Our senses: Seeing, hearing, and smelling the world

SEEING,HEARING,

SMELLINGAND

THEWORLD

A R E P O R T F R O M T H E H O WA R D H U G H E S M E D I C A L I N S T I T U T E

NEW FINDINGS

HELP SCIENTISTS

MAKE SENSE OF

OUR SENSES

The Howard Hughes Medical Institute

was founded in 1953 by the aviator-

industrialist Howard R. Hughes. Its charter

reads, in part:

“The primary purpose and objective of the

Howard Hughes Medical Institute shall be

the promotion of human knowledge within

the field of the basic sciences (principally

the field of medical research and medical

education) and the effective application

thereof for the benefit of mankind.”

This is the fifth in a series of reports about biomedicalscience. For further information, please contact theHoward Hughes Medical Institute, 4000 Jones Bridge Road, Chevy Chase, Maryland 20815-6789

© 1995 Howard Hughes Medical Institute

SEEING, HEARING, AND SMELLING THE WORLD • 3

F O R E W O R D

I t is a pleasure to introduce the latest of the biomedical researchreports that the Howard Hughes Medical Institute publishes forgeneral readers. Seeing, Hearing, and Smelling the World, like

the four previous publications in the series, takes us to the frontiersof science. It guides us on a journey into the fascinating world of thesenses and the nervous system, where researchers are working tounderstand problems of great potential benefit.

The most routine, everyday occurrences, such as recognizing afriend on the street and exchanging greetings, demonstrate the bio-logical complexity of the puzzles that scientists are attempting tosolve. Although such encounters seem simple, they require hun-dreds of millions of cells to act in precise ways to receive the sightsand sounds and translate them into electrical impulses. Theseimpulses flow through the nervous system to carry the messages tothe brain, where they can be understood and acted upon at aston-ishing speed.

Centuries of effort by thousands of scientists in laboratoriesthroughout the world have been required to bring us to our current,deepening understanding about how we hear, see, and smell.Thanks to the new analytical tools provided by molecular biology,progress toward understanding the senses and the nervous systemhas been rapid during the past decade. Indeed, many neuroscien-tists believe that biomedical science is poised to make substantialprogress toward understanding how the brain works, not only interms of the senses, but also complex functions like learning andmemory. It is an exciting prospect.

This series is published by the Institute as a public service inorder to make the results of current biomedical research availableto readers who are not scientists. It is clear that a basic grasp ofbiology is increasingly essential for citizens who have to make diffi-cult decisions about health care, drug abuse, the environment, andother critical issues.

Teachers are particularly enthusiastic about these reports, andsurveys tell us that they preserve their copies and use them yearafter year. Nearly 4,000 class sets have been requested by highschool, college, and even medical school teachers in the UnitedStates and abroad; altogether, more than 400,000 copies of the publications have been printed.

The Institute’s interest in science education continues to deepenand its commitment to education reform to grow. Its grants pro-gram, which was established in 1987, has now become the largestprivate science education effort in U.S. history. Through its finan-cial support and other activities, the Institute is seeking to makescience come alive for today’s students, which is exactly what wehope Seeing, Hearing, and Smelling the World will do.

Purnell W. Choppin, M.D.PresidentHoward Hughes Medical Institute

4 • SEEING, HEARING, AND SMELLING THE WORLD

A nerve cell that can detect in whatdirection an object is moving branches

out to make contact with many othercells in a rabbit’s visual system. The cell

glows yellow because it was injectedwith fluorescent dye.

Labo

rato

ry o

f Ric

hard

Mas

land

, Mas

sach

uset

ts G

ener

al H

ospi

tal


SEEING, HEARING, AND

SMELLING THE WORLDNew Findings Help Scientists Make Sense of Our Senses

Foreword by Purnell W. Choppin, M.D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

Our Common Senses by Maya Pines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

A Language the Brain Can Understand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11

Breaking the Code of Color by Geoffrey Montgomery . . . . . . . . . . . . . . . . . . . . . . . . . . .12

A Narrow Tunnel of Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

How We See Things That Move by Geoffrey Montgomery . . . . . . . . . . . . . . . . . . . . . . . .22

The Urgent Need to Use Both Eyes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26

Brain Scans That Spy on the Senses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29

The Quivering Bundles That Let Us Hear by Jeff Goldberg . . . . . . . . . . . . . . . . . . . . .32

On the Trail of a “Deafness” Gene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36

Locating a Mouse by Its Sound by Jeff Goldberg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40

Help from a Bat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45

The Mystery of Smell by Maya Pines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46

A Secret Sense in the Human Nose? by Maya Pines . . . . . . . . . . . . . . . . . . . . . . . . . . . .54

The Next Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .57

A R E P O R T F R O M T H E H O WA R D H U G H E S M E D I C A L I N S T I T U T E

A LIFE-SIZEHUMAN BRAINEach of the five senses activates aseparate area of the cerebral cortex, thesheet of neurons that makes up the outerlayer of the brain’s hemispheres. Thisbrain, shown in actual size, is a computerreconstruction based on data from mag-netic resonance imaging (MRI).Approximate locations of the primarysensory areas are shown in color.Most of the activity takes place withinconvolutions that cannot be seenfrom the surface of the brain.

Vision

Hearing

Touch

Smell

Taste

We can recognize a friend instantly—full-

face, in profile, or even by the back of his

head. We can distinguish hundreds of

colors and possibly as many as 10,000 smells. We

can feel a feather as it brushes our skin, hear the

faint rustle of a leaf. It all seems so effortless: we

open our eyes or ears and let the world stream in.

Yet anything we see, hear, feel, smell, or taste

requires billions of nerve cells to flash urgent mes-

sages along linked pathways and feedback loops in

our brains, performing intricate calculations that

scientists have only begun to decipher.

“You can think of sensory systems as little scien-

tists that generate hypotheses about the world,”

says Anthony Movshon, an HHMI investigator at

New York University. Where did that sound come

from? What color is this, really? The brain makes

an educated guess, based on the information at

hand and on some simple assumptions.


times “hear things” that are not reallythere. But suppose a leopard approached,half-hidden in the jungle—then our abilityto make patterns out of incomplete sights,sounds, or smells could save our lives.

Everything we know about the worldcomes to us through our senses. Tradition-ally, we were thought to have just five ofthem—vision, hearing, touch, smell, andtaste. Scientists now recognize that we haveseveral additional kinds of sensations, suchas pain, pressure, temperature, joint posi-tion, muscle sense, and movement, butthese are generally included under “touch.”(The brain areas involved are called the“somatosensory” areas.)

Although we pay little attention to them,

each of these senses is precious and almostirreplaceable—as we discover, to our sor-row, if we lose one. People usually fearblindness above all other disabilities. Yetdeafness can be an even more severe handi-cap, especially in early life, when childrenlearn language. This is why Helen Keller’sachievements were so extraordinary. As a

When you look at the illustration below,for instance, you see an X made of spheressurrounded by cavities. But if you turn thepage upside down, all the cavities becomespheres, and vice versa. In each case, theshapes seem real because “your brainassumes there is a single light source—andthat this light comes from above,” saysVilayanur Ramachandran, a professor ofneuroscience at the University of California,San Diego. As he points out, this is a goodrule of thumb in our sunlit world.

To resolve ambiguities and make senseof the world, the brain also creates shapesfrom incomplete data, Ramachandran says.He likes to show an apparent triangle thatwas developed by the Italian psychologist

Gaetano Kanizsa. If you hide part of thispicture, depriving the brain of certain cluesit uses to form conclusions, the large whitetriangle disappears.

We construct such images unconsciouslyand very rapidly. Our brains are just as fer-tile when we use our other senses. Inmoments of anxiety, for instance, we some-


ILLUSIONS REVEAL SOME OF THE BRAIN’S ASSUMPTIONS

The shaded circles seem to form an X made ofspheres. But if you turn the page upsidedown, the same circles form an X made ofcavities, since the brain assumes that lightcomes from above.

Are these triangles real? They appear to be,because the brain automatically fills in linesthat are missing. But if you block out parts ofthe picture, the triangles vanish.

…the patients

were awake,

since the brain

does not feel

what is hap-

pening to it.


result of an acute illness at the age of 19months, she lost both vision and hearingand sank into a totally dark, silent universe.She was rescued from this terrible isolationby her teacher, Anne Sullivan, who man-aged to explain, by tapping signs into thelittle girl’s palm, that things have names,that letters make up words, and that thesecan be used to express wants or ideas.Helen Keller later grew into a writer (herautobiography, The Story of My Life, waspublished while she was still an undergrad-uate at Radcliffe College) and a well-knownadvocate for the handicapped. Her remark-able development owed a great deal to herdetermination, her teacher, and her family.But it also showed that when a sense (or

two, in Helen Keller’s case) is missing,another sense (in her case, touch) may betrained to make up for the loss, at least inpart.

What we perceive through our senses isquite different from the physical character-istics of the stimuli around us. We cannotsee light in the ultraviolet range, though

bees can, and we cannot detect light in theinfrared range, though rattlesnakes can.Our nervous system reacts only to a select-ed range of wavelengths, vibrations, orother properties. It is limited by our genes,as well as our previous experience and ourcurrent state of attention.

What draws our attention, in manycases, is change. Our senses are finelyattuned to change. Stationary or unchang-ing objects become part of the scenery andare mostly unseen. Customary soundsbecome background noise, mostly unheard.The feel of a sweater against our skin issoon ignored. Our touch receptors, “so alertat first, so hungry for novelty, after a whilesay the electrical equivalent of ‘Oh, thatagain,’ and begin to doze, so we can get onwith life,” writes Diane Ackerman in A Nat-ural History of the Senses.

If something in the environmentchanges, we need to take notice because itmight mean danger—or opportunity. Sup-pose an insect lands on your leg. Instantlythe touch receptors on the affected leg fire amessage that travels through your spinalcolumn and up to your brain. There it cross-es into the opposite hemisphere (the righthemisphere of the brain receives signalsfrom the left side of the body, and viceversa) to alert brain cells at a particularspot on a sensory map of the body.

This map extends vertically along a stripof cerebral cortex near the center of theskull. The cortex—a deeply wrinkled sheetof neurons, or nerve cells, that covers thetwo hemispheres of the brain—governs allour sensations, movements, and thoughts.

The sensory map in humans was origi-nally charted by the Canadian neurosur-geon Wilder Penfield in the 1930s. Beforeoperating on patients who suffered fromepilepsy, Penfield stimulated different partsof their brains with electrodes to locate thecells that set off their attacks. He could dothis while the patients were awake, sincethe brain does not feel what is happening toit. In this way, Penfield soon learned exactlywhere each part of the body that wastouched or moved was represented in thebrain, as he showed in his famous“homunculus” cartoons of the somatosensory

The black line in the back seems much longerthan the one in the front because your brainassumes it is seeing the effects of perspective.Take a ruler to find out for yourself.

brains of animals. In the 1930s and 1940s,scientists applied electrodes to the surfaceof the brain or placed them on the skull ofhumans to study “evoked responses,” thechanging rhythms of electrical signals inthe brain in response to specific stimulisuch as light or sound. Unfortunately, thesesignals from billions of brain cells provedalmost impossible to unscramble.

When extremely thin microelectrodesbecame available in the late 1950s,researchers implanted them into the brainsof living animals to spy on the activity ofindividual cells. Sharp popping soundscould be heard as specific neurons fired, andthe scientists tried to find out what pro-voked these electrical discharges.

This is how David Hubel and TorstenWiesel, who were then at Johns HopkinsUniversity, began the groundbreaking

and motor areas.Surprisingly, these maps do not accu-

rately reflect the size of body parts butrather, their sensitivity. Arms and legs takeup very little space, despite their length.The face and hands, which have greatersensitivity, are given more space—especial-ly the tips of the fingers. Nevertheless, thesignal that a mosquito has landed on theback of your left leg comes through loud andclear. In a fraction of a second, through adecision process that is not yet understood,this signal leads you to swat the insect atjust the right place.

Ever since humans have wonderedabout where their thoughts came from, theyhave tried to understand the senses. Muchwas learned from observing the results ofhead injuries and tumors, as well as by dis-secting postmortem human brains and the


These famous maps by Wilder Penfield show that each part of the body is represented on two strips of thebrain’s cerebral cortex, the somatosensory cortex (left), which receives sensations of touch, and the motorcortex (right), which controls movements. Fingers, mouth, and other sensitive areas take up most spaceon both maps. Penfield called these cross sections the “sensory homunculus” and the “motor homunculus.”

Somatosensorycortex

Motorcortex

experiments on the visual cortex of cats andmonkeys, for which they later won a Nobelprize. They discovered that one neuron in theprimary visual cortex at the back of a cat’sbrain might fire only when the animal’s eyewas exposed to a line at a particular locationand angle, while another next to it would fireonly in response to a line at a slightly differentangle. No one had suspected that these neu-rons would dissect a scene—and respond toparticular elements of it—with such amazingspecificity. Hubel and Wiesel’s success led to ageneral focus on the abilities of single neu-rons, especially in the visual system.

The past decade has seen an explosion ofresearch on all the senses, partly because ofthe new tools supplied by molecular biology.Scientists can now analyze sensory neuronsfar more precisely, down to the level of specificgenes and proteins within these neurons. Thispublication will describe some recent researchon three of our senses—vision, hearing, andsmell—in which there have been particularlyinteresting developments.

The visual system, which involves roughlya quarter of the human cerebral cortex, hasattracted more research than all the othersensory systems combined. It is also the mostaccessible of our senses. The retina, a sheet ofneurons at the back of the eye that any physi-cian can see through an ophthalmoscope, isthe only part of the brain that is visible fromoutside the skull. Research on the visual sys-tem has taught scientists much of what theyknow about the brain, and it remains at theforefront of progress in the neurosciences.

Research on hearing is also gatheringmomentum. One group of scientists recentlydiscovered how receptor neurons in the ear—the so-called “hair cells”—respond to sounds.Another group explored how animals usesounds to compute an object’s location inspace. This may be a model of similar opera-tions in the auditory system of humans.

The olfactory system, which was almost atotal mystery until a few years ago, hasbecome the source of much excitement. Thereceptor proteins that make the first contactwith odorant molecules appear to have beenidentified with the help of molecular genetics,and researchers are beginning to examine howinformation about smells is coded in the brain.

The use of molecular biology has enabledscientists to discover just how receptor neu-rons respond to light, to vibrations in the air,to odorant molecules, or to other stimuli. Thereceptor neurons in each sensory system dealwith different kinds of energy—electromag-netic, mechanical, or chemical. The receptorcells look different from one another, and theyexhibit different receptor proteins. But theyall do the same job: converting a stimulusfrom the environment into an electrochemicalnerve impulse, which is the common languageof the brain (see p. 11). Recently, researchershave uncovered many of the genes and pro-teins involved in this process of sensory trans-duction.

From their understanding of this first stepon the sensory pathway, researchers haveedged up to analyzing how messages about asensory stimulus travel through the brain tothe cerebral cortex and how these messagesare coded.

They know that nearly all sensory signalsgo first to a relay station in the thalamus, acentral structure in the brain. The messagesthen travel to primary sensory areas in thecortex (a different area for each sense), wherethey are modified and sent on to “higher”regions of the brain. Somewhere along theway, the brain figures out what the messagesmean.

Many factors enter into this interpretation,including what signals are coming in fromother parts of the brain, prior learning, overallgoals, and general state of arousal. Going inthe opposite direction, signals from a sensoryarea may help other parts of the brain main-tain arousal, form an image of where the bodyis in space, or regulate movement.

These interactions are so complex thatfocusing on the activity of single neurons—oreven single pathways—is clearly not enough.Researchers are now asking what the centralnervous system does with all the informationit gets from its various pathways.

In more authoritarian times, scientistsbelieved that the brain had a strictly hierar-chical organization. Each relay station wassupposed to send increasingly complex infor-mation to a higher level until it reached thevery top, where everything would somehow beput together. But now “we are witnessing a


Sharp

popping

sounds could

be heard as

specific neu-

rons fired…

senses evolved

“to help

animals

solve vital

problems...”

paradigm shift,” says Terrence Sejnowski,an HHMI investigator who directs the Com-putational Neurobiology Laboratory at theSalk Institute in La Jolla, California.Instead of viewing the cortex as “a rigidmachine,” scientists see it as “a dynamicpattern-processor and categorizer” that rec-ognizes which categories go together with aparticular stimulus, as best it can, everystep of the way. “There is no ‘grandmothercell’ at the top that responds specifically toan image of Grandma,” Sejnowski empha-sizes. “We recognize a face by how its fea-tures are put together in relation to oneanother.”

Sejnowski, a leader in the new field ofcomputational neuroscience, studies neuralnetworks in which the interaction of manyneurons produces surprisingly complexbehavior. He recently designed a computer

model of how such a network might learn to“see” the three-dimensional shape of objectsjust from their shading, without any otherinformation about where the light came from.After being “trained” by being shown manyexamples of shaded shapes, the networkmade its own generalizations and found away to determine the objects’ curvature.

Vision and the other senses evolved “tohelp animals solve vital problems—forexample, knowing where to flee,” saysSejnowski. Large populations of sensoryneurons shift and work together in thebrain to make this possible. They enable usto see the world in a unified way. They linkup with the motor systems that control ouractions. These neurons produce an output“that is more than the sum of its parts,”Sejnowski says. Just how they do it is aquestion for the next century.

Maya Pines, Editor


Rod

Meissner Corpuscle

Free NerveEnding

Rod and cone cells in the eye respond to elec-tromagnetic radiation—light.

The ear’s receptor neurons are topped byhair bundles that move in response to vibra-tions—sound.

Olfactory neurons at the back of the noserespond—and bind—to odorant chemicals .

Taste receptor cells on the tongue and backof the mouth respond—and bind—to chemi-cal substances.

Meissner corpuscles are specialized forrapid response to touch, while free nerve end-ings bring sensations of pain.

SPECIAL RECEPTOR CELLS FOR EACH OF THE SENSES

Cone

VISIONHEARING

SMELL

TOUCH

●

TASTE

A LANGUAGE THE BRAIN CAN UNDERSTAND


Almost at the very instant that light hits a cell inthe retina, or a sound wave nudges the tip of areceptor cell in the ear, the receptor cell convertsthis stimulus into an electrical signal—the lan-guage of the brain.

This conversion, or transduction, is swift andprecise. But it is also surprisingly intricate—sointricate that the process is not yet fully under-stood for most of the senses. In the past decade,however, it has been worked out quite thoroughlyfor vision.

It begins when a photon of light meets one ofthe photoreceptor cells of the retina (either a rodor a cone cell). A photon that strikes a rod cell isimmediately absorbed by one of the 100 millionmolecules of a receptor protein—rhodopsin—thatare embedded in the membranes of a stack ofdisks in the top part, or “outer segment,” of eachcell. These rhodopsin molecules have a snakelikeshape, crisscrossing the membrane seven times,and contain retinal (a form of vitamin A), whichactually absorbs the light. In the dark, the retinalfits snugly into a binding pocket in rhodopsin. But on exposure to light, it straightens out. Thisalters the three-dimensional structure of theentire rhodopsin molecule, activating it and trig-gering a biochemical cascade.

The activated rhodopsin then stimulatestransducin, a protein that belongs to the largefamily of so-called G proteins. This in turn acti-vates an enzyme that breaks down cyclic GMP, a“second messenger,” dramatically lowering itslevel. Cyclic GMP carries signals from the disks,

where light is absorbed, to the cell’s surface mem-brane, which contains a large number of channelsthat control the flow of ions (charged atoms) intothe cell. As ions move into the cell, they alter itselectrical potential.

“In the dark, the channels are constantly openbecause of a high level of cyclic GMP. This allowssodium and calcium ions, which carry positivecharges, to flow into the cell,” explains King-WaiYau, an HHMI investigator at the Johns HopkinsUniversity School of Medicine who played animportant role in deciphering the transductionprocess. “But in the light, the channels close. Thenthe electrical potential inside the cell becomesmore negative. This reduces the amount of neuro-transmitter that is released from the base of thecell to act on other cells”—and thus alerts neuronsin the next layer of retinal cells that a photon oflight has arrived.

This complex cascade of transduction events isrepeated in a remarkably similar way in olfactoryreceptor cells, which respond to odors, says Yau.But the receptor cells that respond to sound use avery different system: their channels open andclose as a direct response to a mechanical force—either tension or relaxation.

Whatever the means, the end result of trans-duction is the same: the cell generates an electri-cal signal that flashes through a dense thicket ofnerve cell connections in the brain, bringing newsfrom the outside world in a Morse-code-like lan-guage the brain can understand.

Rod MembraneRod CellDisc Membrane

Rhodopsin

Enzyme

Light

Transducin (a G protein)

Open Channel Closed ChannelCyclic GMP

Ion Channel

Breaking the C


ode of COLOR

b y G e o f f r e y M o n t g o m e r y

A bright red beach ball comes whirling

toward you. You see its color, shape, and motion all

at once—but your brain deals with each of these

characteristics separately.

“We need parallel processing because neurons

are relatively slow computing machines,” says Jere-

my Nathans, an HHMI investigator at the Johns

Hopkins University School of Medicine. “They take

several milliseconds to go from input to output. Yet

you see things in a fraction of a second—time for no

more than 100 serial steps. So the system has to

have a massively parallel architecture.”

Nathans adjusts a slide projector to show the colors that aredetected by receptor proteins in red and green cone cells.The proteins were made from human DNA in his lab. Thepeaks in the graph indicate the wavelengths (in nano-meters) of light best absorbed by each protein.

Wald himself had wished to do 40years earlier: find the receptorproteins in the retina thatrespond to color.

Rod cells function only in dimlight and are blind to color. “Getup on a dark moonlit night andlook around,” suggests DavidHubel of Harvard MedicalSchool, a winner of the Nobelprize for his research on vision.“Although you can see shapesfairly well, colors are completelyabsent. It is remarkable how fewpeople realize that they do with-out color vision in dim light.”

But the human retina alsocontains another kind of photore-ceptor cell: the cones, which oper-ate in bright light and areresponsible for high-acuity vision,as well as color.

Rods and cones form anuneven mosaic within the retina,with rods generally outnumbering

cones more than 10 to 1—except in the reti-na’s center, or fovea. The cones are highlyconcentrated in the fovea, an area thatNathans calls “the most valuable square mil-limeter of tissue in the body.”

Even though the fovea is essential forfine vision, it is less sensitive to light thanthe surrounding retina. Thus, if we wish todetect a faint star at night, we must gazeslightly to the side of the star in order toproject its image onto the more sensitiverods, as the star casts insufficient light totrigger a cone into action.

In bright light, then, when the cones areactive, how do we perceive colors? Thisquestion has attracted some of the finestminds in science. As early as 1672, byexperimenting with prisms, Isaac Newtonmade the fundamental discovery that ordi-nary “white” light is really a mixture oflights of many different wavelengths, asseen in a rainbow. Objects appear to be aparticular color because they reflect somewavelengths more than others. A red appleis red because it reflects rays from the redend of the visible spectrum and absorbsrays from the blue end. A blueberry, on the

The First Glimmer of ColorNathans became interested inhow we see in color the day heheard of new discoveries abouthow we see in black and white. Itwas 1980, and he was a student atStanford Medical School, herecalls, when Lubert Stryer andDenis Baylor, both of Stanford,described their remarkable find-ings about the workings of rodcells. These cells—one of twokinds of photoreceptor cells in theretina—enable us to see in dimlight, even by the muted starlightof a hazy night.

“Baylor showed that rod cellsachieve the ultimate in light sensi-tivity—that they can respond to asingle photon, or particle of light,”says Nathans. “It was a beautifulexperiment.” (Baylor’s work wasdone in collaboration with TrevorLamb and King-Wai Yau.)

Then Stryer explained howrhodopsin, the light-sensitive receptor pro-tein in the disk membranes of rod cells,announces the arrival of this tiny pulse oflight to the signaling machinery inside thecell. Stryer had found that rhodopsin coulddo this only with the help of an intermedi-ary, called a G protein, which belonged to afamily of proteins that was already knownto biochemists from their study of how cellsrespond to hormones and growth factors.

Nathans immediately realized this meantthat the structure of rhodopsin itself mightbe similar to that of receptors for hormones.His mind began racing with possibilities.“And I ran—literally ran—to the library andstarted reading about vision,” he says.

Until then, Nathans had been studyingthe genetics of fruit flies. But as he read apaper by Harvard University biologistGeorge Wald—a transcript of Wald’s 1967Nobel prize lecture on “The Molecular Basisof Visual Excitation”—Nathans set off on adifferent course. He determined to do what


The intricate layersand connections of nerve cells in the retina were

drawn by the famedSpanish anatomistSantiago Ramón yCajal around 1900. Rod and cone cells

are at the top. Optic nerve fibers

leading to the brainmay be seen at bottom right.

Geoffrey Montgomery, a New York-basedscience writer, is working on a book aboutvision and the brain.

FROM FRONT

OF THE EYE TO

BACK OF

THE BRAIN

ed that this was not an intrinsic property oflight, but arose from the combined activityof three different “particles” in the retina,each sensitive to different wavelengths.

We now know that color vision actuallydepends on the interaction of three types ofcones—one especially sensitive to red light,another to green light, and a third to bluelight. In 1964, George Wald and Paul Brownat Harvard and Edward MacNichol andWilliam Marks at Johns Hopkins showed thateach human cone cell absorbs light in only oneof these three sectors of the spectrum.

Wald went on to propose that the recep-

other hand, reflects the blue end of the spec-trum and absorbs the red.

Thinking about Newton’s discovery in1802, the physician Thomas Young, wholater helped decipher the hieroglyphics ofthe Rosetta Stone, concluded that the retinacould not possibly have a different receptorfor each of these wavelengths, which spanthe entire continuum of colors from violet tored. Instead, he proposed that colors wereperceived by a three-color code. As artistsknew well, any color of the spectrum (exceptwhite) could be matched by judicious mixingof just three colors of paint. Young suggest-


PrimaryVisual Cortex (V1)

The visual pathway:Light rays reflected by an

object—for example, a pencil—enter the eye and pass through its

lens. The lens projects an invertedimage of the pencil onto the retina at the

back of the eye. Signals produced by rod and conecells in the retina then start on their way into the

brain through the optic nerve and reach a major relaystation, the LGN (lateral geniculate nucleus).

Signals about particular elements of the pencil then travel to selectedareas of the primary visual cortex, or V1, which curves around a deep fis-sure at the back of the brain. From there, signals fan out to “higher” areasof cortex that process more global aspects of the pencil such as its shape,color, or motion.

Surprisingly, light rays must penetrate two transparent layers of neu-rons in the retina before reaching the precious rods and cones at the back: amiddle layer of bipolar cells, and a front layer of ganglion cells whose longaxons (fibers that transmit electrical impulses to other neurons) form theoptic nerve leading into the brain.

Lateral GeniculateNucleus (LGN)

Fovea

Light

RetinaOpticNerveOpticNerve

Retina

Axons

GanglionCell

BipolarCell

RodCell Cone

Cell

Light

couldn’t go any further.” Nathans’ ambi-tious plan to isolate the genes that coded forthe three color receptor proteins dependedon Wald’s view that the genes all evolvedfrom the same primordial ancestor. Theonly visual receptor protein that had beenstudied with any intensity at that time wasbovine rhodopsin—from the rod cells ofcows’ eyes. Scientists had purified bovinerhodopsin and deduced the sequence of afragment of the DNA that coded for it.Nathans used this information to constructa lure—a single strand of DNA—with whichhe fished out the complete gene for bovinerhodopsin from a sea of bovine DNA.

Next he used part of this bovine gene asa lure to catch the gene for humanrhodopsin from the jumble of DNA in ahuman cell. This took less than a year“because the genes for human and bovinerhodopsin are virtually identical, despite anevolutionary distance of 200 million yearsbetween cattle and humans,” Nathans says.

Finding the human genes for the colorreceptors proved more challenging, howev-er, since these genes are less closely relatedto the gene for rhodopsin. Nathans began tosift through DNA from his own cells. “I fig-ured I’d be an unlimited source of DNA aslong as I kept eating,” he says. Eventuallyhe fished out some pieces of DNA thatbelonged to three different genes, each ofthem clearly related to the rhodopsin gene.“This coincidence—three genes, three typesof cones—didn’t escape our notice,” he said.Furthermore, two of these genes were onthe X chromosome—”exactly what onewould expect,” says Nathans, “since it haslong been known that defects in red andgreen color vision are X-linked.”

Some 10 million American men—fully 7percent of the male population—either can-not distinguish red from green, or see redand green differently from most people.This is the commonest form of color blind-ness, but it affects only 0.4 percent ofwomen. The fact that color blindness is somuch more prevalent among men impliesthat, like hemophilia, it is carried on the Xchromosome, of which men have only onecopy. (As in hemophilia, women are protect-ed because they have two X chromosomes; a

tor proteins in all these cones were built onthe same plan as rhodopsin. Each proteinuses retinal, a derivative of vitamin A, toabsorb light; and each tunes the retinal toabsorb a different range of wavelengths.Wald believed that the three receptor pro-teins in cones probably evolved from thesame primordial gene—and so didrhodopsin. They were all “variations on acentral theme,” Wald wrote in his Nobel lecture.

This evolutionary message was music toNathans’ ears. It meant that if the geneencoding only one receptor protein could belocated, the genes encoding the other recep-tor proteins could be found by the similarityof the sequence of bases in their DNA.

“I realized while reading Wald’s lecture,”says Nathans, “that Wald had laid out thewhole problem of the genetic basis of colorvision, and that this problem was now solv-able, completely solvable, by moleculargenetic methods.” Wald had taken the prob-lem as far as he could, Nathans pointed out.“But lacking these molecular methods, he


The many rods (small circles) and

fewer cones (dark outlines) inmost of the retina

form a mottled pattern, as shown in

this photomicrograph.The central retina,

or fovea, has only cones.

Labo

rato

ry o

f Ric

hard

Mas

land

, Mas

sach

uset

ts G

ener

al H

ospi

tal

normal gene on one chromosome can oftenmake up for a defective gene on the other.)

Wald and others had found that in color-blind men, the green or red cones workedimproperly or not at all. Wald suggested thatthe genes for the red and green receptorswere altered in these men. He also thoughtthat these genes must lie near each other onthe X chromosome. This tandem arrange-ment—which Nathans confirmed—probablyresults from the duplication of a DNA frag-ment in primates that occurred some 40 mil-lion years ago. The New-World monkeys ofSouth America, which broke from the conti-nent of Africa at about that time, possessonly a single functional copy of a red or greengene, much like color-blind men. But in OldWorld primates—the monkeys and apes ofAfrica and the ancestors of humans—a pri-mordial red-green gene must have duplicat-ed and then diverged slightly in sequence,leading to separate receptors of the red andgreen type. In keeping with this picture,

Nathans found that the DNA sequences ofthe genes for red and green receptors differby only 2 percent—evidence of their commonorigin and recent divergence.

Nathans himself is not color-blind. Beforeusing his own DNA, he thoroughly tested hiscolor vision to ensure that it was normal.Nevertheless, one of his initial findings pre-sented a puzzle: Lying head to tail along hisX chromosome were not just the two genesfor the red and green receptors, but also anextra copy of the green receptor gene.

Here was the explanation for the preva-lence of color blindness, he realized.Because the DNA sequences of the red andgreen receptor genes are so similar, andbecause they lie head to tail, it is easy formistakes to occur during the development ofegg and sperm, as genetic material is repli-cated and exchanged between chromosomes.One X chromosome—like Nathans’—mayreceive an extra green receptor gene, forinstance, or maybe even two. This does no


Rhodopsin, the receptor protein in rodcells, crosses the disk

membrane seventimes; its odd shape is

shared by the threereceptor proteins incone cells. Retinal

(which absorbs light)is shown in purple.The other colored

balls represent aminoacids that make up the rhodopsin

structure.

Membrane Retinal

A NARROW TUNNEL OF LIGHT

Inability to see well in the dark may be an ominoussign in a child. Though it could just signal a need toeat more carrots or take vitamin A supplements, formore than a million people around the world (oneout of every 4,000), it is the first symptom of retini-tis pigmentosa (RP), a genetic disorder that mayleave them totally blind by the age of 40.

“Sometime between their teens and their thir-ties, depending on the family, their retinas begin todegenerate,” says Jeremy Nathans, who has beenstudying the genetic errors that cause the disease.First, the rod cells die at the retina’s periphery.Then these zones of cell death slowly expand, leav-ing only a small patch of functioning retinal cellsnear the center of vision. The patients’ visible worldcontracts to a narrow tunnel of light. Finally, thedying tissue may take everything with it, includingthe precious cones in the central retina, which areresponsible for high-acuity vision.

“The retina doesn’t regenerate,” explains Nathans.“If any part of it goes, you won’t get it back. And so

far, there is no effective therapy for RP.”Until five years ago, no one even knew the cause

of RP. Thinking the disease might be related to adefect in rhodopsin, the receptor protein of rod cells(which are responsible for night vision), Nathansbegan to collect blood samples from patients so hecould study their DNA. In 1989, Peter Humphriesat Trinity College in Dublin, Ireland, found thelocation of a gene defect in a very large Irish familythat had a dominant form of the disease (in whichthe inheritance of a mutated gene from just oneparent causes the disease). Remarkably,Humphries mapped the defect to the same region ofchromosome 3 in which Nathans had located thegene for rhodopsin.

Since then, two teams of scientists—ThaddeusDryja at the Massachusetts Eye and Ear Infirmary,and Nathans and HHMI associate Ching-HwaSung at Johns Hopkins—have shown that aboutone-fourth of patients with the dominant form ofthe disease have mutations in their gene forrhodopsin. Other forms of RP result from mutationsin different genes. Most of the errors in therhodopsin gene cause the protein to be unstable,


A healthy retina (below), as seen through an ophthalmo-scope, has a firm, regular structure. In the retina of aperson with retinitis pigmentosa (right), cells die, starting

at the periphery. Cells laden with the black pigmentmelatonin invade the dead retinal tissue, producingblack deposits that are characteristic of the disease.

Labo

rato

ry o

f Sam

uel J

acob

son,

Uni

vers

ity o

f Mia

mi (

2)


Nathans says. “Either it doesn’t fold cor-rectly to start with, or once it folds, itfalls apart.” There seems to be a correla-tion between the kind of mutation andthe severity of the disease.

To find out how these mutationsdamage the retina and what drugs mightbe designed to prevent this process, sev-eral groups of researchers recentlyinserted defective genes for rhodopsininto mice, where these mutant genescause an RP-like disease. They hope touse this mouse model to develop newtreatments.

Nathans points out that the retinanormally consumes more energy per

gram than any other tissue in the body.A rod cell that needs to dispose ofmutant rhodopsin must expend furtherenergy still, which may “push the cellover the edge, so that it runs out of ener-gy and dies,” taking along the adjoiningcones. Nathans speculates that perhapsa drug that reduced energy consumptionin the rod cells might minimize or delaythe retina’s degeneration—and therebysave the patient’s cones. “RP is a slowdisease,” says Nathans. “It may take 30years to develop, so if we can delay itsprogression by another 30 years, that’svirtually a cure.”

Cones are tightly packed inthe fovea, which is special-ized for high-acuity vision.In a healthy retina (top),cones appear tall andstraight. In the retinas ofpeople with advancedretinitis pigmentosa, coneslose their light-sensitiveouter segment (shown withan *) and then die.

So many cones havedied in the retina seen inthe bottom picture that onlyone layer of cones remains(n indicates the cones’nuclei) and the whole areahas shrunk. The two pic-tures were taken under amicroscope at the samemagnification.

Labo

rato

ry o

f Ann

Mila

m, U

nive

rsity

of W

ashi

ngto

n (2

)

“Mondrian” after its resemblance to theworks of the Dutch painter Piet Mondrian.They used three projectors that beamedlight matching the wavelength-sensitivity ofthe three human cone types. With theseprojectors, the wavelength compositionreflected from any given patch on the Mon-drian could be exactly controlled.

“Land pointed out a patch on the Mon-drian that looked orange in the context ofthe surrounding colors,” Nathans recalls.“Then he gave me a tube, like the tubeinside a paper towel roll, and had me lookat this patch in isolation. And it wasn’torange anymore. It was a perfect red.”

The patch was in fact painted orange,but Land had beamed a high-intensity long-wave light from the red end of the spectrumon it so that it reflected a high proportion ofred light. Under normal viewing conditions,however—when the patch was surroundedby other Mondrian colors—Nathans stillsaw the orange figure by its true color.Somehow, by comparing a patch of colorwith the surrounding colored region, thebrain is able to discount the wavelength ofthe illuminating light and reconstruct thepatch’s color as it would be seen in daylight.

“Color constancy is the most importantproperty of the color system,” declares neu-robiologist Semir Zeki of University College,London. Color would be a poor way of label-ing objects if the perceived colors kept shift-ing under different conditions, he pointsout. But the eye is not a camera. Instead,the eye-brain pathway constitutes a kind ofcomputer—vastly more complex and power-ful than any that human engineers havebuilt—designed to construct a stable visualrepresentation of reality.

The key to color constancy is that we donot determine the color of an object in isola-tion; rather, the object’s color derives from acomparison of the wavelengths reflectedfrom the object and its surroundings. In therosy light of dawn, for instance, a yellowlemon will reflect more long-wave light andtherefore may appear orange; but its sur-rounding leaves also reflect more long-wavelight. The brain compares the two and can-cels out the increases.

Land’s “Retinex” theory of color vision—a

harm. But then the other chromosome withwhich it is exchanging bits of genetic infor-mation is left with only a red receptor gene.The man who inherits this slightly truncat-ed chromosome will be color-blind, bereft ofthe genetic information needed to make agreen receptor.

More than 95 percent of all variations inhuman color vision involve the red andgreen receptors in men’s eyes. It is very rarefor anyone—male or female—to be “blind”to the blue end of the spectrum. Nathansprovided a genetic explanation for this phe-nomenon. He showed that the gene codingfor the blue receptor lies on chromosome 7,which is shared equally by men and women,and that this gene does not have any neigh-bor whose DNA sequence is similar. Bluecolor blindness is caused by a simple muta-tion in this gene.

What Color Is It?Seeing a color involves making compar-isons. “All that a single cone can do is cap-ture light and tell you something about itsintensity,” Nathans points out; “it tells younothing about color.” To see any color, thebrain must compare the input from differ-ent kinds of cone cells—and then makemany other comparisons as well.

The lightning-fast work of judging acolor begins in the retina, which has threelayers of cells. Signals from the red andgreen cones in the first layer, for instance,are compared by specialized red-green“opponent” cells in the second layer. Theseopponent cells compute the balance betweenred and green light coming from a particu-lar part of the visual field. Other opponentcells then compare signals from blue coneswith the combined signals from red andgreen cones.

On a broader scale, comparisons ofneighboring portions of an image lead to ouramazing ability to see colors as constants inan ever-changing world. Nathans vividlyremembers demonstrations of this “colorconstancy” by the late Edwin Land, theinventor of instant photography andfounder of the Polaroid Corporation. Landand his colleagues had made a large collageof multi-colored geometric shapes, called a


But the eyeis not acamera.

mathematical model of this comparison pro-cess—left open the question of where in thepathway between retina and cortex color con-stancy was achieved. This issue could only beaddressed by studying the brain itself.

Working with anesthetized monkeys inthe 1960s, David Hubel and Torsten Wieselof Harvard Medical School had shown thatthe primary visual cortex or area V1, a cred-it-card-sized region at the back of the brain,possesses a highly organized system of neu-rons for analyzing the orientation of anobject’s outlines. But in their early studiesthey found relatively few color-sensitivecells. Then in 1973, Semir Zeki identified aseparate area that he called V4, which wasfull of cells that crackled with activity whenthe monkeys’ visual field was exposed to dif-ferent colors.

A few years later, Edwin Land paid Zekia visit in London. “He showed me hisdemonstration, and I was much taken bythat,” Zeki says. “I was converted, in fact.So I used his Mondrian display to study thesingle cells in area V4.”

In this way, Zeki discovered that some ofthe cells in area V4 consistently respond tothe actual surface color of a Mondrianpatch, regardless of the lighting conditions.He believes these are the cells that performcolor constancy. More recently, with the aidof PET scans, he found an area similar inlocation to the monkeys’ V4 that is specifi-cally activated in humans when they look atMondrian color displays. The color displaysalso stimulate the primary visual area andan area that is adjacent to it, V2.

Much controversy exists about allaspects of the color pathway beyond theretina, however. Researchers disagree aboutthe exact role of cells in human V1 and V2,about the importance of V4, about the simi-larities between monkey and human brains.

To resolve such issues, scientists awaitthe results of further experiments onhumans. The new, noninvasive imagingtechniques that can show the brain at work(see p. 30) may supply key answers. Withina few years, researchers hope, these tech-niques will reveal the precise paths of theneural messages that make it possible forus to see the wealth of colors around us.


A microelectrode recordsthe firing of individual cellsin a monkey’s visual cortex.

●

M O V E24 • SEEING, HEARING, AND SMELLING THE WORLD

HOWWE SEETHINGS

THAT

BY

GEOFFREY

MONTGOMERY

tors and their prey depend upon

being able to detect motion rapidly.

In fact, frogs and some other

simple vertebrates may not even

see an object unless it is moving. If

a dead fly on a string is dangled

motionlessly in front of a starving

frog, the frog cannot sense this

winged meal. The “bug-detecting”

cells in its retina are wired to

respond only to movement. The frog

might starve to death, tongue firm-

ly folded in its mouth, unaware that

salvation lies suspended on a string

in front of its eyes.

street and then upon her, without

ever seeming to occupy the interven-

ing space.

Even people milling through a

room made her feel very uneasy, she

complained to Josef Zihl, a neuropsy-

chologist who saw her at the Max

Planck Institute for Psychiatry in

Munich, Germany, in 1980, because

“the people were suddenly here or

there but I did not see them moving.”

The woman’s rare motion blind-

ness resulted from a stroke that

damaged selected areas of her brain.

What she lost—the ability to see

objects move through space—is a key

aspect of vision. In animals, this abil-

ity is crucial to survival: Both preda-

The patient had great difficulty

pouring coffee into a cup. She could

clearly see the cup’s shape, color,

and position on the table, she told

her doctor. She was able to pour the

coffee from the pot. But the column

of fluid flowing from the spout

appeared frozen, like a waterfall

turned to ice. She could not see its

motion. So the coffee would rise in

the cup and spill over the sides.

More dangerous problems arose

when she went outdoors. She could

not cross a street, for instance,

because the motion of cars was

invisible to her: a car was up the


Unable to see motion,Gisela Leibold feels anxiousas she rides down an esca-lator in Munich.


While the retina of frogs can detectmovement, the retina of humans and otherprimates cannot. “The dumber the animal,the smarter its retina,” observes Denis Bay-lor of Stanford Medical School. The large,versatile brain of humans takes over thejob, analyzing motion through a highly spe-cialized pathway of neural connections.

This is the pathway that was damagedin the motion-blind patient from Munich.Compared with the complex ensemble ofregions in the visual cortex that are devotedto perceiving color and form, this motion-perception pathway seems relativelystreamlined and simple. More than anyother part of the cortex, it has yielded toefforts to unveil “the precise relationshipbetween perception and the activity of asensory neuron somewhere in the brain,”says Anthony Movshon, an HHMI investi-gator at New York University. By studyingthe reactions of humans and monkeys todifferent moving stimuli and probing theparts of the visual cortex that are arousedat such times, researchers have begun tobuild a bridge between the objective worldof electrically signaling neurons that can beobserved in a laboratory and the subjectiveworld of perception accessible only to anindividual’s own consciousness.

One way to visualize the key challengesfor the motion-perception system, suggestsThomas Albright of the Salk Institute, is toconsider what happens when we watch amovie. Each of the 24 frames projected persecond on the theater screen is a still photo-graph; nothing in a movie truly movesexcept the film. The illusion of movement iscreated by the motion-processing system inour brains, which automatically fuses, forinstance, the images of legs that shift posi-tion slightly from frame to frame into theappearance of a walking actor. The Munichpatient is unable to perform this fusion. Inlife or in the movie theater, she sees theworld as a series of stills.

“The motion system must match up imageelements from frame to frame, over space andtime,” says Albright. “It has to detect whichdirection a hand is moving in, for instance,and not confuse that hand with a head whenit waves in front of someone’s face.”

Researchers have now traced the path ofneural connections that make up the motionpathway and tested the responses of cells atdifferent steps along this path. This hasrevealed the basic stages by which the motionsystem senses which way a hand is waving.

Starting in the retina, large ganglioncells called magnocellular neurons, or Mcells, are triggered into action when part ofthe image of a moving hand sweeps acrosstheir receptive field—the small area of thevisual field to which each cell is sensitive.The M cells’ impulses travel along the opticnerve to a relay station in the thalamus,near the middle of the brain, called the lat-eral geniculate nucleus. Then they flash tothe middle layer of neurons in the primaryvisual cortex. There, by pooling together theinputs from many M cells, certain neuronsgain a new property: they become sensitiveto the direction in which the hand is movingacross their window of vision.

Such direction-sensitive cells were firstdiscovered in the mammalian visual cortexby David Hubel and Torsten Wiesel, whoprojected moving bars of light across thereceptive fields of cells in the primary visualcortex of anesthetized cats and monkeys.Electrodes very close to these cells pickedup their response to different moving lines,and the pattern of activity could be heard asa crackling “pop-pop-pop” when the signalswere amplified and fed into a loudspeaker.

“Listening to a strongly direction-selec-tive cell responding,” Hubel has written,“the feeling you get is that the line movingin one direction grabs the cell and pulls italong and that the line moving in the otherdirection fails utterly to engage it, some-thing like the feeling you get with a ratchet,in winding a watch.”

The keystone of the motion pathway isan area of the cortex that lies just beyondthe primary and secondary visual areas (V1and V2)—a largely unexplored wildernessthat used to be known as the “sensory asso-ciation cortex.” “It was thought that some-where in this mishmash of association cor-tex visual forms were recognized and associ-ated with information from other senses,”says John Allman of Caltech. But studies inthe owl monkey by Allman and Jon Kaas

Nothing

in a

movie

truly

moves...

(who is now at Vanderbilt) and in the rhe-sus monkey by Semir Zeki revealed that thearea was not a mishmash at all. Instead,much of it was made up of separate visualmaps, each containing a distinct representa-tion of the visual field. In 1971, Zeki showedthat one of these visual maps was remark-ably specialized. Though its cells did notrespond to color or form, over 90 percent ofthem responded to movement in a particu-lar direction. American scientists usuallycall this map MT (middle temporal area),but Zeki called it V5. He also nicknamed it“the motion area.”

“This very striking finding of this littlehot spot, this little pocket, in which almostall the cells are sensitive for the direction ofmovement,” says Anthony Movshon, wasthe impetus for many vision researchers toturn their attention to motion. Nowhereelse in the visual cortex was there an areathat seemed so functionally specialized.

The cells of this motion area, MT, aredirectly connected to the layer of direction-sensitive cells in the primary visual area,V1. And the two areas have a remarkablysimilar architecture. Hubel and Wiesel haddiscovered that V1 is organized into a seriesof columns. The cells in one column may fireonly when shown lines oriented like an hourhand pointing to one o’clock, for instance,while the cells in the next column fire mostreadily to lines oriented at two o’clock, andso on around the dial. Amazingly, MT hasthe same kind of orientation system as V1,but in addition the cells in its columnsrespond preferentially to the direction ofmovement.

“When you see that an area, like V1 orMT, has this highly organized columnarstructure,” says Wiesel, “you get a sense ofuncovering something fundamental aboutthe way the cells in the visual area work.”

In perceiving motion, as in determiningcolor, the brain constructs a view of theworld from pieces of information that canthemselves be mistaken or ambiguous.Suppose you paint an X on a piece of paperand then move that paper up and down infront of someone’s eyes. Direction-selectivecells in the motion-pathway layer of V1—each of which sees only a small part of the

scene—will respond to the diagonal orien-tation of each of the lines making up the Xbut will not register the movement of the Xas a whole. How, then, is this overall move-ment sensed?

There must be two stages of motionanalysis in the cortex, suggested Movshonand Edward Adelson, who was then a post-doctoral fellow at New York University (heis now a professor at the MassachusettsInstitute of Technology). At the secondstage, certain cells must integrate the sig-nals regarding the orientation of movinglines and produce an overall signal aboutthe motion of the whole object.

When Movshon presented this idea atan annual meeting of vision researchers in1981, William Newsome, then a postdoctor-al fellow at the National Institutes ofHealth (he is now a professor of neurobiolo-gy at Stanford University School ofMedicine), approached him. A lively three-hour dinner ensued and the two menresolved to collaborate. Together with Adel-son, they would search for such cells in themotion area.

The researchers soon found that one-third of MT’s cells could, in fact, signal thedirection in which a hand waves throughspace. Later on, Albright’s research groupshowed that MT cells can detect “transpar-ent” motion, such as a shadow sweepingacross the ground.

Then Allman and his colleagues discov-ered that many MT cells are able to inte-grate motion information from a largeswath of the scene. “Even though an MTcell may respond directly to just one spot inthe visual field,” says Allman, “the cellshave knowledge of what’s going on in theregion surrounding them.” Using a comput-er display with a background texture thatlooks vaguely like a leafy forest, Allmanshowed that some MT cells will fire particu-larly furiously if the leafy backgroundmoves in a direction opposite to a movingobject—the sort of visual pattern a cheetahwould see when chasing an antelope alonga stand of trees. If, however, the back-ground moved in the same direction as themoving object, the cell’s firing was sup-pressed. The cell acted as a large-scale



When you look at yourself in the mirror, “you arelooking into a predator’s eyes,” writes Diana Acker-man in A Natural History of the Senses. Predatorsgenerally have eyes set right on the front of theirheads so they can use precise, binocular vision totrack their prey, she explains, whereas prey haveeyes at the sides of their heads so they can beaware of predators sneaking up on them.

Binocular vision lets us see much moresharply—but only if our two eyes work smoothlytogether, starting early in life. Most of the timethere is no problem. Each year, however, at least30,000 babies in the United States develop strabis-mus, which means that their left and right eyes failto align properly in the first few months after birth.

Until the 1970s, doctors did not realize theurgency of doing something about this condition.Treatment was generally delayed until the childrenwere 4 or older—too late to do much good.

The need for earlier intervention became clear asa result of David Hubel and Torsten Wiesel’s experi-ments with kittens. They showed that there is a crit-ical period, shortly after birth, during which thevisual cortex requires normal signals from both eyesin order to develop properly. In kittens, the criticalperiod lasts for about a month or six weeks. Inhumans, it continues until the age of 5 or 6.

A curious feature of cells in the visual cortex isthat those responding to information from the leftand right eyes form separate “ocular dominancecolumns,” one for each eye. Normally these columnsare arranged in a series of alternating bands that canbe labelled by injecting a marker into one eye. Thisproduces a pattern that resembles the black-and-white stripes of a zebra. But the columns are not fullywired at birth; they take shape during the firstmonths of life, in response to visual experience.

If vision through one eye is blocked during thecritical period, the ocular dominance columnsresponding to the open eye expand in the cortex,while the columns that would normally respond tothe blocked eye progressively shrink. An adult wholoses his vision because of a cataract (a clouding ofthe lens of the eye) will generally see normallyagain if the opaque lens is removed and replacedwith a clear artificial lens. But a child whosecataract is not removed until the age of 7 will beblind in the eye that was blocked by the cataract,even though the cataract is gone and the retina of

THE URGENT NEED TO USE BOTH EYES

LeftVisualField

RightVisualField

Eye

OpticNerve

OpticChiasm

LGN

Primary Visual Cortex

Primary Visual Cortex

The two eyes provide slightly different views of the same scene.Information from the left visual field goes to the right side of theretina in both eyes. At the optic chiasm, half the nerve fibers fromthe left eye cross over to the right hemisphere and the rest stayuncrossed, so that all the information from the left visual fieldends up in the right hemisphere. In this way, a given hemispheregets information from the opposite half of the visual world—buteach hemisphere gets input from both eyes.


detector of motion contrast, performingexactly the sort of operation an animalwould need to sense a figure movingthrough the camouflage of the forest.

While MT cells do not respond to staticforms and colors, Albright has found thatthey will detect a moving object much moreeasily if its form or color strongly contrastswith its background.

“Imagine you’re looking down the con-course in Grand Central Station and you’resupposed to find the woman in the reddress,” says Albright. “There are hundredsof surrounding people moving in differentdirections. Yet there’s no problem at all indetecting the woman in the red dress walk-ing along. Your visual system uses thedress’s color to filter out all the irrelevantnoise around it and homes in on the movingobject of interest.”

Suppose scientists could record from theMT cells in a laboratory monkey that lookedat the woman in the red dress crossingGrand Central Station. They could deter-mine that a particular cell fired when thewoman in the red dress passed through itsreceptive field. But how would they knowthat the firing of this specific MT cell—andnot a network of thousands of other cells inthe brain, of which this cell is only onenode—actually causes the monkey to per-ceive the direction of the woman’s move-ment? How could they ever get inside themonkey’s mind and determine what it per-ceives?

Since Hubel and Wiesel’s pioneeringstudies in the visual cortex, most visual sci-entists have assumed that the perception ofform, color, depth, and motion correspondsto the firing of cells specialized to detectthese visual qualities. In a spectacularseries of experiments conducted since themid-1980s, Newsome and his colleagues atStanford have been directly testing this linkbetween perception and the activity of spe-cific neurons.

They use a device that was developed inMovshon’s laboratory at NYU: a blizzard ofwhite dots moving on a computer monitor.When all the white dots are moving ran-domly, the display looks like a TV tuned toa nonbroadcasting channel. However, the

the eye is able to function normally.The same kind of amblyopia—a loss of vision

without any apparent defect of the eye—occurs inchildren whose eyes are misaligned, as well as inthose whose eyes focus at different distances. Toavoid double vision, such children generally favorone eye and stop using the other. The brain thensuppresses the signals coming from the unfavoredor “lazy” eye. The neurons in the ocular dominancecolumns that should receive signals from this eyebecome wired incorrectly, and the child loses hisability to see with the neglected eye. After a fewyears, neither surgery nor exercises nor a patchover the favored eye can restore the lost vision.

Armed with this information, ophthalmologistsnow treat infants who have visual defects as earlyas possible, with either spectacles or surgery, sincenormal vision can be restored if treatment beginsbefore the age of 3 or 4.

Anthony Movshon and other researchers havestudied monkeys with artificially produced ambly-opia. They found that the cells in these monkeys’retinas and lateral geniculate nuclei (LGN), thevisual system’s relay stations in the center of thebrain, are all normal. But in the case of the neglect-ed eye, “the signals that go from the LGN to the pri-mary visual cortex don’t make it,” Movshon says.The loss occurs because the connections betweencells in the LGN and cells in the corresponding eyedominance columns fail to develop or be main-tained.

Ocular dominance columns form azebra-like pattern of stripes in

part of the primaryvisual cortex.


experimenters can gradually increase thepercentage of dots moving in the samedirection. When 10 percent of the dots movecoherently together, their motion becomesapparent. By 25 percent, it is unmistakable.

Movshon had found that whenever ahuman subject could detect the dots’ motionat all, he or she could also tell the directionin which the dots were moving. “This meansthat the part of the visual pathway carryingthe information used for motion detection isalso carrying a label that says what direc-tion is being detected,” says Movshon. Thisis precisely how one would expect MT, withits columns of direction-selective cells, toencode a moving target.

Next, Newsome began to teach rhesusmonkeys to “tell” him what they saw on thecomputer screen. When they saw dots mov-ing downward, for instance, the monkeyswere supposed to move their eyes to adownward point on the screen. Correctresponses were rewarded with fruit juice.Soon the monkeys could signal with eyemovements that they saw the dots move inany of six directions around the clock. Andafter much training on low-percentage mov-ing dot displays, the monkeys were able toperform nearly as well as Movshon’s humansubjects.

Everything was in place. Newsome,Movshon, and their colleagues were readyto study the relationship between the mon-keys’ perception of motion and the activityof cells in particular columns of MT.

“We found, very much to our surprise,”says Newsome, “that the average MT cellwas as sensitive to the direction of motionas the monkey was.” As more dots movedtogether and the monkey’s ability to recog-nize their direction increased, so did the fir-ing of the MT neuron surveying the dots.

If the monkeys were actually “listen-ing” to the cells in a single MT column asthey made their decision about the direc-tion of movement of the dots on the screen,could the decision be altered by stimulat-ing a different MT column, theresearchers wondered. So they electricallystimulated an MT “up” column while themonkeys looked at a downward-movingdisplay. This radically changed the mon-

keys’ reports of what they saw.“The tenth experiment was an unforget-

table experience,” remembers Newsome.“We got the first of what became known inthe lab as ‘whoppers’—when the effects ofmicrostimulation were just massive. Fiftypercent of the dots would move down, andyet if we’d stimulate an ‘up’ column, themonkey would signal ‘up’ with its eyes. Thatwas really a remarkable day.”

The monkeys’ perceptual responses nolonger seemed to be driven by the directionof dots on the screen. Instead, the animals’perceptual responses were being controlledby an electric stimulus applied to specificcells in the brain by an experimenter.“Intellectually,” says Newsome, “it’s likeputting a novel gene into a bacterium andseeing a novel protein come out. We’reputting a signal into this motion circuitry,and we’re seeing a predictable behaviorcome out that corresponds to the signal weput in.”

These experiments, says Movshon,“close a loop between what the cells aredoing and what the monkey’s doing.” All-man calls the finding “the most direct linkthat’s yet been established between visualperception and the behavior of neurons inthe visual cortex.”

It is still possible, however, that when thedots are moving down and the experimentersstimulate an MT “up” column, the stimula-tion changes what the monkey decides with-out actually changing what it sees.

“This is a key question,” says Newsome.“We now know a lot about the first and laststages of this process. But we are almosttotally ignorant about the decision processout there in the middle—the mechanismthat links sensory input to the appropriatemotor output. How does the decision getmade?”

It is a burning question not only forresearch on the visual system, but for all ofcognitive neuroscience, Newsome believes.The answer would provide a bridge from thestudy of the senses, where so much progresshas been made, to the much more difficultstudy of human thought. At long last, New-some says, “we’re now poised to approachthis question.” ●

SEEING, HEARING AND SMELLING THE WORLD • 31

BRAIN SCANS THAT

For centuries, scien-tists dreamed of beingable to peer into ahuman brain as it per-forms various activi-ties—for example,while a person is see-ing, hearing, smelling, tasting, or touching some-thing. Now several imaging techniques such asPET (positron emission tomography) and thenewer fMRI (functional magnetic resonanceimaging) make it possible to observe humanbrains at work.

The PET scan on the left shows two areas ofthe brain (red and yellow) that become particu-larly active when volunteers read words on avideo screen: the primary visual cortex and anadditional part of the visual system, both in theleft hemisphere.

Other brain regions become especially active

when subjects hearwords through ear-phones, as seen inthe PET scan on theright.

To create theseimages, researchers

gave volunteers injections of radioactive waterand then placed them, head first, into a dough-nut-shaped PET scanner. Since brain activityinvolves an increase in blood flow, more blood—and radioactive water—streamed into the areasof the volunteers’ brains that were most activewhile they saw or heard words. The radiationcounts on the PET scanner went up accordingly.This enabled the scientists to build electronicimages of brain activity along any desired “slice”of the subjects’ brains. The images below wereproduced by averaging the results of tests onnine different volunteers.

ON THE SENSES

Laboratory of Marcus R

aichle, Washington U

niversitySEEINGWORDS

HEARINGWORDS

Much excitement surrounds a newer tech-nique, fMRI, that needs no radioactive materi-als and produces images at a higher resolutionthan PET. In this system, a giant magnet sur-rounds the subject’s head. Changes in thedirection of the magnetic field induce hydrogenatoms in the brain to emit radio signals. Thesesignals increase when the level of blood oxygengoes up, indicating which parts of the brain aremost active.

Since the method is non-invasive,researchers can do hundreds of scans on thesame person and obtain very detailed informa-tion about a particular brain’s activity, as wellas its structure. They no longer need to averagethe resuts from tests on different subjects,whose brains are as individual as fingerprints.

Here a normal volunteer prepares for a fMRI study of face recognition. She will have to match one of the two faces at the bottom of the display with theface at the top. James Haxby, chief of the section onfunctional brain imaging at the National Institute of Mental Health in Bethesda, Maryland, adjusts themirror that will allow her to see the display frominside the magnet.

The volunteer’s brain is particularly active in an areaof her right hemisphere called the fusiform gyrus(arrow) as she matches one of the two faces at the bot-tom of the display with the face at the top. This “slice”of her brain is seen as though looking through her face.

32 • SEEING, HEARING AND SMELLING THE WORLD

A GIANT MAGNETREVEALS THEBRAIN’S ACTIVITY

Sec

tion

on F

unct

iona

l Bra

in Im

agin

g, N

IMH

David Corey and James Hudspeth, who playedmajor roles in discovering how the ear’s hair cellsrespond to sound, discuss the cells’ conversion ofvibrations into nerve signals. In the background, aslide shows hair cells in the inner ear.

Q i e i gU V R N


An unusual dance recital was videotaped in David

Corey’s lab at Massachusetts General Hospital recently.

The star of the performance, magnified many times

under a high-powered microscope, was a sound-receptor

cell from the ear of a bullfrog, called a hair cell because of

the distinctive tuft of fine bristles sprouting from its top.

The music ranged from the opening bars of Beethoven’s

Fifth Symphony and Richard Strauss’s “Thus Spake

Zarathustra” to David Byrne and the Beatles.

As the music rose and fell, an electronic amplifier

translated it into vibrations of a tiny glass probe that

stimulated the hair cell, mimicking its normal stimula-

tion in the ear. The bristly bundle of “stereocilia” at the

top of the cell quivered to the high-pitched tones of vio-

lins, swayed to the rumblings of kettle drums, and bowed

and recoiled, like tiny trees in a hurricane, to the blasts

of rock-and-roll.

BY JEFF GOLDBERG

Q i e i gU V R NTHE

BUNDLESTHAT LET US

HEAR

When this bundle of50 to 60 cilia at the

top of a hair cellvibrates in response

to sound, the haircell (from a bullfrog’s

inner ear) producesan electrical signal.

Tiny tip links can be seen joiningthe tops of shorter

cilia to the sides oftaller ones (arrow).

And the cells are beautiful. I never get tiredof looking at them,” says Corey.

Corey and Hudspeth have explored themicroscopic inner workings of hair cells infiner and finer detail over the past 20 years,gaining a solid understanding of how the cellswork. Some pieces of the puzzle have falleninto place recently with the discovery of aunique mechanism that endows hair cellswith their two most distinctive properties—

The dance of the hair cell’s cilia plays avital role in hearing, Corey explains. Now anHHMI investigator at MGH and HarvardMedical School, Corey was a graduate studentat the California Institute of Technology whenhe began working with James Hudspeth, aleading authority on hair cells. Together, thetwo researchers have helped discover howmovements of the cilia, which quiver with themechanical vibrations of sound waves, causethe cell to produce a series of brief electricalsignals that are conveyed to the brain as aburst of acoustic information.

In humans and other mammals, haircell bundles are arranged in four long, par-allel columns on a gauzy strip of tissuecalled the basilar membrane. This mem-brane, just over an inch long, coils withinthe cochlea, a bony, snail-shaped structureabout the size of a pea that is located deepinside the inner ear.

Sound waves generated by mechanicalforces, such as a bow being drawn across astring, water splashing on a hard surface, orair being expelled across the larynx, causethe eardrum—and, in turn, the three tinybones of the middle ear—to vibrate. Thelast of these three bones (the stapes, or“stirrup”) jiggles a flexible layer of tissue atthe base of the cochlea. This pressure sendswaves rippling along the basilar membrane,stimulating some of its hair cells. Thesecells then send out a rapid-fire code of elec-trical signals about the frequency, intensity,and duration of a sound. The messagestravel through auditory nerve fibers thatrun from the base of the hair cells to thecenter of the cochlea, and from there to thebrain. After several relays within the brain,the messages finally reach the auditoryareas of the cerebral cortex, which processesand interprets these signals as a musicalphrase, a dripping faucet, a human voice, orany of the myriad sounds in the worldaround us at any particular moment.

“The mechanics of the hair cell are fasci-nating—the fact that simply pushing a littlebundle of cilia magically allows us to hear.

Jeff Goldberg, the author of Anatomy of aScientific Discovery, is writing a book aboutfetal tissue transplantation.


Labo

rato

ry o

f Dav

id C

orey

, Mas

sach

uset

ts G

ener

al H

ospi

tal


Because of the hair cell bundles’ uncan-ny resemblance to little antennae and theirlocation in the inner ear, the cells had longbeen suspected of playing an important rolein hearing. This view was bolstered by clini-cal evidence that the majority of hearingimpairments—which affect some 30 millionAmericans—involve damage to hair cells.

There are only 16,000 hair cells in ahuman cochlea, compared to 100 millionphotoreceptors in the retina of the eye, andthey are extremely vulnerable. Life in ahigh-decibel society of pounding jackham-mers, screeching subway cars, and heavymetal rock music can take a devastating tollon them. But whatever the cause—overex-posure to loud noises, disease, heredity, oraging—people tend to lose 40 percent oftheir hair cells by the age of 65. And oncedestroyed, these cells do not regenerate.

Hudspeth’s investigation of these cellswas initially a solitary, frustrating effort. “Iwas struck by the fact that so little wasknown about them,” recalls Hudspeth, whois now an HHMI investigator at the Univer-sity of Texas Southwestern Medical Center.“So I decided to apply myself to solving thisone basic problem.” He wanted to seewhether movements of the ciliabundle ontop of the cell could convert mechanicalvibrations into electrical signals to thebrain, a process known as transduction.

Together with Corey, who joined him in1975, Hudspeth began a series of experi-ments that focused on transduction in haircells. Such experiments, now routine intheir labs, are tricky. Protected deep insidethe skull, hair cells cannot easily be studiedin living creatures—and once removed fromlaboratory animals, these cells quickly die.Even now, Corey acknowledges, “a goodexperiment would be to study three or fourcells for maybe 15 minutes each.”

The measurements are so delicate thatthey are usually carried out on a tablemounted on air-cushioned legs, to reduceany external movements or vibrations; oth-erwise, the building’s own vibration woulddeafen a hair cell in seconds. Hudspethfound that an unused swimming pool builton bedrock in a basement at the Universityof California, San Francisco, where he

extreme sensitivity and extreme speed.This success has attracted a large number

of scientists to the study of the auditory sys-tem. But until the early 1970s, when Hud-speth set out to determine precisely what haircells did and how they did it, research intothe basic biology of the auditory systemlagged so far behind the exciting advancesbeing made in vision that it was dubbed the“Cinderella sense” by some researchers.

worked previously, made the perfect labora-tory for hair cell experiments—especiallyafter he had it filled with 30 truckloads ofconcrete for more stability.

Hair cells from bullfrogs were exposed byremoving the sacculus, a part of the innerear, and pinning the pinhead-sized tissue toa microscope slide. Working under a micro-scope, Hudspeth and Corey were then able tomanipulate an individual hair cell’s bundle ofcilia with a thin glass tube. They slipped thetube over the bundle’s 50 to 60 stereocilia,which are arranged like a tepee on the top ofeach hair cell, and moved the tube back andforth, deflecting the bundle less than a ten-thousandth of an inch. The hair cell’sresponse was detected by a microelectrodeinserted through the cell membrane.

Corey and Hudspeth found that the bun-dle of stereocilia operated like a lightswitch. When the bundle was prodded inone direction—from the shortest cilia to thetallest—it turned the cell on; when the bun-dle moved in the opposite direction, itturned the cell off.

Based on data from thousands of experi-ments in which they wiggled the bundleback and forth, the researchers calculatedthat hair cells are so sensitive that deflect-ing the tip of a bundle by the width of anatom is enough to make the cell respond.This infinitesimal movement, which mightbe caused by a very low, quiet sound at thethreshold of hearing, is equivalent to dis-placing the top of the Eiffel Tower by onlyhalf an inch.

At the same time, the investigators rea-soned that the hair cells’ response had to beamazingly rapid. “In order to be able to pro-cess sounds at the highest frequency rangeof human hearing, hair cells must be able toturn current on and off 20,000 times per sec-ond. They are capable of even more aston-ishing speeds in bats and whales, which candistinguish sounds at frequencies as high as200,000 cycles per second,” says Hudspeth.

Photoreceptors in the eye are muchslower, he points out. “The visual system isso slow that when you look at a movie at 24frames per second, it seems continuous,


Being able to hear speech is taken for granted—“Isit possible for a hearing person to comprehend theenormity of its absence in someone else?” asks Han-nah Merker in her poignant book Listening. “Thesilence around me is invisible....”

Most of the 28 million deaf or hearing-impairedpeople in the United States were born with normalhearing, as was Merker, who became deaf after askiing accident in her twenties. Deafness generallyresults from overexposure to loud noise, disease, orold age. But genetic factors are also an importantcause of hearing loss, especially in children.

It has been estimated that 1 in every 1,000 new-borns is profoundly deaf, while nearly 1 in 20 hassignificant hearing impairment. In more than half ofthese cases, the cause is genetic.

Large families in which a single type of deafnessis clearly inherited are rare, however. GeoffreyDuyk, until recently an HHMI investigator at Har-vard Medical School, often spends hours on the tele-phone with health care workers and geneticists inthe United States and abroad, trying to track downleads on families with similar hearing disorders sohe can search for the genetic error leading to theirdeafness. He recently found an unusually large fam-ily in Worcester, Massachusetts, whose DNA he cananalyze, as well as an entire tribe of Bedouin Arabsin Northern Israel.

Hearing loss has taken an extremely high tollamong the members of both families. Thirteen offifty members of the Worcester family, recentlyexamined by a team of specialists coordinated byDuyk, were going through the same disastroussequence of events: although they could hear well atbirth, they would start losing their hearing in theirteens; by their early forties they were profoundlydeaf. The Bedouin tribe suffered from an even moredamaging kind of hereditary disorder: about a quar-ter of their children were born deaf.

Duyk took samples of blood from both familiesand set out to find mutations in their DNA thatcould account for their hearing loss. The task wasformidable. “Deafness is associated with over 100different genetic disorders, and there are upwards of30 forms of hereditary hearing loss alone, eachcaused by a different mutation,” Duyk says. Yet he

ON THE TRAIL OF A

“DEAFNESS” GENE

is hot on the trail of a genetic error that appears tobe responsible for the Bedouin tribe’s deafness.

Meanwhile, David Corey and his colleague Xan-dra Breakefield at Massachusetts General Hospitalare examining the gene that is defective in Norriedisease, a different disorder that causes not only aprogressive loss of hearing similar to that of theWorcester family but, in addition, blindness at birth.The scientists are now analyzing the protein madeby normal copies of this gene and trying to under-stand its function, which might lead to ways of pre-venting the disorder.

For a deeper understanding of such disorders,

scientists need to work with animal models. Duyk’sresearch group is presently studying two strains ofmice, called “jerkers” and “shakers,” that have beenfound to suffer from inherited forms of progressivehearing loss (as well as the peculiar movement disor-ders that give them their names.) The researchers arelooking for fragments of DNA from these mice thatmight be similar to pieces of DNA from families withgenetic deafness.

“We would like to develop new kinds of treatmentfor hearing loss,” Duyk explains. “But first we need toidentify the proteins and genes that are essential tohearing.”

A two-hour exposureto loud noise—such asthat of loud rockbands—is enough toseriously damage ciliabundles on the haircells of a cat’s innerear. Normal mam-malian hair cell bun-dles have two or threeparallel rows of cilia,one taller than thenext. The tall cilia aremost vulnerable tonoise. After exposureto loud noise, all thetall cilia on the rightof this picture havedisappeared or fusedtogether and fallenover.


Labo

rato

ry o

f Cha

rles

Libe

rman

, Mas

sach

uset

ts E

ye a

nd E

ar In

firm

ary

without any flicker. Contrast 24 frames persecond with 20,000 cycles per second. Theauditory system is a thousand times faster.”

How do hair cells do this?Unlike other types of sensory receptor cells,hair cells do not rely on a cascade of chemicalreactions to generate a signal. Photoreceptorcells in the eye, for instance, require a seriesof intricate interactions with a G protein anda second messenger before their ion channelsclose, sending a signal to the brain. This pro-cess would be much too slow to deal withsounds. Hair cells have to possess a mecha-nism that allows their ion channels to openand close more rapidly than those of anyother sensory receptor cells.

The answer is that hair cells use some-thing very much like a spring to open theirchannels when the cilia bend, without theneed for a time-consuming chemical exchange.

Corey and Hudspeth first theorized thatsuch a “gating spring” mechanism existed inthe early 1980s. They proposed that haircells had a previously unknown type of ionchannel—a channel directly activated bymechanical force. They also developed a bio-

physical theory to account for the hair cells’rapid response. But their theory didn’t tellthem where the channels were or what thespring was.

By painstakingly measuring the electri-cal field around the cilia with an electrode,Hudspeth detected a tiny drop in voltage atthe cilia’s tips, as if the current were beingsucked into a minute whirlpool. This ledhim to conclude that the channels throughwhich charged particles move into the cell,changing its electrical potential, were locat-ed at the cilia’s tips. He then reasoned thatthe gating springs that opened these chan-nels should be there as well.

The springs themselves were firstobserved in 1984, in electron microscopeimages taken by James Pickles and his col-leagues in England. Called tip links, theseminute filaments join each stereocilium toits tallest neighbor. Pickles pointed out thatthe geometry of the cilia bundle would causethe bundle to stretch the links when it wasdeflected in one direction and relax themwhen it was moved in the other. If the tiplinks were the hypothetical gating springs,it would explain everything.


��

A TIP LINK

PULLS UP

THE GATE OF

A CHANNEL.In this sketch, JamesHudspeth suggests howthe movement of a haircell’s cilia bundle (top)opens ion channels atthe tips of the cilia. Whenthe bundle tilts to the right, tip links fromhigher cilia pull up the gates of ion chan-nels on adjoining, shorter cilia.

A close-up shows how a tip link betweentwo cilia opens an ion channel on the shorter cilium.

Even more highly magnified (right), the open channel allows ionsinto the cell. A cluster of myosin molecules in the taller cilium isshown in green and some actin filaments are shown in blue.

doing so with a flamboyant, bouncy, up-and-down motion not found in any other celltype. They outnumber inner hair cells 3 to1. However, the 4,000 inner hair cells areconnected to most of the auditory nervefibers leading to the brain and are clearlythe main transmitters of sound.

The precise function of the outer haircells is still unclear. Auditory researchersspeculate that these cells may serve as anamplification mechanism for tuning up low-frequency sound waves, possibly by acceler-ating the motion of the basilar membrane.

Hudspeth is also intrigued by the possi-bility that outer hair cells may be responsi-ble for something that has puzzledresearchers for years: the fact that our earsnot only receive sounds, but emit them aswell. When sensitive microphones are placedin the ear and a tone is played, a faint echocan be detected resonating back out. Suchotoacoustic emissions are considered nor-mal; in fact, their presence in screeningexams of newborn babies is thought to beindicative of healthy hearing. However, incertain cases, otoacoustic emissions can bespontaneous and so intense that they areaudible without the aid of special equip-ment. “In some people, you can actually hearthem. The loudest ones ever recorded werein a dog in Minnesota, whose owner noticedthe sound coming out of the animal’s earand took the dog to a specialist, who didrecordings and analysis,” says Hudspeth.

“What may be happening is that theamplification system driven by the move-ments of outer hair cells is generating feed-back, like a public address system that’stuned up too high,” he speculates, addingthat such otoacoustic emissions gone awrymay account for certain unusual forms oftinnitus, or ringing in the ear.

Hudspeth and Corey’s research is pro-viding such a detailed picture of the haircell that it is now possible to begin to identi-fy the individual proteins making up the tiplinks, ion channels, and motor mechanismsinvolved, as well as the genes that producethem. Malfunctions in those genes, result-ing in defects in these important structures,may be the cause of inherited forms of deaf-ness (see p. 36)

❑

...our earsnot onlyreceivesounds,but emitthem as

well.❑


“This was a completely new kind ofmechanism, unlike anything ever observedbefore,” says Corey, who provided com-pelling evidence two years ago that the tiplinks pull on the channels. By “cutting” thetip links with a chemical, Corey could stopthe cell’s response cold. “Within less than asecond, as the tip links became unstable,the whole mechanical sensitivity of the cellwas destroyed,” Corey observed.

Recently, both he and Hudspeth havebeen independently investigating anotherproperty of hair cells: their ability to adaptto being deflected. At first, when a hair cellbundle is deflected, the ion channels open.But if the bundle remains deflected for atenth of a second, the channels close sponta-neously. It appears from electron micro-scope images and physiological evidencethat the channels close when the tip linksrelax. This is related to the activity of thetip links’ attachment points, which canmove up and down along the cilia to fine-tune the tension on the channels. When theattachment points move down, the tip linksare relaxed and the ion channels close.

While the researchers are still trying tofigure out what enables the attachmentpoints to move, they strongly suspect thatmyosin plays a role. Myosin is the proteinthat gives muscle cells their ability to con-tract and relax, and Hudspeth’s group hasfound evidence of myosin in cilia bundles.Both labs have now cloned and sequencedthe gene for a myosin molecule in hair cells.A cluster of such molecules in each stere-ocilium could provide the force to move theattachment point up or down.

Slight movements of the attachmentpoints allow the hair cell to set just theright amount of tension on each channel soit is maximally sensitive. They also permitthe cell to avoid being overloaded when it isbarraged by sound.

A second type of hair cell in the highlyspecialized cochlea of mammals may enableus to distinguish the quietest sounds. Theseouter hair cells, which are shaped like tinyhot dogs, look distinctly different from innerhair cells. The outer hair cells also have apeculiar ability to become shorter or longerwithin microseconds when stimulated, ●


L O C A T I N G

A M O U S E

B Y I T S

S O U N D

B Y J E F F G O L D B E R G

In total darkness, a barn owl swoopsdown on a mouse.

While some scientists investigate the mystery

of how we hear from the bottom up, beginning

with the ear’s sound receptors, others search for

answers from the top down, mapping networks

of auditory neurons in the brain in an effort to

understand how the brain processes sounds.

At Caltech in the mid-1970s, Masakazu

(“Mark”) Konishi began studying the auditory

system of barn owls in an effort to resolve a

seemingly simple question: Why do we have two

ears?

While most sounds can be distinguished quite

well with one ear alone, the task of pinpointing

where sounds are coming from in space requires

a complex process called binaural fusion, in

which the brain must compare information

received from each ear, then translate subtle dif-

ferences into a unified perception of a single

sound—say a dog’s bark—coming from a partic-

ular location.

Konishi, a zoologist and expert on the ner-

vous system of birds, chose to study this

process in owls. The ability to identify

where sounds are coming from based on

auditory cues alone is common to all

hearing creatures, but owls—espe-

cially barn owls—excel at the task.

These birds exhibit such extraor-

dinary sound localization abili-

ties that they are able to hunt

in total darkness.

10,000 in all—which would fire only whensounds were presented in a particular loca-tion. Astonishingly, the cells were organizedin a precise topographic array, similar tomaps of cells in the visual cortex of thebrain. Aggregates of space-specific neurons,corresponding to the precise vertical andhorizontal coordinates of the speaker, firedwhen a tone was played at that location.

“Regardless of the level of the sound orthe content of the sound, these cells alwaysresponded to the sources at the same placein space. Each group of cells across the cir-cuit was sensitive to sound coming from adifferent place in space, so when the soundmoved, the pattern of firing shifted acrossthe map of cells,” Knudsen recalls.

The discovery of auditory brain cells thatcould identify the location of sounds inspace quickly produced a new mystery. “The

Working with Eric Knudsen, who is nowconducting his own research on owls atStanford University, Konishi undertook aseries of experiments on owls in 1977 toidentify networks of neurons that could dis-tinguish sounds coming from different loca-tions. He used a technique pioneered byvision researchers, probing the brains ofanesthetized owls with fine electrodes. Withthe electrodes in place, a remote-controlledsound speaker was moved to different loca-tions around the owl’s head along an imagi-nary sphere. As the speaker moved, imitat-ing sounds the owl would hear in the wild,the investigators recorded the firing of neu-rons in the vicinity of the electrodes.

Over the course of several months, Kon-ishi and Knudsen were able to identify anarea in the midbrain of the birds containingcells called space-specific neurons—about


Dan Feldman, anHHMI predoctoral

fellow in Eric Knudsen’s lab, wearsprotective gloves as he prepares an owl for an experiment

that will record theowl’s head-turning

movements inresponse to sounds.

any particular frequency, the more vigor-ously hair cells tuned to that frequencyrespond, while their signaling pattern pro-vides information about the timing andrhythm of a sound.

Konishi hypothesized that such timingand intensity information was vital forsound localization. So he placed micro-phones in the ears of owls to measure pre-cisely what they were hearing as theportable loudspeaker rotated around theirhead. He then recorded the differences intime and intensity as sounds reached eachof the owl’s ears. The differences are veryslight. A sound that originates at theextreme left of the animal will arrive at theleft ear about 200 microseconds (millionthsof a second) before it reaches the right ear.(In humans, whose sound localization abili-ties are keen but not on a par with those ofowls, the difference between a similarsound’s time of arrival in each ear would beabout three times greater.)

As the sound source was moved towardthe center of the owl’s head, these interau-ral time differences diminished, Konishiobserved. Differences in the intensity ofsounds entering the two ears occurred asthe speaker was moved up and down, most-ly because the owl’s ears are asymmetri-cal—the left ear is higher than eye level andpoints downward, while the right ear islower and points upward.

Based on his findings, Konishi deliveredsignals separated by various time intervals

and volume differ-ences through tinyearphones insertedinto the owls’ earcanals. Then heobserved how theanimals responded.Because owls’ eyesare fixed in theirsockets and cannotrotate, the animalsturn quickly in thedirection of a sound,a characteristicmovement. By elec-tronically monitoringthese head-turning


lens of the eye projects visual space ontoreceptors on a 2-dimensional sheet, the reti-na, and the optic nerve fibers project thesame spatial relationships to the brain,”says Konishi. “But in the auditory system,only the frequency of sound waves ismapped on the receptor layer, and the audi-tory nerve fibers project this map of fre-quency to the brain. How can the brain cre-ate a map of auditory space, based only onfrequency cues?”

The answer, Konishi believes, may shedlight on how the brain and the auditory sys-tem process all sounds.

To enable the brain to process efficientlythe rapid stream of impulses emanatingfrom the hair cells in the ear, the auditorysystem must first filter out simple, discreteaspects of complex sounds. Informationabout how high- or low-pitched a sound is,how loud it is, and how often it is heard isthen channeled along separate nerve path-ways to higher-order processing centers inthe brain, where millions of auditory neu-rons can compute the raw data into a recog-nizable sound pattern.

This filtering process begins with thehair cells, which respond to different fre-quencies at different locations along thebasilar membrane. Hair cells at the bottomof the basilar membrane respond morereadily when they detect high-frequencysound waves, while those at the top aremore sensitive to low-frequency sounds.David Corey compares the arrangement tothe strings of a grandpiano, with the highnotes at the base ofthe cochlea, wherethe basilar mem-brane is narrow andstiff, and the bassnotes at the apex,where the membraneis wider and moreflexible.

Hair cells alsoconvey basic infor-mation about theintensity and dura-tion of sounds. Thelouder a sound is at

After wearing prism spectacles for a few months,this owl began to miss auditory targets

because the sound localization system in its braintried to harmonize with the visual system,

which received erroneous cues.

Labo

rato

ry o

f Eric

Knu

dsen

, Sta

nfor

d U

nive

rsity

movements, Konishi and his assistantsshowed that the owls would turn toward aprecise location in space corresponding tothe interaural time and intensity differencesin the signals. This suggested that owls fusethe two sounds that are delivered to theirtwo ears into an image of a single source—inthis case, a phantom source.

“When the sound in one ear preceded thatin the other ear, the head turned in the direc-tion of the leading ear. The longer we delayeddelivering the sound to the second ear, thefurther the head turned,” Konishi recalls.

Next, Konishi tried the same experimenton anesthetized owls to learn how theirbrains carry out binaural fusion. Years ear-lier, he and Knudsen had identified space-specific neurons in the auditory area of theowl’s midbrain that fire only in response tosounds coming from specific areas in space.Now Konishi and his associates found thatthese space-specific neurons react to specificcombinations of signals, corresponding tothe exact direction in which the animal

turned its head when phantom sounds wereplayed. “Each neuron was set to a particularcombination of interaural time and intensi-ty difference,” Konishi recalls.

Konishi then decided to trace the path-ways of neurons that carry successivelymore refined information about the timingand intensity of sounds to the owl’s mid-brain. Such information is first processed inthe cochlear nuclei, two bundles of neuronsprojecting from the inner ear. Working withTerry Takahashi, who is now at the Univer-sity of Oregon, Konishi showed that one ofthe nuclei in this first way station signalsonly the timing of each frequency band,while the other records intensity. The sig-nals are then transmitted to two higher-order processing stations before reaching thespace-specific neurons in the owl’s midbrain.

One more experiment proved conclusive-ly that the timing and intensity of soundsare processed along separate pathways.When the researchers injected a minuteamount of local anesthetic into one of the


Parts of a volunteer’s brain were activated (white box on left of first picture) when he heard a series of sharp but meaningless clicks while inside a fMRI magnet at Massachusetts General Hospital. Some of the same

areas became much more active and several new areas were activated as well (square box on right of second picture)when he listened to instrumental music, reflecting the richer meaning of the sounds.

CAN FUNCTIONAL MRI TELL WHETHER

A PERSON IS HEARING MUSIC OR JUST MEANINGLESS CLICKS?

NM

R C

ente

r, M

assa

chus

etts

Gen

eral

Hos

pita

l

+–

cochlear nuclei (the magnocellular nucleus),the space-specific neurons higher in thebrain stopped responding to differences ininteraural time, though their response todifferences in intensity was unchanged. Theconverse occurred when neurons carryingintensity information were blocked.

“I think we are dealing with basic princi-ples of how an auditory stimulus is pro-cessed and analyzed in the brain. Differentfeatures are processed along parallel,almost independent pathways to higher sta-tions, which create more and more refinedneural codes for the stimulus,” says Kon-ishi. “Our knowledge is not complete, butwe know a great deal. We are very lucky.The problem with taking a top-downapproach is that often you find nothing.”

Konishi has been able to express themechanical principles of the owl’s soundlocalization process as a step-by-stepsequence. He has collaborated with comput-er scientists at Caltech in developing an“owl chip” that harnesses the speed andaccuracy of the owl’s neural networks forpossible use in computers.

At Stanford University, Eric Knudsenhas recently been conducting experimentson owls fitted with prism spectacles todetermine whether distortions in theirvision affect their sound localization abili-ties. Despite their exceptionally acute hear-ing, he has found, the owls trust their visioneven more. When they wear distortingprisms, their hunting skills deteriorate overa period of weeks as their auditory systemstry to adapt to the optical displacement ofthe prisms. “The visual system has ultimatecontrol and basically dictates how the brainwill interpret auditory localization cues,”Knudsen says.

He is also examining a particular net-work of neurons in the animals’ brainswhere he believes auditory and visual sys-tem signals converge. “This network makesit possible for the owls to direct their eyesand attention to a sound once it’s heard,”Knudsen explains. His research is part of anew wave of studies that focus not just onsingle sensory pathways, but on how thebrain combines information it receives frommany different sources.


Perhaps the finest achievement in soundprocessing is the ability to understandspeech. Since this is a uniquely humantrait, it would seem difficult to study inanimals. Yet a researcher at WashingtonUniversity in St. Louis believes it can beexamined—by working with bats.

Bats navigate and locate prey byecholocation, a form of sonar in which theyemit sound signals of their own and thenanalyze the reflected sounds. Nobuo Suga,who has spent nearly 20 years investigat-ing the neural mechanisms used by bats toprocess the reflected signals, is convincedthat such research can shed light on theunderstanding of human speech.

When Suga slowed down recordings ofthe high-frequency, short-duration soundsthat bats hear, he found that the sounds’acoustic components were surprisinglysimilar to those of mammalian communi-cation, including human speech. Therewere some constant frequencies and noisebursts, not unlike vowel and consonantsounds, as well as frequency-modulatedcomponents that were similar to those incombinations of phonemes such as “papa.”

Each of these acoustic elements isprocessed along a distinct pathway tohigher-order neurons, which combine andrefine different aspects of the sonar pat-tern in much the same way that space-specific neurons combine the timing andintensity cues of sound signals.

Suga also identified maps of neuronsin the bats’ auditory cortex which registerslight variations in these components ofsound. Humans may use similar maps toprocess the basic acoustic patterns ofspeech, though speech requires additional,higher-level mechanisms, he points out.

“The ability to recognize variations insound is what enables us to understandeach other. No two people pronouncevowels and consonants in exactly thesame way, but we are able to recognizethe similarities,” says Suga. He believesthat neuronal maps may also play a rolein human voice recognition—the abilityto recognize who is speaking as well aswhat is being said.

HELP FROM A BAT

●

fter taking a mixture of mind-alteringdrugs one night, Stephen D., a 22-year-old medical student, dreamedthat he had become a dog and was

surrounded by extraordinarily rich, meaningful smells.The dream seemed to continue after he woke up—hisworld was suddenly filled with pungent odors.

Walking into the hospital clinic that morning, “Isniffed like a dog. And in that sniff I recognized, beforeseeing them, the twenty patients who were there,” helater told neurologist Oliver Sacks. “Each had his ownsmell-face, far more vivid and evocative than anysight-face.” He also recognized local streets and shopsby their smell. Some smells gave him pleasure and

others disgusted him, but all were so compelling thathe could hardly think about anything else.

The strange symptoms disappeared after a fewweeks. Stephen D. was greatly relieved to be normalagain, but he felt “a tremendous loss, too,” Sacksreported in his book The Man Who Mistook His Wifefor a Hat and Other Clinical Tales. Years later, as asuccessful physician, Stephen D. still remembered“that smell-world—so vivid, so real! It was like a visitto another world, a world of pure perception, rich,alive, self-sufficient, and full...I see now what we give

48 • SEEING, HEARING AND SMELLING THE WORLD

M Y S T E R YThe

A

smell—and often suppress our awareness of what ournose tells us. Many of us have been taught that thereis something shameful about odors.

Yet mothers can recognize their babies by smell,and newborns recognize their mothers in the sameway. The smells that surround us affect our well-being throughout our lives. Smells also retain anuncanny power to move us. A whiff of pipe tobacco, aparticular perfume, or a long-forgotten scent caninstantly conjure up scenes and emotions from thepast. Many writers and artists have marveled at thehaunting quality of such memories.

In Remembrance of Things Past, French novelistMarcel Proust described what happened to him after


up in being civilized and human.”Being civilized and human means, for one thing,

that our lives are not ruled by smells. The socialbehavior of most animals is controlled by smells andother chemical signals. Dogs and mice rely on odorsto locate food, recognize trails and territory, identifykin, find a receptive mate. Social insects such as antssend and receive intricate chemical signals that tellthem precisely where to go and how to behave at alltimes of day. But humans “see” the world largelythrough eyes and ears. We neglect the sense of

of S M E L L BY MAYA PINES

Linda Buck (right)sniffs an odorant usedto study the sense ofsmell. She and RichardAxel (left) discoveredwhat appear to be thelong-sought odorantreceptor proteins.

from the olfactory cells in the nose reach theolfactory area of the cortex after only a sin-gle relay in the olfactory bulb. The olfactorycortex, in turn, connects directly with a keystructure called the hypothalamus, whichcontrols sexual and maternal behavior.When scientists tried to explore the detailsof this system, however, they hit a blankwall. None of the methods that had provedfruitful in the study of vision seemed towork.

To make matters worse, very little wasknown about the substances to which theolfactory system responds. The averagehuman being, it is said, can recognize some10,000 separate odors. We are surroundedby odorant molecules that emanate fromtrees, flowers, earth, animals, food, indus-trial activity, bacterial decomposition, otherhumans. Yet when we want to describethese myriad odors, we often resort to crudeanalogies: something smells like a rose, likesweat, or like ammonia. Our culture placessuch low value on olfaction that we havenever developed a proper vocabulary for it.In A Natural History of the Senses, poet andessayist Diane Ackerman notes that it isalmost impossible to explain how somethingsmells to someone who hasn’t smelled it.There are names for all the pastels in a hue,she writes—but none for the tones and tintsof a smell.

Nor can odors be measured on the kindof linear scale that scientists use to measurethe wavelength of light or the frequency ofsounds. “It would be nice if one smell corre-sponded to a short wavelength and anotherto a long wavelength, such as rose versusskunk, and you could place every smell onthis linear scale,” says Randall Reed, anHHMI investigator at the Johns HopkinsUniversity School of Medicine who has longbeen interested in olfaction. “But there is nosmell scale,” since odorous molecules varywidely in chemical composition and three-dimensional shape.

To find out how these diverse odorantstrigger our perception of smell, researchersneeded to examine the olfactory cells andidentify the receptor proteins that actuallybind with the odorants. This task was mademore difficult by the awkward location of

drinking a spoonful of tea in which he hadsoaked a piece of a madeleine, a type ofcake: “No sooner had the warm liquid mixedwith the crumbs touched my palate than ashudder ran through my whole body, and Istopped, intent upon the extraordinarything that was happening to me,” he wrote.“An exquisite pleasure had invaded mysenses...with no suggestion of its origin....

“Suddenly the memory revealed itself.The taste was of a little piece of madeleinewhich on Sunday mornings...my AuntLeonie used to give me, dipping it first in herown cup of tea....Immediately the old grayhouse on the street, where her room was,rose up like a stage set...and the entire town,with its people and houses, gardens, church,and surroundings, taking shape and solidity,sprang into being from my cup of tea.”

Just seeing the madeleine had notbrought back these memories, Proust noted.He needed to taste and smell it. “Whennothing else subsists from the past,” hewrote, “after the people are dead, after thethings are broken and scattered...the smelland taste of things remain poised a longtime, like souls...bearing resiliently, onimpalpable droplets, the immense edifice ofmemory.”

Proust referred to both taste and smell—and rightly so, because most of the flavor offood comes from its aroma, which wafts upthe nostrils to sensory cells in the nose andalso reaches these cells through a passage-way in the back of the mouth. Our tastebuds provide only four distinct sensations:sweet, salty, sour, and bitter. Other flavorscome from smell, and when the nose isblocked, as by a cold, most foods seem blandor tasteless.

Both smell and taste require us to incor-porate—to breathe in or swallow—chemicalsubstances that attach themselves to recep-tors on our sensory cells. Early in evolution,the two senses had the same precursor, acommon chemical sense that enabled bacteriaand other single-celled organisms to locatefood or be aware of harmful substances.

How we perceive such chemical sub-stances as odors is a mystery that, untilrecently, defeated most attempts to solve it.Anatomical studies showed that signals

“Suddenly

the memory

revealed

itself…”



HOW ODOR AND

TASTE SIGNALS

REACH THE BRAIN

OlfactoryNeurons

OlfactoryTract

Smell SensoryCortex

TasteReceptor

cells

Taste SensoryCortex

OlfactoryBulb

Snyder and his colleagues, includingJonathan Pevsner, Randall Reed, and PaulFeinstein, did find a receptor protein in thisway, but it was not the one they were look-ing for. “It turned out we had discovered aprotein that seems to function as a kind ofamplifying device for odorants,” Snydersays. This protein, an abundant constituentof mucus, is “made in a gland in the noseand sprayed, as by an atomizer, into theinhaled air, where it can bind to the maxi-mum number of odorant molecules,” heexplains. Snyder believes this odorant-bind-ing protein gathers the molecules into largeenough concentrations to activate the senseof smell and then deposits them at the backof the nose, where the olfactory neurons are.If so, it might explain why we are able toperceive a few molecules of odorants in atrillion molecules of air, even though olfac-tory neurons are not activated until the con-centration of odorant is 1,000 times greater.

This did not lead to the long-soughtodorant receptor protein, however. The bigbreak came in 1991. “In the past five years,we have gone from an era in which we knewnothing about the biochemical processinvolved in perceiving odors to knowingnearly all the biochemical steps involved—as well as how they generate electrical sig-nals to the brain,” says Reed. During thisperiod, Reed himself discovered a protein,Golf (the olfactory G protein), that is acti-vated early in the cascade of biochemicalevents leading to electrical signals. Then hecloned one of the genes for the ion channelthat opens in the cell membrane in responseto this cascade, generating an electrical sig-nal. He is now trying to reconstruct theentire signaling pathway in lines of cul-tured cells.

The string of discoveries that totallychanged the study of olfaction resulted froma new emphasis on genetics. Instead ofhunting for the receptor proteins directly,Richard Axel and Linda Buck, who wasthen a postdoctoral fellow in Axel’s groupand is now an HHMI investigator at Har-vard Medical School, looked for genes thatcontained instructions for proteins foundonly in the olfactory epithelium. Theirefforts produced nothing at first. “Now we

the olfactory cells.“We think that we smell with our noses,

[but] this is a little like saying that we hearwith our ear lobes,” writes Gordon Shep-herd, professor of neuroscience at Yale Uni-versity. “In fact, the part of the nose we cansee from the outside serves only to take inand channel the air containing odorousmolecules.” The neurons that sense thesemolecules lie high up in the nose, in a patchof cells called the olfactory epithelium.

Perched behind a sort of hairpin turn atthe very top of the nasal cavity, theolfactory epithelium is only a few centi-meters square. It contains some five millionolfactory neurons, plus their supportingcells and stem cells. Actually, there are twosuch patches—one on each side of thenose—lying in a horizontal line just belowthe level of the eye. Each olfactory neuronin the epithelium is topped by at least tenhairlike cilia that protrude into a thin bathof mucus at the cell surface. Somewhere onthese cilia, scientists were convinced, theremust be receptor proteins that recognizeand bind odorant molecules, thereby stimu-lating the cell to send signals to the brain.

The receptor proteins would be the keyto answering two basic questions aboutolfaction, explains Richard Axel, an HHMIinvestigator at Columbia University. First,how does the system respond to the thou-sands of molecules of different shapes andsizes that we call odorants? “Does it use arestricted number of promiscuous receptors,or a large number of relatively specificreceptors?” And second, how does the brainmake use of these responses to discriminatebetween odors?

In the mid-1980s, Solomon Snyder of theJohns Hopkins University School of Medicine,whose research team had identified manyother receptor proteins, took up the challenge.He decided to use the same approach that hadworked so well for him before. He wouldattach radioactive tags to a set of molecules (inthis case, odorant molecules that smelled ofbell pepper), mix the tagged molecules withcells (in this case, olfactory neurons) in a testtube, and examine where the molecules boundto the cells. The cell’s binding site would bethe odorant receptor.


know why our initial schemes failed,” saysAxel. “It’s because there are a large numberof odorant receptors, and each wasexpressed only at a very low level.”

Finally, Buck came up with what Axelcalls “an extremely clever twist.” She madethree assumptions that drastically nar-rowed the field, allowing her to zero in on agroup of genes that appear to code for theodorant receptor proteins, though the finalproof is not yet in.

Her first assumption—based on bits ofevidence from various labs—was that theodorant receptors look a lot like rhodopsin,the receptor protein in rod cells of the eye.Rhodopsin and at least 40 other receptorproteins criss-cross the cell surface seventimes, which gives them a characteristic,snake-like shape. They also function in sim-ilar ways, by interacting with G proteins(see p.11) to transmit signals to the cell’sinterior. Since many receptors of this typeshare certain DNA sequences, Buckdesigned probes that would recognize thesesequences.

Next, she assumed that the odorantreceptors are members of a large family ofrelated proteins. So she looked for groups ofgenes that had certain similarities. Third,the genes had to be expressed only in a rat’solfactory epithelium.

“Had we employed only one of these cri-teria, we would have had to sort throughthousands more genes,” says Axel. “Thissaved several years of drudgery.”

Buck recalls that “I had tried so manythings and had been working so hard forthree and a half years, with nothing to showfor it. So when I finally found the genes, Icouldn’t believe it! None of them had everbeen seen before. They were all differentbut all related to each other. That was verysatisfying.”

The discovery made it possible to studythe sense of smell with the techniques ofmodern molecular and cell biology and toexplore how the brain discriminates amongodors. It also allowed researchers to “pullout” the genes for similar receptor proteinsin other species by searching throughlibraries of DNA from these species. Odor-ant receptors of humans, mice, catfish, dogs,

and salamanders have been identified inthis way.

The team’s most surprising finding wasthat there are so many olfactory receptors.The 100 different genes the researchersidentified first are just the tip of the iceberg,according to Axel. He thinks there must bea total of “about 1,000 separate receptorproteins” on rat—and probably human—olfactory neurons.

“That’s really a lot of genes,” Axel says.“It’s 1 percent of the genome! This meansthat, at least in the rat, 1 out of every 100genes is likely to be engaged in the detec-tion of odors.” This staggering number ofgenes reflects the crucial importance ofsmell to animals.

Large as the number of receptors may be,however, it is probably smaller than the num-ber of odors we can recognize. “Most likely,the number of odorants far exceeds the num-ber of receptor proteins—by a ratio of at least10 to 1,” Axel says. “In that case, how doesthe brain know what the nose is smelling?”

The visual system needs only three kindsof receptors to distinguish among all the col-ors that we can perceive, he points out. Thesereceptors all respond to the same thing—light. Light of different wavelengths makesthe three kinds of receptors react with differ-ent intensity, and then the brain comparestheir signals to determine color. But theolfactory system must use a different strate-gy in dealing with the wide variety ofmolecules that produce odors.

To figure out this strategy, Axel beganby asking how many kinds of receptor pro-teins are made by a single olfactory neuron.“If a single neuron expresses only one or asmall number of receptors, then the prob-lem of determining which receptors havebeen activated reduces to determiningwhich neurons have been activated,” hesays.

He thought he would make more rapidprogress by working with simpler organ-isms than rats. So he turned to fish, whichrespond to fewer odorants and were likely tohave fewer receptors. From studies withcatfish, whose odorant receptors provedvery similar to those of rats, Axel and hisassociates soon concluded that a given


“When

I finally

found the

genes,

I couldn’t

believe

it!”

tracks for long distances over varied ter-rain, have larger olfactory bulbs thanhumans do—even though humans are morethan twice the total size of these dogs andhave brains that are several times as large.

In the olfactory epithelium of the nose,Axel’s group found, neurons that have agiven odorant receptor do not cluster togeth-er. Instead, these neurons are distributedrandomly within certain broad regions ofthe nasal epithelium. Then their axons con-verge on the same place in the olfactorybulb, Axel believes.

“The brain is essentially saying some-thing like, ‘I’m seeing activity in positions 1,15, and 54 of the olfactory bulb, which cor-respond to odorant receptors 1, 15, and 54,so that must be jasmine,” Axel suggests.Most odors consist of mixtures of odorantmolecules, so other odors would be identi-fied by different combinations.

Buck, who has been trying to solve thesame problem at Harvard, recently foundthat the olfactory epithelium of mice isdivided into regions that she calls expres-sion zones, each of which contains a differ-

olfactory neuron can make only one or, atmost, a few odorant receptors. (Buck andher colleagues have come to the same con-clusion from their work with mice.)

The next step was to find out how theseodorant receptors—and the neurons thatmake them—are distributed in the nose.Also, what parts of the brain do these neu-rons connect with? “We want to learn thenature of the olfactory code,” Axel says.“Will neurons that respond to jasmine relayto a different station in the brain than thoseresponding to basil?” If so, he suggests, thebrain might rely on the position of activatedneurons to define the quality of odors.

Each olfactory neuron in the nose has along fiber, or axon, that pokes through atiny opening in the bone above it, the cribri-form plate, to make a connection, orsynapse, with other neurons in the olfactorybulb, which is a part of the brain. A round,knob-like structure, the olfactory bulb isquite large in animals that have an acutesense of smell. It decreases in relative sizeas this ability wanes. Thus, bloodhounds,which can follow the scent of a person’s

Separate zones of theolfactory epithelium of mice are shown in

red, blue, and yellow.A different set

of odorant receptorgenes is expressed

in each zone


Labo

rato

ry o

f Lin

da B

uck,

Har

vard

Med

ical

Sch

ool

ent set of odorant receptors. These zones aresymmetrical on the two sides of the ani-mals’ nasal cavities (see p.52). “This sug-gests that there may be an initial broadorganization of sensory information thatoccurs in the nose, even before the informa-tion is sent on to the brain,” she declares.

Earlier researchers had traced theanatomical connections between neurons inthe olfactory epithelium and the olfactorybulb, using radioactive labels. When Buckand her associates examined this olderwork recently, “we were amazed,” she says,“because their patterns [of connections]looked just like our zones.” Putting the twosets of findings together, she has produced atentative map of the connections betweenexpression zones in the olfactory epitheliumof mice and certain parts of the olfactorybulb. She believes the initial organization ofinformation about smells is maintained as itreaches the bulb. She has preliminary evi-dence that once the axons get to the bulb,they reassort themselves so that all thosethat express the same receptor converge ata specific site.

And so the first stages of olfaction arebeginning to yield to researchers. But manymysteries remain.

One riddle is how we manage to remem-ber smells despite the fact that each olfactoryneuron in the epithelium only survives forabout 60 days, and is then replaced by a newcell. “The olfactory neurons are the only neu-rons in the body that are continuallyreplaced in adults,” points out Randall Reed.Other neurons die without any successors,and it is thought that we lose increasingnumbers of brain cells as we age. The olfacto-ry neurons are far more exposed and vulner-able than other neurons, since they come intodirect contact with the outside environment.But as they die, a layer of stem cells beneaththem constantly generates new olfactoryneurons to maintain a steady supply.

“Then how can we remember smells?”asks Buck. “How do we maintain perceptualfidelity when these neurons are constantlydying and being replaced, and new synapsesare being formed? You’d have to recreatethe same kind of connections between olfac-tory epithelium and bulb over and over

again, throughout life, or you wouldn’t beable to remember smells in the same way.”

An even deeper mystery is what hap-pens to information about smells after ithas made its way from the olfactory epithe-lium to the olfactory bulb. How is it pro-cessed there, as well as in the olfactory cor-tex? How does it go into long-term memory?How does it reach the higher brain centers,in which information about smells is linkedto behavior?

Some researchers believe that suchquestions can best be answered by studyingthe salamander, in which the nasal cavity isa flattened sac. “You can open it up more orless like a book” to examine how its olfacto-ry neurons respond to odors, says JohnKauer, a neuroscientist at Tufts MedicalSchool and New England Medical Center inBoston, Massachusetts, who has been work-ing on olfaction since the mid-1970s.

Salamanders will make it possible toanalyze the entire olfactory system—fromodorant receptors to cells in the olfactorybulb, to higher levels of the brain, and evento behavior, Kauer thinks. His researchgroup has already trained salamanders tochange their skin potential—the type ofbehavioral response that is measured in liedetector tests—whenever they perceive aparticular odor. To study the entire systemnon-invasively, Kauer uses arrays of pho-todetectors that record from many sites atonce. He applies special dyes that revealvoltage changes in the membranes of cells.Then he turns on a videocamera that pro-vides an image of activity in many parts ofthe system.

“We think this optical recording will giveus a global view of what all the componentsdo when they operate together,” says Kauer.He hopes that “maybe 10 years from now, or20 years from now, we’ll be able to make avery careful description of each step in theprocess.”

This would be amazing progress for asensory system that was relatively neglect-ed until five years ago. Axel and Buck’s dis-coveries have galvanized the study of olfac-tion, and scientists now flock to this field,aroused by the possibility of success, at last,in solving its mysteries.

“how

can we

remember

smells…

when these

neurons are

constantly

dying

and being

replaced?”


●

In addition to our sense of smell, do we havethe ability to sense certain chemical signalsemitted by people around us—without beingaware of it? Many other mammals use a sep-arate set of sensory receptor cells in theirnose to receive social and sexual informationfrom members of their species, and there isgrowing suspicion that we do, too.

A whiff of airborne chemicals from afemale mouse, for instance, may spur a malemouse to mate immediately. Certain chemi-cal messages from other males may makehim aggressive. Other messages may pro-

duce changes in his physiology—as well as inthat of the responding female.

The effects of such messages would befar less obvious in humans. If we do receivechemical signals from people in our vicinity,these signals must compete with manyother factors that influence our behavior.Yet our physiology may be just as respon-sive to chemical messages as that of othermammals. It is known that certain chemicalmessages from other mice lead to the onsetof puberty in young males, while a differentset of signals brings young female mice intoestrus. Similarly, there are some sugges-tions that women may alter their hormonalcycles when exposed to chemical signalsfrom other people.

In the past five years, scientists havebecome extremely interested in these sig-nals, as well as in the “accessory olfactorysystem” that responds to them in many ani-

mals. This system starts with nerve cells in apair of tiny, cigar-shaped sacs called thevomeronasal organs (VNOs), where the sig-nals are first picked up.

“The VNO appears to be a much moreprimitive structure that uses a different setof molecular machinery than the main olfac-

tory system,” says Richard Axel, who recent-ly became intrigued with this system. “Itseems to work in a different way—and wedon’t know how.”

The VNOs are located just behind thenostrils, in the nose’s dividing wall (theytake their name from the vomer bone, wherethe nasal septum meets the hard palate). Inrodents, at least, signals travel from theVNO to the accessory olfactory bulb (ratherthan the main olfactory bulb) and then, asSally Winans of the University of Michiganshowed in 1970, to parts of the brain thatcontrol reproduction and maternal behavior.

“It’s an alternate route to the brain,”explains Rochelle Small, who runs the chem-ical senses program at the National Institute

SE C R EA

on Deafness and Other Communicative Dis-orders in Bethesda, Maryland. If the acces-sory olfactory system functions in humansas it does in rodents, bypassing the cerebralcortex, there is likely to be no consciousawareness of it at all.

This system is particularly important toanimals that are inexperienced sexually.Experiments by Michael Meredith, a neuro-scientist at Florida State University in Tal-lahassee, Charles Wysocki, of the MonellChemical Senses Center in Philadelphia,and others have shown that the VNOs playa key role in triggering sexual behavior innaive hamsters, mice, and rats.

A virgin male hamster or mouse whosevomeronasal organs are removed generallywill not mate with a receptive female, evenif the male’s main olfactory nerves areundamaged. Apparently, the VNOs areneeded to start certain chains of behaviorthat are already programmed in the brain.

Losing the VNOs has a much less dras-tic effect on experienced animals, says

Wysocki, who has been studying the VNOsfor nearly 20 years. When male mice havebegun to associate sexual activity withother cues from females, including smells,they become less dependent on the VNOs.Sexually experienced males whose VNOsare removed mate almost as frequently asintact males.

Do human beings have VNOs? In theearly 1800s, L. Jacobson, a Danish physi-cian, detected likely structures in a patient’snose, but he assumed they were non-senso-ry organs. Others thought that althoughVNOs exist in human embryos, they disap-pear during development or remain “vesti-gial”—imperfectly developed.

Recently, both VNOs and vomeronasal

pits—tiny openings to the VNO in the nasalseptum—have been found in nearly allpatients examined by Bruce Jafek, an oto-laryngologist at the University of Coloradoat Denver, and David Moran, who is now atthe University of Pennsylvania’s Smell andTaste Center in Philadelphia. Last yearThomas and Marilyn Getchell of the Uni-versity of Kentucky College of Medicine inLexington and their colleagues found thatthe cells lining these organs have severalmolecular markers in common with theolfactory neurons that respond to odors.

“This has opened up the possibility of anew sensory system in humans,” saysRochelle Small. “We were often told that theVNO does not exist in adults, so we havetaken a big step just to show that the struc-ture is there.” She cautions that we stilldon’t know whether this organ actually hasconnections to the brain, however. “Thequestion now,” she says, “is what its func-tion might be.”

Just what do the VNOs of rodents—or,perhaps, humans—respond to? Proba-bly pheromones, akind of chemical sig-nal originally stud-ied in insects. Thefirst pheromone everidentified (in 1956)was a powerful sexattractant for silk-worm moths. Ateam of Germanresearchers worked20 years to isolate it.After removing cer-

tain glands at the tip of the abdomen of500,000 female moths, they extracted acurious compound. The minutest amount ofit made male moths beat their wings madlyin a “flutter dance.” This clear sign that themales had sensed the attractant enabledthe scientists to purify the pheromone. Stepby step, they removed extraneous matterand sharply reduced the amount of attrac-tant needed to provoke the flutter dance.

When at last they obtained a chemicallypure pheromone, they named it “bombykol”for the silkworm moth, Bombyx mori, fromwhich it was extracted. It signaled, “come tome!” from great distances. “It has beensoberly calculated that if a single femalemoth were to release all the bombykol in


T S E N S EIN

THE

HUMAN

NOSE?

B Y M AYA P I N E S


and spent a lot of time together graduallydeveloped closer menstrual cycles. Though thewomen’s cycles were randomly scatteredwhen they arrived, after a while their timingbecame more synchronized.

McClintock is now doing a new study ofwomen’s menstrual cycles, based on herfindings from an experiment with rats.When she exposed a group of female rats—let’s call them the “A” rats—to airborne“chemosignals” taken from various phasesof other rats’ estrous cycles, she discoveredthat one set of signals significantly short-ened the A rats’ cycles, while another setlengthened them. Now she wants to knowwhether the same is true for humans—whether there are two opposingpheromones that can either delay oradvance women’s cycles. In this study, sheis focusing on the exact time of ovulationrather than on synchrony.

The most direct scientific route to under-standing pheromones and the VNO may,once again, be through genetics. Manyresearchers, including Axel, Buck, andReed, are now racing to find the genes forthe receptor proteins that actually bind topheromones in the VNOs of rodents. Thesegenes would lead them to the first receptorsfor pheromones ever identified in mam-mals—a prize tool for studying the mecha-nism and function of the VNO.

Once the genes for such receptors areidentified, it should be easy to find outwhether equivalent genes exist in humans.Scientists could then determine, once andfor all, whether such genes are expressed inthe human nose. If they are, the receptorsmay provide a new scientific clue to themystery of attraction between men andwomen—some evidence of real, measurablesexual chemistry.

her sac in a single spray, all at once, shecould theoretically attract a trillion males inthe instant,” wrote Lewis Thomas in TheLives of a Cell.

In dealing with mammals, however, sci-entists faced an entirely different problem.Compared to insects, whose behavior isstereotyped and highly predictable, mam-mals are independent, ornery, complex crea-tures. Their behavior varies greatly, and itsmeaning is not always clear.

What scientists need is “a behavioralassay that is really specific, that leaves nodoubt,” explains Alan Singer of the MonellChemical Senses Center. A few years ago,Singer and Foteos Macrides of the Worces-ter Foundation for Experimental Biology inMassachusetts did find an assay thatworked with hamsters—but the experimentwould be hard to repeat with larger mam-mals. It went as follows: First theresearchers anesthetized a male goldenhamster and placed it in a cage. Then theylet a normal male hamster into the samecage. The normal hamster either ignoredthe anesthetized stranger or bit its ears anddragged it around the cage. Next theresearchers repeated the procedure with ananesthetized male hamster on which theyhad rubbed some vaginal secretions from afemale hamster. This time the normal malehamster’s reaction was quite different:instead of rejecting the anesthetized male,the hamster tried to mate with it.

Eventually Singer isolated the proteinthat triggered this clear-cut response.“Aphrodisin,” as the researchers called it,appears to be a carrier protein for a smallermolecule that is tightly bound to it and maybe the real pheromone. The substanceseems to work through the VNO, since malehamsters do not respond to it when theirVNOs have been removed.

Many other substances have powerfuleffects on lower mammals, but the pheromonesinvolved have not been precisely identified andit is not clear whether they activate the VNO orthe main olfactory system, or both.

Humans are “the hardest of all” mammalsto work with, Singer says. Yet some studiessuggest that humans may also respond tosome chemical signals from other people. In1971, Martha McClintock, a researcher who isnow at the University of Chicago (she wasthen at Harvard University), noted that col-lege women who lived in the same dormitory

…some

evidence

of real,

measurable

sexual

chemistry.

●

The opening of anadult woman’s VNOis seen as a small pit(arrow) in the picture

below, which wastaken with an angledtelescope. The VNOsare narrow sacs, only

a few millimeterslong. They lie oneither side of the

nasal septum, quitefar from the olfactory

epithelium.

Pit

OlfactoryEpithelium

VNOPit


essages from the senses travel so swiftly through the brain that imaging machines

such as PET and fMRI cannot keep up with them. To track these messages in real

time, scientists now use faster methods—electrical recording techniques such as MEG

(magnetoencephalography) or EEG (electroencephalography). These techniques rely

on large arrays of sensors or electrodes that are placed harmlessly on the scalp to

record the firing of brain cells almost instantaneously. Their data are then combined with anatomical information

obtained by structural MRI scans.

One of the first experiments in which structural MRI was used

jointly with MEG produced a three-dimensional map of the areas of the

brain that are activated by touching the five fingers of one hand

(below). A New York University research team headed by Rodolfo

Llinás found this map to be distorted in the brain of a patient who had

two webbed fingers since birth. A few weeks after the man’s fingers

were separated by surgery, however, parts of his brain reorganized and

the map became almost normal.

MTHE

NEXTGENERATION

THENEXT

GENERATIONEach of the color-coded areas in this combined MRI/MEG image ofthe brain responds to the touch of a different finger of the right hand.

Laboratory of Rodolfo Llinas, New York University


Comparing(High load)

Comparing(Low load)

Updating(High load)

The rapidly-shifting patterns of activity inthe six images below reflect what goes on inthe brain of a woman who is looking at aletter on a screen during a test at the EEGSystems Laboratory, a private research cen-ter directed by Alan Gevins in San Francis-co. The woman’s task is to decide whetherthe letter is located in the same place as aletter she has seen before.

In the “low load” test she compares thenew letter’s location to a previous one. Inthe “high load” test she compares the newlocation to three previous ones, and thebrighter colors reflect a higher degree ofbrain activation.

The images are based on data from 124recording electrodes positioned in a softhelmet that covered the woman’s head.The scientists used an MRI-derived model

In this high-techversion of EEG, thepositions of 124recording electrodes(attached to a softhelmet) are careful-ly plotted on anMRI model of thehead.

M A T C H I N G A L O C A T I O N

These computer-generated images recreate the electrical signals that flash across the brain ofa volunteer during the matching test. A strong electrical signal (first image) sweeps acrossthe frontal cortex of her right hemisphere 320 milliseconds after a new letter has appeared onthe screen, as she compares the letter’s location to three locations that she has seen before.The same areas of her brain are activated—but less intensively—in the second image, as shecompares a new letter’s location to only one location that she has seen before.

Only 140 milliseconds later, a dif-ferent set of electrical signals isrecorded from the volunteer’s brainand recreated in these images. Thistime the frontal cortex of her left

EEG Systems Laboratory, San Francisco

EE

G S

yste

ms

Labo

rato

ry, S

an F

ranc

isco

Updating(Low load)

Rehearsing(High load)

Rehearsing(Low load)

of her head to correct for any distortions inelectrical transmission that might be causedby variations in the thickness of her skull.

The resulting images clearly show thatvarious areas of the woman’s brain are acti-vated in turn. However, these images arelimited to the brain’s surface.

The next generation of imaging technologywill use functional MRI in various combina-tions with MEG and EEG, predicts JohnBelliveau, director of cognitive neuroimag-ing at the Massachusetts General Hospitalin Cambridge. Functional MRI shows activi-ty deep in the brain with high spatial reso-lution, but is relatively slow since it is basedon the blood-flow response, which takesabout 450 milliseconds. “If you do a visualstimulation experiment, four to five differ-

ent areas may have turned on within thattime,” Belliveau says. “We know wherethose areas are, but we don’t know whichone turned on first.” By contrast, EEG’sspatial resolution is relatively poor, butbecause of its speed it may reveal thesequence of events. His group has alreadydone some EEG recordings right inside themagnet of an fMRI machine, to get simulta-neous measurements.

Together, such techniques will offer scien-tists a glimpse of how information from thesenses is processed in different parts of thebrain. Building on the studies shown here, thenew hybrids may then begin to tackle neuralnetworks. They may help researchers exam-ine how various parts of the brain exchangeinformation and—most intriguing—how sen-sory information leads to thought.


After the screen goes blank, the volunteer rehearses the new memory. As thenext two images show, this activity produces yet another electrical signal overher right hemisphere. The signal is stronger in the high load than in the lowload condition, but in both cases it is maintained until a new letter appears onthe screen.

hemisphere is activated as sheenters the location of the new letterinto her working memory. The sig-nals are more intense in the highload than in the low load condition.

●

TRUSTEES

Alexander G. Bearn, M.D. Adjunct Professor The Rockefeller University Professor Emeritus of MedicineCornell University Medical CollegeFormer Senior Vice President Merck Sharp & Dohme, International

Helen K. Copley Chairman of the Corporation andChief Executive Officer The Copley Press, Inc.

Frank William Gay Former President and Chief Executive OfficerSUMMA Corporation

James H. Gilliam, Jr., Esq. Executive Vice President Beneficial Corporation

Hanna H. Gray, Ph.D. President Emeritus andProfessor of the Department of History and the CollegeThe University of Chicago

William R. Lummis, Esq. Chairman of the Board of Directors The Howard Hughes Corporation

Irving S. Shapiro, Esq. Chairman Of Counsel Skadden, Arps, Slate, Meagher & Flom Former Chairman and ChiefExecutive Officer E.I. du Pont de Nemours and Company

George W. Thorn, M.D. Chairman Emeritus Professor Emeritus Harvard Medical School

James D. Wolfensohn President James D. Wolfensohn Incorporated

OFFICERS

Purnell W. Choppin, M.D. President

W. Maxwell Cowan, M.D., Ph.D. Vice President and Chief ScientificOfficer

Joan S. Leonard, Esq. Vice President and General Counsel

Joseph G. Perpich, M.D., J.D. Vice President for Grants and Special Programs

Jon C. Strauss Vice President and Chief Financial Officer

C. F. Wolfe Vice President and Chief InvestmentOfficer

Editor ................................................................... Maya PinesCopy Editors ........................................... Jennifer Ackerman

...........................................................William T. Carrigan

Production Editor ............................................. Gail MarkleyPermissions Editor .......................... Kimberly A. Blanchard

Art Direction and Design: RCW Communication Design Inc.

Editorial Director: Robert A. Potter

Illustration and Photography Credits:Cover: Head-A. Gevins and M.E. Smith, EEG Systems Laboratory, San Francisco; retinal cells-Richard H. Masland, Massachusetts General Hospital; nerve cell-Guang Yang and RichardH. Masland, Massachusetts General Hospital; hair cell-John Assad, Gordon Shepherd and David Corey, Massachusetts General Hospital; retina-Photographic Laboratory of BascomPalmer Eye Institute, University of Miami; profile-Michael Freeman; sensory receptor cells-Eade Creative Services, Inc./George Eade illustrator (cone cell adapted from Scientific American Vol. 256, No. 2, page 42, 1987)

Page Source2-3 Guang Yang and Richard H. Masland, Massachusetts General Hospital4-5 Photo of head-Michael Freeman; MRI scan of brain-John Belliveau, NMR Center,

Massachusetts General Hospital 6-7 Mark R. Holmes, ©National Geographic Society8 Brain-John Belliveau, NMR Center, Massachusetts General Hospital

Homunculus-Reprinted with permission of Simon & Schuster, Inc. from The Cerebral Cortex of Man by Wilder Penfield and Theodore Rasmussen ©1950 Macmillan Publishing Co.; renewed ©1978 Theodore Rasmussen

10 Eade Creative Services, Inc./George Eade illustrator (rod and cone cells adapted from Scientific American Vol. 256, No. 2, page 42, 1987)

11 Eade Creative Services, Inc./George Eade illustrator (rod cell adapted from Scientific American Vol. 256, No. 2, page 42, 1987; transduction adapted from figure 28-3, page 404, Principles of Neural Science by Eric R. Kandel, James H. Schwartz and Thomas M. Jessell ©1991 Elsevier Science Publishing Company, Inc.)

12-13 Kay Chernush14 Instituto Cajal. CSIC. Madrid.15 Eade Creative Services, Inc./George Eade illustrator (visual system adapted from a

drawing by Laszlo Kubinyi and Precision Graphics on page 9 of Images of Mind by Michael I. Posner and Marcus E. Raichle, Scientific American Library ©1994; retinalcells adapted from a drawing by Carol Donner/Tom Cardamone Associates on page 37 of Eye, Brain and Vision by David H. Hubel, Scientific American Library, ©1988)

16 Richard H. Masland, Massachusetts General Hospital17 Eade Creative Services, Inc./George Eade illustrator (adapted from figure 28-4,

page 405, Principles of Neural Science by Eric R. Kandel, James H. Schwartz andThomas M. Jessell ©1991 Elsevier Science Publishing Company, Inc.)

18 Photographic Laboratory of Bascom Palmer Eye Institute, University of Miami19 Ann H. Milam, University of Washington, Seattle21 Fritz Goro, Life Magazine ©Time Inc.23 Joe McNally/Sygma26 Eade Creative Services, Inc./George Eade illustrator (adapted from a drawing by

Carol Donner on page 60 of Eye, Brain and Vision, by David H. Hubel, Scientific American Library, ©1988)

27 Simon LeVay, Journal of Neuroscience Vol. 5, No. 2, February 1985. 30 Vincent Clark and James Haxby, Section on Functional Brain Imaging,

Laboratory of Psychology and Psychopathology, NIMH, NIH30-31 Kay Chernush32-33 Kay Chernush34-35 John Assad, Gordon Shepherd and David Corey, Massachusetts General Hospital37 Michael J. Mulroy and M. Charles Liberman, Massachusetts Eye and Ear Infirmary38 Jennifer Jordan/RCW Communication Design Inc. (adapted from a sketch by

James Hudspeth, HHMI, University of Texas Southwestern Medical Center)40-41 Masakazu Konishi, California Institute of Technology42-43 Kay Chernush43 Eric Knudsen, Stanford University44 Randall R. Benson and Thomas Talavage, NMR Center, Massachusetts

General Hospital46-47 Kay Chernush49 Scott T. Barrows, ©National Geographic Society52 Kerry J. Ressler, Susan L. Sullivan, and Linda B. Buck from Cell Vol. 73, page

601, figure 4(5), May 7, 1993 ©Cell Press54 The Lovers by Pablo Picasso, National Gallery of Art/Chester Dale Collection56 Photo-The Journal of NIH Research Vol. 6, January 1994. Reprinted with

permission. Illustration-Eade Creative Services, Inc./George Eade illustrator (adapted from The Journal of NIH Research Vol. 6, January 1994)

57 Urs Ribary, Rodolfo Llinás et al from the Proceedings of the National Academy of Sciences Vol. 90, page 3594, April 1993, ©National Academy of Sciences

58 A. Gevins and M.E. Smith, EEG Systems Laboratory, San Francisco58-59 A. Gevins and M.E. Smith, EEG Systems Laboratory, San Francisco

Printing: S&S Graphics Inc.

02-95-002-110K62 • SEEING, HEARING, AND SMELLING THE WORLD

The views and opinions expressed in this publication are not necessarily those of the Trustees or management of the Howard Hughes Medical Institute.

The Howard Hughes Medical Institute is a nonprofit scientific and philanthropic organization incorporated in Delaware. It is dedicated to the promotion of human knowledge within the basic sciences, principally medical research and education, and the use of such knowledge for the benefit of humanity. The Institute is directly engaged in medical research and is qualified as a medical research organization under the federal tax law. It is an equal opportunity employer.

This publication is a project of the Institute’s Office of Communications, Robert A. Potter, director, (301) 215-8855. Published February 1995. Printed in the USA.

Our senses: Seeing, hearing, and smelling the world

Documents

Transcript of Our senses: Seeing, hearing, and smelling the world