Chapter 24 Chapter Fourteen Perceptual Psychologist-2

In nature, unlike in a laboratory, no form or movement can exist without being three-dimensional.In order to understand shape and movement perception in everyday life, it is crucial to understand depth perception.Psychologists have long considered this to be the central question about perception.All their literature sources on depth perception would fill the space of a book. The basic question has always been obvious and simple: how do we see the three-dimensional world when our source of information, the images on the retina, is essentially two-dimensional?Why don't we see a flat world, like in a color photograph, where the distance and three-dimensional nature of each object can only be hinted at by its size, viewpoint, shadows, and other cues?

In fact, these hints are the answers provided by a set of theories.These theories come in various forms, but they all hold that depth perception is not automatic and innate.Some theories say it's the result of experiences that make us associate depth with cues; others say it's a learned mental process by which we infer depth from cues. The idea that depth perception is a product of our association of cues with depth experience begins with Locke and Berkeley.From their day to the present, psychologists in the associative-behaviourist tradition have held that we all associate, consciously or not, cues of two-dimensional images of the retina with experiences of how far objects are from us, so that Generated some hints.

Another view is that our ability to perceive depth is a result of logical reasoning about what we see.It was first proposed by J. S. Mill in 1843. When he mentioned perception, he said that one-tenth of what we observe is observation, and nine-tenths is reasoning.Later in the century, Helmholtz proposed in more detail that we unconsciously infer three-dimensional reality from two-dimensional images on the retina.Psychologists of a range of cognitive orientations, then and now, have argued that perception, including depth perception, is partly, or largely, a product of higher mental functions—"sort of like thought processes" Irving Rock put it this way - reasoning from cues is only one of these processes.

No matter which term one prefers, cues to depth are extremely familiar in everyday life, and their role in perception has been shown by hundreds of experiments.Here are some main leads and some representative experiments: - Apparent size: The farther away an object is, the smaller it appears to be, however, if we already know how big it is - say a person - we will infer the distance from its displayed size, even if it is at a Same goes for the featureless plane, where we don't get any hints.In an experiment carried out in 1951, a researcher made playing cards ranging from half to double the normal size, and then asked subjects to look at the cards under laboratory conditions, so there was no relevant information. Tips for distance.The subjects thought the double-sized cards were closer to them, while the half-sized cards were farther away.All cards are actually within the same distance.

- Intervention: When an object is partially occluded by another object, we realize that the occluded object will be farther away than the occluded object.When we look at the cityscape from a distance, we can easily feel the distance of a distant high-rise building, because some buildings on the lower floors block the floors below the high-rise building; on the other hand, in the sea, the distance of a floating object is very Difficult to judge. ——Point of View: Parallel lines going outward from the viewer's side, such as railway tracks or corners of walls, floors and walkways, will converge with distance.How much we are influenced by such cues can be seen in the pattern below.

The viewpoint slope is that it is impossible to see the columns as being the same size, even though they are. - the texture of the surface of the object - a patch of grass, a cement sidewalk - is constant, however, the finer and finer lines are seen from a distance as an important clue that anything is on the surface on the distance. ——Buildings or hills in the distance look plain and blurrier than nearby objects, because there are a lot of air layers in between. - Parallax: the changing relationship of objects to each other as we move - it is an important depth information, especially when looking at near objects relative to distant ones.

- Convergence and adaptation: When we look at something close to us, the eye angles inward, and the muscles next to each lens fail to focus it.When we look at distant objects, the eyes are parallel and the lenses are in a relaxed state.Symbiotic visceral sensations are some important cues that tell us the distance of objects at 10 feet and closer. ——Binocular difference: When we look at a relatively close object, its image will fall on the center of each eye’s vision—the center of the retina, while the images of some objects that are equally far away will fall on both eyes. where the retina corresponds.Regardless of the distance of the object, its image will fall on different places on the retina, as shown in the following figure:

Differences between the two retinal images are interpreted by the brain to indicate which object is farther away.The binocular difference is most pronounced in close-ups from 800-1900 feet.Some theories of perception suggest that this is the most important cue for depth. All of the aforementioned depth cues can be explained by intrinsic mechanisms or learned behaviors.The innate element of depth perception, however, is supported by other, more convincing evidence. Depth perception is shown to be instinctual in a series of historic experiments conducted at Cornell University in the 1950s and early 1960s by Alina Gibson on high-level reading of pronounceable and unpronounceable words. The work, which we mentioned earlier, she did these experiments with her colleague Richard Volker.Gibson, who had a lifelong fear of cliffs, and Volker, who had trained troops jump off high platforms during World War II, teamed up to create a "visual cliff" to establish whether rats had learned depth perception or innate There is such a skill.The visual cliff is a large pane of glass under which half is covered with corrugated wallpaper, and the other half is also covered with wallpaper, but a few feet below.The problem is that animals with no depth experience—that is, things that have never jumped down from any kind of height—will automatically avoid things that look like they are going to jump down.

Researchers kept chickens, rats, and other animals in the dark, preventing them from having any experience of depth, and placed them on a board that ran across the glass from the shadowed side to the other that looked deeper The place.The results were dramatic.The animal, although never experiencing depth, always avoided the darker side and backed up on the board to the shaded side.Gibson and Volker then experimented with human infants.As Gibson recalled not long ago: We cannot nurse babies in the dark and have to wait until they are able to move on their own, using their avoidance of edges as our depth differentiator.However, crawling babies do avoid the "deep" side.They may have learned something before crawling, but whatever it was, it wouldn't be externally reinforced because the parents never reported that the baby had ever fallen from a height Pass.

The mother of each baby will stand to the left or right of the installation and beckon to the child.In almost all cases, the babies crawled step by step towards their mothers when they were in the shadows, but only 3 out of 27 babies dared to crawl to the deep side. (Recent experiments by others tend to weaken Gibson and Walker's conclusions, showing that human infants' fear of heights is acquired—not through the experience of falling, but through movement in general gained from experience.) The extremely convincing evidence that depth perception is built into the nervous system came in 1960 from an unlikely source, AT&T's Bell Laboratories, and an equally unlikely researcher, a man working on television signals. Young electrical engineer launching specialist.Bella Juletz was born in Hungary and received education there. After the failure of the revolution in 1956, he came to the United States and was hired by Bell Labs in the newly translated West Murray Mountain. He mainly solved the problem of channel width used by compressed TV line numbers.However, Jules was attracted by some more interesting problems, and from 1959, with the tacit approval of Bell Labs, decided to specialize in problems of human vision.Although he never had a degree in psychology, he soon became a multi-awarded perceptual psychologist. Director of the Vision Research Lab at Jess University.

Juletz had an idea that made him instantly famous in psychology circles when he was just beginning to study vision.While reading books on physical depth perception, he discovered to his surprise that the overall reception of physical vision was the result of the brain comparing some cues with shape and depth in each eye's image.This is thought to form the coincidence of imagery and depth perception.Jules, who had previously worked as a radar engineer in Hungary, felt that this was clearly wrong: After all, to unmask aerial reconnaissance, one would look at the aerial image (from two slightly different positions) through a kind of solid mirror, and the masked target would pop out with striking depth.Of course, in real life, ideal masks do not exist, and after observation through a solid mirror, one can detect with one eye faint cues that might distinguish an object from its background.I therefore used the largest computer, the IBM 704, which had just arrived at Bell Labs at the time, to simulate the ideal camouflaged entity image. This consists of images made randomly using black and white dots, like these two patterns: If one of the patterns is looked at alone, there is no depth cue in the two patterns.However, although they are mostly the same, there is a small area in the center that has been shifted slightly to one side by the computer, so that when the patterns overlap when one looks at each image with one eye A type of binocular parallax is created and appears to float from the background. (To see this amazing effect, hold a 4-by-4-inch piece of cardboard or paper upright in front of you, perpendicular to the page, so that each eye is looking at only one of the patterns. Watch One corner of the pattern, and after a while, the two patterns will move toward each other and coincide. At this point, the center square will appear to lift about an inch off the page.) This elevation of random dots is much more than just a fun trick.It proves that physical vision does not rely on cues on each retina to form the experience of three-dimensional qualities, and that, in turn, the brain superimposes meaningless images to reveal hidden cues of three-dimensional qualities.This is not a cognitive process, a matter of learning how to interpret depth cues, but an innate neurophysiological process that takes place in a particular layer of the visual cortex.It is here that some closely organized interacting cells interact with the midpoint of the pattern, and the superimposed three-dimensional effect perception is obtained. (Physical effects aren't the only way we get depth perception. Jules' work doesn't rule out other approaches, including those involving learning.) Juletz was proud to see that his discovery led Huber and Wiesel and others to turn their attention from shape perception to the investigation of binocular parallax, but he said modestly: I have never considered my role in introducing random point-of-view into psychology a major intellectual achievement, although many of its results have been beneficial to the study of the brain.It's just a lucky coincidence, a collision between two cultures, an association between two languages ​​in the mind of a bilingual person (the language of psychologists and engineers). There is another modern theory of depth perception that is neither specifically neurological nor specifically cognitive.Not that its proponent cleverly merged the two; rather, he effectively dismissed neurological and cognitive theories as unnecessary and based on false assumptions. Only a daring geek could throw away a century of depth perception research and declare that he has found a new and correct approach.Only a true nonconformist would be able to say with certainty that we perceive depth neither through neural detectors nor through inference using cues, but "directly" and automatically.Only a hothead would come up with a radical epistemology that argues that the physical properties of light are such that what comes into the eye is an accurate and actual experience of depth, and that we don't need to explain what we see because what we see Just the actual stuff. One such man was the late James J. Gibson (1904-1980), considered by his admirers "the most important scholar of visual perception in the 20th century" and "the most original theorist in the world of perceptual psychology." , however, his theory is regarded as "extremely implausible" by most perception experts (one reviewer even thinks his theory is too "stupid" to be worth discussing at all), and no expert advocates his theory . Gibson did little to help his cause when he formulated his radical conception of perception between 1950 and 1980, and he was even more contemptuous of existing theories of perception in mainstream psychology in general.Typical of his comments are: "Psychology, or at least American psychology, is an inferior discipline . Answers to the wrong questions; the questions chosen for study are those that are convenient for the study, not the relevant ones".But despite his blustery-sounding opinions and severe hearing problems, he was always peaceful, charming, and very friendly, liked by almost everyone he came in contact with, a passionate editor The chronicler found him to be "very charming, lively and lovable". Gibson was born in a small riverside town in Ohio and grew up throughout the Midwest, the son of a railroad sightfinder.This gave him many opportunities to ride trains and experience many things on the railroads that would later become central elements of his theory of visual perception.As he said in a short autobiography: "When I was 8 years old, I learned what the world looked like from the railway. When I stood at the back of the train and looked at the world, it seemed to flow inward. And when viewed from the locomotive, it expands outward.” Gibson went to Princeton University, but he felt that some people here were not satisfactory, and he preferred to be close to some people he called "crazy".For a while he tossed back and forth between philosophy and social life (he was curly-haired, square-faced, and handsome enough to be a leading figure), but after his senior year he took psychology, which he liked immediately. This subject is gone.He was influenced by behaviorist psychology when he was a graduate student in psychology at Princeton.In 1928, however, he accepted an appointment as a teacher at Smith College, where he met Kurt Koffka, and then, though he did not become a Gestalt scholar, he was inspired by Gestalt psychology on organization and structure. Great influence on the concept. For some years, Gibson has been interested in both social psychology and more traditional studies of perception.Then, during World War II, he was hired by the Army Aviation Personnel Aeronautical Psychology Program to develop a depth perception test to determine who had the visual acuity necessary to fly, especially for successful takeoffs and landings. Perhaps because of some of Gibson's early experiences on trains, he found traditional depth perception cues, including shadows and points of view, to be of little value.In his view, these are based on oil paintings and stereoscopic cameras in the living room, rather than the three-dimensional qualities of reality, based on static images rather than movement.Far more useful and realistic to him were two cues: texture gradients, like the jagged roughness of a runway seen by pilots when they take their final foot close to the ground; Point of view, or the flow of changing relationships between objects as one moves through the environment, includes everything a pilot sees during takeoff and landing.These cues were quickly accepted and are still part of the cue-based theory of depth perception today. Gibson's work with aviation personnel contained the essence of his later views.The most crucial mechanism in depth perception (in all perception, according to Gibson) is not the retinal image, despite its many cues, but the ever-changing flow of relative relationships between objects and their surroundings in which the perceiver moves. the surface among them. This concept would come to dominate Gibson's thinking in the 1950s and 1960s, when he did a considerable amount of research at Cornell University and tested his belief in texture gradients.In some tests, he placed divergent feeding bottles on the viewer and the textured surface; in others, he caused the viewer's eyes to swell to avoid focusing too much on the texture.In other experiments, he cut a Ping-Pong ball in half to make goggles, so that his subjects saw things like frogs, without surface or volume.From experiments like these, and from careful consideration of his studies of aviation personnel, Gibson slowly discarded texture gradients, emphasizing the movement of the observer through the environment as the key to depth perception.No matter how big or small the movement is, it will bring about a change in the visual arrangement - the structured pattern of light coming from the environment into the eye - such as in the next image: Visual alignment is rich in information seen from any angle, and it becomes infinitely richer as the observer moves.Even the slightest change in the head alters this arrangement, which alters the appearance of the object seen, resulting in this or that optical flow.Gibson came to believe that there was enough, far more than enough information in the visual arrangement, and flow, to convey depth and distance directly, without mental calculations or extrapolation from cues at all. The following is the depth perception explained by Gibson in his general theory of "immediate perception" and "ecology".Alas, Gibson, the outsider and geek, who was, in the words of one of his fellow psychologists, "extremely stubborn and uncompromising," was determined to throw the baby out with the bathwater.For, it is possible to admit that the neural and cognitive perspectives of depth perception correctly explain different aspects of a phenomenon, and that Gibbonson's views are complementary to both.However, for Gibson J. James, this is impossible. "Visual perception," Bella Jules recently said, "is in the same state as physics before Galileo, or biochemistry before Watson and Crick discovered the double helix of DNA." Somewhat harsh, but the two main approaches—three if we juxtapose Gibson's theory with the others—both explain only some aspects of these phenomena, and there is no comprehensive theory of visual perception.This may mean that some large organizing concept has not yet been discovered, or that visual perception is so complex that no one theory can explain all its aspects, and different methods explain different complex problems. We've seen a few different approaches.Here, we do the finishing work and make a comprehensive outline of their explanation of visual perception in general. Neuron theory: This method answers the question that physiologists in the 19th century have always been fascinated by: Although the sensory nerves are all the same in structure, how does it transmit different sensations to the brain? The answer, given in great detail, is that the nerve impulses themselves don't make a difference; instead, the receptors that respond to certain stimuli send their signals separately to the striatum, the main area of ​​the visual cortex.Huber and his current colleague Margaret Livingstone (with whom Wiesel split off after 20 years and is now president of Rockefeller University) have recently shown that a combination of form, movement, depth and The color-generated impulses reach the cortex via different parallel channels, where they later form a whole. (Color is a much-studied subject; it is only a peripheral part of the focus of this book and is therefore ignored.) The primary visual cortex covers only about 15 square centimeters, but its internal architecture is extremely complex.Neurophysiologists have been exploring this structure and the wiring within it for decades.They learned that incoming information goes first to "simple" cells, where it is tuned to respond to specific stimuli.These cells send their impulses to "complex" cells through extremely complex circuitry that is largely genetically determined.Complex cells begin to integrate the individual pulses and mix the information from both eyes.As a result, retinal images are sent to the visual cortex as "maps" of collective firing of complex neurons, but the patterns of these firings do not resemble images on the retina or scenes outside the eye.According to Huber and Wessel: What does the visual scene actually look like when it is projected into the visual cortex?Suppose an animal is staring at a certain point, and the only object in the field of vision is slightly above and to the left of this line, and the animal's gaze is fixed on this point.What would the pattern look like if every active cell were to fire, and if a person could stand on top of the cortex and look down?To make the problem more interesting, suppose the pattern is seen with only one eye...the pattern turns out not to be a straight line, but just a set of regularly spaced strings. In other words, it is not an image, but a coded image expression, which is a bit like the pattern of the magnetic field on a tape is not a sound, but a coded sound expression.However, this expression is not yet perception; as Huber and Wiesel put it: "The primary visual cortex is by no means the end of neural transmission. Maybe it's just an early stage." Partially aggregated and integrated information is sent from the striatal area to another area of ​​the visual cortex, and on to higher cortical areas beyond it.Here, the information is finally seen by the mind and recognized as belonging to something familiar or not seen before.How this happens has so far not been figured out, and most neurophysiologists see it that way.A few, however, have ventured to conjecture that, at higher levels of the brain, there are "traces" containing previously seen objects in the form of synaptic connections or molecular Only when the information matches the trace will it react.This response to the cooperative partner is an awareness ("I've seen the face"); non-cooperation does not elicit a response, which is also an awareness ("I haven't seen the face.") Research The authors half-jokingly, half-seriously refer to these hypothetical neurons of the visual system as "grandmother" cells, because some of them only respond to seeing their own grandmother-coded version of themselves, while others respond. Any signal is ignored. Beyond these imaginary things, neural theory tells us a lot about how visual perception works on a micro level, but no macro theory; says a lot about the mechanisms of vision, but not about its owners and operators, A lot is said about neuronal responses, but not about perceptual experience.As one cognitive theorist put it: "Understanding perception by studying only neurons is like understanding bird flight by studying only feathers." Cognitive Theory: This approach deals primarily with the mental processes at work in perceptual phenomena like shape conformity, feature recognition, shape recognition, depth perception from cues, and recognition of people when much information is lost. The mental processes that lead to these results are made up of billions of neural phenomena, yet cognitive theorists believe that macro- rather than micro-theories are needed to explain these processes.A physicist studying how a wave changes shape and crumbles as it approaches a beach would not derive the laws of wave mechanics from the interactions of countless water molecules, even with a supercomputer.These laws express collective effects, and they exist in an entirely different organization.The sound a person makes when he talks to us is made of the vibrations of molecules in the atmosphere, but the meaning of the words can never be expressed in these molecular forms. The same is true of the psychological process of visual perception.They are the organizing effects of a collective of neural phenomena, expressed by psychological rather than neurophysiological laws.We have seen evidence of this, but one particularly interesting example is worth discussing.What happens, and at what level, when we call up an image from memory and look at it in the mind's eye?Some experiments performed by cognitive theorists have shown that this can only be explained in terms of advanced cognitive theories.The most fluent and expressive experiment is the "mental rotation" experiment conducted by Roger Sheppard of Stanford University.Sheppard asked subjects to say whether the objects in the following three groups were the same: Most people, after studying the patterns, will recognize that the objects in A are the same as the objects in B. Objects inside C are not.When asked how they came to their conclusions, they said that these things were spinning inside their heads as if real things were spinning in the real world.In another experiment, Sheppard showed how this process mirrors real rotation.In this experiment, the viewer sees a given shape from an angled angle.For example, the following set of experiments shows a single shape at a range of locations: When subjects were shown these patterns, the time it took them to recognize that the objects were identical to each other was proportional to the angular difference in the positions of the patterns.That is, the more one pattern needs to be turned to compare it to another, the longer it will take to recognize it. This is just one example of many perceptual phenomena that involve higher mental processes that must operate with internalized symbols of the external world.In recent years, a number of perception researchers have attempted to formulate a comprehensive theory of cognition to explain what these processes look like and how they produce these perceptions. There are two approaches to their theory.One is to use concepts and processes from a branch of computer science called artificial intelligence.The basic assumption of artificial intelligence is that human mental activities can be imitated by computer programs step by step, and also occur step by step according to the same steps.On the one hand, artificial intelligence experts want computers to recognize what they are looking for, and on the other hand, they want to gain a better understanding of human perception. They have written a lot of shape recognition programs.To obtain basic form-recognition elements—such as recognizing triangles, squares, and other regular polygons—a procedure might proceed in a series of if-then steps.If there is a straight line, then it follows the straight line, and measures it, to the end; if another line continues from here, then it calls the point an angle, and measures the angle, changes through the angle Orientation; if this other line is a straight line, it traces it until... etc., until the numbers of sides and angles have been counted and paired with a polygon and its features. The main argument in favor of artificial intelligence methods for visual perception is that there are no projectors or screens inside the brain, and no imaginary little gnomes see things in it; therefore, the mind must not be processing images, but coded images. Data, it processes step by step, and that's how computer programs work. The main argument against artificial intelligence methods is that, compared with human programs, there is no machine vision program that can match it. There is no ability to recognize two-dimensional shapes, let alone three-dimensional images. Can sense three-dimensional distribution in the world around it, can't understand if it's in the environment, can't recognize possible physical properties of rocks, chairs, couches, water, bread, or anything it sees.As Ulrike Reiser sums it up: Many of the difficulties encountered in machine vision design can be attributed to a simple source: the lack of effective theory for the work performed by designers.Most of them thought that visual perception was a matter of recognizing a particular pattern of stimuli.If this were all the case, we should be able to produce seeing computers long ago.Failure to form a model based on these theories, the reason for the failure, can serve as evidence that the theories themselves do not work, and thus the need for another approach. Another school of thought about how cognitive perceptual processes work relies on laboratory studies of human minds rather than the imitation of minds performed by machines.The origins of this idea go as far back as Helmholtz's time, when the conventional view was that perception is the result of unconscious reasoning from incomplete information, including another class of conscious thought processes.Its most prominent proponent was the aforementioned Irving Rock of the University of California.His 1983 book "The Logic of Perception" was described in the 1991 Annals of Psychology as "the most comprehensive and experimentally feasible account of perceptual effects that seem to require intellectual activity on the part of the observer". Although Rock was a brilliant sensory psychologist, he was far from it in his early student years.In fact, in an intellectual family, he was a black sheep.However, during World War II, his unit was bombed by enemy planes, and he felt that he would be killed. "I swore to myself that if I could survive, I would do more things in my lifetime than before."After the war, he became a top university student.He began his graduate school in physics but turned to psychology when he realized that the opportunities for making a greater contribution to knowledge were much greater in the young field of study. At the New School for Social Research, Rock was heavily influenced by the Gestalt schools, of which he himself became an enthusiastic researcher.Some of the basic organizational and relational thinking laws of Gestalt psychology are still a part of his theory.These laws, however, described basic automatic processes, and Rock came to believe that many perceptual phenomena could only be explained by a mental process like a thought. He started the idea in an experiment he conducted in 1957, as described above, in which he tilted a box to make it look like a diamond, and then tilted the observer.Since the observer still perceives the box as diamond-shaped, Rock reasoned that the observer must be using visual and visceral cues to explain what is seen.Rock spent many years designing and conducting other experiments to test the hypothesis that perception often requires processes at a higher level than the visual cortex.These studies eventually led him to his present thesis, that "perception is an intellectual activity because it is based on operations like those that constitute thought." Indeed, Roark says, perception may be what gives rise to thought; perhaps it is an evolutionary connection between the low-level sensory processes of primitive organisms and the high-level cognitive processes of more complex life forms.He argued that if what the eye sees is a ambiguous and deformable representation of reality, some mechanism would have to evolve to arrive at a reliable and faithful understanding of reality.按他的话说，“智力操作也许就是为了服务于知觉而进化出来的”。 这并不是说，所有的知觉都是类似于思想的。罗克特别引用了瀑布错觉，作为在低水平的神经形式上可以解释的例子。可是，有关运动知觉和其它形式的知觉的大部分事实对他都好像需要高水平的过程。无意识的推论，比如我们利用纹理阶度提示来感觉距离，只是其中的一种。对这种解释的结果的描述是另外一种。在由波林绘制的含义模糊的老巫婆与少妇图案中，人们看到的不仅仅是简单辨认一个图像的结果，而是对自己解释一个特别的曲线是什么样子的结果：像一只鼻子，或像脸。许多物体被知觉出来的形式并非立即可以辨认出来的；辨认事物究竟是什么东西是通过这样一个过程得来的。 知觉还经常需要这种或那种问题求解。人们很少认为知觉是问题求解，可罗克已经掌握了大量的证据——很多是从别人早期的研究中得来的，有些是从他自己的原创实验中得来的——可以显示，在很多情况下，我们寻找一个假说来解释我们看到的东西，把这种假设与别的可能性进行比较，然后选择好像能够解决使我们看到的东西产生意义这个问题的那一个。所有这些通常都发生在几分之一秒的时间内。 一个例子：在一个自亥姆霍兹时代以来就知道的一项实验室现象中，如果一条像波浪的曲线横向通过一个细孔，如下图所示； 大多数观察者首先看到小的元素在上下移动，可是，过一会儿后，其中一些人会突然看到这条曲线以直角在孔的前后移动。是什么东西产生了这些发生了变化的正确的知觉的呢？罗克发现，他们使用的一条提示是线条在通过小孔时不断变化的曲率；另一个是曲线的末尾，如果它能够为人们所看到的话。这些提示给思维提示了另一种假设——一条曲线平行通过小孔，而不是一个小元素在上下移动。这个假设好得多，思维很快就接受了，并认为这条线真的就是这个样子的。 罗克是这样总结他的理论的： 在理论水平上，至少按照提供在这里的一套理论来说，知觉和思想包含着推理。在有些情况下，总括或者规则是通过归纳在知觉中形成的。这些规则接着就被演绎加以利用，作为推出结论的前提。某些情形下的知觉可以概括为创造性的问题求解的结果，因为它也是在寻找一个基础（或者内部的解）从而得出一个具体的解释。知觉包含着决定，正如思想一样。最后形成知觉经验的操作与概括思维的东西是同一类型。 直接或生态知觉理论：吉布森及其追随者的直接或生态知觉理论不仅仅试图解释深度知觉，如我们前面已经说过的一样，而且要解决总体的视知觉问题。吉布森的理论，即观察者通过环境的移动会产生一个连续变化的视觉排列，不仅仅是他的深度知觉解释的中心，而且也是他对形式、大小、距离和运动知觉的中心。 吉布森于1979年在《视知觉的生态学方法》一书中阐述了他的理论，他的阐述深奥难解、穷根究底，很难读下去，部分原因是因为他创制了许多新词。可是，这本书在知觉心理学家中极其有限的接受程度，也许更多的是由于它否定了由其他人到目前为止在知觉研究中取得的几乎一切成果。吉布森典型的话如下所示（摘自他去世后发表的一篇文章）： 一个世纪以来从对知觉的研究中得出的结论是微不足道的。从一个世纪以来对感觉的研究中获取的知识是不连贯的。我们没有足够的知觉理论，我们在寻找感觉当中所发现的东西是一串混杂的错觉、生理怪事和身体的感觉。这些含义是令人沮丧的。在解决知觉的问题上，必须有一个全新的开始。 吉布森之所以决定要另起炉灶的理由听上去是足够充分的：他用这个问题（跟科夫卡一样）开始“为什么事物就是它们看上去的样子”，然后说： 我们如何看出自己处于环境之中的什么地方？我们如何知道自己是不是在移动？如果是在移动，我们是在向哪里移动？我们如何知道事物的长处在哪里？我们如何知道怎样处理事情？ 为了正儿八经地重起炉灶，吉布森排除掉了大多数基本的假设，这都是哲学家和心理学家们在视知觉方面一向采用的假想。其中有：感觉是知觉的基础；来自外界的刺激会引起有机体的反应；有机体能产生反应；大脑可以处理、整合并解释信息；对视觉的合适研究从视网膜和大脑开始；光学的运动是视觉系统必须加以补偿的东西；环境只是由视网膜上的信息部分再现出来；知觉在很大程度上取决于推论等等。吉布森可谓是一把新扫帚。 对前述假设加以全盘扫荡之后，吉布森提出了自己的新理论，这种理论多一些哲学意味，少一些实证基础，更多的是概念上的，而不是以资料以基础的，而且很难掌握。尽管如此，他还是通过实验达到他自己的结论的，首先是运动知觉，然后是形式和其它现实世界特点的形式。如我们在前面已经说过的，第一线曙光是从他参加航空人员研究中得出的，如他所言： 我们了解了更多有关物体知觉的东西，我认为这比通过外形知觉进行的实验室标准实验多得多。一方面，我有一个挥之不去的感觉，即从来没有人真正在生活中看到过一个二维的东西，也就是说，一个事物的图片。人们看到的是一连串视点的转换，无穷多的外形，它们可以确定物体变化了的外形。 从这里开始，他后来不仅排斥所有以图片和错觉为基础的研究，而且排斥就知觉作为思维对视网膜上的二维图片不完全的信息进行的解释的所有理论： 最终，我慢慢理解了，图片形式的知觉与自然的知觉方式是何等的不同。前者是二手的知觉；后者是一手的知觉。从图片中来到眼睛里的有框架的视觉排列，与来自世界而到达眼睛里的自然的视觉排列完全不同……眼睛之所以进化；是因为它们要看这个世界，而不是看一个图片。自从我意识到这些以后，不管是什么意义上的用途，我一概回避使用“视网膜图象”这个词了。 吉布森的结论我们已经在前面看到，就是说，光学排列包含了我们需要的所有有关现实世界的信息。他承认，我们的对这个世界的知识由于神经系统的特性而有限，可是，他坚持认为，这些特性取决于对这个世界的进化适应，我们知道必须了解的所有环境特点。进化会在每种动物中产生一个知觉系统，会使它直接辨认对其有用的环境的特点——按吉布森的说法是“可利用性”，即任何东西的特征的利用都会使这个物种受益。因此，事物相对于观察者的大小和生理装备来说，看上去是可食用的、可饮用的、可以在上面行走的、可以在里面游泳的等等。 简短地说，知觉不是对一种降格视网膜图象的解释过程，而是通过光学排列和光学流动直接和真实的现实体验。这对吉布森而言，就是知觉的核心机制——而不是由胡贝尔和威塞尔（他们进行的工作他认为是无关的）记录下来的神经现象，也不是认知过程，他认为认知过程是以错误和人工的假设为基础的。 直接知觉也是吉布森对贝克莱的回答。我们知道，世界就在那里，在我们之外独立地存在着，因为当我们在环境中移动时，我们以连续变化的形式看见事物，而我们会体验到事物的连续性、真实感，且不依赖我们作为观察者而存在。其它的所有动物亦是如此。只有哲学家们才会去怀疑世界不是我们看上去的样子。吉布森的视知觉理论因而就大胆突破了对视觉的研究，从而进入认识论的领域。 That's not all.到他生命快要终结的时候，吉布森慢慢相信，知觉是全部心理学的支柱，而他的知觉理论可以给这门科学的许多领域带来巨大的变化。像思维、意识、学习和驱动力这样一些概念，都可以被生态心理学所代替，而生态心理学是以动物对地点、事件和物体有用的、危险的特征的意识为基础的，也是以他们对自己的动作进行组织和控制，以达到他们在现实世界里所欲求的结果为基础的。 吉布森经由何种过程以期达到这种至高无尚的目标的，我们不得而知。《生态心理学》出版两年后他就去世了，尽管他有关视知觉，特别是光学流动的一些思想已成为被接受的知识的一部分，可是，这些知识和他更为雄心勃勃的一些概念却并没有给知觉带来一场革命，更不用说总体意义上的心理学。 很可惜，吉布森会变得这样不能宽容，这样自负，他的光学流动概念当然是很有价值的，尽管它并不会使他认为的许多东西毫无必要，可是，他爱走极端的性格使他对心理学作出的贡献不能像它应该的那样广为接受。 所有这些会把我们引向哪里？ 《心理学年报》1991年对知觉问题的回顾提出了全部三种方法——神经生理学的、认知学的以及吉布森式的方法——并认为这三种共同存在的理论和知识体系彼此并非互相排斥。以刺激为基础的神经学方法、以人类思维进行的认知方法和光学流动直接法都描述了全部现实的不同部分。他们并非彼此冲突和矛盾的，而是互补的。 从这个立场出发，知觉好像就是心理学知识当中相对发达的一个领域，尽管有许多谜尚未解开。对这些有趣的问题，还有太多而不是太少的答案，但是，现代心理学其它的研究领域也是这般情景。未来是否会产生一个包罗万象的综合理论，这还需要时日来证明。同时，我们知道，我们已经了解了比以前多得多的知觉方面的情况，而且知道，还有更多的知觉知识尚需了解。