Graphics and Interactive Techniques in the U.S., attended by nearly 30,000 people, about a dozen VR exhibits always had long lines oI visitors. However, the lines to two oI these exhibits were much longer than any oI the others. One was called Dome, the other -- Virtual Reality Cave; in both cases, to see the show, the visitors had to go inside some cave-line structures. It did not matter that one oI the exhibits Ieatured a scientiIic visualization display, oI interest only to specialists. Clearly, the Iact that in order to see the spectacle one had to go inside a dark, cave-like space, diIIerent Irom normal space, provided enough attraction at the end oI this century, just as it did at its beginning, when millions Ilocked into the dark caves oI movie theaters.
theory, theories oI the mind were inIluenced by thermodynamics (as in Freud) and mechanics (Hobbes). 137
It can be also claimed, however, that human imagination about the mind's operations is limited by the current visual technologies. LakoII's assertion that "natural reasoning makes use oI at least some unconscious and automatic image-based processes such as superimposing images, scanning them, Iocusing on part oI them" (1986: 149) as well as Johnson-Laird's proposal that logical reasoning is a matter oI scanning visual models, would have been impossible beIore television and later, computer graphics. These visual technologies made operations on images such as scanning, Iocusing, and superimposition seem natural. Even more telling are the models oI cognitive psychologists who, in the last two decades, have systematically scrutinized the role played by mental images in reasoning. These models, which deIine mental images in terms oI such characteristics as spatial resolution, speed oI access, basic graphic operations (rotation, translation, copy), seem to be describing Iirst oI all computer imaging systems. Psychologists argue among themselves which imaging systems better resemble mental processes, but they do not doubt the basic metaphor. As Paul Virilio notes, "now the virtual images oI the computer screen seem to conIirm not only the existence oI certain Iorms oI representation but, more immediately, the objective presence oI mental images." 138
137 While this theory is well known and widely accepted, other Iacts suggest that, at least, sometimes, the inIluence runs in the opposite direction -- biological and psychological theories oI body and mind providing paradigms Ior theories oI mechanisms. For instance, it appears that Norbert Wiener's cybernetics was inspired by the concept oI homeostasis developed in biology: "Physiologist Walter B. Cannon viewed the animal body as a selI-regulating machine. Building on the work done by Claude Bernard in the nineteenth century, Cannon developed the concept oI 'homeostasis' -- the process by which the body maintains itselI in a state oI internal equilibrium. Cannon's ideas were well known to Norbert Wiener. In Iact, Cannon's Wisdom oI the Body (1932) may be read as sort oI an introduction to Wiener's Cybernetics (1948)." Charles Eames and Ray Eames, A Computer Perspective: Background to the Computer Age (Cambridge: Harvard University Press, 1990), 99. Another example is provided by the turn oI ArtiIicial Intelligence (AI) in the 1980s Irom trying to simulate the disembodied mind to the simulation oI a collective oI primitive organisms, having the Iunctionality oI insects. Drawing directly on research in biology, the researchers in AI hope that intelligence will emerge as a product oI the collective behavior oI machines simulating simple biological organisms. 138 Paul Virilio, Lost Dimension (New York: Semiotext(e), 1991), 114.
Similarly, in the earlier period, when Freud, in The Interpretation oI Dreams, described the mechanisms by which dream-thoughts and the logical relations between them are represented in dreams, he and his Iellow psychologists relied on available visual technology Ior their understanding oI the mind. Not surprisingly, Galton's composites, the earliest Iorm oI image processing beIore digital computers, provided a particularly attractive model. Freud compared the process oI condensation with one oI Francis Galton's procedures which became especially Iamous: making Iamily portraits by overlaying a diIIerent negative image Ior each member oI the Iamily and then making a single print. 139 Writing in the same decade, the American psychologist Edward Titchener opened the discussion oI the nature oI abstract ideas in his textbook oI psychology by noting that "the suggestion has been made that an abstract idea is a sort oI composite photograph, a mental picture which results Irom the superimposition oI many particular perceptions or ideas, and which thereIore shows the common elements distinct and the individual elements blurred." 140 He then proceeds to consider the pros and cons oI this view. We should not wonder why Titchener, Freud and other psychologists take the comparison Ior granted rather than presenting it as a simple metaphor -- contemporary cognitive psychologists also do not question why their models oI the mind are so similar to the computer workstations on which they are constructed. Thus, not only reason has always been conceived by philosophers in visual terms; as psychologists begin to Iuriously take on the questions philosophers only wondered about, subjecting mental processes to controlled, scientiIic, laboratory study, their models begin to reIlect, more and more, the external visual Iorms made possible by whatever visual technology dominates the period. As beIore, vision is the mirror reason looks into.
139 Sigmund Freud, Standard Edition oI the Complete Psychological Works (London: Hogarth Press, 1953), 4: 293. 140 Edward BradIord Titchener, A Beginner's Psychology (New York: The Macmillan Company, 1915), 114.
Given the reliance oI psychological theories oI the mind on contemporary visual technologies, is there any "progress" between the turn oI the century and today, except that the imagination oI contemporary psychologists depends on the more sophisticated visual technologies oI computer graphics? In Iact, during this century, the assumption oI an isomorphism between the mental and the objective became even more prominent; and externalization oI reasoning has been taken much Iurther, both technologically and theoretically. On the one hand, the reIinement oI various medical imagining techniques in the 1980s made possible an increasingly precise imaging oI brain activity, including the visualization oI reasoning -- in a literal sense. It is now possible to ask the experimental subject to concentrate on solving a problem and to see which parts oI the cerebral cortex are active. The question oI whether reasoning in Iact depends on the operations normally involved in perception becomes more and more the question which, according to a number oI researchers, can be answered through experimentation -- it is enough to show that the part oI a cortex normally dedicated to the processing oI visual inIormation is activated in the process oI reasoning. 141
More importantly, in their theories, many cognitive psychologists have accepted as given the equivalence between internal mental processes and the operations which can be done with external, objectively existing visual representations and objects. Consider the debates about the nature and role oI mental imagery, which have constituted one oI the most active areas oI research in cognitive psychology in the last two decades. 142 On the one hand, there are those (such as Zenon Pylyshyn) who argue that mental imagery simply consists oI the use oI general thought processes to simulate physical perceptual events. In this view, iI the subjects report the presence oI mental imagery during reasoning and problem solving, this is simply a side eIIect, a
141 Martha Farah, "Is Visual Imagery Really Visual? Overlooked Evidence Irom Neuropsychology," Psychological Review 95, no. 3 (1988): 307-317. 142 For a summary oI diIIerent positions, see Ronald A. Finke, Principles oI Mental Imagery (Cambridge: The MIT Press, 1989).
by-product oI real mental computations which do not involve visual representations. On the other hand, there are those psychologists and neurophysiologists who, through experiments and direct imaging oI brain activity, want to prove that reasoning takes place through the construction and manipulation oI mental images (Alan Pavio, Roger Shepard, Stephen Kosslyn, Martha Farah). One oI the most well-known experiments in deIense oI the latter view has been done by Roger Shepard and Jacqueline Metzler oI Harvard University. 143 They presented subjects with
143 S. Shepard and J. Metzler, "Mental Rotations oI Three-dimensional Objects," Science 171 (1971): 701- 703.
pairs oI perspective line drawings oI three-dimensional Iorms constructed Irom small cubes. The subjects' task was to determine whether or not the Iorms were identical in shape, despite the diIIerence in orientation. Shepard and Metzler have Iound that the reaction time was proportional to the degree oI rotation which is required to bring the two objects into a similar position. These results were taken as prooI that in solving the problem, the subjects mentally rotated representations oI three-dimensional objects, and that imagined rotations corresponded to actual physical rotations oI objects: "Imagined rotations and physical transIormations exhibit corresponding dynamic characteristics and are governed by the same laws oI motion." 144 Thus, a mental process was equated with an operation one would perIorm with real, objectively existing objects. Other experiments in deIense oI the position that many kinds oI reasoning involve manipulation oI mental imagery entail the comparison oI abstract qualities. When subjects were asked to recall two animals and to judge which one was larger, the reaction time decreased proportionally to the diIIerence in estimated size. In another experiment, one group oI subjects was asked to rate animals in intelligence on a scale Irom one to ten, while another group had to compare the intelligence oI pairs oI animals. Again, the reaction time decreased as the distance in rated intelligence increased. It was concluded that when the subjects tried to discriminate between two objects, reaction time was shorter the greater the diIIerence between two objects, regardless oI whether the objects were really presented (Ior instance, two lines oI diIIerent length), or were imagined (size oI animals), or whether the qualities to be compared were abstract (intelligence oI animals). 145
We shall leave to psychologists the debates whether these and numerous related experiments indeed prove that internal mental processes involve the manipulation oI pictures
144 Finke, Principles oI Mental Imagery, 93. 145 John Robert Anderson, "Mental Imagery," in Cognitive Psychology and Its Implications, (W.H. Freeman and Company, 1980).
similar in their qualities to real images. But Ior my purpose it is signiIicant in itselI that in imagining what mental processes are like, contemporary psychologists have assumed, without any reservations, an equivalence between the internal and the external, between the mental objects and the real ones.
Modernization, visualization, externalization. In order to externalize the internal, the invisible, it was Iirst equated with the visual. Once this was accomplished, it was simple and only logical to equate the visual inside with the visual outside, the objectivity and standardization oI images drawn on a classroom blackboard, on the screen oI a movie theater, or, most recently, on the computer terminal.
In this chapter I am concerned with the answer to the problem oI how to represent, communicate, induce or perIorm reasoning more eIIiciently. This problem Iirst became important with the emergence oI mass society which required an eIIicient means to communicate abstract ideas to large audiences. It became even more important with the shiIt to post-industrial society when logical reasoning and other cognitive skills attained central economic importance. The answer was to adopt vision as a more eIIicient medium. It can be perceived in Freud's theoretical speculations oI how ideas and logical arguments can be visualized, in the development oI new techniques oI visual representation (Venn's pictorial logic, diagrammatic graphic system), and in the claims made about new visual technologies (Galton, MYnsterberg, Eisenstein). Vision, traditionally considered to be inIerior to reason, is now seen as the proper, indeed, as the ideal medium Ior it. And this is not only a question oI the eIIiciency 8. AnaIog Engine
and eIIectiveness oI communication, although, with visual reasoning, these two demands are Iully satisIied. Indeed, in Eisenstein's montage sequences, reasoning is speeded up -- to twenty- Iour Irames a second. And, with pictorial statistics and diagrammatic visual language, reasoning becomes more eIIective: complex conceptual hierarchies are reduced to boxes and arrows, huge sets oI data become marks on paper which can be easily manipulated. But what is equally important is that visual reasoning Iits perIectly with the demand oI mass society Ior standardization. The subjects have to be standardized, and the means by which they are standardized need to be standardized as well. Hence the objectiIication oI internal, private mental processes, including ideas and logical connections, and their equation with external visual Iorms which can be easily manipulated, mass produced, and standardized on its own. Whether we are dealing with the conventions oI Venn's diagrams, oI signs oI diagrammatic sign system, or oI cinematic techniques, the private and individual is translated into the public and becomes regulated. This new use oI vision as a means oI reasoning is accompanied by its rationalization on a conceptual level. II previously vision was thought to be conIused, incapable oI Iorming generalizations and unsuitable Ior representing logical Iorms, now this is all Iorgotten. Vision is accepted as being Iully capable oI representing logical relations (Freud), abstract ideas (Galton), and even dialectics (Eisenstein). This equation oI vision with logic, accomplished by the beginning oI the twentieth century, is the Iirst stage in vision's rationalization. When industrial society turned into the society oI inIormation, computers became the dominant technology, both economically and intellectually, and behaviorism (eIIiciency oI the body) was replaced by cognitive psychology (eIIiciency oI the mind), this rationalization entered into a second stage. Now, with the questions oI how inIormation can be coded, stored, retrieved, and processed more eIIiciently becoming a matter oI economic survival, cognitive psychologists, together with computer scientists, began to
scrutinize vision Irom this new perspective. In the process, they came up with very precise descriptions oI what vision is. However, what they describe is not some ahistorical, essential vision, but vision already conceived in a historically and socially speciIic way -- as an eIIicient computational system. Why is it that when cognitive psychologists started to wonder about the properties oI data structures involved in human reasoning, vision began to take a central place in their models? Because every mental process is conceived as a computation, and it is assumed that this computation has to be eIIicient. The criterion oI eIIiciency is a very general and pervasive notion in contemporary cognitive psychology. This criterion is the main Iactor in deciding which oI the competing models should be Iavored. The criterion is justiIied by a reIerence to biological evolution -- it is said that in the process oI evolution the most eIIicient method, or the most eIIicient representational code had to be adopted. However, the real justiIication oI this criterion, in my view, is economic, coming Irom the close connection between cognitive psychology and its master discipline, to which cognitive psychology ultimately serves -- computer science, and behind it, the computer industry. Just as computer scientists would not want to build an ineIIicient computer, psychologists cannot imagine an ineIIicient mental process. Once psychologists realized that images possess some unique properties, which make them a very eIIicient medium Ior computation, it was only natural that they entered into their theories. How does Johnson-Laird justiIy his theory oI mental models, according to which logical reasoning involves the construction and manipulation oI spatial models, analogous to Venn's diagrams? A model represented in a dimensional space can be constructed, manipulated or scanned in any direction, while propositional representations can only be processed in one direction, making them much less eIIicient. Why do some cognitive psychologists argue that human long-term memory stores not only knowledge in the Iorm oI propositions, but also as images? Because iI we store appearances
oI objects then later we can compute those properties that could not be anticipated when we saw the objects initially. 146 In contrast to propositional (linguistic) representation, images turn out to be capital which can be used over and over in the mind's economy. Finally, why all the debates in cognitive psychology about the role oI mental imagery in reasoning in the last two decades? Because many problems can be solved more eIIiciently by analog rather than digital computer, and vision has unique properties that enable it to Iunction as such a computer:
II images are representations in a medium with certain Iixed properties, and can be subjected to transIormations such as rotation and scaling, then imagery could be used as an analog computer, to solve problems whose entities and relations are isomorphic to objects and spatial relations. That is, certain abstract problems could be best solved by translating their entities into imagined objects, transIorming them using available image transIormations, detecting the resulting spatial relations and properties, and translating those relations and properties back to the problem domain. 147
What are these "certain properties" which make vision into an analog engine? On the one hand, images can represent continuously varying inIormation. On the other hand, images can represent topological relations such as inclusion, "next to," and others. Taken together, these two properties make vision a very powerIul computer, better suited Ior solving many problems than language. As Steven Pinker explains, in a visual medium, "it is impossible to represent the Iact that one object is next to another without also committing oneselI to which is to the leIt. (This is not true Ior abstract propositional representations, where the two-place predicate 'next to' can be asserted oI a pair oI objects with no such commitment.)" 148
In principle, it is always possible to solve analog problems using propositional representations but this simply would not be as eIIicient. So eIIiciency becomes the Iinal reason why vision, in spite oI the arguments oI Descartes, Locke or Leibniz, is today equated with
146 Steven Pinker, "Visual Cognition: An Introduction," Cognition 18 (1984): 65. 147 Ibid., 66. 148 Ibid., 66-67.
reasoning. However, in this equation, vision has little to do with what these philosophers meant by it -- it became an analog code, a "data structure" more proIicient Ior reasoning than language. What remained the same, although recast in new terms, is the idea that vision came beIore verbal language. II Ior Freud vision was the original, primordial language oI humankind, today cognitive psychologists speculate that the continuous, analog representational system oI mental images emerged Iirst, Iollowed by the system Ior propositional representation (language). I once saw, thereIore I think.
The year 1991 saw two events, oI diIIerent importance and seemingly unrelated. One was the long awaited publication in English oI one oI the most inIluential essays oI modern art history -- Chapter 3. Mapping Space 1. VisuaI NominaIism
Erwin PanoIsky's Die Perspektive als 'symbolische Form'. 149 The interest generated around the re-emergence oI this legendary essay, written in 1924-1925, 150 demonstrates that the problem oI
149 Erwin PanoIsky, Perspective as Symbolic Form (New York: Zone Books, 1991) .
perspective is still Ielt to be relevant to contemporary culture. Joseph Leo Koerner starts his review oI Perspective as Symbolic Form with the Iollowing claim: "Perspective is again the rage." 151 The second event was the GulI War, the outcome oI which was largely predetermined by Western superiority in the techniques oI perspectival representation.
150 Erwin PanoIsky, "Die Perspektive als 'symbolische Form'," Vsrtrage der Bibliothek Warburg (Leipzig & Berlin: 1927), 258-330. 151 Joseph Leo Koerner, "The Shock oI the View," The New Republic (April 26, 1993): 32.
The images extensively televised during the GulI War, conIirmed Paul Virilio's thesis that modern warIare has become a matter oI the "logistics oI perception." 152 True, broadcasts have included more traditional views oI soldiers, planes, and tanks as seen Irom the outside, by a reporter's video camera. But what we also saw were not just images oI the war, but endless images through which the war was carried out: video images Irom an inIrared camera mounted on a plane; video images Irom a camera installed on a weapon guided by a laser sensor; video in its role as "battle damage assessment" where a weapon equipped with an imaging device Iollows a weapon oI destruction and records details oI the damage. This was no longer a traditional reporter's view oI a battle. We saw what the soldiers themselves saw: the images that were their only inIormation about their targets. More oIten, in a strange case oI identiIication, we witnessed what was "seen" by a machine, a bomb, or a missile. Television reports became catalogs oI various vision technologies used to plan a battle, to Iight it, and to assess the results. They included satellite imagery oI Iraqi territory; diagrams oI Iraqi radar activity over the course oI a week; 3-D visualizations oI weapons; a 3-D animation computed Irom satellite data oI a mountain region oI Iraqi thought to contain a secret sight oI
152 Paul Virilio, War and Cinema: the Logistics oI Perception (London: Verso, 1989).
biological weapons production. We saw U.S. soldiers equipped with night vision devices as well as U.S. pilots and soldiers training in Ilight and battle simulators. We also saw military advisors scrutinizing and interpreting aerial photographs and satellite imagery. We were also shown Iraq's weak, antiquated resistance against the superior Western technologies oI vision, image making, and image analysis. Iraq Iought back with camouIlage weapons, decoys, and Iinally, according to Western journalists, with oil Iires set to conIuse U.S. surveillance satellites and planes. The GulI War was the combat oI surveillance against camouIlage, visibility against invisibility, human eye against computer eye. This warIare was indeed based on the "logistics oI perception," but we can describe its visual techniques even more precisely. Vision was employed in a limited way as an instrument to capture and represent inIormation about shapes oI objects and their positions in three-dimensional space. The productive employment oI this Iunction oI vision can be seen most dramatically in the example oI radar. A radar screen shows nothing but distances between the radar and other objects in its Iield oI vision. The colors and textures are not important. The guiding laser sensor oI a "smart" missile tracks the same inIormation -- the distance between the missile and the target. Similarly, it is the layout oI a scene, distances between points, dimensions oI objects which are recorded by a satellite photograph. The multitude oI vision technologies which only "see" distances and dimensions are paralleled by the reduction oI human vision to the same single Iunction. Thus, the images Irom traditional or inIrared video cameras, which contain inIormation about various Ieatures oI the visual world, are used by a soldier or a pilot in a much more limited way -- as a source oI inIormation about distances and shapes in a scene. What is the shape oI an object on the horizon -- is it an enemy or a Iriend? What is the distance between me and the enemy? Finally, the technologies (such as radar) and techniques (using human sight in a Iunction oI a radar) oI obtaining inIormation about shapes and distances in the real world are
complemented by 3-D computer graphics that simulate the real world on the basis oI such inIormation. In their turn, simulated images generated with 3-D computer graphics are used in Ilight and battle simulators, where the military eye undergoes its training to learn how to reduce the world to geometry and topology. The use oI technologies which capture and visualize inIormation about shapes and distances today extends beyond warIare into all spheres oI industry and science. The pilots oI commercial airlines utilize the same radar screens and are trained in the same Ilight simulators as Air Force pilots. 153 Engineers and architects use 3-D graphics to visualize their designs. Oil-rich
153 As I write this in early 1993, many companies which yesterday supplied very expensive simulators to the military are busy converting them into entertainment arcade-based systems. In Iact, one oI the Iirst such systems already commercially operating in a number oI major cities, including Chicago and Tokyo -- Battletech Center Irom Virtual World Entertainment, Inc. -- is directly modeled on SIMNET (Simulation Network) developed by DARPA (DeIense Advanced Research Projects Agency). SIMNET can be thought oI as the Iirst model oI cyberspace, the very Iirst collaborative VR environment. SIMNET consists oI a number oI individual simulators, networked together, each containing a copy oI the world database and the virtual representation oI all other participants in the conIlict such as the Kuwaiti theater oI operations. Similarly, a Battletech Center comprises a networked collection oI Iuturistic cockpit models with VR gear. For $7 each, seven players can Iight each other in a simulated environment. In another example, in 1992 Lucas Arts has teamed up with Hughes AircraIt, combining the expertise in computer games oI the Iormer with the expertise in building actual Ilight simulators oI the latter, in a joint venture aimed at theme-park type rides. On the connection between SIMNET and Battletech Centers, see Tony Reveaux, "Virtual Reality Gets Real," New Media (January 1993): 36-41. On VR entertainment systems in the
countries pay a high price Ior satellite images oI their territories to help them discover potential oil Iields. And, in automated Iactories, tireless robotic arms move parts under the control oI computerized vision systems trained to recognize the shapes oI these parts.
In this chapter I will discuss twentieth century automation oI the use oI vision to record and visualize geometric and topographic inIormation -- to chart the distance and shape oI real and imaginary scenes and objects. This Iunction oI vision complements another Iunction which was the subject oI the previous chapter: the visual representation oI abstract ideas and their logical relations. Here we will deal with the opposite problem -- how to represent a single, concrete object or concrete scene. How, Ior instance, to represent not an idea oI a chair but a particular chair with a unique shape and dimensions. These two kinds oI reIerents -- abstract ideas and arguments, on the one hand, and concrete three-dimensional shapes, on the other hand, can be related to two long-standing philosophical positions on what constitutes reality. For Plato, sensible particulars were but pale reIlections oI Ideas or Forms. Aristotle criticized Plato, declaring the primary substances were not Ideas but the individual things such as particular men or animals. These opposing views have continued to be debated in scholastic philosophy, with Plato's view giving rise to realism and Aristotle's to nominalism. 154
context oI location-based entertainment -- arcades and theme parks -- see Richard Cook, "Serious Entertainment," Computer Graphics World (May 1992): 40-48. 154 Allan Sekula relies on this distinction between the philosophical positions oI realism and nominalism in his discussion oI diIIerent systems oI police photography in the end oI the
In scholastic philosophy the terms "realism" and "nominalism" reIer to two opposing ontological positions: realism claiming the primacy oI Ideas and nominalism -- the primacy oI individual things. 155 Although in contemporary thought the meanings oI "realism" and "nominalism" are somewhat diIIerent, the original meanings are helpIul in distinguishing two Iundamental kinds oI representations: abstract ideas versus concrete objects. I have suggested in the previous chapter that the desire to visualize abstract ideas and logical relations was new to modernity. But it would not be reasonable to propose that the Iunction oI vision which can be called visual nominalism -- to capture the identity oI individual objects and spaces by recording distances and shapes -- originates in the twentieth century as well. AIter all, maps were used Ior many centuries; and the sophisticated graphic language oI engineering drawings was well developed by the nineteenth century. Yet, as this chapter will try to demonstrate, in this century the automation oI visual nominalism, which began with the Renaissance perspective, entered a new stage. The sign oI this automation is the multitude oI new technologies used to capture and visualize three-dimensional reality that have emerged since
nineteenth century. Allan Sekula, "The Body and the Archive," October 39 (1987): 3-64. 155 Antony Flew, ed., A Dictionary oI Philosophy (London: Pan Books Ltd., 1984), 250, 299.
the middle oI the twentieth century, such as radar, inIrared imaging, laser sensors, CAT scans, magnetic resonance imagining, 3-D computer graphics, and computer holography. Finally, since the early 1960s work has been under way to automate vision completely, to create computer vision systems that will recognize objects and interpret scenes automatically. The development oI these technologies has been accompanied by massive research into the general problems oI visual nominalism in computer science, experimental psychology, and neuroscience. New Iormal mathematical techniques were developed to analyze images as a source oI depth inIormation and, vice versa, to transIorm this inIormation into realistic images. The work on the automation oI visual nominalism has also lead to new attention to particular aspects oI human perception. A new paradigm Ior the study oI human vision emerged during the 1970s at MIT associated with the name oI David Marr. 156 Within this paradigm, the goal oI human perception is taken to be the recognition oI shapes, leading researchers to study algorithms by which the brain "computes" shapes oI objects Irom retinal input in the hope that these algorithms can then be used by computer vision systems. This chapter brings together seemingly unrelated visual technologies and techniques: radar, 3-D computer graphics, and computer vision. I will suggest that they are ultimately related by the Iunction they all serve -- visual nominalism. Until this century, this Iunction has been perIormed by techniques oI perspectival representation such as perspective and descriptive geometry and, later, photography. Rather than thinking oI them as artistic techniques, such writers as William Ivins and Bruno Latour emphasize their signiIicance Ior the development oI science, modern technology, and the economy. What these techniques allow best is to represent precisely three-dimensional inIormation on a two-dimensional plane, thus recording quantitatively and eIIiciently the knowledge oI the surrounding space. New twentieth century techniques Iurther extend the useIulness oI these kinds oI representations. Radar allows to obtain
156 David Marr, Vision (New York: W.H. Freeman and Company, 1982).
spatial inIormation without the limitations oI visibility (section 3); 3-D computer graphics add the dimension oI interactivity (section 4); and computer vision automates visual nominalism altogether (section 5).
While there is no history oI the burgeoning techniques oI visual nominalism in the twentieth century, the earlier history oI these techniques has been studied in detail by art historians. The problem oI representing three-dimensional space upon a two dimensional plane has been a subject oI art historical research Irom at least the beginning oI the century. It is quite remarkable that the people who are today recognized as the Iounders oI modern art history (Alois Riegl, Heinrich WslIIlin, and Erwin PanoIsky), all deIined the Iield as the history oI the representation oI space. Working within the paradigms oI cyclic cultural development and racial topology, they related the representation oI space in art to the spirit oI entire epochs, civilizations and races. In his 1901 Die SpStrsmische Kunstindustrie, Riegl characterized humankind's cultural development as the oscillation between two extreme poles, two ways to understand space, which he called "haptic" and "optic." Haptic perception isolates the object in the Iield as a discrete entity, while optic perception uniIies objects in a spatial continuum. 157 Riegl's contemporary, Heinrich WslIIlin, similarly proposed that the temperament oI a period or a nation expresses itselI in a particular mode oI seeing and representing space. WslIIlin's Principles oI Art History (1913) plots the diIIerence between Renaissance and Baroque on Iive dimensions: linear -- painterly, plane -- recession, closed Iorm -- open Iorm, multiplicity -- unity, and clearness -- unclearness. Finally, another Iounder oI modern art history, Erwin PanoIsky, contrasted the
157 Michael Ann Holly, PanoIsky and the Foundations oI Art History (Ithaca: Cornell University Press, 1984), 73. 2. "The most important event of the Renaissance."
"aggregate" space oI the Greeks with the "systematic" space oI the Italian Renaissance in the essay Perspective as a Symbolic Form (1924-1925). 158 PanoIsky established a parallel between the history oI spatial representation and the evolution oI abstract thought. The Iormer moves Irom the space oI individual objects in antiquity to the representation oI space as continuous and systematic in modernity; in PanoIsky's neologisms, Irom "aggregate" space to "systematic" space. Correspondingly, the evolution oI abstract thought progresses Irom ancient philosophy's view oI the physical universe as discontinuous to the post-Renaissance understanding oI space as inIinite, ontologically primal in relation to bodies, homogeneous, and isotropic -- in short, as "systematic." PanoIsky's 1924-25 essay has been recognized as the Iirst in what became an ever- growing series oI interpretations oI perspective. 159 These interpretations related perspective to every known characteristic oI the modern period: economic, social, and philosophical. Some oI these interpretations acquired the character oI dogmas, selI-evident truths. For instance, the correlation between Cartesian ideas oI rational subjectivity in philosophy and Renaissance perspective in visual art has been presented as one oI the central metaphors in interpreting
158 PanoIsky, "Die Perspektive als 'symbolische Form.' 159 The recent very important work on perspective is Hubert Damisch, L'origine de la perspective (Paris: Flammarion, 1987).
Western modern culture. This concept became so prevalent that in his critical analysis Martin Jay Ieels justiIied to name it "Cartesian perspectivalism." 160
Most interpreters have singled out an aspect oI perspective, which then become Ior them the core oI its symbolic Iunction. For instance, Pierre Francastel, and Iollowing him, John Berger, and many Iilm theorists oI the 1960s and 1970s, Iocused on the mechanical character oI perspective and its capacity to organize the representation oI reality around a single point oI view. They theorized a connection between Renaissance perspective and the development oI early capitalism, the emergence oI instrumental consciousness which measures everything, which is concerned with solidity and extension, with numbers and equivalents -- the consciousness oI market and accounting. Representations oI cosmic, religious, unmeasurable space were replaced by the pictures oI cultivated, appropriated landscapes or oI rooms cluttered by precisely drawn objects signiIying wealth. Perspective presented the world as ready to be mastered, consumed, colonized -- the world originating in the eye oI the spectator. 161
160 Martin Jay, "Scopic Regimes oI Modernity," Vision and Visuality, ed. Hal Foster (Seattle: Bay Press, 1988), 4. 161 Andrew Dudley, Concepts in Film Theory (OxIord: OxIord University Press, 1984), 31.
Other writers Ioregrounded diIIerent characteristics oI perspective in their interpretations. For instance, Ior RudolI Arnheim, a pyramidal world portrayed by perspective signiIies a hierarchical conception oI human existence. 162 On the other hand, Ior Fernande Saint-Martin, perspectival representation which can only show objects at a distance, corresponds to the "conceptual rather than sensorial, 'ideological' rather than concretely experienced" relation oI a subject to the world. 163
PanoIsky's own account oI perspective is more subtle than many oI the consequent interpretations it inspired. PanoIsky correlates a mode oI visual representation with a philosophical system, but he is careIul not to read one as an expression oI the other. His conclusion is open: perspective can be equally interpreted as subjective and objective, since it presents space as an objective phenomenon but inevitably perceived Irom a subjective point oI view. Calling perspective "a two-edged sword," PanoIsky writes:
Perspective subjects the artistic phenomenon to stable and even mathematically exact rules, but on the other hand, makes the phenomenon contingent upon human beings, indeed upon the individual: Ior these rules reIer to the psychological and physical conditions oI the visual impression, and the way they take eIIect is determined by the Ireely chosen position oI a subjective "point oI view." 164
162 RudolI Arnheim, Art and Visual Perception. The New Version (Berkeley: University oI CaliIornia Press, 1974), 295. 163 Fernande Saint-Martin, Semiotics oI Visual Language (Bloomington and Indianapolis: 1990), 141. 164 And while modern perspective, as a coherent unity, is contrasted by PanoIsky to the modes oI representation in antiquity and the Middle Ages, he also suggests that within the perspective mode there exist many distinct variations. Working within the same mode oI perspective, diIIerent cultures and epochs take its meaning to be diIIerent: the Dutch represent "near space," Germans Iavor "oblique space," Italians explore "high space."
While PanoIsky's essay had to wait almost six decades beIore its publication in English, another equally important essay on perspective, "Obratnaya Perspektiva" (The Inverted Perspective) still remains untranslated and unknown to Western art historians. 165 In his article Joseph Leo Koerner brieIly reIers to the essay and regrets its unavailability in English (this is the only reIerence I am aware oI). 166 The essay was written by the Russian philosopher, linguist, mathematician and art historian Father Pavel Florensky, who headed the department oI the
Ervin PanoIsky, Perspective as Symbolic Form, 68-70. 165 Pavel Florensky, "Obratnaya Perspektiva (The Inverted Perspective)," Sobranie Sochineniy, ed. N.A. Struve (Paris: YMCA-PRESS, 1985), 1: 117-192. 166 Koerner, "The Shock oI the View," 37-38.
Analysis oI Spatial Representation in the Arts at VKhUTEMAS Irom 1921-1924. 167 Florensky's essay can be justly considered as the counterpart to PanoIsky's work, completed in 1922, two years beIore PanoIsky's work. Both PanoIsky and Florensky interpret any perspectival system as convention and oppose the view that any particular system is natural. 168 At the same time, their arguments tend to Iavor antithetical perspectival systems and world views: linear perspective and Renaissance individualism in the case oI PanoIsky; inverted perspective and medieval religious collectivism in the case oI Florensky. PanoIsky narrates the evolution oI the representation oI space as an inevitable progression towards the "systematic" yet subjective space oI the Renaissance; thus, the Renaissance perspective is given a privileged status, its Iinal triumph as inevitable as the progress oI history itselI. The discussion oI medieval representation oI space becomes Ior
167 VKhUTEMAS (State High Art- Technical Studious) was the leading school oI art and design in the USSR in the 1920s and the center oI avant-garde artistic culture. 168 Florensky writes: "Is it true that perspective, as it is claimed by its supporters, expresses the true nature oI things and thereIore should be everywhere and always understood as the absolute condition oI artistic truth? Or is it only a scheme, only one among many schemes oI representation, corresponding not to the universal world view but only to one possible understanding oI the world, connected with a particular sensibility and cognizance?" Florensky, "Obratnaya Perspektiva," 123.
PanoIsky nothing more than a brilliant narrative device, a dramatic detour in what otherwise is presented as a linear course Irom Antiquity to Renaissance: "When work on certain artistic problems has advanced so Iar that Iurther work in the same direction, proceeding Irom the same premises, appears to bear Iruit, the result is oIten a great recoil, or perhaps better, a reversal oI direction." 169 In PanoIsky's interpretation, in other words, medieval representation oI space serves as a dialectical antithesis between ancient and Renaissance representations. Thus, PanoIsky claims, while it may appear that medieval representation gave up the advance oI Antiquity in repressing three-dimensionality, in Iact it approached the Renaissance understanding oI space as inIinite and homogeneous: "For iI Romanesque painting reduced bodies and space to surIace, in the same way and with the same decisiveness, by these very means it also managed Ior the Iirst time to conIirm and establish the homogeneity oI bodies and space." 170
Florensky, on the other hand, is mostly interested in the medieval representation oI space as maniIested by the spatial constructions oI Russian icons between the Iourteenth and sixteenth centuries. According to Florensky, seemingly heterogeneous spatial constructions oI icons were neither a result oI the lack oI artistic skill nor a step towards the Renaissance perspective. 171
169 PanoIsky, Perspective as Symbolic Form, 47. 170 Ibid., 51. 171 The view oI medieval art as unsystematic representation oI space, against which Florensky revolted, continues to persist. Thus, Samuel Edgerton writes in 1975: "Unlike the Renaissance painter depicting his scene in perspective, the medieval artist viewed his world quite subjectively. He saw each element in his composition separately and independently, and thus
Rather, they are result oI a conscious and coherent system Ior representing reality which corresponds to a coherent world view. It is not hard to see that Ior the orthodox and slavophile Florensky, this world view is superior to Western post-Renaissance individualism. Russian artists, like the artists oI ancient Egypt and China, were aware oI perspective but consciously reIused its power, choosing instead "religious objectivity and super-personal metaphysics." For
paid little attention to anything in the way oI systematic spatial relationship between objects."
Samuel Edgerton, The Renaissance Rediscovery oI Linear Perspective (New York: Basic Books, 1975), 21.
Florensky, the Western Renaissance represents the beginning oI the Fall: "When religious stability oI the world view begins to disintegrate, and the sacred metaphysics oI collective popular consciousness is segmented by individual vision oI an individual subject Irom an individual point oI view, and only in a particular moment oI time -- only then perspectivalism, symptomatic oI disunited consciousness, emerges." 172
Florensky's Slavophilism strikes us today as being politically biased; yet, this bias makes us realize that PanoIsky's seemingly more objective account is also biased in its privileging oI what Florensky labeled "bourgeois individualism." While Florensky oIIered a theoretical critique oI linear perspective, Ior his compatriots at the time this critique became a practical problem: what new visual Iorms shall be suitable Ior the "Iirst people's state"? Similarly identiIying linear perspective with "bourgeois individualism," El Lissitsky and Kazimir Malevich relied in their designs on other types oI perspectival construction such as parallel projection. In this type oI projection, all projection rays are parallel to each other, instead oI originating Irom a single point oI view. Parallel projection, with its absence oI a single point oI view, was presented as a more suitable way to symbolize the vision and knowledge oI a collective. Paradoxically, this "objective" point oI view oI a collective was expressed in the image (commonly encountered in
172 Florensky, "Obratnaya Perspektiva," 125-126. JustiIying the value oI reverse perspective as a symbol oI "super-personal metaphysics," Florensky oIIered a critique oI the arguments later to be Iound in PanoIsky's "Die Perspektive als 'symbolische Form.'" For instance, he precisely articulates the notion oI "systematic" space, which, according to PanoIsky, represented the triumph oI the Renaissance and was still absent in Medieval thought: "II we summarize everything which is said in a Iormal sense against Medieval art, they are reduced to one criticism: 'There is no understanding oI space.' This criticism means that there is no spatial unity, no scheme oI Euclid's and Kant's space, the latter reduced, within the Iramework oI painting, to linear perspective and proportionality."
Malevich's writings) oI the individual artist looking at the Earth Irom an inIinitely Iar away point in outer space, so that the convergence oI rays is replaced by a parallel projection. 173
173
A.G. Rappoport, "El Lissitsky i ego 'Pun- geometiya'" (El Lissitsky and his 'Pan-geometry'), Sovetskoe Iskusstvoznanie 25 (1989): 118- 119. Four decades later, LANDSAT satellites start bringing back images oI Earth objects in axonometric projections (because they were shot Irom high orbits) thus IulIilling Malevich's designs.
The reIusal oI linear perspective by such artists as Malevich or Lissitsky is usually seen as just one episode in what by now became a widely accepted narrative -- the shattering and negation oI perspectival space by modernist artists. According to this narrative, perspective was already dead by the time art historians such as PanoIsky had begun writing its history. Such narrative is announced, Ior instance, in the very title oI Pierre Francastel's Painting and Society. Birth and Destruction oI Plastic Space Irom the Renaissance to Cubism (1952). The opening section oI The Production oI Space by Henri LeIebvre is equally authoritative:
The Iact is that around 1910 a certain space was shattered. It was a space oI common sense, oI knowledge (savoir), oI social practice, or political power...a space, too, oI classical perspective and geometry, developed Irom the Renaissance onwards on the basis oI the Greek tradition (Euclid, logic) and bodied Iorth in Western art and philosophy, as in the Iorm oI the city and town. 174
Yet, iI perspective disappeared Irom modern art, it survived as one oI techniques oI the visual nominalism, a method Ior precisely representing the three-dimensional world on a two- dimensional surIace. In this role, it extended into many new domains and became the Ioundation oI new kinds oI automated technologies oI remote sensing and image synthesis. To consider perspective in its role as a technique oI visual nominalism we should turn to another interpreter oI perspective -- William Ivins and his short 1939 essay On the Rationalization oI Sight. 175 II PanoIsky connects the development oI perspective with the idea oI inIinite space, abstract space existing prior to objects, Ivins on the contrary emphasizes that perspective allows the creation oI precise maps oI three-dimensional reality, to record the shapes oI concrete objects and the layout oI concrete spaces. It is the tool oI a businessman and a scientist rather than an artist. In Ivins' deIinition, perspective is "a practical means Ior securing a rigorous two-way, or reciprocal, metrical relationship between the shapes oI objects as deIinitely located in space and
174 Henri LeIebvre, The Production oI Space (OxIord: Blackwell Publishers, 1991), 25. 175 William M. Ivins, On the Rationalization oI Sight (New York: Da Capo Press, Inc., 1975).
their representations." 176 This deIinition sums up a number oI unique characteristics oI perspectival projection. First, the size oI graphic marks in a perspectival image is proportionally related to the dimensions oI represented objects. For instance, a Ioot-long object would be represented by a line oI three inches, while an object twice as long would become a line oI six inches, and so on. Second, the ratio oI sizes between these lines is proportional to the ratio oI sizes between the corresponding objects. Third, the perspectival image does not only record the shapes and dimensions oI objects but also contains inIormation about their distance Irom one another (in contrast to parallel projection, Ior instance). Finally, all these relationships between the perspectival symbols on the one hand, and between the symbols and the corresponding objects on the other hand, are controlled by a single rule or, to use the language which would become common within a decade aIter the publication oI On the Rationalization oI Sight, by a single algorithm. Given the coordinates oI the point oI view and the projection surIace, the projection oI any point in space can be determined by mechanically Iollowing a sequence oI steps. The most important quality oI perspective Ioregrounded in Ivins' deIinition is the precise and reciprocal relationship it establishes between objects and their symbols. We can go Irom objects to symbols (two-dimensional representations); but we can also go Irom such symbols to three-dimensional objects. Using the rules oI perspective, an architect could create a visual representation oI an already existing building, change this representation, and the workers would execute the speciIied changes in the building itselI. Better yet, our architect can design a brand new building -- not by painstakingly constructing a small model, where each change takes time to implement -- but by creating a perspectival drawing, where each Iorm is just a line and change requires only a single movement oI an eraser (or a single press oI a key on a computer keyboard, iI the architect is working with a CAD system).
176 Ibid., 9.
Bruno Latour extended Ivins' idea by pointing out that a reciprocal relationship made possible by perspective allows us not only to represent reality but also to control it. 177 Latour sees perspectival representations as the "most powerIul instrument oI power," deIined as the ability to mobilize resources across space and time, to manipulate these resources at a distance. For instance, we cannot measure the sun in space directly, but we only need a small ruler to
177 Bruno Latour, "Visualization and Cognition: Thinking with Eyes and Hands," Knowledge and Society: Studies in the Sociology oI Culture Past and Present 6 (1986): 1-40.
measure it on a photograph (the perspectival image par excellence). 178 And even iI we could Ily around the sun, we would still be better oII studying the sun through its representations which we can bring back Irom the trip -- because now we have unlimited time to measure, analyze, and catalog them. We can also represent absent things and plan our movement through space by working on representations: "One cannot smell or hear or touch Sakhalin Island, but you can look at the map and determine at which bearing you will see the land when you send the next Ileet." 179 We can move objects Irom one place to another by simply moving their representations: "You can see a church in Rome, and carry it with you in London in such a way as to reconstruct it in London, or you can go back to Rome and amend the picture." Finally, as Latour points out, "the two ways become a Iour-lane Ireeway! Impossible palaces can be drawn realistically, but it is also possible to draw possible objects as iI they were utopian ones." Real and imagined objects can meet on a Ilat space oI perspectival representation. All in all, perspective is more than just a sign system, reIlecting reality -- it makes possible the manipulation oI reality through the manipulation oI its symbols.
178 Ibid., 22. 179 Ibid., 8.
Citing perspective as a perIect example oI a new instrument oI power and domination, Latour proposes that the history oI science and technology should be understood as a cascade oI such innovations in representation. Yet, his argument is already anticipated by Ivins. Ivins concludes his essay by stating that the beginning oI the rationalization oI sight through the discovery and the development oI perspective "was the most important event oI the Renaissance." 180 It was more important than the Iall oI Constantinople or the ReIormation or the Counter ReIormation because it made all other discoveries possible by launching a constantly accelerating series oI innovations. 181 The invention oI perspective propelled modern empirical science, Ior instance biology, which could now represent Iorms oI nature with mathematical precision. It also stimulated the rise oI modern engineering and manuIacturing by making Ieasible the distribution oI identical designs to Iar away places. 182
Ivins' approach stands in sharp contrast to the more traditional art historical analyses oI perspective by PanoIsky and Francastel. They are concerned with perspective as an artistic Iorm and do not look beyond its history in art. Ivins, on the contrary, is concerned with visual culture - - the techniques and technologies oI visual representation available to a society at a given moment and the Iundamental role they play in shaping every aspect oI society. Because he sees these techniques and technologies 183 as ultimately determining social development, his approach is deeply materialistic. Perspective, Ior Ivins, is not an artistic Iorm reIlecting social or philosophical modernity; it is the very condition which makes modernity and modernization possible. 184 For instance, he writes: "Many reasons are assigned Ior the mechanization oI liIe
180 Ivins, On the Rationalization oI Sight, 13. 181 Ibid., 7, 12. 182 Ibid., 13. 183 Ivins Iocuses on two Iundamental developments: Iirst, the invention oI perspective and its development into descriptive and perspective geometry; second, the invention oI technique Ior printing pictures which made possible easy duplication oI exact copies oI the same image. 184 This is the theme which Latour also greatly extends, proposing "a new kind oI materialism" which will account Ior social and scientiIic history through the analysis oI available Iorms oI representation.
and industry during the nineteenth century, but the mathematical development oI perspective was absolutely prerequisite to it." 185 Ivins is also a materialist in the most extreme Marxist sense because he suggests that the available Iorms oI representation, such as perspective, determine the development oI abstract thought. He points out that as important as the development oI perspective was Ior picture making, "it is doubtless even more important to general thought, because the premises on which it is based are implicit in every statement made with its aid." 186
This idea is repeated again at the end oI the essay:
The constant extensions oI the Iields oI useIulness oI the pictorial symbol that is precisely duplicable |through printing| and oI the grammars oI its use |development oI perspective and descriptive geometry| have had a most astonishing eIIect not only upon knowledge but upon thought and its basic assumptions or intuitions...Relativity, which now in one Iorm or another runs throughout contemporary thought and practice, is in large measure a development oI ideas that were evolved through the study and use oI projective transIormations. 187
It is Ivins' general Iocus on visual culture, rather than art, that makes him attentive to the current use oI perspective and its continuing development and expansion. While PanoIsky or Francastel write about perspective in the past tense aware oI its demise in modern art, Ivins on the contrary notes "that today there are Iew sciences and technologies that are not predicated in one way or another upon this power oI invariant pictorial symbolization." 188 Modern designers, scientists or engineers, oI course, do not simply use perspective as it was Iormulated by Alberti in the IiIteenth century; they use much more sophisticated techniques. According to Ivins, the rationalization oI perspectival sight proceeded in two directions. On the one hand, perspective became the Ioundation Ior the development oI the techniques oI descriptive and perspective geometry which became the standard visual language oI modern engineers and architects (Iig. 11). On the other hand, the photographic technologies automated the creation oI perspectival
185 Ivins, On the Rationalization oI Sight, 12. 186 Ibid., 9. Emphasis mine - L.M. 187 Ibid., 13. 188 Ibid.
images. Both were accomplishments oI the nineteenth century; in Iact, both were developed more or less simultaneously. Indeed, as Ivins points out, Ni pce and Talbot, the Iounders oI photography, were contemporaries oI Monge and Poncelet, decisive Iigures in the development oI descriptive and perspective geometry.
Writing On Rationalization oI Sight between 1936 and 1938, Ivins mentions such examples oI the contemporary use oI perspective as aerial photographic surveillance, classiIication in the Iield oI archeology, and criminal detection. 189 However, all these applications oI perspectival techniques already existed in the nineteenth century and, by the 1930s, did not represent the "latest" developments. While photo reconnaissance was Iirst employed systematically on a mass scale during World War I, the interest in using photography Ior aerial surveillance existed since its invention. FZlix Tournachon Nadar, one oI the most eminent photographers oI the nineteenth century, Iamous Ior his portraits, succeeded in exposing a photographic plate at 262 Ieet over Bi vre, France in 1858. He was soon approached by the French Army to attempt photo reconnaissance but rejected the oIIer. In 1882, unmanned photo balloons were already in the air; a little later, they were joined by photo rockets both in France and in Germany. The only innovation oI World War I was to combine aerial cameras with a superior Ilying platIorm: the airplane. 190
189 Ibid., 12, 13. 190 Beaumont Newhall, Airborne Camera (New York: Hastings House, Publishers, 1969). For critical histories oI photo reconnaissance see Allan Sekula, "The Instrumental Image: Steichen at War," in Photography against the Grain: Essays and Photo Works, 1973-1983 (HaliIax: The Press oI the Nova Scotia College oI Art and Design, 1984); Paul Virilio, War and Cinema: the Logistics oI Perception (London: Verso, 1989); 3. Radar: Seeing Without Eyes
In 1858, Albrecht Meydenbauer, a director oI the German Government Building OIIice, published a proposal to use photographs Ior scale measurement. His proposal was based on the existence oI a geometrical relationship between the photographic image and the object being photographed. Why, Ior instance, climb a Iacade oI a cathedral in order to measure it (as Meydenbauer had to do, nearly once getting killed) when it is much saIer to measure a photograph? Additionally, wrote Meydenbauer, "some may Iind it hard to believe, but
Manuel De Landa, "Policing the Spectrum," in War in the Age oI Intelligent Machines (New York: Zone Books, 1991) .
experience has proven than one can see, not everything, but many things, better in scale measurement than on the spot." In 1885 the Royal Prussian Institute Ior Scale Measurement was Iounded and the measurement oI photographs oI historic monuments became a Irequent practice. 191
In the Iield oI criminal detection, perspectival technology oI photography was routinely employed decades beIore the time oI Ivins' writing. In his 1844 Pencil oI Nature William Fox Talbot, one oI photography's Iounders, comments on a beautiIul calotype depicting several shelves oI china: "should a thieI aIterwards purloin the treasures -- iI the mute testimony oI the picture were to be produced against him in court -- it would certainly be evidence oI a novel kind." 192 In 1883, Alphonse Bertillon, director oI the IdentiIication Bureau oI the Paris PreIecture oI Police, developed a system Ior classiIication and identiIication oI criminals that relied on photographs. By 1893, Bertillon's system was already employed in the United States, Belgium, Switzerland, Russia, much oI South America, Tunisia, the British West Indies, and Romania. 193
Modern perspective and descriptive geometry were Iully developed in the Iirst halI oI the nineteenth century. Photography was already utilized by a number oI proIessions Ior measurement, identiIication, and classiIication by the end oI the century. What was the next step in the "rationalization oI sight"? In Iact, while Ivins was writing his essay on perspective, across the Atlantic, in England, work was already underway to install twenty radar stations on the east and southeast coasts to provide surveillance oI these air approaches. These radar installations turned out to be absolutely essential in the coming war, allowing Ior the severely outnumbered Royal Air Force to deIeat the
191 Harun Farocki, "Reality Would Have to Begin," Documents 1/2 (1992): 136-146. 192 Qtd. in Sekula, "The Body and the Archive," 6. 193 Sekula, "The Body and the Archive," 25, 35.
LuItwaIIe in the Battle oI Britain. Radar, the latest technology oI visual nominalism, became Britain's most important weapon. 194
The Iollowing anecdote about the "discovery" oI radar, regardless oI its historical accuracy, is useIul in understanding how radar works. 195 In 1922, two civilian scientists employed by the Navy set out to conduct a routine communication experiment along the arm oI the Potomac river. As usual, they set a short-wave radio receiver and transmitter along the two banks oI the river. This time, however, a passing steamer interrupted the signal, creating an electronic echo. The scientists immediately realized that such echoes could be used to deduce the locations oI ships entering a harbor.
This anecdote illustrates the main principle oI radar operation. Radar is an acronym Ior Radio Detection and Ranging. Like sound waves, radio waves create echoes when they are interrupted by objects in their path. Radar transmits a radio wave in a particular direction. The signal reIlected back Irom the objects is picked up by an antenna. The time between the transmission and the reception oI the echo indicates the distance to the object; the direction the antenna is pointing when the echo is received reveals the object's position in relation to the radar. Detected objects appear as bright spots on a display watched by a radar operator. 196
Radar exempliIies the Iurther rationalization oI visual nominalism. All it sees and all it shows are the positions oI objects, 3-D coordinates oI points in space, points which correspond to submarines, aircraIt, birds or missiles. Color, texture, even shape are disregarded. Instead oI Alberti's window, opening onto the Iull richness oI the visible world, a radar operator sees a
194 For the history and technology oI radar, I mainly relied on two sources: Echoes oI War (Boston: WGBH Boston, n.d.), videotape; McGraw-Hill Encyclopedia oI Science & Technology: an International ReIerence Work in Twenty Volumes Including Index (New York: McGraw-Hill, 1992.) 195 Echoes oI War (Boston: WGBH Boston, n.d.), videotape. 196 Numerous variations oI basic radar technology exist. For instance, in addition to active radars which send a signal and detect energy reIlected by objects there are also passive radars which do not send a signal themselves. However, all radars have in common the use oI electromagnetic radiation (radio waves) to detect and measure objects in their vicinity.
screen, a dark Iield with a Iew bright spots. Here, the Iunction oI visual nominalism, which perspectival image perIormed along with many other Iunctions (representation oI light, color, shade, texture, etc.), is isolated and abstracted. Radar serves a single Iunction -- but it perIorms it more eIIiciently than any previous perspectival technique or technology. First, the detection oI objects' positions in space is no longer limited by conditions oI visibility. Vision is no longer conIined by the distance resolvable by the human or camera eye, since radar can detect objects hundreds oI miles away, and at any time oI day or night. Second, this recording takes place instantly. Previously, military commanders had to wait until pilots come back Irom surveillance missions and the Iilm was developed. The inevitable delay between the time oI the observation and the delivery oI the Iinished image limited its useIulness: since by the time the image was produced, enemy positions could have changed. Now, the imaging is instantaneous. With radar, its vision not limited by visibility, its image literally tracking the reIerent, rationalization oI visual nominalism reaches a new stage. The useIulness oI vision and oI imaging are greatly extended. This expansion is achieved by Iundamentally redeIining the very Ioundations oI vision and imaging technologies. While radar came to supplement perspectival representation and photography in terms oI its Iunction, it also represents a Iundamental break with older technologies. Like the photographic camera, radar captures the positions oI objects and displays them on a Ilat surIace. But how it sees has little resemblance to what was commonly meant by vision as deIined by older optical apparatuses -- Irom human eyes to photography. Instead oI "looking" in a particular direction, radar sees all around, its antenna scanning the area in a circular sweep. Instead oI passively receiving light reIlected by objects, radar sends energy into the environment. Its vision is active; it can be compared to exploration oI the
environment by a blind person. 197 Instead oI relying, like photography, on the small region oI the electromagnetic spectrum to which our eyes are sensitive, it uses other regions, sending and receiving waves oI diIIerent lengths. Vision is no longer limited by the spectral capacity oI the human or camera eye; it is extended to include the whole oI electromagnetic spectrum. The visible becomes a small part oI a larger Iield oI sensory exploration oI the environment. Images produced by radar are also Iundamentally diIIerent than those oI previous imaging technologies, Irom drawing paper to photographic Iilm. The photographic image is an imprint corresponding to a single reIerent or to a limited time oI observation. With the invention oI the radar screen, the image surIace is no longer static. For the Iirst time it becomes constantly updatable in real time. Here, with radar, we see Ior the Iirst time the mass employment oI a Iundamentally new kind oI display, which soon comes to dominate modern visual culture, Iound today in video or computer screen. What is new about such a display is that its image can change in real time reIlecting changes in the reIerent, be it the position oI an object in space (radar), any change in visible reality (live video) or changing data in computer memory (computer screen). The radar screen changes, tracking the reIerent. But while it appears that the element oI time delay always previously present in imaging is eliminated, in Iact time enters radar image in a new way. With older photographic technologies, all parts oI an image are exposed simultaneously. Now the image is produced through sequential scanning, which means that its contiguous parts in Iact correspond to diIIerent moments in time. Like in an audio record, moments in time become circular tracks on a surIace.
Along with radar, many other technologies oI visual nominalism came into existence Iollowing the advances in electronics and computers during World War II: ultrasonic imaging,
197 There also exist passive radars which have no transmitters but are equipped to measure signals Irom targets themselves. The active radar can be thought as a historical continuation oI projectors previously used by the military to artiIicially light up the scene oI a battle to increase visibility.
multispectral photography, multispectral imaging, inIrared, sonar, magnetic resonance imaging, and so on. 198 As radar, these technologies are eIIectively used to record distance, position, layout, shape, and volume. Sonar, Ior instance, detects objects in the water by using sound waves. Ultrasonic computer tomography uses sound waves and computer graphics to construct images oI body tissues. Multispectral photography isolates energy reIlected Irom surIaces in a number oI given wavelength bands. Coming to supplement perspectival techniques and technologies, the new technologies oI visual nominalism are more eIIicient but also more specialized. II older optical technologies indiscriminately capture a number oI dimensions oI visual reality (color, shape, texture, tonality), now a single dimension can be isolated. More precisely, a single dimension, in this case 3-D inIormation, is broken into distinct components recorded by diIIerent imaging technologies. The technique oI perspective allows one to record both the shape oI objects and their relative positions in relation to each other. Radar only records positions, it is blind to shape. On the other hand, a laser or ultrasound range Iinder produces a depth map, treating as shapes both the actual objects and the empty space between them. This specialization can be seen as a side eIIect oI the automation oI visual nominalism. This automation requires a priori assumptions about the nature oI reality which become embedded in diIIerent technologies. In the case oI radar, reality is reduced to a set oI geometric points which have no extension. In the case oI a range Iinder, reality is treated as iI a skin, a membrane, a continuous 3-D Iorm. We may recall here the Iar-seeing prediction oI the eIIects oI photography published in 1859 by Oliver Wendell Holmes: "Every conceivable object oI Nature and Art will soon scale oII its surIace Ior us. Men will hunt all curious, beautiIul, grand objects,
198 See McGraw-Hill Encyclopedia oI Science & Technology.
as they hunt cattle in South America, Ior their skins and leave the carcasses as oI little worth." 199
Holmes' words seem to come true not only metaphorically (modern culture which values only images) but also literally, as range Iinders methodically skin reality. Along with the separation between technologies that record positions and technologies that record shapes, another specialization emerges. Depending upon whether the position oI the observer in relation to objects in space is important or not, the technologies oI visual nominalism specialize in creating two diIIerent types oI representations which can be called viewer-oriented and viewer-independent. 200 The Iirst kind oI representation is exempliIied by the radar screen, a record oI distances between enemy targets and the subject -- the point in the center oI the screen. Here, even more so than in the traditional perspective, the world is represented in relation to the observer who becomes the center oI the coordinate system. In the second type oI representation, the distance between the observer and the object is unimportant and in Iact, the position oI the observer cannot even be deduced Irom the image. These are images which employ types oI projection other than perspective: maps, engineering and architectural plans, and aerial and satellite photography, whether conventional, inIrared or multispectral. A third specialization, a third division oI labor brought about by new technologies is that between vision and imaging, which now become distinct processes. Engineering textbooks and encyclopedias group many new technologies oI visual nominalism under the term "remote sensing," deIined as the gathering and imaging oI inIormation without actual physical contact with the object or area being investigated. 201 This deIinition clearly separates the two
199 Oliver Wendel Holmes, "The Stereoscope and the Stereograph," in Photography, Essays and Images: Illustrated Readings in the History oI Photography, ed. Beaumont Newhall (New York: Museum oI Modern Art, 1980), 60. 200 This distinction is related to but not synonymous with the distinction between "viewer- centered" and "object-centered" representations, common since David Marr's Vision (New York: W.H. Freeman and Company, 1982). It is also related, but again, not equivalent to, the diIIerence between perspectival and orthographic projections. The Iirst records reality Irom a single point oI view, the second does not have a single point oI view. 201 McGraw-Hill Encyclopedia oI Science & Technology, 15: 311.
operations: the gathering oI inIormation (vision) and its presentation (imaging). The Iirst operation may have nothing to do with what is visible to the human eye, but in the second operation the eye eventually comes into play since the gathered inIormation has to be presented to the human observer in visual Iorm in order to be useIul. Technologies oI remote sensing convert the data obtained by technologically augmented biological senses (Ior instance, we can think oI radar, which sends waves in order to detect objects, as an augmented sense oI touch) into inIormation perceivable by the human eye. They map other parts oI the electromagnetic spectrum into a small visible part. The invisible (sound waves, microwaves, ultraviolet waves) is converted into the visible. This is why these technologies are also oIten called imaging technologies -- multispectral imaging, ultrasound imaging, inIrared imaging, and so on.
I have emphasized the continuity oI Iunction between perspectival drawing, photography, and newer technologies such as radar and ultrasound imaging. Yet, not only these newer technologies oI visual nominalism serve the same Iunction as the old -- to capture the identity oI individual objects and spaces by recording distances and shapes -- but they rely on the same principle oI perspective. We can justiIiably reIer to them as perspectival technologies iI we understand perspective as extending beyond the domain oI the visible. In his seminar "OI the Gaze as Object Petit a" Jacques Lacan argued that this is how perspective should be understood. 202 He starts by reminding us that an image is anything deIined "by a point-by-point correspondence oI two unities in space." To obtain an image oI something we do not have to rely on light or to operate in the domain oI the visible. Nor do we have to limit images to 2-D representations oI 3-D reality. We can represent an object by another
202 Jacques Lacan, "On the Gaze as Objet Petit a," in The Four Fundamental Concepts oI Psycho-Analysis, ed. Jacques-Alain Miller (New York: W.W. Norton & Company, 1981), 67- 122.
object or represent a 2-D Iorm by another Iorm. All that is required is a rule to establish the correspondence between the points oI the object being imaged and the points on the image. Similarly, says Lacan, "what is an issue in geometric perspective is simply the mapping oI space, not sight." 203 Perspective is one such rule, a particular method to establish a correspondence between the object and its image. The method oI perspective consists oI connecting a single point in space (usually reIerred to as subject's point oI view) with a number oI points on the object by straight lines; the intersection oI these lines with a plane creates an image. It is coincidental that perspective, whether as a part oI the human sight apparatus or as a part oI the photographic apparatus, works through light. Light travels in straight lines, thereIore it can be used to create perspectival images. But one can construct such images without light: "In Descartes, dioptrics, the action oI the eyes, is represented as the conjugated action oI two sticks." 204 As Lacan points out Iurther on in the seminar, this idea that perspective is not limited to sight alone but Iunctions in other senses as well deIines the classical discourse on perception: "The whole trick, the key presto!, oI the classic dialectic around perception, derives Irom the Iact that it deals with geometric vision, that is to say, with vision in so Iar as it is situated in a space that is not in its essence the visual." 205
Lacan's clariIication that the principle oI perspective is not limited to the visible helps us understand that the technologies oI remote sensing Iunction on the principle oI perspective. Regardless oI their lengths, all waves travel in straight lines, and thereIore points in space are connected by straight lines to a point oI reception (such as radar antenna) or recording (such as a photographic camera). Radar, inIrared imaging, sonar, and ultrasound are all part oI what Lacan called "geometric vision," perspectival vision extending beyond the visible.
203 Ibid., 86. 204 Ibid., 87. 205 Ibid., 94.
The technologies oI remote sensing made it possible to map space without the limitations oI visibility. In the next section I will discuss the technology oI 3-D computer graphics, which Iurther extended the useIulness oI perspectival representations by making them interactive.
From the moment oI adaptation oI perspective, artists and draItsmen have attempted to aid the laborious manual process oI creating perspectival images. 206 Between the sixteenth and the nineteenth century various "perspectival machines" were constructed. They were used to construct particularly challenging perspectival images, to illustrate the principles oI perspective, to help students learn how to draw in perspective, to impress artists' clients, or to serve as intellectual toys. Already in the Iirst decades oI the sixteenth century, DYrer described a number oI such machines. 207 One device is a net in the Iorm oI a rectangular grid, stretched between the artist and the subject. Another uses a string representing a line oI sight (Iig. 12). The string is Iixed on one end, while the other end is moved successively to key points on the subject. The point where the string crosses the projection plane, deIined by a wooden Irame, is recorded by two crossed strings. For each position, a hinged board attached to the Irame is moved and the point oI intersection is marked on its surIace. It is hard to claim that such a device, which gave rise to many variations, made the creation oI perspectival images more eIIicient, however the images it helped to produce had reassuring mechanical precision. Other major types oI perspectival machines that appeared subsequently included the perspectograph, pantograph, physionotrace, and optigraph.
206 For a survey oI perspectival instruments, see Martin Kemp, The Science oI Art (New Haven: Yale University Press, 1990), 167-220. 207 Ibid., 171-172. 4. 3-D Computer Graphics: Interactive PerspectivaIism
Why manually move the string imitating the ray oI light Irom point to point? Along with perspectival machines a whole range oI optical apparatuses was in use, particularly Ior depicting landscapes and in conducting topographic surveys. They included versions oI camera obscura Irom large tents to smaller, easily transportable boxes. AIter 1800, the artist's ammunition was strengthened by camera lucida, patented in 1806. 208 Camera lucida utilized a prism with two reIlecting surIaces at 135. The draItsman careIully positioned his eye to see both the image and the drawing surIace below and traced the outline oI the image with a pencil. Optical apparatuses came closer than previous perspectival devices to the automation oI perspectival imaging. However, the images produced by camera obscura or camera lucida were only ephemeral and considerable eIIort was still required to Iix these images. A draItsman had to meticulously trace the image to transIorm it into the permanent Iorm oI a drawing. With photography, this time-consuming process was Iinally eliminated. The process oI imaging physical reality, the creation oI perspectival representations oI real objects was now mechanized. However, this mechanization did not aIIect other uses oI perspectival representation. According to Latour, the greatest advantage oI perspective over other kinds oI representations is that it establishes a "Iour-lane Ireeway" between physical reality and its representation. We can combine real and imagined objects in a single geometric model and go back and Iorth between reality and the model. By the twentieth century, the creation oI a geometric model oI both existing and imagined reality still remained a time consuming manual process, requiring the techniques oI perspectival and analytical geometry, pencil, ruler, and eraser. Similarly, iI one wanted to visualize the model in perspective, hours oI draIting were required. And to view the model Irom another angle, one had to start all over again. The mechanization and automation oI geometrical modeling and display were yet to come.
208 Ibid., 200.
Nothing perhaps symbolizes mechanization as dramatically as the Iirst assembly lines installed by Henry Ford in U.S. Iactories in 1913. The assembly line relied on two crucial principles. The Iirst was the standardization oI parts, already employed in the production oI military uniIorms in the nineteenth century. The second, newer principle, was the separation oI the production process into a set oI repetitive, sequential, and simple activities that could be executed by workers who did not have to master the entire process and could be easily replaced. It seemed that mechanical modernity was at its peak. Yet, in the same year the Spanish inventor Leonardo Torres y Quevedo had already advocated the industrial use oI programmed machines. 209 He pointed out that although automatons existed beIore, they were never used to perIorm useIul work:
The ancient automatons...imitate the appearance and movement oI living beings, but this has not much practical interest, and what is wanted is a class oI apparatus which leaves out the mere visible gestures oI man and attempts to accomplish the results which a living person obtains, thus replacing a man by a machine. 210
With mechanization, work is perIormed by a human but his or her physical labor is augmented by a machine. Automation takes mechanization one step Iurther: the machine is programmed to replace the Iunctions oI human organs oI observation, eIIort, and decision. The term "automation" was coined in 1947; and in 1949 Ford began the construction oI the Iirst automated Iactories. Mass automation was made possible by the development oI digital computers during World War II and thus became synonymous with computerization. A decade later, the automation oI the process oI constructing perspectival images oI both existent and non- existent objects and scenes was well underway. 211 By the early 1960s Boeing designers already
209 Charles Eames and Ray Eames, A Computer Perspective: Background to the Computer Age (Cambridge: Harvard University Press, 1990), 65-67. 210 Qtd. in ibid., 67. 211 I am not aiming here by any means to provide a Iull account oI the history oI 3-D computer graphics or its various uses. I am concerned with computer graphics as one development, among others, in the general move toward the rationalization oI perspectival imaging. For a more
relied on 3-D computer graphics Ior the simulation oI landings on the runway and oI pilot movement in the cockpit (Iig 13). 212
By automating perspectival imaging, digital computers completed the process which began in the Renaissance. This automation became possible because perspectival drawing has always been a step-by-step procedure, an algorithm involving a series oI steps required to project coordinates oI points in 3-D space onto a plane. BeIore computers the steps oI the algorithm were executed by human draItsmen and artists. With a computer, these steps can be executed automatically and, thereIore, much more eIIiciently. The details oI the actual perspective-generating algorithm which could be executed by a computer were published in 1963 by Lawrence G. Roberts, then a graduate student at MIT. 213
The perspective-generating algorithm constructs perspectival images in a manner quite similar to traditional perspectival techniques. In Iact, Roberts had to reIer to German textbooks on perspectival geometry Irom the early 1800s to get the mathematics oI perspective. 214 The algorithm reduces reality to solid objects, and the objects are Iurther reduced to planes deIined by straight lines. The coordinates oI the endpoint oI each line are stored in a computer. Also stored are the parameters oI a virtual camera -- the coordinates oI a point oI view, the direction oI sight, and the position oI a projection plane. Given this inIormation, the algorithm generates a perspectival image oI an object, point by point.
comprehensive account oI 3-D computer graphics techniques, see J. William Mitchell, The ReconIigured Eye: Visual Truth in the Post-Photographic Era (Cambridge, The MIT Press, 1992), 117-162. 212 Jasia Reichardt, The Computer in Art (London and New York: Studio Vista and Van Nostrand Reinhold Company, 1971), 15. 213 L.G. Roberts, Machine Perception oI Three-Dimensional Solids, MIT Lincoln Laboratory TR 315, 1963; L.G. Roberts, Homogeneous Matrix Representations and Manipulation oI N- Dimensional Constructs, MIT Lincoln Laboratory MS 1405, 1965. 214 "Retrospectives II: The Early Years in Computer Graphics at MIT, Lincoln Lab, and Harvard," in SIGGRAPH '89 Panel Proceedings (New York: The Association Ior Computing Machinery, 1989), 72.
The subsequent development oI computer graphics can be seen as the struggle to automate other operations involved in producing perspectival stills and moving images. The computerization oI perspectival construction made possible the automatic generation oI a perspectival image oI a geometric model as seen Irom an arbitrary point oI view -- a picture oI a virtual world recorded by a virtual camera. But, just like with the early perspectival machines described by DYrer, early computer graphics systems did not really save much time over traditional methods. To produce a Iilm oI a simulated landing, Boeing had to supplement computer technology with manual labor. As in traditional animation, twenty-Iour plots were required Ior each second oI Iilm. These plots were computer-generated and consisted oI simple lines. Each plot was then hand-colored by an artist. Finished plots were Iilmed, again manually, on an animation stand. 215 Gradually, throughout the 1970s and the 1980s, the coloring stage was automated as well. Many algorithms were developed to add the Iull set oI depth cues to a synthetic image -- hidden line and hidden surIace removal, shading, texture, atmospheric perspective, shadows, reIlections, and so on. 216
Today, these algorithms make possible the simulation oI almost any object in such a way that its computer image is indistinguishable Irom the photograph. Yet, this by itselI does not represent a radical break with older techniques oI visual nominalism. A painter can paint the
215 This mixture oI automated and pre-industrial labor is characteristic oI the early uses oI computers Ior the production oI images. In 1955 the psychologist Attneave was the Iirst to construct an image which was to become one oI the icons oI the age oI digital visuality -- random squares pattern. A pattern consisted oI a grid made Irom small squares colored black or white. A computer generated table oI random numbers has been used to determine the colors oI the square -- odd number Ior one color, even number Ior another. Using this procedure, two research assistants manually Iilled in 19,600 squares oI the pattern. Paul Vitz and Arnold B. Glimcher, Modern Art and Modern Science (New York: Praeger Publishers, 1984), 234. Later, many artists, such as Harold Cohen, used computers to generate line drawings which they then colored by hand, transIerred to canvas to serve as a Ioundation Ior painting, etc. 216 For Iurther discussion oI the problem oI realism in computer graphics, see Lev Manovich, "'Real' Wars: Esthetics and ProIessionalism in Computer Animation," Design Issues 6, no. 1 (Fall 1991): 18-25; Lev Manovich, "Assembling Reality: Myths oI Computer Graphics," AIterimage 20, no. 2 (September 1992): 12-14.
same image, although it would take much longer. What does represent a radical break is interactive computer graphics. In 1962 Ivan Sutherland designed his now legendary Sketchpad program. With Sketchpad, a human operator could create graphics directly on the computer screen by touching the screen with a light pen. In the same year, ITEK began marketing its Electronic DraIting Machine, an apparatus similar to Sketchpad. 217 Although both programs dealt only with 2-D graphics, they introduced a new paradigm oI interactive graphics: by changing something on the screen, the operator changed the data in the computer's memory. 218
When this paradigm oI interactive editing was combined with the algorithms oI 3-D graphics, a radically new way oI using perspectival images emerged. This development was more revolutionary than the automation oI perspective construction per se. Indeed, traditional draItsman could have accomplished what the computer at Boeing was doing -- generating plots in perspective given 3-D database -- only more slowly. But now it became possible to change the point oI view oI a virtual camera and see the corresponding changes in the perspectival image aIter a short delay. It also became possible to build and modiIy 3-D models interactively and observe the changes on the screen. The emergence oI interactive 3-D computer graphics started the race to eliminate the time delay between the action oI an operator and the displayed results. To be able to move through virtual world in real time a computer has to generate 30 Irames a second. The amount oI
217 "Retrospectives II: The Early Years in Computer Graphics at MIT, Lincoln Lab, and Harvard," 51. 218 In Iact, interactive computer graphics technology appeared earlier, although it was not publicized. Already in the 1950s the Air Force used interactive CRT displays and light pens in order to more eIIiciently process inIormation obtained by radar. Both CRT displays and light pens were designed at Lincoln Laboratory as part oI the SAGE project. Using this technology, Lincoln researchers created a number oI computer graphics programs. They included programs which made posible the display oI brain waves (1957), oI simulated planet and gravitational activity (1960), as well as the creation oI 2-D drawings (1958). "Retrospectives II: The Early Years in Computer Graphics at MIT, Lincoln Lab, and Harvard," 42-54.
calculations involved increases proportionally as the model becomes more complex and as more depth cues are added such as shading and shadows. In this race Ior speed, which accelerated in the 1970s as synthetic images began to be utilized in Ilight simulators, the algorithms oI 3-D graphics were gradually transported Irom soItware into hardware, each algorithm becoming a special computer chip. In 1966 Ivan Sutherland and his colleagues started research on the Iirst head-mounted display, a prototype oI Virtual Reality -- the ultimate interIace Ior 3-D computer graphics. The research was cosponsored by ARPA (Advanced Research Projects Agency) and the OIIice oI Naval Research. 219 Describing the project, Sutherland wrote in 1968:
The Iundamental idea behind the three-dimensional display is to present the user with a perspective image which changes as he moves. The retinal image oI the real objects which we see is, aIter all, only two-dimensional. Thus iI we place suitable two-dimensional images on the observer's retinas, we can create the illusion that he is seeing a three-dimensional object...The image presented by the three-dimensional display must change in exactly the way that the image oI a real object would change Ior similar motions oI the user's head. 220
The images seen in 1970 by the Iirst users oI Sutherland's system, such as a wireIrame cube or a square room, looked remarkably similar to those utilized by Renaissance artists in their demonstration oI perspective. Demonstrations oI perspective always Iavored such simple geometrical objects because they most clearly showed the achievement oI perspective -- the creation oI a 2-D image oI a 3-D object with Ioreshortening that matches our visual experience. 221 The diIIerence was that now the perspectival image oI a cube was not Iixed on a sheet oI paper; it appeared to an observer as Iloating in the air in Iront oI her, changing its view
219 Howard Rheingold, Virtual Reality (New York: Simon & Schuster, Inc., 1991), 105. 220 Qtd. in ibid., 104. 221 Later, Renaissance artists would deliberately insert checkerboard Iloors in perIect perspective in their paintings to impress their clients -- a move repeated in computer graphics, where such checkerboard background receding towards inIinity became a Iavorite motiI.
in correspondence with the slightest movement oI the observer's head. Perspectival imaging became interactive. In order to achieve this interactivity, special hardware was built. To speed up the calculations necessary to update the synthetic image 30 times a second, Sutherland's group broke down the process oI 3-D image synthesis into a series oI discrete steps. Each step was delegated to a special electronic circuit: a clipping divider, a matrix multiplier, a vector generator. Later on, such circuits became specialized computer chips, connected together to achieve real-time, high resolution, photorealistic 3-D graphics. Silicon Graphics, one oI the major manuIacturers oI computer graphics hardware, labeled such a system "geometry engine." The term appropriately symbolizes the second stage oI the automation oI perspectival imaging. At the Iirst stage, the photographic camera, with perspective physically built into its lens, mechanized the process oI creating perspectival images oI existing objects. Now, with the perspectival algorithm and other necessary geometric operations embedded in silicon, it becomes possible to display and interactively manipulate models oI non-existent objects as well. In the next section, I will discuss the Iinal stage in the automation oI perspectivalism -- the development oI computer vision.
In his papers, published between 1963 and 1965, Roberts Iormalized the mathematics necessary Ior generating and modiIying perspective views oI geometric models on the computer. This, writes William J. Mitchell, was "an event as momentous, in its way, as Brunelleschi's perspective 5. Computer Vision: Automation of Sight
demonstration." 222 However, Roberts developed techniques oI 3-D computer graphics having in mind not the automation oI perspectival imaging but another, much more daring goal -- "to have
222 Mitchell, The ReconIigured Eye , 118.
the machine recognize and understand photographs oI three dimensional objects." 223 Thus, the two Iields were born simultaneously: 3-D computer graphics and computer vision, automation oI imaging and oI sight. The Iield oI computer vision can be seen as the culmination oI at least centuries-long histories. The Iirst is the history oI mechanical devices designed to aid human perception, such as Renaissance perspectival machines. This history reaches its Iinal stage with computer vision,
223 "Retrospectives II: The Early Years in Computer Graphics at MIT, Lincoln Lab, and Harvard," 57.
which aims to replace human sight altogether. The second is the history oI automata, whose construction was especially popular in the seventeenth and eighteenth centuries. Yet, despite similarity in appearance, there is a Iundamental diIIerence between Enlightenment automata which imitated human's or animal's bodily Iunctions and the modern day robots equipped with computer vision systems, artiIicial legs, arms, etc. As noted by Leonardo Torres, old automata, while successIully copying the appearance and movement oI living beings, had no economic value. Indeed, such voice synthesis machines as Wolgang von Kempelen's 1778 device which directly imitated the Iunctioning oI the oral cavity or AbbZ Mical's T tes Parlantes (1783) operated by a technician hiding oIIstage and pressing a key on a keyboard were used only Ior entertainment. 224 When in 1913 Torres called Ior automata that would "accomplish the results which a living person obtains, thus replacing a man by a machine" he was expressing a Iundamentally new idea oI using automata Ior productive labor. A Iew years later, the brother oI the Czech writer Karel Capek coined the word robot Irom the Czech word robota, which means "Iorced labor." 225 Capek's play R.U.R. (1921) and Fritz Lang's Metropolis (1927) clearly demonstrate this new association oI automata with physical industrial labor. ThereIore, it would be erroneous to conclude that, with computer vision, twentieth century technology simply added the sense oI sight to eighteenth century mechanical statues. But even to see computer vision as the continuation oI Torres', Capek's or Lang's ideas about industrial automation which replaces manual labor would not be Iully accurate. The idea oI
224 Remko Scha, "Virtual Voices," MediaMatic 7, no. 1 (1992): 33. Scha describes two Iundamental approaches taken by the developers oI voice imitating machines: the genetic method which imitates the physiological processes that generate speech sounds in the human body and the gennematic method which is based on the analysis and reconstruction oI speech sounds themselves without considering the way in which the human body produces them. While the Iield oI computer vision, and other Iields oI artiIicial intelligence, Iirst clearly Iollowed gennematic method, in the 1980s, with the growing popularity oI neural networks, there was a shiIt towards the genetic method -- direct imitation oI the physiology oI the visual system. In a number oI laboratories, scientists begin to build artiIicial eyes which move, Iocus, and analyze inIormation exactly like human eyes. 225 Eames and Eames, A Computer Perspective, 100.
computer vision became possible and the economic means to realize this idea became available only when automation entered its second stage aIter World War II. The attention turned Irom the automation oI the body to the automation oI the mind, Irom physical to mental labor. This shiIt to the automation oI mental Iunctions such as vision, hearing, reasoning, problem solving is exempliIied by the very names oI the two new Iields that emerged during the 1950s and 1960s -- artiIicial intelligence and cognitive psychology. The latter gradually replacing behaviorism, the dominant psychology oI the "Fordism" era. The emergence oI the Iield oI computer vision is a part oI this cognitive revolution, a revolution which was Iinanced by the military escalation oI the Cold War. 226 This connection is solidiIied in the very term "artiIicial intelligence" which may reIer simultaneously to two meanings oI "intelligence": reason, the ability to learn or understand, and inIormation concerning an enemy or a possible enemy or an area. ArtiIicial intelligence: artiIicial reason to analyze collected inIormation, collected intelligence. In the 1950s, Iaced with the enormous task oI gathering and analyzing written, photographic, and radar inIormation about the enemy, the CIA and the NSA (National Security Agency) began to Iund the Iirst artiIicial intelligence projects. One oI the earliest projects was a Program Ior Mechanical Translation, initiated in the early 1950s in the attempt to automate the monitoring oI Soviet communications and media. 227 The work on mechanical translation was probably the major cause oI many subsequent developments in modern linguistics, its move towards Iormalization; it can be discerned in Noam Chomsky's early theory which, by postulating the existence oI language universals in the domain oI grammar, implied that translation between arbitrary human languages could be automated. The same work on mechanical translation was also one oI the catalysts in the development oI the Iield oI pattern recognition, the precursor to computer vision. Pattern recognition is concerned with
226 De Landa, "Policing the Spectrum," 194-203. 227 Ibid., 214.
automatically detecting and identiIying predetermined patterns in the Ilow oI inIormation. A typical example is character recognition, the Iirst stage in the process oI automating translation. Pattern recognition was also used in the U.S. Ior the monitoring oI Soviet radio and telephone communication. Instead oI listening to every transmission, an operator would be alerted iI computer picked up certain words in the conversation. As a "logistics oI perception" came to dominate modern warIare and surveillance and as the space race began, image processing became another major new Iield oI research. Image processing comprises techniques to improve images Ior human or computer interpretation. 228 In 1964, the space program Ior the Iirst time used image processing to correct distortions in the
228 The Iirst paper on image processing was published in 1955. L.S.G. Kovasznay, and H.M. Joseph, "Image Processing," Proceedings oI IRE 43 (1955): 560-570.
pictures oI the Moon introduced by a on-board television camera oI Ranger 7. 229 In 1961, the National Photographic Interpretation Center (NPIC) was created to produce photoanalysis Ior the rest oI the U.S. intelligence community and, as Manual De Landa points out, by the end oI the next decade computers "were routinely used to correct Ior distortions made by satellite's imaging sensors and by atmospheric eIIects, sharpen out-oI-Iocus images, bring multicolored single images out oI several pictures taken in diIIerent spectral bands, extract particular Ieatures while diminishing or eliminating their backgrounds altogether..." De Landa also notes that computer analysis oI photographic imagery also became the only way to deal with the pure volume oI intelligence being gathered: "It became apparent during the 1970s that there is no hope oI keeping up with the millions oI images that poured into NPIC...by simply looking at them the way they had been looked at in World War II. The computers thereIore also had to be taught to compare new imagery oI a given scene with old imagery, ignoring what had not changed and calling the interpreter's attention to what had." 230
229 RaIael C. Gonzalez, and Paul Wintz, Digital Image Processing (Reading, Mass.: Addison-Wesley Publishing Company, 1977), 2. 230 Qtd. in De Landa, "Policing the Spectrum," 200.
The techniques oI image processing, which can automatically increase an image's contrast, remove the eIIects oI blur, extract edges, record diIIerences between two images, and so on, greatly eased the job oI human photoanalysts. And the combining oI image processing with pattern recognition made it possible in some cases to delegate the analysis oI photographs to a computer. For instance, the technique oI pattern matching used to recognize printed characters can also be used to recognize objects in a satellite photograph. In both cases, the image is treated as consisting oI two-dimensional Iorms. The contours oI the Iorms are extracted Irom the image are then compared to templates stored in computer memory. II a contour Iound in the image matches a particular template, the computer signals that a corresponding object is present in a photograph. A general computer vision program has to be able to recognize not just two-dimensional but three-dimensional objects in a scene taken Irom an arbitrary angle. 231 Only then it can be used to recognize an enemy's tank, to guide an automatic missile towards its target or to control a robotic arm on the Iactory Iloor. The problem with using simple template matching is that "a two-dimensional representation oI a two-dimensional object is substantially like the object, but a two-dimensional representation oI a three-dimensional object introduces a perspective projection that makes the representation ambiguous with respect to the object." 232 While pattern recognition was working Ior images oI two-dimensional objects, such as letters or chromosomes, a diIIerent approach was required to "see" in 3-D. Roberts' 1965 paper "Machine Recognition oI Three-dimensional Solids" is considered to be the Iirst attempt at solving the general task oI automatically recognizing three-dimensional
231 Within the Iield oI computer vision, a scene is deIined as a collection oI three-dimensional objects depicted in an input picture. David McArthur, "Computer Vision and Perceptual Psychology," Psychological Bulletin 92, no. 2 (1982): 284. 232 Paul R. Cohen and Edward A. Feigenbaum, eds., The Handbook oI ArtiIicial Intelligence (Los Atlos, CA: William KauImann, Inc., 1982), 3: 139.
objects. 233 His program was designed to understand the artiIicial world composed solely oI polyhedral blocks -- a reduction oI reality to geometry that would have pleased CZzanne. Using image processing techniques, a photograph oI a scene was Iirst converted into a line drawing. Next, the techniques oI 3-D computer graphics were used:
Roberts' program had access to three-dimensional models oI objects: a cube, a rectangular solid, a wedge, and a hexagonal prism. They were represented by coordinates (x, y, z) oI their vertices. A
233 L.G. Roberts, "Machine perception oI three-dimensional solids," Optical and Electo Optical InIormation Processing, ed. J.T. Tippett (Cambridge: The MIT Press, 1965) .
program recognized these objects in a line drawing oI the scene. A candidate model was selected on the basis oI simple Ieatures such as a number oI vertices. Then the selected model was rotated, scaled, projected, and matched with the input line drawing. II the match was good, the object was recognized, as were its position and size. Roberts' program could handle even a composite object made oI multiple primitive shapes; it subtracted parts oI a line drawing Irom the drawing as they were recognized, and the remaining portions were analyzed Iurther. 234
Was this enough to completely automate human vision? This depends upon how we deIine vision. The chapter on computer vision in The Handbook oI ArtiIicial Intelligence (1982) opens with the Iollowing deIinition: "Vision is the inIormation-processing task oI understanding a scene Irom its projected images." 235 But what does "understanding a scene" mean? With computer vision research Iinanced by the military-industrial complex, the deIinition oI understanding becomes highly pragmatic. In the best tradition oI the pragmatism oI James and Pierce, cognition is equated with action. The computer can be said to "understand" a scene iI it can act on it -- move objects, assemble details, destroy targets. Thus, in the Iield oI computer vision "understanding a scene" implies two goals. First, it means the identiIication oI various objects represented in an image. Second, it means reconstruction oI three-dimensional space Irom the image. A robot, Ior instance, need not only recognize particular objects, but it has to construct a representation oI the surrounding environment to plan its movements. Similarly, a missile not only has to identiIy a target but also to determine the position oI this target in three- dimensional space. It can be seen that Roberts' program simultaneously IulIilled both goals. His program exempliIied the approach taken by most computer vision researchers in the Iollowing two decades. A typical program Iirst reconstructs the three-dimensional scene Irom the input image and then matches the reconstructed objects to the models stored in memory. II the match is good, the program can be said to recognize the object, while simultaneously recording its position (Iig. 14).
234 Cohen and Feigenbaum, The Handbook oI ArtiIicial Intelligence, 3: 129. 235 Ibid., 127.
A computer vision program thus acts like a blind person who "sees" objects (i.e., identiIies them) by reading their shapes through touch. As Ior a blind person, understanding the world and the recognition oI shapes are locked together; they cannot be accomplished independently oI one another. With computer vision, visual nominalism, which, aIter all, is just a particular use oI vision, attains unprecedented importance. Now, to see means to see shapes and distances. Visual nominalism and vision itselI have become equated.
I have presented the new twentieth century technologies oI visual nominalism as building upon perspective, extending its powers in space and time. Technologies oI remote sensing, such as radar, extended perspective beyond the realm oI the visible. 3-D computer graphics speeded up and automated design and perspectival display oI the models oI both real and imagined objects. With the emergence oI the Iield oI computer vision, perspectival sight reaches its apotheosis and at the same time begins its retreat. At Iirst computer vision researchers believed that they could invert the perspective and reconstruct the represented scene Irom a single perspectival image. Eventually, they realized that it is oIten easier to bypass perspectival images altogether and use other means as a source oI three-dimensional inIormation. Latour points out that with the invention oI perspective it became possible to represent absent things and plan our movement through space by working on representations. To quote him again, "one cannot smell or hear or touch Sakhalin island, but you can look at the map and determine at which bearing you will see the land when you send the next Ileet." 236 Roberts' program extended these abilities even Iurther. Now the computer could acquire Iull knowledge oI the three-dimensional world Irom a single perspectival image! And because the program determined the exact position and orientation oI objects in a scene, it became possible to see the reconstructed scene Irom another viewpoint. It also became possible to predict how the scene
236 Latour, "Visualisation and Cognition," 8.
would look Irom an arbitrary viewpoint. 237 Finally, it also became possible to guide automatically the movement oI a robot through the scene. Roberts' program worked using only a single photograph -- but only because it was presented with a highly artiIicial scene and because it "knew" what it could expect to see. Roberts limited the world which his program could recognize to simple polyhedral blocks. The shapes oI possible blocks were stored in a computer. Others simpliIied the task even Iurther by painting all objects in a scene the same color. However, given an arbitrary scene, composed Irom arbitrary surIaces oI arbitrary color and lighted in an arbitrary way, it is very diIIicult to reconstruct the scene correctly Irom a single perspectival image. The image is "underdetermined." First, numerous spatial layouts can give rise to the same two-dimensional image. Second, "the appearance oI an object is inIluenced by its surIace material, the atmospheric conditions, the angle oI the light source, the ambient light, the camera angle and characteristics, and so on," and all oI these diIIerent Iactors are collapsed together in the image. 238 Third, perspective, as any other type oI projection, does not preserve many geometric properties oI a scene. Parallel lines turn into convergent lines; all angles change; equal lines appear unequal. All this makes it very diIIicult Ior a computer to determine which lines belong to a single object. Thus, perspective, which until now served as a Ioundation oI new technologies oI visual nominalism, becomes the drawback which needs to be overcome. Perspective, this Iirst step towards the rationalization oI sight (Ivins) has eventually become a limit to its total rationalization -- the development oI computer vision. The realization oI the ambiguities inherent in a perspectival image itselI came aIter years oI vision research. In trying to compensate Ior these ambiguities, laboratories began to scrutinize
237 Cohen and and Feigenbaum, The Handbook oI ArtiIicial Intelligence, 3: 141. 238 Ibid., 128.
the Iormal structure oI a perspectival image with a degree oI attention unprecedented in the history oI perspective. In 1968 Adolpho Guzman classiIied the types oI junctions that appear in line representations aIter he realized that a junction type can be used to deduce whether regions oI either side oI a junction line were part oI the same object. 239 In 1986 David Lowe presented a method to calculate the probability that a particular regularity in an image (Ior instance, parallel lines) reIlects the physical layout oI the scene rather than being an accident due to a particular viewpoint. 240 All other sources oI depth inIormation such as shading, shadows or texture gradients were also systematically studied and described mathematically. Despite these advances, a single perspectival image remained too ambiguous a source oI inIormation Ior practical computer vision systems. An alternative has been to use more than one image at a time. Computer stereo systems employ two cameras which, like human eyes, are positioned a distance apart. II the common Ieature can be identiIied in both images, then the position oI an object can be simply determined through geometric calculations. Other systems use a series oI continuous images recorded by a video camera. But why struggle with the ambiguity oI perspectival images at all? Instead oI inIerring three-dimensional structure Irom a two-dimensional representation, it is possible to measure depth directly by employing various remote sensing technologies. In addition to video cameras, modern vision systems also utilize a whole range oI diIIerent range Iinders such as lasers or ultrasound. 241 Range Iinders are devices which can directly produce a three-dimensional map oI an object. The same basic principle employed in radar is used: the time required Ior an electro- magnetic wave to reach an object and reIlect back is proportional to the distance to the object. But while radar reduces an object to a single point and in Iact is blind to close-by objects, a range
239 Ibid., 131. 240 David Lowe, Three-Dimensional Object Recognition Irom Single Two-Dimensional Images, Robotics Report 62 (New York: Courant Institute oI Mathematical Sciences, New York University, 1986). 241 Cohen and Feigenbaum, The Handbook oI ArtiIicial Intelligence, 3: 254-259.
Iinder operates at small distances. By systematically scanning the surIace oI an object, it directly produces a "depth map," a record oI an object's shape which can be then matched to geometric models stored in computer memory thus bypassing the perspectival image altogether (Iig. 15).
What began as separate developments -- to automate the recording oI three-dimensional inIormation, the generation oI perspectival views, and the identiIication oI objects -- eventually converged. Remote sensing, 3-D computer graphics, and computer vision Iorm a closed circle. As we have seen, algorithms oI 3-D computer graphics were Iirst developed by Roberts in order to solve the general problem oI computer vision. From that point on, the two Iields developed in parallel. The Iield oI computer graphics was gradually learning to simulate more and more "realistic" images oI reality, adding shading, shadows and texture to initial wireIrame drawings. Meanwhile, the Iield oI computer vision was learning to deal with shading, shadows, and texture as sources oI depth inIormation. In eIIect, the goal oI computer vision is to undo what 3-D computer graphics aims to achieve. The Iormer tries to reconstruct a scene Irom its photographic image, to deduce viewer independent inIormation about the objects: their shapes and dimensions. The goal is to "produce an object centered representation that is independent oI the particular details oI the viewing angle or photographic process." 242 The aim oI 3-D computer graphics is exactly the opposite: given the objective inIormation about the scene (shapes and positions oI objects, their materials, direction oI light) to produce its image which is virtually indistinguishable Irom a photograph. At the Ioundation oI this mutual dependence is the shared geometrical model oI reality as a set oI surIaces having a particular orientation, position, and curvature. The circle is completed by the technologies oI remote sensing that, by producing depth maps oI real scenes, can
242 David Peat, ArtiIicial Intelligence: How Machines Think (New York: Baen Enterprises, 1985), 230.
automatically reduce them to geometric models. Portable shape acquisition cameras are already being developed that, instead oI taking snapshots oI a scene, will automatically generate geometric models which then can be used Ior computer vision or 3-D computer graphics. 243
Technologies oI remote sensing can automatically create a depth map oI reality. Using such depth map as input, 3-D computer graphics can be used to generate an image which, in its turn, can act as input to a computer vision system. Computer vision then can reconstruct "objective" inIormation Irom an image, a model which can act as a source Ior Iurther 2-D computer graphics synthesis. A series oI mirrors, constructions and reconstructions are triggered by each other, with speed only limited by the processing power oI "geometry engines."
In this chapter I discussed the twentieth century automation oI the Iunction oI vision which I have called visual nominalism -- capturing the inIormation about shapes and distances and representing this inIormation in two-dimensional displays. Radar and other technologies oI remote sensing extend the capacities oI human vision beyond the limits oI visibility, making it possible to track positions oI objects in real time. Techniques oI 3-D computer graphics allow the interactive display and construction oI geometric models oI both real and non-existent objects. Finally, computer vision automates the recognition oI objects and the reconstruction oI three- dimensional scenes Irom their images. In this discussion, the concept oI perspective played a special role Ior a number oI reasons. I relied on the Iamiliarity oI this concept to explain the principles oI operations oI newer twentieth century technologies oI visual nominalism. I have also relied on Ivins' account oI perspective and its extension by Latour to suggest that the Renaissance's adaptation oI
perspective represented the Iirst step in the automation oI visual nominalism. While other cultures used sophisticated methods oI space mapping, the importance oI perspective lies not in its representational superiority but in its algorithmic character. This algorithmic character enabled the gradual development oI visual languages oI perspective and descriptive geometry and, in parallel, oI perspectival machines and technologies, Irom a simple net described by DYrer to photography and radar. And when digital computers made possible mass automation in general, automation oI perspectival vision and imaging Iollowed soon. It is this automation oI vision and imaging which gave the West superiority in the "logistics oI perception," and contributed to its victory in the GulI War. A two-dimensional representation containing inIormation about geometry and topography allows one to control reality across space and time. The advantage oI perspective is that it makes it possible to present this inIormation in an easily readable Iorm within a single image. In the West, until this century, perspective Iormed the Ioundation oI techniques and technologies oI visual nominalism. New twentieth century technologies oI visual nominalism extend perspective, utilizing to the extreme its inherent qualities such as the algorithmic character and the reciprocal relationship it establishes between reality and representation. For instance, the perspective algorithm, a Ioundation oI both computer graphics and computer vision, is used to generate perspectival views given a geometric model and to deduce the model given a perspectival view. In another example, technologies oI remote sensing, such as radar, extend perspectival vision beyond the visible, relying on the Iact that "what is an issue in geometric perspective is simply the mapping oI space, not sight" (Lacan). Yet, while giving rise to new technologies oI "geometric vision," perspective also becomes a limit to the Iinal automation oI sight -- recognition oI objects by a computer. Finally, it is displaced Irom its privileged role, becoming just one among other techniques oI space mapping and visualization.
In order to automate sight, to replace human vision with machine vision, a new understanding oI vision became necessary. David Marr's Vision (1980) has been the most inIluential account oI this paradigm, shared by computer scientists and psychologists. The book opens with this statement:
What does it mean, to see? The plain man's answer (and Aristotle's, too) would be, to know what is where by looking. In other words, vision is the process oI discovering Irom images what is present in the world, and where it is. 244
There is nothing "plain" about this deIinition oI vision. Marr projects the goals oI computer vision onto human vision: the identiIication oI objects represented in an image and the reconstruction oI their positions in three-dimensional space. The larger part oI the book is devoted to the discussion oI algorithms by which the human nervous system may accomplish this reconstruction. In this way, nature was redeIined as the most useIul instrument in its own replacement. What David Marr, and many others have assumed, is that the key to successIul computer vision systems lies in emulating the algorithms hidden in human nervous system. II only these algorithms, created by nature during millions oI years oI evolution, could be understood, the way to replace human vision by computer vision will be open. The automation oI vision is a part oI the overall process oI industrial automation aIter World War II. Automation aIIects the modern understanding oI vision in two crucial ways. On the one hand, automation entails the replacement oI human cognitive Iunctions by a computer, such as the substitution oI vision by computer vision. Here, visual nominalism becomes the dominant paradigm Ior the study oI vision. On the other hand, the automation also involves the integration oI human and machine in new man-machine systems, such as radar installations or pilot cockpits. The need to describe the perIormance oI human and machine components in the
244 Marr, Vision, 3.
same terms leads to the emergence oI another paradigm -- vision as inIormation processing -- which will be discussed in the Iollowing chapter.
Writing in 1927, Lszl Moholy-Nagy makes a comparison between engineering and design:
A modern engineer, iI his goal and the Iunctional purpose oI his work are clear, can without any great eIIort make a product that is Iormally adequate and perIect in its economic construction. But the photographic advertisements oI our time are not so easy to deIine. They don't come with "user's instructions." Research into the physiological and psychological laws oI visual eIIectiveness is still Iar behind the times, compared to the study oI the physical laws. 245
In the opening chapter, where this quotation already appeared, I have traced one line oI research into "the physiological and psychological laws oI visual eIIectiveness," suggesting that investigations oI the psychological eIIects oI basic colors and elemental Iorms, conducted in experimental psychology since the second halI oI the nineteenth century made possible the idea oI a rational visual language composed Irom simple elements. In theory, this model called Ior the enumeration oI visual elements and the description oI their eIIects in order to compose dictionaries oI "visual eIIectiveness." In practice, it translated into visual works which at Iirst only contained and then gradually became Iully composed oI simple abstract Iorms. Such theory and practice were already pursued in the nineteenth and early twentieth centuries by a number oI artists (Seurat, Signac, Kandinsky, Klee, Mondrian) and critics interested in "scientiIic aesthetics" (Blanc, Henry). However, in the 1920s, when artists Iound themselves in the roles oI the designers oI mass communication, the idea oI a scientiIically grounded visual language acquired new urgency and importance. It was no longer a question oI scientiIic aesthetics, oI a work oI art producing an aesthetic eIIect in an individual spectator. Now it became a question oI rationalizing mass communication -- a question oI economic and political importance. The design oI mass communication was too important to be leIt to an artist's intuition. ThereIore,
245 Lszl Moholy-Nagy, "Photography in Advertising," in Photography in the Modern Era, ed. Christopher Phillips (New York: Aperture, 1989), 87. Chapter 4: The Engineering of Vision from INKhUK to MIT 1. Not Artists but Engineers
Moholy-Nagy wrote about the need to precisely "engineer" photographic advertisements, and the Bauhaus welcomed experimental psychologists. The trend reached its extreme in Soviet Russia, where artists thought that they could control not just the consumer habits oI segments oI society (as in the West), but the consciousness oI the whole country --"aIIecting mass psyche, organizing the will oI the class" (Tretyakov). Here the discourse oI engineering visual communication became most systematic, translating into the establishment oI a number oI psychological laboratories at various art institutions, such as VKhUTEMAS, GAKhN, and the State Institute oI Artistic Culture. Their goal was to create a Iully rationalized visual language oI mass communication where each visual element is capable oI communicating a meaning, producing an emotion or causing a behavioral response. In the words oI a paper entitled "Engineerism" presented at INKhUK: "visual sensations as such will concertedly shape the human being as an organized unit." 246
This is one development, one trajectory -- Irom Blanc's theories and Fechner's psychophysical experiments to INKhUK and VKhUTEMAS -- where the dream oI controlling the masses through perception was systematically pursued. This trajectory leads Irom Wundt's psychological laboratory to the "laboratory" explorations oI constructivists and rationalists; and Irom the laboratory to the sites oI mass communication -- a movie hall, a printed page, an architectural space, a stadium, a city street with posters, advertisements, and placards. In this concluding chapter I will Iollow the next part oI this trajectory, the next stage in the rationalization oI visual communication: Irom the avant-garde's Iantasies oI psychophysical control oI the masses to inIormation theory, engineering psychology, and human-machine interIaces. The visual language advocated by Rodchenko or Lissitsky -- elemental abstract Iorms, each having a predictable eIIect -- seemed the most eIIective solution to the problem oI
246 Nicholas H. Allison, ed., Art Into LiIe: Russian Constructivism 1914-1932 (New York: Rizzoli International Publications, 1990), 81.
rationalizing vision. But aIter World War II, a new kind oI rationalization became necessary. A new paradigm was now needed to encompass not only human communication but also communication between a human and a machine, the two linked in a human-machine system. It is the emergence oI this paradigm which is the subject oI this chapter. The new paradigm, called inIormation theory, emerged in the 1920s in response to the growth oI modern telecommunications. The theory was originally developed by engineers in order to measure the eIIiciency oI communication systems, such as telephone, radio, and television. However, in the late 1940s, the theory leIt its engineering context and was embraced by aestheticians, linguists, social scientists, and semioticians. Most importantly, it was taken up by psychologists Iaced with the practical task oI designing human-machine systems which increasingly came to dominate both the battleIield and the workplace aIter World War II (radar screen, aircraIt controls, computer terminals oI the automated Iactory). As human vision became the main channel oI communication between human and machine, it also came to be understood in terms oI inIormation theory which now provided a new quantitative model oI a human observer. Thus the dreams oI the avant-garde Ior an engineering approach to visual communication were IulIilled.
The avant-garde was concern with rationalizing vision, with creating visual communication according to scientiIic principles. In this avant-garde artists likened themselves to modern engineers, as can be seen Irom Moholy-Nagy's statement. But what is a modern engineer? The engineering proIession emerged in the 1870s as the liaison between science and industry. The job oI an engineer was to put any industrial or corporate process on a scientiIic basis in order to achieve the maximum output with a minimum investment oI time, materials, and energy. In other words, the job oI an engineer was to make a process eIIicient, regardless oI what industry s/he worked in. A 1933 book on engineering deIines it in this way: "Engineering is
the science and art oI eIIicient dealing with materials and Iorces...it involves the most economical design and execution oI a vast number oI important undertakings, assuring, when properly perIormed, the most advantageous combination oI accuracy, saIety, durability, speed, simplicity, eIIiciency and economy possible Ior the design and service." 247 The avant-garde artists aimed to apply these principles oI engineering to the Iield oI mass visual communication. In 1921, during a heated discussion at INKhUK (Institute oI Artistic Culture) over how to turn their Iellow artists into "constructivists," who would utilize modern engineering methods Ior construction -- be it the construction oI consumer goods or mass psyche, industrial design or mass propaganda -- participants expressed doubts about whether traditional artists could accomplish this transition Irom "art into liIe." Boris Arvatov suggested that accomplished artists should become political activists, and the young ones would have to go to the polytechnic. 248
More categorical was the opinion oI Sternberg, who, aIter asking the same question, "What are artists to do?," answered it in these words, "They're good Ior nothing, they should be dealt with the way the Cheka deals with counterrevolutionaries." 249 A Iew minutes later in the discussion he suggested a more humane solution, the solution which was repeatedly put into practice in regard to Soviet non-conIormist artists decades later -- their conIinement to a mental institution. "All the artists who conduct youth groups, and who are going to continue their work and perhaps teaching the way they have in the past, we should really stuII a hospital with these artists." 250
Arvatov's request that artists should get their training in an engineering school, and Sternberg's warning that those who reIuse to comply would be shot, illustrate the grave seriousness with which many Soviet avant-garde artists oI the 1920s took the ideal oI engineering as a model Ior their own practice. Consider the term constructivism itselI. In
247 Qtd. in Theodore Hoover and Charles Fish, The Engineering ProIession (StanIord: StanIord University Press, 1941), 416. Emphasis mine -- L.M. 248 Allison, Art Into LiIe, 74. 249 Ibid. 250 Ibid., 77.
Russian, konstruktor (the one who constructs) is a synonym Ior engineer. This allowed a number oI artists to call themselves konstruktors and to reIer to their movement as constructivism. The term conveniently emphasized that, like engineers, they would construct utilitarian objects or propagandistic works on the basis oI scientiIic, rational principles, in contrast to traditional artists who were said to rely on inspiration. Translated into English, the word constructivism loses its close association with engineering, leading to a somewhat diIIerent interpretation oI the entire movement. In the early 1920s, constructivist artists competed to be the best at embodying the ideal oI an engineer. Vertov announces his desire to become "kinok-engineer, controlling cameras over distance." 251 Eisenstein, who, like Sternberg, graduated Irom an engineering school beIore the 1917 Revolution, compares the theatrical apparatus (actors' perIormances, sound eIIects, lighting) to the machinery (orudie obrabotki) which a director should use to Iorm the viewer -- as a milling machine Iorms machinery parts. 252 Later, he likens the eIIect oI a Iilm to the work oI an engine -- a signiIicant analogy since engineering was historically identiIied with the design oI engines. 253 Tatlin photographs himselI next to his already legendary design Ior the monument to the Third International. The monument, with its two intertwining lattice spirals 400 meters high, looks like an EiIIel Tower put at an angle, eternally moving into the sky, a symbol appropriate to signiIy the idea oI permanent revolution, the (Iuture) industrial might oI the new state and, at the same time, the engineering skills oI its designer. Finally, Lissitsky and his students design another triumph to engineerism -- an inclined lattice girder entitled "Lenin Tower."
251 Dziga Vertov, "Kinoki. Perevorot" (Kinoki. A revolution), LEF 3 (1923): 143. 252 Sergei Eisenstein, "Montazh atraktsionov (Montage oI attractions)," LEF 3 (1923): 71. 253 A book on the engineering proIession published in 1941 quotes these dictionary deIinitions oI engineering: "engineering: originally, the art oI making engines; in its modern and extended sense, the art and science by which mechanical properties oI matter are made useIul to man in structures and machines;" "engineering: the art and science oI making, building, or using engines and machines, or oI designing and constructing public works or the like requiring special knowledge oI materials, machinery, and the laws oI mechanics." Hoover and Fish, The Engineering ProIession, 416.
Here is another device: similar to traditional portraits where the identity oI a person was established by depicting the instruments oI his or her craIt, constructivists' portraits and selI- portraits prominently Ieature engineering instruments. Lissitsky's selI portrait: on the background oI graphed paper -- his Iace and through it -- a hand holding a compass, its leg touching a perIectly drawn square. The title oI this photomontage is Konstruktor. Even more dramatic is another Lissitsky's montage, a portrait oI Tatlin (Iig. 16). Tatlin is standing on a chair, holding a long pole -- a ruler. By his Ieet are mathematical symbols (an integral, a square root), a curve crossing axis lines. Behind him is a blackboard with more symbols. And, a monstrosity: a huge compass, widely open, growing straight Irom Tatlin's eye. A man-compass -- measuring, calculating, constructing. Constructing what? "My reaction to the participation oI the artist in production is positive," declared architect Vesnin during one INKhUK discussion, "but an artist must look aIter his own aIIairs, and his aIIair is the |psycho-physiological| eIIect oI Iorm on |human| consciousness." In other words, Vesnin is saying: let us leave the engineering oI machines and mass-produced objects to real engineers. Our business, the business oI designers is the engineering oI perception, the engineering oI visual communication -- the creation oI architecture, mass spectacles, posters, designs, photomontages, which would produce calculated bodily responses, emotions, and meanings in the viewers. "The modern engineer has created brilliant objects: the bridge, the steam engine, the airplane, the crane,...the modern artist must create objects that are equal to them in power, intensity, and potential in the context oI their psycho-physiological impact as an organizing element in man's consciousness." 254 The job oI an artist is to engineer vision, and to engineer vision means to aIIect the viewer with engineering precision, predictability, and eIIectiveness.
254 Qtd. in Anatole Senkevitch, Jr., "The Sources and Ideals oI Constructivism in Soviet Architecture," in Art Into LiIe: Russian Constructivism 1914-1932, ed. Nicholas H. Allison (New York: Rizzoli International Publications, 1990), 173-174.
In this statement Irom the early 1920s, Vesnin envisions an engineer as the designer oI physical objects -- the bridge, the steam engine, the airplane, the crane. But already in the same decade a new kind oI engineer appears, whose job is to design not objects but communication systems -- telephone, radio, television. A new paradigm appears as well -- inIormation theory which becomes the scientiIic basis oI communication engineering. In the next section, I will Iollow its emergence.
In 1922 Moholy-Nagy created his Iamous "Telephone Pictures." Here is his account:
In 1922 I ordered by telephone Irom a sign painter Iive paintings in porcelain enamel. I had the Iactory's color chart beIore me and I sketched my paintings on graph paper. At the other end oI the telephone the Iactory supervisor had the same kind oI paper, divided into squares. He took the dictated shapes in correct position. (It was like playing chess by correspondence.) 255
There is no conclusive prooI that this event actually took place. At any rate, the story helped Moholy-Nagy get a teaching job at the Bauhaus the next year, where he could Iully pursue his vision oI using the latest communication systems as artistic tools. His colleague wrote oI the Bauhaus atmosphere: "There is incessant talk oI cinema, optics, mechanics, projection and continuous motion...Is this the atmosphere in which painters like Klee and some others oI us can go on developing? Klee was quite depressed yesterday when talking about Moholy." 256
The story oI "Telephone Pictures" became one oI the myths oI modern art, or rather, oI the modern struggle against the Romantic notion oI art: to reIuse subjectivity, the unique touch
255 Lszl Moholy-Nagy, The New Vision (New York: Wittenborn, 1947), 79. The three existing accounts oI this episode are discussed in Eduardo Kac, "Aspects oI the Aesthetics oI Telecommunications," in SIGGRAPH '92 Visual Proceedings, ed. John Grimes and Gray Lorig (New York: The Association Ior Computing Machinery, 1992), 52-53. 256 Qtd. in Otto Stelzer, "Moholy-Nagy and His Vision," in Lszl Moholy-Nagy, Painting, Photography, Film (1925; reprint, Cambridge: The MIT Press, 1973), 146. 2. Information Theory: an Engineer AnaIyzes Communication
oI the artist's hand and the artist's eye (exempliIied in Bauhaus by Klee) and to embrace the mass-produced, the industrial, the mechanical, electrical, electronic, and digital -- Irom Duchamp's ready-made to Op Art, Kinetic Art, copy art, telecommunication art, Iax art, computer art -- in short, the tradition oI industrial art, throughout the twentieth century Ieeding on the latest technical developments in communication and imaging. And Moholy-Nagy intended the story this way, as a myth, a beginning more important than Duchamp's ready- mades. Indeed, not just art as chess, as a logical game (as with Duchamp), but -- art as "playing chess by correspondence," a communication over distance between a sender and a receiver who do not occupy the same physical space. Let us analyze Moholy-Nagy's story Irom the point oI view oI telecommunication -- the communication over distance. It mentions three diIIerent communication systems. First, the most ancient: communication with the help oI signs, such as traIIic or street signs (Moholy-Nagy orders his pictures at the sign Iactory). Second, more modern: the postal service. Third, still a recent development Ior the 1920s: the telephone. As communication systems developed, they made possible the transmission oI more inIormation over longer distances in a shorter period oI time. The eIIectiveness oI road signs was limited by the distance at which they were visible, and the amount oI inIormation they carried was minimal. With the postal service, this distance limit was overcome, but a signiIicant time delay remained. This delay was Iinally eliminated by the telephone. In the story, Moholy-Nagy communicated his design to the Iactory supervisor over the telephone, dictating coordinates and colors oI squares. He was able to do this only by Iitting his design to a pre-existing Iactory grid, treating the grid's squares as minimal units oI his composition. (Moholy-Nagy, in other words, Iully embraced digital visuality and computer art long beIore digital computers.) But could he, in 1922, transmit this design as a picture -- and in
real time? And what iI the picture was an arbitrary image, Ior instance a photograph oI a real scene, rather than a by-product oI the communication code? The rapid transmission oI photographic pictures was just becoming possible, utilizing the same method employed by Moholy-Nagy -- digital representation. In 1921, digitized newspaper photographs began to be regularly sent between London and New York over a submarine cable in less than three hours. On one end, a picture was segmented using a grid, and the tone oI each cell was coded; on another end, the received code would drive a special printer Iitted with typeIaces simulating diIIerent tones. 257
A Iew years later, Moholy-Nagy included the examples oI this "wireless telegraphed photography" in his Painting, Photography, Film, proudly putting them at the very end oI the photographic section oI the book to represent the latest achievement in mechanized tele-seeing (Iig. 17). Yet, this system was not Iast enough Ior the age oI "the Iilm; the electric sign, simultaneity oI sensorily perceptible events." 258 The speed oI wireless telegraphed photography was limited by the channel capacity oI an older communication system -- telegraphy. Images contained much more inIormation than telegraphic cable was originally designed to carry, so speed had to be sacriIiced. Throughout the 1920s, engineers in Russia, the U.S., England, and Germany were working on a new system to transmit images in real time called television -- vision over distance. The explosive development oI modern communication systems -- telegraphy, telephone, radio, and television, along with their growing economic importance -- urged the creation oI a general communication theory. 259 With telephone, radio, and television becoming the Ioundations oI whole new industries in the earlier decades oI the century, the question oI
257 M. D. McFarlane, "Digital Pictures FiIty Years Ago," Proc. IEEE 60, no. 7 (1972). 258 Moholy-Nagy, Painting, Photography, Film, 39. 259 For a history oI the developments leading to the creation oI mathematical theory oI communication, see Colin Cherry, On Human Communication, 2nd ed. (Cambridge: The MIT Press, 1968), 41-52.
eIIiciency came to the Ioreground. Engineers needed a quantitative theory which could deIine and measure the commodity that was communicated over telegraph, telephone, and radio networks. Quantitative measures were also required in order to compare the capacities oI various communication systems (Ior instance, wire versus radio transmission) and to evaluate their perIormance. The distant precursor to modern communication theory can be Iound in the centuries-old interest in inventing secret codes Ior military and diplomatic communication. But with the appearance oI modern telecommunications, the problem oI designing codes to make the deciphering oI messages more diIIicult Ior the enemy was joined by a new problem, that oI designing codes which could make everyday communication more eIIicient. Already by 1825, telegraphists designated certain commonly used messages, such as commercial expressions and greetings, by single numbers, thus making coded messages shorter. The Iamous dot-dash code, introduced by S.F.B. Morse in 1832, associated the most commonly used letters with the shortest dot-dash symbols, thus making coded messages use even Iewer symbols. 260
The problem oI how to devise the most eIIicient methods oI encoding, transmitting, and decoding signals was greatly intensiIied by the development oI television in the 1920s. Television, the "Far Seer," required the simultaneous transmission oI a much greater amount oI inIormation than in previous communication technologies: "masses oI inIormation had to be read oII at high speed at the camera end, transmitted, and reassembled in the receiver." 261 The great channel capacity required Ior television spurred new theoretical studies, and by the end oI the 1920s, the Iundamental ideas oI the modern mathematical theory oI communication (which also came to be called inIormation theory or statistical communication theory) were Iormulated. (SigniIicantly, it was the great promise and also the great diIIiculty in the development oI a
260 Cherry, On Human Communication, 36-37. 261 Ibid., 42.
system Ior visual telecommunication -- television -- which was decisive in the rise oI the new paradigm oI a quantitative approach to communication.) The essential ideas oI inIormation theory were presented in 1927 at the International Congress oI Telegraphy and Telephony by R.V.L. Hartley oI the Bell Telephone Laboratories in a paper entitled "Transmission oI InIormation." Hartley proposed "to set up a quantitative measure whereby the capacities oI various systems to transmit inIormation can be compared" and to "discuss its application to systems oI telegraphy, telephony, picture transmission and television over both wire and radio paths." 262 Hartley's theory conceptualized the transmission oI a message as the successive selection oI signs Irom a set, known to both the sender and the receiver. II the receiver knows beIorehand which oI the symbols will be transmitted, no new knowledge (no inIormation) is communicated. The amount oI inIormation is proportional to the uncertainty about which sign can be received. II a particular message is made up Irom N signs which can be chosen Irom a total set oI S signs, then there are S N distinct possibilities. Hartley proposed that the "quantity oI inIormation" (H) oI such a message is HN log 2 S. Hartley also established one oI the basic laws oI inIormation theory: the total amount oI inIormation a system can transmit "is proportional to the product oI the Irequency range which it transmits by the time available Ior transmission." 263 In other words, the more signals a system transmits simultaneously (Irequency range, or bandwidth) and the longer the time available Ior transmission, the greater its inIormation capacity. With Hartley's paper the two Iundamental tools oI inIormation theory were now in place: a quantitative measure oI inIormation and an equation speciIying the inIormation capacity oI a communication system. It may seem that the theory deIined inIormation in an abstract and counter-intuitive way by associating it with the Ireedom to make a choice (on the part oI the
262 R.V.L. Hartley, "Transmission oI InIormation," Bell System Tech. Journal 7 (1928): 535. 263 Ibid., 554.
sender) and with uncertainty (on the part oI the receiver). Yet, the deIinition oI inIormation is quite pragmatic Irom the point oI view oI an engineer concerned with making a communication system eIIicient. What the measure oI inIormation (H) deIines is the minimum number oI binary choices which enables the receiver to reconstruct the message, or conversely, which is required to transmit it. 264 In other words, "it is a measure oI the minimal eIIort by which message can be transmitted." 265
The development oI inIormation theory was Iurther motivated by the work oI scientists and engineers on communication and electronic systems during World War II, the most important oI which was radar. In its modern Iorm, the theory was Iormulated by Claude Shannon; in 1949, his papers, together with a non-technical overview by Warren Weaver, were published in what became the bible oI the coming inIormation age -- Mathematical Theory oI Communication. 266
In his contribution, Weaver extended Shannon's model oI inIormation transmission, created to describe physical systems oI telecommunications, such as telegraphy or radio, to a more general model oI any communication situation (Iig. 18). To emphasize this generality, his description oI the model deliberately juxtaposes examples oI telecommunication systems and human communication:
The inIormation source selects a desired message out oI a set oI possible messages...The selected message may consist oI written or spoken words, or oI pictures, music, etc. The transmitter changes this message into the signal which is actually sent over a communication channel Irom the transmitter to the receiver. In the case oI telephony, the channel
264 Each binary choice constitutes one bit oI inIormation. Weaver writes: "iI one has available say 16 alternative messages among which he is equally Iree to choose, then since 162 4 so that log 2 164, one says that this situation is characterized by 4 bits oI inIormation." Claude E. Shannon and Warren Weaver, The Mathematical Theory oI Communication (Urbana: The University oI Illinois Press, 1949), 100-101. 265 D. Gabor, "A Summary oI Communication Theory," in Proceedings oI a Symposium on Applications oI Communication Theory, London, 1952, ed. Willis Jackson (London: Butterworth ScientiIic Publications, 1953), 2. 266 Shannon and Weaver, The Mathematical Theory oI Communication.
is a wire, the signal a varying electrical current on this wire; the transmitter is the set oI devices (telephone transmitter, etc.) which change the sound pressure oI the voice into the varying electrical current. In telegraphy, the transmitter codes written words into sequences oI interrupted currents oI varying lengths (dots, dashes, spaces). In oral speech, the inIormation source is the brain, the transmitter is the voice mechanism producing the varying sound pressure (the signal) which is transmitted through the air (the channel). In radio, the channel is simply space (or the aether, iI any one still preIers that antiquated and misleading word), and the signal is the electromagnetic wave which is transmitted. The receiver is a sort oI inverse transmitter, changing the transmitted signal back into a message, and handing this message on to the destination. When I talk to you, my brain is the inIormation source, yours the destination; my vocal system is the transmitter, and your ear and the associated eighth nerve is the receiver. In the process oI being transmitted, it is unIortunately characteristic that certain things are added to the signal which were not intended by the inIormation source. These unwanted additions may be distortions oI sound (in telephony, Ior example), or static (in radio), or distortions in shape or shading oI picture (television), or errors in transmission (telegraphy or Iacsimile) etc. All oI these unwanted changes in the transmitted signal are called noise. 267
It is not accidental that the principal examples in this description -- telegraphy, telephony, and oral speech -- all have to do with written or verbal communication. By the middle oI the century, both the engineer and the public still associated electronic communication with telephone and radio, and this is why the example oI human communication which naturally comes to Weaver is speech. It would be Iew decades beIore electronic communication would become synonymous with the communication oI images (television, teleconIerencing, multimedia, teleoperators, ISDN, cyberspace); beIore the terms oI inIormation theory, such as channel, bit, bandwidth will enter the speech oI journalists -- telling us everyday about yet another development in visual telecommunication -- as well as the speech oI policymakers, endlessly debating who will control these channels and how wide their bandwidth will be.
The Mathematical Theory oI Communication appeared in 1949 and within years the theory was taken up by linguists, psychologists, social scientists, and even art historians. Thus, Roman
267 Ibid., 98-99. Emphasis in the original. 3. The InfIuence of Information Theory or the IdeoIogy of the Code
Jakobson, who as a proIessor at MIT was close to the source oI the paradigm, adopted the model Ior linguistics and then semiotics. 268 Even quicker to embrace the paradigm were experimental psychologists. 269
As inIormation theory was transIormed Irom a tool oI the communication engineer into a broad intellectual paradigm, three crucial developments took place. First, the theory was extended to understand not just communication within an electronic system, but also communication between humans as well as between humans and machines. Second, the theory was adopted to describe the meaning and eIIect oI communication -- its semantic and pragmatic aspects. Third, the basic assumption oI the theory, that communication is a one-way process, was extended to theorize social communication. In this section, I will Iirst consider these three developments in turn. The Iirst development is connected with the Iigure oI Norbert Wiener, a proIessor oI mathematics at MIT who supervised Shannon's research. In 1948 Wiener, published Cybernetics or Control and Communication in the Animal and the Machine which proposed that seemingly diIIerent entities -- a machine, an animal, a human, an animal collective, a human collective -- can be conceived oI as communication systems. Wiener also equated communication and control, thus opening a way to think oI social, political, and economic problems in inIormation- cybernetic terms. He justiIied this equation in this way:
When I communicate with another person, I impart a message to him, and when he communicates back with me he returns a related message which contains inIormation primarily accessible to him
268 Roman Jakobson, (1956), "Metalanguage as a Linguistic Problem," Presidential address delivered at the Annual Meeting oI the Linguistic Society oI America, 1956, published in The Framework oI Language (Ann Arbour: Michigan Studies in the Humanities, 1980): 81-92; Roman Jakobson, (1960), "Closing Statement: Linguistics and Poetics," in Semiotics. An Introductory Anthology, ed. Robert Innis (Bloomington: Indiana University Press, 1985). 269 G.A. Miller, Language and Communication (New York: McGraw-Hill Book Co., Inc., 1951); D.E. Broadbent, Perception and Communication (OxIord: Pergamon Press, 1958); H. Quastler, ed., InIormation Theory in Psychology: Problems and Methods (Glencoe, IL.: Free Press, 1955); F. Attneave, Applications oI InIormation Theory to Psychology (New York: Holt, 1959).
and not to me. When I control the actions oI another person, I communicate a message to him, and although this message is in the imperative mood, the technique oI communication does not diIIer Irom that oI a message oI Iact. Furthermore, iI my control is to be eIIective I must take cognizance oI any messages Irom him which may indicate that the order is understood and has been obeyed. 270
In this explication Wiener talks about human communication, but his realization that the "problems oI control engineering and oI communication engineering were inseparable" emerged during the war when he worked on anti-aircraIt artillery systems. Wiener realized that the key to successIul control lies in Ieedback -- the modiIication oI behavior oI a system (be it human or a machine) by taking into account the results oI the actual (rather than predicted) perIormance. The results oI the perIormance are communicated back to the control mechanism, making it adjust the system's behavior. A prototypical example oI a Ieedback mechanism is a selI-guided shell: "When we desire a motion to Iollow a given pattern, the diIIerence between this pattern and the actually perIormed motion is used as a new input to cause the part regulated to move in such a way as to bring its motion closer to that given by the pattern." 271
The authors oI inIormation theory always insisted that the theory was only concerned with the conditions Ior the correct and eIIicient transmission oI signals constituting the message, and not with its meaning, interpretation, or eIIect. Hartley points out that "in estimating the capacity oI the physical system to transmit we should ignore the question oI interpretation." 272
Shannon also begins by insisting that "semantic aspects oI communication are irrelevant to the engineering problem." 273 Finally, Weaver starts his essay included in The Mathematical Theory oI Communication by distinguishing between the three levels oI a communication problem: technical (how accurately the signs can be transmitted), semantic (how accurately signs convey
270 Norbert Wiener, (1950), The Human Use oI Human Beings. Cybernetics and Society (New York: Avon Books, 1967), 24-25. 271 Norbert Wiener, Cybernetics. OI Control and Communication in the Animal and the Machine (New York: John Wiley & Sons, Inc., 1948), 13. 272 Hartley, "Transmission oI InIormation," 538. 273 Shannon and Weaver, The Mathematical Theory oI Communication, 3.
the desired meaning), and pragmatic (how eIIective the message is in aIIecting the behavior oI the receiver). Weaver reIers to these levels as A, B, and C. 274 He next emphasizes that the mathematical theory oI communication as developed by Shannon applies only to the Iirst level, yet also immediately suggests that it is deeply relevant Ior the second and third. Impressed by the Iact that the technical aspect oI a communication process is Iinally controlled by a precise mathematical theory, Weaver clearly hopes that the theory can somehow be extended to handle the elusive questions oI semantics and pragmatics with the same quantitative precision. Throughout the essay he promises a detailed discussion oI how this can be done. Weaver never keeps his promise; instead, in a number oI places he suddenly announces that perhaps no modiIications will be necessary. The announcements appear abruptly at the end oI sentences, explosions in what is otherwise the dry rhetoric oI a scientiIic discourse. "It is the purpose oI this concluding section to review the situation, and to see to what extent and in what terms the original section was justiIied in indicating that the progress made at Level A is capable oI contributing to levels B and C, was justiIied in indicating that the interrelation oI the three levels is so considerable that one's Iinal conclusion may be that the separation into the three levels is really artiIicial and undesirable." "It is almost certainly true that a consideration oI communication on levels B and C will require additions to the schematic diagram on page 97 |Iig. 18|, but it seems equally likely that what is required are minor additions, and no real revision." 275
The temptation to extend inIormation theory to handle the meaning and eIIect oI communication proved too great to resist. This was understandable given the historical context oI the Cold War during which the theory matured and attracted public attention. This context is not hard to see in Weaver's descriptions oI diIIerent levels oI communication: "The semantic
problem has wide ramiIications iI one thinks oI communication in general. Consider, Ior example, the meaning to a Russian oI a U.S. newsreel picture." "The problem oI eIIectiveness involves...all the psychological and emotional aspects oI propaganda theory." 276 Even more revealing is the opening paragraph oI the essay, where within a Iew lines we are transported Irom artistic communication, painting, theater, and ballet, to the theater oI the new cybernetic warIare, which emerged during World War II and now protected the U.S. against a Soviet attack:
The word communication will be used in a very broad sense to include all oI the procedures by which one mind may aIIect another. This, oI course, involves not only written and oral speech, but also music, the pictorial arts, the theatre, the ballet, and in Iact all human behavior. In some connections it may be desirable to use a still broader deIinition oI communication, namely, one which would include the procedures by means oI which one mechanism (say automatic equipment to track an airplane and to compute its probable Iuture positions) aIIects another mechanism (say a guided missile chasing this airplane). 277
In 1952 Y. Bar-Hillel, who was working in the Iield oI automatic language translation (which, as I mentioned in the previous chapter, also developed during Cold War with the promise to automate the surveillance and analysis oI Soviet military communication and media), extended inIormation theory to deIine and measure "semantic inIormation" oI the propositions in a language system. In the same years, the Shannon-Weaver model (Iig. 18) was adopted in the U.S. by the growing Iield oI mass communication studies as the theoretical Ioundation oI the Iield. Weaver's hope that the model, originally designed to deal solely with the technical aspects oI telecommunication, could be applied to describe semantic and pragmatic levels as well, was Iully realized: the model was adopted to describe the process oI mass communication: sender (Iilm studio, TV station, publisher) -- message (a Iilm, a television program, a newspaper story) - - receiver (Iilm viewer, television viewer, newspaper reader). 278
276 Ibid., 97. 277 Ibid., 95. 278 Wilbur Schramm, "How Communication Works," in The Process and EIIects oI Mass Communication (Urbana: University oI Illinois Press: 1954): 3-26.
In this transIer, the basic postulate oI the inIormation theory became an ideology. Shannon begins his exposition oI inIormation theory by stating that "the Iundamental problem oI communication is that oI reproducing at one point either exactly or approximately a message selected at another point." 279 American scholars oI mass communication applied the same postulate to the semantic level. Even a textbook on mass communication published in the 1970s, still states: "iI the meaning oI the destination is isomorphic with the meaning oI the source which originated the act, then communication can be said to have taken place." 280 With such a deIinition, any discrepancy between the codes oI a sender and a receiver becomes undesirable "noise." As Stuart Hall pointed out, the ideology lies in assuming that encoding and decoding codes are or should be the same: "What are called 'distortions' or 'misunderstandings' arise precisely Irom the lack oI equivalence between the two sides in the communication exchange." 281 Thus, while American social scientists contrasted American democracy with Soviet totalitarianism, they simultaneously adopted the theoretical model according to which mass communication was synonymous with Iollowing the prescribed meaning. In short, communication was deIined as control.
Despite the Iact that modern mass communication crucially depends on the visual channel, American scholars oI mass communication never seriously conIronted its visual aspects. Others, however, have tried to use inIormation theory to understand visual communication. I will next
279 Shannon and Weaver, The Mathematical Theory oI Communication, 3. Shannon includes "approximately" because one oI the main concerns oI a communication engineer is to increase the eIIiciency oI a communication system by making it transmit only the inIormation which is absolutely essential Ior the correct identiIication oI a message by the receiver. For instance, the telephone does not carry the whole range oI Irequencies oI the human voice but only those suIIicient Ior understanding what is said. 280 Melvin L. De Fleur and Sandra Ball-Rokeach, Theories oI Mass Communication (New York: Longman, 1975), 127. 281 Stuart Hall, "Encoding/Decoding," in Culture, Media, Language, ed. Stuart Hall et al. (Hatchinson, 1980), 131.
discuss these attempts which took place in the late 1950s and in the 1960s in semiotics, experimental aesthetics, and art history. In the early 1960s Roland Barthes published The Photographic Message and Rhetoric oI the Image where he applied a semiotic apparatus to the analysis oI mass media images, in his case, publicity photographs. 282 However, it may be surprising to discover that The Photographic Message opens by Iormulating the situation oI mass communication in terms oI inIormation theory, exactly in the same way as it was done in communication studies:
The press photograph is a message. Considered overall this message is Iormed by a source oI emission, a channel oI transmission and a point oI reception. The source oI emission is the staII oI the newspaper, the group oI technicians certain oI whom take the photo, some oI whom choose, compose and treat it, while others, Iinally, give it a title, a caption and a commentary. The point oI reception is the public which reads the paper. As Ior the channel oI transmission, this is the newspaper itselI...
Barthes, however, was not going to propose, as communication researchers did, that the system oI mass communication also contains a Ieedback mechanism -- the right oI readers to write letters to newspapers with complaints or suggestions -- thus assuring that mass communication truly serves the purposes oI Democracy and Free Speech, rather than unquestioned ideology. In Iact, Barthes' whole purpose in his articles on the semiotics oI mass media images was to reveal the mechanisms by which images communicate or subvert an ideological meaning. Communicate or subvert: Barthes' position oscillates between these two mutually exclusive positions. On the one hand, Barthes proposes in The Rhetoric oI the Image that the speciIicity oI a photograph's codes is what allows it to naturalize the ideological message in a particularly transparent manner; on the other hand, in the same article he also claims that an image is essentially polysemous and thereIore a caption is required to anchor a single meaning. In a later article The Third Meaning Barthes makes an even stronger claim Ior the subversive
282 Roland Barthes, (1961), "The Photographic Message," in Image, Music, Text, ed. Stephen Heath (New York: Hill and Wang, 1977); "Rhetoric oI the Image," (1964), in Image, Music, Text.
nature oI the visual code in relation to both ideology and interpretation by suggesting that a Iilm still contains an excess, an "obtuse meaning" not reducible to its "inIormational" or "symbolic" meanings. 283
The code. More than any other, it was this particular concept oI inIormation theory that entered semiotics in the 1960s. And Ior semioticians concerned with visual signs, it became the main preoccupation oI their research: is there a unique visual code, an autonomous "language oI pictures," and iI so, what is its speciIicity? Now, aIter we have Iollowed the development oI inIormation theory, we can see the historical speciIicity oI approaching vision as a code by Barthes, J.M. Floch, Felix ThYrlemann, the members oI the Groupe Mu, Fernande Saint-Martin, and other semioticians. 284 The notion oI the code, as inIormation theory itselI, has emerged in response to the practical problem Iaced by engineers: how to make telecommunication systems more eIIicient and how to measure this eIIiciency. Encoding is one oI the principal ways to achieve this eIIiciency. Morse code condensed transmitted messages by associating the shortest combination oI dots and dashes with the most Irequently used letters. Later, communication engineers were able to achieve even greater economy by encoding only as much inIormation as necessary Ior the receiver to decode the message with suIIicient accuracy. 285 When carried over to visual semiotics, the notion oI the code retained its original engineering meaning, its association with the issue oI eIIiciency. Indeed, to ask what kind oI inIormation can be obtained through seeing and/or represented as images; how visual codes diIIer Irom other kinds oI codes; whether visual representations have any unique qualities in terms oI their eIIect on the viewer -- in short, to propose the uniqueness oI the visual code or to deny this uniqueness is already to
283 Roland Barthes, (1970), "The Third Meaning," in Heath, Image, Music, Text. 284 For a critical discussion oI visual semiotics, see Gsran Sonesson, Pictorial Concepts. Inquiries into the Semiotic Heritage and its Relevance Ior the Analysis oI the Visual World (Lund, Sweden: Lund University Press, 1989). 285 A principal contribution oI Shannon was his Iamous Sampling Theorem which speciIied the absolute minimum oI the signal which needs to be retained beIore inIormation is irreversibly lost.
conceive oI vision as productive, as potentially a special way oI transmitting inIormation, to conceive oI it as a communication engineer would. 286
The semioticians did not go as Iar as to apply the mathematical apparatus oI inIormation theory to the analysis oI visual signs. However, the Iield oI experimental aesthetics, which, Irom its origins in the nineteenth century always tried to reduce the eIIect oI an image to a mathematical Iormula, welcomed the quantitative aspect oI the theory. But how to apply a theory developed Ior the purpose oI engineering communication to a very diIIerent question oI aesthetic pleasure? Experimental psychologists interested in aesthetics assumed that a measure oI inIormation in an image could represent its aesthetic eIIect. In the 1960s the German aesthetician Max Bense developed a theory oI "inIormation aesthetics" and tried to measure the "inIormation" contained in such classical images as Rembrandt's etchings. It was not at all obvious, however, how to measure the "inIormation" in an etching or a painting, so other researchers went on to construct simple abstract patterns where "inIormation" was easier to quantiIy. They reasoned that since the amount oI inIormation in a message is directly proportional to its unpredictability and inversely proportional to its redundancy, a pattern made Irom small squares, where the tone oI each square is determined randomly, would be the most unpredictable and thereIore would contain the most inIormation. On the contrary, in a pattern like Mondrian's or Malevich's painting there is much more redundancy (more predictability) and consequently, the amount oI inIormation is low. In a number oI experiments, subjects rated their "aesthetic pleasure" when looking at diIIerent patterns (Iig. 19); psychologists then tried to
286 This carrying over oI assumptions oI inIormation theory into semiotics reaches its extreme in Jurij Lotman's Structure oI the Artistic Text (1970). Lotman was strongly inIluenced by cybernetics which in the 1960s was more popular in the Soviet Union than it ever was in the West and which became as important Ior Soviet semioticians as Saussure or Peirce. (In part, it was used to justiIy semiotics in the eyes oI the authorities by presenting it as a part oI cybernetics, concerned with the study oI diIIerent artiIicial languages, Irom computer languages to the languages oI art.) Lotman claimed that the artistic text is a message coded in a particular way and that art is the most economical way to transmit inIormation because oI the unique Ieatures oI the artistic code. Jurij Lotman, (1970), The Structure oI the Artistic Text (Ann Arbour: Department oI Slavic Languages and Literatures, The University oI Michigan, 1977).
determine the level oI visual organization that was biologically most pleasurable. 287
Psychologists, turned aestheticians. Aestheticians, turned communication engineers. Finally, one oI the most inIluential texts oI post World War II art history is also heavily indebted to inIormation theory. 288 In Art and Illusion (1960) Ernst Gombrich provides a revealing analysis oI Greek vases and mosaics as precursors to modern digital codes; he compares the reading oI pictures to the decipherment oI codes; and he uses various concepts oI inIormation theory throughout the book. 289 However, Gombrich does more than simply pay tribute to a Iashionable theory. During the war, Gombrich was employed Ior six years by the British Broadcasting Corporation to listen Ior radio transmissions. 290 He was conIronted daily with the central problem oI modern inIormation theory -- the identiIication oI signal in the presence oI noise. As he himselI admits, it was this experience which led to his deep interest in inIormation theory as a paradigm Ior understanding perception. Rather than being accidental, the paradigm oI inIormation theory plays a central role in Art and Illusion. One oI Gombrich's key concepts is "the beholder's share": the importance oI inIerence in interpreting the image. According to Gombrich, a representational image is essentially incomplete. 291 As a representation oI three-dimensional space, a two-dimensional
287 See, Ior instance, D.E. Berlyne, "Novelty, Complexity, and Hedonic Value," Perception and Psychophysics 8 (1970): 279-86; D. DorIman and H. McKenna, "Pattern PreIerence as a Function oI Pattern Uncertainty," Canadian Journal oI Psychology 20 (1966): 143-53; P.C. Vitz, "PreIerence Ior DiIIerent Amounts oI Visual Complexity," Behavioral Science 2 (1966): 105-14. It is interesting to compare this work in experimental aesthetics with Lotman's view oI art as communication (see Iootnote 278). Lotman proposed that a unique Ieature oI art is that it carries more inIormation than any other kind oI communication, while experimental aestheticians proceeded under the assumption that there is some optimal level between the complete lack oI inIormation and the complete saturation by it which characterizes the most pleasurable works oI art. 288 I am grateIul to Michael Ann Holly and Norman Bryson who brought to my attention the relevance oI Art and Illusion to the problematic oI this chapter. 289 E.H. Gombrich, Art and Illusion (Princeton: Princeton University Press, 1960), 39-40, 88, 205. 290 Ibid., 204. 291 Ibid., 208, 211.
image is always ambiguous, potentially corresponding to numerous spatial conIigurations. It is equally ambiguous as a representation oI other aspects oI visual reality, substituting a Iew brush strokes Ior a Iield oI grass, a Iew spots oI color Ior a human Iorm. The job oI a beholder is to supplement the partial inIormation in the image with her or his own projection. In other words, Gombrich understands visual communication in terms oI inIormation theory: as the decoding oI a signal in the presence oI noise. According to inIormation theory, this decoding is always probabilistic: there can never be the absolute certainty than an original signal is correctly reconstructed. Similarly, Gombrich describes interpretation as the process oI Iorming and rejecting hypotheses, a process which is never Iinished. For an engineer, this uncertainty oI communication is a problem to be solved. Gombrich, however, interprets it as the source oI our aesthetic pleasure. We derive pleasure Irom art precisely because in trying to "identiIy a signal" we have to use our imagination. There is an unexpected agreement here between Gombrich and experimental aestheticians oI the 1960s, such as Bense. Both start Irom the assumption that the aesthetic pleasure is proportional to the amount oI inIormation in a message, i.e. its unpredictability. As Gombrich explains it, "the greater the probability oI a symbol's occurrence in any given situation, the smaller will be its inIormation content." 292 Because oI their ambiguity, representational images in Western art carry signiIicant inIormation -- and, correspondingly, aesthetic pleasure.
Mass communications, visual semiotics, and aesthetics are just some oI the Iields that adopted concepts oI inIormation theory in order to understand visual communication. The avant-garde oI the 1920s, which dreamed oI engineering visual communication, would welcome all these applications oI inIormation theory. The theory, which was developed by a communication engineer in response to the growth oI television, was now used to rationalize the idea that mass
292 Ibid., 205.
communication is synonymous with control, and that to understand is to decode an intended meaning (mass communication studies). The theory, designed to measure only the eIIiciency oI telecommunications networks, was now adopted to describe all other levels oI communication as well. Visual semiotics and experimental aesthetics extended inIormation theory to the levels oI meaning and eIIect, the Iormer addressing the speciIicity oI visual signiIication, the latter attempting to quantiIy the eIIect oI an image by measuring its "inIormation" content. However, it was the Iield oI experimental psychology that was most transIormed by inIormation theory. AIter World War II, two crucial developments took place. The Iirst is that psychologists, by adopting the approach oI inIormation theory, essentially became engineers who approach human vision as a communication system, studying its channel capacity, bandwidth, and other engineering characteristics. The second is that work, more and more, became a matter oI the mental processing oI visually presented inIormation. And this made experimental psychology, Ior the Iirst time since the discipline emerged, practically important; in Iact, as important as communication engineering. In the next three sections we will see how these developments came about and what their impact was on modern understanding oI vision.
Why was it that even though the essential ideas oI inIormation theory were articulated in the 1920s, it was only in the late 1940s, aIter the publication oI The Mathematical Theory oI Communication, that the theory was suddenly widely accepted in so many Iields? Is there a relationship between this acceptance and the social and economic changes which were taking place in modern society aIter World War II? The best way to answer these questions is to compare the role oI inIormation theory in the second halI oI the twentieth century to the role 4. From "Human Motor" to "Human Information Processing"
previously occupied by thermodynamics. First, a historical parallel. Hermann von Helmholtz Iormulated the universal law oI the conservation oI energy in 1847. Exactly a century later, the theory oI inIormation in its modern Iorm was developed by Shannon and Wiener. Helmholtz was instrumental in turning the theory oI thermodynamics into a more general paradigm which treated the work oI nature, the work oI machines, and human labor on the same terms. Similarly, Wiener extrapolated the concepts oI inIormation theory and the techniques oI control engineering into the science oI cybernetics which described the organization and behavior oI human, machine, and society using a single set oI terms as well -- inIormation, Ieedback, and entropy. The concept oI entropy, in Iact, directly connects thermodynamics and inIormation theory. In 1929 the physicist Leo Szilard identiIied entropy with inIormation, and the measure oI inIormation with the negative oI the measure oI entropy. 293 Entropy is the central notion oI thermodynamics, reIerring to the tendency oI a closed system to change Irom an organized, diIIerentiated, and less probable state to a chaotic, undiIIerentiated, and more probable state. Since entropy is the measure oI randomness, i.e. the unpredictability oI a thermodynamic system, and since inIormation similarly measures the unpredictability oI a message, the equations Ior entropy and inIormation are the same. As Weaver explains, "Ior a communication source one can say, just as he would also say it oI a thermodynamic ensemble, 'This situation is highly organized, it is not characterized by a large degree oI randomness or oI choice -- that is to say, the inIormation (or the entropy) is low.'" 294
But beyond these connections, there is a more crucial parallel. Just as thermodynamics provided the model Ior understanding human work in the latter nineteenth and Iirst halI oI the
293 Howard Resnikov, The Illusion oI Reality (New York: Springer-Verlag New York Inc., 1989), 5. 294 Shannon and Weaver, The Mathematical Theory oI Communication, 103.
twentieth century, inIormation theory has been crucial Ior a new post-industrial model oI work, which gradually emerged aIter 1940.
In The Human Motor: Energy, Fatigue and the Origins oI Modernity Anson Rabinbach demonstrated how the scientiIic ideas oI thermodynamics, Iormulated in the middle oI the nineteenth century, became central Ior the conception oI work in modernity. Helmholtz, who discovered the law oI the conservation oI energy, promoted this law as the universal principle which equally applies to nature, machines, and humans. Helmholtz "portrayed the movements oI the planets, the Iorces oI nature, the productive Iorce oI machines, and oI course, human labor power as examples oI the principle oI conservation oI energy." 295 All work was understood as the expenditure oI energy, with a crucial consequence oI redeIining human labor as labor power, the expenditure oI the energy oI a body. Thus a worker was redeIined as a "human motor." This, in turn, lead to the emergence, towards the end oI the century, oI the movement which Rabinbach calls the European science oI work, "the search Ior the precise laws oI muscles, nerves, and the eIIicient expenditure oI energy centered on the physiology oI labor." 296 In manual labor, the energy stored in the body where it was accumulated through the intake oI Iood, sleep, and rest is transIerred into muscular Iorce -- hammerer striking a blow, Iiler Iiling a machine part, and so on. ThereIore, psychologists, physiologists and industrial experts searched Ior methods to maximize both the accumulation oI a worker's energy (through proper nutrition, shorter working hours, appropriate breaks) and its expenditure in labor. Just as an engineer designing an engine was concerned with the most eIIicient transIer oI Iuel energy into movement, European work experts aimed to maximize worker eIIiciency and to eliminate possible waste. Central to the quest Ior the eIIiciency oI the human motor was the struggle
295 Anson Rabinbach, The Human Motor: Energy, Fatigue, and the Origins oI Modernity (Basic Books, Inc., 1990), 3. 296 Ibid., 10.
against Iatigue, understood as the equivalent oI entropy. "As entropy revealed the loss oI energy involved in any transIer oI Iorce, so Iatigue revealed the loss oI energy in the conservation oI KraIt to socially useIul production. As energy was the transcendental, 'objective' Iorce in nature, Iatigue became the objective nemesis oI a society Iounded on labor power." 297
The European science oI work may appear to be very similar to the American scientiIic management movement pioneered by Frederick Winslow Taylor, a Iormer engineer turned management consultant. As a part oI his program, Taylor aimed to minimize and standardize the time required by a worker to perIorm each operation. He employed the method oI time studies whereby the best workers were timed and the results became the norm to be Iollowed by the rest. 298 Later, Frank and Lilian Gilberts (he -- an engineer, she -- a psychologist) popularized another method oI motion study. 299 They argued that maximizing worker productivity is best achieved by the elimination oI unnecessary movements and making the necessary more eIIicient. Although both time and motion studies and the European science oI work were concerned with the eIIiciency oI manual work, there was a Iundamental diIIerence between the two approaches. 300 Taylorism aimed Ior maximum productivity, and had no concern Ior the exhaustion and deterioration oI the human motor. In contrast, European scientists aimed Ior optimum productivity, and thereIore were concerned not only with the rationalization oI the workplace, but also with the workers' health, nutrition, saIety, and the optimal length oI a workday. In short, Taylorism had no reservations about replacing one exhausted human motor with another -- the philosophy which in the U.S. seems to go hand in hand with the emerging ethics oI the consumer society and with immigration policies which assured the constant supply
297 Ibid., 68. 298 Frederick Winslow Taylor, The Principles oI ScientiIic Managment (New York, 1967). 299 William R. Spriegel and Clark E. Myers, eds., The Writings oI the Gilbreths (Homewood, IL., 1953). I am grateIul to Lisa Cartwright Ior introducing me to the work oI the Gilbreths and its relevance Ior the history oI modern vision. 300 Rabinbach, The Human Motor, 117, 277.
oI a cheap labor Iorce. Europeans, on the other hand, were committed to caring Ior and servicing the human motor. The two paradigms converged aIter World War I, when European industrialists partly adopted the more brutal, but ultimately more eIIective Taylorist methods, while U.S. managment experts became more sensitive to workers' physiology and psychology. Taylorism reduced the worker's body to a mechanical machine and had no concern Ior her or his mind. Indeed, as Marta Braun points out, Taylorism aimed to systematically rob the worker oI any degree oI independence or even understanding oI the overall work process by "separating responsibility Ior the execution oI work Irom its planning or conception." 301 This disdain Ior the mind was shared by behaviorism, which matured at the same time as the European science oI work and Taylorism, and which equally well characterizes the imaginary oI hard-edged social engineering oI the Iirst halI oI the twentieth century. In 1913, J.B. Watson, the Iounder oI behaviorism, explicitly deIined it as the science oI social control: "Psychology as the behaviorist views it is a purely objective experimental branch oI natural science. Its theoretical goal is the prediction and control oI behavior." 302 Behaviorism approached the human subject as an input-output system oI stimulus and response to be controlled through conditioning. Concerned with controlling the body, it almost completely suppressed any studies oI perceptual or mental processes between 1920 and 1950 in the U.S. It was a psychology well suited Ior controlling the subject already reduced to the brainless human motor.
301 Marta Braun, Picturing Time: the Work oI Etienne-Jules Marey (1830-1904) (Chicago: The University oI Chicago Press, 1992), 337. 302 Qtd. in Eliot Hearst, "One Hundred Years: Themes and Perspectives," in The First Century oI Experimental Psychology, ed. Eliot Hearst (Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1979), 27. 5. Communication Engineer AnaIyzes Human Vision
AIter World War II, the reIerences to time and motion studies disappear. And by 1960, the dominance oI behaviorism is over as well. Industrial society became post-industrial society. In this shiIt, the concepts oI manual labor, production oI goods, and Iatigue were replaced by new concepts oI cognitive labor, inIormation processing, and noise. Taylorism, Gilberts' motion studies, and behaviorism gave way to engineering psychology, "human inIormation processing," and cognitive science. In short, with the transIormation oI industrial society into post-industrial society, the disciplines oI the eIIiciency oI the body were replaced by the disciplines concerned with the eIIiciency oI the new instrument oI labor -- the mind. One obvious and perhaps the earliest sign oI this transition was the speed with which inIormation theory was disseminated into so many Iields by the end oI the 1940s and early 1950s. But, along with being a symptom oI post-industrial society which was yet to Iully maniIest itselI, inIormation theory also provided actual techniques Ior deIining and studying the human subject as an "inIormation processing unit," and its visual apparatus as a crucial part oI "human inIormation processing." InIormation theory thereIore is central Ior understanding the new ideas and techniques oI vision that emerged aIter the World War II. The signiIicance oI the adaptation oI inIormation theory as a new model oI the human observer, which took place in the 1950s, can be best seen through the Iollowing analogy: Industrial society: manual labor, energy, Iatigue. Post-industrial society: mental labor, inIormation processing, noise.
1. Manual versus mental labor. The point is not whether corporeal labor was indeed universally displaced by mental labor: this is diIIerent Irom country to country, Irom industry to industry. What is important is that the obsession with the rationalization oI corporeal work (Taylorism, European science oI work, psychotechnics) disappeared, displaced by new obsession with the rationalization oI the mind (cognitive psychology, artiIicial intelligence, cognitive
engineering). Regardless oI the percentage oI the work Iorce that still may be engaged in manual labor, society is no longer concerned with spending more intellectual resources to perIect workers' movements. What comes under scrutiny since the 1950s, when cognitive psychology begins to displace then dominant behaviorism, are mental Iunctions: perception, attention, text comprehension, memory, and problem solving. This replacement oI manual work by cognitive work is directly related to automation. Already in 1961, in an inIluential study oI automation in French industry, Pierre Naville and his Iellow sociologists had described the transition Irom the "work oI the laborer to the work oI communication," work which became primarily "cognitive or semiotic." 303 In his summary oI this study Rabinbach writes, "The appearance oI the cerebral worker whose material and product is 'inIormation' is emblematic oI the vast distance traversed between the worker who surveys complex technologies oI communication and the 'man-beeI' oI Taylor." 304 It is important to note that automation does not lead to the replacement oI human by machine. Rather, the worker's role becomes one oI monitoring and regulation: watching displays, analyzing incoming inIormation, making decisions, and operating controls. And it is the corresponding human Iunctions oI perception, attention, memory, and problem solving which become the subject oI research by new cognitive sciences. What Taylor's scientiIic management was Ior the age oI industrialization, cognitive sciences are Ior the age oI automation. In the 1940s, Herbert Simon worked on theories oI management, the Iield oI research originated by Taylor. Having recognized the increasing importance oI mental skills in the corporate workplace, Simon became one oI the pioneers oI cognitive science with his work on automatic reasoning by computer. In 1964 he wrote that "the
303 Qtd. in Rabinbach, The Human Motor, 298. 304 Ibid., 298.
bulk oI productive wealth consists oI programs...stored in human minds." 305 Another pioneer oI cognitive science was Jerome Bruner. ReIlecting back on his work in the 1950s, he noted in 1983: "It seems plain to me now that the 'cognitive revolution'...was a response to the technological demands oI the 'post-industrial revolution.' You cannot properly conceive oI managing a complex world oI inIormation without a workable concept oI mind." 306
2. Just as thermodynamics in the nineteenth century provided a model Ior the laboring body, inIormation theory provided a new model Ior the mental processes, becoming particularly inIluential in the research on human senses, including vision. In 1958 the English psychologist D.E. Broadbent published Perception and Communication, which was the Iirst attempt to understand human sensory-motor perIormance as inIormation processing, or, as he put it, to describe it "in terms originally developed Ior telephone engineering." 307 This statement may at Iirst remind us oI the comparison between nerves and telegraphy, Irequently evoked by nineteenth century physiologists and psychologists, including Helmholtz. 308 Yet, the approach oI Broadbent and other psychologists who Iollowed him was Iundamentally new. It was less concerned with human physiology than with the limits oI human perIormance understood as a communication system. In other words, just as a modern telephone engineer is concerned with the capacity oI the telephone system, a psychologist is concerned with the amount oI visual or auditory inIormation that can be adequately processed by a human system; sensed, recognized, stored in memory, recalled, acted upon. What Ior Helmholtz was a metaphor became a systematic program oI research a century later.
305 Qtd. in Douglas Noble, "Mental Materiel: The Militarization oI Learning and Intelligence in U.S. Education," in Cyborg Worlds: the Military InIormation Society, ed. Les Levidov and Kevin Robins (London: Free Association Books, 1989), 34. 306 Qtd. in Ibid., 34-35. 307 Broadbent, Perception and Communication, 36. 308 Jonathan Crary, Techniques oI the Observer: on Vision and Modernity in the Nineteenth Century (Cambridge: The MIT Press, 1990), 93-94.
For a communication engineer, inIormation theory provides both a standard set oI terms to speciIy quantitatively the perIormance oI a communication system and a set oI principles which speciIy the limits oI this perIormance. For instance, the capacity oI a system depends on the code used. Morse code makes transmitted messages shorter by associating the most Irequently used letters with the shortest dot--dash symbols. The time saved in transmitting, however, is oIIset by the additional time now required to encode and decode the message. Another example is the principle, already established by Hartley, that the maximum amount oI inIormation a system can transmit is the product oI its bandwidth and the time available Ior transmission. In the 1920s, it was possible to transmit images over telegraph wires -- a small bandwidth system -- by sacriIicing time. In addition to the adoption oI the terms oI inIormation theory, such as bandwidth, inIormation capacity, and Iilter, psychologists also applied its principles to understanding the limits oI human perIormance. One oI Broadbent's contributions was the idea that the nervous system cannot process all oI the sensory stimulation which may be available and that there is "a Iilter at the entrance to the nervous system which will pass some classes oI stimuli but not others" 309 (Iig. 20). In other words, because oI its limited bandwidth, a nervous system can only process a certain amount oI inIormation in a given unit oI time. How much can be processed depends on the nature oI this inIormation since diIIerent human senses such as vision and hearing, now understood as separate communication channels, have diIIerent inIormation capacities. Once Iiltered through the senses, the inIormation is Iurther processed through a series oI stages. Each stage is understood as a speciIic communication system with its own capacities determined by its codes and bandwidth. Figure 21, taken Irom 1986 Handbook oI Perception and
309 Broadbent, Perception and Communication, 42.
PerIormance, summarizes the current knowledge oI the temporal capacities oI diIIerent stages oI sensory and mental processing. So just as the productivity oI a laboring body was limited by the available energy, it is now assumed that the productivity oI the mind is limited by its inIormation-processing capacity. In Iact, the notion oI a limited reservoir oI the energy in the body has Iound a direct correlatation in the notion oI there being a limited amount oI human inIormation-processing resources which can be used in diIIerent ways to deal with the incoming inIormation. For instance, the subject may concentrate all his/her resources on processing visual inIormation, thus sacriIicing his or her attending to auditory stimuli.
3. Fatigue versus noise. Just as Iatigue is the opposite oI energy, the limit to the body's productivity, noise, always present in a communication system, whether natural or man-made, is the limit to perIect communication and, consequently, the limit to Iaultless human inIormation processing. Noise is the source oI error. And error is the worst enemy oI the human-machine system in post-industrial cybernetic society. The late nineteenth century science oI work came about as a struggle against Iatigue, this "subversive element in the human motor, the body's IiIth column...Iatigue was the physiological limit oI even the most perIectly executed work, the horizon oI the metaphor oI the human motor." 310 As physiologists and psychologists were preoccupied with Iighting the body's Iatigue, cognitive psychologists became preoccupied with preventing human cognitive and perceptual errors. A textbook on human Iactors begins a chapter on error by optimistically asking: "Are errors a basic part oI human nature that cannot be changed, or is there some way to eliminate or reduce error and the damage errors cause?" The experts on human Iactors classiIied
310 Rabinbach, The Human Motor, 118.
various types oI human errors, came up with mathematical Iormulas to calculate human reliability, and compiled tables oI Human Error Probability (HEP) Ior diIIerent tasks. 311
What are the sources oI human error? In contrast to a manual worker oI the industrial age, an operator in a human-machine system is primarily engaged in the observation oI displays which present inIormation in real time about the changing status oI a system or an environment, real or virtual: a radar screen tracking a surrounding space (see chapter 3); a computer screen updating the prices oI stocks; a video screen oI a computer game presenting an imaginary battleIield; a control panel oI an automobile showing its speed, etc. 312 From time to time, some inIormation causes an operator to make a decision and to intervene in the system's operation: tell the computer to track an enemy bomber noticed on the radar screen; buy or sell a stock; press a joystick; change gears. It is not essential that in some situations these interventions may be required every second (a pilot engaged with an enemy, a computer game player, a Iinancial analysis monitoring stock prices), while in others they are needed very rarely (a technician monitoring an automated plant, power station, a nuclear reactor; a radar operator monitoring a radar screen, waiting Ior potential enemy planes). This is why a human-machine system is deIined as "an equipment system, in which at least one oI the components is a human being who interacts with or intervenes in the operation oI the machine components oI the system Irom time to time." 313
In a Iamous passage Walter Benjamin characterized modern experience as a constant periodic rhythm oI perceptual shocks; the experience shared by an assembly line worker, by a
311 Barry H. Kantowitz and Robert D. Sorkin, Human Factors: Understanding People-System Relationships (John Wiley & Sons: 1983), 30-57. 312 A 1965 textbook on human-machine systems calls an automobile "a Iirst rate example oI a true man-machine system...a highly complex system in which the operator plays a commanding role or actively intervenes in the system Irom time to time." Alphonse Chapanis, Man-Machine Engineering (Bemont, CA: Wadsworth Publishing Company, Inc., 1965), 16. 313 Chapanis, Man-Machine Engineering, 16. Emphasis mine -- L.M.
pedestrian, and by a Iilm viewer. 314 This experience, characteristic oI modernization, Iinds a direct continuation in one type oI cybernetic workplace: the constant overwhelming amount oI inIormation; the constant cascade oI cognitive shocks which require immediate interventions (a pilot engaged with an enemy, a player oI a computer game). Now, however, these shocks arrive exclusively through the visual channel (dials, computer screen, head-mounted display). Thus oI the roles mentioned by Benjamin, it is the Iilm viewer rather than the assembly line worker who directly anticipates the experience oI an operator in this type oI human-machine situation. However, there is also a second type oI work experience, new to post-industrial society: work as waiting Ior something to happen. A radar operator waiting Ior a tiny dot to appear on the screen; a technician monitoring an automated plant, power station, or nuclear reactor, knowing that a soItware bug will eventually maniIest itselI, making a pointer on one oI numerous dials shoot into the red... II a manual worker is eventually overcome by Iatigue, an operator monitoring displays in situations where nothing may happen Ior hours eventually loses vigilance. In terms oI the inIormation processing approach, when a signal occurs inIrequently and randomly over time, an operator is more likely to miss the signal. 315 This is one source oI human error. Another source
314 "Whereas Poe's passers-by cast glances in all directions which still appeared to be aimless, today's pedestrians are obliged to do so in order to keep abreast oI traIIic signals. Thus technology has subjected the human sensorium to a complex kind oI training. There came a day when a new and urgent need Ior stimuli was met by the Iilm. In Iilm, perception in the Iorm oI shocks was established as a Iormal principle. That which determines the rhythm oI production on a conveyer belt is the basis oI the rhythm oI reception oI Iilm." Walter Benjamin, "On Some Motives in Baudelaire," in Illuminations, ed. Hannah Arendt (New York: Schochen Books, 1969), 175. 315 In the extreme case oI sensory deprivation the subject becomes totally disoriented, experiences illusions and even Iaints. It is interesting that in the 1950s, when automation was making this new work-as-waiting-Ior-signal commonplace, experimental psychologists began to conduct experiments in artiIicial sensory deprivation. In 1953 L.A. Riggs and his colleagues discovered that iI the image is stabilized on the retina, thus eliminating miniature eye movements normally accompanying vision, the image vanishes. In 1954-57, W. Heron constructed an experimental cubicle to study the eIIects oI total perceptual isolation. Subjects, deprived oI all visual, auditory, and tactile stimuli, experienced extreme boredom, restlessness and, eventually, visual hallucinations. L.A. Riggs, et al., "The Disappearance oI Steadily Fixated Visual Tests
is visual illusions, limits to the reliability oI human vision, which preoccupied experimental psychologists Irom the early days oI the discipline. Still another source is the inherent limitations oI a human as a decision maker. When cognitive psychologists began to study human reasoning, they made a grave discovery: human judgments do not Iollow statistical probabilities and ignore inIormation about the prior probability oI an event. 316 So even iI the operator detects a signal in time, s/he may still make a wrong decision.
Along with providing a model that allowed psychologists to conceive oI human vision as a communication system where perIormance may be described in quantitative terms, inIormation theory also provided a general concept to speciIy an inherent limit to this perIormance -- noise. The incorporation oI the concept oI noise is what distinguishes modern inIormation theory, as developed by Shannon, Wiener, KolmogoroII, and others, Irom the original Iormulations by Hartley. In any real telecommunication system noise is always present due to the random motion oI electrons in all electrical conductors. ThereIore, in practice, the signal at the receiving end is always to some degree diIIerent Irom the signal originally sent. One oI Shannon's principal contributions was an equation Ior determining the maximum capacity oI a communication channel to transIer inIormation in the presence oI noise. As Colin Cherry notes, "This channel capacity concept is somewhat analogous to the notion oI conservation oI energy; it is a deIinite limit which no practical system can exceed, and its principal value is that it provides a standard against which eIIiciencies may be assessed." 317
Just as entropy limits the perIormance oI any thermodynamic system (such as a motor), leading to some oI the energy being lost unproductively, noise sets the limit on the eIIiciency oI
Objects," J. Optical Soc. America 43 (1953): 495; W. Heron, "The Pathology oI Boredom," ScientiIic American 196 (1957): 52-56. 316 Gillian Cohen, The Psychology oI Cognition, 2nd ed. (London and New York: Academic Press, 1983), 190. 317 Cherry, On Human Communication, 52.
communication. And just as attempts to rationalize manual perIormance stumbled across the problem oI Iatigue, the "physiological limit" to the eIIiciency oI work, psychologists recognized noise as the inherent limit in the perIormance oI a human inIormation processing system. II the source oI noise in a telecommunication apparatus is the random motion oI electrons, the source oI noise in the nervous system is the random Iiring oI neurons. Always present, these random Iirings are assumed to generate a small but constantly present level oI neural activity which interIeres with human inIormation processing. The observer may miss a signal (Ior instance, a Ilash oI light in a psychophysical experiment or a dot on a radar screen in a real situation) or, on the contrary, detect a signal where none was present; or s/he may make a mistake in estimating the signal's strength or its position, and so on. In Iact, the noise in man-made components and biological noise become combined. Due to the noise in its electronic parts, radar may display a dot when nothing is present or the location oI the dot may be inaccurate. Additional noise is added by the imperIections oI a display, such as glare, inadequate contrast, or placement. Finally, neural noise may cause the radar operator to miss the dot, to misjudge its size and, consequently, either to miss an enemy plane or to send an intercepting weapon towards a Iriendly aircraIt. In this way, the overall perIormance oI a human- machine system, such as a radar installation, is equally limited by the perIormance oI its electronic and human parts.
A radar operator? A human-machine system? Was not I talking about inIormation theory as a paradigm Ior experimental psychology, a purely academic discipline? A discipline whose practitioners, according to William Estes, "developed the tradition and the reputation oI being 6. Human Engineering
among the purest oI pure scientists" during its Iirst decades? 318 Practitioners who Ior decades never leIt the seclusion oI the university laboratory while their colleagues -- physiologists, Iatigue experts, behaviorists -- seemed to hold the knowledge necessary to maximize the productivity oI industrial society? The Iact that aIter 1950 experimental psychologists displaced their colleagues in importance, oI course, Iits in with the larger shiIt Irom industrial to post-industrial society and the new image oI work: visual and mental processing oI inIormation rather than corporeal labor. The problem with such global explanations is that they explain things too well. It is easy to connect the interest oI experimental psychologists in human inIormation processing to the shiIt to post-industrial society. In Iact, it is as easy as to connect the very emergence oI experimental psychology in the nineteenth century to the processes oI modernization, industrialization, or rationalization. For instance, discussing Fechner's psychophysics, developed in the 1850s, Jonathan Crary writes that Fechner's equations were "a means oI rendering a perceiver manageable, predictable, productive and above all consonant with other areas oI rationalization." 319 Similarly, in his discussion oI reaction time experiments, which became standard in experimental psychology by the 1890s, Didier Deleule points out that "the measured time is only the sign oI the living machine's capacity Ior increasing its productivity." 320
Both Crary and Deleule are right in seeing the social utility oI experimental psychology. It should be also noted, however, that this social utility remained unIulIilled Ior many decades. In Iact, the enormous amount oI knowledge accumulated by psychologists did not Iind its systematic application until World War II.
318 William Estes, "Experimental Psychology: an Overview," in The First Century oI Experimental Psychology, ed. Eliot Hearst (Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1979), 629. 319 Crary, Techniques oI the Observer, 147. 320 Didier Deleule, "The Living Machine: Psychology as Organology," in Incorporations, ed. Jonathan Crary and SanIord Kwinter (New York: Zone Books, 1992), 215.
The gradual expansion oI the practical applications oI experimental psychology provides a precise map oI the new occupations and new conditions oI modern experience which call Ior perceptual skills. During World War I, England, Germany, and France utilized experimental psychologists to design and administer tests Ior aviation pilots, aeronautical, and airplane observers, hydrophone operators, and submarine "listeners-in." 321 During peacetime, a number oI psychologists published papers on the readability oI written text and oI highway signs and on the visibility oI lights at sea. 322 However, in the industrial world which conceived oI the worker as a human motor and was largely concerned with the productivity oI manual rather than perceptual labor, these studies were an exception rather than the mainstream rule. It was World War II which Iinally put to use the expertise oI experimental psychologists. Why did this happen? The Iirst textbook on applied experimental psychology (1949) opens by describing the recent origins oI the Iield:
For years experimental psychologists have worked diligently in academic laboratories studying man's capacities to perceive, to work, and to learn. Only very slowly, however, have the Iacts and methods which they have assembled been put to use in everyday liIe. A particularly glaring gap in modern technology, both industrial and military, is the lack oI human engineering -- engineering oI machines Ior human use and engineering oI human tasks Ior operating machines. Motion-and- time engineers have been at work on many oI these problems, but the experimental psychologist is also needed Ior his Iundamental knowledge oI human capacities and his methods oI measuring human perIormance. The recent war put the spotlight on this gap. The war needed, and produced, many complex machines, and it taxed the resources oI both the designer and operator in making them practical Ior human use. The war also brought together psychologists, physiologists, physicists, design engineers, and motion-and-time engineers to solve some oI these problems. Though much oI their work began too late to do any real good, it has continued on a rather large scale into the piece. Today, there are many groups busy with research on man-machine problems. They use diIIerent names to describe the work in its various aspects: biotechnology, biomechanics, psychoacoustics, human engineering, and systems research. Other names may be appropriate and may appear in the Iuture. In casting about Ior a title Ior this book, we tried to select one that would describe the subject matter without the restrictive connotations attaching to some oI the names
321 Morris Viteles, Industrial Psychology (New York: W.W. Norton & Company, Inc., 1932), 43. 322 Paul Fitts, "Engineering Psychology and Equipment Design," in Handbook oI Experimental Psychology, ed. S.S. Stevens (New York and London: John Wiley & Sons, Inc., 1951), 1287- 1340.
mentioned above. Applied Experimental Psychology seems best to Iill these requirements, because the traditional data and subject oI experimental psychology are Iundamental to this Iield. 323
Already beIore the war, experimental psychologists assisted in selecting military personnel Ior such jobs as pilot or airplane observer by administering special aptitude tests. During the war, a much greater number oI pilots, radar operators and other similar personnel became needed. The emphasis was shiIted, thereIore, Irom selecting personnel with particularly good perceptual and motor skills to designing the equipment (controls, radar screens, dials, warning lights) to match the sensory capacities oI an average person. 324 And it was the Iield oI experimental psychology that possessed the knowledge about the sensory capacities oI an average, statistical person: how visibility and acuity vary between day and night; how the ability to distinguish colors and brightness vary with illumination or distance; what the smallest amount oI light is which can be reliably noticed; and so on. 325 All this data was now utilized Ior designing better displays and controls oI the Iirst modern human-machine systems such as high-speed aircraIts or radar installations. The development oI these new human-machine systems during the war pushed human perceptual and mental perIormance to the limit and this was the second reason why experimental psychologists were called in. The perIormance oI a human-machine system was limited by human inIormation capacity to process inIormation. In the words oI the authors oI Applied Experimental Psychology,
We can make a machine that will do almost anything, given enough time and enough engineers. But man has limits to his developments, at least as Iar as we can see it. When we think how much a single radar can do in a small Iraction oI a second, and then realize by comparison that even the simplest Iorm oI reaction Ior a human being requires about a IiIth oI a second, we realize what we are up against... The Iull potential oI radar, Ior example, lagged Iar behind physical developments
323 Alphonse Chapanis, Wendell R. Garner, and CliIIord T. Morgan, Applied Experimental Psychology (New York: John Wiley & Sons, Inc., 1949), v. 324 Ibid., 8. 325 Estes, "Experimental Psychology: an Overview," 630.
because human operators could not master the complex operation oI this machine system. We had to worry about such things as a new kind oI visual signal -- very small and not very bright. 326
Considering that the authors described the work oI time-and-motion engineers as directly leading to applied experimental psychology, this rhetoric can be expected. Taylor was impatient with the limitations oI the body; now there was a similar impatience with the limitations oI human inIormation processing. With Taylor, it was the question oI the speed oI muscular movements; now, it became the question oI reaction time: the minimum time in milliseconds required Ior an operator to detect a signal, to identiIy it, to press a control. In order to measure normal human sensory capacities, experimental psychologists have always put subjects in, so to speak, boundary conditions. They measured sensory thresholds, such as the least amount oI light which can be detected. They also measured just noticeable diIIerences (j.n.d.), the smallest diIIerence between two stimuli which can be detected. Finally, they measured reaction times, the measure which became the main tool to deduce the time taken by diIIerent mental processes. In order to measure these characteristics, a number oI standard experiments were designed, and they remained largely unchanged Irom the times oI Weber, Fechner, and Wundt. In a detection experiment, the task oI an observer is to detect the presence oI barely visible stimuli, Ior instance a tiny light brieIly Ilashed in the dark (did I see something?). In an identiIication experiment, the task is to identiIy which oI possible stimuli was presented, Ior instance which oI two colors (which one did I see?). In a recognition experiment, the task is to not only detect something, but to recognize what it is, Ior instance: what was the shape that brieIly appeared (what did I see?) During World War II, the radar operator, the anti-aircraIt gunner, the aircraIt pilot Iound themselves in the same situations in which nineteenth century psychologists put their experimental subjects. The setups oI psychophysical experiments became, in all details, the
conditions oI military work; the tasks devised by psychologists to study human vision became the actual tasks Iaced by the operators oI human-machine systems. Like the subject oI a detection experiment, a radar operator scans the radar screen Ior a barely noticeable dot oI light. 327 Like the subject oI an identiIication experiment, he has to try to guess whether this dot is the same or diIIerent Irom another dot which Irom his previous experience he knows to correspond to a Iriendly airplane. An anti-aircraIt gunner is subjected to a recognition experiment, trying to identiIy a plane by its shape. And all oI them, especially the pilot, are engaged in a sort oI reaction time experiment. Thus, nineteenth century psychophysical setups became the military, and soon, civilian workplaces oI post-industrial society; Irom there, they traveled back into laboratories, leading to such close interrelations between basic research in experimental psychology and its practical applications that they were no longer separable. 328 The terms "applied experimental psychology," "human engineering" and "man-machine engineering" were replaced by another term standard today -- "human Iactors." The radar operator who in the 1940s and 1950s was the prototypical example oI a human-machine system, was replaced by the 1980s by a new prototypical Iigure, the computer user. Thus, reIerences to "human-machine systems" became reIerences to "human-computer systems." The same amount oI intellectual energy and research which in the middle oI the century went into theorizing the perIormance oI a radar operator and adapting him and radar display to each other, today goes into the work on computer interIaces. In
327 As Paul Fitts notes in his 1951 overview oI engineering psychology, "radar operators are oIten Iorced to search Ior weak signals at near-threshold levels." Fitts, "Engineering Psychology and Equipment Design," 1290. 328 For example, a 1947 article in American Psychologist describes the work oI Naval Research Laboratory as Iollowing these three directions: "the design oI gun Iire control and missile control instruments Irom the point oI view oI ease and eIIiciency oI operation; the design and evaluation oI synthetic gunnery and missile control trainers; and basic psychological research." But what is meant here by "basic research"? We read that "at present, all basic research studies are aimed at the eye-hand coordination problem involved in target tracking." "Target tracking" is just one example oI a military task which traveled into a psychological laboratory, and gradually become a standard psychophysical experiment. Franklin Taylor, "Psychology at the Naval Research Laboratory," American Psychologist 2, no. 3 (1947): 87, 91.
retrospect then, we should recognize the radar operator as the central Iigure standing at the origins oI post-industrial society, the Iigure which put directly into motion the new disciplines oI the eIIiciency oI the mind: engineering psychology, human inIormation processing, and cognitive science. We are now in a position to see that the ready acceptance oI inIormation theory by psychologists in the 1950s was not simply a part oI a "Zeitgeist" oI the coming inIormation age. Rather it was a practical necessity. The 1986 Handbook oI Perception and Human PerIormance describes the rationale Ior adopting the inIormation processing approach in psychology in this way:
Because the language oI inIormation processing provides an objective and quantitative way oI describing the basis oI human perIormance, it has proven useIul in applications. Indeed, much oI the impetus Ior the development oI this kind oI empirical study stemmed Irom the desire to integrate description oI the human within the overall systems. In the design oI cockpits, control systems Ior weapons, and other military applications, it has been useIul to have descriptions oI the speed, accuracy, and reliability oI the human in terms comparable to those used Ior other mechanical parts oI the system. 329
As psychologists became involved in designing human-machine systems, they needed a language to characterize human perIormance quantitatively. But not just any language. II the human is conceived oI as part oI a human-machine system (Ior instance, in the role oI a radar operator), it is convenient to use the same language to describe the perIormance, capacities, and errors oI both the electronic and human components. Put diIIerently, iI the human-machine system is conceived oI as a machine, and its perIormance is to be characterized in engineering terms, the same terms need to be applied to its human components iI the perIormance oI the overall machine is to be characterized at all. Since inIormation theory already provided a language to
329 Michael I. Posner, "Section V: InIormation Processing. Overview," in Handbook oI Perception and Human PerIormance, ed. Kenneth BoII, Lloyd KauIman, and James P. Thomas (New York: John Wiley and Sons, 1986), II: V-6.
characterize the perIormance oI electronic components, it was logical to apply it to the human component as well.
Modernization brought with it a special discipline concerned with eIIiciency -- engineering. The job oI an engineer was to ensure maximum perIormance with a minimum investment oI energy, materials, and time, be it the perIormance oI machines (mechanical engineering), communication systems (communication engineering) or human bodies (scientiIic management, time and motion studies). Inspired by modern engineering, the avant-garde oI the 1920s tried to systematically apply its principles to vision. To engineer vision meant to eliminate waste, to use minimal material resources. Thus, constructivist graphic design streamlined typography, eliminating complicated typeIaces in Iavor oI block letters consisting oI straight lines; it also eliminated illustrations and "wasteIul" decorations by making type itselI the main element oI design. The goal: maximum impact with minimum use oI ink. To engineer vision also meant to minimize the psycho-physical resources required oI the viewer. Vertov writes in his Iamous 1923 maniIesto: "The least advantageous, the least economical communication oI a scene is theatrical communication." 330 In contrast, montage Iorces the eye to see the right thing at the right time, thus eliminating the visual waste oI theater, ballet, painting, and other traditional Iorms. In montage, "camera drags the eyes oI a Iilm viewer Irom hands to legs, Irom legs to eyes and the rest in the most advantageous order..." 331
330 Vertov, "Kinoki," 139. 331 Ibid., 139. Emphasis in the original -- L.M. 7. ConcIusion: the Labor of Perception
To engineer vision also meant to ensure perception in the shortest possible time. Here as well, the avant-garde promoted montage as an example oI possible economy, in this case economy oI time. Maud Lavin describes the 1930 maniIesto oI the group oI leading German designers headed by Kurt Schwitters: "Walter Dexel writes that modern man has the right to expect communications in the shortest possible time. Willi Baumeister points out that photomontage is eIIicient, allowing Ior the quick grasp oI several images at once...Similarities between photomontage and Iilm are oIten emphasized, with photomontage being considered a quicker, more eIIicient medium." 332
Finally, to engineer vision also meant to be able to measure its eIIiciency, or, to use the language oI a communication engineer, to measure "system perIormance." Eisenstein, Iresh Irom engineering school, invented his Iirst theory oI artistic communication, the Iamous "montage oI attractions": "...Laboratory analyses and diagrams...Mendeleev's Periodic Table and the laws oI Gay-Lussac and Boyle Mariotte in the realm oI art!...Let us thereIore search Ior the unit which will measure the inIluence exerted by art! Science has its "ions," its "electrons," its neutrons." Art will have -- attractions!" 333
In its desire to engineer vision, the avant-garde was ahead oI its time. The systematic engineering oI vision took place only aIter World War II when vision become the major instrument oI labor, the most productive organ oI a worker in a human-machine system. To ensure the maximum perIormance oI such a system, it became necessary to engineer it around the capacities and the limitations oI human vision. It also became necessary to understand vision in a new way: as inIormation processing.
332 Maud Lavin, "Photomontage, Mass Culture, and Modernity. Utopianism in the Circle oI New Advertising Designers," in Montage and Modern LiIe: 1919-1942, ed. Matthew Teitelbaum (Cambridge: The MIT Press, 1992), 54. 333 Qtd. in Jacques Aumont, Montage Eisenstein (London and Bloomington: BFI Publishing and Indiana University Press, 1987), 41. Emphasis mine -- L.M.
This is the history oI the engineering oI vision Irom INKhUK to MIT: Irom the Soviet art institute oI the 1920s which gathered the avant-garde artists united by the idea oI engineering vision -- to the premier engineering school where much oI this engineering later occurred through the work on inIormation theory, on radar and computer graphics, on new human- machine interIaces, on algorithms oI human vision.
This dissertation begins at the scene oI heated debates at the Soviet art institutes in the early 1920s. It ends with the recent research by cognitive scientists and engineers at MIT. What justiIies this temporal continuity, on the one hand, and disciplinary shiIt, on the other hand -- Irom the Soviet artistic avant-garde to one oI the premier science institutions in the world, which today is in the IoreIront oI research on robotics, artiIicial intelligence, and human-computer interIaces? What can be in common between Soviet constructivism and American cognitive science? Both are the product oI the age oI engineering. Engineering, the quintessentially modern proIession, appeared in the 1870s. The job oI an engineer was to ensure maximum perIormance - - oI machines, human bodies or communication systems -- with minimum investment oI energy, materials, and time. The engineer was the new specialist in the eIIiciency oI any productive process, regardless oI its nature. And this ideology oI "engineerism" became the modern myth, the religion oI a society aspiring to total rationalization. 334 In order to employ vision eIIiciently in its new roles, whether as a medium oI propaganda and advertising or as the channel oI communication between human and machine, artists, designers, and scientists adopted the engineering approach as well. It is not surprising that Soviet constructivists, the pioneers oI modern propaganda design, insisted on calling themselves engineers rather than artists. It is also not surprising that the archeology oI twentieth century vision would uncover numerous similar solutions to the problem oI its rationalization in such seemingly distant Iields as Iilm and robotics, graphic design and experimental psychology. For instance, let us compare two statements: one by a designer, the other by a psychologist. El Lissitsky wrote in the 1920s regarding his new geometric style oI poster design: "The most unambiguous and immediately
334 Max Weber proposed rationalization as an umbrella concept to describe the social processes oI modernity. See, Ior instance, From Max Weber: Essays in Sociology (New York: OxIord University Press, 1958). Max Weber, General Economic History (New York, 1927). ConcIusion
recognizable Iorms are geometric Iorms. No one will conIuse a rectangle with a circle, or a circle with a triangle." 335 In 1951, the already mentioned experimental psychologist Paul Fitts echoed practically the same words in his summary oI research on the use oI graphic symbols Ior human- machine interIaces: "Geometrical Iigures diIIer considerably in legibility. Straight lines have been Iound to be more legible than curved ones. The triangle, the rectangle, and the square have been reported to be more easily recognized under conditions oI low illumination and in peripheral vision than circular and hexagonal Iorms." 336 Both Lissitsky and Fitts were concerned with the same problem -- reliability and speed in the recognition oI visual symbols -- and, thereIore, it is logical that they recommended the same solution. Whether Iilmmakers or psychologists, engineers or interIace designers, twentieth century proIessionals have understood vision as a medium oI communication and have attempted to put its use on a scientiIic basis. II in the early part oI the century this research Iocused on human communication (the human as the subject oI mass propaganda and mass entertainment), aIter World War II, the Iocus shiIted to human-machine interIace. Human vision became the key instrument oI post-industrial labor as the channel oI communication between human and machine. This history can be summarized by three images. The Iirst image: a portrait oI Tatlin by Lissitsky (Iig. 16). A compass, extending straight Irom Tatlin's eye, a metaphor oI vision Ior work. The second image, a popular icon in the late 1980s and early 1990s, promoted virtual reality interIace designed at NASA/Ames Human Factors Research Center (Iig. 22). 337 Instead
335 Qtd. in Igor Golomstock, Totalitarian Art (New York: HarperCollins Publishers, 1990), 24. 336 Paul Fitts, "Engineering Psychology and Equipment Design," in Handbook oI Experimental Psychology, ed. S.S. Stevens (New York and London: John Wiley & Sons, Inc., 1951), 1298. 337 On NASA/Ames virtual reality research in the 1980s, see Scott S. Fisher, "Virtual InterIace Environments," in The Art oI Human-Computer InterIace Design, ed. Brenda Laurel (Reading, Mass.: Addison-Wesley Publishing Company, 1990): 423-438.
oI the metaphor oI the eye-compass, a reality: video monitors strapped to the eyes. The notion oI vision as work is now Iully realized: the operator wearing the gear works by mentaly processing visually presented inIormation. The gear is designed using all the available knowledge accumulated by experimental psychology about human vision. In the photograph we see the last leItover Irom the age oI manual labor -- an arm in a DataGlove. It will soon disappear since through gaze tracking the operator can control the system by merely looking at diIIerent points in virtual space. What is the third image which stands between these two? Let me quote Irom the description oI the history oI engineering psychology Iound in an 1965 overview oI the Iield:
The primary emphasis in time-and-motion engineering has been on man as a worker; that is, as a source oI mechanical power. It was not until World War II that a new category oI machines appeared -- machines that made demands not upon the operator's muscular power, but upon his sensory, perceptual, judgmental, and decision-making abilities. The job oI a radar operator, Ior example, requires virtually no muscular eIIort, but makes severe demands on sensory capacity, vigilance, and decision-making ability. This new class oI machines raised some intricate and unusual questions about human abilities: How much inIormation can a man absorb Irom a radar screen? 338
The World War II radar operator is this third image, a historical intermediary between Lissitsky's man-compass (1920s) and NASA's virtual reality user (1980s). The Iigure oI a radar operator stands at the gates to the post-industrial society oI perceptual labor, at the origins oI research on human-machine interIace, and the understanding oI human vision as inIormation processing (Iig. 23). "I am a mechanical eye," wrote Dziga Vertov in 1923. 339
Allison, Nicholas H., ed. Art Into LiIe: Russian Constructivism 1914-1932. New York: Rizzoli International Publications, 1990.
Anderson, John Robert. Cognitive Psychology and Its Implications. W.H. Freeman and Company, 1980.
ArgYelles, JosZ. Charles Henry and the Formation oI a Psychophysical Aesthetics. Chicago: University oI Chicago Press, 1972.
Arnheim, RudolI. "A Plea Ior Visual Thinking." In New Essays on the Psychology oI Art. Los Angeles and London: University oI CaliIornia Press, 1986.
Arnheim, RudolI. Art and Visual Perception. The New Version. Berkeley: University oI CaliIornia Press, 1974.
Attneave, F. Applications oI InIormation Theory to Psychology. New York: Holt, 1959.
Aumont, Jacques. Montage Eisenstein. London and Bloomington: BFI Publishing and Indiana University Press, 1987.
Barthes, Roland. "Rhetoric oI the Image." In Image, Music, Text, edited by Stephen Heath. New York: Hill and Wang, 1977.
Barthes, Roland. "The Photographic Message." In Image, Music, Text, edited by Stephen Heath. New York: Hill and Wang, 1977.
Barthes, Roland. "The Third Meaning." In Image, Music, Text, edited by Stephen Heath. New York: Hill and Wang, 1977.
Baxandall, Michael. Painting and Experience in FiIteenth Century Italy. OxIord: OxIord University Press, 1972.
Beliajeva, E.L. Arkhitekturno-prostranstvennaya sreda soroda kak obyekt zritelnogo vospriyatiya (Architectural and spacial city environment as an object oI visual perception). Moscow: Stroyizdat, 1977.
Benjamin, Walter. "On Some Motives in Baudelaire." In Illuminations, edited by Hannah Arendt. New York: Schochen Books, 1969.
Bergin, Thomas Goddard and Max Harold Fisch. The New Science oI Giambattista Vico. Ithaca: Cornell University Press, 1968.
Berlyne, D.E. "Novelty, Complexity, and Hedonic Value." Perception and Psychophysics 8 (1970): 279-86.
Borch-Jakobson, M. The Absolute Master. StanIord: StanIord University Press, 1991.
Braun, Marta. Picturing Time: the Work oI Etienne-Jules Marey (1830-1904). Chicago: The University oI Chicago Press, 1992.
Broadbent, D.E. Perception and Communication. OxIord: Pergamon Press, 1958.
Bryson, Norman. Looking at the Overlooked: Four Essays on Still LiIe Painting. Cambridge: Harvard University Press, 1990.
Buck-Morss, Susan. The Dialectics oI Seeing: Walter Benjamin and the Arcades Project. Cambridge: The MIT Press, 1989.
Bulgakova, Olga. "Sergei Eisenshtein i ego 'psikhologicheskiy Berlin' -- mezhdu psikhoanalizom i strukturnoy psikhologiey" (Sergei Eisenstein and his 'psychological Berlin': between psychoanalysis and structural psychology). Kinovedcheskie Zapiski 2 (1988): 174-191.
Chapanis, Alphonse, Wendell R. Garner, and CliIIord T. Morgan. Applied Experimental Psychology. New York: John Wiley & Sons, Inc., 1949.
Chen, Jack. Soviet Art and Artists. London: The Pilot Press, Ltd., 1945.
Cherry, Colin. On Human Communication. 2nd ed. Cambridge: The MIT Press, 1968.
Cohen, Gillian. The Psychology oI Cognition. 2nd ed. (London and New York: Academic Press, 1983).
Cohen, Paul R. and Edward A. Feigenbaum, eds. The Handbook oI ArtiIicial Intelligence. Los Atlos, CA: William KauImann, Inc., 1982.
Cook, Richard. "Serious Entertainment." Computer Graphics World (May 1992): 40-48.
Crary, Jonathan. Techniques oI the Observer: on Vision and Modernity in the Nineteenth Century. Cambridge: The MIT Press, 1990.
Damisch, Hubert. L'origine de la perspective. Paris: Flammarion, 1987.
Danziger, Kurt. Constructing the Subject: Historical Origins oI Psychological Research. Cambridge: Cambridge University Press, 1990.
De Fleur, Melvin L. and Sandra Ball-Rokeach. Theories oI Mass Communication. New York: Longman, 1975.
De Landa, Manuel. War in the Age oI Intelligent Machines. New York: Zone Books, 1991.
Deleule, Didier. "The Living Machine: Psychology as Organology." In Incorporations, edited by Jonathan Crary and SanIord Kwinter. New York: Zone Books, 1992.
Descartes, RenZ. "Meditations on the First Philosophy," In The Rationalists. Garden City, NY: Doubleday & Company, 1960.
DorIman, D. and H. McKenna. "Pattern PreIerence as a Function oI Pattern Uncertainty." Canadian Journal oI Psychology 20 (1966): 143-53.
Druckrey, Timothy. "Revenge oI the Nerds. An Interview with Jaron Lanier." AIterimage (May 1991): 5-9.
Dudley, Andrew. Concepts in Film Theory. OxIord: OxIord University Press, 1984.
Dulac, Germaine. "Aesthetics, Obstacles, Integral CinZgraphie." In vol. 1, French Film Theory and Criticism, edited by Richard Abel. Princeton: University oI Princeton Press, 1988.
Eames, Charles and Ray Eames. A Computer Perspective: Background to the Computer Age. Cambridge: Harvard University Press, 1990.
Echoes oI War. Boston: WGBH Boston, n.d. Videotape.
Edgerton, Samuel. The Renaissance Rediscovery oI Linear Perspective. New York: Basic Books, 1975.
Eduardo, Kac. "Aspects oI the Aesthetics oI Telecommunications." In SIGGRAPH '92 Visual Proceedings, edited by John Grimes and Gray Lorig. New York: The Association Ior Computing Machinery, 1992.
Edwards, Paul, ed. Encyclopedia oI Philosophy. New York: Macmillan, 1967.
Eisenstein, Sergei. "A Dialectical Approach to Film Form." In Film Form: Essays in Film Theory, edited by Jay Leyda. New York: Harcourt Brace and World, 1949.
Epstein, Jean. "On Certain Characteristics oI PhotogZnie." In vol. 1, French Film Theory and Criticism, edited by Richard Abel. Princeton: University oI Princeton Press, 1988).
Estes, William. "Experimental Psychology: an Overview." In The First Century oI Experimental Psychology, edited by Eliot Hearst. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1979.
Farocki, Harun. "Reality Would Have to Begin." Documents 1/2 (1992): 136-146.
Finke, Ronald A. Principles oI Mental Imagery. Cambridge: The MIT Press, 1989.
Fisher, Scott S. "Virtual InterIace Environments." In The Art oI Human-Computer InterIace Design, edited by Brenda Laurel. Reading, Mass.: Addison-Wesley Publishing Company, 1990.
Fitts, Paul. "Engineering Psychology and Equipment Design." In Handbook oI Experimental Psychology, edited by S.S. Stevens. New York and London: John Wiley & Sons, Inc., 1951.
Flew, Antony, ed. A Dictionary oI Philosophy. London: Pan Books Ltd., 1984.
Florensky, Pavel. "Obratnaya Perspektiva" (The Inverted Perspective). In vol. 1, Sobranie Sochineniy, ed. N.A. Struve. Paris: YMCA-PRESS, 1985.
Florensky, P.A. "Symbolarium (Slovar simvolov). Predislovie" |Symbolarium (The dictionary oI symbols). Introduction|. Trudy po znakovym sistemam V. Tartu: Uchebnye zapiski Tarturskogo Universiteta 284, 1971.
Foster, Hal, ed., Vision and Visuality. Seattle: Bay Press, 1988.
Freud, Sigmund. The Interpretation oI Dreams. New York: Avon Books, 1965.
Freud, Sigmund. Standard Edition oI the Complete Psychological Works. London: Hogarth Press, 1953.
FriedhoII, Richard Mark and William Benson. The Second Computer Revolution: Visualization. W.H. Freeman and Company, 1991.
Gabor, D. "A Summary oI Communication Theory." In Proceedings oI a Symposium on Applications oI Communication Theory, London, 1952, edited by Willis Jackson. London: Butterworth ScientiIic Publications, 1953.
Gardner, Martin. Logical Machines and Diagrams. 2nd ed. Chicago: The University oI Chicago Press, 1982.
Golomstock, Igor. Totalitarian Art. New York: HarperCollins Publishers, 1990.
Gombrich, E.H. "Pictorial Instructions." In Images and Understanding, ed. Barlow Horace, Colin Blakemore and Miranda Weston-Smith. Cambridge: Cambridge University Press, 1990.
Gombrich, E.H. Art and Illusion. Princeton: Princeton University Press, 1960.
Goodman, Nelson. Languages oI Art. Indianapolis: Hackett, 1976.
Hall, Stuart. "Encoding/Decoding." In Culture, Media, Language, ed. Stuart Hall et al. Hatchinson, 1980.
Hansen, Miriam. "The Hieroglyph and the Whore: D.W. GriIIith's Intolerance." The South Atlantic Quarterly 88, no. 2 (1989): 361-392.
Hartley, R.V.L. "Transmission oI InIormation." Bell System Tech. Journal 7 (1928): 535-563.
Hearst, Eliot. "One Hundred Years: Themes and Perspectives." In The First Century oI Experimental Psychology, edited by Eliot Hearst. Hillsdale, NJ: Lawrence Erlbaum Associates, 1979.
Heidegger, Martin. What is Called Thinking? New York: Harper & Row, Publishers, 1968.
Heron, W. "The Pathology oI Boredom." ScientiIic American 196 (1957): 52-56.
Hochberg, Julian. "Sensation and Perception." In The First Century oI Experimental Psychology, edited by Eliot Hearst. Hillsdale, NJ: Lawrence Erlbaum Associates, 1979.
Holly, Michael Ann. PanoIsky and the Foundations oI Art History. Ithaca: Cornell University Press, 1984.
Holmes, Oliver Wendel. "The Stereoscope and the Stereograph." In Photography, Essays and Images: Illustrated Readings in the History oI Photography, edited by Beaumont Newhall. New York: Museum oI Modern Art, 1980.
Hoover, Theodore and Charles Fish. The Engineering ProIession. StanIord: StanIord University Press, 1941.
"Institut Khudozestvennoy Kultury" (The Institute oI Artistic Culture). Russkoe Iskusstvo 2-3 (1923): 85-88.
Ivins, William M. On the Rationalization oI Sight. New York: Da Capo Press, Inc., 1975.
Jakobson, Roman. "Closing Statement: Linguistics and Poetics." In Semiotics. An Introductory Anthology, edited by Robert Innis. Bloomington: Indiana University Press, 1985.
Jakobson, Roman. "Metalanguage as a Linguistic Problem," Presidential address delivered at the Annual Meeting oI the Linguistic Society oI America, 1956. In The Framework oI Language. Ann Arbour: Michigan Studies in the Humanities, 1980.
Jay, Martin. Downcast Eyes: the Denigration oI Vision in Twentieth-century French Thought. Berkeley: The University oI CaliIornia Press, 1993.
Jay, Martin. "Scopic Regimes oI Modernity." In Vision and Visuality, ed. Hal Foster. Seattle: Bay Press, 1988.
Johnson-Laird, Philip. Mental Models: Towards a Cognitive Science oI Language, InIerence, and Consciousness. Cambridge: Cambridge University Press, 1983.
Joravsky, David. Russian Psychology: A Critical History. Cambridge, Mass.: Basil Blackwell, 1989.
Kandinsky, Wassily. Point and Line to Plane. New York: Solomon R. Guggenheim Foundation, 1947.
Kandinsky, Wassily. "Programma raboty Instituta khudozhestvennoy kultury" (The program oI the Institute oI Artistic Culture). In Sovetskoe iskusstvo za 15 let. Materialy i dokumentatsiia (FiIteen years oI Soviet art: materials and documentation), ed I. Matsa (Moscow, IZOGIZ: 1933).
Kantowitz, Barry H. and Robert D. Sorkin. Human Factors: Understanding People-System Relationships. John Wiley & Sons: 1983.
Keller, Evelyn Fox and Christine R. Grontkowski. "The Mind's Eye." In Discovering Reality: Feminist Perspectives on Epistemology, Metaphysics, Methodology, and Philosopohy oI Science, edited by Sandra Harding and Merrill B. Hintikka. London: D. Reidel Publishing Company.
Kemp, Martin. The Science oI Art. New Haven: Yale University Press, 1990.
Kenez, Peter. The Birth oI the Propaganda State. Soviet Methods oI Mass Mobilization, 1917- 1929. Cambridge: Cambridge University Press, 1985.
Khan-Magomedov, S.O. "INKhUK: vozniknovenie, Iormirovanie i pervyy period raboty. 1920" (INKhUK: appearance, Iormation and Iirst period oI its work, 1920). Sovetskoe Iskusstvoznanie (1981): 332-368.
Koerner, Joseph Leo. "The Shock oI the View." The New Republic (April 26, 1993): 32-38.
Kogan, P.C. "O zadachakh akademii i yeyo zhurnala" (About the goals oI the academy and its journal). Iskusstvo 1, no. 1 (1923).
Kovasznay, L.S.G. and H.M. Joseph. "Image Processing." Proceedings oI IRE 43 (1955): 560- 570.
Kozulin, Alex. Psychology in Utopia: Toward a Social History oI Soviet Psychology. Cambridge: The MIT Press, 1984.
Krauss, Rosalind. The Optical Unconscious. Cambridge: The MIT Press, 1993.
Lacan, Jacques. "On the Gaze as Objet Petit a." In The Four Fundamental Concepts oI Psycho- Analysis, edited by Jacques-Alain Miller. New York: W.W. Norton & Company, 1981.
LakoII, George. "The Invariance Hypothesis: Is Abstract Reason Based on Image-Schemas?" Cognitive Linguistics 1, no. 1 (1990): 39-74.
LakoII, George. "Cognitive Linguistics." Versus 44/45 (1986): 119-154.
Latour, Bruno. "Visualization and Cognition: Thinking with Eyes and Hands." Knowledge and Society: Studies in the Sociology oI Culture Past and Present 6 (1986): 1-40.
Lavin, Maud. "Photomontage, Mass Culture, and Modernity. Utopianism in the Circle oI New Advertising Designers." In Montage and Modern LiIe: 1919-1942, edited by Matthew Teitelbaum. Cambridge: The MIT Press, 1992.
LeIebvre, Henri. The Production oI Space. OxIord: Blackwell Publishers, 1991.
LZvy-Bruhl, Lucien. Les Iunctions mentales dans les sociZtZs primitives. Paris: 1910.
Locke, John. An Essay Concerning Human Understanding. Edited by A.S. Pringle-Pattison. OxIord: Clarendon Press, 1924.
Lotman, Jurij. The Structure oI the Artistic Text. Ann Arbour: Department oI Slavic Languages and Literatures, The University oI Michigan, 1977.
Lowe, David. Three-Dimensional Object Recognition Irom Single Two-Dimensional Images. Robotics Report 62. New York: Courant Institute oI Mathematical Sciences, New York University, 1986.
Mandler, Jean Matter and George Mandler, eds. Thinking: From Association to Gestalt. New York: John Wiley & Sons, Inc., 1964.
Manovich, Lev. "'Real' Wars: Esthetics and ProIessionalism in Computer Animation." Design Issues 6, no. 1 (Fall 1991): 18-25.
Marr, David. Vision. New York: W.H. Freeman and Company, 1982.
Matsa, I., ed. Sovetskoe iskusstvo za 15 let. Materialy i dokumentatsiia (FiIteen years oI Soviet art: materials and documentation). Moscow, IZOGIZ: 1933.
McFarlane, M. D. "Digital Pictures FiIty Years Ago." Proc. IEEE 60, no. 7 (1972).
McGraw-Hill Encyclopedia oI Science & Technology: an International ReIerence Work in Twenty Volumes Including Index. New York: McGraw-Hill, 1992.
Michelson, Annette. "Reading Eisenstein Reading 'Capital'." October 2 (1976): 27-38; October 3 (1977): 82-88.
Miller, G.A. Language and Communication. New York: McGraw-Hill Book Co., Inc., 1951.
Mitchell, J. William. The ReconIigured Eye: Visual Truth in the Post-Photographic Era. Cambridge, The MIT Press, 1992.
Mitchell, W.J.T. Iconology. Image, Text, Ideology. Chicago: The University oI Chicago Press, 1985.
Moholy-Nagy, Lszl. "Photography in Advertising." In Photography in the Modern Era, edited by Christopher Phillips. New York: Aperture, 1989.
Moholy-Nagy, Lszl. The New Vision. New York: Wittenborn, 1947.
Lszl Moholy-Nagy. Painting, Photography, Film. 1925. Reprint. Cambridge: The MIT Press, 1973.
Molok, Yu. A. "'Slovar simvolov' Pavla Florenskogo. Nekotorye margonalii" (Pavel Florensky's 'dictionary oI symbols.' A Iew margins). Sovetskoe Iskusstvoznanie 26 (1990): 322-343.
MYnsterberg, Hugo. The Photoplay: A Psychological Study. New York: D. Aplleton & Co., 1916.
Newhall, Beaumont. Airborne Camera. New York: Hastings House, Publishers, 1969.
Nietzsche, Friedrich. "On Truth and Falsity in their Ultramoral Sense." In Early Greek Philosophy and Other Essays. New York: The Macmillan Company, 1924.
Noble, Douglas. "Mental Materiel: The Militarization oI Learning and Intelligence in U.S. Education." In Cyborg Worlds: the Military InIormation Society, edited by Les Levidov and Kevin Robins. London: Free Association Books, 1989.
Nsth, WinIried. Handbook oI Semiotics. Bloomington and Indianapolis: Indiana University Press, 1990.
PanoIsky, Erwin. Perspective as Symbolic Form. New York: Zone Books, 1991.
Peat, David. ArtiIicial Intelligence: How Machines Think. New York: Baen Enterprises, 1985.
Pinker, Steven. "Visual Cognition: An Introduction." Cognition 18 (1984): 1-63.
Posner, Michael I. and Gordon L. Shulman. "Cognitive Science." In The First Century oI Experimental Psychology, edited by Eliot Hearst. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1979.
Posner, Michael I. "Section V: InIormation Processing. Overview." In vol. 2, Handbook oI Perception and Human PerIormance, edited by Kenneth BoII, Lloyd KauIman, and James P. Thomas. New York: John Wiley and Sons, 1986.
Quastler, H., ed. InIormation Theory in Psychology: Problems and Methods. Glencoe, IL.: Free Press, 1955.
Rabinbach, Anson. The Human Motor: Energy, Fatigue, and the Origins oI Modernity. Basic Books, Inc., 1990.
Rappoport, A.G. "El Lissitsky i ego 'Pun-geometiya'" (El Lissitsky and his 'Pan-geometry'). Sovetskoe Iskusstvoznanie 25 (1989): 113-130.
Reichardt, Jasia. The Computer in Art. London and New York: Studio Vista and Van Nostrand Reinhold Company, 1971.
Resnikov, Howard. The Illusion oI Reality. New York: Springer-Verlag New York Inc., 1989.
Reveaux, Tony. "Virtual Reality Gets Real." New Media (January 1993): 36-41.
Rheingold, Howard. Virtual Reality. New York: Simon & Schuster, Inc., 1991.
Riggs, L.A. et al. "The Disappearance oI Steadily Fixated Visual Tests Objects." J. Optical Soc. America 43 (1953): 495.
Roberts, L.G. "Machine perception oI three-dimensional solids." In Optical and Electo Optical InIormation Processing, edited by J.T. Tippett. Cambridge: The MIT Press, 1965.
Roberts, L.G. Homogeneous Matrix Representations and Manipulation oI N-Dimensional Constructs. MIT Lincoln Laboratory MS 1405, 1965.
Roberts, L.G. Machine Perception oI Three-Dimensional Solids. MIT Lincoln Laboratory TR 315, 1963.
Rose, Jacqueline. Sexuality in the Field oI Vision. London: Verso, 1986.
Saint-Martin, Fernande. Semiotics oI Visual Language. Bloomington and Indianapolis: 1990.
Sampson, GeoIIrey. Schools oI Linguistics. StanIord, CA: StanIord University Press, 1980.
Sardarov, A.S. "5000 let evolutsii doroznogo znaka" (5000 years oI the evolution oI a traIIic sign). Tekhnicheskaya Estetika 9 (1984): 14-19.
Schramm, Wilbur. "How Communication Works." In The Process and EIIects oI Mass Communication. Urbana: University oI Illinois Press: 1954.
Sekula, Allan. "The Instrumental Image: Steichen at War." In Photography against the Grain: Essays and Photo Works, 1973-1983. HaliIax: The Press oI the Nova Scotia College oI Art and Design, 1984.
Sekula, Allan. "The Body and the Archive." October 39 (1987): 3-64.
Shannon, Claude E. and Warren Weaver. The Mathematical Theory oI Communication. Urbana: The University oI Illinois Press, 1949.
Shepard, S. and J. Metzler. "Mental Rotations oI Three-dimensional Objects." Science 171 (1971): 701-703.
SIGGRAPH '89 Panel Proceedings. New York: The Association Ior Computing Machinery, 1989.
SIGGRAPH '92 Final Program. New York: The Association Ior Computing Machinery, 1992.
Sonesson, Gsran. Pictorial Concepts. Inquiries into the Semiotic Heritage and its Relevance Ior the Analysis oI the Visual World. Lund, Sweden: Lund University Press, 1989.
Spriegel, William R. and Clark E. Myers, eds. The Writings oI the Gilbreths. Homewood, IL., 1953.
Stelzer, Otto. "Moholy-Nagy and His Vision." In Lszl Moholy-Nagy, Painting, Photography, Film. 1925. Reprint, Cambridge: The MIT Press, 1973.
Taylor, Brandon. Art and Literature under the Bolsheviks. London: Pluto Press, 1991.
Taylor, Franklin. "Psychology at the Naval Research Laboratory." American Psychologist 2, no. 3 (1947): 87-92.
Taylor, Frederick Winslow. The Principles oI ScientiIic Managment. New York, 1967.
Titchener, Edward BradIord. A Beginner's Psychology. New York: The Macmillan Company, 1915.
Todorov, Tzvetan. Theories oI the Symbol. Ithaca: Cornell University Press, 1984.
Tretyakov, Sergei. "Otkuda i kuda? (Perspektivu Iuturizma)" |Where Irom and where to? (Perspectives oI Iuturism)|. LEF 1 (1923): 192-203.
Vasilieva, Yu. A. "Eisenstein: Kontseptsija 'Agressivnogo Iskusstva'" (Eisenstein: the concept oI 'aggressive art'). Kinovedcheskije Zapiski 3 (1989), 205-209.
Vitz, Paul C. and Arnold B. Glimcher. Modern Art and Modern Science. The Parallel Analysis oI Vision. New York: Praeger, 1984.
Vygotsky, Lev. Psikhologija Iskusstva (Psychology oI art). Moscow, 1968.
Weber, Max. From Max Weber: Essays in Sociology. New York: OxIord University Press, 1958.
Weber, Samuel. The Legend oI Freud. Minneapolis: University oI Minnesota Press, 1982.
Wiener, Norbert. The Human Use oI Human Beings. Cybernetics and Society. New York: Avon Books, 1967.
Wiener, Norbert. Cybernetics. OI Control and Communication in the Animal and the Machine. New York: John Wiley & Sons, Inc., 1948.
FIGURES
Figure 1. Humbert de Duperville's "Synoptic Table," 1827-32. Figure 2. Charles Henry's aesthetic protractor, 1988. Figure 3. Diagram illustrating Georges Seurat's theory oI expression. Figure 4. The exercises developed by Rodchenko Ior his course on graphics at VKhUTEMAS, 1921-22. Top: examples oI basic Iorms given by Rodchenko. Bottom: example oI a student work. Figure 5. Example oI computer generated visualization. Figure 6. Representations oI propositions as Euler's diagrams, an earlier Iorm oI Venn's diagrams. Max MYller, The Science oI Thought (London: Longmans, Green, and Co., 1887), 543. Figure 7. Composite photography. Frontispiece Irom Francis Galton, Inquiries into Human Faculty (London: Macmillan, 1883). Figure 8. Freud's diagram illustrating the relation between two triads oI concepts: id, ego, and superego, and unconscious, preconscious, and conscious. Figure 9. Top: Sigmund Freud's schematic picture oI sexuality. Bottom: example oI a modern diagram. Figure 10. Example oI diagrammatic graphic language. Figure 11. Figure Irom Monge's GZomZtrie descriptive, 1799. Figure 12. Two draughtsmen plotting points Ior the drawing oI a lute in Ioreshortening, Irom Durer's Underweysung, 1525. Figure 13. WireIrame computer graphics by Boeing, early 1960s. Figure 14. ACRONYM computer vision system processes an aerial photograph, late 1970s. Figure 15. Laser range Iinder, early 1980s. Figure 16. El Lissitsky. Tatlin at Work. Photomontage. Illustration Ior Ilia Erenburg's Six Tales with Easy Endings. 1922. Figure 17. Wireless telegraphed photography. Figure 18. Schematic diagram oI a general communication system. Figure 19. Example oI a stimulus used in experiments on preIerence as a Iunction oI the amount oI inIormation. Figure 20. One oI the earliest models oI a human as inIormation processing system. Figure 21. The model oI a human as inIormation processor which summarizes the available data (by 1986) about human sensory, cognitive and motor perIormance. Figure 22. NASA Ames Virtual Reality Environment Workstation, late 1980s. Figure 23. SAGE (the "Semi-Automatic Ground Environment") -- the Iirst human-machine interactive display system, mid 1950s.