Chapter

Introduction to Perception
O U T L I N E Perception Is a Biological Process 2 Perception Involves Action 7 Why Study Perception? 9 Complementary Approaches to the Study of Perception 14 Recurring Themes 25 Summary and Preview 27 Key Terms 28

he world is filled with objects and events that generate a torrent of potential information. Though much of that information is irrelevant to peoples daily needs, some is absolutely essential. To exploit this information effectively, human beings are equipped with specialized machinery that captures the information and translates it into the language of the nervous system. The brain refines this translated information into neural descriptions of behaviorally relevant objects and events in the environment. Some of these descriptions reach conscious awareness, allowing us to formulate deliberate plans for subsequent intractions with those objects; other descriptions guide immediate or reflexive reactions to objects and events. Perception puts us in contact with the world we live in; it shapes our knowledge of that world, and knowledge is power. Our chances of survival improve markedly if we can detect objects and events in our environment and if we can, then, distinguish the safe from the dangerous, sort out the desirable from the undesirable. Knowing about our world allows us to predict the consequences of our actions, a critical skill in a constantly changing world. Perception doesnt have to provide us with an accurate view of the world, perfectly detailed in every respect. What is crucial is that perception provide us with a useful

view of the world, where useful means being able to interact safely and effectively within our environment. As you will learn, perception accentuates the important and diminishes, or even ignores, the irrelevant. Perception may even misrepresent an objects true appearance, if that misrepresentation improves our chances of interacting effectively with that object. Sensory illusions represent an example of misrepresentation, and youll be seeing many examples of sensory illusions as we move through the chapters. These illusions can be construed as perceptual mistakes that, paradoxically, work in our favor. The environment generates a powerful stream of sensory information. In fact, there are so many objects and events that our perception cannot possibly process and respond to each one. Nonethelessand fortunately for usperception does work. Our perceptual systems overcome this potential sensory overload in several ways. For one, the world in which we live is full of regularities dictated by the physical nature of matter and energy. Those regularities make it easier to detect objects and to discriminate one object from another. For example, our visual system evolved in a world where light nearly always comes from above and, consequently, we unconsciously use that knowledge to interpret shapes of objects based on shadows. (You will see an example of
1

Chapter One Introduction to Perception

this constraint in Chapter 8.) Perception can exploit this and many other environmental regularities to make educated guesses about what in the world gave rise to a particular pattern of sensory stimulation. There is a second effective way that perception deals with the environments overwhelming complexity: it simply ignores much of what is going on in the world. Much of what the environment has to offer is simply of no interest to us. It is not important, for example, to sense the minute electrical fields generated by other biological creatures (including other people). Humanskeen sense of vision makes electroreception superfluous for members of our species. Consequently, our perceptual systems are tuned to those sensory events that are biologically relevant (or, more correctly, were biologically relevant to our primate ancestors). As you will learn throughout this book, animal species differ in their behavioral goals, and the goals of some species require sensitivity to sensory stimulation that falls outside our reality. Perception arises from a complex interplay of mutually interdependent events. To understand perception completelyand no one yet doesrequires knowing all the components involved in the process and the ways those components interact. To begin, we must specify the nature of the environment in which we live, for this environment determines what there is to perceive. Aspects of the environment are specified using terms derived from physics, because stimulation comes in various forms of physical energy: thermal, mechanical, chemical, acoustic, and electromagnetic. The physical energy that initiates the chain of events is called a stimulus (plural, stimuli). Next, it is necessary to understand how the nervous system converts patterns of physical energy into neural events. Known as sensory transduction, this conversion process requires an understanding of the specialized sensory receptors (such as those contained in eyes and ears) that convert physical energy into bioelectrical signals. Once this transduction has been achieved, objects and events are represented solely as patterns of neural impulses within the various sensory nerve fibers. From this point on, all further elaboration and editing of the sensory information must be performed using this neural representation. Of course, the brain plays a central role in perception. So a full understanding of perception also requires knowing about various brain areas specialized for processing patterns of neural impulses arising from the various senses. A full understanding also requires knowledge of how the activity distributed among those many areas is combined to form our unified sense of the world. How, for example, do the neural signals conveying information about the sounds from an object combine with the signals generating the visual impressions of that object? In addi-

tion, we need to discover how neural activity signaling the presence of objects is used to control our behavioral and emotional reactions to those objects. A complete account of perception must incorporate a thorough description of the appearances of objects and events: we have to be able to describe systematically the sights, sounds, smells, and tastes that populate our conscious experiences. In addition to describing how things appear to us, we must also specify how our abilities to detect, discriminate, and recognize objects are governed by the information available to our senses. And, in a similar vein, we must understand the behavioral consequences of sensory stimulation, for our actions will modify those very patterns of sensory stimulation. These tasks represent formidable challenges. Not surprisingly, scientists have developed diverse techniques for systematically cataloging the performance of our perceptual systems and relating that information to patterns of physical stimulation. The enterprise of relating physical stimulation to perceptual events is known as psychophysics. By specifying the relation between physical and perceptual events, psychophysics provides important clues to understanding the various steps leading from objects and events to perception. Perception constitutes a whole sequence of events, beginning with things that happen in the physical world external to the perceiver. From that start, perception proceeds through the translation of external events into patterns of activity within the perceivers nervous system, culminating in the perceivers experiential and behavioral reactions to those events. Those reactions, in turn, can affect the very same sensory events that triggered those reactions. All of this forms a closed loop in which perception alters behavior, and behavior, in turn, alters perception. This dynamic, continuous loop is schematized in Figure 1.1. Lets now consider several important implications of this way of thinking about perception.

Perception Is a Biological Process

In this book, we approach perception as a biological process. To be perceived, any information about events in the world must be registered by the sensory nervous system. The noted neuroscientist Vernon Mountcastle has vividly described this constraint:
Each of us lives within . . . the prison of his own brain. Projecting from it are millions of fragile sensory nerve fibers, in groups uniquely adapted to sample the energetic states of the world around us: heat, light, force, and chemical composition. That is all we ever know of it directly; all else is logical inference. (1975, p. 131)

Perception is a Biological Process

FIGURE 1.1 | Perception registers and interprets sensory information from the environment, in this case light, that guides behavior, which, in turn, shapes the nature of input to the senses.

Reflected light

Eye

Image in eye

Neural activity

Motor activity

Reaches for cat

Cat moves

Reflected light

External object

Sense organ

Brain

Muscles

External object

Mountcastle points out that sensory nerve fibers provide our only link to the external world; they alone represent our communication channels to reality. If environmental events fall outside the sensitivity range of our sensory channels, we will not experience those events directly. Now, it may be possible to detect some of these events indirectly, using specialized instruments that work in one of two ways. Some instruments amplify physical energy, making otherwise weak, undetectable signals strong enough to stimulate the senses. A microscope, for example, can magnify objects too small to be seen by the naked eye. This is the only way were able to know what bacteria actually look like. Other instruments convert energy that is outside the normal bounds of any of our senses into a form that falls within those bounds. Geiger counters can warn about the presence of radioactivity, a form of energy that cannot be sensed directly by most creatures; the Geiger counter converts imperceptible radiation energy into audible sound or visible deflections of a gauge. Box 1.1 actually shows you what ordinarily invisible optical information may look like. Whether amplifying or converting energy, these specialized instruments all perform the same functionthey extend the reach of our sensory systems into realms of physical reality that are normally beyond our perceptual grasp.

It may be difficult to accept that your rich perceptual world encompasses only a tiny, restricted portion of the objects and events in the natural environment. Because ones conception of reality is so intimately determined by subjective experience, it seems unnatural to distinguish between ones perception of the world and the world itself. Roger Sperry (1964) cast this distinction with beautiful clarity: Before brains there was no color or sound in the universe, nor was there any flavor or aroma and probably little sense and no feeling or emotion. To understand perception fully, then, you have to make this distinction. Perhaps a few examples will enable you to appreciate what we mean by the limited scope of your perceptual world. Consider, for instance, how different species of animals experience the world. It is well documented that not all animals have the same sensory systems. Consequently, various species have access to different universes of physical events (Hughes, 1999). You probably know that dogs can hear sounds in regions of the frequency spectrum where humans are deaf. You may not know, however, that bees can navigate using a quality of light, polarization, that is outside the realm of human visual experience; sharks hunt their prey by following electrical trails given off by their potential meal; bats use

BOX 1.1

Seeing the Invisible

is about as insensitive to infrared as the human eye. That is why pictures taken with such film look normal. The photograph on the right shows the same scene taken with film that is sensitive to infrared; it reveals things in the scene (areas of heat and cold) that humans ordinarily would not see. Thus, for instance, the water in the right-hand photo appears dark because it is cold. Although the differences between these pictures are interesting, we cant really claim that the photographs provide much insight into the experiences of those infrared-sensitive snakes. In some cases, their infrared-sensitive organs are not even part of their eyes. Its more likely that they feel infrared energy rather than see it. The photographs do remind us, though, that the human perceptual world is limited to that sample of objects and events with physical energy falling within the range of sensitivity of our sensory systems.

Have you ever heard the adage, Theres more to the world than meets the eye ? Its certainly true, for our world is awash with radiant energy that falls completely outside the range of sensitivity of our eyes. Take, for example, electromagnetic radiation in the portion of the spectrum called infrared. This form of radiation is usually associated with heat, including the body heat of living creatures. If infrared is sensed at all, it is experienced as warmth on the skin. Although some animals, notably certain snakes, have specialized sense organs that allow them to detect and respond to objects on the basis of the infrared energy radiated by those objects, humans are fairly insensitive to infrared. To give you some idea of what it might be like to see infrared radiation, we have prepared the two accompanying photographs. The photograph on the left shows a scene taken with ordinary black-and-white film; this film

self-produced, ultrasonic echoes to navigate in complete darkness; snakes can detect and orient toward infrared radiation; trout have tiny biomagnets in their heads that let them exploit the earths magnetic field to orient their navigation; zebra finches rely on ultraviolet light to select mates; moths can sense chemical substances, that are entirely odorless to humans. In general, there is no single environment in which all animals live. Members of different species interact with their physical worlds in ways that reflect their own unique requirements and capabilities.
4

How can we get a glimpse of the perceptual world of other animals? In a now famous essay, philosopher Thomas Nagel (1974) reasoned that one must actually become that creature in order to understand what it is like to be that creature. (Nagels essay is available on the Web, and can be accessed from [Link]/blake5.) In fact, Nagel wasnt the first to endorse this idea. In T. H. Whites book The Sword in the Stone (1939), Merlyn the magician wanted Arthur, future King of England, to experience a variety of perspectives on the world he would eventually rule. To provide that experience, Merlyn

Perception is a Biological Process

magically turned Arthur into a bird, a fish, an ant, and a badger. Although a magician living in the Middle Ages couldnt possibly have read Nagels twentieth-century essay, Merlyn nonetheless was anticipating Nagels philosophy. Although we are not magicians like Merlyn, we want to enable you to imagine what its like to be another creature, particularly a creature with perceptual abilities very different from yours. So throughout the books chapters youll be challenged to imagine the perceptual world of creatures very different from humans. For that matter, not even all humans experience the same perceptual world. Some people, for example, have eye defects that prevent them from experiencing the full range of colors that most people see; youll meet some of these people in Chapter 7. Many elderly people are unable to hear some high frequency sounds that are clearly audible to younger individuals; youll learn about the consequences of this deficit in Chapter 11. Certain people cannot taste one of the bitter substances in coffee or grapefruit juice; Chapter 15 tells you why these people have limited taste perceptions. These and similar examples appearing throughout this book underscore the dependence of perception on the function of the sensory nervous system. Recognition of perception as a biological process underscores another important point: perception entails symbolic representations. A symbol is something that stands for something other than itself. Hermann Helmholtz, an influential nineteenth-century contributor to physics, physiology, and perception, emphasized this point when he wrote: Our sensations are for us only symbols of the object of the external world, and correspond to them in such way as written characters or articulate words correspond to the thing they denote (quoted in Park, 1999, p. 8). A road map, for instance, is a symbol of the highways and terrain over which you may wish to travel. Running your finger along some highway on a map is very different from traveling the actual road, but the map symbolizes the reality. The words on this page are symbols, denoting objects, actions and relations among them. When you read the letters Z-E-B-R-A, they symbolize a kind of animal with which youre familiar. The letters dont magically conjure up an actual zebra, but they surely might conjure up in your minds eye the image of a horselike animal with black and white stripes. Words are useful for spoken and written communication because they have a sufficiently narrow range of referents that can be recognized by large groups of human beings. Your perceptual experiences are associated with characteristic patterns of neural activity in your brain (hence we say that perceptual states are produced by brain

states). This fact makes perception a symbolic process. Suppose you hear a bird chirping outside your window. Your perceptual experience of that sound is certainly not the same as the actual event that gives rise to the sound, in this case the acoustic energy produced by the bird. However, your perceptual experience does represent important qualities of that acoustic event. In the case of perception, the symbols are not the sort we usually think ofa screech of brakes, the trill of a bird, the crescendo of an orchestra. Instead, the symbols are the various brain states that stand for these sounds. Just like other kinds of symbols, however, the properties of these symbols are not the same as the properties of the things being symbolized: in your brain, the representation of a loud explosion is neither loud nor explosive. What the neural responses can retain, however, is important information about the spatial or temporal structure of the objects and events the neural symbols represent. Thus, the beginning and end of neural patterns in your brain evoked by the birds chirping will generally coincide with the beginning and end of the birds call; the temporal structure of the neural activity mirrors the temporal structure of the event. Viewing a single, isolated star in the sky will create a more spatially compact pattern of neural activity in your brain than will viewing a whole cluster of stars; the spatial structure of the neural activity thus mirrors the spatial structure of the objects being viewed. At least while youre awake, your brain expects to receive more or less continuous sensory input stimulated by events in the external world. When that input is reduced or eliminated, the sensory systems lapse into a kind of disorderly conduct that can be quite bizarre. Input to the brain can be shut off by placing a person in an environment that virtually eliminates sensory stimulation (Bexton, Heron, and Scott, 1954; Siegel, 1984). Although this may sound like a relaxing, pleasant situation, its not: placed in sensory deprivation, people become anxious and begin to hallucinate. A less severe but nonetheless disturbing condition arises when a particular sensory channel has diminished function because of disease. For example, in about 10 to 15 percent of people with serious eye disease, the impaired vision evokes realistic and complex visual hallucinations (Schultz and Melzack, 1991). In this condition, known as Charles Bonnet syndrome, the hallucinations come from impaired sensory systems, not from dementia, as was once assumed. Similarly, following amputation of a body part, many people experience phantom limbs, compelling and very painful hallucinations that the missing body part is still present. The visual hallucinations and the phantom

Chapter One Introduction to Perception

limbs are both generated by activity within the brain, in this case, activity uncoupled from normal sensory input. But lets return to brain activity that produces normal perception, triggered by sensory events. It is really remarkable how deceptively simple and self-evident our perceptions of the world seem. If you were asked to describe how you solve a jigsaw puzzle, you might give an account of the steps involved: Turn all the pieces right side up, arrange the pieces into little piles according to color, and so on. Likewise, many of you can probably describe what you do when brewing fresh coffee (Grind the beans, place a filter in the coffeemaker, etc.). But what answer would you give to the following: How do you read the words on this page? Most people would probably say, I open my eyes and look at the page. Of course, this simple description belies the truly complex nature of the process, but it does dramatize the point we wish to make: normal perception occurs rapidly and without much effort. But we shouldnt be fooled. Even the simplest perceptual experiences result from a complex series of neural events involving extensive interactions among numerous brain cells. These interactions, which bear a formal resemblance to interactions in an electronic circuit, can be thought of as computations. The computations that shape the symbolic representations in the brain work on environmental information picked up by the eyes, ears, and other sensory organs. Using this information, the brain computes the properties of objects and events (such as their size or their distance from the point of observation). The philosophy that guides contemporary work in perception is termed materialismit asserts that perceptual experience depends on the operation of the nervous system, with no requirement for the involvement of some noncorporeal force. The materialistic viewpoint has been well expressed by the late Roger Sperry, a Nobel Prizewinning brain scientist. According to Sperry, perceptual experience is a functional property of brain processing, constituted of neuronal and physicochemical activity, and embodied in, and inseparable from, the active brain (1980, p. 204). Although materialism holds that perception is based on neural events in the brain, it does not imply that one could dissect a brain and thereby locate those experiences. Again, Roger Sperry put it quite well:
Once generated from neural events, the higher order mental patterns and programs have their own subjective qualities and progress, operate and interact by their own causal laws and principles which are different from and cannot be reduced to those of neurophysiology. (1980, p. 201)

To illustrate further what he had in mind, Sperry offered the example of a wheel rolling downhill. The wheel
carries its atoms and molecules through a course in time and space and to a fate determined by the overall system properties of the wheel as a whole and regardless of the inclination of individual atoms and molecules. The atoms and molecules are caught up and overpowered by the higher properties of the whole. One can compare the rolling wheel to an ongoing brain process or a progressing train of thought in which the overall organizational properties of the brain process, as a coherent organizational entity, determine the timing and spacing of the firing patterns within its neuronal infrastructure. (1980, p. 201)

In other words, though ones experiences have a physical basis, they cannot be entirely reduced to a set of physical components; equally important are their spatial organization, what they communicate to one another, and how both spatial organization and communication change with time. If another analogy would help, consider what would happen if you took a television set completely apart and examined all its components in an effort to understand how it worked. The proper function of the television set demands a particular spatial arrangement of parts as well as a certain sequence of signals in time. The secret of the sets operation would have completely eluded you and could not be found in the pile of parts left after the set had been dismantled. And certainly from the parts alone, it would be impossible to deduce the function of a television set. Not everybody agrees with the materialistic perspective. Some prominent scientists, including the late Sir John Eccles (1979), another Nobel Prize winner, subscribe to an alternative view. This alternative, dualism, is often associated with the seventeenth-century French philosopher Ren Descartes. Dualism holds that perceiving (like any mental function) is not solely a phenomenon of the physical brain. Instead, it also entails some special, nonphysical substancethe mind or the soul that interacts with the brain. Many people find dualism persuasive because they are unconvinced that perception, a personal, subjective experience, can be fully explained by brain processes, which are certainly not experiences. They object to materialisms basic claim: a quantity of one sortneural activitycan cause a quantity of so different a sortperception. According to philosopher John Searle (1987), though, no logical barrier prevents cause-and-effect relationships between entities of radically different sorts. In fact, denying that such relationships are possible betrays a misun-

Perception Involves Action

derstanding of cause and effect itself. To drive his point home, Searle draws on examples from physics. Physicists commonly distinguish between large-scale macro phenomena and smaller-scale micro elements, postulating causal relationships between the two, even though macro and micro entities are quite different from one another. Take some examples offered by Searle. Heat and lightning are macro-level phenomena; molecular movements and electrical discharges are elements on the micro level. Physics teaches us that a macro phenomenon can be caused by the behavior of micro elements: we say that heat is caused by molecule movements or that lightning is caused by electrical discharge. Moreover, either of these macro phenomenon can be equated to the behavior of its micro elements. Therefore, we can say that heat is the mean kinetic energy of molecule movements or that lightning is an electrical discharge. Paul Churchland elaborated the main arguments against dualism (1988). Here well mention two of them. Against the claim that perception is independent of what happens in the brain, Churchland cites numerous instances in which changes in the brains condition dramatically alter the content and quality of perception. Throughout the following chapters, we give many examples of perception disordered by brain damage. Against the claim that perception is far too complicated to be the product of things as simple as nerve cells, research on neural networks shows that one can create extraordinarily complex, sophisticated systems out of very simple components, undercutting the need to postulate other, more intelligent agents. As a result, one can account for complex, intelligent aspects of perception without recourse to elements that are themselves complex or intelligent. Searle expresses the view most investigators in the field of perception have adopted:
Mental phenomena, whether conscious or unconscious, whether visual or auditory, pains, tickles, itches, thoughts, and all the rest of our mental life, are caused by processes going on in the brain. Mental phenomena are as much a result of electrochemical processes in the brain as digestion is the result of chemical processes going on in the stomach and the rest of the digestive tract. (1987, p. 220)

Perception Involves Action

Perceiving usually requires some action on the perceivers part. You look around in order to see, searching the visual environment until the desired object of regard is located. Likewise, to make a faint sound audible, you may turn your ear in the direction of the sound. When touching an object, youre better able to identify it if you actively explore it by moving your fingers over it. All these examples remind us that perception is an active process, an idea especially championed by James J. Gibson (1966), who expressed this idea succinctly when he wrote, We must perceive in order to move, but we must also move in order to perceive (1979, p. 223). Active perception accomplishes several goals. First, we sample our environment purposefully, rather than waiting on sensory events to drop into our laps, so to speak. Our sampling behaviorlooking, listening, touching, and so onis usually guided by our needs (Are you hungry, sleepy, afraid, or what?). Perception, in other words, has purpose. Once active sampling uncovers an object of potential interest, we must decide whether to approach it or to avoid it. Here, again, active perception helps us make intelligent decisions. By actively exploring an object were able to improve the quality of the sensory information we receive. As you investigate an unfamiliar object with your hand, your fingers are directed to the most informative parts of that object, looking for telltale signs of the objects shape, size, texture, and, ultimately, its identity. Having identified the object, you may elect to discard it or to keep it. The same kinds of exploratory activities occur with the other senses, including vision and hearing. Your behavior depends on what is perceived, and what is perceived depends on your behavior. Active perception, useful as it is, introduces a potential confusion that the brain must sort out, namely distinguishing self-produced patterns of stimulation from externally produced ones. An example should make this point clear. Hold the index finger of one hand in front of your face and look straight at it (see Figure 1.2). Now shift your gaze back and forth, looking to the left and the right of your finger. This action causes the image of your finger to sweep back and forth over the light-sensitive surface on the back of your eye. Now modify the exercise by moving your finger back and forth while staring straight ahead without moving your eyes. This also causes the image of your finger to sweep back and forth over the back of your eye. So here are two distinct situationsa stationary object and a moving objectthat can produce comparable patterns of eye stimulation. How

Within contemporary psychology and neuroscience, perception is an area of scientific research that has made major advances in explaining the relation between brain and mind. Adoption of the materialist position has greatly facilitated those advances.

Chapter One Introduction to Perception

FIGURE 1.2 | While holding your finger in front of your face, shift your gaze from the left to the right of your finger. As you do this, the image of your finger sweeps back and forth over the back of your eye, much the same as it does when you hold your gaze steady and move your finger back and forth.

Moving eyes

Moving finger

does the visual nervous system distinguish one from the other? Well postpone details of the answer until Chapter 9, but suffice it to say the brain solves the potential dilemma by keeping track of motor commands sent to the eye muscles. A neural copy of those commands can be sent to visual areas of the brain thereby vetoing the implication of signals arriving from the eye. This kind of sensorimotor feedback is probably used whenever we distinguish self-produced stimulation from externally produced stimulation. Perceptions links to action orientation produces an interesting distinction among the various senses that has to do with the proximity of the perceiver to the object of perception. Touch and taste require direct contact between the perceiver and the source of stimulation. Because of this restriction, taste and touch can be considered near senses. The sense of smell is also effectively a near sense, at least for humans. Volatile chemicals from an

odorous substance are diluted with distance, so smell works more effectively for substances in the general vicinity of the nose.1 In contrast, for us humans, seeing and hearing can be thought of as far senses, or distance senses. The eyes and ears can pick up information originating from remote sources. In this respect, they function like a ships radar. They allow one to make perceptual contact with objects located too far away for immediate grasp. They extend your perceptual grasp out into the world beyond your fingertips and your nose. Vision and hearing serve as able substitutes for actual locomotor exploration of the environment. These two far senses let you explore your surroundings vicariously. They provide ad-

Actually, this is not true for all species: some animals possess a highly acute sense of smell that allows them to detect odors over great distances, as youll learn in Chapter 14.

Why Study Perception?

vance warning of approaching danger, and they guide the search for friends and desired objects. In general, hearing and seeing open up to you the large world that lies outside your reach. Imagine how vulnerable you would feel if you were denied access to all information picked up by your far senses. Your whole world would shrink to the area within arms reach. You would be able to sense objects only when you touched them or when they touched you. It is not surprising, therefore, that blindness and deafness, losses of the far senses, are considered so devastating. Incidentally, this distinction between the near and far senses has an important behavioral consequence. Any crucial reaction called for by taste or touch must be executed swiftly. There is no time to decide whether a bitter substance is toxicyou spit it out reflexively. Nor do you first try to judge what is causing a burning sensation before you remove your hand from a hot object. In these cases, you act first and then consciously think about what it was that triggered your reflex action. However, in the case of the far sensesseeing and hearingyoure often dealing with objects located some distance away. This distance permits you the luxury of evaluating the potential consequences of your actions. One final note about active perception. The ability to explore the world actively requires fine motor control over the body parts used for exploration. You could think of your fingers as finely tuned calipers that adjust their grip force to suit the object youre holding. Youre able to do this because the muscles in your fingers contain specialized sensory receptors that gauge the strength of contraction and, hence, the tightness of your grip. Without those internal sensory receptors to monitor grip strength, youd have trouble squeezing the right amount of toothpaste from a tube, and you could crush a delicate flower before you knew what you were doing. Called kinesthesia, this internal sense of muscular contraction is intimately involved in coordinating all sorts of motor activities, including walking, dressing, eating, and typing. We wont dwell on the mechanisms of kinesthesia, but you should keep in mind that the nervous system also includes specialized sensory systems that monitor the internal perceptual states of your bodys interior parts.2
2

Why Study Perception?

Over the years, people have studied perception for a variety of reasons. Some of these reasons, as you will see, stem from practical considerations, such as the need to solve a particular problem. Other reasons do not involve practical concerns, but arise from simple intellectual curiosity about ourselves and the world we live in.

Practical Reasons for Studying Perception

The human senses evolved under environmental conditions in many ways quite different from those we now live in. Many of the challenges confronting the human senses today didnt exist in the more primitive environments for which these senses were designed. Its very important to know just what kind of perceptual demands can reasonably be placed on the human senses without compromising safety and sanity. There is an optimum range of sensory stimulation within which the majority of people work and play most effectively. Intense stimulationsuch as excessive noise, glaring light, and harsh smellscan impair immediate performance as well as damage the sensory nervous system. Through the study of perception, one can identify and correct potentially hazardous environmental conditions that threaten the senses and impair the ability to make decisions. In a related vein, studying perception enables one to design devices that ensure optimal perceptual performance. Just think how often each day you come in contact with devices designed to communicate some message to you. Traffic lights, alarm clocks, telephones, and video displays are just a few of the myriad inventions that people rely on during work, play, study, even sleep. To be effective, these devices should be tailored to human sensory systems. It would be unwise, for example, to use a highpitched tone as a fire alarm in a hotel because most elderly people have difficulty hearing such tones. Similarly, a traffic sign with blue lettering on a green background would be inefficient, because the contrast between these two colors makes letters more difficult to distinguish than, say, red letters on a green background. In general, we want the signs and signals in the environment to be easy to see and hear, and this requires an understanding of human perceptual capacities and limitations. Understanding perception also makes it possible to design devices to help individuals with impaired sensory function. Take hearing aids as an example. For decades hearing aids amplified not only the sounds the user wanted to hearsuch as a persons voicebut also other, unwanted sounds, such as traffic noises. Recognizing this

One could also construe internal regulatory mechanisms as sensory systems, in that they monitor physiological states such as hormone and glucose levels and blood oxygen. When those levels drift out of balance, compensatory reactions are triggered (e.g., a change in heart rate). Thus, these internal mechanisms also form sensory/action loops. We wont include internal monitoring systems in this book, for were concerned with sensory systems that allow us to interact with our external environment.

10

Chapter One Introduction to Perception

problem, Richard L. Gregory developed a procedure that selectively amplifies just speech sounds (Gregory and Drysdale, 1976). This invention, which is now in wide use, grew out of earlier work on the ears ability to respond selectively to particular sounds. And recently weve seen the development of devices that boost hearing by direct stimulation of the auditory nerve, using implants driven by speech processors embedded on microchips (Wilson et al., 1991). This device has been a boon to thousands of hearing-impaired children, but the design of such aids requires a solid understanding of mechanisms of normal perception. Also in a practical vein, companies in the food and beverage industry carefully test the perceptual appeal the taste, smell, and appearanceof their products before marketing them. Advertising, too, capitalizes on perception research to package and market products in ways that will bring those products to the attention of consumers. There are even claims that subliminal sensory messagespictures or words presented too briefly or too faintly to be consciously seen or heardcan improve ones memory or enhance self-esteem, although these claims are questionable (Greenwald et al., 1991). So far, our practical reasons have focused on human perception. But as the following examples show, there are solid reasons for studying animal perception, too. For one thing, animals can be trained to perform jobs that are beyond the sensory limits of humans. Dogs, because of their keen sense of smell, are adept at detecting odors too faint for the human nose. This is why dogs are frequently employed to sniff out illegal drugs or to trace the footsteps of a fleeing suspect. In other instances, knowledge of an animals sensory apparatus allows one to control that animals behavior. For instance, agricultural scientists control cotton bollwormsmoth larvae that damage cotton cropsby spraying crop fields with a chemical that fools adult male moths into mating with moths of a different species. This chemical overwhelms the smell cues that normally guide mating behaviors. As a result, the moths engage in promiscuous and ineffective mating behavior. As a final example, scientists study animals whose sensory capacity is impaired by congenital disorders or by some experimental manipulation such as sensory deprivation (Kaas and Florence, 1996). These studies, in turn, lead to valuable ideas concerning the bases and treatment of comparable sensory disorders in humans.

tinguish friends from foes and trying to locate the next meal. As civilization developed, these pressing demands relaxed. Consequently, civilized people enjoy the freedom to develop pastimes, such as the visual arts, music, and cuisine, that engage their perceptual machinery in pleasurable, amusing, and creative ways. Each of these pastimes involves the stimulation of the senses. Besides their immediate aesthetic and sensual qualities, these kinds of sensory experiences play an important role in the cultural heritage of societies. Through various forms of art, people share the joys and pains experienced by others, and they can savor vicariously the thrill of discovery that originally inspired the artist. In brief, art embodies much of a cultures wisdom and transmits that wisdom from one generation to the next by means of shared sensory experience. Interestingly, however, artistic creations can also provide insight into the nature and mechanisms of perception. After all, artistic works are the creations of people whose perceptual systems obey the same principles as yours; the artists eyes, ears, and brain are not fundamentally different from yours, either. So the artistic creations of those talented individuals, inspired as they may seem, must be guided by the same rules of perception that govern the way the rest of us experience the world. If art is going to communicate ideas and feelings to an audience, that artwork must work within the constraints of the perceptual systems of the audience members. At the same time, artists can exploit perceptual tricks to create capitivating effects, both in visual art and in music. Youll see examples of these tricks in Chapters 5, 7, 8, 11 and 12. In general, it has become increasingly evident that we can learn a lot about perception by studying art (Livingstone, 2002; Zeki, 1999).

Perception and Intellectual Curiosity

Practical and pleasurable concerns aside, learning about perception satisfies an intellectual curiosity about ourselves and the world in which we live. Perception can be regarded as each individuals personal theory of reality, the knowledge-gathering process that defines our view of the world. Because this perceptual outlook guides our mental and behavioral activities, we naturally find it fascinating to inquire about the bases of perception. At the same time, people hold some odd theories about perception that need to be replaced by knowledge grounded in fact, not in intuition. We were surprised to learn, for example, that a large number of college students believe that our eyes emit the light that illuminates objects in the environment (Winer et al. 2002).

Perception and Pleasure

In more primitive lifestyles (such as those of nonhuman primates) the lions share of perceptual processing was probably devoted to survivalbeing on the alert to dis-

Why Study Perception?

11

Natural curiosity leads to a variety of conjectures about perception. When looking at a newborn child, for instance, one cannot help speculating about what that infant sees and hears. Likewise, one is curious to know whether blind people really can hear sounds that escape the ears of sighted people. You may have wondered why colors seem to change depending on the time of day. As dusk approaches, greens take on a deeper richness, while yellows and reds lose some of their brilliance. And why does everyone effectively become color-blind in dim light? One would like to know what sensory cues enable a displaced pet to journey hundreds of miles, eventually returning to its old home. And one grudgingly marvels at how adept mosquitoes are at locating a persons bare skin in total darkness. People are intrigued by their everyday experiences and are curious about the bases of those experiences. This curiosity was long ago formalized in philosophy. For centuries, philosophers argued about how human beings can know the external world. Their arguments reflected a concern about the validity of sense experiences. Though our concept of the world derives from the information of our senses, can we rely on those senses to tell the truth? Might we not be deceived about the world? Perhaps, as Plato suggested in Book VII of The Republic, we are like prisoners in a cave, cut off from the world, so that we can see only shadows created by objects and events outside. In fact, from earliest times on, people have known that their senses were fallible. Realizing that sensory information was not totally dependable, philosophers became increasingly skeptical about anyones ability to know the world as it really is. This skepticism reached full bloom during the late seventeenth and early eighteenth centuries. During that time, the British philosopher John Locke (1690/1924) made a crucial observation: water in a basin can feel either warm or cool to the touch, depending on where your hand has just been. If your hand has been in cold water, the basins water would feel warm; if your hand has been in hot water, the water in the basin would feel cool. The apparent warmth or coolness of the water does not reside in the water itself; it is a quality that depends on the perceivers own state. Because, to him, some perceived qualities of the external world seemed more subjective than others, Locke distinguished between primary qualities (that is, real qualities, actually present in the object) and secondary qualities (that result from an objects power to produce various sensations in us). Primary qualities include the bulk, number, motion, and shape of objects. Lockes secondary qualities include an objects color, sound, taste, and smell. Locke thought we can rely on primary qualities to reflect accurately the nature of ob-

jects in the real world, but we must be cautious, or skeptical, about relying on secondary qualities in the same way. This skepticism about the information of the senses was carried to greater extremes by David Hume in A Treatise of Human Nature (1739/1963). Hume rejected the distinction between primary and secondary qualities, banishing all sense experiences to the realm of the subjective and unreliable. Humes pessimism about the possibility of ever understanding perception is represented quite well by the following comment from his Treatise:
As to those impressions which arise from the senses, their ultimate cause is, in my opinion, perfectly inexplicable by human reason, and it will always be impossible to decide with certainty whether they arise immediately from the object, or are produced by the creative power of the mind, or are derived from the author of our being. (Book I, Part III, Section V, p. 75)

There is good reason, though, to question Humes skepticism (Schlagel, 1984). As knowledge of our senses has deepened, we have come to understand that lawful processes are responsible for what previously seemed to be mysterious sensory caprice. We know that holding ones hand in a basin of hot water initiates a process called adaptation, an alteration in the skins temperature receptors. This process produces the thermal paradox that so perplexed Locke. If we understand adaptation how it grows with time, how long it lasts, and so on Lockes paradox becomes less of a reason for skepticism. Or suppose you put on a pair of colored sunglasses. If theyre strongly tinted, theyll temporarily alter the overall color appearance of the world. But thats no reason to dismiss vision as inherently undependable. If you understand how the sunglasses alter the light that reaches your eyes, and if you understand enough about color vision itself, youll be able to explain this change in the worlds color appearance. In fact, the senses are actually quite dependableas long as you understand how they operate. For instance, you would be able to take any pair of sunglasses and predict quite accurately how the world will look through those glasses, or youd be able to take any persons hand and any basin of water and predict exactly how warm that water will feel. The scientific study of perception overcomes some of the skeptics doubts. Inasmuch as we are discussing attitudes about the relation between perception and reality, this is a good time to introduce a view well mention from time to time in this book. This view, called naive realism, is common among laypersons and beginning students of perception. Naive realism is the view that what we know about the world is both unadulterated and unexpurgated with respect to even

12

Chapter One Introduction to Perception

FIGURE 1.3 | The perspectives of these two transparent figures appear to fluctuate. The darkened surface in each figure seems sometimes to be an inner surface and sometimes an outer surface. The fluctuations of the rhomboid (left) were first noticed by L. A. Necker about 150 years ago while he was examining some crystals. Today these figures are known as Neckers rhomboid and The Neckers cube.

its most subtle details (Shaw and Bransford, 1977, p. 18). In other words, the world is always exactly as it appears. A simple test can determine whether someone is a naive realist. When asked, Why does the world look to you the way it does? the naive realist will answer, Because it is that way. In other words, the properties of experience can always be completely and easily explained by the properties of the world itself. But this simple view of perception is mistaken. For one thing, it cannot explain why different people experience the same environmental event differently. And this does happen, as youll discover throughout this book. We know, for example, that infants cannot see small objects that adults can see; that young adults can hear some sounds that older adults cannot; and that certain people are completely oblivious to odors that others have no trouble smelling. These individual differences challenge naive realism. We all live in the same physical world. If naive realism were a valid viewpoint, wouldnt our perceptual worlds be identical? Theres another reason to reject naive realism: a single, unchanging physical stimulus can fluctuate in appearance from one moment to the next. A classic example of this is the Necker cube, named after Swiss naturalist Louis Albert Necker (see Figure 1.3). Note how the rhomboid at the left seems to switch between two alternative perceptual interpretations. At one moment line segment AB appears closest to you, while at another moment segment CD appears closest. The appearance of the figure at the right undergoes similar fluctuations. This figure is said to be perceptually bistable. Now if perception were determined solely by the physical properties of the figure, its appearance should be un-

changing. Such examples clearly demonstrate that ones perceptions have qualities not present in the physical attributes of the stimulus. Youll see more examples of bistable figures in the forthcoming chapters. At the other extreme from naive realism is subjective idealism, the view that the physical world is entirely the product of the mind, a compelling mental fiction. This philosophical position is associated with the Irish philosopher George Berkeley, who capsulized the idea in the phrase to be is to perceive. In other words, the world exists only as a result of perception; no perception, no world. Carried to its extreme, this position leads to solipsism, the notion that only your mind exists and all other worldly objects are perceptions of your mind. This position can be entertaining to discuss among friends but is scientifically sterile. If the world in which we exist were not real, there would be no reason to study the relation between perceiving and that (imaginary) world. Having rejected naive realism and solipsism, what do we propose about the relations between human perceptions and the real world? As stressed earlier, we readily acknowledge the existence of the real world and assert that its existence does not depend on a perceiver. At the same time, we recognize the perceivers special contribution to the process of perception. The perceivers view of the world is not perfectly accurate, of course, because the perceivers sensory system both limits the information that is available and augments that information. To show you more exactly what we mean by the perceivers contribution, consider a familiar question: Does a tree falling in the forest make a sound if there is no one around to hear it? According to a solipsist, no tree, no forest, and no sound would exist in the absence of a perceiver. But according to our view, not only would the falling tree still exist even though no perceiver happened to be around, its fall would create acoustic energy in the form of air pressure waves. But would this constitute sound? If the term sound means a perceptual experience, then clearly the falling tree would not produce a sound. For the tree to produce a sound requires the presence of some organism with a sensory system capable of registering the available acoustic energy. But even this does not guarantee that the resulting experience would be what is normally called sound. Its conceivable that the organism that is present might not be able to hear because it has no ears, but instead could feel the energy produced by the falling tree (in the same way that you can literally feel the beat of a bass drum). To qualify as sound, the energy must strike the ears of a humanor some other creature with a nervous system like that of a human. What this boils down to is that the quality of ones sensory experience depends on events within the nervous system, as Box 1.2 underscores.

Hearing Lightning, Seeing Thunder BOX 1.2

Seeing and hearing are qualitatively different perceptual experiences. This is shown by the fact that people never confuse sight and sound. The same can be said for touch, taste, and smell. In fact, these qualitative differences form the basis for the classic five-part division of the sensestouch, taste, smell, hearing, and seeing. Our assumption in this book is that these subjectively different experiences are products of neural events within the brain. And yet those events, it is known, all boil down to patterns of nerve impulses within the brain. Since different experiences are represented by the same sort of events, how does the brain manage to distinguish one type of experience from anothersight from sound, and taste from smell? Lets consider this question as it applies to sight and sound. It is tempting to answer by pointing out that sound waves, the stimulus for hearing, are fundamentally different from light energy, the stimulus for seeing. However, this argument is not adequate because the brain does not directly receive either sound waves or light energy. It receives only tiny electrical signals called neural impulses. In other words, from the brains perspective, all incoming signals are equivalent. But, you might point out, although they resemble one another, those neural impulses arise from different sources, namely, the eyes and the ears. And, you might continue, those sources are fundamentally differentthey are specially designed to respond only to particular kinds of physical stimulation. Because of their specialized receptors, the eyes respond to light but not to sound, while the opposite is true for the ears. So, you might well conclude, the distinctiveness of seeing and hearing depends on the difference between the eyes and the ears. This explanation, however, is not adequate either because sensations of light and sound can be produced without the participation of eyes and ears. One can bypass them and stimulate the brain directly. During the course of brain surgery on awake, alert humans, neurosurgeons sometimes need to stimulate the brains surface electrically to determine exactly where they are working. Depending on the area of the brain stimulated, patients report vivid sensations that seem quite real (Penfield and Perrot, 1963). For instance, stimulation at a point in the back of the brain can elicit sensations of light flashes, whereas stimulation at the proper spot on the side of the brain can cause the patient to hear tones. Here, then, are examples of qualitatively distinct sensations that arise from exactly the same sort of stimulationa mild electric current. Note, though, that the patients did not feel the electric current, they heard it or saw it, depending on the brain region stimulated. These observations force a surprising conclusion: the critical difference between hearing and seeing depends not so much on differences between the eyes and the ears but on where in the brain the eyes and ears send their messages and how those brain areas are organized. This is actually a very old idea, dating back to Johannes Mller, a nineteenth-century German physiologist. Mllers theory, called the doctrine of specific nerve energies, states that the nature of a sensation depends on the particular set of nerve fibers stimulated. According to this doctrine, activity in the nerve from the eye will invariably produce visual sensations, regardless of how that activity is instigated. Nowadays, it is recognized that sensory nerves travel to specific brain areas: the nerve from the eye travels to one place, the nerve from the ear to another. Thus, the emphasis has shifted from the nerves themselves to their projection sites in the brain. It is now widely believed that the distinctiveness of sight and sound is related to the unique properties of the neural connections within different regions of the brain, connections that are established during early brain development (von Melchner, Pallas, and Sur, 2000). Mllers doctrine suggested a provocative thought experiment to William James (1892, p. 12). To paraphrase his idea, suppose you were able to reroute the nerve from your eye, sending it to the part of your brain that normally receives input from your ear. Suppose that while you were at it, you also rerouted the nerve from your ear, sending it to that part of your brain that normally gets visual information. Now imagine that with this revised nervous system, you are caught in a thunderstorm. A flash of lightning, which stimulates the eyes, should evoke auditory sensations, while the subsequent sound of thunder, which stimulates the ears, should evoke visual sensations. But dont expect the lightevoked auditory sensations to sound like thunder or the sound-evoked visual sensations to resemble lightning the train of neural impulses carried by the optic nerve and the auditory nerve are unlikely to establish the precise patterns of brain activity associated with thunder and lightning.

13

14

Chapter One Introduction to Perception

To sum up: In order to understand perception as fully as possible, one must study not only the properties of the physical world but also those of the perceiver.

Complementary Approaches to the Study of Perception

At the outset of this chapter we stated that perception entails a sequence of interrelated events that mutually influence one another. Furthermore, we said that to understand perception requires knowing something about each component of the sequence. Understanding these components requires the combined knowledge from several different scientific disciplines, ranging from biophysics to psychology. These disciplines use different levels of analysis, from the microscopic (studying the behavior of molecules) to the macroscopic

(studying the behavior of whole organisms and groups of organisms). For a complete picture, then, one needs to analyze perception at several different levels, each offering a unique and necessary perspective. To illustrate metaphorically what we mean by levels of analysis, look at Figure 1.4. It shows an aerial photograph taken over the Peruvian desert from a very great height. From this altitude, one can see a mammoth sand carving thought to be a thousand years old. The carving, made by people lost to history, is a figure nearly 1,600 meters long. It is so huge that it can be recognized only from a great height. Standing on the ground, you would be able to see only small portions of the carving, never the entire thing. And at a really close level of scrutiny, you would see only the hills and valleys in the grounds surface. Thus, to appreciate the carvings entire pattern requires a particular level of analysis, namely, far above

FIGURE 1.4 | Aerial photograph of sand carving on Peruvian desert.

Complementary Approaches to the Study of Perception

15

the carving. Suppose, though, that you wanted to understand just how the carving had been done. Then you would need to examine the details of the carving and what it was made of. Such an examination would require a radically different perspective, one focused on the details of the carving.3 This beautiful and mysterious desert carving dramatizes a point about perception: one must adopt different levels of analysis in order to answer all the significant questions about the subject. Consequently, well be adopting various levels of analysis in our examination of perception. The three main levels of analysis that well explore are the psychological, biological, and theoretical. Distinguishing among the psychological, biological, and theoretical approaches will help organize the discussion that follows. We repeat, though, that these approaches are not mutually exclusive, but are complementary. One simply cannot learn all one wants to know about perception from just one approach.

describe some of those techniques as the need arises. Now, however, lets not focus on the details of particular behavioral techniques. Instead, lets analyze the methods along more general lines, grouping them according to their degree of formality. By formality we mean the extent to which stimuli and reactions to them are structured or controlled. The Phenomenal/Naturalistic Approach The least formal is the phenomenal/naturalistic method. Phenomenal (sometimes called phenomenological) means that the data used to learn about perception consists of ones own conscious experiences which, of course, can be communicated to others only using verbal descriptions. Naturalistic means that the evidence pertains to perceptual experiences occurring within our everyday environment; there is no effort to induce, modify, or control the objects or events triggering perceptual experience. This informal approach to perception has some strengths. For one thing, it relies on the most readily available data, the steady stream of perceptual experiences evoked by naturally occurring events. Such experiences might include the deeply saturated colors experienced around sunset, the rising and then falling pitch of an ambulances siren as the vehicle speeds past you, the funny taste of orange juice after youve brushed your teeth. We all have countless experiences like these throughout our waking hours. To study perception, then, you could collect and organize these experiences. Going one step further, you could discuss your perceptual experiences with other people, for purposes of comparison. But what limitations would you encounter by following this program? First, by restricting yourself to phenomenal descriptions, you would be unable able to study perception in animals and preverbal infants, a serious limitation. Second, even working with humans who can verbalize their experiences, you would need to be wary. Verbal reports can be fallible and misleading. For one thing, not all people use words in the same way. For instance, many colorblind people have learned to label colors much as color-normal people do, even though their experiences are surely very different. Because verbal descriptions are typically made with great confidence (after all it is your experience youre describing), one can be misled into believing verbal descriptions provide an accurate and complete picture of perceptual experience. That assumption, however, is unwarranted: there is reason to doubt whether individuals

Psychological Approaches
Psychological approaches can take many different forms, but all have in common the use of some behavioral measure as a gauge of perception. Those behavioral measures can be verbal responses (Yes, I hear it), manual reactions (Press this button when you hear it), or reflexive reactions (Did he flinch when the sound was produced?). Those behavioral reactions to stimuli are taken as indices of whether those stimuli can be detected or whether they can be discriminated from other stimuli. For instance, one might attempt to train a bird to fly to a red perch but not to a green perch. If the bird succeeds in learning this task, one might infer that the bird can discriminate red from green. To justify this conclusion, however, requires additional tests to ensure that the bird is actually relying on color, not other visual cues such as position or brightness. Similarly, a human being can be instructed to push one button whenever a red object is presented and to push another button whenever a green object is presented. For birds and humans, behavior is used to infer something about perception. There are actually many specific techniques for studying perception, and well

It is strange to realize that the creators of this carving were never able to see the entire fruits of their labor, since they could never enjoy a view from a perspective anything like the one shown in Figure 1.4.

16

Chapter One Introduction to Perception

FIGURE 1.5 | Look at this drawing for a while. If you cannot discern an animal, look at Figure 1.6 on page 18.

can accurately describe their experiences, motives, and thought processes (Nisbett and Wilson, 1977). Factors of which were completely unaware can profoundly influence what we perceive. As you will learn as we move through the books chapters, these unconscious influences include expectations, prior experience, and motivations (see Figure 1.5). Although such factors are integral to the nature of perceptual experience, the phenomenal approach would foreclose identifying them: you cannot possibly describe aspects of perception that transpire outside your awareness. In a related vein, verbal reports of perceptual experience can unwittingly force someone to categorize a perceptual experience in a way that belies the true nature of that experience. Take, for example, the simple question, Did you hear that faint, rattling sound when you started the car? The question calls for a binary decisionyes or nobut the sensory experience upon which that decision must be based is probably far more subtle. Your binary answer can thus depend on the possible implications of your answer as well as on your prior knowledge about the status of your car. Fortunately, there are procedures allowing us to assess the influence of these criterion effects on sensory decisions (see Box 1.3). Theres another reason why verbal reports cannot always be trusted. In some instances, people are motivated to avoid telling what they consider to be the truth about their experiences. Heres one example. A malingerer is someone who pretends to have an illness or disability to get some special gain or to avoid some re-

sponsibility. Feigned deafness is one form of malingering. Ask such a malingerer if he can hear, and youll get no answer unless the question is communicated in writing, lip reading, or sign language. Then, the malingerer will assure you that he cant hear (a misleading verbal report). But there is a very clever, foolproof way to catch the malingerer: delayed auditory feedback. While the person is reading aloud, record his speech and, following a very short delay, play it back into his ears. If he is genuinely deaf, and not a malingerer, delayed auditory feedback will have no influence on his reading. But if he can hear, the delayed feedback of his own voice will invariably disrupt his speech. The word feigns denotes that someone is purposely lying or pretending. But there are instances where a persons erroneous verbal reports dont really constitute lying. One such instance is Antons syndrome, which was first described by Gabriel Anton more than a hundred years ago (Foerstl, Owen, and David, 1993). This syndrome, as rare as it is bizarre, involves complete blindness coupled with denial (the blind person denies that he or she is blind). The condition supposedly arises because two different areas of the brain have been damaged: the one needed for seeing and the one needed for knowing that youre seeing. This damage to the brain occurs quite suddenlyusually as the result of a cerebral vascular accident (a stroke)and the victim of Antons syndrome may walk around for quite some time bumping into things and having other mishaps until he or she becomes convinced that something is wrong. But immediately following damage to the brain, victims of Antons syndrome confidently insist that they can see. Asked to describe what they see, the victims may give very detailed answers, which are utter fabrications, as evidenced by their lack of correspondence to objective reality. Antons syndrome not only underscores the potential unreliability of verbal reports about perception, but also points up a more general fact: perceptual experiences and knowledge of those experiences are two quite separate things. Despite its limitations, the phenomenal/naturalistic approach to perception has an important role to play. For more than a hundred years, careful and thoughtful observers have used this informal approach as a basis upon which to build a more formal study of perception. Although this book will emphasize these more formal approaches, many of the ideas for formal study derive from this less formal method.

Should You Answer the Phone? BOX 1.3

Just about everyone has had this maddening experience. While taking a shower, you faintly hear what sounds like the telephone ringing. Because of the showers steady noise, though, youre not sure it is the phone. So do you decide to run, dripping wet, to answer it? Or do you conclude that its only your imagination? Your behavior in this situation depends on factors other than the loudness of the ringing sound. For instance, if you are expecting an important call, you will in all likelihood scurry out of the shower to see whether the phone is actually ringing. If, in contrast, youre not expecting a call, youre more likely to attribute the ringing sound to the showers own noises. Your decision about the reality of the sound, then, is influenced by your expectations. This example illustrates a significant principle, namely, that ones interpretation of sensory data depends significantly on nonsensory factors. This dependence affects the way in which results from perceptual studies are interpreted. Imagine testing a persons hearing by presenting faint sounds and having the person say whether or not she could hear the sound. Performance on such a test can vary from one person to the next, and not just because some people have better hearing than others. Some people are simply more willing to take a gamble, asserting that they heard something even if theyre not 100 percent certain (these people might also want to impress the tester with their keen hearing, say, if the hearing test is part of a job application). There are also more conservative people, who are not gamblers. In the hearing test, such people might require a much louder sound before they are willing to say they heard it. Suppose that two people took a hearing test, one a conservative type, the other a gambler. On the basis of their performance on the hearing test, the tester might mistakenly conclude that the conservative had inferior hearing. People do differ in the sensitivity of their sensory systems. Some individuals, for instance, have a keener sense of smell than do others. But people also differ in their motivations, expectations, and willingness to gamble. As an aggregate, these latter differences can be labeled motivational differences. In studies of perceptual abilities, it is important to distinguish between an individuals sensitivity and motivation. Toward this end, psychologists have developed several strategies for separating the two. To tell whether a person can really hear an extremely faint sound, one needs more to go on than the fact that she is constantly saying that she hears a sound. Logically, one must also ensure that she does not make exactly the same claim when no sound whatever has been presented (Goldman, 1976). Many experiments on hearing, then, randomly intermix two types of test trials. On one type of trial, a weak sound is presented; on the other, no sound is presented. After each trial, the person says whether or not she heard a sound. Someone really interested in impressing the tester might say, Yes, I hear it, after every single trial. Of course, shed be right on every trial in which a sound actually occurred, but shed be wrong with respect to every trial in which no sound occurred. From this result, the tester should realize that this person could not discriminate the presence of sound from the absence of sound. Omitting the sound and noting the subjects failure to recognize that omission allows the tester to separate the persons sensitivity to sound from other possible factors, such as the motivation to impress. This general strategy is not limited to the study of hearing; similar methods are used with the other senses as well. To implement the strategy, psychologists have developed a set of sophisticated statistical techniques collectively known as signal detection theory. The appendix provides additional details of signal detection theory. For a more thorough treatment of this topic, we suggest you consult Wickens (2001), MacMillan and Creelman (2004), or Swets, Tanner, and Birdsall (1961). Meanwhile, before showering the next time, take the phone off the hook.

Experimental Approaches The phenomenal/naturalistic approach works with objects and events as they occur in nature, without trying to control or manipulate them. This approach is simple but not entirely satisfactory. For one thing, to study a particular aspect of perception often requires creation of a class of stimuli that is not available naturally. To give an example, nearly all sounds in

the natural world comprise a broad range of acoustic energy spread throughout the audible frequency spectrum. But to understand our ability to hear natural sounds, it is essential to study hearings sensitivity to different, isolated frequencies, and that requires generating those sound frequencies artificially in the laboratory. Moreover, it is sometimes necessary to use a whole series of
17

18

Chapter One Introduction to Perception

FIGURE 1.6 | Outline drawing of the same animal depicted in Figure 1.5. Note how seeing this figure helps you interpret Figure 1.5. Surprisingly, this effect lasts for months.

stimuli to compare each to the effects produced by its fellows. And the specific series you want might never occur naturally. For example, suppose you want the members of a series of sound tones to differ in only one attribute (such as pitch) with all others (such as loudness) held constant. This would make it easy to ascribe any resulting change in perception to the varied attribute. If several attributes varied simultaneously (as is usually the case with naturally occurring stimuli), you would have trouble knowing how much each attribute contributed to perception. In a related vein, naturalistic stimuli do not often repeat themselves precisely. We mentioned previously that the study of perception should be general: It should measure the perceptual responses of more than one person (generalizability), and it should measure those responses multiple times in the same person (replicability). To satisfy these needs requires repeated measures of perception in response to the same stimuli presented multiple times under controlled conditions. Control and careful manipulation of the stimulus also allow one to identify exactly what aspect of the stimulus underlies some perceptual experience. Heres one illustration: Some rare individuals are able to discriminate among thousands of different wines by taste alone. To determine the basis of this remarkable ability, you could create a series of specially constituted wines that varied in their composition, and using this set of controlled stimuli you could isolate the cues enabling such individuals to distinguish what most people cannot. Control over sensory stimuli represents a key requirement for two kinds of experiments that are the

foundation of the scientific study of perception: matching experiments and detection experiments. Matching experiments ask people to adjust one stimulus until it appears identical to another. This obviously requires stimuli that can be manipulated precisely. Detection experiments measure the weakest stimulus that a person can detect. Again, such experiments require stimuli whose intensities can be controlled. The appendix describes some of the formal, structured methods for performing these kinds of experiments. In addition, throughout the book, youll encounter descriptions of various methods in the context of the research problems for which they were designed. So, to control and manipulate stimuli, one usually synthesizes specially designed stimuli not found in the natural environment. Such stimuli are sometimes criticized as being nonecological, because they are not the objects and events for which perceptual systems evolved. Researchers widely agree, however, that simple, artificial (nonecological) stimuli can be valid because they often clarify the effects of more complex, naturally occurring stimuli. This point is well documented throughout this book. There is merit, though, to studying perception using ecologically representative stimuli. In recent years, perception research has increasingly tried to relate simple, laboratory-created stimuli to the objects and events ordinarily encountered in the natural environment. In fact, one relatively new branch of perception focuses explicitly on quantitative descriptions of the visual world and the auditory world in which we live. Called natural scene statistics (Nelken, 2004; Kersten, Mamassian, and Yuille, 2004), these descriptions can provide powerful links between psychophysical data and results generated using any of several biological approaches that youll learn about in the next section. Another non-ecological characteristic of the experimental approach is the tendency to study a given sensory modality in isolation of the others. In fact, entire journals are devoted to research on single modalties (e.g., Vision Research, Hearing Research, Chemical Senses). This parceling of perception into sensory systems is also reflected in the organization of this textbook, with separate chapters devoted to vision, hearing, touch, taste, and smell. Yet our perceptual experiences in the natural environment are nearly always multimodal in character, with information from the separate modalities merged to form a coherent perceptual experience. For that reason, throughout this book you will repeatedly encounter examples underscoring the interdependence of perceptual experience on information from multiple sensory systems.

Complementary Approaches to the Study of Perception

19

So despite its limitations, the psychological approach is essential in the study of perception. Nonetheless, it leaves unanswered important questions about neural processes underlying perception. In the next section we turn to a complementary approach that addresses those questions.

Biological Approaches
Throughout the history of scientific research on perception, an enduring theme has been the dependence of perceptual events on neural events, within sensory receptors and within the brain. An overriding goal, then, has been to investigate the linkages between perceptual and biological phenomena. Though such investigations can be challenging to carry out, their outcomes often yield information fundamental to understanding perception. Here well sketch out some of the strategies used to relate physiology to behavior, without going into the details or the outcomes. In subsequent chapters, youll learn how these strategies provide answers to specific questions about perception. Lesion Technique A wound or an injury to a delimited part of the brain can destroy neurons within that part of the brain. This loss of brain cells produces what is called a lesion, and by studying the consequences of that lesion one can draw inferences about the function of the neural tissue in that region of the brain. Lesions can occur naturally, from disease or trauma, or lesions can be created in experimental animals specifically for research purposes. Restricted lesions can be produced by applying electrical current to a targeted area of the brain, by injecting a chemical that kills nerve cells, or by surgically removing brain tissue. Usually, such lesions are permanent, and irreversible. Regardless of how the lesion occurs, a researcher can measure resulting changes in perceptual function. Interpreted with proper caution, these studies can help identify the anatomical locus of neural structures crucially involved in a given perceptual ability. There are, however, limitations to the conclusions that can be drawn from lesion studies, especially because lesions disrupt more than just the neural operations associated with the lesioned brain region. Consider this analogy: A mammoth snowfall on the Massachusetts Turnpikes eastern end can seriously interfere with the economic activity throughout the Boston area, but we wouldnt conclude that economic activity is localized on the Massachusetts Turnpike. Likewise, lesioning a particular brain area may destroy an animals ability to recognize a certain class of

visual objects such as faces, but it would be misleading to state that this brain region was solely responsible for analyzing that class of objects. By the same token, an animal may recover function following the destruction of brain tissue, but this doesnt mean that the lesioned area does not normally participate in that function. Other brain areas may have taken over for the destroyed area. The perceptual consequences of naturally occurring lesions in humans have been used for centuries to gain insights into the neural basis of human perception (Boring, 1942). Lesion results from human studies, however, can be especially difficult to interpret (Rorden and Karnath, 2004). Disease or trauma usually creates brain damage in multiple areas of the brain, so naturally occurring lesions may offer only limited help in pinpointing the neural basis of perceptual function. So, results from lesion studies can be strongly suggestive but seldom definitive. Before turning to the next biological approach, we want to mention a relatively new technique, transcranial magnetic stimulation (TMS), that can be construed as a safe, reversible form of the lesion technique. As you probably know, brain activity consists of tiny electric signals generated by the brains billions of neurons. A lesion, of course, destroys that electrical activity by destroying the neurons themselves. TMS, in contrast, disrupts neural activity by briefly but potently altering the biophysical events underlying generation of those electrical signals. Specifically, TMS involves the delivery of a brief, strong pulse of magnetic energy through a probe resting on a persons scalp. These bursts of magnetic radiation easily penetrate the scalp and skull, thereby inducing small electrical currents within the brain tissue directly below the probe. Those electrical currents interfere with normal, ongoing neural activity within cells located at the site of the TMS pulse and, thereby, disrupt whatever function those neurons ordinarily subserve. Laboratories are now beginning to use TMS to study the role of different brain regions in perception (Stewart et al., 2001). It is crucial, of course, to use this technique with great caution, to ensure that the currents induced by TMS remain within safe levels and to be certain that people with any history of epileptic seizure do not participate in a TMS study (Anand and Hotson, 2002). Evoked Potential Technique Another widely used procedure relates perceptual judgments made in response to a given stimulus with the electrical brain activity evoked by that same stimulus. Measured through small electrodes placed on the surface of a persons scalp, this brain activity is termed an evoked potential (EP) or,

20

Chapter One Introduction to Perception

FIGURE 1.7 | Photograph of a human volunteer with a multi-electrode array attached to the scalp. Each electrode is capable of picking up tiny electrical signals generated by the brain.

The ERP technique is also useful for studying nonverbal perception in infants and in animals, individuals unable to report their perceptions verbally. When ERPs to a given stimulus can be recorded from the brain of an infant or an animal, we can safely conclude that the brain is responsive to that stimulus. We cannot, of course, conclude that the individual actually perceives the stimulus, but we can be certain that a necessary condition for perception has been satisfied. Using ERPs, researchers have harvested important facts about the visual world of infants (e.g., Atkinson and Braddick, 1998) and of animals (e.g., Berkley and Watkins, 1971). This technique also has its drawbacks. For one, the failure to record an ERP doesnt necessarily mean the brain fails to register the evoking stimulusthe neural signals may just be too weak to be picked up or the recording electrodes may be misplaced on the scalp. For another, it is very difficult to pinpoint in exactly what region of the brain ERP activity is arising. So, it is important to realize that the evoked potential provides a rather diffuse measure of brain activity picked up from tiny signals generated by large numbers of brain cells; the evoked potential reflects omnibus brain activity. Brain Imaging Techniques In the past decade, perception research has profited substantially from imaging techniques that generate detailed pictures of the human brain in action. When measured under appropriately designed experimental conditions, images produced by these methods can reveal specific regions of the human brain uniquely activated while a person is engaged in a specific perceptual task. Currently in use are three powerful forms of brain imaging: positron-emission tomography (PET) scan, functional magnetic resonance imaging (fMRI), and magnetoencephalography (MEG). Each of these techniques has produced remarkably revealing snapshots of brain regions engaged in various acts of perception. MEG exploits the fact that active nerve cells produce tiny electrical currents, which, in turn, generate minute, localized magnetic fields. The skull and scalp are essentially transparent to these magnetic fields, so an array of several hundred ultra-sensitive detectors around the head can pick up the magnetic fields.4 Sophisticated software then analyzes the spatial and temporal distribu-

alternatively, an event-related potential (ERP). This procedure requires sensitive electronic amplifiers to boost the electrical signals measured from the scalp and skull. These signals are called massed responses because they reflect the varied activity among the thousands of brain cells located near each electrode. In recent years, ERPs have been measured using an array of electrodes distributed over a large region of the scalp (see Figure 1.7). By recording electrical signals from all these electrodes simultaneously, researchers can more accurately localize the source of brain activity associated with a given stimulus. Moreover, using multiple electrodes makes it possible to monitor how brain activity builds and spreads from one location to another. Over the years, ERPs have been measured in a variety of tasks. In some instances, researchers want to know whether the magnitude of the ERP correlates with a persons subjective impression of the stimuluss intensity. In other cases, researchers measure the ERP in response to unanticipated or novel stimulation. A particularly intriguing application of the ERP technique looks for brain activity in response to visual or auditory events that, in fact, are not perceived. For example, Luck, Vogel, and Shapiro (1996) found that unexpected words still elicited a strong novelty response in the ERP, even when the presence of those words could not be reported because the persons attention was temporarily distracted.

The brains magnetic fields are less than one hundred millionth of earths normal magnetic field. As a result, MEG requires a magnetically shielded room and ultra-sensitive detectors that must be cooled to the temperature of liquid helium, about 270 degrees Celsius.

Complementary Approaches to the Study of Perception

21

tion of the magnetic field, and can localize the brain regions that gave rise to the field. Because MEG does not depend upon relatively slow metabolic changes in the brain, as PET and fMRI do, it can resolve very fast changes in brain activity. With a temporal resolution about a thousand times better than that of the other techniques, MEG can tell apart distinct stages of processing that contribute to perception (Finney et al., 2003; Liu, Harris, and Kanwisher, 2002), including stages involved in short-term memory of what has been perceived (Lu, Williamson, and Kaufman, 1992). Compared to fMRI, MEG is less often used as an imaging technique, in part because of the great expense of the necessary facilities and the extreme vulnerability of the measured signals to extraneous noise. Also, compared to fMRI, MEG lacks the spatial precision to pinpoint activity to a small area in the brain. Technological advances will undoubtedly reduce these problems, however, which means we can expect MEG to play a growing role in the study of perception. PET capitalizes on the fact that activated regions of the brain temporarily require additional amounts of glucose and oxygen. PET tracks this increased metabolic demand by applying a radioactive tag to glucose or to oxygen atoms in water, both of which are required for increased metabolism. Sensors can then register the temporary accumulation of this radioactively labeled chemical in particular brain regions. In one version of the procedure, a person first inhales a mixture of air and radioactively labeled carbon dioxide. From the persons lungs, trace amounts of the radioactive material enter the bloodstream and then rapidly dissipate. For a few minutes or so after inhalation, regions of the brain with heightened metabolic activity will attract a strong local flow of blood, bringing with it additional radioactive material. Using radioactive sensors arrayed around the head, one can record an image of the distribution of radioactively tagged blood within the brain. Going one step further, the person can now engage in a particular perceptual task (e.g., detect changes in the color of a geometric form) while the PET scan is being performed. It thus becomes possible to relate neural activity in specific brain regions with the execution of particular perceptual tasks (Corbetta et al., 1991; McIntosh et al., 1999). PET has some noteworthy limitations, however. First, the technique has limited spatial resolution: PET cannot distinguish differences in activity between small brain areas that adjoin one another. Second, safety and health regulations limit the amount of radioactive exposure an individual can receive, making it impossible to test the same person repeatedly un-

FIGURE 1.8 | Photograph of an MRI scanner, a device capable of measuring tiny fluctuations in a magnetic field produced by increased blood flow in the brain. Called functional magnetic resonance imaging, this technique has developed into the premier method for studying brain activations associated with cognitive function, perception included. The volunteer participant lies on the table (shown here outside of the scanner), and the table is moved into the scanner so that the persons brain is centered inside the donut-shaped bore of the scanner.

der different conditions. For these reasons, recent brain imaging work has favored another technology that we turn to next, fMRI. fMRI exploits the same principle as PET: increased neural activity triggers a temporary increase in metabolic demand at the site of activation. But unlike PET, fMRI tracks the consequences of this increased demand without the need for a radioactive labeling agent. fMRI capitalizes on the fact that oxygenated blood has different magnetic properties from deoxygenated blood. Thus, its possible to detect and localize brief surges in oxygenated blood within the brain, thereby inferring sites of increased neural activity. An fMRI scanner, then, allows researchers to measure fluctuations in the magnetic fields of well-defined areas of the brain (see Figure 1.8). These measurements can be made at the same

22

Chapter One Introduction to Perception

FIGURE 1.9 | This is an image of the left hemisphere of the brain of an adult human (that human happens to be Randolph Blake, a coauthor of this textbook). The checkered overlay at the back of the brain (the right-hand portion of image) denotes regions activated when this individual viewed a rotating checkerboard figure. Other areas within interior regions (not visible in this picture) were also activated by this visual pattern. Regions of activation were identified by the brain imaging technique called functional magnetic resonance imaging, or fMRI.

time that an individual engages in a perceptual task, making it possible to correlate fluctuations in an fMRI signal with performance of the task (see Figure 1.9). A nice primer on fMRI can be found on the Web (accessible at [Link]/blake5). Depending on the scanners sensitivity, fMRI can pinpoint regions of neural activation with very high spatial accuracy (Wandell, 1999). As fMRI requires no radioactive material, it can be repeated on the same individual, which makes it possible to examine aspects of perception, such as learning, that may take days to develop. As powerful as it is, fMRI does have its limitations, however. First and foremost is the relatively poor temporal resolution of the technique. The hemodynamic changes in response to increased neural activity are themselves relatively slow, meaning that it can take several seconds for blood flow to increase in those parts of the brain where neural activity has gone up. This makes it difficult to pinpoint with high accuracy exactly when activity increased and whether activation in one area

started before activation in another area. Besides its relatively poor temporal resolution, fMRI also suffers from our incomplete understanding of the nature of the signal being measured by the technique. Theres no doubt that fMRI is indexing neural activity, but at present there is no way to know what proportion of that activity corresponds to output signals from a brain region compared to input signals to that brain region. For that matter, we arent sure about the exact nature of the physiological events triggering increased blood flow. Those events could involve any number of component processes that go into the generation of neural activity, and the interpretation of fMRI results can depend on what those components are (Logothetis, 2002). Fortunately, these questions are going to be answered in the near future, thereby sharpening the conclusions we can draw from fMRI experiments. Despite these limitations, the technique remains quite popular and is highly useful in the study of perception; throughout the coming chapters, youll read about many results obtained using fMRI.

Complementary Approaches to the Study of Perception

23

Single Cell Techniques The techniques discussed so far all operate on a fairly coarse scale, in which the grain comprises thousands of neurons. Techniques that provide finer grain analysis are excellent supplements to what we have characterized as coarser techniques. In one finegrain approach, the physiological responses of individual neurons are recorded while an alert behaving animal, usually a monkey, engages in some perceptual task. For example, while varying the complexity of the task, the researcher tries to identify single neurons whose responses are strongly correlated with the animals performance. Correlated activity suggests that those neurons form part of the neural machinery involved in the perceptual task. To be successful, this approach must draw upon prior evidence about the locus of neurons thought to register information utilized in the task. Otherwise, the research would degenerate into an unguided search for a needle in a haystack. Once an investigator has identified a set of neurons that is believed to mediale a particular perceptual judgment, that hypothesis can be tested more directly. One can artificially stimulate the neurons by passing weak electrical current into them. If activity in those neurons is crucial, this boost in activity should alter an animals performance on an associated perceptual task (Newsome, Britten, and Movshon, 1989; Cohen and Newsome, 2004). You will see examples of revealing results using this technique in several chapters to come. Note, however, that electrical stimulation activates not only those neurons in contact with the electrode but also other neurons nearby. So it would be a mistake to assume that activity in those neurons alone is affecting the animals behavior. Incidentally, just because these recording techniques gather information from individual neurons it doesnt mean that researchers are limited to gathering that information from just one neuron at a time. In recent years, it has become possible to perform multielectrode recordings, capturing the activity of dozens or more neurons simultaneously (Nicolelis and Ribeiro, 2002; Katz, Nicolelis, and Simon, 2002). This great advance allows us to examine entire neural circuits in action, and to understand the flow of neural information from one part of the circuit to another.

these sections dealt with the input for perception, the hardware of the perceptual process, and its output. This metaphor, however, ignores a crucial element: the program. Here, program refers to the set of instructions or rules that transform input into output. Sometimes people confuse the program with the hardware on which it runs. This confusion obscures an important distinction between the two. To reinforce the distinction between hardware and program, consider philosopher Daniel Dennetts discussion of an abacus.
Its computational task is to do arithmetic: to yield a correct output for any arithmetic problem given to it as input. At this level, then, an abacus and an electronic calculator are alike; they are designed to perform the same information-processing task. The algorithmic description of the abacus is what you learn when you learn how to manipulate itthe recipe for moving the beads in the course of adding, subtracting, multiplying, and dividing. Its physical description depends on what it is made of: It might be wooden beads strung on wires in a frame; or it might be poker chips lined up along the cracks in the floor; or something accomplished with a pencil and good eraser on a sheet of lined paper. (1991, p. 276).

Theoretical Approaches
The previous two sections outlined strategies by which perception and its underlying neural events can be studied and ultimately related to one another. Borrowing terminology from computer science, we might say that

In his influential book Vision, David Marr (1982) argued the value of studying perception on three complementary levels of abstraction. From most to least abstract, the levels are as follows: one can analyze perception as an information-processing problem; one can examine the set of rules (the program) used to solve the information-processing problem; finally, one can study the neural machinery of perception and how that machinery executes the program. Marr insisted that some effort on the most abstract level ought to precede work on the other two levels. If, like Marr, one describes perception as a solution to a problem, one needs to appreciate exactly what the problem is. An analogy may help. Suppose a friend tells you that she is thinking of a number between 1 and 20. Your job is to identify that number. You get one clue: the number is odd. Obviously, the clue does not give enough information to identify the number with certainty; it under specifies the solution (as opposed to being told, for instance, that the number is the largest prime number in the set). Generally speaking, the information provided to our sense organs under specifies the true nature of objects in the world. Somehow, though, despite significant under specification, the

24

Chapter One Introduction to Perception

perceptual process manages to yield high-quality, useful representations of those objects. A researcher may sometimes want to identify not only the processes that make it possible to perform a particular perceptual task, but also how efficiently humans perform that task. The theoretical goal is to identify any aspect of the perceptual process that fails to exploit fully whatever information might be available. One strategy is to compare human performance of a perceptual task against what could be expected from an ideal perceiver who was able to utilize all the available information. Often the ideal perceiver is a computer-generated model that processes information optimally, with none of the quirks, biases, blindspots, or inefficiencies of human perception (Geisler, 1989). Comparisons of human performance against theoretically ideal benchmarks make it possible to identify aspects of perceptual processing that limit actual performance. Any computation, whether carried out by an ideal observer, by an electronic device, or by a biological device, is only as good as its data and its processing rules. The brains perceptual computations may be accurate or they may be in error, depending on the quality of information supplied by the senses and on the brains predisposition to process that information in certain ways. But if certain processing rules can lead to errors, why would the brain be predisposed to use those rules? The information picked up by the senses is not merely a series of unrelated, incoherent data. Instead, sensory information closely conforms to predictable, structured patterns (Snyder and Barlow, 1988). These patterns arise from the very nature of the physical world itself, the world in which our senses have evolved. For instance, in our world, objects tend to be compact. The various parts of any object tend to be near one another, not scattered at random all over the landscape (Geisler et al., 2001). In our world, the surface color or texture of most natural objects tends to change gradually rather than abruptly (Kersten, 1987). In our world, light tends to come from above rather than from below (Gregory, 1978). In our world, the hardness of a surface determines how sound energy is reflected from that surface (Handel, 1989). If the brains processing rules embodied these regularities, or constraints, that characterize the natural world, the brains perceptual operation could be more efficient and rapid (Kersten, Mamassian, and Yuille, 2004). Just as processing rules can be embodied within the microchips of an electronic device, rules that assume these regulari-

FIGURE 1.10 | Subjective, or illusory, square.

ties of the natural world could be embodied in the hardware of our brains (Ramachandran, 1988). Well illustrate this with an example. Look at Figure 1.10, one of many interesting figures devised by Gaetano Kanizsa (1976). This Kanizsa figure conveys a strong impression of a white square resting atop four black circles. However, in creating the figure, we had only to make the four sectored disks. Your perceptual system did the rest by creating the white square. It is widely believed that subjective figures, like the one seen in Figure 1.10, occur because the visual system makes the quite reasonable assumption that nearer objects tend to occlude objects located farther away. Now, in principle, theres no reason why perception has to include a square in its interpretation of Figure 1.10a perfectly reasonable alternative would be for you to see four sectored disks, period. In fact, thats exactly what you do see when you look at Figure 1.10, a set of four objects seen against a white background. So why does perception fabricate a square in Figure 1.10? Because perception takes into account the likelihood of the layout of objects within a scene when interpreting that scene. For you to see just the four sectored disks and no square in the figure, those disks would have to be precisely arranged in the world with their missing sectors forming perfect right angles relative to one another. Because such an experience is unlikely, our perceptual apparatus prefers the alternative interpretation of four complete disks partially occluded by a square. Perception, in other words, behaves as if it knew the likelihood of various environmental configurations. In the words of Horace Barlow (1998, page 886), Learning the ways of the world consists of learning what does NOT happen as much or more than learning what does.

Recurring Themes

25

FIGURE 1.11 | Photographs of natural objects occluding on another.

Because vision is designed to deal with partial occlusions within naturally occurring scenes (see Figure 1.11), visual perception automatically treats Kanizsas figure as though it were one object (a white square) occluding other, more distant objects (four black disks). In the preceding paragraphs, we focused on perception from an information-processing approach, one aimed at specifying the computational problems facing perception. This represents one form of theorizing within the field of perception. In general, the development of theories of perception, whether computational or not, sharpens our thinking, often translating qualitative observations into quantitative statements. These kinds of quantitatively explicit theories then serve to guide the design and implementation of experiments in perception. Good theories tell us what to look for and, often, where to look.

Recurring Themes
Running throughout the following chapters are some recurring themes that transcend a given perceptual modality. To provide a framework for integrating the material you encounter in the following chapters, well finish our

introductory comments with a summary of five of those major themes. Sensory Transduction Each of the various sensory modalities we will examinevision, hearing, touch, taste, and smellis designed to pick up information in the environment about a particular class of stimulus objects and events. The physical nature of the information depends on the modality under consideration. Vision, for example, is triggered by photons of electromagnetic energy (light as we call it) that stimulate the eyes light-sensitive photoreceptors. Smell is triggered by an entirely different class of stimulation, volatile molecules that interlock with proteins within the nasal cavity. These are just two examples of the process known as sensory transduction: the conversion of some form of physical energy (photons and volatile molecules, in the examples just given) into electrical signals within biological tissues called sensory receptors. Your understanding of perception is not complete without an appreciation of the nature of the physical stimulus associated with a given modality and the nature of the biological processes that register the presence and nature of that stimulus.

26

Chapter One Introduction to Perception

Variations in Perception among People and Animals The study of perception tends to focus on normal adult human beings. Given that students of perception nearly all fall in this category, that focus is understandable and sensible. But some of the most revealing lessons about perception can be achieved by studying individuals who fall outside of the normal adult category. For example, we can examine perception developmentally by studying newborn infants, young children, and, at life spans other end, elderly people. Systematic changes in perception accompany developmental changes in the sense organs and central nervous system. Thus, studying developmental changes in perception offers one way to understand the relation between sensory systems and perceptual experience (Teller, 1997). Normal perception can be compromised by inherited disorders, by disease, or by injury. And just as studying pathology illuminates the processes of health, studying the abnormal perception can illuminate the normal processes of perception. Some of our most remarkable discoveries about perception have come from clinical case studies in which a perceptual disorder has been traced to a physiological abnormality. A classic example of this strategy comes from color vision, where the genetics of color vision were revealed by studying people with deficient color vision. Even when the cause of a perceptual disorder is not fully understood, one can still acquire useful clues about possible causes by determining what other aspects of perception are unaffected. For example, a person with misaligned (crossed) eyes can have diminished depth perception but still be able to see objects and to judge whether those objects are moving or stationary. These kinds of selective perceptual losses indicate that the affected perceptual ability depends on structures within the nervous system different from those mediating the unaffected perceptual abilities (Hebb, 1949; Brindley, 1970). The study of human perception can also profit from learning about perception in nonhuman animals. Although special, sophisticated techniques are required to study animal perception (Blake, 1998), such work expands our appreciation of the mechanisms of perception. In this book, some of what you will learn about human perception comes from animal studies. A good deal is known about the anatomy and physiology of the sensory systems of several nonhuman species.

FIGURE 1.12 | Zllners illusion of orientation.

This knowledge takes on added significance when it can be related to studies of perception in the same species. Illusions and Errors of Perception It is natural to assume that perception works best when it provides an accurate, veridical view of the world. Yet there are numerous, fascinating examples where perception makes mistakes, sometimes glaring ones. In the case of vision, we may grossly misjudge the size of an object; in hearing, we may swear that a person is uttering one word when, in fact, hes saying something different; and in touch, we may judge a stimulus as cold when touching it with one hand but hot when touching it with the other hand. In all the chapters in this book youll run into these so-called illusions of perception, where the appearance of a stimulus departs notably from the actual nature of the evoking stimulus. Figure 1.10 illustrates one such illusion, and Figure 1.12 illustrates another. Here the long horizontal lines appear tilted relative to one another, but in fact they are perfectly parallel. This illusion of orientation, invented by Franz Zllner in 1860, disappears when the radially oriented lines are removed leaving just the horizontal ones, and this reveals an important clue to illusions: when they occur, illusions are usually attributable to the context in which the illusory figures appear. In fact, illusions and the conditions producing them provide genuine insights into the operations of normal perception. Illusions, in

Recurring Themes

27

other words, highlight the ordinary processes underlying normal perception. Throughout the book, youll have opportunities to experience these fascinating curiosities. You also should visit some of the websites on illusions available at [Link] Perception Is Modifiable Many of our perceptual skills improve with experience, and these improvements occur without special effort on our part: the mere act of using perception sharpens our vision, hearing, touch, taste, and smell. For example, youre able to distinguish subtle differences in speech sounds (e.g., the utterances bill and bell) that can baffle the ears of a nonnative English speaker. Moreover, we can purposefully set about educating our perceptual systems, achieving levels of discriminability that far exceed those of the average individual. Among wine aficionados and aficionadas exist individuals whose broad repertoire of tasting experiences allow them to identify accurately the grape from which a wine is made, the region within a given country where that grape was grown, and the year in which the grapes were harvested. These individuals have spent years learning all the nuisancescolor, odor, tastethat distinguish wines from one another. Perception psychologists have long recognized that the senses can be educated to a remarkable degree, and you will see evidence of this theme in many of the coming chapters. You will also learn what kinds of changes occur within the nervous system to promote those perceptual changes. Perception Is Supplemented by Cognitive and Affective Influences The purpose of perception is to inform us about the objects and events in our im-

mediate environment that can intelligently guide our behavior in a busy, potentially dangerous world. Should the information guiding our perceptual decisions be limited solely to the sensory data gathered by the peripheral receptors? Or should sensory data be tempered or amplified by expectations? Reflexive actions based on sensory signals work adequately for some species, including frogs, whose very limited behavioral repertoire requires little in the way of interpretation of that sensory data. If a frog is hungry and a small black moving object is within its field of view, the frog will mindlessly flick its long tongue at the black object on the assumption that its an insect. But humans constantly make more refined behavioral decisions based on the perceptual data bombarding us. Those behavioral decisions are guided by knowledge that places the perceptual information in context and by emotional factors that help us anticipate the impact of different perceptually guided decisions. Perception, in other words, can be shaped by knowledge and expectations and can be tinged by affective reaction. Thus, your perceptual reaction to the appearance of a person at your door depends on whether the persons visit was expected and whether the person is a friend or a stranger. Perception is susceptible to all sorts of nonsensory influences, and we will highlight those influences throughout the upcoming chapters. With these introductory comments in place, were now ready to tackle a question that has fascinated and puzzled philosophers, scientists, and curious laypeople for centuries: how is it that we establish perceptual contact with the world in which we live?

S U M M A R Y

A N D

P R E V I E W

This chapter lays the framework for our analysis of perception, including the philosophical assumptions well be making. We have mentioned some of the practical and theoretical reasons for wanting to know more about perception. And we have outlined three distinct, though complementary, ways of understanding perception: the psychological, biological, and theoretical approaches. Now that this general framework is in place, the next chapter will

begin to fill in the pieces, starting with some fundamentals about the organ we use for seeingthe eye. Although this book also covers hearing, taste, smell, and touch, it devotes more space to seeing. We know more about vision than the other senses, and we believe vision represents the richest source of environmental information. This preeminence of vision is mirrored in the large portion of the human brain devoted to vision.

28

Chapter One Introduction to Perception

K E Y
Antons syndrome Charles Bonnet syndrome dualism event-related potential (ERP) evoked potential (EP) far or distance senses fMRI (functional magnetic resonance imaging) illusions Kanizsa figure kinesthesia

T E R M S
psychophysics sensory transduction solipsism specific nerve energies stimulus subjective idealism transcranial magnetic stimulation (TMS)

lesion materialism MEG (magnetoencephalography) naive realism natural scene statistics near senses Necker cube perception PET (positron-emission tomography) scan