The Sensory Necessity for
Ornament.
Department of Mathematics, University of Texas at San
Antonio, San Antonio, Texas 78249, USA.
E-mail: salingar@sphere.math.utsa.edu
Ornament is a necessary component of any architecture that aims to connect to human beings. The suppression of ornament, on the other hand, results in alien forms that generate physiological and psychological distress. Early twentieth-century architects proposed major stylistic changes -- now universally adopted -- without having any idea of how the human eye/brain system works.
- Introduction
- Visual meaning
- How the eye scans a picture
- Neurophysiology of the eye/brain system
- Visual ordering and patterns
- Color and intelligence
- The necessity for ornament
- Ornament and writing
- Conclusion
1. Introduction.
This paper argues that ornament is necessary for us to experience architectural form in a positive way. An earlier article (Salingaros, 2000) presented mathematical reasons for why ornament is necessary. The visual coherence of a complex form as defined by systems theory requires ordered substructure on all scales: from the overall size of the building, down to the detailed grain in the materials at 1mm. Natural structures have this (fractal) property. If a man-made form lacks ordered structure on one or more obvious scales, it is perceived by human beings as being visually incoherent, and consequently as alien to our conception of the world. A building's substructure on the range of scales from 1m to 1mm is usually achieved via traditional ornament, demonstrating its necessity.
Here I wish to discuss how our neurophysiology requires that visual input from our surroundings contain many of the characteristics of traditional ornament. Our visual and mental make-up is linked through human evolutionary processes to the informational richness of the environment. This biological background will help to explain some aspects of why human beings create ornament. I present several rules derived from our cognitive apparatus -- these rules are intended to help understand how we conceive a form as being visually coherent. I then discuss the relationship between the cognitive rules and the creation of ornament.
I propose two arguments against minimalist and random design of built forms. One is that they cause anxiety and physiological distress, because they inhibit human beings from connecting mentally with a given structure through meaningful information. This point will be discussed in detail. The other reason is the resemblance between a minimalist built environment, and the perception of a normal, visually complex environment by persons with a damaged perceptual apparatus. I shall indicate how different types of injury to the eye and brain result in precisely the same effects offered by either minimalist or intentionally disordered design.
2. Visual meaning.
A form's visual organization communicates information to people through the surfaces and geometry it presents. Environmental experience involves an intimate interaction of human beings with surfaces and spaces, which influences our emotions and physiological state and consequently our actions. A building's exterior and interior surfaces either connect in an emotionally positive manner with the user; remain neutral by having no effect; or act in a negative fashion so as to repel. This connection resides in the information content of space and its transitions. Even though surface qualities are usually assumed to be separate from the spatial geometry in a building, they are in fact interdependent, and both contribute to how people respond to their surroundings.
Traditional architecture uses this mechanism to establish a positive connection. In the twentieth century this mechanism was neglected so as to focus on pure geometrical form. Nevertheless, the emotional link established between people and built structures in many cases leads us, through feedback, to produce those structures. Human emotional response is based on neurophysiology and information input. An environment lacking in texture, color, and ornament can be punishing for a human being, as exemplified in the design of prisons throughout history. Going to the other extreme, an environment that is supercharged in uncoordinated visual stimuli -- such as the Las Vegas strip lit up by neon lights -- exceeds the visual input that can be consistently tolerated.
We seek meaning from our environment and are repelled by environments that convey no meaning, either because they lack visual information, or because the information present is disorganized (Klinger and Salingaros, 2000). Information has driven our evolutionary development: human vision and intelligence increased our capacity for processing information. The eye and the brain form a single mechanism. Design is itself a product of human vision and intelligence, therefore the complexity of traditional designs satisfies cognitive needs of the human brain. This observation makes the underlying reasons of why we build complex things less mysterious. People are motivated to build so as to extend their consciousness to a wider domain outside their own mind.
3. How the eye scans a picture.
In classic experiments on human eye motion while scanning a picture (Hubel, 1988; Noton and Stark, 1971; Yarbus, 1967), the eye is observed to focus most of the time in the regions of a picture that have the most detail, differentiations, contrast, and curvature (the experiments referred to did not include color). These are clearly the high-information regions in the picture. Eye fixations establish a fairly narrow "scan path" where the eye spends about one-third of its time, with random excursions to low-information (i.e. plain) regions of a visual. The brain thus selects informative details such as convoluted, detailed contours and contrasting edges for recognizing and remembering an object. Our visual system is built to select those items of concentrated information that can provide the most complete response in the shortest possible time.
The information content of visuals lies precisely where the eye spends its energy scanning; the rest of the picture is easily reconstructed by extrapolation (Nicolis, 1991). This is how the brain stores information via a compression algorithm. Selective weighting of information to minimize coding also provides the basis for information storage in artificial systems such as computer graphics (Klinger and Salingaros, 2000). We comprehend an object by seeking to define its boundaries and any characteristic internal details. Note, however, that a discontinuity or sharp interface such as occurs when two edges come together has no width or dimension, and so does not provide any detail. A precise straight edge has no information. The more complex a curve is, the more information it contains.
Detail is nothing more than contrast on the smallest scales. In theory, therefore, contrast (coupled with hierarchy) includes detail. Nevertheless, as the universality of hierarchical scales (Klinger and Salingaros, 2000) is not so widely known, I will continue to refer to contrast and detail as separate concepts.
There is another reason why our eye/brain system has evolved to prefer detail, and that is our capacity to predict future events (Llinás, 2002). An intelligent, mobile animal focuses on details that give crucial information about an adversary during combat; about changing physical conditions crucial to survival; the recognition of familiar animals and their facial expression; visual cues from a prey being hunted; etc. All of this information comes directly from telling details. Our cognitive system normally has no time to process all available visual information in those instances, and has to rely on first input in order to make almost instantaneous decisions.
Recent work (VanRullen and Thorpe, 2002) reveals that a first, rough image is created using only the salient parts of a retinal image -- that is, regions of high contrast and detail. In this first wave of signals, it is contrast that encodes sufficient information to make a decision on our need to respond. The rapidity of this first image, which is entirely subconscious, is faster than our motor response time. This is an absolutely necessary feature, which makes possible a rapid response to any potential threat while the full image is still being processed. Our "instinctive" response to form is therefore based on contrast and detail. Further information from the retina is processed more slowly, and any kind of rational analysis of the form can begin only after that is completed.
The above considerations suggest two cognitive rules on how we perceive our world. I propose that artificial structures ought to in general follow such rules, precisely because our perceptual apparatus has evolved to use them. The understanding is that our cognitive mechanism implies analogous rules for constructing the man-made world.
Rule 1. Every structure must have some subregion with a high degree of contrast, detail, and curvature.
Rule 2. Plain surfaces require either their interior, or their borders, to be defined through contrast and detail.
Large, plain objects or surfaces disturb the observer by presenting no information -- the most disturbing being surfaces of glass or mirrors that prevent the eye from even focusing on them. We instantly look for reference points, either in a form's interior, or at its edge. We need to comprehend a structure as quickly as possible, to make sure that it poses no threat to us. Large uniform regions with abrupt, ill-defined boundaries generate physiological distress as the instrument (namely, the eye/brain system) seeks visual information that isn't there, thus frustrating our cognitive process.
Rule 2 makes sense in the context of message transmission. In sending a message, it is necessary to indicate its limits. For example, a one-dimensional piece of information needs to be identified as such by noting where it begins and ends. This requires additional coding for the message's boundaries. Without those boundaries, the receiver has no idea of what it is receiving, and cannot distinguish a message from other portions of a signal. In ordinary writing, a sentence begins with a capital letter and ends with a period.
4. Neurophysiology of the eye/brain system.
Starting from a light-sensitive spot on protozoans and primitive worms capable of judging direction, the primitive eye developed a sense for various degrees of light intensity so as to perceive distance, or the shadow of an aggressor. Finer and finer tuning corresponds to an increase in information channels. Researchers believe that the brain developed concurrently with the eye in order to handle the increasingly complex optical information input from the evolving eye. A startling proof of this co-evolution lies in the left/right reversal of functions in the two brain hemispheres, corresponding to the reversal of an optical image on the retina (Fischler and Firschein, 1987).
Uniformity is decoupled from our neurophysiology, because a majority of cells in both the retina and visual cortex will not fire in response to a uniform field (Hubel, 1988; Zeki, 1993). Visual receptors in the retina (either single cells, or groups of cells) compare the characteristics of adjacent regions -- they spatially differentiate the signal. Color wavelength is determined by comparing the output from three different types of cone cells due to the response from a single point. Neurobiologists have identified specialized neurons and clusters of neurons that perceive angles, curvature, and contrast (Hubel, 1988). The latter work via lateral inhibition (i.e. signal comparison) and are successfully simulated in artificial (computational) visual systems to achieve edge detection (Fischler and Firschein, 1987).
Particular brain cells, and some groups of cells, have a preference for all possible oblique orientations in addition to vertical and horizontal. The directional preference of successive cells in a cortical region distinguishes between angles of 10 to 20 degrees (Hubel, 1988; Zeki, 1993). The existence of orientation-specific cells in the visual cortex proves the importance of angular information. In addition, "end-stopped" cells in the visual cortex respond to lines of a distinct orientation up to some maximum length, beyond which the response drops to zero. End-stopped cells are biological receptors that are directly sensitive to corners, curvature, and to discontinuities in lines.
All these anatomical details of our perceptual apparatus support Rules 1 and 2. Our eye/brain system evolved to perform a very specific function, and it doesn't make sense for us to present it with structures that frustrate what it is supposed to do. Minimalist surfaces and edges violate the way human beings process information. When we go against our neurophysiology for whatever reason, then our body reacts with physical and psychological distress. Such effects are measurable, and include raised blood pressure, raised level of adrenaline, raised skin temperature, contraction of the pupils -- all symptoms of triggering our defensive mechanisms against a threat.
Degradation of our ability to see fine detail signals the onset of different pathologies of the eye itself rather than the brain. The first group of problems occur with the lens -- either the lens can no longer focus, or it becomes opaque due to a cataract. The second group of problems have to do with the retina; in particular, with the macula, the central region of the retina where cone cells that are responsible for seeing fine detail and color are concentrated. The retina can be damaged by detachment, or the macula can degenerate because of inadequate blood flow. The loss of visual information cuts us off from our environment, and creates anxiety by lowering our ability to respond to it.
5. Visual ordering and patterns.
Rules 1 and 2 explain the necessity of visual information. Now we turn to the opposite problem: the case when there exists too much information. The first two rules are by themselves not sufficient to explain the geometry of form, since they say nothing about how visual information may be ordered. We know very well, however, that our cognitive system craves ordered information and is overloaded with disordered (i.e. random) information. Ordering via patterns is discussed in (Klinger and Salingaros, 2000), where we propose a complexity index to measure visual coherence. This leads to additional rules that govern how visual information can be organized. The easiest way to group information is along some curve.
Rule 3. Visual information can be ordered via linear continuity.
This corresponds to the simplest possible expedient of lining up high-contrast objects on end; not necessarily in a straight line, but on some sort of curve. The units do not need to repeat in Rule 3. What this lining-up does is to significantly narrow the scan path that the eye needs to follow in order to grasp the information encoded in the components, since now there are fewer excursions to farther regions. Lining-up corresponds to a condensation of information, yet there exist other techniques of organizing information spatially without condensing it along a line.
The alternative is to organize the high-information units using symmetry, which leads to patterns in space. A further savings of effort is accomplished in visual compression, by repeating a similar unit. Repetition can give rise to the wide range of traditional symmetries, such as reflectional, rotational, translational, and glide symmetries (Washburn and Crowe, 1988). High-contrast objects on the small scale can be arranged in a symmetrical pattern, and the smaller units made similar to cut down the total information.
Rule 4. Symmetries and patterns organize visual information without significantly increasing the computational overhead.
A well-defined unit that is repeated does not need to be processed by our mind each time anew. We apparently have the means to recognize similarity very easily, so the eye/brain system can encode a pattern in terms of one or more basic units, plus their positional distribution. If the units are repeated in some symmetric fashion -- i.e., the units' positions are themselves symmetric -- then only a little additional information is needed to specify the pattern. For this reason, patterns tend to be preferred over a random distribution of repeated units. In the absence of any symmetry or ordering, our eye/brain system has to compute the position of each unit separately, which increases effort and comprehension time.
Organization endows structural information with meaning, which in turn connects that object with the human mind without the need for conscious reflection. Here is where hierarchy, the topic of (Salingaros, 2000), comes into play in an essential manner. A symmetric arrangement of units is perceived on a higher level of scale than the units themselves. As soon as one starts to do this, then recursion can be applied to define increasingly higher levels of scale, with each coherent arrangement on a particular level being very easily comprehended.
Readers might also note a relationship between the cognitive rules and the Gestalt laws (Fischler and Firschein, 1987), as follows. Rule 3 relates to "Proximity" and "Good Continuation", while Rule 4 relates to "Similarity", "Closure", and "Symmetry".
Failure to perceive patterns indicates a pathology of the brain; in particular, the failure of different specialized regions and mechanisms that process visual information to integrate their functions (Zeki, 1993). Specific causes of such disintegration include Carbon Monoxide poisoning and cerebral lesions due to strokes. In what is known as "visual agnosia", a person perceives detail but cannot integrate this information to recognize an overall form. This could be manifested as an inability to recognize objects or faces. Such persons can see but cannot understand their environment, and the trauma makes them anywhere from mildly to severely dysfunctional.
6. Color and intelligence.
Color vision represents a significant information increase over monochromatic vision found in otherwise intelligent animals such as dogs and cats. The sensation of color resides just as much in the computational part of the brain as it does in the optical mechanism of the eye (Hubel, 1988; Zeki, 1993). This is shown by "color constancy", which is the ability of the eye-brain system to adjust a biased color illumination and reconstruct a faithful color image. In the experiments of Edwin Land, a color painting or collage illuminated by red, green, and blue lights together appears the same regardless of the relative intensities of the three different lamps used for illumination. Color photographs of an object under different lights, however, look very different.
Neurological studies indicate that intelligence evolved to support the perceptual process occurring in the human brain. Color perception takes place in the most evolutionary developed region of the brain's cortex. Positron Emission Tomography (PET) can measure the varying blood flow to those cortical regions, which correlates with the level of neuronal activity corresponding to the eye's sensation of color. Blood flow to the region of the brain responsible for color vision increases by three-fold when subjects first view a picture only in shades of gray, then again in full color (Zeki, 1993). This corresponds exactly to what one would expect from an increase in information due to adding the three color dimensions.
Three different types of cone cells are needed in order to perceive color hue or wavelength, and to distinguish color intensity from white (colorless) (Hubel, 1988). Interestingly, the cone cells in the retina responsible for color vision are also responsible for our ability to see fine detail (Hubel, 1988), thus linking color with geometry in our perceptual apparatus. Contrary to what is frequently assumed, therefore, color and linear design are intimately related. This leads us to the final rule.
Rule 5. Color is a necessary component of our environment.
Three arguments support this claim: first, the existence of our highly-developed color sensitivity; second, the neurophysiological coupling between our ability to see detail -- something that is necessary for our survival -- and our ability to see color; third, psychological experiments demonstrating how colors affect us profoundly. Not only does color have the ability to change our mood (with the greater pleasure offered by the more saturated hues); it can also directly affect our physiological state (Mehrabian, 1976). Finding the appropriate color, however, is a very difficult problem, which will not be treated here. A significant portion of the world's economy -- that driven by the advertising and fashion industries -- is based on the connection between human beings and color.
Loss of color occurs in a pathology known as "cerebral achromatopsia" (Zeki, 1993). Cortical lesions in the specific region of the brain responsible for color vision destroy the ability to see in color, usually as a result of a stroke. Alternatively, transient achromatopsia can be caused by inadequate blood supply to this region. As a consequence, the world is seen entirely in shades of grey, but the ability to distinguish detail is not affected. Patients who are stricken with this condition describe their surroundings as "drab" and "depressing", and frequently live lives of despair after their injury (Zeki, 1993). Organic objects (such as foods and person's faces) are now repellent, since a gray coloration is normally associated with decay and death.
7. The necessity for ornament.
In order to satisfy the five rules given above, buildings should either have a continuous swath of high-density visual structure that the eye can follow in traversing their overall form, or focal points of intense detail and contrast arranged in the middle or at the corners. These could include a thick border or edge of the building; a thick boundary around openings and discontinuities; concentrated structure in the centers or corners of walls; etc. The visually-intense structure should organize information via patterns and symmetries. Color can appear throughout the structure, and can help to define the visually-intense regions. Our neurophysiology thus resurrects an element of architecture that was arbitrarily condemned a century ago -- that is, ornament (Alexander, 2001; Bloomer, 2000).
This conclusion invalidates a basic assumption of twentieth-century architecture: that a building could be conceived in an abstract design space unrelated to human beings. People actively seek perceptual connection with their physical environment to satisfy a fundamental physiological need. This is consistent with the view of buildings and people forming a unified, interacting system (Alexander, 2001). Buildings do not exist in isolation from nature; the complexity of natural structures establishes a lower threshold value for information. This threshold is part of us. A building is successful or not after it is erected, for many different reasons. In addition to its strictly utilitarian aspects, liking a building depends on establishing visual and tactile connections with it.
Ornament is an indispensable part of this connection, but we have forgotten how ornament is generated. Since we no longer think about ornament as an integral part of architecture, most ornament created today fails to register. Its detail is too small or indistinct, its differentiations are too faint or excessively subtle, or its components are random. Successful ornamentation requires the recursive capacity of only the most highly-developed brains, those of human beings. Different types of recursion include rhythm and repetition that generate translational and rotational symmetries; the iteration of structure on smaller and smaller scales that generates fractal patterns; and iteration on the same scale that generates denser and denser connections. The human capacity for spoken and written language is in fact made possible by our capacity for recursive logical thought.
8. Ornament and writing.
Ornament presents organized information that is entirely distinct from text as encoded in letters and signs. Ornament does not communicate a message in written language, but instead in a subconscious language. I will use the example of typography to discuss this difference. When early typeface fonts for printing were cut by hand, they were created with the aim of having maximal legibility, guided by aesthetic considerations. They were serif fonts (in which open lines end with a dot or T-stroke), like present-day Times and Garamond.
The introduction of radically new typefaces at the beginning of the twentieth century confirms that removing the ornamental serifs also removes a level of meaning. Sans-serif fonts such as Helvetica were popularized along with the modernist Bauhaus design style. They were promoted for their mathematical simplicity. It has been experimentally established that sans-serif fonts degrade legibility. People's reaction to these stripped-down typefaces was strongly negative; so much so that the ancestral sans-serif font was called "grotesque" by the Berthold foundry, which introduced it commercially (the sans-serif typeface Berthold Akzidenz-Grotesk eventually gave rise to Helvetica).
The transition from sans-serif to serif fonts shows clearly how ornament works to make form clearer, sharper, hence more distinguishable. Classic serif fonts go much further in establishing a positive emotional connection with the reader. In (Salingaros, 2000) I argue for the necessity of detail from hierarchical arguments. It is not just any added detail that improves the legibility of the font, however. Adding dots or small cross-strokes anywhere other than at the terminals of open lines (and even there, at some arbitrary angle) would degrade the font.
Ornament organizes detail in a very precise and sophisticated fashion in order to make a larger form more comprehensible. Adjustments are necessary for a better comprehension of letters. The most effective serif fonts are vastly more complex mathematically than a similar sans-serif font. They show substructure on a hierarchy of decreasing scales. A serif typeface doesn't simply add end-strokes; the entire font is adjusted so that new, more detailed elements cooperate to define a coherent whole. Correcting an old misunderstanding, ornamentation does not superimpose unrelated structure; rather it is an operation that generates highly-organized internal complexity. It therefore has to be extremely precise in order to be effective.
9. Conclusion.
This paper reviewed results from neurobiology and experimental psychology, which together provide evidence of an informational connection between people and structures. Visual information input helps to create a physiological state in the user, triggered by the design of the environment. The quality of information and its organization affects the emotional connection that human beings establish with forms and surfaces. Traditional architecture uses the interaction between human beings and environmental information to connect people with a building. Detail, differentiations, curvature, and color appear necessary in at least some part of a building, implying that ornament is vital after all. Without it, buildings tend to be perceived as having alien qualities.
References.
Alexander, Christopher (2001). The Phenomenon of Life, Oxford University Press, New York. [The Nature of Order, Book One]
Bloomer, Kent (2000). The Nature of Ornament, W. W. Norton, New York.
Fischler, Martin A. and Firschein, Oscar (1987). Intelligence: The Eye, the Brain, and the Computer, Addison-Wesley, Reading, Massachusetts.
Hubel, David H. (1988). Eye, Brain, and Vision, Scientific American Library, New York.
Klinger, Allen and Salingaros, Nikos A. (2000). "A Pattern Measure" in Environment and Planning B: Planning and Design, vol. 27, pp. 537-547.
Llinás, Rodolfo (2002). I of the Vortex, MIT Press, Cambridge, Massachusetts.
Mehrabian, Albert (1976). Public Places and Private Spaces, Basic Books, New York.
Nicolis, John S. (1991). Chaos and Information Processing, World Scientific, Singapore.
Noton, David and Stark, Lawrence (1971). "Eye Movements and Visual Perception" in Scientific American, vol. 224, No. 6, June pp. 35-43. [Reprinted in: Image, Object and Illusion, Richard Held, Editor, Scientific American, Freeman, San Francisco, 1974, pp. 113-122]
Salingaros, Nikos A. (2000). "Hierarchical Cooperation in Architecture, and the Mathematical Necessity for Ornament" in Journal of Architectural and Planning Research, vol. 17, pp. 221-235.
VanRullen, Rufin and Thorpe, Simon J. (2002). "Perception, decision, attention visuelles: ce que les potentiels evoques nous apprennent sur le fonctionnement du systeme visuel", in: L'Imagerie Fonctionnelle Electrique et Magnetique: Ses Applications en Sciences Cognitives, Edited by: B. Renault, Hermes, Paris, to appear.
Washburn, Dorothy K. and Crowe, Donald W. (1988). Symmetries of Culture, University of Washington Press, Seattle.
Yarbus, Alfred L. (1967). Eye Movements and Vision, Plenum Press, New York.
Zeki, Semir (1993). A Vision of the Brain, Blackwell Scientific Publications, Oxford.