In January 2004 a group of scientists visiting a hostel for the blind in New Delhi met a young man known in their papers as S.K. Born in Bihar, he had grown up blind because the lenses of his eyes had almost entirely gone. Surgery was recommended when he was twelve, but the family’s money ran out, so he went to a school for the blind, learned Braille, and was studying by correspondence for a master’s degree in political science. All that time, what his eyes needed most was a pair of strong glasses. That summer the scientists bought him a pair.
His sight was still far from sharp, but within two weeks he could tell colors apart, judge brightness and see which way things moved, almost as well as anyone. Then the scientists showed him two outlined squares, one overlapping the corner of the other, and asked how many objects he saw. He said three.
He counted the two L shapes and the little square where they overlapped. Shown a picture of a cube shaded as if lit from one side, he saw three objects, one for each face. Of 50 photographs of everyday things, he named about a quarter. His eyes were sending his brain everything they could. What his brain hadn’t yet learned was how to turn that into objects.
Looking feels like opening a window, but it isn’t. Your eyes, ears, nose and skin pick up thin signals, a few kinds of energy and chemicals, and your brain builds a world out of them, mostly by making good guesses. Psychologists split this into two jobs. Sensation is the detecting: special cells, called receptors, respond to light, sound, chemicals or pressure and turn them into nerve signals. Perception is what the brain makes of those signals: organizing them, interpreting them, and turning them into the experience of a room, a voice or a smell of rain. S.K. had sensation. For a while, he had very little perception.
Catching the signals
Every sense starts the same way. A receptor is a cell built to respond to one kind of thing: light in the eye, vibration in the ear, certain molecules in the nose, pressure or heat in the skin. When it responds, it turns that energy into the electrical signals nerve cells use, a step called transduction. The brain never meets light or sound itself. It gets coded messages and has to work out what sent them.
The old count of five senses, sight, hearing, smell, taste and touch, leaves a lot out. You also sense your balance, where your arms and legs are without looking, temperature and pain.
How faint can a signal be and still be noticed? Psychologists measure this as a threshold. The absolute threshold is the weakest signal a person detects half the time. Half, because at the edge the same faint light is seen on one try and missed on the next. A few years ago a team in Vienna sent single particles of light, photons, into the eyes of three men who had sat in the dark for about forty minutes. Each trial had two brief moments, and the men had to say which one held the light. On about 2,400 trials that sent exactly one photon, they were right 51.6% of the time, against 50% for pure guessing, and 60% of the times they felt sure. That is a small edge, from three people, in one study that nobody seems to have repeated, and the overall result just misses the usual test of being more than chance, passing it only with a second set of trials added. So the evidence is modest, but it suggests that, in perfect darkness, a person can sometimes see one photon.
The other threshold is about change: the smallest difference you can notice, called the just noticeable difference or difference threshold. Nearly two hundred years ago the German scientist Ernst Weber noticed that it isn’t a fixed amount. It’s a fixed fraction of what’s already there. A torch switched on in a room during load-shedding is obvious to everyone; the same torch switched on in a sunny courtyard goes unnoticed. This rule is called Weber’s law.
The senses also tune out what doesn’t change. When you walk into a kitchen where onions are frying, the smell is strong; ten minutes later you barely notice it, though the air is the same. This is sensory adaptation.
Eyes, ears, nose and skin
Light enters the eye through the clear front, the cornea, which starts the focusing. It passes through the pupil, the hole in the colored iris, which widens in dim light and narrows in bright. The lens behind it changes shape to focus near or far, and throws an upside-down image onto the retina, the thin lining at the back of the eye. The full anatomy, layer by layer, is in Grade 12 Biology.
Eye: traced from Gray’s Anatomy (1918), Fig. 869, H. V. Carter, public domain
The retina holds two kinds of receptor. Cones work in good light, see color and fine detail, and are packed into a small pit, the fovea, where you look straight at something. Rods are spread round the rest of the retina; they see no color but work in dim light, which is why a dark room slowly appears as they take over. There are three kinds of cone, most sensitive to short, medium and long wavelengths of light, and every color you see comes from comparing how strongly each responds. Beyond the retina, colors are handled in opposing pairs, red against green and blue against yellow, which is why staring at red leaves a greenish afterimage. A missing or unusual kind of cone gives color blindness, most often between red and green; about 5% of Asian men and 8% of men of European descent have it, and far fewer women.
The retina’s nerve fibers gather and leave the back of the eye as the optic nerve, and where they leave there are no receptors at all. So each eye has a blind spot, a hole in what it sees. You can find yours.
You never notice the hole, partly because each eye covers the other’s, and partly because the brain fills it in with whatever is around it. What you see is already the brain’s account of the scene, gaps patched. The optic nerves cross partly on their way, so the left half of what you see goes to the right side of the brain, and on to the occipital lobe at the back.
In the ear, sound waves run down the ear canal and shake the eardrum. Three tiny bones in the middle ear, the hammer, anvil and stirrup (malleus, incus and stapes), carry the shaking to a window in the cochlea, a fluid-filled tube coiled like a snail’s shell. There waves in the fluid bend the hairs of hair cells on a strip of tissue, the basilar membrane, which fire signals up the auditory nerve (part of the vestibulocochlear nerve) to the temporal lobe. High notes move the hair cells near the base of the coil most, low notes those near its tip. Young ears hear from about 20 to 20,000 vibrations a second. Next to the cochlea, three fluid-filled loops, the semicircular canals, sense how your head turns and keep your balance.
Servier Medical Art (smart.servier.com), CC BY 4.0
Molecules in the air dissolve in the mucus high up in the nose, where they fit receptors on the hair-like tips of smell cells, and those cells send signals straight to the olfactory bulb on the underside of the brain. It is the one sense that skips the thalamus, the brain’s relay station, going on to the limbic system, which works in emotion and memory. People have about 390 kinds of smell receptor, against about 1,100 in a mouse. Taste, on the tongue, gives only sweet, salty, sour, bitter and the savory taste called umami, and perhaps a taste for fat. Most of what you call flavor is smell, which is why food tastes of little when you have a cold.
Servier Medical Art (smart.servier.com), CC BY 4.0
People often say humans have a poor sense of smell. The idea comes less from tests than from the French surgeon Paul Broca, who in the 19th century saw that the human olfactory bulb is small beside the rest of the brain, and tied that to free will. In fact the bulb is large in absolute size, with about as many nerve cells as other mammals’, and people turn out to be more sensitive than dogs or mice to some smells, and can follow a scent trail across a field.
The skin is several senses in one. Its nerve endings respond to light pressure, to vibration, to stretching, to warmth and cold, and to harm. The signals go up the spinal cord to a strip of the parietal lobe (the Going Further topic on touch follows one), and on a fingertip, packed with receptors, two points 2 or 3 mm apart feel like two, while on the forearm points 3 or 4 cm apart can feel like one.
Pain is where sensation and perception come apart most plainly. Nerve endings that detect damage send signals, but pain itself, in the definition the international association of pain researchers uses, is “an unpleasant sensory and emotional experience”, shaped by the body, the mind and the situation. The rare people born unable to feel pain show what it’s for: they feel touch and temperature, yet often injure themselves badly. Two researchers proposed that pain signals pass a kind of gate in the spinal cord, which signals from touch, as when you rub a bumped elbow, and messages coming down from the brain can partly close. Their wiring turned out to be wrong in its details, but the main idea, that pain is adjusted on its way up and doesn’t simply measure damage, changed how pain is understood. People who lose an arm or a leg often go on feeling it, a phantom limb, and in a review of many studies about two in three felt pain in it.
Making a world
Perception is what turns them into things. Only some of what reaches you becomes experience at all (Chapter 7 is about that), and what does is put in order, patches into objects and sounds into words. It draws on what you know and expect, so two people can make different things of the same signal, and it adds what the senses leave out, as it fills in the blind spot. And it holds things steady, so they keep their size and color as conditions change. The cognitive approach, which studies the mind as a system that takes in information, works on it and acts, treats perception as the first step.
Two kinds of work go into it at once. Some runs up from the signals, edges, colors and movement combined; some runs down from what you know. You read a doctor’s handwriting partly by knowing what medicine is likely, and you hear your name in a noisy bus because you’re ready for it. Psychologists call these bottom-up and top-down processing.
A century ago psychologists of the Gestalt school, in Germany, began writing down the rules the brain seems to use to group what it sees. We see a figure standing out from a background; things close together seem to belong together, and so do things alike in color or shape; a line seems to carry on smoothly behind something in front of it; and a shape with gaps is seen as whole. The famous picture of a vase that is also two faces flips between two figures because it gives the brain two equally good choices.
This is where S.K. comes back. These grouping rules feel automatic, perhaps built in. But when the team tested S.K. and two boys treated for cataracts at 7 and 13, the rules that should have joined a shape’s parts, smooth lines carrying on and the way edges meet, did almost nothing for them. What helped was movement. When the overlapping squares slid about, each moving as one piece, all three patients counted them correctly most of the time, and the objects they could name in photographs tended to be things that are often seen moving. Ten to eighteen months later they could separate many still shapes too. The team suggested that things moving together may be how a young brain first learns what belongs to one object, and that the still-picture rules are learned from it. These were only three people, so this is a hint rather than a settled answer, but it suggests that some of what the Gestalt psychologists described is learned, and can be learned late. These patients came through Project Prakash, a program in India that finds children whose blindness can be treated, has carried out more than five hundred surgeries, and studies how each learns to see.
Depth, size and color
The image on the retina is flat, yet you see depth. The brain judges distance from clues, and a few need two eyes. The two eyes see from slightly different places, and the nearer something is, the more their views differ: hold up a finger and close one eye, then the other, and it jumps. Most clues need only one eye, which is why a painting can show depth: nearer things hide farther ones, things of known size look smaller far away, parallel lines like a road’s edges seem to meet, a field’s texture gets finer as it recedes, and things nearer the horizon seem farther. And distant things look hazier and paler, which is why the Himalaya can seem startlingly close from Kathmandu on a clear winter morning, with the haze gone.
Babies seem to read some of these clues early. The American psychologist Eleanor Gibson and a colleague built a “visual cliff”: a sheet of thick glass over a checkered floor, with the floor dropping away under one half. They tested 36 babies old enough to crawl, each set on a board in the middle, with its mother there to coax it across. Of the 27 who left the board, only three crawled out over the drop. Since the babies were already months old, this can’t show whether depth perception is inborn or learned in those months.
As you walk away from a friend, the image of them on your retina shrinks to half, then a quarter, yet they don’t seem to shrink. A door swinging open throws a narrowing shape on the retina, yet you see a rectangle. A white page looks white in shade and in sunlight. These are the constancies, of size, shape and color: the brain allows for distance, angle and lighting and shows you what it judges the thing itself to be like. The checker shadow catches it at work: two squares of the same gray look quite different because the brain corrects one for a shadow.
Some years ago a photograph of a striped dress spread online, and people fought over whether it was white and gold or blue and black. (It was blue and black.) In one survey a few people even said blue and brown. The photograph gave few clues to the lighting, so each brain had to assume some: the researchers’ best explanation was that people who assumed bluish daylight discounted the blue and saw white and gold, while those who assumed warm indoor light saw blue and black. As they put it, a color percept is the visual system’s best guess.
Why illusions happen
How much of perception is guessing? There are two broad answers. The British psychologist Richard Gregory argued that perceptions are hypotheses, educated guesses tested against the signals as a scientist tests a theory, and that an illusion is a good guess meeting an odd picture. The American psychologist James Gibson, Eleanor Gibson’s husband, argued that the light reaching a moving observer is far richer than a single flat image, full of patterns that change as you move, and that much of the world can be picked up directly, without guessing. The two are usually taught as rivals.
What you expect, want and are used to can tilt the guess, which psychologists call a perceptual set: someone waiting for an important call hears the phone ring in the shower.
An illusion, then, isn’t a fault. It’s a rule that usually works, caught in a situation built to fool it. That’s why an illusion persists even when you know the answer: knowing that two lines are equal doesn’t stop them looking different. Optical Illusions has over a hundred to try.
The best known is the Müller-Lyer illusion: two equal lines, one with arrowheads at its ends and one with the heads turned outward like fins, the first looking shorter. Gregory explained it as a good guess misapplied: the arrowheads look like the near corner of a building seen from outside, the fins like the far corner of a room seen from inside, so the brain allows for a distance that isn’t there. If that habit is learned from straight walls, it should differ between places. Some sixty years ago a team led by the American psychologist Marshall Segall tested people in 15 societies and found large differences, and the idea grew that people raised in a “carpentered world” of straight walls and square corners are fooled more.
It became the textbook example of culture shaping sight, and it is now doubted. A recent review argued that the idea is almost certainly false: other animals see the illusion, it doesn’t need straight lines or even any lines, it turns up in touch, it appears even in people born blind, and the old studies’ data were inconsistent. And Project Prakash tested nine newly sighted children, aged 8 to 16, within 48 hours of their first surgery: they saw the Müller-Lyer illusion just as sighted children did, after almost no experience of seeing shapes. Culture may still change perception in other ways. Himba children from remote villages in Namibia were found to be fooled much less than British children by a different illusion, in which a circle looks bigger among small circles, and children from the nearest town were fooled more than those in the villages. But as an account of the Müller-Lyer lines, the straight-walls story no longer stands up well.
So when you look round a room, think of S.K. counting three shapes where you see two. Your eyes and ears catch only a few thin kinds of signal, with holes and blurs in it. Your brain turns them into steady objects at their right distances and colors, using rules it partly brought with it and partly learned, most of them so long ago that you can’t remember not knowing them. Most of the time its guesses are excellent. An illusion is just a moment when you can catch it guessing.
Found a mistake? Please tell us.
