Chapter 5 Chapter 3 Seeing

"seeing is believing". At the dinner table, non-scientific people often ask me what I'm working on, and when I reply that I'm thinking about something about the mammalian visual system—how we see—they tend to be a little surprised. Embarrassed silence.The questioner is often puzzled why something as simple as seeing something should be difficult.After all, when we open our eyes, we can see an open and clear world full of colorful objects without much effort.Everything seems effortless, so what's the problem?Of course, if I'm now delving into the mentally demanding problems of mathematics, chemistry, or even economics, there may be something worth talking about.However, see...?

Plus, many people think that since their brains are working so well, why bother?The main problem with the brain, they argue, is how we treat it when it goes wrong.Only a scientifically minded few would go any further and ask: How exactly does the brain work when we look at an object? Two aspects of our existing knowledge of the visual system are quite surprising. First, the amount of knowledge we already possess is huge no matter what standard is used. The school has a complete set of visual psychology (such as: The conditions under which still images presented in rapid succession on a movie screen produce smooth motion), Physiology of Vision (the structure and behavior of the eye and related brain regions) and Molecular and Cellular Biology of Vision (nerve cells and their constituent molecules).This knowledge is the result of years of painstaking work by many experimentalists and theorists working on humans and animals.

Another surprise is that, despite all this work, we really don't have a clear idea of ​​how to see things.This fact is often concealed from those students who pursue these courses.Of course, if after all this careful research and exhaustive discussion we still lack a clear, scientific understanding of the visual process, it probably shouldn't be.By strictly scientific (e.g., physics, chemistry, molecular biology) standards, we don't even have a general understanding of how the brain produces vivid visual awareness, we just take it for granted.We know some bits and pieces of this process, but we lack the data and ideas to answer some of the simplest questions: How do we see color?What happens when I recall an image of a familiar face?etc.

But there is a third oddity.You probably already have a rough idea of ​​how you see things.You think of each eye as a tiny television camera, using a corneal lens to focus the outside world onto a special retinal screen behind the eye.Each retina has millions of "photoreceptors" that respond to photons entering the eye.Then, you put together the images coming into your brain from both eyes so that you can see.Before considering these questions, you may have certain ideas about what might happen.But, perhaps to your surprise, even though scientists don't yet know how we see, it's easy to show that you're simplistic, and in many cases, completely wrong about how we see.

The picture most of us imagine is that somewhere in our brains there is a little man trying to mimic what the brain is doing. We call this the Fallacy of the Homunculus. In Latin for homunculus meaning dwarf).Many people do feel this way (at a certain point, this fact itself needs explanation).But our "surprise hypothesis" doesn't think so.Roughly, he says, it thinks that "all of this is done by neurons." With this assumption, the problem of seeing takes on entirely new properties.In short, there must be some structure or operation in the brain that behaves as if it corresponds in some mysterious way to the mental image of the "dwarf".But what would they be?In order to study this puzzle, we must understand the task involved and the biological machinery in the mind that accomplishes it.

Why do you need a vision system?An ingenious answer is that it enables you or helps your relatives reproduce.But this answer is too general, and we don't get much out of it.In fact, animals need to use the visual system to find food, avoid predators and other dangers, and to mate, raise offspring (for some species) and so on.So a good vision system is invaluable. According to John Allman, a neurobiologist at the California Institute of Technology, mammals need to conserve more heat than reptiles because of their constant activity and relatively high and constant body temperature .This is especially true for small mammals because of their large surface area compared to their bulk, resulting in soft fur (a uniquely mammalian attribute) and highly developed neocortex.He believes that the development of this brain region made early mammals smarter and able to find enough food to maintain their body temperature.

Despite their high intelligence, mammals, as a class of animals, do not have a special visual system.This may be because they evolved from small nocturnal animals whose vision was far less important than their senses of smell and hearing.Primates (monkeys, apes, and humans) are the exception.Most of them have highly evolved vision, but like humans, their sense of smell may be poor. After the extinction of the dinosaurs, these early mammals quickly developed and replaced the ecological vacuum left by the dinosaurs.The brighter brains of mammals helped them to perform these tasks efficiently, which eventually led to the emergence of humans, the most intelligent of all mammals.

What is the purpose of mammalian eyes?The photons that enter our eyes can only tell us the brightness and certain wavelength information of a certain part of the field of view, but what you want to know is what is there, what it is doing and what it may be doing.In other words, you need to look at objects, their movement, and their "meaning".I.e. what they usually do, what are they used for, in what context have you seen them or similar in the past, etc. You need more than this information in order to survive and reproduce.In computing terms, you have to be "real-time", that is, quickly enough to act on this information before it becomes obsolete.It doesn't make much sense to be highly accurate if it takes a week to calculate tomorrow's weather forecast.Therefore, it is extremely important to extract vivid information as soon as possible.This is especially important for predators and prey when animals are trying to hunt other animals.

Therefore, the eye and brain must analyze the light information entering the eye in order to obtain all this important information.How does it accomplish this task?Before describing in more detail what is involved in watching, first let me make the following three basic comments. 1.You can easily be tricked by your visual system. 2.The visual information provided by our eyes can be ambiguous. 3.Seeing is a constructive process. Although the three are not related, we will describe them in turn. You can easily be tricked by your visual system.For example, many people believe that they can see everything with equal clarity.Just as when I look at a garden through a window, I get the impression that the shrubs in front of me are as clear as the trees on the right, and if I hold my eyes still for a short time it is easy to see that this perception is false.Only when I get close to the center of gaze, can I see the details of objects. When I deviate from the center of gaze, my vision becomes more and more blurred. When I reach the outermost periphery of my field of vision, I have difficulty even distinguishing objects. In daily life, the reason why this limitation is not obvious Apparently, it is our tendency to move our eyes constantly, giving us the illusion that objects everywhere are equally clear.

Pick up a colored object, such as a blue pen or a red playing card, and place it completely out of sight behind the side of your head.Then, slowly move it forward so that it just enters the edge of your field of vision, keeping your eyes still!At this time, if you shake the object, you will feel that there is something moving there before you can see what it is.Before you can tell what color that pen is, you can tell if it's horizontal or vertical.Until you move it very close to the center of gaze, even though you can see its shape and color, you still can't see the details of the object.My pen has an "extra fine point" logo on it.It's printed very small, but I can read it quite clearly with my glasses on and hold it a foot away.However, if I put my finger next to the pen and fixate on the fingertip instead of the pen, I can't read what's written on the pen, even though they are very close to the center of fixation.My visual acuity decreases rapidly with distance from the center of fixation.

To demonstrate in a simple and straightforward way how the visual system can trick us, let's look at Figure 1, where you immediately see a horizontal strip of texture surrounded by a background.The left side of the background is black, then gradually whitens from left to right, the horizontal strip itself, the left side looks obviously brighter than the right side, but in fact, the brightness of its texture is all the way across the width of the horizontal strip is even.You can easily see this if you block out the background with your hand. Our visual system can also trick us in more subtle ways.Figure 2 is the famous Kanizsa triangle, named after Gaetano Kanizsa, an Italian psychologist who worked in Trieste.You will see a large white equilateral triangle appearing in front of three black discs ①.And this white triangle may appear brighter than the rest of the graph. The outline of this illusion white triangle is often referred to as the "illusion outline" because there is no real outline there.When you block most of the graphics with your hand and only expose a short "outline", you will find that the paper surface that originally had a visible outline now appears to be of uniform brightness without any outline. My second general comment is that any kind of visual information our eyes give us is often ambiguous.By itself it does not provide enough information to allow us to give a definite interpretation of objects in the real world.In fact, there are often multiple plausible alternative explanations. An obvious example is seeing objects in three dimensions.If you hold your head still and close one eye, you can still get some degree of depth perception.The only visual information at this point comes from the two-dimensional image on the retina of your open eye.If the object directly in front of you is a square frame with a uniform white background at a certain distance (Figure 3a), you will of course see it as a square. However, this wireframe figure may not actually be a square at all, but instead is produced by a slanted quadrilateral of a certain shape (see Figure 3b), and its retinal image is exactly the same as the square facing you exactly the same.In addition, there will be a large number of distorted other wireframe graphics that can form the same retina image. This example may seem a bit too special, because a person rarely closes one eye and fixes his head to observe the world.If you observe a photo or a sketch of a scene, at this time, even if you turn your head and use your eyes, you can only see a flat photo or picture.But in most cases, you can still see the three-dimensional information expressed in the drawing. Certain simple line drawings may have several equally possible interpretations.Please see Figure 4.The figure consists of twelve consecutive black straight lines drawn on the surface of the paper.But almost everyone sees it as a three-dimensional cube outline map. This particular shape, known as a Necker cube, has an interesting property.If you look at the figure more steadily for a while, the cube will flip, as if the viewing angle has changed.After a while, the perception will switch back to the original one.In this case, there are two equally possible three-dimensional interpretations of the image, and the brain cannot decide which is preferable.But it's worth noting that there can only be one explanation at a time, not a fancy mix of the two. Different interpretations of visual images are examples of what is called an "ill-posed problem" in mathematics.There are many possible solutions to any ill-posed problem.They are equally plausible without any additional information attached.To get the real solution, the closest to what's really there (sometimes measured by other tests, like walking up to it and touching it), we need to use what's called "constraints" in mathematics.In other words, the vision system has to come up with inherent assumptions about how to best interpret input information. The reason why we usually see things without uncertainty is that the brain combines information provided by many salient features of the visual scene, such as shape, color, movement, etc., and takes all these different visual cues into account. The most reasonable explanation was proposed. My third general comment is that seeing is a constructive process, i.e. the brain does not passively record visual information entering the eyes, as the examples above show, the brain actively seeks interpretation of this information.Another prominent example is the "fill" process.One type of filling is associated with the blind spot, which occurs because the optic nerve fibers that connect the eye to the brain need to leave the eye at a certain point, so there are no photoreceptors in a small area of ​​the retina.Please close or cover one eye and stare straight ahead.Hold a finger vertically and place it about a foot from the tip of the nose so that the tip of the finger is approximately at the same level as the center of the eye, and move the finger horizontally so that it is about 15 degrees away from the center of the gaze.A little searching and you'll find a place where you can't see your fingertips (be sure to stare straight ahead).This small area within your field of vision is the blind spot. Despite the blind spot here, there doesn't appear to be a noticeable hole in your field of vision.For example, as I said before, when I look at the lawn outside from the window at home, even if I close one eye and look straight ahead, I can't feel that there is a hole in the lawn.As surprising as it may seem, the brain tries to fill in what should be in the upper blind spot with accurate guesses.Exactly how the brain makes this inference is what psychologists and neuroscientists are trying to figure out. (I discuss the filling process more fully in Chapter 4.) At the beginning of this chapter I gave a phrase "seeing is believing".In common parlance it means that if you see something, you should believe it exists. I'm going to offer a completely different interpretation of this mystical idiom: what you see doesn't necessarily exist, It's your brain thinking it exists.In many cases, it does match the properties of the visual world.But in some cases, blind "belief" can lead to mistakes.Seeing is an active construction process.Your brain interprets it best based on previous experience and the limited and vague information your eyes provide.Evolution has ensured that the brain is usually quite successful at these kinds of tasks, but that's not always the case. Psychologists are so interested in studying optical illusions because some of the defects in the visual system can reveal exactly what's going on in that system. The way the organization is organized provides some useful clues. So how should we think about vision?Let's take as a starting point the naive point of view of people who don't take vision seriously) It was clear that I seemed to have a "picture" of the world in front of me in my head.But few believe that somewhere in the brain there is a real screen that produces patterns of light that correspond to the outside world.We all know that devices such as televisions can do this kind of work.However, in the open skull, we did not find regular arrays of brain cells emitting light of various colors.Of course, TV picture information is not just displayed on its screen.If you use a special computer program to create art, you will find that the information needed to form a picture is not stored in the pattern of light.Instead, it is stored in the computer's memory as a sequence of charges in a memory chip, where it may be stored as a regular sub-array of numbers, with each number representing the intensity of light at that point.This memory doesn't look like graphics, yet a computer can use it to generate images on the screen. Here we give an example of a symbol: the information stored by the computer is not an image, but a symbolic representation of the image.A symbol is like a word, in that it represents one thing for another.The word dog signifies an animal, but no one would regard the word itself as a real animal.Symbols are not necessarily words, such as a red traffic light for "Stop".Clearly, what we expect to find in the brain is some sort of symbolic representation of the visual scene. So, you might ask, why don't we have a symbolic screen in our brain?Suppose the screen consists of an ordered array of nerve cells, each operating on a specific "point" in the image, with an intensity of activity proportional to the light intensity at that point.If the spot is bright, the cell is active, and if there is no light, the cell stops active. (Combinations of three cells per dot allow simultaneous processing of colors.) In this way, representations are symbolic, and the cells on the imaginary screen produce not light but some electrical activity representing symbols of light .Isn't that all we want? The problem with this arrangement is that it cannot "perceive" anything but each small spot of light.It can see no more than your TV can see.You can say to your friend, "Please let me know when that nice girl starts reading the news." But trying to get your TV to do that is futile.We cannot design a television to recognize a woman, let alone a particular woman performing a certain action.But your brain (or your friend's brain) does it without breaking a sweat. So the brain can't be just a collection of cells that merely represent what kind of light intensity there is.It must produce a higher-level symbolic description, presumably a sequence of higher-level symbolic descriptions.As we have seen, this is not a one-shot thing, since it has to find the best interpretation of the visual signal with the help of previous experience.Therefore, what the brain needs to construct is a multi-level interpretation of the external visual scene, usually in terms of objects, events and their meanings. Since an object (such as a face) is usually composed of various parts (such as eyes, nose, mouth, etc.), These sections are in turn composed of their subsections, so symbolic interpretation is likely to occur at several levels. Of course, these higher-level interpretations are already implicit in the light patterns on the retina.But that is not enough, the brain must also make these explanations explicit.A clear representation of an object is symbolic and requires no further processing.Implicit representations already contain this information, but must undergo further processing to make it explicit.It is easy to make a TV give a signal when a red dot appears somewhere on the screen, just add a small device to the TV, but if you want to design a TV that uses It flashes when it sees a woman's face anywhere on the screen, requiring more complex information processing.This is so difficult that today we cannot build complex devices for such a task. Once something is symbolized in a clear form, that information can easily become universal.It can be used both for further processing and for an action.In neural terminology, "clarity" probably means that the firing of nerve cells must be able to represent this kind of information relatively directly.It seems plausible, therefore, that to "see" a scene we need its explicit, multi-level symbolic1 interpretation. For many people, it is difficult to accept that what we see is only a symbolic interpretation of the world.Because everything seems to be "the real thing", in fact, we do not have direct knowledge of the various objects in the world around us.This is nothing more than an illusion created by the highly efficient visual system, because, as we have seen, our interpretations occasionally go wrong.Yet people prefer to believe in the existence of a disembodied soul who, by means of the ingenious apparatus of the brain, produces actual vision in some mystical way.These people are called "dualists" (dualists), they believe that matter is one thing, and spirit is another thing entirely.In contrast, our surprising hypothesis suggests that this is not the case.All of this is done by nerve cells.What we are considering is how to decide between the two experimentally. ① A more accurate term should be stimulus field (stimulus field).But for most readers, I think Visual field, field of vision, visual scene would be more appropriate.What is important, of course, is the distinction between objects in the external world and the corresponding processes in your mind when looking at them. ①The actual shape of the single black area in the picture - the notched disc, often called "Pacmen". ①The use of the word sign does not imply that there is really a homunculus.It merely shows that neuron firing is closely related to some aspect of the visual world, and whether this notation should be considered a vector (rather than just a scalar) is a tricky question which I will not consider here.In other words, how are individual symbols distributed?