I have stated that perception is achieved by the processing of absorbed energy gradients whose attributes are causally connected to the attributes of objects by virtue of physical laws determining the interactions of objects and different forms of energy. An understanding of the physiological basis of perception requires that we identify:
(1) How energy gradients are obtained by the active organism and applied to the energy sensitive receptor tissue, and
(2) How the object-specifying attributes of energy gradients are isolated and segregated by the sensory system at the physiological level. {17}
A. The Sampling and Display of Energy Structure.
In general there are two ways to obtain energy gradients necessary for object perception: The receptor may, by direct contact, pick up the energy gradients existing at the surfaces of objects. I will call this the pick-up of contact gradients. The second major category of energy gradient pick-up consists of the receptor interception of patterns of ambient energy traveling through the medium between the organism and the objects composing its environment. This latter method is referred to as the pick-up of ambient energy gradients.
Contact Gradient Pick-up
The mode of obtaining contact gradients that is almost universal for perceiving organisms is by means of mechanical probing and sweeping. Generally some appendage of the organism containing receptors sensitive to mechanical deformation is applied to the object, either by a series of mechanical contacts and withdrawals, i.e., a series of probing actions, or by a drawing of the sensory appendage across the object while maintaining constant contact. The object probed does not necessarily have to be brought into direct contact with the organism’s body. Man, for instance, can probe objects by means of a stick or cane. In all cases of mechanical probing and sweeping, the source of the mechanical energy responsible for the changing pressure patterns is provided by the organism’s own muscular efforts. These gradients are, therefore, organism energized, and can be contrasted with externally energized gradients.
The pressure patterns can be displayed over time, as in the case of probing with antennae or with a stick or cane; or the pressure patterns can be displayed as spatially distributed gradients, as in the case of probing {18} with one’s hand or having an object brought into cutaneous contact with the body.
In either case, contact gradients possess the same sequential order as the objects boundaries and margins that cause them. For example, in Figure 1, note that margin M2 is fixed between margin M1 and margin M3. As a probe is drawn continuously over the surface of this irregularly shaped object, the temporal order of the “deflections” registered at the organism’s end of the probe will always correspond to the spatial order of the margins (or depressions”) regardless of the length of the probe or of the angle at which the probe is applied. If the sweep of the probe is reversed in direction, the order of the encountered margins is correspondingly reversed, and so, therefore, is the temporal order of the registered deflections.
The same facts of correspondent ordering hold, of course, regardless of whether the probe is a mechanical sensor, a thermal sensor, a chemical sensor, or a photo sensor.
Furthermore, when the object is swept by multiple probes, such as by drawing ones fingertips over the object or by drawing the object over ones body surface, all of the stimulus “edges” caused by the object’s margins move across the skin together and at the same rate.
The Pick-Up of Ambient Contact Gradients
Ambient energy refers to the energy waves converging upon every point in an environment from multiple sources in the environment. The sources of ambient energy are either primary (energy emitters) or secondary (energy reflectors and transmitters). Furthermore, ambient gradients may be either organism energized, such as in the case of the high-frequency sound emissions of bats and porpoises, or externally energized, such as the light emitted by the sun, or environmentally produced sound, which are the sources most widely available to organisms living above ground or in the water. {19}
For purposes of our analysis, forms of wave energy may be arranged along a continuum, according to the degree to which each form is or is not diffracted by the edges of objects. In general, the property of wave energy is dependent upon wavelength (or frequency). Short wavelength (high frequency) wave energy, such as light, is negligibly diffracted by the edges of objects, and is said, therefore, to travel in “straight lines.” Long wavelength (low frequency) wave energy, such as sound, is strongly diffracted by, and therefore easily bends around, the edges of objects.
Figure 1.
The Pick-up of Contact Gradients. The spatial order at the object surface is registered as a corresponding temporal order of deflection at the organism’s end of the anatomical probe. (See text page 18.)
Angular position of the anatomical probe registers relative direction to each remote object boundary or margin. (See text page 26.) {20}
The Pick-Up of Ambient Light Gradients
As a consequence of the fact that light travels in straight lines, the object boundaries across which the absorption or reflection of radiant energy changes abruptly, produces energy margins , which are carried away from the object in straight lines. These margins can be picked up and displayed onto a receptor surface or a neural array by an organism stationed at a point in the environment exposed to the ambient radiant energy. Gibson (1966) calls such a point in the environment available to the organism for the potential pickup of ambient energy: a station point.
A necessary condition for displaying the energy margins, which define the optical structure at a station point, is an anatomical structure capable of registering intensity or frequency change as a function of directional differences or changes in receptor orientation. The chambered eye of the vertebrate and the compound eye of the insect are two anatomical structures subserving the function of displaying ambient optical structure obtained at a station point. {21}
In the chambered eye, the energy margins composing the retinal “image” are directionality-specific registrations of the spatial relationships of the projected object boundaries, which are their remote cause. Unlike contact gradients, ambient gradients produce margins whose order does not, in itself, correspond to the spatial order of the object boundaries which are their remote cause. Movement of the eye does not, furthermore, yield uniform movement of optical margins. Optical margins move at different rates (relative to one another). The differences in rates of margin movement are caused by and specify the relative change in directionality from each object boundary to each successively sampled station point. Coincidence of margins specifies coincidence of direction; .to each corresponding object boundary. Furthermore, the rate of directional change, which causes the relative rate of margin movement, specifies differences in object boundary distances. These latter differences specify spatial layout of object to object and of the object surfaces and parts to one another.
Figure 2.
The Pick-Up of Ambient Light Gradients. As the chambered eye in the circle) moves from position P1 to P2, the displayed margins A”, B”, C”, D” are displaced across the retina at different rates. These margins are the product of object boundaries structuring ambient light. (The rectangular shapes O1 and O2 represent remote objects.) (See text page 22.){22}
I have said that the function of the eye is to display optical energy structure defined in terms of differences in intensity and frequency values in different directions. I have also said that the so-called retinal “image” represents a spatial display of this optical structure. But optical structure can be picked-up and registered without producing an “image;” the compound eye of the insect is a case in point. This version of an eye captures energy differences in different directions without yielding an image. This is accomplished by means of an array of optical “funnels”, called omatidia, each pointed in a different direction. The array of omatidia are arranged so that their proximal ends, if extended, would converge at a common point at the center of the insect’s eye, and {23} so that their distal ends collectively compose the spherical external surface of the insect’s eye. The compound eye is a dramatic illustration of the fact that the retinal image as such is, as Gibson first identified, totally incidental to the essence of visual perception. The retinal image represents merely one way of displaying the changing or flowing optical gradients sampled at a station point, and does not represent a mapping of the rigid structure “of objects “imaged.”
The Pick-Up of Ambient Sound Gradients
The pick-up and display of sound patterns differs from the pick-up and display of optical patterns by virtue of the fact that sound bends easily around edges of objects, as a consequence of which, sound travels with almost equal energy in all directions from an object edge or boundary. The boundaries of objects do not produce sharp acoustical margins. This precludes the specification of object boundaries in terms of acoustical gradients (at least for the frequency range of humanly audible sound, viz. 2OHZ to 20,000Hz). A further implication of the absence of sharp acoustical margins is that the directionality of a remote acoustical event is not specifiable at a single station point. On the other hand, unlike light, sound travels at a relatively low velocity (about 750mph in air). This velocity is low enough to permit neural tissue to differentially respond to the difference in arrival time of the same sound event registered at two (or more) simultaneously occupied station points. Of course, the intensity difference of the sound event can also be registered at the simultaneously occupied station points. The two ears, operating as an integrated sensory organ, register the frequency spectrum of the sound as well as the differences in intensity and latency of the remote sound event at the station points occupied by each of {24} the two ears. The inner ear converts a temporally distributed energy pattern (at the oval window) into a spatially distributed pattern or gradient (across the tympanic membrane). The time, frequency, and energy dimensions of the acoustical event we displayed in terms of a moving gradient along the basilar membrane.
If, with the head rotating, the interaural latency differences and interaural intensity differences are jointly considered, the following specifications obtain:
1. Unchanging interaural latency differences plus unchanging inter; aural intensity differences specifies a sound source inside or on the head.
2. Changing interaural latency differences or changing interaural intensity differences specifies a sound source outside of and away from the head.
3. The rates at which self-movement correlated interaural latency differences and interaural intensity differences change, specify both the relative directions and the relative distances from the head of multiple sound sources: (a) the greater the rate of change with head rotation, the nearer the sound source; (b) as the head rotates toward the direction of the sound the intensity of the sound at the leading ear decreases and at the trailing ear increases, while the latency of the sound at the leading ear increases and at the trailing ear decreases; (c) under these conditions of head rotation, when the interaural latency and intensity differences reach zero, the head is facing in the direction of the sound source.
Ultrasonic Pick-Up and Display. Bats, porpoises, whales, and various species of birds and rodents employ self-emitted high-frequency sounds to aid them in navigating about their environments. Foremost among such {25} creatures is the bat, some species of which can emit ultrasonic pulses in the frequency range of 450,000hz (Milne &’Milne, 1962). At such frequency ranges, sound behaves somewhat like light, traveling in almost straight lines producing reasonably sharp acoustical margins at the edges and corners of objects. Ultrasonic energy, therefore, is capable of carrying the specifications of object boundaries. By means of ultrasonic sound alone, bats and porpoises are able to differentiate objects having different shapes or different textures (Van Bergeijk, et al.,1960; Stevens & Warshofsky, 1965).
The organ for ultrasonic pick-up is the ear, which register’s object boundary specification in terms of changing acoustical patterns that are sampled over time by rotating the head or moving the body through air or water.
The performances of bats and porpoises prove that there is nothing in the nature of sound or auditory display, per se, which precludes object perception by means of sound energy picked up by the ears.
Order as the Essence of Display.
The receptor apparatus in general is a device for “capturing” and displaying the pattern or order contained in changing energy gradients. Since it is the order that is fundamental, and not the form in which the order is carried, the sensory system may capture patterns carried as temporal order in an energy event, and display these patterns in terms of spatial adjacencies (as occurs at the basilar membrane). Or, spatial order in an energy pattern may be displayed in terms of temporal adjacencies (as-in optical, tactile, and acoustical scanning). The capacity to “trade space for time” (Gibson, 1966) seems to be an inherent property of sensory systems. {26}
Specification of Space
All spatial attributes of objects, including: (a) distance from object to organism, (b) inter-object distances, (c) spatial form of objects, (d) environmental layout, and (e) the layout of the organism’s own body parts, are specified at the receptor in terms of directionality differences associated with different components of-the registered energy patterns. This principle of spatial specification by registered differences in directionality cuts across all sense modalities. We have already noted how directionality differences specify spatial attributes for ambient energy pick-up. The same principle applies to the pick-up of contact gradients. In the case of contact pick-up, directionality is registered in terms of changes and rates of change in the angular positions of the organism’s receptor appendages relative to one another and relative to the organism’s body as the organism probes or sweeps the object (see Figure 1).
Space/Object Specification and Self-Initiated Movement.
The specification of spatial attributes and, therefore, the specification of objects depends upon the pick-up of changes in energy directionality as the organism moves his body or body parts through space. Object specification requires active, self-initiated movement as well as active self-initiated resistance to movement. This requires that the organism be, for the most part, free from subjection to passive motion imposed on the organism by its environment, which means that the organism must possess a body of sufficient mass and design to resist forces exerted by normal turbulence and flow of its natural medium, be it air or water. As Flores (1973) properly identified:
Moving with external pressures precludes awareness of them. Awareness requires resistance, the ability not to yield, a disparity between the body’s own movements and the movements in the space around it. (p. 124) {27}
Since the pick-up of object-specific information requires that organisms be capable of moving parts of their bodies, either organs or appendages, while maintaining their place in the environment, as well as be capable of moving their bodies through their environment, organism’s require a contractile or muscular system. And, because the organism’s motor actions must be functionally integrated with and regulated by the energy sensitive sensory systems, the motor system and the sensory system must jointly act through the nervous system. The nervous system, therefore, must be capable of being affected by energy gradients, on the one hand, and be capable of effecting motor action and control, on the other hand. In short, the nervous system must be the organism’s sensory-motor integrator.
In this section, I have discussed the problem of how, in principle, object-specifying energy gradients are picked-up or intercepted by a receptor apparatus, and how the specifications of objects are displayed at the receptor’s energy sensitive tissue. The existence of object-specifying energy gradients, and the interception and display of these gradients are necessary conditions for the perception of objects. But the fact that the specifications of objects are displayed at the receptor does not explain how these specifications are physiologically isolated. This is the problem to be discussed in the next section.
B. Physiological Isolation and Segregation of Object-Specific Energy Attributes.
Physiological Explanation in Psychology
Most of our current knowledge of the physiology of perception consists of inferences from psychological data. This is the necessary and proper approach to the study of {28} perceptual physiology, for there is no other way to determine what constitutes a physiological mechanism then to first determine what function is to be achieved. A physiological unit is a functional unit. What the physiologist must look for is the mechanism explaining the functional achievement. One cannot build a psychology out of physiological “primaries”, for without prior knowledge of the principles ruling the functional end, there can be no real physiology. For this reason, physiology is not an independent discipline within biology. The physiologist cannot work alone. There are no principles of physiology, as such. There are only principles [of the physiology of metabolism, of respiration, or reproduction, of growth, of immunology, of aging, etc. This is equally true for the physiology of perception. As Robert Efron concludes in his paper entitled “What Is Perception”:
The phenomena of consciousness must be understood before one can hope to ‘explain’ them in terms of neural action. The attempt by many neurophysiologists to reverse this order — to study the neural mechanisms underlying perception prior to any adequate definition or conceptualization of perception — is doomed to failure. (Efron 1969, p. 171)
Perception and Psychophysics
It was pointed out, in an earlier section, that the psychological experiences of brightness, loudness, color, pitch, hardness, etc. are the category of conscious forms in which we appreciate something about the energy-dependent attributes of objects. This category of sensory qualities has been the central concern of a sub-discipline of psychology, called psychophysics. The aim of the psychophysicist is to quantify the capacity of the sensory system to differentially respond to energy variables absorbed at the receptors, and to discover the laws by which energy differences underlie differences in the various dimensions of {29} our sensory experiences.
Psychophysics has established, for instance, that for stimuli of sufficiently long duration holding all other factors constant, and changing only the intensity of a visible light source produces changes in experienced “brightness” that are in the same direction as the intensity changes. Increasing intensity produces increasing brightness; decreasing intensity produces decreasing brightness. Similar relationships were found to hold for other sensory qualities as well, such as for example, between increasing sound pressure and heard loudness; increasing temperature and felt warmth; increasing sugar concentration in solution and tasted sweetness; and so on. Furthermore, psychophysics has established that continuous qualitative changes in the applied energy produce continuous qualitative changes in the corresponding sensory experience. For example, a sequence of discrete pulses of light of sufficient duration and intra-pulse separation, arranged in sequential order of increasing wavelength produces a sequence of sensory experiences of discrete pulses of color, which can be arranged in the same order in terms of the experiential dimension of “similarity of hue.” Thus, if the subject were asked to arrange the colored lights in the order of their perceived similarity, the order would be the same as the physicist’s order of the lights according to increasing wavelength. Similar relationships were found to hold between acoustical frequency and experienced pitch. From these observations it is clear that the sensory system yields separate experiences for separate forms of energy, and that these experiences possess dimensions which parallel the dimensions of the registered energy form. Differences in energy quality, such as frequency, produce {30} differences in sensory quality, such as color. We can conclude, therefore, that our sensory qualities are the forms in which certain attributes of energy patterns are isolated and consciously retained. To understand the relevance of these facts for perception, it is necessary to identify: (1) What are the isolated features of energy patterns which determine the at- tributes of our sensory qualities? and (2) How does the isolation of these particular energy features make possible our perception of objects? In other words, what is the functional value, for object perception, of the physiological principles underlying the isolation of energy attributes to yield sensory qualities?
The Sensory Registration of Energy Features
In order to answer our first question, it is not enough to know, for example, that brightness is the conscious form in which we appreciated the intensity of a luminous source. For this fact does not tell us what it is the intensity of a luminous source that we appreciate when we are aware of its brightness. To answer this question, we must know how intensity is registered as brightness. {31}
Figure 3.
Sensor Registration of Energy Magnitudes: Diagram of the three basic ways to register energy magnitude: Registration in terms of Absolute local magnitudes; Registration in terms of Differences in local magnitudes; and Registration in terms of Ratios of local magnitudes. (See text page 32.)
In general, there are three basic ways to register energy attributes: “Absolute” registration, “Difference” registration, and “Ratio” registration. (See Figure 3.) An absolute register is one which always yields the same indication or output for the same absolute energy and whose output is directly proportionate to the absolute magnitude of the incident energy.
A difference register is one which yields the same indication or output for the same energy difference between any two points and whose output is directly proportionate to the difference in magnitude between the incident energies at the two points of absorption. {32}
A ratio register is one which yields the same out-put for the same ratio of energies between any two points and whose output is proportionate to the ratio of the energy magnitudes at the two points of absorption. The primary limiting factor for all energy registers is the amount of energy they can tolerate without damage or overload. The energy absorbing elements can be protected from energy overload by means of a mechanism that automatically attenuates intense energy before it reaches the absorbing elements or before its effects reach the registering mechanism. Any attenuating mechanism, however, must change the indicator readings obtained from absolute energy responders and from difference responders.
In Figure 3, this fact is diagrammatically illustrated. Notice that when two energy sources, a and b are attenuated so as to reach the absorbing elements with energy values reduced to a’ and b’, absolute energy responders and difference energy responders will produce different responses to unattenuated as compared to attenuated conditions. But a ratio responder will not alter its response. For the ratio responder, both conditions produce the same output. Thus, ratio responders are the only category of energy registers that allow for overall energy attenuation without affecting their output. Of course, ratio responders pay for this advantage by being unable to provide information relating to the absolute energy or to energy differences at their receptor elements.
Let us turn now to the psychological data in order to determine what category of energy response is represented by our sensory experiences. Observations of sensory experiences produced by single stimuli will not suffice to identify how energy magnitudes are registered by our senses. We must use multiple stimuli and determine the experiential effects of each of these {33} stimuli when they are presented simultaneously to the observer.
Psychophysical data employing multiple stimuli strongly suggest that our sensory systems are largely ratio responders. Consider, for example, the classic demonstrations of Gelb (reported in Ellis, 1939), in which a flat black disk is suspended in the doorway connecting two darkened rooms. The disk is illuminated by a white light source not visible to the observer seated in one of the rooms facing the black disk. The observer reports seeing a bright white disk against the dark background of the adjoining room. A small white sheet of paper is then brought into the light and placed against the black disk. Instantly, the subject reports that the disk has now turned black and the slip of paper is seen as white. This experiment demonstrates that our perception of objects as light or dark does not represent a sensory registration of local absolute energy, for our perception of the lightness of the same local region can change when no energy change occurs in that region. Our perceptions of the brightnesses of objects jointly depends in some way upon the energy reaching us from the object as well as the energy reaching us from surrounding objects or background. The now classic experiments of Wallach (1948) are the result of an attempt to identify the exact relationship between stimulus and surround, which determines the perceived brightness of each. In these experiments, two disks and two surrounds were used. The intensities of the light at either of the disks or the surrounds were independently adjustable. When the first disk/surround display, which served as the standard, was adjusted to any disk illumination value, and different surround illumination value, and with the surround illumination of the test {34} display set at some value, Its different from ISS, subjects, when asked to match the test disk to the standard disk by adjusting the illumination of the former, consistently adjusted the illumination of the test disk so that Isd/Iss = ltd/Its (approximately). Wallach concludes from these experiments on what he calls “Brightness Constancy and Achromatic Colors,” that the colors, which come about under these circumstances, depend in close approximation on the ratios of the intensities involved and seem independent of the absolute intensity of local stimulation. The region of higher intensity will assume the color white and that of lower intensity will show a gray (or black), which depends on the intensity ratio of the two regions. (p. 241)
A similar dependency between the light composition of an object or visual region and the composition of surrounding regions is indicated by the well-known phenomenon of “simultaneous color contrast.” This phenomenon reveals that the perception of a color is, at least partially, determined by the wavelength composition of the surround or background. As an example of this phenomenon, if a red and a white circular spot of light are projected onto a screen so that the two spots partially overlap, the region of overlap will be perceived as white, the physically white I crescent, i.e., the crescent receiving light from the white projector only, will be perceived as blue-green, and the red crescent will be perceived as red. If a blue spot replaces the red one, the perceptions of the three areas will be: a) white, where the spots intersect, b) yellow where only the physically white light reaches the screen, and c) blue where only the blue light reaches the screen. In general, the white light crescent region is perceived as complementary is color to that of the other projector, with the overlap region being perceived as white. Furthermore, the saturation of the perceived color in the region receiving only the white light is {35} proportionate to the intensity of the light reaching the screen from the colored projector, and is independent of changes in overall intensity.
The experiments of Edwin Land (1959, 1965) on color perception suggest that perceived color depends, at least in part, upon the joint effects of lightness scaling, in terms of ratio of local energy to background energy and to overall illumination, carried out by two or more receptor systems, which are each selectively responsive to different (but overlapping) regions of the visible spectrum. Land has demonstrated, for example, that mixing white light with monochromatic light such that there is produced a pattern of different ratios or combinations of white and monochromatic mixtures across the displayed projection, will yield the perception of a wide array of colors, most of which cannot be obtained by the standard procedures of isolated color mixtures of the white and monochromatic light. Furthermore, the colors obtained by Land remained stable when the overall energy from the projectors was increased or decreased. Changing overall illumination has no effect on the local frequency ratios.
Ratio Response and Object Perception
In answer to the second of the questions raised at the beginning of this section, the value of ratio response is that energy ratios tell us more about objects than do energy differences. These latter, on the other hand, tell us more about energy than they do about objects. To understand what this means, we must recall that an object is a spatially bounded, cohering, segregated “thing.” The way in which we perceive an object is by registering the fact that the reaction to energy at the region in space occupied by the object is different from the reaction to energy of the region surrounding the object. The boundaries of the object define the regions across which the reactions to {36} energy are different. These object boundaries cause energy margins in ambient light (Gibson, 1966) and these energy margins are displayed as lines of demarcation defining the patterns in the energy gradients distributed across the retina.
It is the structural differences between objects, which accounts for their respective reactions to energy. These structural differences are not caused by the energy; they are energy-independent. As I have said earlier, and object’s structure, shape, and spatial relationship to other objects are all energy-independent object attributes.
Bearing these facts in mind, we can now see why a sensory system that registers the difference in the degree to which an object reflects or absorbs energy as compared to the degree to which the background reflects or absorbs energy, independent of the actual intensity of the overall incident energy, is superior to a sensory system that registers the differences in the degree of energy reflected or absorbed by the object as compared to the reflection or absorption by the background. For the former (ratio responsive) system, the object boundaries will be registered equally well, i.e., with equal sharpness, under conditions of both intense and weak illumination. For the latter (difference responsive) system, object boundaries are registered with decreasing sharpness as illumination is decreased. Ratio thresholds, operative in the former system, allow objects boundaries to be registered under a much wider range of prevailing energy conditions than is allowed under systems ruled by difference thresholds, operative in the latter system.
Subthreshold and suprathreshold energy ratios, being independent of incident energy intensity, tell us something about objects themselves. {37} Subthreshold and suprathreshold energy differences, being energy-dependent, may tell us more about the prevailing energy level than about the objects. For a sensory system operating by difference response, object boundaries will alternately appear and disappear as overall illumination increases and decreases. But, for sensory systems operating in terms of ratio response, object boundaries remain highly stable regardless of changes in illumination. For this reason, ratio response is preferred over difference response for the appreciation of object boundaries. Conversely, difference response is preferred over ratio response for the appreciation of energy boundaries.
Since optimal registration of object boundaries is essential for optimal object perception, for vision, lightness scaling in terms of energy ratios is the best way to optimize the registration of object boundaries.
The Value and Significance of Color Constancy.
“Chromatic and achromatic lightness scaling are made possible by virtue of the energy-dependent properties of objects, viz., their surface reflectancies and pigmentations. But the value of lightness scaling lies in the fact that it permits the registration of energy-independent object attributes, viz., object boundaries. The fact that energy-dependent attributes of objects are also appreciated is incidental to object perception. The important thing about objects, for perception, is not their local colors, but their color contrasts with their surrounds; not their lightnesses, but their reflectancy contrasts with their surrounds. The fact that the perceived colors and lightnesses of objects appear to remain relatively constant regardless of changes in overall illumination is a separate and secondary issue. What is primary is the registration of object boundaries in terms 9; brightness differences. In his paper on color constancy, Helson (1943) identifies {38} the fundamental importance of brightness-differences for contour and boundary perception:
Contrast gradients, especially of brightness-differences, establish the boundaries between objects and their surroundings. To see objects in the third dimension requires separation of planes and formation of borders. Gradients help in separating objects from surrounds
. . . (p. 262)
. . .
The presentation of lightness serves to demarcate the object from its surround and to make it a thing just as much as it tends to color constancy. (p. 26b)
In fact, the whole issue of color constancy is a muddle of confusion. The very terms “color constancy” and “lightness constancy” are highly misleading. These terms are usually meant to imply that our senses tend to preserve the “true” or constant local colors or brightnesses of objects. But this is false. In the first place, the phenomenon of lightness and color contrast reveal that the perceived lightness and color of an object are definitely altered by the surround contexts in which the objects are presented. In the second place the photo-reactive properties of objects, such as their absolute local achromatic or chromatic reflectancies, are not directly appreciable by visual perception and are not the correspondents of lightness and color, respectively. On the other hand, it is true that if the wavelength composition of the source of illumination is held constant, and the source in changed only in intensity, the relative lightnesses and colors will remain constant. But this does not represent the effect of a sensory mechanism that tends to hold local color and lightness values constant. The colors and lightnesses are perceived as constant because these sense qualities are determined by energy ratios, which, under these conditions of intensity changes in the illuminating source, are constant. {39}
There is a very subtle fallacy in the concepts of lightness and color constancy. These terms are products of a theory of vision, which holds that the perception of lightness and color are the forms in which we directly appreciate local facts about light dependent properties of objects. In fact, however, color and lightness are the forms in which we perceive the relationships between the light-dependent attributes of an object as compared to and contrasted with the light-dependent attributes of the background in which that object happens to be found. Consequently, this relationship changes as the objects background changes, which results in the perception of different object colors under different background conditions. In any case, our perception of the object’s color is equally valid, for what we perceive, in the form of changing local colors, are changes in the object/ background light-reactive relationship to which perceived color, after all, corresponds.
The problem of resolving the apparent conflict between the facts of color constancy and the facts of color contrast only arises if we assume that brightness corresponds to local light intensity and color corresponds to local wave-length composition, and if we assume that “lightness” corresponds to local reflectancy. These assumptions force us to interpret the data on color and brightness perception to mean that our visual system is able to overcome changes in illumination in order to provide us with accurate information about the local luminances of objects, an achievement called “brightness constancy” and that our visual systems are unable to overcome the effects on perceived brightness produced by differences in background or surround illumination, a distortion called “simultaneous brightness contrast.” But the objectivity of perception can be reaffirmed {40} if we abandon this interpretation of the data by abandoning the unwarranted assumptions about the correspondents of visual perception, upon which this interpretation is based, and recognize the fact that both brightness and color correspond to complex relationships between the photo-reactive properties of objects and the photo-reactive properties of the objects surroundings.
If we are to understand perception, we must not begin with any preconceptions about what the correspondents of sensory qualities should be. We must discover empirically what these correspondents are. If we make the mistake of taking the former approach, i.e., prejudging what our sense qualities should correspond to, when we discover that our senses to not reliably register what they “should” be registering, we may leap to the unwarranted and epistemologically self-contradictory conclusion that our senses are unreliable and subject to all sorts of “illusions.” This disastrous route has been followed by most writers on the subject of sense perception.
Sharp Energy Boundaries as the Prerequisite for Object Perception
The perception of objects depends upon the awareness of object contours or boundaries, which depends upon the sensory registration of the ratios of stimulus energies defining the energy margins which compose the displayed energy gradients and which specify the external object boundaries. Objects are not merely spatially bounded; they are, for the most part, sharply defined by their outer surfaces. It is at the surfaces of objects that organisms interact with objects. Furthermore, objects having ethereal or nebulous margins are ecologically rare (e.g., clouds or fog) and are spatially unstable. It is for these reasons that the perceptual system {41} has evolved into a highly selective boundary register. Since object boundaries are specified at the receptor in terms of relationships between energy boundaries, the perception of objects requires the registration of energy ratios developed across sharply defined energy boundaries. Vague or gradual boundaries produce unstable form-perceptions or, if gradual enough, no form perception at all. Two examples of this principle in the visual domain follow:
The first example consists of an experiment that can be easily carried out by the reader. Place a pencil between a large diffuse light source and a flat surface so that a vague shadow of the pencil is cast upon the surface. Now fixate the center of this vague shadow, and notice how quickly the shadow disappears. This disappearance.is accompanied by a slight decrease in the apparent brightness of the surface, an effect that is possibly due to some sort of energy averaging by the visual system. The shadow can be recovered by sharpening the lightness gradient at the retina. This can be accomplished without changing the steepness of the energy gradient at the shadowed surface by quickly moving the pencil sideways, i.e. at right angles to the line connecting the light source and the surface, or by moving the eye so that the shadowed region is displaced across the retina. Either of these movements causes the rate differences at which energy is absorbed at adjacent retinal regions to increase, thus reestablishing a gradient of adequate sharpness for boundary and, therefore, form perception.
My second example refers back to Land’s work on color perception. You will recall that Land (1959) was able to produce the perception of a large array of colors by using only red and white lights mixed in different ratios across the illuminated screen. This was accomplished by means of two black and white transparencies of a scene photographed once through a red filter, {42} and once through a green filter. The two resulting negatives were made into two positive transparencies called respectively the long and the short record. Red light was projected through the long record and white light was projected through the short record. When the two resulting images were superimposed onto the projection screen, a multicolored array resulted containing the same colors present in the original scene, but with considerable loss in saturation.
After considerable experimentation with different types of black and white film, I was able to prepare a set of long and short records which replicate Land’s findings quite well.* I was able to perceive, using only red and white light; red, pale blue, blue, green, pale yellow, orange, brown, black, gray, and white. These corresponded to the colors in the original scene, but with considerable desaturation, particularly for the blues and yellows.
The thrust of our latter example, however, concerns a second experiment by Land, reported in the same article. In this experiment Land attempted to produce every possible ratio of red and white light by replacing the two_ transparencies with two neutral density wedges oriented at right angles to one another. Thus, the white light continuously decreased in intensity from the top of the screen to the bottom, whereas the red light continuously decreased in intensity from left to right on the screen. Land describes the results as follows:
With both projectors turned on we now have an infinite variety of red-to-white ratios on the screen, duplicating all those that could possibly occur in a colored image. However, they are arranged in a strictly ordered progression. There is no randomness. And on the screen there is no color only a graded pink wash. (p93)
* I would like to thank Mr. Jay Conne for his invaluable suggestions and for his photographic work in preparing the slides. {43}
Land’s conclusion that colors can only be perceived when chromatic ratios change in a random manner from point to point, is not quite sufficient to explain the lack of color under these conditions of continually graded ratio differences. Land was still trying to deal with color perception in terms of pointal or local analysis of frequency composition. The fact that color perception requires what Land calls a random distribution of local ratios is far more intelligible when described in terms of the hypothesis which I am advocating here, namely, (as Gibson might agree) that color perception requires sharp chromatic energy boundaries, i.e., boundaries across which chromatic ratios sharply change.
Another dramatic example of the importance of sharp energy boundaries for the awareness of color differences can be found in the Ganzfeld experiments of Hochberg, Triebel, and Seaman (1951). The Ganzfeld consists of homogeneous, textureless visual field, usually produced experimentally by uniformly illuminating a white textureless surface occupying the entire visual field. Hochberg and his colleagues employed halves of table tennis balls cemented comfortably but firmly over the eyes and illuminated from the outside. Two general findings of the Ganzfeld experiments are relevant to our present discussion.
The first significant finding is that in the total absence of visual texture, i.e., visual contour or boundary patterns, all perception of color disappears. If, for example, the field is illuminated with red or green light, subjects report that the perception of red or of green only occurs during the moment that the illumination is turned on after which the color gradually disappears. When the color has disappeared, the introduction of a visual gradient, created by casting a shadow of the experimenter’s finger {44} onto the central region of the eye-cap, produced the following results:
In Group A1, in which the incident light was red, all five Ss reported a black shadow surrounded by a red halo, and both the shadow and the halo quickly disappeared with the removal of the finger. In Group A2, in which the incident light was green, four of the five Ss reported a black shadow against a green background or halo. The remaining S reported a purple shadow against a green halo. (Hochberg, Triebel, and Seaman ,1951 p. 66)
These results suggest that color perception can only be maintained when there exist energy gradients across the visual field. Furthermore, two facts obtained in these experiments seem to counter indicate that the disappearance of color is the passive result of retinal fatigue to monochromatic light. First of all, the introduction of the shadow causes the immediate return of color to the region surrounding the shadow, even though there is no change in the incident light energy falling on the retinal regions surrounding the retinal image of the shadow. Secondly, nine of the ten subjects tested reported that when the diffuse illumination was interrupted for intervals of about 2 seconds they experienced momentarily the complementary color, even though the stimulation of the complementary retinal receptors must be assumed to have been either unaltered or, at best, slightly decreased, by interrupting the illumination. In any case, the effect of interrupting the illumination should have produced, according to the simple local retinal fatigue or excitation hypotheses, either no appreciation of complementary color, or at best a decreased appreciation of the color complement. We can conclude that in the absence of energy patterns, the chromatic properties of light energy are inadequate to produce sustained color perception.
A second major significant finding coming out of Ganzfeld studies by both Hochberg and by Cohen (1957) is that in the complete absence of visual texture, perception radically breaks down to such a degree that some subjects {45} report intense anxiety and fear of “going blind” (Hochberg, et al.). On this phenomenon, Cohen reports:
During the course of adaptation, five of the sixteen observers reported a complete cessation of visual experience. Two reported a blackout almost every time they were confronted with prolonged, homogeneous stimulation._ This was a-unique experience which in- involved a complete disappearance of the sense of vision for short periods of time, and not simply the presence of a dark, undifferentiated visual field. The following description is representative: “foggy whiteness, everything blacks out, returns, goes. I feel blind. I’m not even seeing blackness. This differs from the black when lights are out.” It may be conjectured that the perceptual mechanism has evolved to cope with a differentiated field and in the absence of differentiation there is a temporary breakdown of the mechanism… (p. 21) (Emphasis added.)
Cohen’s conjecture is consistent with the thesis offered here, namely that our perceptual systems are not designed to deal with energy as such but are designed to process energy gradients or patterns which are a sine qua non for perception. Homogeneous energy impinging on our receptors does not yield any sort of pure sensational awareness, but produces instead cognitive breakdown experienced as a loss of contact with reality. When homogeneous energy is applied to all of the senses, i.e., diffuse light to the eyes, white noise to the ears, uniform temperature and pressure to the skin, the degree of cognitive breakdown is dramatic (see Solomon et al., 1961).
The Perceptual Significance of Sensory Adaptation
Sensory adaptation is a general term referring to the progressive and reversible change in responsiveness of the sensory system to components of the available energy. By sensory adaptation, I am referring to the physiological adaptation to-energy components, and do not include perceptual adaptations, such as those achieved during the wearing of inverting or displacing prisms, which involve sensory-motor learning. Sensory adaptation does not involve {46} learning, inasmuch as no conscious action on the part of the perceiver is required. (This psychological requirement for perceptual learning will be discussed at length later.) Sensory adaptation is said to have occurred whenever a previously adequate stimulus now fails to have an effect on perception, or when a previously inadequate stimulus is now adequate to effect perception. I will give examples of each.
A simple example of an adequate stimulus becoming inadequate is the adaptation to homogeneous color in the Ganzfeld experiments. Here the presence of monochromatic light gradually fails to produce the perception of a colored surface, and produces instead either the experience of an indeterminate fog or produces no visual awareness at all.
Another example in this category is that of cutaneous pressure adaptation, whereby a weight placed on the skin gradually becomes undetectable. Both of these examples must be regarded as cases of stimulus failure and not of neural failure. I have already mentioned this point in my earlier discussion of the Ganzfeld experiments. This point is equally applicable to cutaneous pressure adaptation. As Geldard (1972) clearly notes in his discussion of the results of work carried out by Nafe and Wagoner (1941):
Their results show that a weight, once placed on the skin surface, does not rest there but continues to move downward for a surprisingly long time. Pressure is felt just so long as a supraliminal rate of movement is maintained. When tissue resistance reduces motion to an undetectable level the sensation fades out. The end point of “adaptation” has been reached. The conclusion is obvious: “complete adaptation” represents stimulus failure. The fact that the answer is not to be found in “fatigue” or depletion of the resources of the end organ is easily proved by increasing the load on the tissue once adaptation has apparently run its course. In such a situation there is a fresh response, a new sensation accompanying the further compression, until the stimulus once again “fails.” (Geldard. p. 299) {47}
These two examples of sensory adaptation actually represent the effects of stimulus failure and suggest that the minimum conditions for perception require more than merely the display of energy at the receptor, and more than the display of a sharp energy gradient at the receptor. What is required is a changing energy distribution, i.e., an energy display that changes at the receptor over time. This perceptual requirement is dramatically illustrated in the visual domain by the experimental investigations on fixed retinal images. This body of data which has been reviewed by Heckenmueller (1965), is concerned with the perceptual disappearance of an object whose optical image is displayed on the retina in such a way that the image’s position relative to the retina remains fixed regardless of how the eyes move relative to the object. In this case, object perception fails even though there exists a sharply defined pattern of energy at the receptor. Perceptual failure under these conditions of fixed retinal images strongly suggests that visual perception requires energy patterns that are constantly displaced across the retina. This requirement of adequate visual stimulation seems to indicate the existence of a physiological mechanism that enables the visual system to separate out neural effects, which are not caused by externally originating excitation patterns. The precise mechanism probably employs the microgitter of the eye, i.e., the constant and naturally occurring physiological tremor of the eyeball. Any boundary or margin of an energy gradient is normally continuously displaced as a unit across the retina. The perception of a boundary might require a continuous synchronous firing of different sets of retinal elements in the same relational pattern. Thus, as the energy “edge”{48} is oscillated back and forth across the retina, different sets of retinal elements are made to fire synchronously, but the percept remains constant because the relational attribute of the different sets of activated neural elements are constant. If these assumptions are correct, then microgitter is not a physiological “problem” for the visual system, but is actually essential for adequate visual perception. The perceptual “neutralization” of fixed retinal patterns may reflect the actions of an automatic physiological mechanism which treats as “receptor-specific” any local neural activation pattern which is not systematically spatially transposed to another set of neural elements.
Regardless of what the specific neurological mechanisms turn out to be, the general principle suggested by experiments on fixed retinal images, Ganzfelds, color and brightness contrast, and cutaneous pressure sensitivity, is that adequate sensory stimulation requires constantly changing displays of energy patterns.
Philosophical Note On the Epistemological Import of Stimulus Failure
The requirements for adequate sensory stimulation, i.e., stimulation that will yield a stable perception, may be thought of as conditions designed into perceptual systems in order to insure, insofar as it is physiologically possible, that perceptions are caused by exogenous patterns of neural excitation, and are unaffected by endogenous patterns of excitation. In this n regard, epistemologically it is more important that no endogenous neural activity be allowed to effect perceptual content than that all exogenous excitations be allowed to effect perceptual content. Direct awareness of external reality requires the existence of a physiological mechanism that specifically isolates externally caused patterns of neural excitation {49}
Let us turn now to the category of sensory adaptation in which a previously inadequate stimulus becomes adequate. From the data on this category of adaptation, another principle of sensory functioning seems to emerge, viz. that our sensory systems constantly adjust their responses to the components of energy gradients so as to maximize the range of perceptual differentiations possible under the given conditions of stimulation.
Consider, for example, the classic phenomenon of color adaptation. If the visual field is flooded with light of a single color, or if a segment of the visible spectrum is filtered out, for example by wearing colored spectacles, the initial perceptual effect is a dramatic loss in the range of discriminable colors. Red spectacles (or red flood-lighting) at first produce a perception of the visual field in terms of differing saturations and brightnesses of red. Gradually, however, the red “wash” over everything seems to fade and other colors reappear, although with considerable desaturation. In effect, the sensory system has become, to some degree, desensitized to red light, with the result being that a greater range of frequency- difference ratios can now be registered yielding a wider array of colors and, thereby, an increased registration of boundary differences. The effects of color adaptation are illustrated diagrammatically in Figure 4. {50}
Figure 4.
Diagram of a hypothetical case of color adaptation (See text page #9.)
If we assume, as the experimental data seem to indicate, that color perception is in part determined by the ratios of the chromatic response differences at regions A and B, viz., the-ratio of the differences between the red and green response at A, expressed as RA-GA, and the red and green response at B, expressed as RB-GB, then decreasing the overall amount of green light by filtering will decrease the difference ratio and thus decrease the chromatic contrast. This will cause regions A and B to become {51} less discriminable by virtue of their appearing more similar in color. If, however, the sensory system responds to this situation by decreasing its sensitivity to red light, by the process of adaptation, then the difference ratio will increase, and the consequence will be that regions A and B will become more discriminable again by virtue of their appearing more dissimilar in color.
Color adaptation, like the adaptations which yield stimulus failure, is an active sensory mechanism for insuring optimal adequate sensory stimulation, and is not merely the passive result of receptor “fatigue,” or “exhaustion” of the red (or green) receptors. That this latter is not an adequate explanation of color adaptation, is dramatically illustrated by the experiments of Kohler(1962) with colored goggles. Kohler’s subjects wore, for an extended period of time, spectacles each monocle of which was divided in half vertically, so that the left field was tinted yellow and the right field was tinted blue. By virtue of this arrangement, when the subjects rotated their eyes to the left, the visual field was overlaid with a yellowish color. Conversely, by rotating the eyes to the right, the visual field was perceived as bluish. Gradually, these rotation-specific color effects disappeared when the goggles were removed. Furthermore, after removal subjects reported that when their eyes are rotated to the right, the entire visual field appeared yellowish (opposite of the effect produced when wearing the goggles in the early stages of adaptation). In addition, if the same object is fixated, and the body rotated so that the eyes now faced left, the subjects report that the entire visual field now appears blue, even though the retinal image has not “moved.” Receptor fatigue is wholly inadequate to explain these results, which must be regarded as the product of an active motor-specific {52} adjustment of the visual system’s criterion of adequate color information.
From the experimental evidence, evaluated within the context of the concept of perception offered in this paper, several general conclusions appear to be warranted:
1. Our sensory systems are ingeniously engineered so as to maximally respond to those features of energy gradients that are caused by the actions of external objects. In other words, our sensory systems are designed to register, i.e., yield perceptual effects for object-specifying energy attributes reaching our receptors.
2. Our sensory systems achieve their function of isolating and registering object-specific energy attributes by isolating-the attributes of changing energy gradients that are energy-independent. For example, our sensory systems tend to isolate energy ratios, since these are relatively independent of changes in overall energy. Our sensory systems isolate sharp energy boundaries, since these are also independent of changes in overall energy. Our sensory systems isolate systematic changes in the directionality of the energy gradients displayed at the receptor, since directionality is also independent of changes in overall energy. Except insofar as energy displays consist of changing energy ratios, changing energy boundaries, and changing energy directionalities, energy yields no stable perceptual effects. Perception depends upon the registration and isolation of object-dependent, and, therefore, energy-independent and receptor-independent relational attributes of changing energy patterns. It is these relational invariants that specify the constancies, viz. objects, which are their causes.
3. The perceptual system specifically employs the principle of pattern transposition, in order to isolate externally caused excitation {53} patterns. The excitatory effect produced by a pattern contains features that are invariant over pattern translation across the neural substrate. It is these relational invariants and not the neural elements themselves that constitute sensory units. Our perception of even the “simplest” sensory fact, such as lightness differences, must depend: upon the creation of neural patterns containing relational features that are invariant over translation of the stimulation pattern or gradient.) Thus, the neurophysiology of perception must rest on the principle that a sensory unit consists of relational invariants in patterns or neural events.
So far we have discussed some of the purely physiological mechanisms responsible for the isolation and extraction of object-specific energy effects. We have seen that the sensory system has built-in processes that automatically overcome the effects of non-informative energy events and endogenous physiological disturbances. These results are achieved by the sensory system’s dealing with the relational properties of changing energy patterns. Although we have identified what are the general physical and physiological preconditions for perception, these are not in themselves sufficient to explain perception as a conscious awareness of objects external to and independent of the perceiver. Perception is not the passive product of physical and physiological events, but depends for its full functional achievement upon an interaction between conscious processes and automatic physiological processes. To understand this interaction, we must make an abrupt shift from the physiological level to the psychological level. {54}
Latest page revision: April 19, 2014 3:58 pm
bioperipatetic 