Any account of the facts of perceiving must include the facts of error - the failures to notice as well as the noticing, the overlooking as well as the looking. Actually, the deficiencies of perception are much more familiar to us than its successes. We take the latter for granted, but we are naturally curious about the causes of our misperceptions, misjudgments, and mistakes. We have a special curiosity about a class of inaccuracies that are called illusions. They are usually not serious enough to be called misperceptions. Often we are aware of the illusion, as we are of the image in a mirror, the bent stick in water, the circular coin that looks elliptical, and the after-sensation "in front of the eyes". But these are still failures of perception, to be exact, and they are very interesting.

How does a theory of information-based perception as distinguished from the theories of sensation-based perception account for misperceptions and illusions? Since the present theory is primarily a theory of correct perception, it must explain incorrect perception by supplementary assumptions.

The classical theories of perception, on the contrary, explain both perception and illusion, with the same assumptions. The influence of past experience on sensory data, for example, is supposed to be sometimes one of correcting the data and sometimes one of distorting them. The effect of sensory organization in the brain on the inputs of the nerves is supposed to be one that makes the forms in the brain like the forms in the world, but also one that makes them unlike the forms in the world. There is a lack of logic here. If misperception is the opposite of perception, the law of association or the law of sensory organization cannot apply to both at the same time. The same principle should not be used to explain why perceiving is so often correct and why it is so often incorrect. A theory of perception should certainly allow for misperception, but it can hardly at the same time be a theory of misperception.

In the theory of information pickup, clearly, the pickup may fail when [p. 288] the available information is inadequate, or it may fail when the information is adequate but is not picked up. The former is no fault of the observer; the latter is. The two cases are theoretically separate, even if they are practically sometimes hard to distinguish. In this chapter the attempt is made to list the ecological inadequacies of information first and the psychological deficiencies of perception second. The perceptual systems, as we shall note, do the best they can with what they get, but in some circumstances they get very little to work on.

Inadequate Information

There are some natural circumstances in which the obvious information in light specifies a fact that is false. A straight stick looks bent (Figure 14.1) when part of it is submerged in water because its corresponding margins in the optic array are bent by the refraction of rays. A mirage of trees and buildings appears because of reflection from air layers. The green foliage of distant mountains appears blue because of the differential transmission of wavelengths through air. There can even be misinformation in mechanical stimulation, as when a jet of water feels solid instead of liquid because the information for an unyielding instead of a yielding surface is present. But these natural situations cannot be treated experimentally, and what the psychologist knows about perception comes largely from experiments. A large class of these are studies in which the stimulus information from objects or events has been artificially reduced by a curious investigator.

Experiments on perception with reduced information are very frequent in psychology. They have always been thought of, however, as experiments with reduced stimulation. It has been the hope of the investigators to cut back the sensory basis of perception so as to allow the perceptual process to come into its own - to reveal in relatively pure form the laws of its operation. We shall have to reinterpret the work in terms of reduced information.

In this chapter an attempt will be made to classify reduced stimulus information under seven headings: minimal energy, blurring, masking, conflicting information, interval cutoff, narrowing down of an array, and operations on structure. In some of these experiments, especially the last, the information may not have been reduced or diminished but only altered in form.

Minimal Energy and the Concept of Threshold

Photoreceptors, mechanoreceptors, and chemoreceptors require a certain amount of energy to be excited (Chapter 2), although the amount may be very small. Conceivably, a rod cell of the human retina when it is [p. 289] dark-adapted can be discharged by one quantum of light energy (Pirenne, 1956). The absolute intensity thresholds for sensation of imposed stimulation, however, are not as simple as is implied by the theory of a receptor mosaic composed of cells, for the area of a stimulus combines with its intensity to determine the threshold and so does the short-term duration of a stimulus. The thresholds of the receptive units interspersed in the retina depend on the area stimulated and on the length of time. The size of the "goad," as it were, and the duration of its application, help to determine its effectiveness as a stimulus. All we can be sure of is this: a sufficiently distant lighthouse or a sufficiently distant star, or a sufficiently brief flash of either, will cease to be seen at night. If a source of vibration is far enough away its pressure waves will cease to excite the ear; and if a contact with the body surface is sufficiently light, small, and brief, it will cease to affect the skin. Obviously no information can be obtained about the lighthouse, the star, the sounding object, or the touching thing if the stimulus is ineffective.

With respect to the ear and the skin, however, the mechanoreceptors [p. 290] are so sensitive that the absence of detected sound pretty well guarantees the absence of any vibrating object in the vicinity, and the absence of detected touch guarantees the presence only of air (or the medium) adjacent to the body. I have argued that these facts are themselves information, albeit of the most primitive sort.

The obscuring of the structural information in ambient light by a low level of illumination is a deficiency in the eye, not in the light. The obscuring of the information in sound waves by a low level of amplitude is a deficiency in the ear, not in the sound. The proof is that the information can be made available by operations of enhancement and amplification, e.g., by an image-magnifier or an audio-amplifier. The human eye declines in "acuity" as the light weakens, and the human ear fails to make out the words as the sound weakens, but the information is physically present, that is, theoretically available.

It is important to remember that the concept of a physiological threshold - a certain minimal amount of energy absorbed by a sensory surface over a given area during a given time - refers only to energy measurements and not to information, that is, not to the variables of higher order that contain information. Fixed thresholds apply to the theoretical sensitivity of passive receptors but not to the sensitivity of active organs, since the latter depends on the development and education of the perceptual system to which the organ belongs. The attempt to measure absolute thresholds, accordingly, can be carried out when sensory impressions are reported by a passive observer but not when he actively seeks to obtain perceptual information. Absolute thresholds of pure sensation, if they could be established, would probably not be lowered by learning. The trouble is that such fixed thresholds have not been established experimentally, for the notion of a wholly passive observer, an "ideal" observer as he is called by sensory physiologists, is a myth. Real observers in real experiments have to be motivated to observe, and their attention fluctuates with their degree of motivation. Consequently the idea of a statistical threshold has had to be substituted for the idea of a physiological threshold. But some theorists do not find this satisfactory and have suggested that a so-called threshold is actually the probability of detecting a signal in the presence of noise (see below). The very concept of a sensory threshold has become uncertain in recent years, although it is fundamental to the theory of sensations.

The attempt to measure intensities that will just excite neurons is a useful endeavor for physiologists. The attempt to measure intensities that will just arouse sensations is practically useful for such purposes as the design of lighthouses. But it is clear that the latter measures are not absolutes, and the former, although they set limits to the activity of perceptual systems, will never explain how they work. [p. 291]

.................................................

The Muddle of Subliminal Perception

An inconclusive controversy about social perception has recently arisen (Bevan, 1964) which illustrates the confusion into which we are led by assuming that information pickup must have a fixed threshold if supposedly basic sensation has a fixed threshold. Certain experiments purported to show that an observer could perceive meanings or suggestions unconsciously, or could discriminate them without awareness of the sensory difference between them. This seemed to imply unconscious defense mechanisms governing perception as well as motivated behavior - wishful perceiving. But to say that one can perceive in order not to perceive is a logical contradiction. Something is wrong somewhere.

When perception is conceived as the detection of information, the weakness of physical stimulation may cause it to be piecemeal, partial, and dependent on personal motivation. But this does not imply subthreshold perception or "subception"; it only suggests that a perceptual system may be sensitized to one level of information and not to another.

................................................

The Blurring of Structure

The blurring of an optic array by fog, smoke, or haze should not be confused with the blurring of the retinal distribution by, say, myopia (nearsightedness), the loss of structure in the first case is incurable while that in the latter case can be corrected by eyeglasses. The blurring of an optic array by an imperfectly transparent medium can occur in varying degrees. The fine structure or texture of the array is the first to disappear. This yields what is called aerial perspective (Figure 14.2). Then, as the linear projection of the network of rays (Chapter 10) gives way to the dispersion or scattering of rays, the coarse structure may also disappear. In this situation the Londoner may ultimately complain that the fog is so thick he cannot see his hand in front of his face in full daylight. With fog in the air, the determinants of the features of environmental layout that do or do not remain visible (in accordance with the laws of size and distance) are extremely complex, as witness the difficulty of measuring visibility for the safe landing of airplanes. It is very hard for an observer in a control tower to tell whether or not there is enough structure in the manifold of perspectives in the air mass above an airport to enable a pilot to see what he needs to see. The structuring of light by the layout and reflectance of surfaces is itself complex, as we have noted. The fragile information with which we so confidently get about in the world is wholly at the mercy of atmospheric conditions. The nature of this information is such that it is physically weakened by blur. It is not, however, physically weakened by low intensity.

The ultimate degree of blur is found in a homogeneous optic array, that is, one with no structure at all. This is what the cloudless sky presents to [p. 292] an eye, or a fog of the highest density. This mode of stimulation is best achieved experimentally by covering the eyes of an observer with hemispheres of diffusing plastic. Halves of table-tennis balls serve nicely for this purpose, and I have repeatedly done this experiment (Gibson and Waddell, 1952, see also Cohen, 1957). The observer says he sees fog, or sometimes sky, or often "nothing." In a certain sense he is right. In the same sense that the absence of contact specifies nothing but air, the absence of optical texture specifies nothing but air. Under natural conditions a textureless sector within the total array of ambient light guarantees a space into which a bird can fly without danger of collision, into which one can proceed indefinitely, for no surface lies in that direction. So in a sense even the absence of structure conveys information as the alternative of nothing to something. [p. 293]

The appearance of sky is produced, as every theatergoer knows, by a finely textured curved surface at the back of a stage which can be flooded with illumination. It is called a cyclorama. The actual surface may be only a few feet behind the garden wall of a stage setting, but to the audience 50 feet away the illusion of depthless space will be compelling. There are other ways of causing a surface to look filmy and insubstantial (to be described later in this chapter), but making the grain of the optical texture too fine to be detected is one.

The occurrence of "whiteout" in the environment of a level snowplain under certain special weather conditions is instructive in this connection (see also Chapter 10, p. 212). It is analogous to a "blackout," in which case also nothing is visible. Blackout provides no information about the world because energy is absent; whiteout provides no information about the world because, although energy is present, structure is absent. It is said to be a very alarming experience for those who drive vehicles about in arctic regions. The undifferentiated light specifies an empty medium before the observer but this information is false; the snow-covered terrain with its potential obstacles exists although it seems to have vanished.

The Masking of Structure

In the study of auditory sensations a well-known effect is expressed by saying that one sound can mask another if the two are concurrent. It this effect physical or physiological? It is widely assumed to be physiological because physical vibrations do not cancel one another out - or do they? For us the question is whether information can cancel out other information. In the study of auditory communication, where the notion of information is introduced, the fact is that a signal is progressively harder to detect as the level of noise is increased. Speech, for example, eventually becomes unintelligible in the presence of "static," or the hissing sound of a "white noise." On the assumption that the intelligible signal and the random noise are reciprocals of one another, the noise does objectively cancel the signal. This theory works very well for problems of communication by telephone or radio. There is some question, however, whether it should be applied to the broadcasting of information by natural event in a terrestrial environment.

It is a very interesting puzzle to decide whether, for example, the information broadcast by a bird call is present in the air at a station-point where a nearby waterfall fills the air with pressure waves of much higher amplitude. I am no expert in acoustics and may be wrong, but I am inclined to think that it is not present. However, in the "cocktail party phenomenon," where overlapping fields of speech sounds tend to make bedlam, it is possible that wave-front information as distinguished from [p. 294] wave-train information (Chapter 5) may help to sort out the vocal signals. In any event, whether the information is not available or is available but not registerable, one sound source can in effect mask another.

A rather similar kind of masking can occur in stationary optical structures. Examples are found in the hidden figures contained in line drawings, such as the one shown in Figure 14.3 (Gottschaldt, 1926; Metzger, 1953, Ch. 2). This masking has been compared to that which seems to occur in nature when animals "freeze" and presumably thereby reduce their visibility relative to the background, as for example, in Figure 14.4. Both kinds of masking have been compared to the art of military camouflage. In the drawings, the information for detecting the part figures is present in the optic array from the pictures, since they can in fact be perceived after considerable visual searching. So can the information for detecting the animals, but the question arises whether optical information can in other cases be so thoroughly imbedded in optical "noise" that it ceases to exist. For this puzzle, too, I have no certain solution. In any event, the information in the structure of pictorial optical stimulation, like that in acoustical stimulation, can become so intertwined with other information that observers cannot perceive it. Whether they can always be trained to do so is a theoretical question.

Visual masking as described is not the same thing as the "veiling" of contour and texture that occurs with high illumination. Presumably the latter is due to glare, so called, and this is a subjective phenomenon - that is, the failure to detect is a failure of the visual system, not of the structure of light. When the system is swamped by too much energy despite the moderating effect of the pupils, the situation can be remedied by wearing dark glasses. [p. 295] [p. 296]

In Chapter 11 (Figure 11.7), four illustrations were given of the balanced opposition of stimulus information from a pictorial display, two causing an apparent reversal of figure-ground and the other two an apparent reversal of perspective. In one picture a goblet was seen to alternate with a pair of faces. The information for the goblet might be said to mask the information for the faces. This seems to me a more fundamental assertion than saying that the brain process corresponding to the goblet interferes with the brain process corresponding to the faces, as Gestalt theory does.

It should be recalled that in all such cases of equivocal perceptions from the same frozen array, the ambiguity of edge and of depth would be resolved at once if the array underwent transformation. Static structure does not convey as much information as kinetic structure does. I believe that all cases of visual masking are confined to the static situation.

Conflicting or Contradictory Information

The reversible figures raise the question of conflicting stimulus information, or what is traditionally called "conflict of cues." In these figures the conflicting cues were both visual, but they need not be. Information is usually available to more than one perceptual system at the same time. Experimenters on perception have often devised situations where the information for one system, e.g., the visual does not coincide with the information for another, e.g., the vestibular. An example is the perception of the vertical-horizontal framework of the environment, described in Chapter 4. Ordinarily, the main lines of the ambient array specify the true vertical and the pull of gravity on the weights of the inner ear also specifies it. Moreover, the upward pressure of the surface of support usually specifies it. But if a whole room is artificially tilted the visual and vestibular directions of up and down no longer coincide.

What exactly does this discrepancy or non-coincidence consist of? I would call it a discrepancy of information, not of sensations. Let us consider what this implies. A traditionalist would argue that the input of the hair cells of a statocyst organ (Chapter 4) and the input of a retinal image are mere arbitrary signals that must be associated before they have meaning. But this argument neglects what is important - namely the range of inputs of a statocyst as the head goes from horizontal through vertical to horizontal again. A given input has meaning by virtue of its place in this range of inputs. Moreover, in ordinary life this is coincident with the range of inputs of the retina as one lies on the left side, sits up, and lies on the right side. The normal upright of haptic-somatic space coincides with the normal upright of visual space because the differentiated inputs of these two organ systems are covariant. Whatever one's posture, the line of the horizon as registered visually remains coupled with the line of gravity as registered by the body. These two kinds [p. 297] of perceptual development, differentiation and covariation, were described in the last chapter. The discrepancy introduced in the above experiments, therefore, means that a genuine biological invariant has been destroyed. The input of one system contradicts that of another. When the visual and the postural determinants of the phenomenal vertical disagree (see Figure 14.5), there is no longer any single unitary phenomenal vertical, I conclude, and this seems to the fit the evidence (Gibson, 1952b). The observer in these experiments, e.g., a man seated in an artificially tilted room, must either accept the visual information and reject the postural, or accept the postural information and reject the visual, or alternate between the two, or compromise between the two. Of course, he may sometimes be just confused. All of these outcomes show up in the results of the experiment shown in Figure 14.5 (left).

The problem of conflicting cues in space perception has been studied [p. 298] in a great many experiments, by many investigators, but always with the aim of trying to solve the puzzle of how one sense could validate another, or provide criteria for another. The puzzle goes back to Bishop Berkeley, who maintained that visual sensations could only get spatial meaning from touch. Efforts to prove or disprove this hypothesis continued up to the present. The issue disappears, however, on the assumption that any perceptual system can pick up information inasmuch as its inputs are differentiated with respect to its possible inputs. Usually this information is covariant, coincident, or correlated with the information got by another perceptual system, and it is therefore redundant or equivalent. It can be made contradictory, however, by an experimenter, with various interesting consequences for perception.

Interval Cutoff with a Tachistoscope

A favorite device for impoverishing visual stimulation is the tachistoscope, which presents a display for only a brief interval of time to a human eye fixated on the window. The effect of this device on the information available to the eye is complex, not simple. At very short intervals, measured in milliseconds, the energy needed for vision is minimized by virtue of the law of photosensitivity that trades intensity for duration. This reduces information perforce. At longer intervals, up to about a tenth of a second, the pickup of information is reduced to that obtained with a single fixation. Exploring or scanning is thus prevented, and the human eye, being highly foveated, unlike that of the horse, must explore in order to perceive fully. The eye is thus treated like a camera and its intake of information is unnaturally limited. The rationale of this experiment is that the tachistoscope, by limiting perception to what can be seen in a single glance, enables the experimenter to isolate a simpler and purer form of perception. The assumption that pictorial perception is simpler than transformational perception has already been discussed in Chapter 12. At still longer intervals, around half second, applying the fovea to details of structure becomes possible, but the system is still frustrated to the degree that the sequence of fixations is cut short. Time is required for primate vision to reach its full scope. With the tachistoscope, therefore, available information is impoverished by limiting the time during which it is available.

It is instructive to contrast this method with another that is sometimes used by experimenters, one that reduces the ability to register the information. The subject may be required to fixate a mark on a screen and the display is then presented for a long interval at some angular distance peripheral to the clear center of the visual field. (The method demands disciplined and practiced subject, for the urge to fixate on items of interest is very strong and must be [p. 299] inhibited). In this situation the structural and temporal information from the display is fully available, but only the crudest features of it can be detected because the periphery of the human retina does not have the neural mechanisms for high acuity.

Narrowing Down of an Array

If the array entering an eye is reduced in angular scope to a few degrees by a tube or by a small aperture in a large screen, the surface reflecting the light within this narrow cone becomes invisible and the observer sees only a film at the end of the tube or in the aperture. Katz (1935) described it as a film color instead of a surface color, and thought of it as a "reduced" color. The experience seems to be the result of reducing the information for the detection of a surface. The area of optical texture projecting an area of surface texture has been so diminished that it no longer yields information about the layout of surface. A certain minimum angular size of an array seems to be required for such detection. My observations suggest the following stages. When the angle is large one sees a surface extending behind the window at a certain orientation, of a certain color, and in a certain illumination. When the angle is reduced, these differential properties begin to be indefinite. When the angle is quite small, none of these properties is visible and the lack of thingness may be described by saying that it looks like a film stretched across the window. When the angle of the aperture in a screen is still smaller it may cease to look like an aperture and appear to be merely a spot on the surface of the screen. With the progressive narrowing of the array, what has been reduced is the structuring of the array, I suggest, and the supposed reduction of color from a perceptual mode to a sensory mode of appearance is only part of what happens.

The ultimate reduction of the optic array to a single point of light in a dark room is an even more familiar experiment in psychology. In that case not even the location of the point remains definite for long, and the "autokinetic" phenomenon is the result.

Experimental Operations on Structure

Finally, we should consider some of the various ways of modifying, altering, biasing, or distorting the spatial and temporal structure of stimulation that have been tried by experimenters. Operations on sound and light by electronic and optical means are easy because the energies are in the form of waves and rays outside the observer where the experimenter can intervene between the source and the impinging stimulus. This field of research is relatively new and therefore what can be said about it is provisional.

Electronic distortion of sound waves can now be achieved. Most of it [p. 300] has been done with speech sounds. Various sorts of "clipping" of the wave train have been tried. Interval clipping, for example, involves the introduction of short intervals of silence into the sound. Within limits, this can be done without affecting the intelligibility of the speech; beyond these limits the perception of speech suffers. In much the same way, intervals of darkness can be cut out of the changing optic array from a natural episode; the changing array from a motion-picture screen is clipped in this fashion with what seems no loss of information. As longer blank intervals are introduced into the continuous natural sequence, flicker appears and motion becomes jerky. This is what happens when the standard rate of 24 frames per second of the modern motion-picture projector is reduced to 10 or 12 per second. It is interesting to note, however, that even when the pictorial sequence is reduced to a few samples of still pictures, the major transformations of the episode may still be preserved. This is demonstrated by the fact of storytelling with a picture sequence, a so-called filmstrip, and by the success of comic strips.

Another sort of sound distortion is peak clipping, which alters the wave forms but not their sequence. This can also be imposed on speech sounds without affecting perception. Still other distortions are described in Cherry's book on human communication (1957). As he suggests, when the essential invariants are preserved under distortion, intelligibility remains. Some distortions destroy them; others do not. The "search for invariants," as he puts it, is the fundamental fact of perception (p. 297).

What can be done to the simultaneous structure of an optic array? Gaps can be introduced without much loss, as when one looks at a scene through a picket fence, or photographs such a scene. The half tone reproduction of a photograph is full of small gaps that do not affect perception. The natural optic array carries much more information than anyone is ever likely to pick up, and much of it can be sacrificed. It is highly redundant, in the terminology of information theory.

A similar introduction of gaps into the outlines of a representative drawing has been studied by the Gestalt psychologists and their followers. The interesting discovery here is that the information is destroyed if the gaps occur in certain critical locations but not if they occur in others. A classic example from Koffka (1935) of a drawing that contains barely enough information for a familiar perception is shown in Figure 14.6. What are the critical locations in a drawing that convey the essential information? That, of course, is the question. An answer is being sought by Hochberg (1964, Ch. 5).

The most interesting experimental operations on the structure of an optic array, however, come not from pictures but from what I call the spectacle-wearing experiments. These operations are imposed on the [p. 301] cone of rays entering an eye from the environment by means of optical distortion. The structure of this array can be wholly transformed in any of several ways by putting a refracting piece of glass, such as a prism or a lens, in front of the the eye. Ordinary eyeglasses and contact lenses do not do this at all. These kinds of lenses, when properly fitted, are in effect merely adjuncts of the ocular system, correcting its anatomical defects and enabling it to register the finer details of an optic array. Experimental spectacles, on the contrary, alter the available information coming to each eye by introducing an inversion, reversal, or bias of its overall structure. A lens system will invert it; a wedge prism will bias it; a right-angle prism will reverse right and left (or up and down).

The first of the spectacle-wearing experiments was that of Stratton (1896, 1897), who inverted the field of view. The most comprehensive of them is the series of experiments at Innsbruck by Köhler (1964), who reversed or biased the field of view for long periods of time and tried other types of deformation. He also used colored spectacles, which bias the spectral structure of the array but not its geometrical structure. This kind of bias changes the colors of things at first, but as we know, the wearer soon adapts to the change. For spectacles altering geometrical structure, depending on the kind of optical alteration imposed, the perception of the environmental layout is correspondingly falsified. The information available in most of these experiments is not so much impoverished as deformed. It would be impoverished if the array were blurred by the spectacles, or if its adjacent order were permuted or disrupted as by transmission through pebbled glass, but the types of optical alteration so far used are not as radical as this. Wedge prisms, for example, introduce curvatures and compressions and spectral bands at the edges and corners of things, but the main features of the environment look the same as before.

The layout of surrounding surfaces is wrongly perceived in these [p. 302] experiments, and the acts of reaching and walking are accordingly made in the wrong direction. There is a conflict of information between the visual system and the haptic system. But the remarkable result is, first, that behavior soon adjusts to the altered information and, second, that the perception of layout gradually tends to become correct, at least with respect to certain variables of layout. The phenomenal curvatures of edges, the compressions of shape, the spectral bands, and the astonishing apparent motions of the world when one turns his head or walks about, weaken in the course of time. And at the end of the adaptation period, when the experimental spectacles are taken off and the observer looks around, the curvatures, compressions, color fringes, and non-rigid motions, all reappear as opposites of what he saw while wearing the spectacles.

The explanation of this adaptation, the correcting or veridicalizing of perceptual experience together with the readapting that must occur when the optical information returns to normal, is now being actively sought by a number of investigators. It is too soon to say what the final conclusions will be. The theory expounded in this book, however, implies a certain kind of explanation rather than other kinds. The theory was suggested, in part, by an effort to understand the early results of the spectacle-wearing experiments. There must be invariants over time in the flowing array of optical stimulation to specify the rectilinearity, the constancy, and the rigidity of the world. This assumption holds as much for vision without spectacles as for vision with spectacles. When they are first put on, the observer must learn what the new constants are in the stimulus flux. When they are taken off the observer must relearn the old constants again. The extraction of invariants by the perceptual system is taken to be the crux of the explanation of phenomenal adaptation.

.................................................................................

The Adjustment of Visual Proprioception to Spectacle Wearing

The reasons the apparent non-ridgid motions of the world disappear during adaptation (and reappear as opposite motions when the spectacles are removed) is that the novel transformations, being in fact self-produced, come to be taken as self-produced, and therefore cease to be taken as specifying motions of the environment. The rule is that total optical transformations are propriospecific and that invariants under transformation are exterospecific, and this rule holds with or without spectacles. The altered transformations with spectacles demand a sort of relearning of what constitutes visual kinesthesis and what does not, but the visual system seems to be capable of this (Köhler, 1964; Held, 1965).

................................................................................

[p. 303]

The Consequences of Inadequate Information

To conclude the first part of this chapter, we may ask, what happens to perception when the information is inadequate? In general, the answer seems to be that the perceptual system hunts. It tries to find meaning, to make sense from what little information it can get.

The observer keeps trying to see even in a dense fog, and he also does so at night in anything less than complete darkness. Similarly, he tries to hear even with little or no sound in the air. In darkness and silence, men and other diurnal animals may, of course, simply go to sleep and relax all attention, but so long as the individual stays awake and alert his exploratory attention persists.

In the complete darkness and utter silence of the so-called sensory deprivation experiments carried out recently with human subjects, the effort to see and hear is completely blocked. The subject cannot be deprived of all stimulation, however. He can always fall back on the haptic system, feeling the adjacent layout of surfaces. The information available to this system cannot be eliminated, for the subject is necessarily in touch with his surface of support even if not with more distant objects and events. But his "contact with the world" is much restricted. As a last resort he has only residual proprioception to keep his attention active; he can do exercises, feel himself, make noises, or talk to himself. But this is no longer perception; it is an egocentric or introverted kind of activity, and the ordinary person is dissatisfied by it. The subjects of experiments on perceptual deprivation have been paid large sums of money to persuade them to stay in their cubicles for more than a day or two. Despite the strong incentive, some of them report experiences so strange as to approximate hallucinations. Darkness, silence, and social isolation are highly frustrating. The perceptual systems seem to go on trying to function even without input, racing like a motor without a load. Perhaps this tendency explains the semi-hallucinations.

More typical of life than absence of stimulation, however, is the presence of stimulation with inadequate information - information that is conflicting, masked, equivocal, cut short, reduced, or even sometimes false. The effort of apprehension may then be strenuous. With conflicting or contradictory information the overall perceptual system alternates or compromises, as noted, but in lifelike situations a search for additional information begins, information that will reinforce one or the other alternative. When the information is masked or hidden in camouflage, a search is made over the whole array. If detection still fails, the system hunts more widely in space and longer in time. It tests for what remains invariant over time, trying out different perspectives. If the invariants still do not appear, a whole repertory of poorly understood processes [p. 304] variously called assumptions, inferences, or guesses come into play. Merely probable information, clues or cues, is not as satisfying for the perceptual system as the achieving of clarity, the insight that reveals the permanence underlying the change; but guessing does occur in highly complex situations and the individual may sometimes have to be content with it.

The above assertions are consistent with the results of many experiments on the effects of impoverished stimulation and inadequate information. The methods employed were described in the last section. The results have here been reinterpreted but the general formula of the search for meaning seems to fit them all fairly well.

The Deficiencies of the Perceptual Process

One can admire the efficiency of the perceptual process and at the same time study its failures and defects. If the available information about the world is theoretically unlimited, as I have assumed, perception at its best will always be deficient in some ultimate sense. For that matter, if potential scientific information is infinite, scientific knowledge will always be imperfect. From this cosmic and philosophic point of view we can never be absolutely sure of anything. But the point of view adopted in this book is more modest and less demanding. It is the point of view of a scientific psychologist concerned with the perceptual achievements of animals and men. He must formulate the findings of ecology and the physical sciences about the properties of the real environment and then ask why they are sometimes detected and sometimes not. It is a mistake for the psychologist to ask himself at this point how he himself perceives or knows the properties of the real environment. This is not his problem. He can only properly ask how one perceives, not how he perceives. His role as an investigator of the perception of the world should not be confused with his other role of having to know what the world is like so that he can evaluate the process of perception. But the confusion of these roles is a common error, I believe, among psychologists.

Keeping in mind the above considerations, we can ask, what, then, are the deficiencies of the perceptual process commonly found in individuals? We know that perception is deficient in the lower animals as compared to the higher animals, and that it is less efficient in the human child than in the human adult, but let us confine the question to the last case only - to the supposedly normal observer. We exclude from consideration all deficiencies due to disease or injury.

The Failure of Organ Adjustment at High Intensity

The hermit retina does quite well at low levels of illumination but, s we have already noted, it is subject to the effect of glare at very high [p. 305] levels despite the adjusting of the pupil. Full sunlight on snow or water, or the direct rays of the sun itself, constitute an array whose differences of intensity cannot be registered because the intensities are too great. The capacity for adjustment of the system is then exceeded. The sensation of "dazzle," and even pain, accompanies this state of affairs.

The ear picks up the sequential structure of very weak sounds but it fails to do so for very loud sounds. There is a muscular mechanism in the middle ear that can alter the tension of the eardrum, which is thought to offer some protection against high-amplitude pressure waves, but it is not sufficient for the highest of them. Pain supplants perception in this event.

Intense mechanical encounters with objects are likewise painful. So is rapid absorption of heat. The haptic information is then not detectable. One cannot explore the shape of a nettle or finger a hot object. The pain is too obtrusive for that.

In all these cases the sensation o pin is no doubt useful biologically, inasmuch as it dictates the avoidance of injury, but it is nonetheless not a perception. It carries no information about the world, only about the body of the observer, and it interferes with perception.

Physiological After-effects

The way in which the semicircular canals of the vestibular organ presumably register turns of the head was described in Chapter 4. The stopping of a prolonged rotation, however, induces an illusion of being turned in the opposite direction, the experience called vertigo. It is probably caused by the off-center position of the flexible cupula after rotation has ceased. Under rather special conditions, especially those of passive locomotion in vehicles, the capability of the vestibular system to register starts and stops is exceeded. The after-sensation of rotation when real rotation has ceased is a consequence of the structure of the organ and of the way it works. When the cupula regains its null position, the illusion ceases. The purely perceptual illusion, of course, is mixed with motor disabilities of posture, of equilibrium, and of accurate pointing with the hand.

The illusion of the water which feels cold to a warm hand but warm to a cold hand was described in Chapter 7. It probably results from the fact that the information for the perception of warmness-coolness consists of the direction of heat flow at the skin, inward or outward, an form the fact hat the skin tends to reach the same temperature as the medium in which it is maintained. The illusion is an after-effect of temperature adaptation. It is a consequence of the way the system works; the physiological zero between warm and cool being temporarily out of calibration.

The after-sensation of a patch of color in the visual field but not in the [p. 306] visual world, appearing in whatever direction one looks, is an illusion of visual perception that results from the way the retinal photoreceptors work. It is caused either by a high-intensity patch of light on the retina for a short time or a fixation of differential light for a long time. There are several types of this after-sensation, positive and negative, reflecting the complexity of the photoreceptive process. This statement does little justice to the careful study of these sensations and their implications for the photoreceptive theory of color vision, but it is sufficient for our present considerations of perceptual illusions.

In the usual course of events, these after-sensations do not seriously interfere with the getting of information by their respective perceptual systems. They only distract the attention from the registering of objective facts. When they are very strong, however, they can incapacitate the observer.

The Obtruding of Sensation on Perception

We may now consider some often-debated examples of phenomenal experience that seem to be midway between sensory impressions and percepts. They are called cases of incomplete perceptual constancy. One such case is the coin that is really circular but appears elliptical, and another is the railroad tracks that are really parallel but appear to converge. In the theories of sensation-based perception the explanation is offered that the conversion of the pictorial sensation into the tridimensional perception is incomplete, and that a compromise results. These cases of incomplete constancy seem to pose very real difficulty for the present theory since they imply that objective facts cannot be fully registered by the perceptual system.

I have argued that the coin does not always look elliptical and that the tracks do not always seem to converge (surely not to the locomotive engineer!). But I do not seem to win this argument, for most people say they do, and the results of experiments on shape constancy and size constancy commonly bear them out. Must the conclusion be that the shape of objects at a slant and the size of objects receding in the distance is necessarily a compromise between visual sensation and perception? This would contradict the conclusion that perception can be independent of sensation, depending only on the pickup of invariants that specify shape and size. I would prefer, therefore, to interpret the facts of incomplete constancy in another way. I suggest that in certain conditions for the perception of the layout of things, visual sensations obtrude themselves on the perception of true layout and cause the illusions of seeing partially in perspective. Putting it another way, sometimes we attend to the pictorial projections in the visual field instead of exclusively to the ratios and other invariants in the optic array. The pictorial mode of [p. 307] perception (Chapter 11) then asserts itself, since pictorial attention interferes with attention to information. The compromise is not between the raw data and the complete processing of this data, but between two alternative kinds of attention.

Some sorts of visual sensation, especially linear perspective, are very obtrusive, the more so when attention has been educated to it by having learned to draw pictures. The result may be the illusory appearance of foreshortened surfaces and decreasing size with distance. When this attitude is adopted, the information for the slant of the coin becomes a sensation of elliptical shape and the information for the recession of distance becomes a sensation of angular convergence to a vanishing point. I do not know of any good evidence to show that animals or young children are subject to these illusions of perspective.

...............................................................

Do young children see objects in perspective?

When Helmholtz was a grown man he wrote about an event in his childhood supporting his assumption that visual sensations are what the inexperienced eye and brain provide. Having been taken to Potsdam and seeing people high up in the belfry of the church tower, he had exclaimed that they were dolls. From this incident he concluded that the seeing of objects in the distance as small is unlearned; seeing in proper size comes only with learning the clues for distance. But I would draw precisely the opposite conclusion form this story. The infant Helmholtz had been naively perceiving the constant scale of things all along; what this preconscious observer noticed for the first time was that men in the distance could be said to look small. Indeed they can to the sophisticated self-observer. Can we really believe that the young genius took the people to be dolls? Let us rather assume that he saw people but began to be aware of size perspective and of the puzzles of physiological optics. He later wrote a thousand-page treatise on the subject, so brilliant and convincing that his theory of "unconscious inference" still dominates the textbooks a century later.

...............................................................

Another example of a sensory illusion is the finger which "looks double" when held close to the eye. The dual sensations are the consequence of disparity of the two optic arrays (Chapter 9) and they are usually called double "images." A finger held at the tip of the nose with the eyes converging at a considerable distance yields crossed disparity, a maximum degree of it, in fact. Crossed disparity is information for the visual system specifying the nearness of the finger. Usually it is registered simply as information, but if one attends to his retinal sensations, after being trained to do so, one gets the curious illusion of two fingers, knowing full well that there is only one. [p.308]

The doubling of contours in the visual field can only be noticed when it is relatively strong. It is not usually obtrusive and it does not seem to interfere with the pickup of disparity information. In the present theory, the double sensations are an incidental symptom of binocular disparity. But the theories of sensation-based perception must assume a fusion of these double images in the brain, or the "sensorium," and this leads to all sorts of theoretical trouble.

Affter-effects of Habituation

One of the physiological after-effects described above was hat in which a hand is immersed in warm water for a minute. The feeling of warmth diminishes. Then if the hand is put in water at room temperature it feels cold. This is said to be a result of physiological adaptation, but it is analogous to many other kinds of adaptation or habituation that are called psychological. The principle seems to be that whenever opposites can be judged, an experience on one side of the scale tends to shift the judgment of what is neutral toward that side of the scale. There are innumerable instances of this central tendency in perceiving, judging, rating, and evaluating, at various levels of perception. At one level, a moderate illumination seems bright after one has been in darkness but dark after one has been in bright light. At quite another level, ordinary conversation seems brilliant after talking to dull people, but dull after talking to bright people.

The principle applies not only to the qualities of objects that John Locke called secondary, but also to some that he called primary. The secondary qualities were colors, sounds, tastes, smells, and feelings of worth and cold, and they were said to be only in us, not in the physical objects themselves. The primary qualities were shape, size, position, duration, motion, and solidity, and they were said to be in physical objects. Everything in this book, the reader will recognize, goes contrary to this doctrine of Locke's. It is plausible but pernicious, and the attempt to refute it was begun in the first chapters. The point of interest here is this: since after-effects in perception apply not only to colors, tastes, smells, and feelings of temperature, but also to shape, size, position, and motion, one reason for the doctrine breaks down.

Here are the facts. After looking at a curved edge for some time, a straight edge appears to be slightly curved the other way (Gibson, 1933). After looking at a surface slanted backward, a frontal surface appears slanted forward (Bergman and Gibson, 1959). After wearing prismatic spectacles yielding a whole family of abnormalities in surface perception, their opposites appear when the spectacles are removed (Köhler, 1964). The perceptual process for these supposedly objective qualities is not [p.309] different in this respect from the perceptual process for the colors and the temperatures of surfaces.

The illusory negative after-effects in these cases are clearly the consequences of the adaptation or habituation. This might be a tendency to reset the neutral values of perceptual qualities to a running mean of environmental values. Something like this is suggested by Helson's theory of "adaptation level" (Helson, 1964). Whatever the process, it is a realistic one and the occasional errors of judgment are incidental to it.

Overselective Attention

It was said earlier in this chapter that the perceptual systems did not get enough information to work with in some circumstances, such as fog and darkness. In other circumstances, however, they get too much information to work with. In an eventful environment with sights and sounds and smells and touches all around, the individual cannot register everything at once and his perception must therefore be selective. The modes of selective attention, in fact, define the principal perceptual systems. The number of different identifiable objects in different directions may be enormous, and no one can look at them all. The world is often like a three-ring circus to a child - too many things happening too fast for him to comprehend them.

In the face of this situation, an expert perceiver develops a highly economical strategy of perception. This was described at the end of the last chapter. After things are discriminated and their properties abstracted, their number is reduced to a few categories of interest and the subcategories or cross-categories are neglected. At this stage only the information required to identify an object need be picked up and all the other information in the array, whatever makes it unique and special, can be neglected. Hence the percept of the object becomes a mere caricature or schema of what it would be if the perceiver took the time to scan the optical structure of the object. When he gives it only a glance, he neglects available information just as a tenth-second display of an object reduces available information. The tachistoscope forces observation at a glance.

There is great danger of error, we may now note, in this kind of economical perception. The object may in fact be unique or special, that is, an exceptional one that is not in one of the observer's categories of interest. An overselection of information has occurred. What the object really affords may be missed and what the observer perceives it as affording may be mistaken.

The danger of schematic perception is not so much that the percept is a caricature of the object and therefore an imperfect representation. A [p. 310] perception is never a representation in the first place. The danger is that the caricature may not be a good one. A clumsy caricature by a poor artist is not misleading because it is a distortion, but only because the information it conveys is wrong. The prejudgment involved in a skeletonized percept is a necessary result of the selective attunement of a perceptual system. If the prejudgments of conceptual attention are elaborate enough hey will not get the observer into trouble. Prejudices and stereotypes are misleading when the interests of the observer are narrow stereotypes are misleading when the interests of the observer are narrow or malicious. The prejudices of the open-minded observer are another matter.

A Classification of Illusions

When our early ancestors first notice the images in a pool of water, or the shadows of things, and especially when they began to make pictures, we may fairly assume that they became puzzled about the problem of appearance and reality. For these appearances are not real things; they are ghost-like, as the simplest of tests show. They are illusions. Illusions, as the Latin root of the word suggests, mock us. There are many kinds, and some are difficult to explain. It is difficult even to decide just what n illusion is. How, for example, does an illusion differ from an hallucination? Can we now, on the basis of what has been said, define and classify illusions?

On the present theory, illusions, like misperceptions in general, should tend to fall into two major types, objective and subjective. I will suggest that those of the first type are caused by information from artificial sources, by the deflecting of light rays, by contradictory information from pictorial sources, and by obscure combinations of information in geometrical drawings. Those of the second type, on the other hand, arising from deficiencies of perception, seem to be caused by such factors as the after-effects of excitation, insufficient specialization of receptors, and internal excitation of the nervous system. These seven classes of illusions are probably not exhaustive. They illustrate the possibility, however, of a general theoretical approach to a difficult and confusing problem in the study of perception.

Artificial Sources

Perhaps the commonest illusions are representations or reproductions. These were defined in Chapter 11. The faithful picture, the painting, the wax flowers, the statue, and the model are examples. The rule is that if an artificial source of stimulation conveys information equivalent to a natural source, the perceptions will be to that extent equivalent. [p. 311] For vision the same structure (or transformation) of an optic array, whatever its source, will always afford the same perception. The virtual object behind mirror is also the result of this rule.

The motion picture is notoriously a case where the moving event is only apparent. Nothing in the window really moves, but insofar as the optical transformations in the light from the screen are the same as the transformations in the light from the event to the camera, an illusion will result. The virtual event may be highly convincing, as when the young child is distressed by the shadows of the hero and the villain knocking one another down with a display of violent gestures and facial expressions. He presumably has not learned to distinguish illusion form reality by spanning all the available information, including that outside the screen and that prior to or subsequent to the movie.

Onstage fighting, of course, may be even more fighting to the child than motion-picture fighting.

The optical motions produced by the gadgets used in psychological laboratories to isolate, control, and display them are also virtual but not real. The material rotation of a Plateau spiral (Figure 14.7) causes its optical array to undergo expansion or contraction. The device of a rotating spiral behind a slot causes an optical motion of linear translation. The motions of the apparatus are entirely distinct from the motions of the light. The rectangular room seen when trapezoidal room is viewed from the proper peephole is another example. So are all the varieties of sound reproduction. If the fidelity of the system is high, there can be virtual orchestras, singers, or poetry readings.

The Bending of an Optic Array by Reflection or Refraction

The image of oneself in a pool or mirror, in fact the whole virtual scene, is caused by regular reflection of the pencil of rays comprising an array (Figure 10.15). The mirage of unreal trees and buildings in the desert is said to be due to regular reflection at a layer of heated air, following the same principle.

The apparent displacement of the visible environment and its objects by prisms in front of the eyes is due to refraction, which is another sort of bending of a pencil of rays. Apparent reversal of the world can be obtained by refraction and internal reflection in a right-angle prism. The straight stick that appears bent in water is due to refraction.

If the differential scattering of light of different wavelengths by particles in air is reduced to a multitude of reflections and refractions, then the apparent blueness of mountains (which are really green) in the distance is indirectly due to this cause. [p. 312]

Contradictory Information from a Picture

The illusions that can be described as the seeing of two alternative things in the same place were explained in Chapter 11 as cases of equivocal information in the same array. The reversibility of figure and ground and of perspective depth were illustrated. Other examples of hidden representations and various sorts of puzzle pictures are familiar to painters and psychologists. The contradictions of optical structure that can be incorporated in a drawing or painting are endless, and it seems to be fashionable just now for artists to explore them.

The Geometrical Illusions

There is a large class of visual illusions, known to psychologists for a century, that do not involve representation but only the perception of [p. 313] the properties of lines, curves, and geometrical figures. They involve the judgment of apparent lengths, sizes, angles, and areas, and of rectilinearity, parallels, and the like. These are variables of optical structure, of structure as such in the present terminology, but they are variables of relatively low order. They come from plane geometry, which, I argue, carry most of the information about the world. Length, size, and angle are basic variables of visual sensation or perception, but that assumption has here been challenged.

According to Titchener (1906, Vol. I, Part 1), a geometrical illusion is "a perception which differs in some way from the perception which the nature of the visual stimuli would lead us to expect" (p. 151). The stimuli compared in the Müller-Lyer figure are essentially two lines of equal length. We would expect equal lengths to appear equal but they do not. Similarly, we would expect segments of the same line to appear co-linear, and parallels to appear parallel, for these are properties of the stimuli, and presumably of the corresponding sensations. These are illustrated in

But the information for length of line, I have argued, is not simply length of line. To suppose so is to confuse the picture considered as a surface with the optical information to the eye. A line drawn on paper is not a stimulus. The stimulus information for the length of a line is altered by combining it with other lines. We should never have expected equal lengths to appear equal when they are incorporated in different figures. Only if we can isolate the two line segments from the wings and arrowheads in the Müller-Lyer illusion should they appear equal, and this would require a very special kind of selective attention.

The question of why one line looks shorter than the other is no longer of major importance for the theory of perception if line segments as such are not components of perception. The answer depends on discovering combinations of information in line drawings. Of all the many theories of the Müller-Lyer illusion, the one most nearly consistent with our hypothesis is one of this sort: the left-hand figure contains information for the ridge of a roof seen from above, while the right-hand figure contains information for the ridge of a roof seen from below. The apparent sizes of the two ridge-lines depend on their apparent distances in accordance with the general principle of perception of size-at-a-distance illustrated in Figure 14.9.

If this hypothesis is valid, the geometrical illusions are not subjective phenomena as they have always been taken to be, but instead are special cases of the information in variables of optical structure as displayed in drawings. [p. 314] [p. 315] [p.316]

After-effects of Excitation

We now come to those classes of illusions that are genuinely subjective phenomena. The after-sensations of light and color, of warmth and cold, and of rotation were described in the last section, As explained, they result form unusual conditions of stimulation, starting at a source, change of medium, and passive rotation,, respectively. The neural input continues after stimulation has ceased. The perceptual after-effects of habituation, the apparent curvatures and motions of the world described, are a little different. They are ultimately but not so obviously physiological in origin. Distortions of the shape and motion of things look as if they were external but they are just as dependent on neural processes as the patch of color that moves with the eye or the spots that sometimes appear with migraine.

Insufficient Specialization of Receptors

The so-called flashes of light that appear when the skull is severely jarred are a result of mechanical stimulation of the visual nervous system, that is, of cells that are supposed to respond to light. Pressure on the eyeball can also excite the retina. This perceptual system is fairly well cushioned against bumps, but some are too much for it. Electrical stimulation, which fortunately is rare in life, will excite any receptor mosaic or any nervous tissue - eye, ear, skin, or muscle. Illusions of light, sound, touch, or action due to electric shock are not encountered unless one is the subject of n experiment. The photoreceptive equipment is well specialized for photoreception, but not perfectly so, and mechanical or electrical energy can touch it off.

None of the perceptual systems is perfectly specialized for the pickup of information. If there is no limit to potential information, perfection of discrimination is not to be expected. The tasting system, as noted in Chapter 8, is subject to a kind of illusion inasmuch as some harmful poisons do not arouse distaste, and conversely, some harmless emetics do arouse distaste. For an omnivorous species like man, not all environmental substances are differentiated with respect to their nutritive value. There are too many substances. Some things that are not nutritive are palatable, and some things that are nutritive are unpalatable.

Internal Excitation of the Nervous System

The false feeling of pain or other sensation in an amputated limb is an impressive demonstration of the fact that perception depends on impulses in nerve fibers and that nerve fibers fire in sequential chains. Where the excitation begins in the case of the feeling of a "phantom limb" is not known, but it is known that nerves are discharged by abnormal causes, [p. 317] especially after injury, and sometimes long after. Drugs can have a direct effect on nervous tissue, as in the false colors of things observed after taking mescaline. And with alcohol poisoning, or the new "psychomimetic" drugs, full hallucinations may arise as a result of direct action on the brain.

Hallucinations, it is said are accompanied by a "feeling of reality," whereas illusions are not. Just what the feeling of reality consists of has not been established. Probably it is graded in degree. One clue to this gradation might be obtained by considering the continued efforts that have been made since the invention of photography to "add realism" to a picture, to make it lifelike, in short to make this classic type of illusion as much as possible like a perception (Chapter 11). The stationary, black-and-white screen picture produced by the early "magic lantern" was soon given color, and the fidelity of color representation has since been radically improved. The scope of the screen has been increased in the attempt to create a panorama. The stereoscopic picture was invented in the effort to enhance the perception of depth. The representation of sequence and transformation was achieved by cinematography. The sound film supplied auditory representation synchronized with the visual representation in recent times. We now have the panoramic motion picture in color with sound.

These developments have all made available more information to the main perceptual systems, vision and hearing. The perceiver does indeed get an approximation to first-hand experience nowadays, especially if the motion-picture camera takes the position of an actor in the scene.

Nevertheless, the modern motion picture is not an hallucination. It is still mere illusion. All proprioception is absent except for eye movements. The perceiver is passive. He sits in a chair. He is not fully surrounded by the environment represented on the screen. He is not fully surrounded by the environment represented on the screen. He cannot alter what will happen in the virtual world. Even though he may be given the experience of walking, approaching, inspecting, and riding in vehicles, it is not his experience for he did not get it for himself; most of it is imposed, not obtained. The perception of real world cannot and never will be completely imitated, for in the real world the perceiver can always find out things for himself and the more he explores the more he will find.

The malfunctioning of the perceptual systems which leads to true hallucinations, as in serious mental diseases, is probably due to some kind of inhibition of perceptual exploration with a shutting off or rejection of the current input of perceptual information. [p. 318]

Summary

In this final chapter an attempt has been made to explain misperceptions and illusions as deficiencies in a process which usually comes out right but which for various reasons sometimes goes wrong. A tentative list of these reasons has been offered.

First, the available stimulus information for perception can be inadequate. The energy may be minimal, or the structure of an array may be blurred, or it can be masked, or the information in structure may be contradictory. The interval during which energy is available may be cut short, or the angular size of the optic array may be narrowed down. The spatial and temporal structure of light or sound can be displaced, or biased, or distorted. Mirrors, prisms, and lenses can alter the structure of light and modern electronic gadgetry can alter the structure of sound.

Second, the physiological process of information pickup can be deficient. Even normal perceptual organs fail to work at high intensities of energy. Their capacity can be overstrained. They get out of calibration after abnormal stimulation and the recovery process then yields aftersensations. The physiological action of the receptors as such may be so obtrusive as to distract the observer. The action of a whole system may be subject to perceptual adaptation followed by after-effects. And finally there are errors of attention due to false expectations.

These causes of misperception provide at least a systematic basis for the classification of what have been called illusions. Some illusions should be ascribed mainly to external conditions and some mainly to internal. Examples of the former are experiences aroused by artificial sources (images and reproductions, including motion pictures); experiences resulting from the bending of optic arrays by natural or artificial means; the ambiguous experiences from contradictory information in a picture; and the anomalies of the "geometrical" illusions. Examples of the latter are the after-sensations and perpetual after-effects, the sensations caused by inappropriate mechanical or electrical stimulation, and the little understood cases of purely internal excitation of the nervous system. The last-named include hallucinations.

See also:

"Introduction" (pp. 1-6).

"Chapter XIII: The Theory of Information Pickup" (pp. 266-286).