• Author / Uploaded
• Ajay Patel

Citation preview

Brief Contents pa CHAPTER 1

Introduction

CHAPTER 2

The First Steps in Vision: From Light to Neural Signals

2 30

CHAPTER

3

Spatial Vi sion: From Spots to Stripes

52

CHAPTER

4

Perceiving and Recognizing Objects

88

CHAPTER 5

Th e Perception of Color

122

CHAPTER 6

Space Perception and Bi nocular Vision

CHAPTER

7

Attention and Scene Perception

CHAPTER

8

Visual Motion Perception

CHAPTER 9

Hearing: Physiology and Psychoacoustics Hearing in the Environment

CHAPTER 11

Music and Speech Perception

12

CHAPTER 13 CHAPTER

14

CHAPTER 15

200

236

CHAPTER 10

CHAPTER

156

290 320

The Vestibular System and Our Sense of Equi librium 348 Touch

388

Olfaction Taste

426

466

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

260

About the Authors

--.-· ..,-

.. .

---"-

\ ..

JEREMY M- WOLFE is Professor of Ophthalmology and Radiology at Harvard Medical School. Dr. \'Volfe was trained as a vision researcher/ experimental psychologist and remains one today. His ea rly work includes papers on binocular vision , adaptation, and accomnuxla tion . The bulk of his recent work has dea lt with visual sea rch and visual attention in the lab and in rea l-world settings sud\ as airport securi ty and cancer screening. He taught Introductory Psychology fo r over h venty-five years at the tvlass.achu sett.s In stitute of Tedmology, w here he \>\>' O il the Baker Memorial Prize for und ergraduate teachin g in 1989. He directs the Visual Attention Lab a.nd the Center for Ad van ced Med ical fmaging of Brigham and \Vomen's Hospital. has ta ug ht at the University of California, Berkeley since 2001. He is Dean/ Professor in the School of Optometry and Professor at the Helen \ Vills Neuroscience Institute. In the lab, Dr. Levi and colle.Jg ues use p sythophysics, comp utntional m od eling. and brain imaging (fMRI ) to study the neura l mech anisms of normal p a ttern vision in htmtans, and to leam how they rue degraded by abno rma l visual exp erience (amblyopia ). DENNIS M. LEVI

Department Hea d, Professor of Speech, language, and Hea ring Sciences, and Professor of Psych ological Sciences at Purdue University. His research encompnsses: how people h e..i.r complex sounds s ud1 ;;is speech; how experience shapes the w :.iiy we hear; how w ha t \.Ve hear guides our a ctions and conuiHmication j dinical problerns of hearing irnpai rmen t or lan g uage delay; and practical concerns about compute r speed1 recognition a nd hearing aid design.. Dr. Kluend er is deeply conunitted to teaching, a nd h3S taught a wide array of com ses--phiJ osoph.ic.Jl, psych ological, cmd physiological.

KEITH A . KLUENDER is

Bus hnell Professor, Department of Food Science and Huma n Nutrition a t the University of Florida. Her research on taste h as opened up broo d n tJW aven ues for htr th er establis hing the impact of both genetic a nd pa thologic.JJ variation in tas te on food preferences, die t, and health. Sh e d iscovered tha t t..lSte n om1a.lly inhibits other oral sensa tions s uch that da mage to taste lead s to unexpected consequences like weight gain and intensified oral pain . Mos t recently, working \.vi th coJleagues in Hor ticulture, he r group fo und tha t a considerable am0tmt of thesweetr'less in fm.it is actually produced by interactions between taste and o lfaction in the brain. TIUs may lead to a nevv way to reduce su gar in foods and beverages. LINDA M. BARTOSHUK is

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

RACHEL s. HERZ is a n Ad jtmd Assistant Professor in the Depa rtme nt of Psychiatry' and Human Behavior a t Brmvn Universit)'rs Wa rren Alpert Med ical Sch ool and Pa rt-ti1ne Faculty in the Psychology Depa rhnent a t Boston College. Her research focuses on a number of facets of olfactory cogni tion and perception and on e motion, me n1ory1 and motivated behavior. Using an experimental approach grmmd ed in evolutionary theory and incorporating bo th cogn:itive-beha vioral and ne uropsyd1ological techniques, Dr. Herz ahns to lmde rs tand hm·\' b iological mechan isms and cogn itive processes interact to influence percep tion, cognition , a nd beh avior. ROBE:RTA L. KlATZKY is the Charl es J. Q ueena n, Jr. Professor of Psychology at Carnegie Mell on University, w here sh e also holds faculty appoinbnent.s in the Center for the Ne ural Basis o f Cognition a nd the Htunan-Compute r hl terac tion Ins titute. Sh e has d one exte ns ive research on haptic and visual object recogn:ition 1 sp ace perception and sp atial thinking, and mo tor p erformance. Her work has application to haptk interfaces, nav igation aids for the blind, itnage-guid ed surgery, teleoperntion, and virtual environments.

SUSAN J . LEDE:RMAN is Pro fessor Emerira of Psychology a t Q ueen 's Uni versity, w ith cross-ap pointments in the Centre fo r Neu roscience and in the Sch ool of Computing . He r research in terests span both perception and cognition , wi th particular em.phases on p syLhophysics, ha.p tic perception and recognition o f representa tions, mul tisen.sory objects a nd their underlying ne ural p rocesses pe rception, and sensory-g uid ed motor control.She has applied the results of he r resea rch to a nunlber of real -world problems, including the design of ha ptic and multisen.sory inte rfaces for virtual en vironments and

is a Professor of O tology and Laryn gology at the Harvard Med ical Sch ool with appoinhnentsa t the Harvard-NUT Hea lth, Science, and Tech nology p rogram and tl1e Harvard School of Engineering a nd Applied Scien ce. He is also the Director of the Jenks Vestibular Physiology Laborato ry at the Eye and Ea r lnfinrui ry. Mudi. of his research career has been spen t studying how th e brain combines info rma tion from multiple sources, w ith a speci fic focus on how the brain p rocesses ambigu ous senso1y information from the vestibular system in the presen ce of noise. Tran slational work includes rese..u ch d eveloping new meth ods to help d iagnose p a tients experien cin g vestibular symptoms a nd research d evelopin g vestibular implants for pa tien ts \vho have severe p roblem s ,..,;_th their vestibuki.r organs . OANIE:L M.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

l\ppa

Contents u ...... u..,J6 AmmJ App.l

Introduction

3

Welcome to Our World Sensation and Perception

Sensory Neuroscience and the Biology of Perception 1B

3

Thresholds and the Dawn of Psychophysics 5 Psychophysical Methods 8 Scaling Methods 9

CHAPTER

2•

·

Neuronal Connections 21 Neural Firing: The Action Potential Neuroimaging 24

Summary

•

The First Steps in Vision: From Light to Neural Signals 31 A Little Light Physics

31

Eyes That Capture Light

12

Signal Detection Theory Fourier Analysis 15

3

32

29

.•

•

Convergence and Divergence of Information via Bipolar Cells 42 Communicabng to the Brain via Ganglion Cells 43

Dark and Light Adaptation

Focusing 1.Jght onto the Retina 33 The Retina 35 What the Doctor Saw 36 Rebnal Geography and Function 38

Retinal Information Processing

•

Pupil Size 48 Photopigment Regeneration The Duplex Retina 49 Neural Circuitry 49

40

Ught Transduction by Rod and Cone Photoreceptors 40 Lateral Inhibition through Horizontal and Amacrine Cells 42

22

47

48

Sensation & Perception in Everyday Life: When Good Retina Goes Bad 50 Summary

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

51

CONTENTS

Spatial Vision: From Spots to Stripes

Simple and Complex CeDs 72 R.Jrther Complications 73

53

Visual Acuity: Oh Say, Can You See? A Visit to the Eye Doctor 57 Types of Visual Acuity 57 Acuity for Law-Contrast Stnpes \.Vhy Sine Wave Gratings? 61

53

59

Retinal Ganglion Cells and Stripes

62

64

65

The Topography of the Human Cortex 67 Some Perceptual Consequences of Cortical Magnifleation 68

Receptive Fields in Striate Cortex Crientation Selectivity 71 Other Receptive-Field Properties

CHAPTER

4

/ -'

What and Where Pathways

70

·--.'

The Development of Spatial Vision

83

Development of the Contrast Sensitl\/ity Function 84

Sensation & Perception in Everyday Life: The Girl Who Almost Couldn't See Stripes 65

Summary

87

1\

1

I

Sensation & Perception in Everyday

89 89

The Problems of Perceiving and Recognizing Objects 95 Middle Vision

The Site of Selective Adaptation Effects 80 SpatiaJ Frequency-Tuned Pattern Analyzers in Human Vision 81

72

/.

Perceiving and Recognizing Objects

Columns and Hypercolumns 74 Selective Adaptation : The Psychologist's Electrode 77

The Lateral Geniculate Nucleus The Striate Cortex

ix

97

Finding Edges 97 Texture Segmentation and Grouping 101 Perceptual Committees Revisited 104 Agure and Ground 106 Dealing with Occlusion 108 Parts and \.Vho/es 109 Summarizing Middle Vision 109 From Metaphor to Formal Model 110

Life: Material Perception: The Everyday Problem of Knowing What It Is Made Of

111

Object Recognition

112

Templates versus Structural Descnptions 114 Problems with Structural-Description Theories 115 Multiple Recognition Committees? 116 Faces: An Illustrative Special Case 117 The Pathway Runs in Both D1iections: Feedback and Reentrant Processing 118

Summary

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

119

X CONTENTS

The Perception of Color

123

Basic Principles of Color Perception 1 23 Three Steps to Color Perception

Step 1: Color Detection

The Umits of the Rainba;v 134 Opponent Colors 135 Color in the Visual Cortex 138

123

124

Step 2: Color Discrimination

Individual Differences in Color Perception 140

124

The Principle of Univariance 124 The Trichromatic Solution 126 Metamers 127 n.veen sensation (mind ) and the energy (matter) that gave rise to that serisation. He called both his me thods a nd his theory psychophysics (psycho for mind, and physics for matter). In his e ffort, Fechner was inspired by the findin gs o f one of his Leipzig colleagues, En1st \"leber (1 79&-1878) (Figure 1.4), an anatomist and physio logist who was interested in touch. \•Veber tested he of our sense of touch by using a device much like the compass onenui;ht use w hen learning geometry. He . distance betv,reen two poi.n ts that \Vas used this d evice to rne..'lsure the sm.-tllest required fora person to fee.I toudl on two p oints instea d of one. Later, Fechner would call the dista nce behveen the points the two-point touch threshold. We wi ll discuss t\'lo-point touch thresholds, and touch in general, in Chapter 13. For Fechne r, Weber's most important findings involved judgments of lifted weights. Weber ,..,.·ou ld ask p eople to lift one s tandard weight (a weight that stayed the sa me over a series of experimental trial s) and one comparison weight that differed from the standard. \ Veber increased the compru-ison weight in increm ental over the series of trials. He found tha t the ability of a s ubject to detect the differe nce be hveen the standard and comparison weights depended greatly on the weight of the standard. \\'hen the s ta ndmd was relatively light, people were much better at detecting a s mall difference when they lifted a weight. \ Vhen the standard was heavier, p eople need ed rt bigger difference before they could de tect a change. He called the difference required for deta"ting a change in weight the just noticeable dtfference, or JND. Another te rm. for JND, the s ma llest in a sti mulus that can be d etected, is the difference threshold. \ Veber noticed that JNDs changed in a systematic \·..-ay. The smalles t chan ge in weight that coukl be d etected was a lways d ose to one-fortie th of the standard ""'eight. Thus, a 1-gram change could be d etected when the s t(mdard weigh ed 40 grams, but a 10-gram change was required w hen the s tandard weighed 400 gram s. Weber went on to test JNDs for a few other kinds of stimuli, sud1 as tbe lengths of two lines, for which the d etectable change ratio was 1:100. For virtual ly every measure-whethe r brightness, pitd11 or time-a constant ratio between the change and what was being d1anged could describe the threshold of detectable change quite well. Th is ratio rule holds true except when intensities, s ize, and so on are very small o r very large, nearing the min.imum and m aximum of our senses. [n recognition of Weber's discovery, Fechner ca lled these ratios Weber fractions. He also gave Weber's observation a formula. Fechner named the general rule--that the size of the difference (6..l) is a constant proportion (K) of the level of the s timulus (D-Weber's law. In Weber's observations, Fechner found what he was looking a way to describe the relationship beh"1een mind and matter. Fedmer asslm1ed that the s mallest de tectable change in a s timulus (6.I) could be con sidered a unit of the mind becau se this is the smallest bit of change that is perceived. He then mathe mati cally ex tend ed Weber's law to create what became knovvn as Fechner 's law ( Figure 1.5):

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

klog R

INTROOUCTION

FIGURE 1.5

7

This illustratbn of Fechner' s law shows

that as stimulus intensity grows large r, larger changes are required for the changes to OOdatected by a percelv«.

w h e re S is the psychological sen s'1 ti on , which is equal to the logarithm of the physical s timulus leve l (log R} rnul tiplied by a cons tant, k. This e quation d escribes the fact that o ur p sych ologica l experience o f the intensity o f light, sound, s rne ll, taste, or toud1 increases less quickly than the actual physical stimulus increases. With th is equation, Fechner provided a m a them atical expression tha t formally d em onstrated a re la ti onship between psyche and

physics (psychophysics). Consider the si milarity between Fedmer 's law and Albert Einstein's famous equ a tio n:

E Like mind and body, energy (f)nnd mass(m) had, before Eins tein , been thought o f as d istinct things. CT"he le tter c corresp onds to the speed o f light, a lm os t a billio n feet per second .) Just as Einstein sh o\lved how to equa te energy and m ass, Pechner provided us w ith at lea.s t one way to think abo ut mind and ma tter as equivale nt. As you learn about the senses lvhen reading thi s book, you w ill find that we typicaUy make a distinction between un.its of physical entities (light, sound ) a nd measures of people's perception . For exa mple (as we' ll learn in 01apter 9), we ineasure the physical intensity of a squn"d'(sound pressure lev el) in d ecibels, but we refer to o ur sensation as "lo udness." Feclmer invented new l·v ays to measure what people see, hea_r, and feel. All of his me thods are s till in u se today. ln explain.1.ng these rne t:hods here, we will use absolute threshold us an example becau se it is simpler to understand, but we would use the s..1.rne methods to d etermine difference threshold s s uch as An absolute threshold is the minimum intensity of a s timulu s tha t can be detected (Table 1.1 ). This returns us to the question we raised earlie r; W hat is the faintest sound yo u can hea r? Of cou rse, we could ask the same ques ti on about the faintest light, the lightes t touch, and so forth . (See Web Activity 1.1 : Psycho physics.)

TABLE 1.1 Some commonsense absolute thresholds

sense

Threshold

Vision

Stars

Hearing

A ticking

Ves.tibular

A tilt of less than half a minute on a clock face

Taste

A teaspoon of sugar in 2 gallons

Smell

A drop of perfume in three rooms

Touch

The wing of a fly falling on your cheek from a

at night, or a candle flame 30 miles away on a dark, clear night watch 20 feet away, with no other noises

of water of 3 inches

source: From Galanter. 1962.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

absolute tlTeshold The minimum amount of stlrnulatloo necessary fcr a persoo to detect a stimulus 50% of the time.

a Appa

8

CHAPTER 1

8230336 AmMa Appa (b)

(a)

100

JOO

"I hear it:' !!. 75

lfil"

50

____________ _

"Cl

.W

J

"I don't hear it:' 11

JO

25

12

Stimulus le\'el (a rbitrary units)

JO

II

12

Stimulus level {aibitr.uy units)

FIGURE 1.6 The mg.thod o f constant stimuli . (fl) We might expect the threshold to b9 a sharp change in detection from never reported to always r.;:ported , as depicted hGr9, but this is not so. fP) In rQ(l.lity, QXP"1flments ITl9asuring absolute thrashcid produce shallower functions rtilatlng stimulus to response. A somewhat arbitrary point on the rurve. oftm 50% detection , is designated as the threshold.

Psychophysical Methods

method of constalt stimuli A i:Gychopl1yslcal method In which many stimuli , from rareti,i to almost always perceivable (or rarely to almost always perceivably different from a reference stimulus), are presented one at a time. Participants respond to each presentation: "yeafno," •same/different," and so on.

H ow can we measure an absolute threshold in a valid a nd reliable milnner? O ne method is known as the method of constant s·ttmull. This me th od requires creating m a ny stimuli with d ifferent intensities in order to find the tiniest intensi ty tha t ca n be d e tected (Figure 1.6). lf you 've had a hearing test, you you could a nd could n ot hear a tone that the may recall ha ...ing to report m1di ologistpl.ayed to you over headpho nes. In this test, intensities of all of the tones would be relatively low, n ot too far above or below the intensity where your th resh old was expected to be. TI1e tones, varying in intensity, '"''ou.ld be presented randomly, and tones would be presented multiple times at each intensity. The "multiple times" piece is important. Subtle perceptual judgments (e.g., thresh o ld judgrnent.s) are variable. TI1e stimulus va ries for physical reasons. Tile observer varies. Attention waivers and sensory sys tems fluctua te for all sorts of reasons. As a con seque nce, one measu re is almost never enough . You need to repeat the measure over and over and the n average responses or othenvise describe the pattern of re.s Lt! ts. Some experiments require thousands of repetitions (thous ands of "tr ials'1 to establish a s ufficiently reliable data point. Re turning to our auditory exarnple, as the listene r you would report w he ther you hemd a tone or not. You would always report hearin g a tone that was relatively fur above threshold, and almost never re p o rt hearing a tone that was well below threshold . In ben.veen, h owever, you would be more likely to hear some tone intensi ties than not hear the m, and you \Vould h ear other, lower intensities on only a few presentations. In generat the intensity a t w hich a sti mulus would be detected 50% of the time would be chosen as your thresh old. That 50% d efi nition of absolute thresho ld is ra ther interesting. Weren' t we looking for a way to measure the \Veakest detectable s timulus? Using the hearing exa mpl e, .shouldn ' t that be a val ue below w hich \.Ve ju st can 't hear an ything (.see Figure l.6a)? It turns o ut ti-Mt no .sud1 hard lxnmdary exists . Because of variability in the n ervous system, stimuli near thre.shold will be

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

9

INTROOUCTION

d etected sometimes and missed at other times, As a resu.lt, the re la ting the probability of d etection ·with the s timulus level MJI be more grad ual (see Figure l.6b), an d we mus t settle for a somewha t arbitra ry d efinition of an absolute tluesh old,. (We wi ll re turn to this issue when we talk abo ut signa l detedion theory.) The rneth od of constant stinnili is simple to use, but it can be inefficient in an experiment because much of the s ubject's time is spent wi tll stimuli that are clear ly well above or below threshold . A som ewhat more efficient approach is the method of limits (Figure 1.7) . Wi th this method, the exp erime nter begins with the same set of s timuli-in this case, tones that vary in

inte nsity. Instead of random presentations, tones are presented in orde r of increasing or d ecreasin g intens ity. When tones are p resented in ascending order, from faintest to loudest, listeners are asked to repor t w hen they first h ear tile tone. \ Vith descending order, the task is to report w hen the tone is n o longer audible. Da.ta from a..n e xperiment s ud1 as this s how tha t there is some "overshoot" in jud gm e n ts. [t us ual.ly takes more inten sity to re port hearing tile tone whe n in tensity is increas ing, a nd it takes more decreases in intensity before a listene r reports tlla t the tone cannot be heard_\ Ve take the average of tllese crossover points-\'\-hen lis tene rs shift from reporting hearing the tone to not hearing the tone, and vice versa-to be the tlue.sho ld . The third a nd final of these classic measures of thresh olds is the method of adjustment This me thod is just like the method o f limits, excep t the s ubject is the one who steadily increases or decreases the intensity of the s timulus . The method of adjust men t m.1y be the easiest me thod to understand becau se it is mudl like d ay-to-day activities s uch as adjusting the volume dial on a stereo or the dimmer 5"\.'itch for a light. Even tho ugh it's tile easiest to unde rstand, the me thod o f adjus tm ent is not us ua lly used to measure thresh old s. 11le m etl,od wo uld be perfec t if tluesh old data looked Uke tlw.se p lotted in Figure L 6a.. Bu t, real data look m ore like Figure l.6b. The same person will adjust a dial to different p laces on diffe rent trials, an d m easurements ge t even m essier w hen we try to combine the da ta from multip le persons .

Trial series

11

j2

13 14 Is f6 j7 ta

20

y

19

y

y

y

y

y

lS

y

y

y

y

17

y

y

y

y

16

y

y

y

y

y

15

y

y

y

y

y

y

y

y

14

y

N

y

N

y

N

y

y

13

N

N

y

N

y

N

N

l2

N

N

N

N

N

N

11

N

N

N

N

10

N

N

N

N

-- --- - -- - - -- - -- -- --- -y

13.5 14.5 1251 4.5 ! LS 14.5 13.5 12.5 Crossover v allles (average-= 13.5)

FIGURE 1.7 Th9 method of limits. Hef9 the subject attends to multiple series o f trials. Fo r each s9ries, the intensity o f the stirnulus Is gradualty incroosed or decreased until the subject d.:atacts M or fails to detect (N), respectlw ty. s titnulus. For each series. an esti rnate of the threshold (red dashed line) Is taksn to be the avqage of the stimulus level jus t before and aft9r the dlang9 In peroeption.

Scaling Methods Mov ing beyond absolute thresholds a nd difference thresh olds, s uppose we '""' anted to kn ow about tile magni tude of your experiences. For examp le, \Ve might give you a li ght and ask how muc h additional light you would need to make a nothe r light that looks hvke as bright? Though that might seem like an odd ques tion , it tunlS out to be a nswerable. We could give you a knob to adjust so that you could set the second light to appear twice as bright as the first, and you could do it We don 't need to give the observers a light to adjust. A s urprising! y s traightforward way to ad d ress the question of the s trength or size of a sensa tion is to simply ask observers to ra te the experience. For exa mple, we could give observers a series of s ugar solutions and as k the m to assign numbers to each sample. We would jus t tell o ur observers that S\'/eeter solutions should get bigger nun1bers, and if solution Aseen.lS twice as sweet as solution B, the number assigned to A should be twice the number assigned to R This m e thod is called magnitude estimation. l11is approadl actua lly works well, even w hen observers are free to choose their ovro r.1nge of numbe rs . More typically, we might begin the experiment by presenting one solution at a n inte nnediate level and telling the taste r to label this level as a specific val ue-10, for insta nce. All of the responses should then be scaled sensibly above or below this stand a rd of 10. H you d o this for s ugar solutions, you ,viU get d a ta tk1t look like the blue "sweetness" line in Figure 1.8.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

method of limlts A psychophysical method Jn which the particular dllneneJcn of a stimulus . er the dlffererce betvveen two stimuli, Is varloct Incrementally until the participant responds d ifferently. method of adjustment A method of limits In which the subject controls tre change In tre stimulus.

magnitude estimation A psychophysical method In which the pant assigns values accord ing to pef celved magnitudes of the stlrnull.

10

CHAPTER 1

100

FIGURE 1.6 Magnitude eistimatio n. The lines on this graph represent data from magnitude estimation QXpel'"iITIQflts using electric shocks o f diffgrent cutnaots, lines o f diff9t"ent lengths . solutions o f different sweetn9Ss leveils. and lights o f diffQ1'"'11nt brightneSS9S. Th9

90

80

exponents that describe these lings are 3.5. 1.0. 0.8. and 0.3, respectively. Fer exponents greater than 1,

such as for electric shock, Fechner' s law does not hcid ,

70

and Stevens' power law must be used instead.

§"'

60

.s

1l 50 40

30

8230336 Arnrna Ap

20 10 30

40

50

60

70

80

00

100

Stimulus energy

stevens' power law

A prln::::lple

describing the relationship between stimulus and resulting sensatrn that says the mag nitude of subjectt-le sensatlm Is proportional to the stimulus magnitude raised to an exponent.

Harva rd psychologis t S.S. Stevens (1962, 1975) in vented magni tude estimation . H e, his s htdents, and successo rs. measured functions like the one in

Pigure 1.8 for many different sensations. Even tho ugh observers were asked to assign num bers to priva te experien ce, the resul ts were m derly and lawful. H m,.,•ever, they were not the sam e for every type of sensation. Thzit rebtionship betv..·een stimulus intensity and sensation is described by w hat is n ow know n

as Stevens' power law: w h ich states tha t the sensation (S) is rela ted to thes'timulus intensi ty(!) by an exponent (b). So 1 for ex.ample, experienced sen sation might rise with intensity squared (Ix l). That would be an exponent of 2.0. If the exponent is less than 1, this mea ns that the sensat ion grows less rapidly than the s timulus. 11-U.S is w ha t Fechner's law and Weber's law would predict. Suppose you have some lit ca ndles and you li ght 10 more. If you started \.vi th 1 candle, the change from 1 to 11 candles must be quite drainatic. lf we add JO to 100, the change will be modest. Adding JO to 10,000 won°t even be notice1ble. In fact, the exponen t for brightness is about0.3. The exponent for swee tness is about0.8(Bartoshuk, 1979). Properties like length h ave e..xponents n ear 1, so, reasonably enough.1 a 12-inch-lon g s tick looks twice as lon g as a 6-in ch-long s tick (S.S. Stevens and Galanter, 1957). Note tha t th is relations hip is true over only a m oderate range of sizes. An inch added to the size of a spider changes your sensory experience mud\ m o re than an ind1 added to the height of a giraffe. Som e stimuli ha\.-e exponents g reater tha n 1. In the painful case of e lectric sh ock, the pain grows wi th f3.S (Stevens, Ca rton, and Shickman, 1958), so a 4-fo.ld increase in the electrical current is experienced as a 128-fold increase in painJ At this p oint in ou r discussion of psychophysics, it is worth taking a moment to compare the three laws that ha\·e been p resented: Weber's, PedUler 's, and Stevens'.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

INTRODUCllON

• VVeber's fmt1 involves a d ear objective measllrement. \Ve know how

much ,.,,.e va ried the stimulus, and eithe r the observers can te ll that the stim ulus changed or they cannot.

• Fedmer's fnw begins \v:ith the sam e sort of objecti ve mei.1Sureme nts as 1Neber 's, but the law is actually a calculation based on some assumptions :about how sensati on works. Jn particular, Fechner 's law assumes tha t all JNDs are perceptually equi valent. ln foct, this a ssumption turns ot.1t to be incorrec t and feads to some places w here the "law'' is viola ted, such as in the e lectric shock example just given. •

Sff?ivms' power

law describes rating da ta quite well, but notice that rating

11

cross-modality matching The to match the Intensities of sensations that come from different sensory mod811tles. This ability allows Insight Into sensay differenoes. For example. a listener might adjust the bnghtness of a light until It matches the

loudness of a tone.

supertaster Supertasters are th:::E;e lndMduals who experience the most

Intense taste soosatlons: fcr some

stimuli, they are dramatically mae intense than fa rnedlurn tasters or nontasters. Supertasters atso tend to experience more Intense oral burn and cral tc:uci1 sensatlcn s.

data are qualitatively differe nt fro m the data tha t s upported \¥eber 's law. \Ve can record the subjects' ratings a nd we can d)eck w he ther those ratings are reasonable and. consistent, but the re is n o way to know w hether they are objectively right or wrong. A use ful varian t of the scaling method can s h ow us tha t different individ uals can live in different sensory worlds, even if they are exposed to the same stimuli. The me thod is cross-modality matching Q. C. Stevens, 1959). In cross-modality matching, a n observer adjusts a s timulus of one sort to m atd1 the perceived magnitude of a s timulus o f a completely different sort. For examp le, we might ask a listener to adjust the brighh1ess of a light until it matches the loudne:..s of a particular tone. Again, though the task might sound odd, people c..m do this, a nd for the most part, everyone with "nomKtl" vision hearing will produce the same pattern of matches of a sound to a light We still can 't exa nUne someone else's private experience, but a t least the relationship of visua l experience and auditory experien ce appears to be si milar across individuals. Not so w hen it comes to the sense of taste. There is a molecule cal led p ropylthiouracil (PROP) that some people experience as Supa-tastersvery bitter, '"'hil e others expe rie nce it as almos t tu.steless. S till oth ers fall in between. This relationship can be examined formally with cross-modali ty matching (Marks et al., 1988). If observers ore asked to match the bitterness of PROP to oth er .sensations comple tely unrelated to taste, we do not find the sort of agreement that is found w hen observers match sounds and lights (Figure 1.9 ). Some people- we'll call them nontasters-match the taste of PROP to very weak sensation s like the solUld of a watch or a w hisper. A group o f supertasters assert that the bitterness o f PROP is simila r in intensity to the bri°'tness Qf the. sun o r the most inten se pain ever experienced. Medium tasters matdl PROP to weaker stimuli, such as the smell of frying bacon o r the pain of a mild headache (Bartoshuk, Fast, and Snyder, 2005). As \\Te """'ill see in Chapter 15, there is a genetic basis for this variation, and it h .:1.S wide implications for our food preferences and, conseciuently, for health. For the present discussion, this example s hows tfo:tt Medium__.,. we can use sca ling methods to quantify '""hat appear to be rea l d ifferences in individuals' taste experiences.

Matching sensations

l.oudaot"""""

Brightut light

Brightness of the sun

Heat of scalding water Sound of a fire engine

pa

Brightness of low-beam headlights Smell of bacon frying Pain of a mild h eadache

Brightne.ss of the moon/ loudne.ss of a convers.:1tion FIG U RE 1.9 Cross- rTOdality matching. Th9 lwels of bitt9rn8SS o f cortCQntrated PROP perc eived by nontasters, rredium tastgrs, and supartasters of PRO P are shO\M"'l on th9 l9ft. Th9 perceiv9d intensities of a varigty o f everyday SQnsatb ns arg shown ori the right. The arraw from Qach taster typ9 indicates the o f sensation to wl1 ich thosQ tasters matc hed th9 tast9 o f PROP. (Data from Fast, 2004.)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Nonlastn

(cf) Co1Yect rejection NO

el

Criterion

rh

Less -

- More

SoWld s like phone

(fl False almn

(e) Hit YES

NO

I

Sounds like ph one

Ci"it'2rion

rh

LeM> -

YES

- More

Sounds like phonll?

NO

(g) !\fas

Criterion

YES

rh

I Less -

NO

Criterion

YES

I

- More

Sounds like phone

Less -

- Mon

Sounds like phone

FIG URE 1. 11

Det9Cting a s timulus using signal d et ection theory (SDl). (a) SDT assumes that al l pgrceptual decisions must be made against a background of noise (the red curv e) g.:nerat:ed In the w-orkl Of ln thie nervou s system . tp) Your job is t o distinguish neivous systgn rQS?Onsss du.e to noise alonQ (r.ed) o r to signal plus nolS9 (b lue). (c) The best you can do is 9StabHsh a criteflcrl (dotted line) and declare that you detect something if the response is abOV9 that c riti9flon . SDT indud9s four c lasses of responses: rejections (you say "no' and there Is, Indeed, no signa O: (9) hits (you Sf£f and then;1 is a sig nal) ; fal se-alarm i:Ti ting h is very influential Hn11dbook of Physiology during the early 1830s. In this book, in addition to covering most of w hat was then known about physiology,lvli.ille r formulated the doctrine of specific nerve energies. The central idea of this doctrine is that Vv""e cannot be directly aware of the world itself, and we are only 3ware of the activity in our nerves. Further, wha t is most impor t.ant is which nerves are s timulated, and no t haw they are stirnulated. For exa mp le, we exp erience vision because the op tic n erve leading fr om the eye to the brain is s timulated, but it d oes not ma tter whethe r light, or .something else, s timulates the nerve. To prove to yourself that this is true, close yom eyes and press very gently on the outside com er of one eye through the lid. (This works better in a darkened room.) You wi ll see a spot of light toward the insid e of your \·isual field by your nose. Desp ite the lack of s timulation by light, yo ur brain interp rets the input from y01rr optic nerve as in for min g you about some thing visual. The cranial nerves leading into and out of the skull illus trate the d octrine of specific nerve energies (Figure 1.20). The pair o f optic nerves is one o f 12

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

INTRODUCTION

19

FIG URE 1.20 Twetv9 pairs of cranial nerv9S pass through small openings in the bone at the base of the skull, AH of therSe nerves conduct for sensation, motorbehavicr. Cf b oth. (After Breedlove and Watson. 20 13.)

VIL F.1ci.1 l Tongue, soft palate:

pairs of crani..'ll ne rves that pass thro ugh s mall ope nings in the bone atthe base of the s kull . The cra nial nerves o:u e d edicated m ainly to sensory a.n d m otor systerns. Crnnial nerves are labeled both by names and by Roman numerals thilt roughly correspond to the order of their loc.a tions, beginning from the

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

20

CHAPTER 1

oHactory (I) nerves

The first pair of

cranial nerves. The axons of the olfactory sensoiy neurcns bundle together after passing throu;ih the crlbrlform plate to fam the dfactory nerve, which 001ducts impulses fran the olfactory eplthella In the nose to the olfactory bulb. optic (IQ nerves

The ooccnd pair

of cranial nerves, which arise trom the retlra and c arry ;;sual Information to the thalamus and other parts of the brain. vestlbulocochlear (VllQ nerves p air of cranial nerves, which ocnnect tlie Inner ear ,.;111 the brain , transmitting Impulses concerned with hearing ard spatial els, reapter 3).

functlonal magnetic resonance Imaging (IMRI) A variant of magnetic resonance Imaging that makes It pooslble to measure bcallzoo patterns of activity In the brain. Activated neurons provcke Increase:! blood fiow. which can be quantlflecl by measuring changes In the response of oxygenated and dooxygenatoo blood to strong magnetic blood oxygen level-dependent (BOLD) signal The ratio of oxygenated to deoxygenated hemoglobin that permits the lccallzatlon of brain neuIn a task. rons that are most

posrt:ron errisslon tomography (PET) An Imaging techndogy that enablee us to delne locations In trn brain where neurons are especialty actr...-perimen t o r series of experiments. You might use behavioral methods to determine the nature of a particular perceptual but s uch methods mjght take hollfS and htrndreds or thousands of tria ls. If your observers can't spend that much time in the scanner, you might h ave them perform a shorter, stripped-do\vTl version of the study while being imaged. You might do yet an o ther version of the task, specialized for the de mands of BEG recording. By means of these "con vergin g operations," you co uld build up quite a d eta iled picture of what perceives and how her brain gives rise to those perceptions. Now tha t yo u've compl eted this brief tour of the methods of the trade, you .sho uld be ready to ern.bark on n m ore comprehensive examina.t.ion of sen.""3tion and perception. We h ope you enjoy the journey.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

INTRODUCTION

Looking

Li stening

Thinle fighi J i11 actually-\,,.1ch the retina. Much of the light energy will have been lost in space or the atmosphere, becc1 use of absorption and scatterin g. as d escribed already. In addition, a good deal of light becomes lost in the eyeball, so only about half of the that arrives at the cornea actually reach es the retin..1 . The role of the reti na is to detect light and "te ll the brain about aspects of lig ht that are rela ted to objects in the world" (Oys ter, 1999). [n othe r words, the retina is w here seeing really begins.

Focusing Ught onto the Retina To foc us a dis tant star on the retina, the refractive power of the four optic components of the eye mus t be perfectly matched to the length of the eyeball. This perfect matd1, known as emmetropia, i s illus trated in Figure 2.3a . The average eyeba ll is abo ut 24 millimeters (mm) lon g a nd h..1s a power, when unaccommo dated (focu sed on a dis tant object like our star), of about +60 diopters. (We'll explain what accommodation means in a bit.)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

acp.Jeous humor The watery fluid In the anter1or chamber of the f1'10. lens Tue lens Inside the eye that enables the c hanging of focus.

pupil The dark. circular cpenlng at the cent"' of the Ins In the fft9, where light enters the fft9. Ins The colored part of the eye, consisting of a muscular diaphragm surrrundlng the pupil and re>gulatlng the Ilg ht enten ng the eye by expardlng and contracting the pupil.

vitreous humor TI'e transparent fiuld that Ills the Wreom chamber In the posterior part of t11e eye.

retina A light-sensitive membrane In the bac k of the "fe that =italns rods and cones, which recetve an Image frcm the lens ard sand It to the brain thrrugh the optb nerve. ernmetropla The condition In which there Is no refractive error. because the refractive pow... of the eye Is perfectly matched to the length of the eyeball.

diopter (D} A unit of measurement of the cptlc P'.>Wl

0

(b) M yopi.:l

myopia Nearsightedness, a com· rnon cordltlon In which light entering the f!!Ye Is focused ln frcnt of the retina and distant objects cannot be soon sharply. hyperopla Farsightedness. a common cordltlon In which light entering the eye Is focused behlrd the retina and accornrnodatlon Is required In crder to see near objects clearly. astl9'Tlatism A visual defect caused by the unequal of ore or more of the refractive surtacoo of the eye, usl.Blly the cornea.

accommodation The process by which the eye changoo Its focus (In w111c11 the lens gets falter as gaze Is directed toward nearer objects).

.

FIG URE 2.3 Optics o1 the human gye. See the text for details. {After Oyster. 1999.)

Refractive errors occur when the eyeball is too long or too short relative to the power of the optic components. If the eyeba ll is too long for the optics (Ftgure 2.3b), the image of our star will be focused in front of the retina, and the star will th us be seen asa blur rather than a spot of li ght. Th is condition is called myopia (or "nearsightedness"). Myopia can be corrected with negative (mi nus) lenses, wh ich di verge the rays of starlight before they enter the eye (Figure 2.3c). ll the eyeball is too s h or t for the op tics (Figure 2.3d), the image of our star will be focused belrfnd the retina-a condition rnlled hyperopia (or "fars ig htedness"). If the hyperopia is not too severe, a youn g hyperope ca n compensate by accommoda ting, the reby increasing the power of the eye. If accommodation foils to correct the hyperopia, the star's image ,,,,.ill again be blurred. Hyperopia can be corrected vdth positive (plus) le nses, w h ich con verge the rays of s ta rli ght before they en ter the eye. On average, the adult human eye is 24 mm long, abo ut the diameter of a quarter. However, eyeballs can be quite a bit longer or s horter and s till be ernmetropk because eyes generally grow to ma tch the pmver of the optic components we're born with. (Most newborns .u'e' hyperopk because the optic components of their eyes are relatively well developed at birth compa red with the len gth of tl1eir eyeballs.) TI1e m ost powerful refracting surface in the eye is the cornea, w hich con\Vhen the come.a is no t lTibu tes about two-thirds of the eye' s focusing sphericat th e result is astigmatism. With astig matis m, verti ca l lines might be focused sli ghtly in front of the retirn::i1 w hil e horizontal l.ines are focu sed slightly behind it (or vice versa). If you h.:1 ve a reasonable degree of uncorrected astigrnatis m, one or rnore of the lines in Figure 2.4 might appea r to be o ut of foc us whil e other lines appear sharp. Len ses that have h-vo focal poin ts (that is, lenses that provide different amounts of focusing power in the horizontal and vertical planes) can correct astigmatism. So far, we've considered the image of a distant object. But wha t happens when \Ve wa nt to focus on some thing dose by (like the words on this page)? Remember that refraction (light bending) is n ecessary to foc us light rays. Because the cornea is highly curved and has a higher refractive index tha n air (L376 versus 1)1 it for ms the most p owerful refractive s urface in the eye. The aqueous and "itreous humors also help refract light. However, the refractive power of each of these three structures is fi.xed, so they cannot be used to bring close objects into focus. This job is performed by the lens, which can alter the refracti ve power by cha nging its s hape-a process called accommodation. Accommodation {change in foc us) is accomplished through contraction of the cil.iary muscle. ll1e lens is attached to the ciliaiy muscle th.rough tiny fibers (knm'\-T\ as the zonules of Zinn) (see Figure 2.2). When the ciliary muscle is relaxed, the zonules c'.\re stretched and the lens is relatively flat. ln this s tate, the eye w ill be focused on very distant objects (like our s tar). But to focus on something a \-Vristwatch-the ciliary muscle mus t contract. This contracti on red uces the tension on th e zonules and enables the lens to bulge. The fatter the lens is, the m ore p ovver it has.

FIG URE 2. 4 Fan chart for astigmaJ:ism . Tais of accommcdatlon,

which makes It near objects.

to focus on

cataract An opacity of the crystal· lln:> lens.

36

CHAPTER 2

FIG URE 2..6 Fundus of the right eye o f a human. {From Rodi.::ock. 1998.)

li ght energy from o tu star is transduced into neural energy that can be interpreted by the brain.

What the Doctor Saw transck.lce To convert from one form of enerw to anothEf (e.g., from "1)ht to neural electncal erergy, o r fran

rnechanlcal movement to neural elec-

trical energy). fundus The back layer of the retina: what tha aye doctor sees through an cphtl'JalmCSO(lle.

Eye doctors use an instrument cal led thelicun Photorecepto r layer Externcil limiting m embrane

l

"'] }°'""''S"-' } Inner &2,Sments

=

} Photorecepto r nuclei

Outer nuclear layer {

•

°'"". plexiforn>lay"· { lnnernudm Jar"

Inn" p1'xifo•= Jar"

}'

*

U

l j) Bipol.u Asual ac uity and seNeS as the point o f flx.9tbn.

eccentricity The distance between the retinal Image and the fwea.

As we'll see in the next section, together these neurons constitute a minicomputer that begins the process o f interpreting the information contained in \o·isu al images. The transduction of light energy into neural energy begins in the back.most layer of the retina, which is made up of cells called photorec:eptors (see Figure 2$). When photoreceptors sense light, they can s timulate neurons in the intermediate layers, including bipolar cells, horizontal cells, and am acrine cells. These neu rons then connect \o'.'l th the frontrnost layer of the retina, m ade up of gan glion cells, whose axons pass th rough the optic nerve to the brain. Before we describe the function of these layers, we sho uld a ddress an obvious question regarding the structure of the retina (see Figure 2.8): Why are the photoreceptors at the back-that is, in the last layer? This ar ran gement requires ligh t to pass throu gh the gang li on, horizon tal, and amacrin e cells before making contact with the photoreceptors. Hov1i'ever, these n eu rons are mostly transparent, whereas cel ls in the pigment epithelitun, \vhich p rovide vi ta I nutrients to the photo receptors, are opa que. O n ce we see th at the photo-receptors mus t be next to both the pig1r1ent epithelium and the o ther neumns, the layering order makes much m ore sense.

Retinal Geography and Function The retin a con tains roughly 100 million photoreceptors. These are the neurons that capture lig ht and initiate the ac t o f seeing by producing chemical sign als. The hmn an retina contains at least two types of photoreceptors: rods and cones. l11e.se two types not only h(1ve diffe rent shapes ('"'·h ich is how they earned. their namesi see Figure 2.9}, but they have different distributions across the retina and serve different hmctions. Because retinas have both rods and cones, they are cons idered to be duplex retinas. Some animals, su ch as rats and owls, have n10s tl y rod retinas; others (e.g., certain lizard s) have mostly cone retinas. Huma ns have many m ore rods (about 90 million ) than cones (about 4-5 million), a nd the two types of cell s ha ve very different geographic distributions on the retina (Figure 2.10 ). Rods a re completely absent from the center of the fovea, and their d en sity increases to a peak a t about 20 d egreES and then declines again. The cones a re most con centrated in the center of the fovea, an d their dens ity drops off drama ticaJly wi th retin al eccentricity (d i.s tance fr om the fovea). The fo vea is the 11pit" in the inne r retina that is specialized for seeing fine detail. As the photographs of photoreceptors a t different eccentricities in Figure 2.10 illustrate, cones are also smaller a nd more ti ghtly packed in the fovea l center (0.0 in Figure 2 10). This "rod-free" area (about 300 square microme ters [µm] on the retina) s ubtends a visual an gle of about 1 degree, and it is directly behind the center of the pupiL So if we look directly at an object whose image TABLE 2. 1 Properties of the fovea and periphery in human viskm Property

Fovea

Photoreceptor

Mostly cones

Mostly rods

Bipolar cell type

Midget

Diffuse

Periphery

Convergence

Low

High

Receptive-field size

Small

Large

Acuity (detail)

High

Low

Light sensitivity

Low

Hi gh

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE FIRST STEPS IN "1SION

16D

8.0

5.0

1.35

0.0

1.35

5.0

SD

39

16D

10 TcrnpornJ

OUtan.:e from foveal

(mm)

0£ 20 Visual .m gle FIGUR E 2.10 Photoreceptor density across the retina The top pane.ls show sMceis through the photoreceptor innq segmns at different eccentriciti9S (dlstane&s from the fovea). The grar::fi shows the density of rods and oones plottoo as a function of from f0\'9a. Note that In the PElfipheral slices, the cones are always the larger OOls. i;Aft9r Oyster, 1000; rnk::rographs from Curcio et al., 1000.)

is s malle r than 1 degree, the image w ill land on a region of the retina that has only cones. (How big is 1 degree? rule of thumb, illus tra ted in Figure 2.11 : your thmnb, when viewed a t am1's length, subtends an angle of about 2 degrees on the retina, assuming your thumb is about 2 cm across and your

outstretched arm extends about 57 011 from yo ur eye). Table 2.1 illustra tes sorne of the funda mental differences in the properties of the fovea compared

FIG U RE 2.11 The o f thumb": w hen v igWQd at arm 's lgngth, your thumb subt.cnds an anglg o f about 2 dggrQQs on the r&tina

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

40

CHAPTER 2

l6Am wi th the peripheral retina. Most important for us, the fovea has hi gh acuity an d we use it to identi fy objects, to read, and to inspect fine d e tail. O n the o ther hand , we use the periphery w hen d etecting and localizing stimuli tha t we aren't lookfrlg at directly (e.g., seeing a m oving truck out of the "corner of the eye" ).

TI1e cones become larger and more sparse away from the foveal center, and the s mall ce lls that appem outside the fovea (e.g., 1.35 in Figure 2.10) are rods, which a re about the same size as the foveal cones. ln a11 of the micrographs excep t for 0.01 the large cells are always the cones. Rod.sand cones operate bes t under different lighting conditi ons: Ro ds hmction relatively well under conditions of din.1 (scotopic) illumination (whid1 is w hy anitnal.s wi th all-rod retinas are nocturnal), but cones require brighter (pho topic) illumination (e.g., s unlight or room Ligh ts) to opera te efficiently. Having an area at the center of the fovea with n o rods means th at under dim illumina tion the central 1 degree or so around the fovea is e ffectively blind! Ind eed , practiced. s targazers know tha t it is often easier to s po t a dim s ta r by looking o ut o f the comer o f one's eye than by looking directly at it. We '"rill revisit ph otopic and scotopk vision again in 01ap te r 5. Rods and cones d iffer fun ctio n.a.Uy in another important way. Because all rods h ave the same type of photo pig ment, they ca nno t signal differences in color. Ead1 con e, on th e o ther hand , has one of three differentpho topig ments that differ in the wavele ngths at w hich they absorb light m ost efficientl y. Therefore, cones can signal information about wavelen gth, and thus they provide the basis for o ur co lor vision.

Retinal Information Processing The retina contains fi ve major classes of ne urons: photoreceptors, h orizontal cells, bipolar cells, amacrine cells, and ganglion cells mentioned in the prev ious section (see Figure 2.8). Let's take a cl oser look at the fonctions of each of these cell types. (See Web Activity 2.4: Retinal Structure.}

Light Transduction by Rod and Cone Photoreceptors

outer se"11ent

The part of a ph:>-

toreceptor that contains photcplgment

molecules. inner segment The part of a photoreceptor that lies between the outer segment and the cell nucleus.

synaptic terminal The locat bn where axons tennlnate at the synapse for transmission o1infa'matlon by the release of a chemical transmltt€f, chromophore The light-catching part of the >Asual pigments of the retina. rtlodopsln The visual pigment found In rods. melanopsin A photcplgment that Is sensitive to ambient light.

Both types of photoreceptors consist of an outer segTient (lvhich is adjacentto the pig1nentepitheli\.m1)1 an lnnersegrnent,and a synaptic terminal. Molecules called visual pigments are m ade in the inner segment (whid1 is like a little factory, filled with mitochondria) and stored in the outer segment, w here they are incorporated into the membran e. Each visual pigment mo lecule consists of a protein (an opsin ), the s tructure of v; hich determines w hich \Vavelengths o f light the pigment molecule absorbs, and a chromophore, w hlch captures light photons. The chromophore is the part of the molecule respons ible for its color, and it selectively absorbs specific vvavelengths of light. The chromophme, known as retin.:1t is de rived from vitamin A, \vhich is in hun m anufactured from beta-carotene, w hich is why your m other told you to ea t your carrots! The opsin and ch romophore are co nnected . Each photoreceptor has only one of the four types of vis ual pigments fol.ind in the human retina. ll1e p igment rhodopsln is fo und in the rods, concentrated rnainly in the stack of men'\bran ous discs i..n the outer segment. Each cone has one of the other th ree pigments-which respond to lon g, med.i LUn,and short waveleng ths, respectively. Recen t evidence su gges ts that there may be a third type of pho toreceptor--one that ''lives'' a mong the ganglion cells and that is involved in adjusting o ur bi ological rhytlu11s to m.a td1 the day and ni ght of the externa l world {Ba:ringa, 2002). These photorecep tors are sensi ti ve to the ambient light level and con tain the photopigment melanopsln, and they send their signals to the

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE FIRST STEPS IN

suprachiasmatic nucle us (SCN), the home of the brain's circadian clock, \e M gan glion cells.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

43

bipolar cell A retinal cell that synapses "1th either rods cr cones (not both) am with horizontal cells. and

then passes the signals on to ganglion cells. diffuse bipolar cell

A bipolar retinal

cell wt'Dse prooesoos are spread a.rt to re::::etv.e Input frcm multlple cx:::nes. sensttMty

1. The ablllty to PffOOMl

via the sense organs. 2. Extreme rooponsMiness to radlatbn, especially to Ilg ht of a spedl c wavelength. 3. The ability to respond to transmitted elgnals.

visual acuity A measure of the flnest detail that can be resorJed by the eyes. midget bipolar cell

A small bipolar

cell In the central retina that rec:elves lnpLrt frcrn a elfXie cone. ON bipolar cell A bipolar c ell that responds to an lnorease In light c aptured by the cones. OFF bipolar cell A bipolar cell that responds to a decrease In light captured by the cones. cell

A retinal cell that

receives visual lnfamatlon from ph:::ito· receptors via 11..o lntet"medlate neurcn types (bipolar cells and amacrlne cells) and transmits lnformatlcn to the brain

and mldbrain. P ganglion cell A small ganglion cell that recewes excitatory Input elngle midget bipolar cells In the central retina am feeds the paivocellular 1(:('fer of the lateral genlculate nu:::leus. M ganglion cell A ganglion cell resembling a little umbrella that

reOOvoo excitatory Input from diffuse bipolar cells and feeds the magno-

cellular layer of the lateral genlculate nu:::leus.

44

CHAPTER 2

FIG URE 2. 13 Different types o f retinal P and M ganglion OOls . ShO"Ml are ganglion calls in sQCtion. (After Oyst er, 1009.)

1l1e astute reader may have no ticed that M and P ganglion cells togethe r cons titute about 8()Q/u of a ll ga ng lio n cells . Other gang lio n cell types, known as konlocellularcells, project to koniocellular layers in the LG N. Some of these, wi th input from S-cones, may be part of a ''primordial '' blue-yellow p athwc1y (seeChn pter 5), w lUle yet other gan glion cells that p roject to the konlocellular layers a re thou ght to corresp o nd to "n onblu e 11 ko n.iocellular cells. CENTER-SURROUND RECEPTIVE FIELDS

Much o f w ha t w e kn ow about

h O\v retinal gan g lion cells work comes from p ain staking physiological s tudies in wh ich tiny electrod es are used to s tudy the electrical ch anges in individual ga ng lion cells. Gangli on cells fire acti on p o ten tia ls sp ontaneo usly, at il bout on e spike p er second, e ven in the absen ce o f vis uaJ s timul ation. H owever, each gang lion cell has a sm a ll w indow on the world kno\-vTl as its receptive field . The recep tive fie ld is the region on the retina in which vis ual s timuli influence the neu ron' s fir ing rate. Th is influence can be eithe r exci ta tory, increasi n g the gan g lion's firing ra te, o r inhibitory, d ecreasin g the gang lio n's firing rate. FURTHER DISCUSSION of receptiw fields can be found in Chapter 3 on pages 70-74.

cellular layers of the lateral genlculate nucleus. This layer Is knoW11 as the konlocellular layer. receptive field The region on the retina In whic h \/lsual stlmUll n fluence a neuron's firing rate.

\Vork on horseshoe crabs and frogs provid ed some of our earliest infom1ation on the receptive fields of retintil neurons (Ha rtline, 1940). But it w as Stephen Kuffl er vd10 first m apped out the receptive fields of individual retinal gan glion cells in the cat, using s mall spo ts of light (Kuffler, 1953). Flgure 2.14 illustrates Kuffier 1s main find in gs, which also apply to primate retina, and p rovides so me impo rtant insig h ts in to how the retina p rocesses ";stial info m1ati on. Kuffler 1s expe rimen ts a re sllnulated in Web Ac tivity 2.6: Ganglion Receptive Fields. Let's conside r Fig ure 2.14b. Kuffler 's visual s timulus was a srnall spot of light, w hich he m o ved about on the retina, tuming it on and off while recording impulses from a s ingle retinal gang lion cell \ Vhen the spot was placed on a specific s ma ll region of the retina, the ganglion cell increfl'Sed its firin g rate when the li ght W4lS turned on (this response is indicated by a plus s ign in the fi glU'e). This area o f the retina is called the /(center" of the gar1g lion celrs receptive field. \.Yhen the spot '""·as m oved to an a djacent area of the retina, the ganglion cell decreased its firing ra te w hen the light was turned on (indica ted by a minus sign). It is in teres ting tha t tuming the light off in th.is area s urrounding the recepti ve-fie ld renter led to a brief s mge in the cell' s firin g rate, after whid1 the cell settled d own to its spontan eous rate.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE FIRST STEPS IN VISION (a)

Reoord signal from optic nen•e

(b) ON-center ganglion cell

Light on

R.•poru• 11 11111111111111 I I I

Spot in center

(c)

FIGURE 2.14 Retinal ganglion 09!1 r909ptiv€1 fi8lds. (a} Mapping reoeptivg fields. fP) ON -center field. In each image on the left, the small white circle illustrates thEi region o n thQ r9tina whera thQ rc;itinal ganglion CQll increased its firin g rate whoo the spot (small yellow circle) was turn9d on. The largg gray c irc le illus trates the region on the retina where. the reitlnal ganglion ct:iH decreased its firing rate when the spot was turned on and incrQaSlQd its rate whQO th9 spot was turmtd o ff. The pbts on the right illustrate the spikes fired by the retlpal ganglion C911. tc) OFF-center field. In each lmag4':) on the left, the large Wille c irc le Illustrates thQ region on th9 retina v.tiere th9 retinal ganglion cell ilcreasQCi Its firing rate wt1'Wl the spot V\laS turned en The small gple around the wortd suffEr" from blinding diseases in v.tiich the rods and/or cones degenerate. These Include age-related macular degeneration (AMO) and retlnltls plgmentosa (RP). (See Web Essay 2.3: Clinical case: The Man Who Couldn't Read and Web Activity 2.7: Simulated Scotoma.) At present, there are no effective cures top-event the progressive degeneration of the photoreceptors that ocour in these diseases. For patients with AMD this may lead to an lnablllty to read or recognize faces. Fcr patients with long-standing RR this leads nevltably to lrreversltle blindness. Fortunately, there are several excltst-receptor reurons and their connections are largely Intact. One such approach is to substitute an electronic prosthesis (an artlficlBI devloo to repace or augment a missing or Impaired part of the bodYl Into the retina. Typi cally the prosthesis uses a camera to convert light Into energy; an array of electrodes Implanted in the retina generates an e-.ctr1cal stimulation pattern based on the light pattern on the camera and delivers this stimUlatlcn pattern to the Intact p;>st-receptor neurons (Figure 2 .18). Unfortunately, v.tille these retinal prostheses can restcre some sight, there are technical challenges to Implanting them, and they suffer low spatial resolution (Weiland, Che. and Huma'y1.Jn, 2011), allowing only perooptlon of spots of light end very high-contrast edges. Another approach that has had some earty success In animal models Is to use gere therapy to express Nght-actlvated channels lsmus A mlsaJlgnment of the two eyee such that a single object In

space Is Imaged on the fovea of one eye and on a nonfoveal area of the other

aye,

anlsometropia A oondttlm In which the two"""" have d ifferent refractive errcrs (e.g., me aye Is farsighted and the other

LLckily for Jane. her pediatrician found the cataract early, and the cataractous lens was surgically replaced by an artificlal lens l/'lhen she was 3 months old. The visual acuity in Jane's left eye just after the replacement lens was inserted was 20/ 1200. aboLrt four times worse than the no rmal value for a 3-month-old . But when tested ag ain 1 month later, acuity in her left eye had already begun to catch up vvith the acuity in her right "fe. In fact. a recent study of 28 infants (Maurer et al., 1999) found significant acuity improvements only an t1our after correcttve measures had been taken. Not all individuals ·with o::>rQenltal cataracts are as lucky as Jane. For example, in much of the third world , because of poverty, children b orn w tth congenital catarac ts (often in both "f9S) go untreated and grow up essentially blind. According t o the Wortd Health Organization , India Is home to the largest population of blird children in the wor1d. to track how these "blind" chllcten learn to see. These studies are Just be;Jlnning to provide important new Insights Into brain pl asticity. Congenital cataracts are not the mly cause of amb lyopla. Eal1y in life, t'wo other di sorders-strablsmus On vJhlch one f1'/0 Is turned so that it is receiving a view of the wortd from an abnonmal angle) and anlsometropla un w hich the two eyes have very different refractive errors; e. g. , one eye is farsighted and the other not)- may also cause amblyopla. These fonns of amblycpla are typically less severe , and often they have a later onset than congenital cataracts. The standard clinical treatm ent fo r aml:Jyop a, for over 250 years, 11as been to patch the good aye and ''l crce" the ambtyopic eye to work. This treatment is ordinarily perchildren (typically younger than 8 years). However, formed only in several recent studies suggest that there may be hope for recovery of visio n in okJer children. and even in adults tliro ug h "perceptual learning" - repeatoct practice of a d emanding visual task (Levi and Li. 20091 or p laying action video games (Li et al. , 20 11).

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

SPATIAL VISION

87

Summary 1. In this cha pter we followed the path of image p rocessing from the eyeball to the brain. Neurons in the cerebral cortex translate the array of activity signaled by retinal ganglio n cells into the beginnings of forms and patterns. The primary visual cortex is organized into thousands of tiny computers, each responsible for determini ng the orienta tion, w idth, color, and o ther characteri stics o f the s tripes in one small portion o f the visual field. In Chapter 4 we will continue this story by seeing how other parts of the brain combine the outputs from these minicomputers to produce a coherent representation . 2. Perha ps the mos t important feature of image processing is the remarkable transformation o f info rmation from the circular recep tive fields o f retinal ganglion cells to the elongated receptive fields of the cortex. 3. Cortical neurons are highly selective along a num ber of dimensions, including stimulus orientation,. size, direction of motion, and eye of origin. 4. Neu rons with s imilar preferences are often arranged in columns in prim ary visual cortex. 5. Seledi ve adaptation provides a powerful, n oninvasive tool for learning about s timulus speci ficity in human vision . 6. The htunan visual cortex contains pattern an..'1.lyzers that are specific to spatial frequ ency and orienta tion 7. Normal v isual d evelopment requires normal visual experience. Abno nnal visual experience early in life can cause massive changes in cortical physiology tha t result in a devastating and pemlal1ent loss of spa tial \.i sion.

8230336 Aml'la Appa

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Refer to the

Sensation and Perception Companirn Website

sltes.slnauer.comtwoWe4e for activrues, essays, study quesl ons, and othBr study aids.

CHAPTER

4

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Perceiving and Recognizing Objects from the eyes into the brain. By the e nd of Chapte r 3, we had readied primary visua l cortex (Vl , s tri ate cortex), w here we encountered cells that '"'·ere optimally stimulated by bars a nd g ratings of diffe rent o rlenrations. Of course, when you look at the world, you d o not see an a rray of bars and g ratingsi you see co herent o bjects and extended surfa ces. Moreover, you recognize speci fic o bjects even H they are odd objects, as in Figure 4.1). C hapter 4 continues our journey through the visua l system and considers how that visuTI as the what patlnvay. This p athway appears to be the locus for the explici t acts of object recogniti o n tha t a re of particular imp ortance in this chapter (U ngerlei der and Mishkin, 1982). As \o•le m ove d ovvn into the te mporal lobe, receptive fie lds get much bigger. As the pathway's name implies, idrn/ is in view seems m ore important than where it is. However, tho ugh it is a useful organizing principle, one sho uld n ot becom e too addicted to this whnt an d where distinction. For ins tance, some basic object information is represen ted in bo th pathways {Konen and Kas b1er, 2008), a nd som e where information is encoded in the tempora l lobe wha t pathway. Early evidence for a relati onship bernreen the tempo ral lobe and object recogniti on came fro m s tudi es in w h.id 1 large sections of the tempor.ctl lobe were dest royed (lesioned) in m onkeys. When Khiver and Bucy (1938, 1939) dJd this, they fo und that their monkeys behaved as thoug h they could see but did not know ¥•.' ha t they were seeing. This deficit, also seen in som e hu-

FIG URE 4.2 The m ain \'i sual areas o f tt1e macaque m onkey cortex. Humans have comparable visual areas (569 Figura 3. 1). Tha d rawing o f the brain has b99n d is to rted to show ar'*'6 that lie deep In the folds (sulCI) o f tha cortex. Each abbreviatio n refers to a. different visual area, not all o f which w ill be discussed here. Visual cortical i::rocesslng can be divided into two broad stn'.lams. O ne, heading for the parietal lcbe, can be thought of as being lntbfested in whGra things are . The other, heading down into the t€1mporal lo be, is concarned with wMtthings are. (After Par\Th'lt brnin activity achlally corresponds to your recognition of your gran d1nother. It is probably a bit extreme to imagine that this is the work of a single neuron in the lT cortex, and more recen t work has cas t d oubt on the idea tha t IT cells respond to an object independent of its position in the visual field (DiCa rloa nd Maunsell, 2003). Nevertheless, it does seem clear tha t cells in tha t part of the brain are critically involved in object recognition. 1l1e IT cortex maintains d ose connections w ith p arts of the brain in volved in mem ory fomlation (not.''1.bly the hipp ocampus). This is important because those IT cells need to learn th eir receptive-field properties. The recep ti ve-field properties of primary visual cortex could be \'\-Titten into the genetic code in some manner, but n e urons that respond to g randmothers dearly cannot be hardwired, since everyon e's grand m othe r is different. Nikos Logothetis and his coworkers d em onstrated that cell s in IT cortex ha ve p recise ly this type

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEIVING AND RECOGNIZING OBJECTS

FIGURE 4.6 Cells in the lnferotemp:>ral cortex of macaque monk9ys ar9 interested in very specific stimuli. In this case, too cell responds vigorously to a rnooksy faca and to som e othgr stimu fi that seQITI related. (After Gross. Rocha-Miranda. and Bender, 1072.)

11111111111111111

11111

111111111111111111111

11 1 11 111

© .

93

11111111111

i

1111111111 111 1111111 1111

I

I

11111

11111

11 11

Ill

1111

I

of plasticity (Logothetis, Pauls, and Poggio, 1995). A fter training monkeys to recognize novel objects, these researchers found IT neurons that responded with high firing rates to those objects, hut only w hen the objects were seen from viewpoints similar to those from which they had been leanled. Human cortex has areas that appear to be the equivalent of monkey IT cortex (the anatomy of human and macaque monkey brains is not identical, so we talk about homologous regions). One of the more amazing demonstrations of this fact comes from a 2())5 study by Quiroga et al. They made recordings from single cells in the temporal lobe of human observers. Normally we do not put electrodes into the human hrain, but these observers were patients being

prepared for brain surgery to treat epilepsy. Implanting electrodes was part of the treatment plan, and recording visual responses' from these cells involved

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

homologous regions

Brain regions

that appear to have the same furci.bn

In different species.

94

CHAPTER 4

JerutiferAn.i.s ton

Jennifer A.ni.ston

Jenni:kr Anlston and Brad Pitt

l..alU'a Linney

Stimulus presented

. .

lJAL ' ;

Spider

Sydney 01--.er,1 Hou se

I

I

I

I

!

!

lLL ' ; LlL '; ! ! I

I

I

I

!

!

I

I

I

I

Leaning Tower

8230336 Ann

., u ·, Lil ', i i i i i LlLU I

I

I

I

i

t

Stimulus ON

Ce ll's response

I

Stimulus

presented

I

Cell' s response

Stimulus OFF FIGURE 4. 7 R9sults of rgcording the activtty of on.e 001 in the kimporal lob9 of a human patient. This oell responded to pictures of the actress JEf'lnifer Alliston. The c:eill did not respond to pictures o f anything 91se, including thoS9 o f o tt"lQ'" actr9SSQS. CAfti;.r Quiroga et aJ ., 2005.)

no extra risk to the patient or interference \Vith treatment. In the experiment, the observer just looked at a collection of im ages \lvhile the activity of a cell was monitored . As Figure 4.7 sh ows, like the cells that G ross h ad found in monkey IT cortex, some of these cells turned out to have very specific tastes. The cell sh O\vn in the fi gure responded only to th e actress Jennifer Aniston and to nothing else presented to the obsen 1er. Other cells had preferen ces for other people, like former Presid ent Bill C linto n. One ce ll responded to the Sydney Opera Ho use; an othe r, to the Eiffel Tower and the l ea ning Tower of Pisa but not to other londmar ks (Quiroga et al., 2005). \Vhile we d o n ot have a lot of systematic d a ta on the responses of single cells in the human \oisual system , we have a growing volume of fun cti onal imagin g data that d ocuments areas in the huma n brain th a t appear be specialized for different sorts of stimuli . Conveniently, as s how n in Figure 4.4, ma ny o f these h ave been given names tha t ma ke th at proposed specialization dear. Thus, cells in the fu siform face area (FFA ) a re interested in faces (Kanwis her, McDerm o tt, and C hun , 1997) while an area like the parahippocampal place a rea (PPA ) has cells that resp o nd to s paces in the \Vorld like roo ms wi th furniture in them (Eps tein and Ka nw isher, 1998). T1'1e visual system has m an y differen t problems to solve-like the problem of face proces.sing-and it appears to have m odules that are specialized for worki ng on different problem s (Kanwisher and Dilks, 2013). Other e\o;dence

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEl\t1 NG AND RECOGNIZING OBJECTS

com es from brain lesions ca used by s tro kes o r o the r accid en ts. Like K.li.iver and Bucy 's m onkeys, humans w ith lesio ns in the tempo ral lo be often s how symptom s of agn osi';:i,, the ability with o ut the ability to know w ha t is be ing seen. Som etimes these a gn osias can be quite s p eci fi c. Prosop agno.sia, an inability to recognize faces, is one exa mple that h as receive d a lo t of study

and that w ill be disc us.sed at the end of thi s chapter (Dam asio, Oa masio, and Va n H oesen, 1982). There are other interes ting subdivi s ions of agnosia as welt s uch as the ability to recognize anirnate objects (e.g., animals) but not inanimate objects (e.g., tools) (Newcombe ilnd de Ha:m , 1994 ). The implicati ons of specific agnosi:as in s p ecific pa tients sh ould no t be taken too far. While 3reas s pecia lized for faces o r place.s seern to be readily d ocum ented in humans and you mig ht have a reas specialize d for some o ther ca tegories of objects, like tools (Hutchison e t a l., 2014), you p roba bly d o n ot have a separa te area for each and e-very ca tegor y o f object tha t yo u can recognize (Kan wisher and Dilks, 2013). Some p rocessing that lead s to the catego ri zation of objects and scenes can be very fas t. Electri cal ac ti vity from the brain can be recorde d froin e lectrodes placed on the scalp. If we flash a p ic ture to an observer and a s k whether it contains a n animal, we ca n record a s ig na l in the observer that reliably differe ntia tes an im al from nonanimal scenes w ithin 150 milli seconds (ms) from the onset of the s timulus Fize, and tvlar:lot, 1996). That's fas t en ough to mean that there cannot be a lo t of feed back fr om higher visual or m emory processes, suggesting that it must be possible to do sorne rou gh object recognition on the basis of the first wave of ac tivity as it m oves, cell by cel l, synapse by syna pse, from retina to stria te cortex to extras tri::.lte cortex and beyond . Tha t feed-forward process mus t be able to gen erate an "an.im a1" s ignal from a w ide rro, ge of animals in different positions, s izes, and so on (Serre, Oliva, ond Poggio, 2007). To summari ze, two path,-vaysernerge from visual cortex. 11,ewlierepathway "vill be taken up in late r chapters. The -whnt pathway m oves through a s uccession of stages, building a rep resen ta tion of your g randm othe r or the Eiffel Tower out of the very .specific, very locali zed s pots, lines, and ba rs that interest the cells in the retina, lateral geniculate nucleus (LG N), and primary v isual cortex. That' s a but d on 't be lu1led into thinking we fully unders tand the p rocess of object recognition si mply because we ha ve som e n oti on of the ne tual pa th ways involved. It is a really difficult p roblem, and in the rest of the ch apter we will illus trate wh y this problem is ha rd and how the visu al system. attempts to sol ...-e different parts of it.

The Problems of Perceiving and Recognizing Objects Figure 4.8a is d ea rly a picture of a ho use. The imag e may contain som e o ther things as well, but the chief object of in teres t is the house. It is p retty clear that Figure 4.8b also sJ, ows a h ouse, perhaps with a s imilar shape a nd chh1u,ey, though in this case the h ouse is pa rt of a m ore abs tract scen e painted by the Frend ' artis t Georges Braque in the early twentieth century. Figures a re two more pictures of h ouses. It is quite d ear tha t Figure 4..Bc is the sam e ho use as Figure 4.811, w hile Fig ure 4.Sd is not. These seemingl y s imple acts of object identification require a lot from the wJrnt pathway, and they cons titute the main topic of the res t of this chapter. Like m any seemin gly s imple acts, object percepti on is actuall y a collecti on of complex and remarkable accompli shments.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

95

process A process that carries out a computation (e.g .. cb)ect reccgnltlon) one neural step after another 1 without need fa feedback frcm a later stage to an eanler stage.

0336 AMmJ App.l

96

CHAPTER 4

FIGURE 4.8 The problem of object racognition . ThQ house and the oi l painting by Goorges Braque in f/:J} look dlffQ!"&nt, but they are both c lOOTIY houses. {c) This irnage is also vety different from the lmag13 in (a). but they ae both cl90.t1y th9 $i.llTl9 houSQ. (d) This image of a red h ouse cleart-,t shows a hoUS9 that is diffSfent from (8) and (C) .

Consid er Fig ure 4.9a. It s hows yet another h ouse. How do we recognize it as a house? First "'ie nee...1 to gather some bclSic visual features. The preceding chapters showed us how single cells in the early visua l system respond to stimuli suc h as simple lines. In Figure 4.9b each circle is a cartoon of the receptive field of a sllnple corticn l cell. Those cells in striate cortex w ill respond well to the high-contrast lines in the outline of the house. Th(lt limited recepti ve field is like a w indow that allows the cell to "see'' only a small part of the world . None of these simple cells see a house. Tl,ey jus t collect local feahtres like horizo nta l, vertical, and oblique lines. At the very least, the local fea tures''\\dll a house. 1his p rocess could be frnagined as the natural extension of a process \Ve have already discussed. One way to 11 construct" the lines detected by simp le cells

(a)

ODDD CD

DD

FIGURE 4.9 The prot:Xem continued. (a) Another house. (b) Cells in primary visual cortQX raspcnd well to the local fQ.ilturoo (circ led} of the housQ. But how do we go from r.accgnizing a collection of local features t o perceMng a house? (c) In this sligt1tty m ore oornplicatad sCGr10, how do we know ..,.,hich bits belong t ogeth19r?

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEIVING AND RECOGNIZING OBJECTS

is to imagine combining a row of d ots det cteab_y gapglioITcellSi. Dots could be grouped together into lines. A pair of lines, each collected by a different cell, could be combined by another ceU tha t would sense a This process could go on a nd o n tmtil we had a h ouse. On closer examination, however, it is dear that this process would not be easy. Consider, for example, Figure 4.9c. Agai.n, sitnple cells would luwe no problem detecting the lines and edges in this scene, bu t how do \-Ve know which edges go \o\ith w hich objects? How d o we avoid considering the s now man part of the h ouse? The h ou.se a nd s n owman m ake comers, as d o the h ouse and its d oor a nd s teps. The car and ho u se a re separa te. None of thei r lines toudl . But this is also true of the outline o f the ho use and its \o\-indo\.vs.

C learly we must have processes that s uccessfull y combin e features into objects. That is one of the tBsks that defines middle (or mldlevel) vision (as opposed to low-level vision, whidl was the topic of Chapters 2 and 3). The next part of this cha pter looks at sorne of the processes of middle vision. 11-ie followin g part takes the fea ture combin a tions given to us by middJ e-level vision and asks h ow we come to know w h.:1t th e object is. The act of recogniti on must involve matching \vha t we perceive now toa memory of somethin g we perceived in the past. H ow ca n we do that? For exam ple, how d o we knmv that all of the images in Figures4.8and 4.9show houses? Witho ut having seen these exact objects before, how d o we go about placing them in the "house" category? Furthermore, h ow d o we know tha t Figu re 4.& and c portray the same house, g iven tha t the images de livered to our retin as in the n"-o figures are radicaJly different? ln the fi na l section of this chapter we..-U disc uss the high-level visual processes that enable us to recognize familiar objects, nov·el views of familiar objects, and new instances of farniliar object ca tegories.

Middle Vision The goa l of middle vision is to orgartlze the elernents of a visual scene into groups that we can then recognize as objects. Let's begin with the .simples t case of an object isolated on a simple background. Finding the edge,s of this object w ill be a good starting place on the road to recogn izin g the object.

Rnding Edges In Chapter 3 we discussed a t length how n eurons in striate cortex can detect bits of lines, but how d o '"'·e decide which bits belong to wh ich objects? We al ready established that '"'e can' t just g roup a ll the edges tha t touch each other into an object. Because objects a but and overlap other objects, simple connected ness will n o t work . Worse yet, before we can concern ourselves with grouping edges, we n eed to worry about the quality of the raw edge in fo rma tion. In Figure 4 .10, part of the house of Fig ure 4.9 is red u ced to a simple, arrow-shaped outline. It is easy to see th at arrow. Notice, h owever, that in som e places the object is lighter than the background, while in other places it is darker. This means that if we trace the e dge of the object \vi th a finger, we m11st pass through locations w here there is n o differen ce between the luminanc-e of the object and the h uninance of the background. In other words, a t these points the shape has n o edge a t all.

FIGURE 4 .10 In some plaoes this ob)act is darker than the background. In o th.:4" plaoes it Is lighter. If the c hanges an:i oontinuous, it follows that th9f9 must b9 plac9S 'Wher9 the edge of this shape simply disappears, even if you see 9dgee as

continuous.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

97

middle (midevel) vision A defined stage of ,;sua1 prooesslng tmt comes after basic features have been cr extracted from the Image \'lsloril and before dJject recognition ard scene un:Jeretandlri;i (highlevel vlslonl.

98 (a)

CHAPTER 4

(b)

(c)

FIGURE 4. 11 (a) The Mfind Qdges" function in a popular graphk:::s program finds gaps in tt1e bordB""s of the image in Agure 4 . 10 - gaps that we do not see. (b) Whffi a o::>mputB"" algorithm loolAng that complex objects or perceptions could be understoo:J by analysis of the components.

Lnterestingly, this oo:asional lack of an edge d oesn' t seem to bother our visual .system a t all In fact, it m ay be hard to see the gap. Asking sirnpl e computer grnphics sofhvare to find the edges in Figure 4.10 wo uld yield som ethin g like Fagure 4.11a . 11'\e visual system kno\vs tha t the gaps are acd den ts of the lig hting and fills in the contour. 111e computer is n ot so clever. Fig ures 4.11 b,c s how an other example. In Figure 4..1 lc, no tire that the ed ge-finding software find s all sorts of edges. The human visual system , however, is d oing some thing quite In the early s tages of processing, it is figuring out w hich ed ges m ark the bound aries of objects a nd w h ich represent s urfa ce fea hues (like cracks on the rock faces). All these differen t bits of information are then combin ed to ma ke the syste m's best g uess about the presence of a contour. ll-1e inferential nature of con tom perception ca n be nppreciated in the m ore extreme demonstration sho""TI in R g ure 4.12. lhis is an example of a "Kanizsa figure"-named after Gaetan o Kanizsa (1913-1 993t a n Ita lian psychologist who spent m any years investigating s uch stimuli. Here it is still e1sy to see the arrow outline, even th ough the vast majority of the shape's lines are m issing'! Check-it yourself. There really is n o border between the white fig u re and the white background . These edges, called Illusory contours, are perceived because they are the best guess abou t w hat is happening in the world a t that loca tion. It rea lly does seem likely that a con tom is present, even if there is no physica.I evidem:.-e that location.

RULES OF EVIDENCE l et's consid er the sorts of regularities in the s timulus tha t are bk.en as evidence for a contour in the world , be it a real conto ur or a good guess like those made in Figu re 4J2. TI1is tendency of the visua l system to make inferential leaps was problenrnti.c for on e of the ea rliest g ro ups of perceptual the structuralists. Stn1ctma lists sud1 as Wilhe lm

FIGURE 4.1 2 This .. houS9" outline is constructed from illusory contours . Even though the contour is doo11y visible , there is no piys1ca1 differsnce betw.ae.n the w hite background and the vvtli\Q house.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEl\t1NG AN D RECOGNIZING OBJECTS

Wundt (1832-1 920) and Edward Bradford Titchener (1867-1927) argued that perceptions a re the s um of atoms of sensatio n-bi ts of colo r, orienta tion 1 and so forth . In the structuralist view, perception is built up of local sensa tio ns the way a crysta l might be built up of an arrny of atoms. A n i1lusory conto ur challe nges th is view because an extend ed edge is seen brid g ing a gap wh ere no local atom of "edgen ess" can be fo1.md. Over time, it became clear that there are many examples w here the s tructuralist a rgmnent seems to fail. Inspired by th ese exampl es, a second group of

psychologists, led by Max Wertheimer (1880-1943), \ Volfgang Kohler (1887-1 967), and Kurt Koffka (1886-1941), fo rmed the Gestalt school (Wagemans e t a l., 2012). Gestalt theory he ld tha t the perceptual w hole is more than the s urn of its sensory parts. Perhaps the most enduring contributio n of this school was to begin the description o f a set of organizin g principles (some times known as Gestalt grouping rules) tha t d escribe the visual syste m 's interpretation of the raw retin al image. In the next sections, we will d escribe some of those rules. More than the specific rules, it is important to remember the overarching goal. The vis ual syste m is trying to make sense of the vast and often a mbiguous and noisy inputs from the early st.'lge of visua l processing. These rules are useful parts of that effort beca u se they reflect regula.rities in the world. l11ey allo\v the visu.'.ll system to So:'l)i "H the input looks like tha t, 1 ca n infer that this is the state of the visual '"'·odd." With tha t in ntlnd, consid er Figure 4 .13a, w he re we h::we a collection of short line segments. O ne se t of these seen'\S to form a contour, an irregular, dented loop. The " rule,'' cartooned in Fig ure 4.13b, is th a t we ten d to see simiJarly oriented lines as part of the same contour (D. J. Field, Hayes, and Hess, 1992). Pola t and Sagi (19Q3) measured how pieces of a contour s upport each o ther. The effect w ill be hard to see in print, but the fainter lines in the (a)

99

Gestalt In German. In refe re nce to perception, a schocil of thought stressing that the percep-

tual whole could be greate r than the apparent sum of the parts.

Gestalt grouping rules A set of rules describing wrrlch elements In an Image will appear to group togetl1€f . The origh1al list was assembled by members o f the Gestatt school of

thought.

(b) \Vhkh gray line is a likely oonlinuation of the black line?

Not l: _\/ /y-------' /

(c)

(d) Very likely

8230336

f'IGU R E 4.1 3 Contour completion. (a) The roughly c irc ular contour saen in th9 cent9r o 1 this diagram r9sults from the visual system's applic:ttlon o f the rule shown in (p). (c) Polat and Sagi {1993) found that it was QaSiQr to SQQ a faJr1t set o f bars if they were flanked by bars of the sarll9 oriantatbn. W. S. Getsler and J. S. Perry actuall'/ measlJred the relationships beitween qeart?y fine segments In natural SCQOQS. Tho figurJ shows the probability that a horizontal tine SQQn')Qnt 'WOUid CO'"()CCur with othGr line S9Qm9flts in the Image

at different distances and orientations. ill9 most hK91y pars are ooM n9lr or nearly so. (Part c aftQf Polat and Sagi. Very unlikely

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

lQOO; part dafter W. S. Geisler and P'1fty. 2009.)

100 (•)

CHAPTER 4

x x x This...

(d)

Btit this ..

DJ DJ DJ

(I>) .. looks more like this ..

(e) .. looks less like this ..

(r)

(j)

.. . not like this .

FIGURE 4.14 The Gestalt princip les o f goocl continuatiOn and closure: Undt< the Gestalt princ iple of good continuation, pr·e99nted with an image like the one in '8). th9 brain tends to int9rpret It as a pair of intersecting lines. as in (b), rather than reaching any o ther conclusion, suc h as that in (c). But th9 principle o f c losLira can trump good continuation. as shovvn in (d-1). ig) Laura Williams c leverly exploits good continuation by lining up the edge of tM fiek::l w ith the reflected E!dge In the mirror. lhis allows the good oontinuatlon cut1 to trump the ocduslon cue (S99 next section) to produCQ this striking image.

.. .thanlike this .

midd le of the trip lets of lines in Ftgu re 4.13c will be easier to see w hen they are collin ear \vi th the flankin g lin es. ll1ose flankers provide evidence for lines of the sam e orientation in be tween. If a set of lines fo rms a closed s hape like the roughlycircu1u con tom in Figure4,13n, then the sh ort segments support ead1 o ther even m ore s tron gly (Kovacs and Julesz, 1993). Geisler and Perry (2009) d ocumented the regtJ arity in the world that s uppo rts the rule shmvn in Figure 4.13b. They labeled many, man y contours in na hmd scenes. Then they examined pairs of contour pieces and asked , 11 What is the cha nce th at tliis piece lrere at this orienta ti on is o f the sam e contour as that piece tire re at thaf orienk1tion ?'' TI1ese like lihoods are colo r-coded in R g ure 4.13d, and they s h ow that in the real world, if h¥o contour elements are dose to co linear, they are likely to come from the sam e contour. Sharp turns are much rarer. 11,eGestaltists called this the priru:iple of good continuation (Figure 4.14), and they illustra ted it \Vith examples like the one sh O\-vn in Figure 4.14a. \'Ve tend to see this figure a pair of intersecting lines (Fig ure 4.14b). There are many other p ossible orga nizations (Figtue 4.14.c, for example), bu t all else beins eqrml, the tenden cy to see lines continuing in the sam e direction-in other words, the tenden cy to group edges that have the sam e orientation-supports the X-like interpretation ill us trated in Figure 4.13b. TI1e 1' all else being equar' phrase is imp ortant. As the Gesta lt psychologists knew, and as we \Vill see, a host o f rules,. principles, and good guesses contribute to our organized perception of the world. These seem to operate according to a sort of committee model Everyone gets together and voices o pinions about how the s timulus ou gh t to be unders tood . Fo r example, look a t Figme 4.14d. \'\Then we make the contours of Figure 4.1411 part of closed forms, good continuation is at odds with the principle of closure. Now, the sam e intersecting lin es are interpreted differently. In this case clos ure is stronger than goo d. continua tion. 11-ie s upremacy of closure over good continuation is not absolute. \ Vhen the opini ons of different 11comi:nittee m e1nbers" the results can be somewhat unpredictable. Somehow, though, a consensus vielv a lmost a lways quickly emerges and we PERCEPTUAL "COMMITIEES"

good continuation A Gestalt grouping rule stating that two ements ,,,;11 tend to group together If they seem to Ila on the same contour. clostre In reference to perception, closure Is the name of a Gestalt pnnciple that holds that a closecl contour Is preferrecl to an cpen contour.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEIVING AND RECOGNIZING OBJECTS

(a)

FIGURE 4. 15 The making o f illusory contours. The little arrCJWS In fJ:>) and (f:I} repr9Seot the visual system's best guess about what is going on in {a) and (c). The illusory disk in (c) aris4:ls when the v isuru system combines a whole collec tion of guasses about the line rninations . as shown in {d).

(b)

(c)

101

(d)

" /

/

'}>..1.•.4'(

-:

"

;-.(

.-

'" '

settle on a sing le of the visua l scene. TI1is committee me ta phor is \·ery useful and 'vill recur through out this chapter. OCCLUSION Lf an e d ge suddenly s tops in an image, tol1y does tha t e d ge s top ? One reasonable gu ess mi ght be tha t it s tops because som e thing else gets in the wa}i hiding it from our vievv. So, returning to the Kanizs.a figure of Figure 4.12, we can the visual sys tem asking w hy the vertic.J.l line

at the bottom of the figure suddenly s tops. TI1e answer tha t the visual syste m seems to come up v..rith is tha t there is a nother conto ur occl11di11s the vertical

line, with the ocduding edge oriented perpendicularly to the occluded edge. llUsgue.ss, combined with a guess that the notches in the circles represent contours tha t can be extended , lead s to the inference of an illusory contour (Fig ure 4 .1 Sa,b ). (Note that we also interpret the notches as places where one object is occluding another.) Rgure 4 .15c sh mvs another example. II eadl black line genera ted weak illusory contours at right an gles to its e ndpoint, then this figure would contain a set of ro ughly coli.near line segmen ts (Figure 4.1 5d). Those segments could be grouped toge the r to form a con vincing ci rcle in the \vay that the line segmen ts formed a loop in Figure 4,13a. (By the way, illusory con tom s li ke this rnake exce llent doodles when you would rathe r be d oing informal p erception exp erim e nts than paying attention.)

Texture Segmentation and Grouping Connecting s hort line segments w ill get u s only so far in di";ding the raw image into objects. On the left in Figure4.16, you can find a region that differs in the size o f its e lem ents. An edge detector like the one described. in Figure 4.5 will be obliv ious to the difference beh.veen the large r-grain and s malle r-grain textures. lnstead, the regio n is found by the visual system' s sophistica ted

mechanisms for texture segmentation (Beck, 1982; Beri;en and Adelson, 1988; Malik and erona, 19'/0/- Qn the r;iglit-hand side o f Figure 4.16, the texture of the water looks different from the t ture of the stone surround ing it. Here, ra thel' th."ln . deciding w hich feahires go together, the visual system m ay look

FIG URE 4. 16 The problem of texture segmentation. How do we distinguish the two halves o fth9 11gure? How do we Si9gment out the smaller regions inside each halt? On the bottom, hO'W might we know that the large area is stone, surrounding a smaller area of 'Water?

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

texture segmentation

Carving an

Image lnto regions of oomrnon texture properties.

102

CHAPTER 4

FIGURE4.17 ltis easy tos99 the approximate average orientation of the tines in the two regions of this figure. The red lines arie vertical ori average and the blue lines, horizontal. How9V€SQyou see thgsefour squares. \M!at is the state o f the \NOr1d that produced this vieJW? ft>) Ma)'be the eye just happens to b9 in the right position t o see four arbitrary shap9S at arbitrary d9pths lin9 up t o form the pattern. That would be quite a coincidence. That V¥'0Uld be w hat is called an ..accidental viewpoint."

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

105

The Duck-Rabbit MILK STOUT

FIGURE 4.24 A d asslc ambiguous figure gets a ngw rolQ.

accidental viewpoint

A '1ewlng

position that produces some regu· larlty In the visual Image that Is not prooent in the world (e.g., the sldoo of two Independent objects lining up perfecttf).

7'..... ,,.::-·-·-·--."·"/ -"L _ ----· · ---- +

106

CHAPTER 4

in Figure 4.12. Another committee consid ers all the possibilities a nd d eval ues any tha t involve accidental vievvpoin ts,.redudng w hat is initiaUy a unsolvable problem (finding the one rorrect interpretation out of the infini te number of possible ones) to a potentially solvable one. And so we p roceed htrther and further into the visual system, until a si ngle, gen erally correct interpre tation emerges.

Figure and Ground

FIG URE 4.26 What is fi gure and what is ground and why?

figure-ground assignment The process of determining that some regbns of an Image belong to a foreground object (figure) and other regbns are part of the background (ground).

surroundedness A rule fer figureground assignment stating that If ooe reg bi Is entlray surrcunded by another. It Is llkay that the surroo1ded regbn Is the figure.

Armed with this understanding of the g r01.md m.les of oomm_ittee d eliberation, we continue with the effort to go from simple features to recogni zable objects. The edge- and region-finding medrnnism s, discussed earlier, would divide Flgure 4.26in to yellow and red regions wi th out much difficulty. But how s h ould those regions be und erstood? It is extremely likely that \;sual =tex In humans tllat Is speclflcally and reliably activated by Images of the body other than the face.

114

CHAPTER 4

Templates versus Strnctural Descriptions The ide'-1 that we recognize objects by matching every pixel or even matching every low-level feature of the input to a representation in

mem ory is what might he called naive template theory. lt won' t

FIG URE 4.40

A basic template.

A -4A

A

a aA a FIG URE 4.41 Th9 probleim with ternplatoo Is that we need a lot of th€m .

na'fve tempiate theory The proDC'3•1 that the visual system recognizes cbjects by matchlr>J the neural representation of the •nagewith a stored representation of the same

work even for si mple objects. The ba:sic idea of a template is rather like a lock a nd key. As we will see in Chapter 14 and 15, the Jockand-key meta phor is quite apt in s mell a nd ta ste, \vhere the ''key'' to be recognized is a molecule like a specific od orant and where that m olecular key presents itself in mo re o r less the same shap e every tim.e. For instance, look at the "shape--pattem " theory o f olfaction in Chapter 14. Figures 4.40 a nd 4.41 show tha t recognizable visual objects like the letter A are much less well behaved. Figure 4..40 illustrates the lock-and-key template idea. A letter A stimulus falls on an array of spot d etectors. If the A falls on the filled d etectors and not the empty on es, it is identified as an A. Therefore, the array of detectors serves as an A template. The difficulty w ith a nalve template m odel is that too many templates a re required.. All of the objects in Figure 4.41 are As. If \Ve need ed a new te mplate for every lette r in every position and orienta tion, we would nm out of brain before w e ran o ut of alphabet. A11d just think about the p roblem of building a Jennifer Aniston rep resentation i n th is way. One way out of this problem is to no tice tha t all the As--a t least all the capitalAs--share abasics tmcture. Cnstead of ma tching each p oint in the image to 3 point in a te mplate, perhaps we perform 3 more concephMl match. Just be described by the relationship of its three lin es: the about any capital A two flankin g lines me-et, and the third line spans the a ngle cre.:ited by those two lines. Now the image of the A is being m atd 1ed to a structural description of an A, a s pecification of an obiect in tem '\S of its parts and the relationships between the parts. lvlany versions of s truch.ual'1escription hyp otheses have been proposed. A key component of each theory is how object parts are re presented in the s tructum l descripti ons. Marr and Nishihara (1971) used "generalized cylind ers'' that could be scaled longer,. s horter, fatter, o r thinner to rep resent different-shaped parts. Bied errnan (1987) proposed a set of geons ("geometr ic ions''); Figure 4.42 provides exarnples. Geans are specified as collections o f n onacddenta l feah.ues (remember accidental features in Figlue 4.25?), so in theory a visual system s hould be able to recognize a geon equally acctrra tely and quickly, regard less o f how the geon is oriented in space (as long as it's not an accidental view}. Geons are fea h.rred in model, whid1 we ll now describe in a bit more detai l.

In

the brain.

structll'al description A description of an object In terms of the nature of Its constituent parts and the relations.hips between those parts. geon

In Blederman's recognition-by-

components model, any of the +lgeornetrlc Ions· rut of wh lch perceptual ctijects are built .

recognition-by.components model Blederman's model of obje:;t recognition, which holds that objects are recognized by the ldentrues and relationships of their component parts.

FIG UR E4.42 111reeofthe36orso geons in recognition-byc omponents model of object recog nitio n. Each geion is s hown in three orientatb n s, Illustrating the fact that thgy are .:iasily reoognizabkj frorry ·ly. any viewpoint.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEIVING AND RECOGNIZING OBJECTS

115

FIGURE 4.43 Combining geons can c reate a v..ide variety of object represenkillons. (a-c} Three objects •spelled out• by different combinations of the cylinder, noodle, and brCk goons. A d iffer911t vi9w of the coff&e mug (d) and a different instance of this ob)gct class (e) 'NOuld lsoct to th9 same strtciural description re in (a).

Just as the finite set of le tters in the a lphabet ca n be combin ed in vari ous ways to produce an infinite nmn.be.r of words, the finite set of geons can be used to construct a very large numbe r of object representa tions . Attaching a noodle to the side of a cylinder gives us a coffee mug (Figure 4.43a). Moving the noodle to the top of the cy linder produces d pail (Figure 4.43b). Changi ng the cylinder to a brick gives us an a ttache case (Figure 4.43c). Because geonsa nd relationships s uch as'' A is beside B" are designed to be eJect recogni tion . In response to these questions about structurc1l-description models, researchers s uch as lvlich ael Tarr1 Heinrich BUJ.thoff, and others returned to template-like representations tha t they cal l ",,; ews." Support for view-based representa tion s generally comes from experime n ts that use n ovel, ra ther than familiar, objects. For e..xa mple, Tarr and Pinke r (1990) tra ined observers to recognize lette rlike objects. During tralning, the objects were always s hown in the upright orientation . 1llen, in a "surprise" phase of the experi ment, the observers were asked to recognize rotated versions of the objects. 11le res ults revealed a linear relati on ship between o rienta tion s hift and the respo nse time to recognize the object. The mme the object was rotated, the longer the observel's took to name it. This result s u ggests tha t the subjects may have s tored a te mplate-like representation of the object during the training phase, and then recognized. the objects in the s mprise phase by me ntally rotating the m.isoriented objects b£tck to the upright views they h.:1d s tored in m em ory. Objects that can be identified on the basis of their geons can stlll sh ow a dependence on vie,,..-point (Hayward and Williams, 2000). Ga uthier e t al.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

viewpoint Invariance of an object that does

1. A property

not change

when otserver \!1ewpotnt changes. 2. A class of theories of object recog-

nition that proposes r€1)resentatlons of objects that do rot c hanga when viewpoint changes.

116

CHAPTER 4

•••• ••• ,

1111

FIGURE 4.44

R9cognizing thQSB •gr99bl9s" dOQS not SQQ'Tl to be lndeper1dEnt, even If they are made up ofgoonlik.e shap8S. (Fran Gauthi9,r 9t al., 1QQ8.)

(1998) mad e novel objects that they named "greebles'' out of geonlike parts (Figure 4 .44). Recognition of these novel objects also appears to have a vie'"•point-dependentoomponent (Web Activity 4 .S: Viewpoint Effects lets you test yourself in a s hort experiment us ing s uch objects). H owever, view-based m odels have thei r mvn problem.s. While d e bate be tween structmaldescription and view-based theorists has some times been quite lively (Peterson and Rhodes, 2003), '""can imagine that a tndy s uccessful model of object recogni tion w ill incorporate &-peci:s of both kinds of theory.

Multiple Recognition Committees?

lf

d1apter has had one theme, it is tha t no s tep on the road fr om image to recognition is taken by a s ingle process actin g alone. Prom the groupin g of similar pi eces of the image to the segrega tion of figme £1. nd grow1d , every s tep has been based on consens us--a committee decision.. 1t seem s likely that the £1.d o f recognition is simUar. Indeed, recognition may not be a sing le act. We can recognize a n object in multiple ways, perhaps simulta neous ly. As an example, Figure 4.45 shows h vo birds. In the termin ology of Pierre Jolicoeur (a)

FIGURE 4.45

(b)

Two quik1 different

birds (a and b). a yet more diffB'ent animal (c), and a third Mbird' (cl}.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEIVING AND RECOGNIZING OBJECTS

and hi s colleagues Qolicoeur, Gluck, a nd Kosslyn, 1984), 11bird' 1 i.s the enby# level category for these o bjects--the firs t word that com es to mind w he n we're asked to n.:"lme them . But these objects are also quite clearly different At a s ubordinate level -a m ore specific level ben ea th the e ntry level-Figure 4.4511 s hows a fox s pa rrow and Figure 4 .45b shows a cardinal. At a s uperordinate level-a broader level above the entry level-these h¥o objects, as well as the one in Figure 4.45c, a re all a nima ls. \Ve can imagine that each of these acts of recogni tion ("fox s parrow," "bird ," ''a nimal") rnight rely on different stored representations and different analyses of the visual stimulus. And we can imagine a geon accotu1t of "bird/' but a geon d escripti on of "fox sparrow" would be harder to en vision. recognition seem s bette r suited to a system in which the p recise de tails of a particular view of the object are encoded . Recognizing the objects in Figure 4A5a--c as animals seem s like qui te a different act. lndeed, a number of lud'iC.s }lave .!t.htn-...11 that it takes considerably longer to recognize objects a t thesubordinllte or superordinate levels than at the en try level. Even more imp ressively, studies looking at brain recordings provide strong evidence tha t different parts of the bram are more active w hen people are engaged in s ubordinate-level recognition than w hen they are recognizing objects a t the entry level (Pa lmeri and Gauthier, 2004; Rich.ler and Palmieri, 2014). There a !'e several m ore interesting hvists. For exa mple, wh en s hown an a typical member of a category, s uch as the bird in Figure4.45d, people are faster to name the object a t a m ore s ubordinn te level ("os trich'1). And when people becom e experts at recognizing a certain class of objects, s ubordinltelevel recognition becomes as fas t as or fas ter than en try-level recognition (or, looking at this a nother wny, we might say that the entry level shifts d ow n one level when one becomes an exper t). (By the way, bird-watchers and dog show judges seem to be the most ex tensively studied s uch experts {Tanaka a nd Tay lor, 19911-) All these results point to the conclusion that the visual sys te m is e mpl oyin g several different object recogni tion committees, each working w ith its own .sets of tools, at the same time.

117

en1ry-level category For an object, the label that comes to mind most It (e.g., 'bird"). qulcklywhen we At the subordinate level, the object might be more spoomca11y named at the supero rdinate (e.g., level, It might be more generally named (e.g .. 'animal").

Faces: An Illustrative Special Case Faces are an interesting special case of object recognition. Take, for exam ple, the images in Figure 4.46. You sh ould have Little difficulty recognizing these as faces, but you w ill probably have some difficulty quickly identifying \.vhid1 one has been mo dified . lf you turn the faces right si de up, however, one o f the pictures will look quite strikingly '"-'rong (Thompson, 1980).

FIGURE 4.46 Whic h o f these two photos has been altered? As you will see, both faces look fine do"A'n, but o ne of them looks quite strikingty" when turned right side up. (After Thompson, 1QBO.)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

118

CHAPTER 4

prosopagnosia ognize faces.

An lnablllty to rec-

double dissociation The phencm -

errn In which one of two !Unctions. ard l.Jice versa.

ti!'

congenital prosopagnosla A f0011 of 'face blindness" apparenlly present fran birth, as opposed to •acqtirecJ prosopagnosla, • whiCh wa.11cl typically be the result of an injury to the navous system .

lltis is another demonstration of the p oi nt m a de in the previ ous section: that d ifferent levels of rntegorization often seem to besubserved by d ifferent con:mUttees using different types of infom1ation. The p rocesses that recognize a face asa face care little about inversion, and they don' t seern to be d isturbed bpd,jStortio,n of the face. The processes that recognize the face as belonging to a specific individual (in this case, one o f the a uthors of this book) work poorly o n inve rted fores a nd are very con ce rned '\.v ith the precise configuration of eyes, nose, and mouth. (See Web Activity 4.6: The Face Inversion Effect.)

Neuropsychology pro\>ides further evidence that the processes can be separated . Damage to specific areas in the temporal lobe o f the brain can p roduce prosopagnosla, a d isorder in w hich the cannot identify faces. (See Web Essay 4.5: Face Blindness.) Though she m ay be abl e to recognize an object as a face, s he will not know w ho the person might be. TIU.s is n ot to say that the two levels of face recognition are s ubserved by completely different systems (Gauthier, Behrmann, and Tarr, 1999). A neu ropsych ological mark of truly separa te brain mo dules is double dissociation. Two functions, s uch as bearing and sight, are d oubly dissociable if one can be damaged wi thout harm to the othe r a nd vice \·ersa. Thus, you can be blind a nd s till hear or you can be deaf and s till see. In foce recogni tion, it is possible to lose the s ubordinate ability to recogaize specific faces w hil e retaining the ability to recognize an object a fuce_ ft is n ot clear tha t the reverse even makes sense. How could you recognize a foc-e as yom m other without recogn izing it as a face? Interestingly, it is possible to be born with a specific impairment in the ability to recognize faces . The exis tence o f this congenital prosopagnosla is a good in d ication that there is a specific neural mo du le fo r fa ce recogni ti on (Beh rma nn and Avidan, 2005). You may think you're suffering from this disorder w hen you fail to m atch a name to a face. However, tha t is a much more cornm on failure of memory. You can recognize the focei you just can't remember the na me tha t goes wi th it. A prosopagnosic would not kn ow that this particular face """ru; famlHar '"'· h_ile another one was not.

The Pathway Runs in Both Directions: Feedback and Reentrant Processing We've been discussing object recognition as a progression of steps down th e what from Vl into the temporal lobe, an d as mentioned earlier, it d oes seem to be the case that you can perform some acts of object categorization in this feed-fonvMd m anner (Serre, Oliva, and Poggio, 2007). 111at said , it is important to recognize that neura l processin g i.s n othing li ke a one-way street. Most of th e routine acts of object recogni tion that you perform w ill in volve information flowin g up and d ow n these pathways. All of the connections in Figure 4.3 run both ways. Consider that receptive fields get bigger and bigger as you ascend the what pathway. TI1e cells care less and less about where an object is in the visual field or how it is oriented.. They become increasingly interested in the object's identity as, perh aps, the Eiffel Tower. However, you know that the Eiffel Tower is right there and it has that red light in thnt s pot. This precision is p robably achieved by going back do,,..,·n the patlnvay, once you have some in forma tion about the object, and interrogating earlier visua l areas about the details of this ins tan c-e of the object (Eps htein, Lifshitz, and Ullman, 2008). lndeed, there is evi dence tha t interference with this "reentrant'' processing can bl ock conscious awareness of the object (Di Lollo, Enns, and Rensink,2000). (See Web Activity 4.7: Object Substitution Masking.) Theoct of object recogniti on should be seen as a con ve rsation am ong many pieces of the brain r.ithe r than as progression from spo ts and lines to th e activation of a g rand1nother ce ll in inferotemporal cortex.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

PERCEIVING AND RECOGNIZING OBJECTS

119

Before we leave this topic, it is worth mentioning on e very large part of the story tha t we've omitted up to n ow. There is good eviden ce that atfention to an object is critical in recogniti on of that object. For exa mple, a sp ecialized piece of the brain might be responsible for om· ability to recognize a specific face. If so, tha t piece of the brain probably requires a single face as its i.nput-hvo eyes, a nose, and a m o uth. If we're looking at a famil y photo, selective a ttention processes must come into play, lest we deliver a collection of a brother 's eyes, sister 's n ose, a nd \.m.cle's mo uth to the face processor. cells in different parts of the visu al system may behave different! y before and after attention is directe d to a n object (Serences and Yantis, 2006). Fo r example, a receptive field might change its size to wrap itself around a n a tte nded object (Moran a nd Desim one, 1985). This makes it pa rt of the nehvork processing tha t object. The same ne uron may be involved in the processing of a different object w hen attention shifts a m ornent later. Many of the brain s truch ues critical for the deployment of attention seem to lie in the o ther m ain path coming o ut of the primary visual cortex-the on e tha t goes toward the pa rie tal lobe. This p ath way, a nd th e general topic of a ttentio n , is taken up in C ha pte r 7.

Summary 1. A series of ex trastriate visual areas continue the work of visual processing. Emerging from Vl (primary visuaJ cor tex) are h.vo broad streams of processing: one going into the temporal lobe and the o ther into the parie tal lobe_ The temporal pathway seems specifically concerned with wlm t a stimulus might be. TI1is chapter follows that pathway. (The parietal where pa thway v.• ill be considered in later chapters.) 2. After early visual processes extract basic features from the visual input, it is the job of middle vision to organize these features intb the regions, s urfaces, and objects that ca n, in tum, serve as input to object recognition and sceneunders tanding processes. 3. Pen::eptual "rommlttees" serve as an important metaphor in this chapter. The idea is that many semi-independent processes are \\'orking on the input at the same time. Different processes may come to different conclusio ns about the presence of an edge or the relationship be hveen tv.·o elements in the input. Under most circumstances, we see the single conclusion that the committees settle upon. Bayesian models are one way to formalize this process of finding the most likely explanation for input. 4. Multiple processes seek to carve the input into regions and to define the edges of those regions, and many rnles are involved in this parsing of the image. For example, image e lements are likely to group together if they are similar in color or s hape, if tile)' are near E'f'"h other!_ at. if they are connec ted. Many of these grouping principles we re fi rst articu lated by members of the Gestalt school. 5. Other, reL.'l.ted processes seek to de termine whether a region is part o f a foreground figure (li ke this black 0) o r part of the background (like the white area around the 0). These rules of grouping and figure-ground assignment are driven by an implicit understanding of the physics of the world. Thus, events that are very unlikely to happen by chance (e.g., h.¥0 rontours parallel to each o ther) are taken to have m eaning. lThose parallel contours are likely to bepartof the same figure.) 6. The proce5..ses that divide visual input into objects and background have to deal with many complexities. Among these are the fad that parts of otJ;ec ts may be hidden behind other objects (occlusion) and the fuct that objects themselves have a structure. Is your nose an object or a part of a larger whole? What about gL'lSses or hair o r a \•dg? 7. In addition to perceiving the shape of objects and their parts, we are also very adept at categorizing the material that an o bject seems to be made

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Refer to 1he Sensation and Perception C.Ompanlon Website

sltes.slnauer.com/woHa4e tcr activifies. essays, stu:ty questklns. ard other study aids.

120

CHAPTER 4

8230336 Amma App·

of-glass, stone, doth, and so on.. We use material perception to estimate physical properties. \.\'hat would it feel like? Can it be grasped like a bottle or would it s lip through o ur fingers like sand? 8. Template models of object recognition hold that an object in the world is recognized when its image fits a particular representation in the brain in the .,.,•ay that a key fits a lock. It has ah,.,1ays been hard to see how naYve template models could work, because of the astronomical number of templates required: we might need one "lock'' for every object in every orientation in every position in the visual field _ 9. Structural models propose tha t objects are recognized by the relationship of parts. 'Thus, an H could be defined as h.vo parallel lines with a horizontal line joining them bet\..'"een their centers. A ca t \..·ould be more difficult, but sim ilar in principle. In their pure form, such models are vi5vpoint independent. ll1e o rientation o f the H doesn' t matter. Object recognition, however, is often viewpoint-dependent, su ggesting that the cor-rect model lies behveen lhe extremes of naYve template matching and pure s tructural description. 10. Faces are an interesting special case in which viewpoint is very important. Upright face. are much easier to recognize than inverted faces. Moreover, some regions o f the brain seem to be specifically interes ted in faces. They lie neilr regions in the tem poral lobes the eyes( Fig...-e5.7a ). Suppose tha t the light cl>at looks green produces 80 tmits of activi ty in the M-cones and 40 in the L-cones {remember, we are ignoring the S-ames fo r n ow). Jn addition, s uppose that the lig ht that looks red produces 40 units of activity in the M-cones a nd 80 in the L-cones. If we assume that we c.an add the cone responses together, then the "red ,, and "green '' lights produces a resp onse of 120 un'its in ead"t cone. TI1e ab.solute value is not importa nt, because it could change if the intens ity o f the light changes. What is hnportant is these hvo ligh ts, mixed together, produce a mixture that excites the L- and M.-cones equally. The key point is that the rest of the nervous system knows 011/y what the cones tell it. lf the mixture oflightsthat lookredand green prod uces the same cone output as the single wavelength of light that looks yellow (Fig ure 5.7b ),

(a)

Whathapperuif you addthi.slight

that loo ks red to o ne th.:\t looks green r

(b)

-------f'Y\-------11- L-cone

// \ \

I/ \\ I

\

I

R

Wavelength

650

700

a

Th is light that looks yellow pra..iuces equal L- .r1 e per lve., the subtractive color mixture a dar k color Ii e b rown. Actua lly, finger pain t mixtures are ra the r complicat&·t with some pigment pa rticles sitting next to ii through a filt er each and effectively add ing their reflected Ught 2. Pass that absorbs Shorter to the result. O ther particles occlude each other, a nd w.w.d.eng ths. Thi.ll do be tter if the v.rrong choice is on the other side of a color categoric.al boundary. Color boundaries are sharper than yo u might think. lf you s how people a collection of colors and ask, "Which are blue and w hich are green?" people do the task without mud 1 d ifficulty. If you have to remember a 12olor, as in the task s hoW11 in Figure 5.20, you are likely to give it a label like "green" or "blue." [f the next color has the same label, you are more likely to be conf used titan if it has a different label. Rosch found tha t the Dani's perfonnance on s uch tasks reflected the same color bounda ries, even when their language did not recognize the distinction be tween the hvo colors (a Dani might call all the colors in Figure 5.20 mi.Ii but would still do be tte r vvith the task in Figure 5.20b). This finding leads to th e conclusio n tha t color perception is not especially influenced by cu lture an d language; blue a nd green are seen as ca tegorically different, even if o ne's language does not e mploy color ternlS to express this difference. In the late 1990s, Debi Roberson went up the Sepik River in New G uinea to study the Berinmo, whose lang uage, like the Dani 's, has a limited set o f basic color terms. Unlike previously s tudi ed g ro ups, the Berinmo have terms tha t for m novel botmdaries in color space. For example, their nol/wor distinction li es in the middle of colo rs we ca tegorize as green, and may rough ly d istin guish Hve from dead or d ying foliage. Moreover, when the Berirnno did the color me mory task, tl1ey perform ed bet ter across their nol/wor boundary than across the blue/ green boundary. Eng lis h speakers sh m,.·ed the opposite result (Davidoff, Davies, a nd Roberson, 1999). More recently, sirnilar results

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE PERCEPTION OF COLOR

FIGU R E 5.20

(b)

(a)

143

It is 9asief to remern-

OOr w hich of t'NO c d ors you have s.:ien if the choices are cat egorically differ-

Pnt. For example, supposa you had to

H.uderchoice

Eafilerchoice

have been found in comparison of English and Russian subjects across the Russ ian categorical disti nction between light blue:s (golriboy) and dark b lues (s i11iy) (\Vi.nawer e t aL, 2007). And even m ore recen tly, evid ence has eme rged tha t lea rning new color s ubca tegories produces increases in gray m a tter in

parts of the bra in impli cated in color vision (V. Kwok et al., 2011 ). So, can language or culture influence color perception? For many years, we thought the answer was "no." Now, thanks to Robe rson's work, we are no t so s ure. Let's think a little rnore abou t the blue and green patches in Figure 5.20. Looking across the bottom four disks, you can categorize them as blue, blue, blue, and green. At the sa me time, you can tell tha t the first and second blues are not the sarne. ft tu m s o ut that different pa rts of the brain seetn to be involved in these different ways of aM lyzing color. When Bird e t al. (2014) s howe d o bservers in a hmctional magne tic resonance ima ging (fMRl ) e xpe riment a series a f patches of one color, folJowed by a change to a pa tch of a different color, vis ual cortex responde d to the size of the color change (green to "n ear blue" < green to 11 fu r blue"). ln contras t1 a location i.n the frontal lobes seem ed inte res ted in th e ca tegory o f the colo r. G reen to blue produced the sa m e response, regardless of the s ize of the cha n ge. Blue to blue did not produce a cha n ge, e\o·en though the change was p erceptib le to the observer.

Genetic Differences in Color Vision The indh·i dual differences described in the previou s t wo sections are either small or, in the case of inverted qualia., hypothetical. Under most circmns tances, if you d eclare two lights to be m etamerically matched, those arotmd yo u w ill generally agree, even if we can' t make definitive s tatements about their qualia. There will be some varia tion between individuals. For example, unique green cru1 vary between observers from at least495 to530 run (Nerger, \ :Olbrecht, and Ayde, 1995). Some of these differences "111 Pt;due tel fo owlik ag51: which turns the lens o f the eye yell ow Q. S. \'Verner, Pe terzell, and Scheetz, ii 990). To a flrs t a pproxi mation , however1 your pe rfor man ce on s tandard measures of color vision w ill be the sa me as others,,. l-I mvever, there is a s ignificant excep tion to this unh·ersa1ity of color matching . Some 8% o f the ma le population and 0.5% of the fe male population have a form of colo r vision deficiency conm1only known as "colo r blindness," in w hich there is a ma lfuncti on in one or rnore of the gen es coding the tluee cone photopigments. It's il "guy thing'' becau se the genes that code for the M - and L-cone photopigme nts are on the X chromosome {Nathans, 1986). h ave

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

rflirrl9mber the · blue'" patch shown at the top o f each part o f th'9 figuni;i. Pick· ing between two ·blues,• as In part (a), 'NOuld 00 rather hard. The t ask would be QBSier it o n9 o f the chcices \N"Elre and the o tt1er "green.• as in part (b). even If the d lstanoe In color space WQfe the same in the WJo cases.

144

CHAPTER 5

An lrdlvl:Jual wro suffers from OOcr blindness tl1at Is due to

deuteranope

1he aboonoe of M -caies.

wro

protanope An irdMdual suffers frcm color blindness 1hat Is due to the

abserce of L-oones. trltanope An Individual wro suffers

frcm color blindness that Is due to the

absen::;;e of S-cones. color-anomalous

A better term for what Is usually called •color-bllrd. • Most 'color-bllnd" lr-.:JMduals can stlll rnake discrtmlnatlons based on wavelength. Those dlscrlm !nations are different frcm the norm- that Is.

anomak>us.

cone monochromat An Individual with only one oone type. Cone rnonoc hromats are truly color-blind.

rod monoctromat

An lrdMdual

w ith no cones of any type. In addition to being truly coior-bllrd, rod rnomc hromats are badly >Asually Impaired In brtght light.

only one cop y of the X ch romosome, so if one is d efective, the male in question w ill have a problein. Fema les have hvo copie.s and can ha ve n orrna1 color \,.ision even if one cop y is abnorrna l (actually, su ch \"lornen ca n end up with four different con e pigrnents) (.Nagy e t al., 1981). 11'1e S.-cone photopigrnent is coded elsewhere, so everyone h..1s two copies, and therefore S.cone color d eficien cies are rare (Alpern, K.i tah.ara, and Krantz, 1983). (See Web Essay 5 .2: Experiencing Color Blindness.) TI1ere are a number of different types of color blindness. One de te rmining factor is the type of cone affected . A second factor is the ty pe of d efect; either the photopigment for tha t cone type is anoma lous (different from the nom1) o r the cone type is missing a ltogether. Although we call people w h o a re m issing one cone type "colo r-blind,'' it is a mistake to think tha t this m eans they ca nn ot see colors at a ll. As you \¥i ll recal l, if you have all three cones '¥i th their s tand.a.rd pho topigments, you need three priaiary colors to make a meta meric match wi th an arbitrary p,tttd1 of color. If you have t\vo cone types ra ther than three, the nom1'-llly th ree-dimensional space becomes a two-din1ensional space. The world "'ill sti ll be seen in color, but you will have a "fl a tter'' color experience, d ifferent from that of people v.ri th normal color vision. Because M- and L-rone defects are the most common, most color-blind individuals have difficul ty discriminating Lig hts in the middle- to-long-wavelength range. f:or exn mple, consider the wavelen gths 560 and 610 mn . Neither of these lights acti vates $-cones very much, and the L-cones fire a t about the same rnte fo r both . But most of us can distingulsh th e ligh ts on the basis o f the M-cone outpu ts they elicit, '"·hich w ill be hi gher for the560-nm light than for the 610-nm li ght (you can confinn these assertions by consulting Figure 5.5). English-speakin g trichromats wou ld labe l the colors of these two ligh ts as ''green'' and " re ddis h ora nge," respectively. Now consi d er a deuteranope., someone who has no M-cones. H is photoreceptor output to these hv o ligh ts ""-ill be identical. Following our maxim tha t the res t of the visu al syste m knm'\is only "''h a t the p h otorecep tors tell it, 560- a nd 610-nm li ghts mus t a nd will be dassified as the sa me color by our d eu tera nop ic individual. A protanope--someone w h o has n o L-cones- w iU il.."lVe a d ifferen t set of color ma tches based on the outputs of his hvo cone types (Mand S). And a trttanope--\vith no S-cones-""'ill be different again. Gene tic factors can Color-anomalous individ uals ty pi cally also make people have three cone photopigrnen ts, but two o f them are so si mila r th.a t these individ uals experience the world in much the sam e way as individuals lvi th only h vo con e types. \Ve actually have some n otio n of exactly what the world looks like to colord efi cient indhiduals, because the re are a few, very rare cases o f individuals w ho are color-blind in only one eye. TI1ey can compare what they see w ith the color-blind eye to w ha t they see with the n ormal eye, enabling us to reconstruct the appearance of the color-blind world (M acleod and Lennie, 1976). True co lor blindness does occur but it is very tmusuaL ft is possi ble to be a cone monochromat, with on ly o ne type of cone in the re tina. Cone monochromats (who also have rods) li ve in a one-dimensional color s pace, seeing the world only in shad es of gray. Even m ore vis ually impaired are rod monochromats1 w ho are missing cones a ltogether. Because the rods work well only i:n dim light and are genera lly absen t in the fovea, these individuals no t only fail to discriminate colors; they also have very poor acuity and serious difficulties seein g tmder n om1al d ayl.ight conditions. \Ve already mentioned one other very inte resting class of color blindness, 1.."'0 ming not from photoreceptor problems but from damage to the visual oortex.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE PERCEPTION OF COLOR

Lesions of specifi c parts of the visua l cortex beyond primary vis ual cor tex can cause achromatopsia. An achroma top sic ind ividual sees the world as drained of color, even w hile s howing e viden ce that wavelen gth infom1 ation is p rocessed a t ear lier stages in the visu al pa thway. Brain lesion s ca n a lso produce various for ms of color agnosia or a ne mia (Ox bury, Oxbury; an d Humphrey, 1969). In agnosia, the p a tient can see something but fai ls to know w ha t it is. Anomia is a n ina bili ty to n a me-in th is case, an ina bility to n ame colors . A patient w ith a nemia might be able to pick the banan a tha t ulooks right" but tu\Elble to repo rt tha t the banan a is or s hould be yellow.

From the Color of Lights to a World of Color All the work of this chapter1 up t o this point, has concerned the d etection , d iscrimin ati on, and appearance of isola ted lights . However, we live in a world \>•there regions of one color a but regions of r-mo ther, a nd this p roximity chan ges the a p pear an ce of colors, as Figure 5.21 shows. 1n color contrast effects, the color o f one regio n induces the op pone nt color in a neighborin g region. Thus, in Figure S.21a the yell ow s urround weake ns the yellow of a central sq uare and s tren gthens the blue. ln color assimllatlon effects, two colors bleed into each o ther, each taking on some of the chroma tic quali ty o f the other. So, in Figure 5.21b the blu e in the first column loQL reddish or pw:plisl n the to p image and greenish on the bottom . Not only can other colors i.n the scene alter the color of a target region, but scenes can contain colors that canno t be experienced in isolation. Though it may be ha rd to believe unless you try it1 you cannot sit in comp le te da_rkn ess

145

a{J'losia A failure to reocgnlze ct:>)ects In spite of the ability to see due to brain tram . Agnosla is damage. anomla An Inability to name objects In spite of the ablllty to see ard recognize them (as shown by usage). Ano· mla Is l)?lcally due to brain damage.

color contrast A color perceptlm effect In which the ook:lr of one region Induces the opponent ccior In a neighboring rajon. color assimilation A color perception effect In Vlfilch !'MO colors blee:J Into each other, each taking on some of the chromatic quality of the other.

(a) Color oontrMt

• (b) Coloras.simifation

FIG URE 5.21 Cobr oontrast and oolor assimilation. (a) In oolor contrast, thG C9ntrol squ an;;i takes on c hromatk: attributes that are opposite those o f the surround . so the gr99r1 central squar.e lcoks gro9Elner on the r9d background thWI on the green background. (b) In color assimilation, odors blend together locally. So, in the S9cond column th.e yellow sq uares look a bit reddish in the upper squ arG and a bt g reenish in thG bwer squarG. (A:ft9r Stcx::krnan and Brainard, 2010.)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

146

CHAPTER 5

utTelated color

A color that can be

experienced In Isolation. related color

A color. such

as

brown or gray, that Is seai onty In relation to other colors. Fcr example, a •gray" patch In ccmplete darkness appears white.

negative afterimage

An afterimage

w l1ose pdarlty Is the op posite of the crlglnal stimulus. Light stimuli produce dark negative afterimages. ColO!S

are oornplementruy: for example. red produces green. and yellow prodUJ:es blue.

adapting stimulus

A stimulus

w11ose removal produces a change In visual perception or sensitivity. neutral point

ll1e pdnt at which

an opp::nent color mechanism Is generating no signal. If red-green and

blue-yellow medianlsrns are at their neutral points. a stimulus will appear achrcmatlc. (The black-white prooess ' ""no neutral point. )

and see a grny light, all by itself. Tiiot light wi ll look white if seen as an isolated or unrelated color. To be seen as it mus t be seen i.n relations hip to othe r patches of color. Thus it is a related color. \Ve mad e this point earlier, ta lking about brov.rn. In isola ti on, a light mi ght look yellow or orange. Lt w ill only to patches. We can dis tinguish a f e\'\i' thousand un related colors. AIJm'\>'UlS for contex t effects is what boosts the number of distinguishable colors to the millions (Shevell, 2003).

Adaptation and Afterimages Color contrast effects show how the spatial relations between colors can influence color a ppearance. Temporal relations ma tter, too. What you saw before has an influence on the color you see now. You already know this from the discussion of light a dapta tion in Chapter 2. Adapting to a bright light ma kes a moderate light look darker. Adapting to darkn ess would ma ke that same mo derate li ght appear brighter. FURTHER DISCUSSION of the time course of dark adaptation can be

found "'1 Chaptel' 2 on pages 4 7-49.

Now Jet's extend tha t principle to color. Adaptation can be color-specific, as we see in the phe no1nenon of negattve attertmages. If yo u look a t one color for a few seconds, a subsequently viewed ach romatic region w ill appear to take o n a color opposi te to the original color. \ Ve can calJ the firs t colored stimulus the adapting stimulus. The illusory color that is seen aften vard is the negati ve afterimage. (See We b Ac tivity 5.4: Afte rimages.) The principle is i1lus trated in Fig ure 5.22. Figure 5.22a consis ts of a circle of g ray sp ots. Now, s tare at th e black d o t at the cen ter of Pigure 5.22b a nd consider whnt happ ens as you expose one bit of your retina and visual systern to the red dot at the top of Figure 5.22b. The L-cones w ill be more stimulated th an the M- or S-cones. L+/ M- oppo nent processes w ill be s timulated. You w ill see \ Vhen you move your eyes back to fi xate on the black dot a t the center of Figure 5.22111 the red is w ithdra\'\-11 from tha t area o f the visual fi e ld. The L-cones wi ll be mo re adap ted tha n M- or S-cones, as w ill the later processes in the retina a nd brain that were more s timulated by the red spot. Adapted processes be have as tho ugh they a_re somew hat tired. They respond less vigorous ly than unadapted processes. The result is a bit like w hat would happen if you he ld a pe ndulum up and released it. The redgreen oppo nent color mecha nis m swi ngs back toward the neutral point, overs hoots th is point, and slides over to the green si de. As a consequence, the gray s pot appears g reenish until the opponen t mechanism settles back to the n eutral point. If you look at the green d ot at the bottom of Figme 5.22b a nd then look back a t the gray image (Fig ure 5.22n), you w ill see the result of pushing the red-green mechanism in the o ther direction. Other colors '""ill prod uce othe r results, w hich you sho uld n ow be able to predict. In Figures 5.22c--e, you can try this '"''i th a real scene. If you s tare at the black d ot in Figm e5.22c and then quickly m.ove your eyes to the same dot in Figure 5.22d, you will see a tinted version of the real photograph (Figure 5.22e, tho ugh this effect w ill be more dramatic if you look a t the version in We b Ac tiv ity S.4: Afterimages). Notice that we are not attributing negati ve afterimages to just the cones or j11st one set of cone- or color-opponent processes. Adaptation occurs a t multiple sites in the nervous system th ough the p rimary generators are in the retin a {Zaidi et .'1., 2012).

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE PERCEPTION OF COLOR

(b)

(•)

• (c)

14 7

• •• • •••

(d)

(e)

FIGURE 5.22 To understard w hat negatiV9 aft91magesare, study the image in p art (a) and oonvine& yourself that a" the c irclgs are gray. Now star& at the black dot

in part (b). Aft91' 10 seconcls or so, shift your fixation t o the black dot In '8). The cird es should n ow look cdored . This is a negative aft9rlmage. Wrry does it happen? Of it didn't hapPQn , try fixating more rigorously. R9El.lly look at the black dot.) Now try with a real scene. Fixate the black dot In (c), and then fl ick your gyes t o the same spot in (d). You should 899 a vvashed-out version of th9 colors In {a).

Color Constancy You m ay recall fro m Ch apte r 2 tha t ad aptation is one way of dealing with the vast range of light levels in the e nvironme nt. Assuming tha t you are readin g

this on good old-fashioned p aper, w hen you are inside, you see this black text on a w hite page. If you take this book outside, you still see black on white, even though the illumination is so mudl brigh ter that the black text, viewed outside, reflects more light to th e eye tha n does the w hite paper inside. TI1is is an achromatic example o f the m ore general phe no meno n of colo r constancy, the tendency for the colo rs of objects to appear relatively Wlchanged in spite of s ubs tantial changes in the lighting conditions. in this book seem to have more o r less the same colors wherever you read the book (though there is an entire research area hidden in wha t we might mean by " m ore or less''; Fos ter, 2011).

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

color constancy The tendency of a surface to appear the same oolCf trder a fab'ly wide range of lllurrtlants.

148

CHAPTER 5

FIG URE 5.23 The S...."1111Q surface (8) illumir1ated by two lights \b) w ill gen19rate tiNo diff.Qf&nt patterns o{ activity in the s -. M- , and L-cones (c.---e). However, the suriaCQ will appear to bQ the same color Lll"ld€< both illurninants.

(a)

Thesurffice ...\ sutilce in the world re flects different percent,1ges o f diff;:>re n t wavelengths.

Perce.nl llge

of light re fl ected

This phenomenon Is known as color constancy. l;Aft"1r Smithson, 2005.)

(b)

8230336 AmMa App

Illumin.m tl

Illumin.mt 2

Ye llowish sunlight and bluish sk)'light are comIX'sed of

Enersy of light

differen t mixt ures

of w,1velengths.

(c)

Su1faoo x illuminant

ReJ.iti,•e

Surfaoe x ilhmtin,mt

reache..s the

amou n t

eye is th e surface

of light

ret1ect.mce mu ltiplied b)' the illunUnant.

(d) Cone

sensithrity

Theresulti.s !i< S-but wha t we \Vant to know is S. It is as if we \\'ere given a munber (sa.y, 48), told tha t it is the p roduct of h¥o other numbers, and asked to guess w ha t those h\.•o mnnber.s might be. The answer could be 12 and 4. Or it could be 16 and 3. Or 6 and 8. Given just the number 48, we cannot solve the problem. NeYertheless, given the result o f I x S, the visu..'ll system d oes a pretty good job of figuring out S. We smnetirnes talk about "discounting " the illuminant as if our w hole goal '"''ere to throw away the I term an d just see the surface color. However, this is n ot quite right. You can tell the difference behveen a scene lit by the tnorn.ing sun and a scene illtuninated by the s un at high noon. Thus, no t only can you recover the color of the stufcice, bu t you also kn ow somethin g about the illuminant. H ow d o you d o this?

Physical Constraints Make Constancy Possible As noted in the previous section, it is impossible to know whid1 hvo integers are multiplied to produce 48. However, if you a re told that the first number is behveen 9 and 14, you 're saved . The first mm1be r must be 12, and the second, then, mus t be 4. In a nalogous way, color constan cy must be based on some information o r assumpti ons tha t cons train the possible answers. There are rnany possible assu mptions that could help. Suppose we assumed that, in a complex scene, the brightest region was white (Land and McCann, 1971) or th at the ave.rage colo r across the whole scene was gray (Buchsbaw11, 1980). VVe could scale the other colors relative to these white or gray anch ors. H owever, th is can ' t be entirely right. Think wha t would happen if you were in a dark romn w ith two spots of light on the wa ll: a red one and a blue one. Under a simple version of a bri ght-is-white theory, the bright"' SJl .t s h) A honeybee can see UVllght. In W. the flowtt's "black.eye· (or maybe the pupl of that eye) is muc h larg er- a better t arg 9t for th9 bQQ. (Coul19sy o f Tom Elsnar.)

In a dditi on tosea rchi.ng fo r a nd assessin g foo d, a n imals spend signi ficant time and e ffort sea_rching for and assessing po tential m ates. Here, too, color plays a central role. ColorfrJ d isplays-from the dramatic patterns on tropical fish (Figure 5 .29a) to the rail of a peacock (Figure 5.29b) to the lace of the

mandrill (Figure 5.29c)- are all sexual signals. Wha t makes the peacock that has the most colorful tail the most desirable mv;ewer to attribute the distortion to the s lant of the picture surface. llle technique known as anamorphosls, or anamorphic projection, illush·ates that the ability to cope \.vi th distortion is limited. In anamorphic projection, the mi es of linear perspective are pushed to an extreme. Now the projection of three dimensions into two dimensions creates a hNo-cHmensiona1 image that is only from an unusual vantage point (or sometimes with a ctu ved mirror). TI1e res ults are k.nmvn as anam orphic art. As an exa mple, there is an odd diagonal s me.:u in the lower center of Hans Holbein 's sixteenth-cen tury painting Tire Ambassadors (Figure 6.19a). lf you could put your eye in exactly the rig ht p ositi on to view the image, the smear would prove to be the s kull s how n in Figure 6 .19b. Despite its s ua:essful recovery of the shapes in Figure 6.1&, the \·i sual system ca nn ot use knowledge about s urfa ce orientation to rompens.:'1 te for the d istortion in Figure 6.19. In our ovvn d ay, the sidewalk chalk

(a)

FIG URE 6.19 In 1633, Hans Hot· bein painted the double portrait in (a) \11/ith an odd objoct (/)) at the feet of the t'NO men.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

SPACE PERCEPTION AND BINOCULAR VISION

169

FIG URE 6.20 Modern-day anamorphic art . (a) In this photograph, artBt Leon Keer creattJs what appears to be a large three-dirnrensional version of the classic \1doo game, Pacman. But. as shown in (j:J), this is just a clever bit of anam orphic art . It is just a flat image that looks threedhnenslonal when viewed from the correct position.

artist Leon Keer crea tes amazin g anamorphic iinages tha t look spectacula rly real from the ri ght vantage point a nd spectacula rly d isto rte d from elsewhere (Figure 6 .20). More on the role of pictorial depth cues in art can be fmmd in Web Essay 6 .1: Making the lrnpliclt Explicit.

Motion Cues Beyond the pictorial depth cues, a number of additiona l sources of information are available to our visual system w hen we \•;ew rea l-\vor ld scene--cues tha t cannot be reprod uced in a s ta tic two-dimension al pictu re. The first no n pictoits power (and rial d epth cue we w ill discuss is motion parallax. To to und erstand why photograp hs of the forest often don' t come out well), the best thi n g to do is to go outside a nd lie w1der a tree. Gaze up in to the branches and leaves \vi th one eye covered and your head stationary. You w-ill notice that leaves and branches form a rela tively fla t text u re. You can see a U the details, but you may have trouble deciding vvhe the r one little branch lies in front of or behind. anothe r. U you open the o the r eye, bin oc ula r s te reopsis (i ntroduced ea rl ier and d iscussed in deta il later in the chapter) \.vill allow the branches and leaves to fill out a tluee-dim e n.sional volllll1 e that was lacki ng before. Close the eye, and the volume collapses ag'1in. Now, move your head from sid e to si de, and m otion parallax will restore the st>nse of depth.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

motion parallax An Important depth c ue that Is based on head movement. The geometrtc Information cbtalned fran an f!>f0 In two different p:E;itlons at two different ti mes Is eJmllar to the Information from ™> eyes In d iffetent poeltlons in the head at the same time.

170

CHAPTER 6

(a)

App a Time

FIG URE 6.21 Mo tion parallax. As you b ok out the window of a moving train , oqects closer to you the flower in this illustration) shift position m ore quickly than c:b obj9Cls farth.Qf" away (the tr"9) fro m on9 moment (a) t o th1:1 nQXt (b). This r9gularity can b"' explo itf'ld as a depth a.e.

H ow d oes m o tion p rovid e a cue for d epth? S uppose you 're si tting on a train, looking o ut the wind ow a t the countrysid e. At o ne instant you see the scene sketched in Fi g ure 6.21a . A m o m e nt la te r, the scen e has ch anged to the o ne in R gure 6.2 1b . N o tice that as ymu train m oved fro m le ft to right in

optic flow The pattern of apparent

motion of objects In a visual scene produced by the relatwe motrn between the oboorver and the scene.

the figure, a ll the objects shifted from ri ght to left. But note that some things s h.ifted more tha n others. Th e flower (Fin the figu re) m oved a l.most a ll the way across your retinal inwge, the cow (C) m oved a much shorter and the tree (T) hardly chan ged positions a:t all. The term para/fax refers to the geometric relations hip revealed. h ere : when you change y oux vievvp oint while rolling d ovim the tu1cks, obj ects closer to you s hift p osition m ore than objects farther away. Of course, you d on' t need to be on a tra in to experien ce motion pamlla xi jus t movin g your head w ill d o. TI1e geometric info rnmtion obtained from a n eye in two d iffe rent posi tions at hvo d ifferent times (m o tion para llax) is simila r to the information &om hvo eyes in different p ositio ns in the head at the sa me time (binocular stereopsis) (Durgin et al., 1995i Rogers and Colle tt, 1989), Motion pa rallax p rovides rela tive m etrical in formati o n about how far away objects are; and as the experiment with the tree bran ches proves, it can prov ide a compelling sense o f d epth in some situations in which other cues are n ot very effective. 1l1e d ow ns ide of moti on parallax is that it works only if the head moves (j us t rnoving the eyes back and fo rth won ' t d o, as you can eas ily prove to yourself). Now you know why a cat might bob its head back nnd forth as it plans a s pecta.cular leap from the sofa to the table, Other mo tion s ignal s produce in fom1ation ab out depth. For example, objects get bigger and s maller as they get cl oser a nd farthe r awa)'i so an object that is simply gettin g bigger on a scree n ca n appear to be looming towa rd you. If everything is looming at once in a large field .. this optic flow may rna ke you feel like y ou

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

SPACE PERCEPTION AND BINOCULAR VISION

(11) Con\'ergence

171

(b) Oi\•ergem:e

•

a Gets bigger. Convergence Oi\•ergence a Gets smaller.

FIGURE 6.22 (a) As WQ shift foa..1s from a far to a point, our 9yes oonvgge, (b) As we go from near to fur, th.a e')"9S diverge. Th€1 sizie o1 the angle (labek3d Qa") is a cue to depth .

Aduals·often descrt>e, !J\dMdual of

a sense are not just .,missing" a sense. Rather, the\/have develOped an entirely different '""I of oonslng the wood . Upon sensory restitution, a fascinating but rather disturbing experience unfolds as the brain has to

adapt to a new way to function. Even more dramatic is the experience of Stereo Bn..ce (Bruoo Bridgeman). a very perceptive vision scientist who had been stereodeficient all his life. Remarkably, he recovered stereopsls attar watching the 30 movie Hugo (Bridgeman , 2014). Whether this sort of immersive

experience. with very large dispsrities along with many other depth cues, v.; 11 be a generally effectr,,e treatment for abnormal stereopsis remains to be tested. However, these caoo stLdies, along v.;t h lab studies of perceptual l6"mlng that result in the reoovery of stereopsls (Di ng ard Levi,

2011 ), call into question the notion that has been the received wisdom. that rew1ery of stereopsis can only occur during early childhood. The Idea, dating back to the early twentieth century, has been that there is a

"crttical period" of dev8oprnent wtien the visual system is still plastic and capatle of change. After ttat . it was thooght. our basic \'isual capabilities are fixed. This led a number of practitioners to tell Susan Bany and her mother that "nothin;J could be done" about her >Asian (one suggested

that she might need a psychiatrist). binocular neurons are present in the visual oortex of primates wit hin the first v.ieek of life (see the "Development of Bi nocular Vision and Stereopsls" section below), Bany

surmises that some of the innate 1,viring of her binocular connections remaina::l Intact , and that vision therapy taught her to move her eyes into position for stereo visi on, 4 flnally giving these reurons the infonnation they were >Mred to receive" (Bany. 2009) .

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

SPACE PERCEPTION AND BINOCULAR VISION

FIG URE 6.3 2

179

A ster80 photograph

o f a woman's face.

Random Dot Stereograms Fol' 100 years or so after the invention of the stereoscop e, it was generally supposed that s te reopsis ocn1rred relative ly late in the p rocessing of visual stimuli. The id ea was that the first s tep in free-fusi ng images such as those in Figure 6.32 wo uld be to analyze the input as a face. We would then use the slight d isparities be tween the left-eye and right-eye ima ges of the n ose, eyes, chin, and other objects and p a rts to enrich the sen se tha t the n ose s ticks out in front of the fa ce, that the eyes are slightly s unke n, and so on.

random dot stereogram (RDS) A stE!feogram made of a large number (otten In the thousands) of randomly placed dots. Random dot stereograrns contain no monocular cues to depth. Stimuli vtslble stereoscopically In randcm dot stereograms are Cyclopean stimuli.

Bela Jule.sz, a Hungarian radar engineer w h o sp ent m ost of his career a t

Bell Labs in Ne w Jersey, thought the conventi onal v.risd om might be backward . He theorized that s tereopsis might be used to discorer objects a nd s ur faces in the world . Why wo uld this be usef ul ? Julesz thought tha t stereopsis might help revea l objects. A m ouse might be the sam e color as its background, but out in the o pen it would be in front of the bdckgro und . A ca t tha t could use s tereopsis to break the m ouse's ca mouflage would be a m ore s uccessful hunter. (Cats d o have s tereopsis, by the way [Blake, 1988; R. Pox and Blake, 1971] .) To p rove his point, )ltlesz (1964, 1971 ) made use of random dot stereograms (RDSs). A n exa mple is s hmvn in Figure 6 .33. 1f you ca n free-fuse these images, yo u will see a p a ir of squa res, o ne s tickin g out like a bump, the oth er lookin g like a h ole in the tex ture twhich one is the bump and w hich is th e h ole dep en ds, again, on whether you converge or di verge your eyes).

FIG URE 6.33 If you can free-fuse. th is random d ot ster90gram, you VYiH seg

two rgciangular re gion s.: on9 in front of the plan e of the page, the o ther behind th'11 page. Which is which d eplWlds on w heithe- you conV€rld Is In a particular state IA) given a particular observation (0).

lf the chapters of th is book v.:.-ere novels, this d 1apter cou ld be said to have t:l\e ;Jtame p lo t as the discussion of object i·ecognition in C hap ter 4, but w ith different characters. ln Chapter 4 we ta lked about a se t o f cues that en able us to group locaJ features toge the r into p ossib le objects and then to recognize those objects. We d escribed the process as a sort o f committee effort in wh ich different sources of informa tion a Ucontribute their opinions and w here we see the ccmunittee decision without necessari ly knowing how that decision was read1ed . ln this chapter we've covered multiple sources of depth infomtation and they. too, need to be combined . Any or all of these cues might be available to the visual system w hen we' re viewing a ny visual scene. None of the cues are foolproof, and n one work in every possible si tuation . For example, relative height produces inconsistent or misleading information if we see the point a t w hich an object to uch es the ground . All we really have is a collection of guesses about p ossible d epth rela ti ons behveen different objects in our visu..1.I field. By carefuUy combining and \\•eighting these guesses, the visual system generally arrives at a coheren t, a nd more or less accurnte, representation of three-dim ensional space. Helmho ltz, \vriting in the nineteenth century {an d trans lated into English in the t\11,rentleth), called this automatic cu e combination process uunconsdous infe ren re" (Helm.holtz, 1924). ln recent years, a number of vision researche rs have been a ttempting to put this sort of argument on the more rigorous m a thematical footing of the Bayesian approach that we mentioned in C hapter 4 and that is the s ubject of a m ore de tailed essay on the website, Web Essay 4A: Bayesian Analysis.

The Bayesian Approach Revisited Reca ll that the basic insigh t of Reverend Thomas Bayes '"'·as th at prior know led ge could influence estimates of the p robability of a current observation. Let's this idea to a concrete, depth perception example. Suppose our visua l systern is confronted '-vith the retinal image shown in Figure 6.41 . There are infinite possible ways to prod uce this retina l image. Actually, thi s is a bit o f ..i

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

SPACE PERCEPTION AND BINOCULAR VISION

problem for the use of Bayes' theorem. We do n' t n.."Ylfly know the prior probabilities (M. Jones and Love, 2011). Still, we can acknowledge that limitation, and we can s till make good u se of the' basic ins ight that som e hy potheses are m ore likely than others and that the;e prior probabilities can shape o ur interpretation of the world. Tiuee hypotheses nbout o ur pennies are sh own in Figure 6 .42. f\faybe the h.\•o pennies are the same size, but the on e on the left

is slightly farther away than the one on the right (Figure 6.42" ). Maybe the penny on the right is mud' s malle r, but also m\.1ch closer, than the penny on the left (Figure 6.42b). the two pennies are equidistant, but the penny on the left is smaller and has had a bite taken out of it (Figure 6.42c). If you d on' t see h ow the set o f p ossibilities could be infinite, reme1nber that s ize and distan ce can vary continuously over a large range. ln Figure 6A2b, the big penny cou ld be on the rnoon (but it would h.ave to be a really big pe1my). How d oes the visu.:1.l syste1n decide what we' re actua lly seeing? Which interpre tatio n seems 1nos t likely? That is the core of the Bayesia n approach (except thatit'sall automatic; our conscious selves do not get to make th e decision). [n mu: exp e rience, all pennies are the sam e size. This cue of fami liar size is o ne source of prior know le dge in this case. This makes the prior probability of the hyp otheses shown in Pig ure 6.42.n higher than the prio r probabilities of the other hvo hypo theses. Furthermore, for the scene in Figure 6.42r to produce the re tinal image in Figure 6A1 , we wou ld have to be seein g the scene from o ne of those unusual and unlikely 11aocid entc1l viewpoints." It is mud1 m ore likely that the points of contact between the images of the h.-vo p e nnies reflect occlusion. If we were to plug all these probabilities into the math of Bayes' equa ti on, \ Ve would find th.:1t,given the image in Figure 6.41, the m ost like ly answer is the scene d ep icted in Figure 6A2n .

(•)

(r)

(b)

1 Identic.tl pennies ,1t .slightly different d'Ptru

PeruUes of di:ffarent siz.es at very different depths

A penny with a bite

out of it, right next to ai nonnal penny

uuu

FIG URE 6.42 Three of the infinite number of scenes that could generate the retinal imag9 in Flgur9 6.4 1. Which o f thooe ts th£ii most probro!Q?

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

FIGURE 6. 41

Retinal Image of a

slmpe v isual scene.

187

188

CHAPTER 6

!\!\!\!\!\ FIG URE 6.43

In which image are the two horizontal lines the same length?

ln thinking about combining d epth cu es, our choice of the me taphor of a committee is n ot arbitrary. We could have talke d about an election, but that

would have itnplied something like "one cue one vote." On a co mmittee, you might have one member w ho is s tronger than the othe rs and vdn.s all the arguments. You might give more '"'eight to the committee member \"tho comes prepared with the best in.formation. The committee nUght defer to one me mber on one top ic and another mernber on a different topic. Somethlng like this option is d escribed by He ld, Cooper, and Banks (2012). Binocularinfom1atio n can be very precise, but that is only true ne ar the plane of fixation (re.m e mber Panum 's fusional area?) . Blur can be quite a good cue too, but it is actually better aw.J.y from the plane of fi xa tion . When Held e t al made s timuli th.a t had dis pa rity cues, blur cues, o r both, they found th.J.t disparity drove respon ses where di sparity was more reliable and that blur drove responses in parts of the three-dim ensionaJ world where it was more reliable. Th.is is different frorn just letting every cu e have its say, and it makes us realize that the visual system mus t be estimating h ow reliable each depth cue might be .

Illusions and the Construction of Space If oul" visual perception o f the '"'orld is our best guess about the ca uses of

8230336 /.

visual input, then interesting things should happ en when a g uess is w rong. jus t discussed, a guess is wro ng w henIn some sense, as vdth the pennies ever we look at a two-dimensiona l picture and see it as three-dim ensional As noted, however, we are not really fooled into thinking that the picture is three-dimens ional. Lt would be m ore accurate to say that '""'e make a p lausible guess about the three-dimensional world that is being represented in the twodimensional picture. \.Vhat about a situation like tha t shown in Figure 6 .43? One of the five pairs of horizontal lines (and on ly one) shows two lines of the sam e length. Can you p ick the correct pa ir (vd thout a ruler)? ln fact, it is the second from the left _ Odds are you picked the third or fourth pair1 even though the bottom line in both of those images is physically longe r than the top line. This is known as the Ponzo illusion, naine d after Mario Ponzo, who described the e ffect in 1913. What causes this illusion? For many years, a p opular famil y of theories has held that the illus ion is a guess gone wrong-a situation in which '"·e overinterp ret the depth cues in a hvo-dim ensional image. The basic idea is illustrated in Figure 6.44. Maybe the two tilte d lines that induce the illusio n in each image of Figure 6A3 are being interpreted b y the visual sys tem as linear-p erspective cues like the train tracks in Figure 6.44. lf so, then objects that were the sam e s ize in the two-dimensional image \o\-'ould represent objects of diffe rent sizes in the three-clime n.sional world . SUch accounts are very cornpelling and exist for a wide range of visual

illusions (Gillam, 1980). (See Web Essay 6.3: The Moon Illusion.) They are consistent with the view that the job of the visual system is to use available cues to make an intelligent guess about tl'e world (Gregory, 1966, 1970). jus t

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

SPACE PERCEPTION AND BINOCULAR VISION

FIG URE 6.44

189

Th9 t...vo p 90ple t>Jjng across th9S9 train

tracks ar9 the same size in the irnage. You c an vt1orlfy this by rnaasurk1g th.em. the more distant pgrson INOUld n94;1d to b9 muc h larg'-C" in the rM threadlmensional v-rorld to produce thB Image in the twofigure.

becau se a n ansvo.·er is plaus ible, however, doesn' t m ean it's right. ln Figure 6A5a, line B looks longe r than line A That m akes sense if we' re interp reting

these lines as lines lying '1t d ifferent distances on the w all of the colo1made. As in the Ponzo illusion, if line Bis farther away than A, then the same i.rnagesize

implies a larger size in the real war.Id. But what '1 bout lines C a nd D? Surely D would be inte rpreted as farther away than C. Does it look convincing ly larger? Ftgure 6.45b reveals \.\.'hat you probably guessed-that alJ fi ve of the lines in Figure 6A5a are the sa me leng th. Prin zmetal, Shimamura, and Mikolins ki (2001) use a demon::;tration like this as pa rt o f the ir argument that the Ponzo illusion is not reall y a by-product of de pth cues. They argued that it reflects a m ore general asp ect of the visual system 's response to tilted Hnes and is related to iJlusions like the Zollner and Hering Wus ions illustrated in Figure 6.46 {which we w i!J '101 try to exp lain;

(a)

FIG URE 6.45 All o f the red lines in this illustration (e) are the same length, as you can Sie9 In f/;JJ. Does e look bigthan A? Does D look bigger than C? In 98.ch case, which ling is farthi91" awismus A misalignment of the two fi'/es such that a sing le object In space Is Imaged on the fovea of one &ye

and on a nonfoveaJ area ofth9 fi'/0.

other

esotropla Strablsmus In which one eye deviates Inward. exotropla

Strablsmus In which one

"'f0 deviates outward. tilt anereffev esotrop ic infants ted stereop sis. This res ult has an in teresting pa rallel in corti cal physio logy. Chino a nd his colleagu es (Kumagam i e t al , 2000) m.ade otl1erv..·ise nonna1 m.onkeys stra bism ic. They fo und tha t a brief period o f experi me ntal strabi.sm us shortly after the age of onset o f stereop sis produced a grea ter loss o f d ispa rity sen sitivity and more b inocular s up p ression in Vl neurons than did an earlier e p isode of s trabis mus. Th ese physiological and perceptual deficits a pp eil!' to be perman ent a nd have importa nt im p lications fo r the s m gic-a l treatment of infantileesotropia. Alm os t all s urgeons agree tha t treatmen t should be early, but there has been a lot of d eba te about just how early. These resul ts su ggest tha t the treabnen t s ho uld take place before the age a t w hich s tereopsis normally develop s, in ord er to mi nimize the damage d one by esotropb.

Normal F.sotropic

10

12

14

Age (months)

Summary 1. Reconstructing a three-dimensional world from two non-Euclidean, curved, t\vo-dimensiona l re tinal images is one basic problem fa ced by the brain. 2 A nwnber of monOC'Ular cues provide informa tion abo ut three-dimensional space. 1h:se include occl usion, various size and position cues, aerial perspective, linear perspec tive, motion cues, accommcxlation, and convergen ce. 3. Having hi.•o eyes is an advantage for a number of reasons, some of w hich have to d o ....·Hh depth perception. It is important to remember, however, that it is possible to reconstruct the three-dimensional \Vorld fro m a single two-dim ension.al image. Two eyes have other advantages over jus t one: expanding the visual field, permitting binocular summa tion, and p roviding redundancy if one eye is damaged. 4. Having h'io laterally separated eyes romtected. to a single brain a lso provides u s with important information about depth throu gh the geometry of the small differences between the images in each eye. These differences, known as binocular dispa rities, give rise to stereoscopic depth perception . 5. Random d ot s tereograms show that we don' t need to know what we' re seeing before \\o"e see it in stereoscopic depth. Binocular disparity a lone can support shape p erception . 6. Stereopsis has been exploited to add, literally, de pth to entertainmentfrom nine teenth-century photos to t\venty-first-century movies. It has also served to enhance the perception of information in military and medical se ttings. 7. The difficulty of matching an image element in one eye w ith the correct element in the other eye is known as the correspondence problem. The brain uses several strategies to solve the problem. For example, it reduces the initial complexity of the problem by matching large "blobs"' in the low-sp atialh equency in fo rmation before tryi ng to match every high-frequency d etaiL

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Refer to the Sensation and Perception Companion Website Sltes.Slnauer.comtwoHe4e for activmes, essays, sl.Jdy quesl ons, and other study aids.

0336 AMma App

198

CHAPTER 6

8. Single neurons in the primary visual cortex and beyond have receptive fields tha t cover a region in three-d imensional space, not just the twodimensional image plane. Some neurons Si:'eITl to be con cerned with a crude

in-front /behind judgment. Other neu rons are con cerned ....ith more precise, me trica l d ep th perception. 9. Whe n the stimuli on corresponding loci in the two eyes are different, we experience a continual percep tual competition behveen the two eyes knrnvn as binocular rivalry. Rival ry is part of the effort to make the best g uess

about the current sta te of the world based on the current state of the input. 10. All of the variou s monocular and binocular depth cues are combined (unconsciously) according to ....·ha t prior knowledge tells us about the probability of the current event. Making the wron g guess about the cause of visual input can lead to illusions. Bayes' theorem is the basis of one ty pe of

formal understanding of the rules of combination. 11. Stereopsis emerges suddenly at abou t4 months of age in humans, and it can be d isrupted through abnormal visual experience during a critical period early in li fe.

8230336 Aml'la Appa

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

lq

8230336 Arnrna App.l

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

CHAPTER

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

7

Attention and Scene Perception IF YOU'Rt Tl-HS, you are probably a s tudent If you' r e a s tudent, you are probably taking more than one course and are therefore very busy. H ere's an idea: w hy n ot read t\vo books at the .sa me time? Chapter 2 will ha ve told you that the lirnit on peri pheral acuity is on e reason tha t this won' t work. The acuity problern could be overcom e if the size of the print were increased, as in Figure 7 .1 . Nevertheless, even \Vith nice will be clear that you ca nnot look at tl'le column of Xs in the figurennd the tvv sent \Ve just cannot read hvo messages a t the sa me time. Note that you cnt1 read the words on one side or the other of the Xs while looking a t the Xs. You jus t can' t read both sides si multaneously. lltis is a specific example of n more general problen"l-that the retin.11 array oon tains for more infom1.ition th.nn we can process. Figure 7 .2 shO\vs another example. We canno t possibly recognize a.LI the objects in this picture at once. That's why 11Where's Waldo?" a nd 111 Spy" gam es are a di.allenge. This is no t just a visual proble m. All of the senses receive m ore input than we ca n handle (see Chaptel' 10 fo r a discussion o f attel\tion in hearin g). Why can' t we process everything at once? Quite literally1 we don' t the brains for it. Remember from Chapter 4 tha t recognizing a single object like an elephant requires a sizable chunk of the brain and its processing power, especially w hen that elephant could be seen in many different orient.'l tions, under different lighting conditions, a t different reti nal sizes, and so on. Moreover, in order to tmders tand Figure 7.21 ""'e also need to process the re lationships beh\.•een objects-like the fact the SOA is 0 millisecond s (ms ), the cue and probe appear . There is n time for the cue to be used to direct atte ntion, a nd there is n o d ifference beh Yeen the effects of valid or in va lid cues. As the SOA increases to a bout 150 ms, the magni tude of the cu eing effect from a valid peripheral rue increases, as s how n by the red Line in Figure 7.4. After that, the e ffect o f the cue levels off o r declines a bit. Symbolic/ end ogeno us cues, sud 1 as the colored do t, ta ke longer to work, presumably because \ Ve need to d o some work to in terpre t the m (the SOA is shmvn by the blue line in Fig ure 7.4). In teres ting ly, some rathe r symbolic cues beha ve like fas t, periphe ral cu es. For examp le, we are very quick to d ep loy m u· a tte ntion to arrow cues and even faster to resp ond to a pair o f eyes looking in one direction or anothe l' (Figure 7.5), as if we were built to get infom1ation frorn the gaze of o thers (Kuhn and Kings tone, 2009). You can h·y the cueing experit11e nt in some of i ts important va ria tions in Web Activity 7.1 : Attentional Cueing. ln terestingly, if you m ove your attention or your eyes to a loca ti on a nd then look away from that location, it is harde r to look back a t it again for a littl e w hile. This is known as inhibition of retum (Klein, 200J). lnhibitio11 of

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

ATTENTlON AND SC ENE PERCEPT! ON

205

re turn he lps to keep yo u from gettin g stuck continu ally revisiting one spot. For ins tance, s uppose yo u were loo king fo r Wa ldo in Figure 7.2 and your attention was attracted to the S.m'ta in the lower ri ght lf yom a ttention kep t goin g back to Santa, that would be rathe r useless. Inhibition of rehun helps your sean:h to m ove fon•lard (Klein Maclnness, 1999).

The "Spotlight" of Attention In a cueing experirnent, attention star ts at the fi xation point and som ehow ends up at the cued loca tion. But d oes it actual ly move from on e point to the nex t? Attention could be deployed from spo t to spot in a ntmlber of ways. It nl ight tn ove in a rna nn er analogous to the Jnovements of our eyes. \Vhen v.Asual cortex ITT humans that Is specll cally and rlably activated more by Images of places than by other stimuli.

1 11.!J • • , ,• • • ••• I I - . I1 I I

······ ·- ....

, •• 11 · ·· · ··· ·- ·

FIGURE 7.17 Attentional S919ctlon. Rrst, attend to the r9d itQffiSand n oticQ the rough oval that they form. Next, w ithout moW'g your eyes, switch atterrtion t o the blue Items or thQ horizontal Items. Diffenmt aspects of the pcture appear more proml· nent as you shift the property that you setect to attend to.

(•)

(b)

x

to alter what you see, even as the stimulus remains physically unchanged . ln Figure 7 .18, you can see attention change the apparent brightness a nd / or color of a stimulus (Tse, 2005). Attentional selection can also activate parts of the brain that are specialized for the processing for one o r another type of stimulus. Recall from Chapter 4 that fMRI has shown tha t different p arts of the human brain are especially important in the p rocessing of faces (the fusnorm lace area [FFA]; Kan wisher, McDermott, and C hun, 1997) a nd places (the parahlppocampal place area [PPA]; Epstein et al., 1999) (Figure 7.19). If s ubjects view an image of a fa ce s uperimposed on an image of a house ( Figure 7 .20), the FFA becomes mo.re acFusiform face area

FIGURE7.16 Attentional modulation of appearance. '8) Lock at the X and notice that when you attend to one rectangle or the oth91', that one appears Ip} Here, if you attend t o the pink r9Ctangle, the owrlapping area looks pink. It changeis to blue If you attend to the tfoe rQCf:anglB. (Aft91"

Tse, 2005.)

FIGURE 7.19 F\Jnctloni> MRI roveals that different plees o f the cortex are activated by faces and by places. This is true wi"l9n both s timuli are present at the same tlme (as in Figure 7.20) and the obS9fV91' roorety attGrS his mrantal set from facas to places. (M Rls oourtesy of Nancy Kan'if\risher .}

QV9n

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Parahippcc.unpal place aN:a

AITENTlON AND SCENE PERCEPTION

217

FIGURE 7.20 ThQ:s.Q imag9s com · b lne faces and houses. In both cases

you c an use attentio n to enhance thie PQrception of on9 or the o thg. As ycu s'Witch your attention f rom one type of object to th9 o ther , you also c hange the activation of different p arts of the bro.i n . as shown in Figure 7.1 9. (Left image from Downing, Liu, and K.anw Bher, 2001 .)

tive when the s ubject is attendin g to the face, a nd the PPA becom es more active w hen the s ubject is atten din g to the ho use (O'Craven a nd Kan wisher, 2000). FURlH ER DISCUSSION of FFA and PPA can be found in Chapter 4 on page 11 3.

Attention Could Coordinate the Acti vity of Different Brain Areas How might the brain irnplement a solution to the binding problem'? Morespeif di ffe:l'ent brn in areas perform different tasks li ke processing f.:ices or places or color or orien ta tion, how rnigh t those areas be coordi nated if, fo r instan ce, we wan t to think about h ow the \Vind ows and door of some red house look like n-..·o eyes and a m outh? On e possibility is tha t th is bind in g a nd coordiMtion o f areas synchronizing the tempor,'\l pa tterns of dctivity in those are.'Posure. By the wa)'i at that rote we cannot conunit images to mem ory very effectively, but we do have a pretty clear idea o f what we 're seeing as it flas hes by.

But, Memory for Objects and Scenes Can Be Amazingly Bad: Change Blindness From the findings just described, we could conclude that pktmes can be tmderstood very rapidly and that, given enough time-perhaps a seocond-\.ve can code the m into memory in sufficient d etail to be able to recognize them days later. But d on' t endorse tha t conclusion too quickly. In the 1990s, Ron Ren.sink and his colleagues introduced a diffe rent sort of picture rne mory experiment (Rensink, O'Regan, and C lark, 1997). 1l1ey just one picture at a time. The observers would look a t the pic ture tor a while, and then it \'Vould vanish for 80 ms and would be replaced by a si milar irn.age. The observer 1s task was to determine what had changed. The two versions of the same image would flip back and forth, separated by a brief interva l in which a blank screen was presented, until the observer spotted the change or time ran o ut. A s tatic version of the task is illustrated in Figure 7.33. Try to find the differences be t\.veen the two pictures in this figure. TI1e answers a re in Figure 7 .34. You can also try detecting chan ges in some flicker movies in Web Activity 7.5: Change Blindness.

It may have taken y·ou a while to discover all four changes in Figure 7.33. That's what happened in Rens ink's experiment:: participants took several seconds

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

ppa

227

228

CHAPTER 7

FIGURE 7. 32 \l\lithout refQtring back to Figure 7.3 1, tty t o identlff \Nhic h o f these or "new.• You c an pictur9s you afrQady saw. Decide wh9ther 9ach picture is check your ane.wers on the grid to the left. n1e shaded squarf!iS show the locations of ok:t p ictures.

8230336 Arlma Appa

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

ATTENTlON AND SC ENE PERCEPTl ON

FIGURE 7.33 them ?

229

There are four d ifferenoes between thase two lma;;1es. Can you find

on aver age to find the changes, and some never num aged the task at a ll in the time they were given. This phenomenon is known as change blindness. Jf the blank screen between the two itnages is rem oved, the cha nges are obvious because observers experience a kind of apparent-motion effect (see Chapter 8) w h en a n object d um ges position, d isap pears, or changes color. But on ce we1 ve h'tken care o f tha t lo w - level facto r by inserting the blank screen , view ers ca n be quite obliv ious to jet engin es tha t vanish from airplan es, b ra nches tha t jump from tree to tree, and boats tha t cha n ge co lor. Actua lly, we can elimin a te the need for a bla n k screen if the cha nge is ma d e w hi le the eyes are m oving. We are virtuall y blind during a n abrupt (saccadic) eye movement McConkie and Currie (1996) took ad vanh.1ge of this fact to m ake dramatic ch an ges in a scene while an observer was jus t looking around . If they made the chan ges w hile the eye was in moti on, observers often foiled to notice. FURTI-IER DISCUSSION of ete movements can be found in Chapter 8 starting on pa;Je 250.

It is not immed ia te ly obvious how these various facts can all be true of the sam e visual system. This is a topic of current controversy in. the field. How can we remember thous.an ds of objects after a mere second or two of v iewing? And if we can d o that, h ow can we fail to n otice s ubstantive change in a picture that is right in front of us? Does it m ake any sense that we remeinber the p ich.ue but seem oblivious to important details?

What Do We Actually See? In the 1990s, Dan Simons and Dan Levin (1997) conducted a wei rd experiment on the s h·eets of Ithaca, New York. One experim enter asked an uns us pecting passierby fo r d irec tions. \'\fhile the passerby was g iving those directions, a couple of othe r p eople carried a door d own the street and bet\.veen the p asserby and the experimenter. While the d oor was m O\.;ng by, the experi menter ducked d own, s neaked away, and was replaced by a second experimen ter. 1l1en the

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

change blindness

The failure to

notice a change betvJeen Thvo scenes.

If the gist. or meaning , of the scene Is not altered , quite large chan:Jes can

pass unnoticed.

230 CHAPTER 7

FIG URE 7.34 HQre are the locations o f the differences between the two irnageis in Rgure 7.33. \NhQn you go back to figure 7.33 with this information. you will hav9 no troubl9 s99ing

what was alttfOO.

d oor was gone d own the street, and the critkaJ ques tion '"'hethe r the passerby wou ld return to giving instn1ctions. In many c,,15es, the passerby did just tha t, OlS if the d1ange in experimenter had gone u1mohced . How could that be? Perhaps this change was undetected because it didn't ch ange the gist of the scene. The passerby might think, ''l' m giving ins tructions toa guy. Now there is this weird d oor. Now I'rn giving ins tn1ctions to d gu y again," and simply n ever register that the gl1y chan ged. See 11 The Colour Changin g Card Trick'' in Web Activity 7.5: Change Blindness for anoth er exa mple. To make this into a more con v inci ng theory, it would help H we really unders tood w ha t the o f a scene nlight be. Gist is clearly m ore tha n the brief verba l d escripti on \Ve ml ght give if asked to "d escribe the gist. " Look at Figure 7 .35. Now look a t Figure 7 .36 and decided w hich one of these two photographs is the same as Figure 7.35. You ca n p robably d o this quite eas ily, even though the gist descriptions of Figures 7..36a and b \Vould be very similar: a skier on a fairly s teep slope on a s unny d ay \vi th mountains in the backgrotmd. TI1ere is some thing a.bou t 11gist" that we cannot trivia lly put into '"1ords . You can experien ce a much simpler version of c hange blindness in Web Activity 7.6: The Attentional Bottleneck. Participa nts ob served a collection of red and green spots. W hile they watch e d, one of the sp ots s uddenly got brig hte r, a nd on half the trial s the brightened spo t also changed color. T he ch an ge in brightness inv:ariably a ttracted a t tention to the critical s pot. All the observers had to do was determine wh e the r the sp ot had changed color w hen

it changed brightness. l11ey performed horribly; tl1ey man aged to get onJy 52%

8230

correct in a task where they couJd get 50% correct just by guessing. It seemed tha t unless the o bservers happened to be a ttending to the critical spot before the d1an ge occurred , they had no idea \vhether the color ha d cha nged (\ Volfe, Re inec ke, and Bra\oVJ.11 2006). Th is finding is consis tent \vith our change blindness sto ry. The gist of the disp lay di d n ot cha nge wh e n on e sp ot changed, so the ch an ge was no t deteeted t'm1ess tha t spot h appened to be attended a t the time of the d1ange. At any moment, then, we are percei\o;ng the gist of the scene, even if we can 't qui te verbaliz.e w ha t that gist might be; and we a re perceiving the currently attended, bound, a nd recognized products of the seledive path way. The ad -

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

ATTENTION AND SCENE PERCEPTION

231

FIG URE 7.35 Look at this pciure for a couple of seconds. Describe it to yourself in a sentence. Then go find Agure 7 .36.

va ntage of this very simple display is th at we can estimate how many objects ""'ere attend ed at the mornent of the d1e1 nge. TI1e estimate is in the range of :zero to fo ur items, indicatin g, again, that selective attention selects very few objects at any on e time. The problem with this view is that we tlrink \¥e see so much more. As we look around the world, we experience a full-color, in-focus, three-.dimensional scene fill ed with dearly identifiable objects. \ftle d o n ot h ave a pe rcep tua l expe rience of zero to four ob}ects, fl oatin g in a sea o f ensemble \Ve know, from the research described in other chapters, that this experience mus t be a constructi on of the m ind, not an exact copy of the s timulus. ·we know from Chap ter 2 that the world is decently focused only at the fovea and that high -resol ution processing occurs only in the central 1 degree or so of the visual field_We wi ll see in 8 that the scene before us is s m eared across the retina three or fo ur times ead\ second as we m ove our eyes. We know from C hapter 6 th at \o\>'e1 re actu a ll y getting h.voslightly different cop ies of the .scene from our hvo retinas, and that both copi es are rea lly rn.·o-dimensional. Over and over in this book, we say thata particular aspect of perception is an inference a guess abo ut the world . Our co1 s:dous experience of the world is the moth of.::ttll Th i!.)eaJization is hard for us to aa::ept beca use perception doesn' t/eel like a n inference about wha t is. It feels like it must be sho'"ring us what's really out there. It feels like Truth. Pheno mena like change b lindness are important beca use they s how us the gap between percepti on and rea li ty. Experiments investigating ch ange blindness s how t LS that we d on ' t see all of wha t is there. Rather, in some spots, we see \vhnt wns there wh en we last paid attention, and even then we may not remembe r it perfectl y (lrwin 1 Zacks, a nd Brown, 1990). In othe r s pots \·Ve perceive w h at we think sh o uld be there {see Web Essay 7.2: Boundary Extension). Moreover, we may n ot perceive what we d o n ot expect to see--whn t d oes not fit with the current inference. This fail m e to notice the unexpected lead s to the phenomenon of inattentional blindness. TI1e great example of inattentional blindness comes from a n other experime nt of Da n Sim ons, this time in colla boration w ith C hristopher 01abris. Th ey had observers keeping track of ba ll m ovements in a basketball passin g gam e. Many of these subjects failed to no tice an actor inn

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

inattentional blindness A failure to not be -cr at least to report-a stimulus that would be easily reportable If it

'Were atterded.

232 CHAPTER 7 (a)

(b)

FIG URE 7.36 Which of thes9 two lrnag9s is the on9 you saw in Figur& 7.35? You probably know, even tt1ough your wrbal d'9SCt1ptlon o f th9 new Image would be very similar to the d95crlptbn you gave for Figure 7 .36. ApparQfltly, the gist o f a Sc.QOQ contains more than Is captured by our usual verbal descriptions.

gorilla

into the middle o f the scene, waved, and wandered

off (Simons and Chabris, 1999). Outside the lab, the bottleneck be tween the world and. our perception is not u su ally mud1 of a problem, because the world is a pretty stable predictable place-at leas t o n a m o m en t-to-mo m e nt basis. U you put your coffee

8230336 Amma A

mug down on the desk and turn your '1ttention to the computer screen, the mug will be there w hen you choose to attend to it again. Only in the lab does the coffee mug vanish during an eye m ovement. This means tha t the physical \\:orld can serve as a n 11externa.1 memoryu tha tb.:lcks up the perceptual world we create in our m inds (01Regan, 1992). \ Vhen the wor ld does cha nge, it usually changes in very predic table ways. U we' ve ju st driven past Second Avenue a nd then Third Avenue, we probably d on ' t need to look a t the next street sign to know that it says "Fourth.'' At a rate of about 20 objects per second, we can moni tor the relevan t ite ms in th is relatively s table world and be reasonably sure that we are up-to-date. And if some thing smprising does occur (a gorilla leaps into the middle o f Pourth Avenue, for exampl e), the event will probably be marked by a visual transient that grabs our a ttention so that we can update our internal pretty qttickly and maintain o ur grasp o n reality (Yantis, 1993). (See Web Essay 7.3: Attentional Capture.) A certain amount of work needs to be d one in order to induce us to m iss gorillas. Still, we sh o uld n ot fee l t oo complacent a bout o ur inference of the perceptual world. Con.sid er, for example, tha t eyewitness testimony is based on the ass umption that you can report w ha t was there, not '"'hat yo u in ferred, gues.sed, hoped, or feared was there. Sadly, if unsurprisingly, we know that eyewitness testimony is subject to the sa me sorts of effec ts discussed here a nd th..1t these factors are p robably the cause of errors with real conseque nces (G. L. Wells and Olson, 2oro). \Ve need m ore data before the courts and the public v1,i ll understa nd this problem. As a step in th at directi on, C hnbris and Simons (2011) have extended their work on real-word inattentional bl indness to show that people w ho were g iven a task of following a jogger down a track could

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

AITENllON AND SCENE PERCEPTION

233

eas ily fai l to notice a very visi ble apparent bea ting occurring along their ro ute. The researche rs picked this particu la1· scena rio because of a re'>'-'

pp.:i

•

FIGURE 8. 15 Stucfying gye movernents. (a) Axing our gaze on the dot whlle the pencH moves to the left: causes the penc;ll to generate motion across the retina. and we pgrcg,lve mOVQ'TIQnt o f th& pgrlCil In this dlr9Ction. {b) Rxating our gazg on th9 pgial while it mows to the right caJS9S the dot to gen9rate motion In the same direction across the retina. but we do not peroalw movement o f the dot.

the desk in front of you, and draw a small black d ot right in the center o f the paper. Close your le ft eye a nd foc us your right eye's gaze on the dot; the n p osition a pencil so that it is n ear the bottom right com er of the paper. Keeping your eye trained on the dot, m ove the pencil slowly across the s heet to the le ft side, as sh own in Figure 8 .15a. 'What did you perceive? The image of the pencil jus t s""-ept across your retina from right to left, so asswning tha t your rightward-motion detectors are functioning correctly, you should have perceived movement in this direction. (You may have thought tha t the word righhvard in the previous sentence was a typo, but technically the image sweeps across the retina in the op posite direction from the actual m oveme nt, s ince the right side of the world p rojects to the left s ide of the retina and vice versa. To avoid confusion , we'll ignore this inconvenient bit o f physics for n ow and pretend that images move on the retina in the same direction that the objects in the world are m ovin g. Web Activity 6.& Eye Movements provides illustrations of these phenom ena tha t honor physics.) Now try a slightly different demonstration . Start with the pencil near the bottom le ft corner of the paper, fixate the eraser, and track it with your eye as you m ove it back across the sheet of paper to the right corner (Figure 8 .15b). Congratulations! You have justexe.:uted a type of eye movement called smooth pursuit, w hich kept the image of the pencil stationary on the retina w hile it was in motion. But think about what happened to the image o f the dot just above your pencil. When the pencil was on the left side of the page, the dot was to the right of your fixation point. As you tracked the pencil, the dotshifred to the center of your retina, and then it s lid to the left of your fixation point once the pencil read1ed the right s ide of the paper. However, you sho uld not have perceived the do t to be moving in this case, even though the image of the dot made essen tially the same journey across your retina that the image o f the pencil did in Figure 8.15a. We hope you 've convinced yourself that the retinal image movement of the pencil when you keep your gaze centered on the d ot {see Figure 8.15a) is essentially the same as the retinal image m ovement of the d ot when you keep your gaze centered on the pencil (see Figure 8.15b). The question is, Why do we perceive motion of the pencil in the first case, but perceive that the d ot is sta tio nary in the second case? TI1e reason is that in o ne case there's

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

A type of voluntary movement in which the eyes move smoothly to follow a moving dlject.

smooth pursuit eye

252

CHAPTER 8

superior colllculus A structure In the mldbfalnthat Is Important In Initiating and guldklg eye movements.

8230336 Ar

an eye m ovement. Think about how the visual system might accomplish this balancing act while we back up a nd describe eye 1uovements in a bit more detail (By the way, you may want to save the piece of paper with the do t for another exercise later in this chap ter.)

Physiology and Types of Eye Movements As Figure 8 .16a shows,six muscles are attached: to each eye, arranged in three pairs. These muscles are contro Ued by an extensive network of s truch1res in the brain. One way to get some inkling of the role of these brain structures is to stimulate them with small electrical s ignals a nd observe the movements of the eyes. For example, if a cell in the supertorcolllculus (Rgure8.16b) of a monkey is stimulated, the m onkey's eyes will move by a specific a mo unt in a specific direction. Every time that cell is stimulated., the same eye movement wiU res ult. Stimulating a neighboring cell will produce a different eye movement (Stryker and Schiller, 1975). (The s uperior colli culus also gets some input directly from retinal l'f'nglion cells; this input presumably helps with the planning of eye movements.) By contrast, in response to stimulation of some of tl1e cells in the frontal eye fields (Figure 18.16b) (and, indeed, certain superior colliculus cells too), the monkey will move its eyes to fixate a specific spot in space. Depending on where the eyes start, this adjustment may require an eye movement up, d own, left, or right. In this case, it is the d estination and no t the movement that is coded (Mays and Sparks, 1980; Schiller and Sandell, 1983). (a) Latera1 view

/

Top view

./Superior obliqu< • rectus

I

Inferior oblique

lnferiorreictus

FIGURE 8.16 Muscles oftheg.;e. {!!}Six musckisarranged in thr8'1 pairs (superior and inferior sup91'ior and inferior roctus, and medial and lateral rectus)-are attached to each !J'ffJ· '1:;i) Brain c irc uits for visualty guided saccades. {Part b after Wurtz, 2000.)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

VISUAL MOTION PERCEPTION

This d escription merely scratd1es the s urface of a motor system that is not only very complex1 but very active. For example, even when we try to h old our eyes completely s tationary, they continue to execute srnall but important move ments. Specifically, there are involuntary eye drifts and small jerks (microsaccades); if the eye muscles are te mpora rily paralyzed-say, as a result of taking curare (it's amazing what some people will s ubject them se lves to in the name of science]) (Matin e t al., 1982}-the en tire visual world gradually fades from viei,.v, Of course, in n onnal viewing. image velocities are fast enough to prevent fading \vithout the need for microsaccad es (Kowle r and Collewijn, 2010). So what are microsaccades good for? Recent work suggests tha t they m:ay be important for very fine s patial judgments, s u ch as threa din g a needle, because they precisely move the eye to nearby regions of interest (Ko, Poletti, and Rucci, 2010), and for compensating fo r the very rapid falloff of acuity even a fe\.v minutes outs ide the fovea (Poletti, Llstorti a nd Rucci, 2013). In additio n to invo luntary eye m ovements, the re are three types of voluntary eye movements. Most ob vious, perhaps, are the previously d iscussed sm ooth-pursuit 1novemen ts that we make w hen tracking a moving object, TileSe sm ooth-pursuit eye movements are often used by d octors as a simp le screenin g for neurological impairments,and they can he lp distinguish schizophrenic patients from others {Benson e t al ., 2012). Vergence eye m ovements occur w h en we rotate our eyes in\'\:ard (con verging the eyes) o r o utward (divergin g the eyes) to focus on a near or far object. The third type of voluntary movement

253

mlcrosaccade An Involuntary, small, Jerkllke eye movement. vergence A type of eye movement In whlcl1 the two eyoo move in opposite directions; for 9X81llple. both eyes turn toward the nose (convergerce) or

" /'lay l'om the nooe (diVefgenoeJ. saccade A fype of aye movement, made both Vduntarlly and lnvoluntarly. In v.tilch the eyes rapidly c hange fixation from cne cbject or location to another.

is the saccade:a fast jump (up to 1000 degrees persecond)(lla hill and Stark, 1979) of the eye that shifts our fi xatio n p oin t fro m one spot to another. We can d ecide to make a saccad e d elibe rately, but w he ther we' re thinking about it or n ot we w ill ma ke three or four saccades every second of every minute of every wa king ho ur of the day. Tha t's some thing like 3 saccad es x 60 seconds x 60 minutes x 16 h ours= 172,800 saocades per day- and tha t doesn' t include the s::iccad es we make during our dream s in rapid-eye-movement (REM) s leep (Moquet et., l., 1996). \'\ihen we view a scene, our saccades are n ot ra.ndwn. \Ve tend to fi xate the "interesting" pL.1.ces in the image. Thus, theeyes are more li kely to make saccades in response to contours than to broad fea tureless areas of an image (Figure 8 .17). Tnterest also h as a richer sem antic m eaning: We m ake eye movements tha t a re based on the content o f a scene and on our specific interests in tha t

8

F IGURE 8. 17

A classic scan path (.righ1;1 showing the pattern o f eye movements during Inspection of the picture of the girl on th19 li9ft. (From Yarbus, 1Q67 .)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

254

CHAPTER 8

reflexive eye movement A rnovement of the fJ'fe that Is automatic ard lnvolu ntary. optoklnetlc nystagmus (OKN) A reflexive eye movement In which the eyes will Involuntarily track a continually moving objoct.

scene (Yarbus, 1967). O ur pattern of eye movements as we en ter the cafeteria will be different if we' re looking fo r lunch than if we're looking for love. Tilere are also reflexive eye movements-for example, when. the eyes rnove to compensate for head and body movement w hile maintaining fi xa tion on a particu.lar target. These a re known as vestibular eye movements and operate

saccadc suppression The reduction of visual sensltMty that

movin g s moothly in one direction (e.g., right) in pursuit of the object moving in that sa me direction , and the n s nap back. T11e presence of OKN in response to moving stripes has often been used as a measure of visual acuity in infan ts.

occurs when we make saccadlc eye

movements. Sacca.die suppression eliminates tt'0t the bes t (lowest)

absolute thresholds fo r hmnan hearin g are beh\.•een 2000 and 6CXJO H z (2- 6 kilohertz [kHz]; rem embe1· tl'lat tl10se frequencies are enhai1ced by the physica l properties of the aud itory cana l). Tiuesholds rise on bo th sides of this range, meaning tha t h igher- and lower-frequen cy sound waves mus t h ave larger amplitudes in order to be heard .

audibil tty threshold

The lowest

scund pressure level that c an be reliably detected at a given frequency.

eCtml 1.xmtposition of sorni.ds. T11e sound-absorbing qualities of air dampen high frequencie.!! m ore than low frequencies, so when sound sources are far away, higher frequencies d ecrease in en e rgy m ore than lower frequencies as the sound waves travel from the source to the ea r. Thus, the further away a sormd source is, the ''muddier" it sounds. This change in spedral composition is noticeable only for fairly large dis tances, greater than lOCXJ meters. You experience the change in spectral composition when you hear thunder from near your v..-indow . or from far away. You hear th\mder as a loud ''crack# n earby, but thw1der from farther

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

inverse-s(J.Jare law A prin:lple stath)J that as distance from a scurce Increases. Intensity decreases faster such that decrease In Intensity is equal

to the distance squared. This general law also applies to optics ard other

forms of erargy,

302

.·/

CHAPTER 10

./

...... ........, ______ _

/

l=IGUR E 10. 11 The relatll/Q amounts of direct and rev9rb9rant energy coming from the listener's neighbor and the singer wHI Inform the llsten&r about the relative distanC9S of the two sound sour09s.

away sounds more like a ''boom." Note that this auditory cue is analogous to the visual depth cue of aerial perspective (see Chapter 6i aerial perspective involves that fact that more distant objects look more blurry). A final distance cue stems &om the fact that, in most environments, the sound that arrives at the ear is some combination of direct energy (which

arrives directly from the source} and reverberant energy (which has bounced off surfaces in the environment). The relative amounts of direct versus reverberant energy inform the listener about distance because when a sound source is dose to a listener, most of the energy reaching the ear is direct, whereas reverberant energy provides a greater proportion of the total when the sound source is farther away. Suppose you're attending a concert. The intensities of the musician's song and your neighbor 's whispered. comments might be identical, but the singer's voice will take time to bounce off the concert hall's walls before reaching your ear, whereas you will hear only the direct energy from your neighbor's wlllipers (Rgure 10.11 ). (See Web Essay 10.1: ReverberaUons and the Precedence Effect.) As it happens, reverberations appear to be important for judging the loudness of sounds. You do not judge that a coyote is howling softly just because it is far so how do you estima te how loud the howls really are? Recall that the sound to which you are listening rapidly decreases in energy, following the inverse-square law. However, reverberations do not fall off so quickly, because the surfaces that sounds bounc:e off do not move when the sound source becomes closer or farther away. Following a clever set of experiments that allowed them to dise.ntangle listeners' use of direct sound energy and reverberant energy, Zaho:rik and Wighhmm (2001) suggest that listeners maintain constant perceptions of loudness across d1anging distances by scaling direct energy relative to reverberant energy.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

HEARING IN THE ENVIRONMENT

303

Complex Sounds Simple sotmds like sine waves and b ands of noise are ve ry useful for explo ring the fund a mental op erating characterfsticsof a udito ry systems, ju st as sin e wave g ratings a nd sin g le-waveleng th lig ht sources are essentia l tools for vision researchers. But pure s ine wave to nes, like pure sin g le-wavelength lig ht

sources, are rare in the rea l '"'·orld, w here objects and events tha t matter to listeners are m ore cotnplex, m ore interesting.. a nd therefore more chal lengi ng for research ers to s tudy.

Harmonics Many environmen tal sounds1 includin g the human voice and the sounds of m.uskal ins trumen ts, have harm onic structure (Figure 10.12 ). h1 fact, harmonic sounds are a mong the m ost con.u11on typ es of SOlmds in the en vironment. TI1e lowest frequen cy o f a harmonic spectrum is the fundamental frequency. With na tural \·ibra tory sources (as opposed to pure tones prod uced in the laboratory), th ere is also energy at freq uen cies that are integer multiples of the hmdamental freq uen cy. For example, a fe.m::ile speaker may p rod uce a vowel so und \vi th a hmdam en ml fre que ncy o f 250 Hz. Th e vocal cords wi ll prod uce the greatest less at 1000 energy at 250 Hz, less ene.rgy at 500 Hz, less s till a t 750 Hz, Hz, and so on. ln this case, 500 Hz is the second ha nnonk , 750 is the third, and 1000 is the fo urth. For har moni c complexes, the perceived p itch o f the complex is determin ed by the fundamental frequen cy, and the ha rmonics (often called "overtones" by musician s) add to the perceived rk hness of the sound. The a uditory system is acu tely sensi tive to thefiltur.:ll relationships beh'''een frequenfy) is-rem oved harmon ics. [n fact, if the fi rst harmonic from a ser ies of har monics, as s how n in Figure 10.13, and only the others (second, third, fo urth, and so on) are p resented, the pitd1 that lis teners hear corresponds to the fund amental frequen cy---even tho ugh it is no t part of the sound l Listeners hear the missing ftmdamen tal. (See Web Activity 10.2: The Missing-Fundamental Effect.) It is not enm necessary to have all the other ha rmonics present in order to hear the missing fundam ental; just a few w ill do. The most s traightforward explan ations of the m.is.sin g-hmdarnental effect involve the tem poral cod e for pitch d iscussed in Chapter 9. One thing th.a t

fundamental frequency The ICM'est -frequency component of a complex pencxJlc sound.

Fundamental frequ ency

H.u monics

Frequ""')' (Hz)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

FIGURE 10.12 ManyenvifonmentaJ sounds. including vob9s, arG harmonic. The lo¥Vest fre· quency o1a hannonic sound is the fundamental frequency, and there ar9 p eaks of e ne rgy at Integer multiples of tM

304 CHAPTER 10

..

al l harmonics of a funda men.tal h ave in comrnon is fluctu,'\tions in sound pres.sure at regular intervals

=- - - -Missing fwtd.amental

cor responding to the fond am en tal frequency. For exa rnp le, the waveform for a 500-Hz tone has a peak every 20 mil liseconds (m s) (Figure 10.14b). The wavefom1s for 75()..and 1000-H z tones have peaks every 1.3 and 1.0 ms, respectively (Figures 10 .14c and d ). As sh o\o\Tt in Figure 10.14e, these three waveform s com e into aligmnent eveiy 4 ms, w h.ich,conveniently, h appen.s to be the period of the Ji.mdamen ta l for these three hnrm onlcs:

250 Hz. lndeed ,eouy Ho rn1orudoi2S0H z Wi!L have an energy peak every 4 ms. Some neurons in the auditorv nerve and cochlear nucleus '"'ill fire action potentia'.isevery 4 ms to therollec:tion of \Vaves sh mvn

in Figure 10.14e, providing an elegant mechanism to explain why listeners perceive the pitch of tlus

2 3 4 5 6 7 8 9 10111213 1415 16 1718 19 20 21 22 23 24 Hammnics

complex to ne to be 250 Hz, even though the to ne has no 250-H z component.

- - - - - - - - - - - - - - - - -.. 600{)

Frequency (H z)

FIGURE 10.13 If the fundamental (lowest frequency) o f a harrronic sound ls remov'9d , listQflefs still t\Qar the pitch of this "'m issirrg

Timbre Loudness and pitch are reJ ati vely easy to d escribe, bee.:,use they correspond fairly well to simple acoustic d imensions (arnplitu d e and frequent..)', respecti vely). Bu t the richness of co mp lex sow1ds Hke those fo und in our world depends on more than

(a)

(b)

I I

V

M.issing fundament,1-l

Ii>

:i!

(r)

""]

:i!

" ; Harmonics - - - - - - - - F,. -q-uen-o-.v -(H_z)_ _ _ _ _ _._ 6c:m

l AOAAOAAAAAAOAAO

(-dl, listeners still h ear the pitch of the mlssing-fundrun.:rrtaJ frequency (a) because the harmonics share a common B"lergy ftuctuation 4 rns, the perbd o f a 260-Hz signal (e).

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

4ms

HEARING IN THE ENVIRONMENT

simple sensations of loudness and pitch. For e..xarnple, a trombone and a ten or saxoph on e might play the sam.e note (that is, th ei r hvo notes will have the exact sa m.e fundamental frequency) at exactly the same lm.idness (the sound waves will have identical intensities), but we would have no troubl e d iscerning that h vo different instnunents \Vere being played.11-le percephtal quality that differs between these ti-1;·0 rnusical instruments, as well as v°'wel sotmds s uch as those in the words lrot, heat, and hoot, is k:no\oVTl as tinibre. (9ee

305

timbre The psychol::glcal sensation by which a listener can Judge that two

sounds with the same loudi1ess and pitch are dissimilar. Tlrnbre quality Is

con'leyed by harmonics and other high frequencies.

Web Activity 10.3 : Timbre.)

What exactly is titnbre? You won 't find a good answer to this questi on in the dictionary, beca use the official definition of timbre is /(the quality that makes listeners hear n.vo different sotmds even thou gh both sounds have the same pitch and loudness" (Am erican Standards Association, 1960). However, differences in timbre between musical instnunents or vowel sotmds can be estima ted closely by of the exten t to which the overall spectra of two sounds overlap (Plomp, 1976). Perception of visual color depends on the relative levels of energy at different wavelengths (see Chap ter 5), and very si milarly, perception of timbre is related to the relative energies of different acoustic spectral components (R g ure 10 .1 5). For example, the trombone and tenor saxoph one notes \vhose spectra are plotted in Figure 10. lSn share the same fundan1ental frequency (middle C, 262 Hz). However, notice that the trombone's third (786-Hz) component is s tronger tha n its fourth (1048-Hz) component, whereas for the saxophone, the relationship between the energies of these h) producoo byth.:i same female talker, also with a fundamental fre· quency of 262 Hz. sound very 10 listeners.

306

CHAPTER 1 0

Sensation & Perception in Everyday Life Auditory "Color" Constancy Once we take off our headphones and get on out into the ·world , listening environments can dramatically alter the sounds that anive at our ears. ·we already learned ab::>ut the importance of revert:eratlcn energy for perception of distance and lcudness of sounds. Even the way you

da:::crate can charge :sounds on their way to your ears. Higher frequencies are reinforced by hard surtaces such as tile floors and concrete walls, but they are dampened by soft surtaces such as thick carpet

and curtains. Thls is much like the problem of oolcr constarcy that you learned about in Chapter 5. In vision, the goal Is to perceive the same colors even thOLtJh the spectrum of illurnlnation can be quite different

depending on the type of lighting (sunlight, incandescent. and so on). In hearing, sutiaces in the en\'ironment refle:::t and absorb energy at different frequencies In W"fS that change the spectral shape that finally anives at you r ears. Klette and Kluender (2008) used the different spectral shapes of the vowel sounds ·ee· and ·oo' to learn how heating calibrates for changes In the listening environment. They took advantage of the fact that the spectra for 'ee ' and 'oo' differ in t\.vo lmp:>rtant '>N'iflS (Fjgure 1 0.16a). First. the vowel sound ·ee' has a relatr.telyftat spectrum with almost as much energy In high frequencies as In low frequencies (upper lefi in the figure). In cmtrast, energy in 'oo' is dcrnlnated by low frequencies w ith less and less energy as frequency increases (lower right). We can S"f that the "tilt" of the ·ee' spectn.rm is much flatter than the tilt of 'oo. · Second, the sounds 'ee' and 'oo' also differ In where peaks are present ln their spectra. The second peak In the 'ee' spectrum Is at a hi gher frequency than the second peak in the ·oo• spectru m . Listeners use both tilt and the frequen:;y of this second peak to tell ·ee' from 'oo' (Figure 10.1 Gb) (Klefte and Kluender, 2005). To learn how the auditory system adjusts for listening context, Klefte and Kluender (2008) created stimuli that enabled them to separately measure how much tilt and second -peak frequency cmtr1bute to the perception of these vowels. Then they had listeners Identify vowels an after healing a sentence like Myou w ill now hear the vows." but interesting tv.;st. They created some sentences so that the overall tilt

of the sentence was the same as tl1e tilt of the foUowirQ vowel, either

8230336 Amma Appa

'ee'-like or 'oo'- like. To other sentences they added a peak in the spectrum, all the way through the sentence, that matched the frequency of the second peak in the vowel ('ee' or 'oo') that listeners would Identify. Usteners heard the very same vowels in dramatically different ways depending on vl'hldl type of manlpulated sentence the vowels. When tilt ·was the same for both the p-eceding sentence and the vowel, listeners used only the frequency of the second peak t o identify the vowel (Figure 10.16c). \M"ren the second peak was present all the way through th3 p-eceding sentence, listeners relied mostly 01 tilt to identify the vowel (Figure 10.16d). Listeners calibrate for reliable spectral charactertstics of a listening caitext much as observers calibrate for the spe::;trai oomposltion of illuminating light when percei.,;ng color (Alexander and Kluender, 2010 : smp et al., 2010). IM"ren the brain int erprets spectral tilt as a consequence of the em'ironrnent in vJhlch a sound is heard (concrete versus curtains), listeners igncre tilt and use only t he seoond peak. Conversely, when there seems to ah.'v'3fs be energy around the second peak in a listening emironment , listeners use only tilt to decide w hich vo·wel they heard.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

HEARING IN THE ENVIRONMENT

FIG URE 10.16 Listeners use both spectral tilt and the frequencies o f spectral peaks to identify vowels. {a) SPQCtra o f natural ' 9€1' and ·oo' are shown In th t.l>ll'l' a oq right, respectively. The vowel 'ee' has a relatively flat spectrum. a"ld 'oo' has a tilted spQCtrum, SLCh that there is muc h less energy at higher freq.JQn· cies. The second spectral pgak. is lower in fr9QU9"C)' for 'ex>' than for 'lile.' ExpGTimGntally manfpulatOO vowel sounds are also stown with the tilt of ·ee' and second peak of 'oo' in the Llppeir light and the tilt of 'ex>' and second peak of ·ee' in the lower left. (b) Listgners use both tilt and fre· quency of the second peak to identify vow9's.. Lighter ar9aS correspond to sounds that subj.:cis Mar as 'ee'; darker areas represent sounds that listeners h'11ar as 'oo. · (c) Upon hearing a vowel atte< a sentence that has the s ame tilt as the vowel, listena-s use only the frequency o f the second peak to identify the vow9' . lp') Ho\11/ever, when the pr9C9ding sentGonca Includes a peak at the samg frequency as tt1e second peak in the VO'N91, IBteners use mostly tilt to identify thQ \'OW'QI.

lo-

(b) Tes! vowel in isolation

....

(l

·oo•

.....

"' "'

'ee' j

307

"'

20 10

20

10 Hybrid: Natural 'err!

-10

Tilt

0

.

:s ·oo'

a:ro

1000

-10 300:>

4000

.scoo

1000

"°50

"°

40

40

10

20 10

"'

•OO• tilt, 'et!! peak 1000

200)

-10

"""

4000

5000

Frequency (Hz)

-3.9

-3.9

NaturaJ 'od 1000

n:o

?(.(()

4000

5COJ

Frequency (Hz) (d) Test vowelafta- senteru:ewithsame frequency

(c) Test vowel after sentet\C2 with same tilt -2.4

....

Hybrid:

-10

-2.4

axo 30CO 4000 .scoo

"1

] "' 20

j

'el!! tUt, 'oct petn brought a lette C D E" F G A E'> C O E: f C A E'> C O E" P C A B

c DE: r e

A 8 CD[ F

FN!qt1etlC)'

FIGURE 11 .1 The sounds of music ext€f"ld across a freq uQncy range from abc:ut 26 to 4200 hertz {Hz).

octave The Interval between two oound frequencies having a ratio of 2:1.

tone height A round qualitycorrespmdlng to the level of pitch. Tone height Is monotonlcaly related to frequency. tone chroma

A sound quality

s hared by tones that have the same

octave lntervaL

COE F G A B (OE F

nma Appa AB C

(Hzj

TONE HEIGHT AND TONE CHROMA Musical pitch is on e of the characte ristics of musica l n o tes, the soun ds that con sti h.1te m elodies. A very important concep t in lmderstanding musical pitch is the octave. \Vhen we described pitch (or psycho..1cou st ic pitch ) in 0 1apte r 9, it seemed dear that the nearer

any hvo sounds \'.:ere in frequency, the nearer they were in pitch. Here's where octaves come in. \ Vhen one of hvo periodic SOlU\ds is dou ble the frequency of the o ther, those n.vo sounds are one octave apart. For example, midd le C (Cd) h as a hmdamen tal frequen cy of 261.6 he rtz (Hz). Notes tha t are one octave below a nd above midd le Care 130.8 Hz (C3) and 523.2 Hz (C 5), respectively. Not only d o these three so unds have the .same names on the mus ical scale (C), but they also so und similar. In fact, C 3 (130.8 H z) sounds more similar to Cd (261.6 Hz} than to a sotmd with a cl oser free tion between 'aba' 6eft) and 'apa' (right) includes at 19as1 16 acwstic differences. Som e diffEfEf'ICes that e asy to include duration o f the 1irst

vowel. duration of the interval between s)oilables, and the presence of low-fre· quencyenergy in the middle o f 'aba.'

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

MUSIC AN D SPEECH PERCEPTION

339

in the brain d o bes t is integratin g nuJtiple sources of informa tion to recognize

patte rns {Kluende r and Alexnnder, 2008). Computer simulatio ns of neural connectivity (n eural ne t work m odels) d esigned to min1ic the use of multiple attributes and associations be tween a ttributes are arnong the m ost s uccessful artificial pattern recognizers yet invente d . lnterestingly, rnany neural ne h'llork m odels (e.g ., tl1ose d eve loped by J. A. Anderson et a l. [1 977) or by Dampe r and Hamad [20CXJ]) also s how ca tegorical-perception effects w hen the rno dels are trained on en ough exan1ples.

Learning to Listen In our discussion of vision earlier in the book, \•vesm"'· tha t experience is incredibly imp ortant for visual perception, particularly the higher-level perception of objects and events in the world. Experience is every bit as imp orta nt for auditory perception, especially for the perception of speech. And tui.like vision, experience w ith speed1 begin s very early in development. In fact1 babies gain significant experience w ith speech even before they' re bornt Measurements of hemt rate as a n indicato r of the ability to n otice change between s peech sounds have revealed that late-tenn fetuses can discriminate between different vowel .sounds (Lecanu et et a.L, 1986). Prenata l experien ce with speech sounds appears to have consid erable influence on subsequent perception. For one thing, a newborn prefers hearin g he r m other 's voice over other women's voices (DeCasper and Fifer, 1980). When 4.-day-old infants in Paris were tested , they preferred hearing French instead of Russian (Mehler et a l., 1988). Perhaps m ost amazingly, newbo rns prefer heari ng particular child ren's s tories that v..iere read al oud by their m othe rs during the third trimester of p regnancy (DeCasper and Spence, 1986).

As we h 1we seen, speech sounds can differ-fu..inany> a . Aroustfc d.f.fkrences tha t 1na tter critically for one language may be irre levan t or even distrncting in an o ther language. Fo r example, the English language makes use of the distinctio n between thesmmds 'r' and 'I/ whereas these two sounds are both very si milar to only one sound (cal led a ''flap ") in Japanese. As another example, Spa n ish is one of many lang uages that uses only the five vowel sounds 'ee' (as in beet), 10 01 (as in boot), 'ah' {as in bomb), 'a'/ (as and 'oh ' (as in bont), whereas English in employs numerous other vowel sotmds. As an yone w h o has s poken to a native Japanese spea ke r in English knows, the ' r ' / 'l' d istinction is very difficult for Japanese people to pick up when they learn English as a second lan g uage (Figure 11 .19). Beca use the difference between '1' and 'r ' is irreleva nt to native Japanese speakers when they're learnin g their native lan gu age, it is adaptive fo r them to leam to ign ore it1 thus all ow ing them to focus o n speech sound distinction s that are important in Japanese. When people have spen t m ost o f their lives listening to Japanese and not hearing the diffe1-ence between 'r ' and 'I/ we are n o t surprised that they have difficulty learning to produce the 'r ' and 'l ' By the same token, a nati ve Span ish speaker who complains that your d og just "be.:-tt" him is probably not d,Umi ng BECOMING A NATIVE LISTENER

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

FIGURE 11.19

How we hear speech

sounds d0pends on our experienc e w ith the SP9Qeh sourds o f our first language, Because o f experience w ith one language, it is often diffic ult to pef· c eive and produ:::e distinctions in a new language. For @rampe, mo st Japanoo.e piecple learning English as a S9cond language have trouble distinguishing between 'I' and 'r ,' both whai th91 lis· tai and w hen thgy speak.

340

CHAPTER 11

100

tha t Rover threw a punch; rather,, the difference betv.·een 'ee" and 'ih ' is less percep tible to the Spanish s peaker, beca use both of these English so unds are simila l' to the Spanish 'ee.' Interesting ly, s tudies s how tha t infants begin filtering out irrelevant .5"' acous tic differences long before they begin to utter speech sounds (even before their babbling s tage). One study found that by 6 months o f age, infants from Seattle were more likely to notice acoustic differences that di sting uish h vo English vowe ls than to notice el1uiva lent differences behveen Swedish vowels, and infants from Stockholm were more likely to notice the dlfferem.-e . between h'l-·oS\·ved ish vowels than the difference twoJlnglish vowels (Kuhl et al., 1992). Twllng of perception for consonan ts appears to take a bit longer to develop, but by the time infants are 1 ye.u old, they have also begun to ign ore, jus t as their parents do, n - 12 6-8 S-10 10-12 consonant distinctions no t used in their native language (Figure 11.20). months months months months O f course,, it is possible, w ith much training, to learn to perceive and produce speec h sound distinctions that you've spent most of your life Eng.Ush infants Hindi infants igno ring. As yo u might expect, the longer a person uses only her first FIG URE 11 .20 Hindi h as a dental language, the longer it takes to learn to produce and percei ve sounds stop consonant produc&d w ith the from a second language (Flege, Bohn, and Jang. 1997i Imai, Flege, and Waytongue tip touc hing th9 t199th and a 1..'l nd, 2002). Many s tud.i es have been aimed at determining w ha t makes ne\>v retroflErx stop consonant that requires distinctions hard or easy for second-la ng uage learners to pick up. Learning the tongue t o bend up and b ack in the rnouth. Adult speakers of English hear is most difficult w hen both of the sounds in the second lang uage .are similar both o f thoo.e sounds as 't' because to a single sorn'd in the firs t langi.Mge (e.g., 'r' and 'I' for Japanese speakers they are sirnilarto the English 't .' \'VhQn learning English). Learning is easier if the h\'O new soi.mds are both unlike Weirker and TMs (1 9 84) testoo infants any sound in the native lang uage. Fo r example, na ti ve English listeners have from English·spgaking famili9 s In Vanno probletns distinguishing dick sounds from Zulu because Zulu clicks are so couver, the youngest (6-8 m onths old) re liably respondeid to the difference unlike any English sounds. Leaming a lso is easier if t\.,,.·o new so unds from a betwwn thQiSQ two Hindi soun ds. By 1 new language diffe r in the same vvay thLlt two sounds from the firs t lan guage year o f age, h owel/l'::f. the infants were differ. mimicking th9ir p arents' b 9havior by Picking up the distinctions in a second languas;e is easiest if the second ignoring the distinction betwee) for a si11gh3 semK;;ircutar-canal afferent n.euror1 at 0.05 Hz. Note that the neural fir1ng rate

inc r98S'9s and decreasEas nearly in tandem with the oscillating stimulus. (c) When the sensitMty calculation is repeated at differi:int stimulus frequ€'e see with o ur eyes and hear \>1r-i th o ur ears, the primary som ce for o ur perception of spa tial orien tation rem ain ed a mys tery. In fuct, in the eigh teenth cenhlI)'i gross fluid shifts in the head were accepted as an ex plan.a ti on for the source of otlI sense of spatial orien ta tion. O n ce it was established that this sense degmd es w hen the vestibulu system. is dam aged, it became dear that the vestibular system p rovides crud.11 information regarding spatial orientation. For ex ample, w hen patients with vestibular loss are rnoved in the dark, they have a mud1 m ore difficu lt time correctly perceiving the ir motion than do peop le w ith normal vestibu lar function. Today, three different techniques are frequently used to investigate s patial orientation perception: thresholds, magnltude estima ti on, and matching. For a threshold study, we ca n ask, \Vht' t is the minimum m o tio n (the threshold) required for cor rectly perceiv ing the direction we are m oved? Note that this is different from simply reporting whether we've m oved , sin ce vibration can pro·dde a mo tion cue wi thout info rmin g .about the d irection of th e m o ti on . The vestibular system tells us m ore th an w hether motion is p resent or n ot; it actually informs us of the direction of m otion- fo r example, \'\>·hether we moved to the left or to the right. In a magnitude estimation stud )i subj ects might be asked to give verbal reports of h ow much they tilt, rotate, or trans late, usin g ph ysica l mUts like the num ber o f degrees they rota ted . Alterna tively, magnitude est im ation may utilize arbitrary scaling. For example, subjects may be trained to rotate a knob in proportion to their perceived velocity o r provide a verbal indicator of their velocity on a scale of 1-10. In a matching task, s ubj ects might be asked to align a visu al line w ith perceived Earth-vertical (wlUd1 wny is "down''?). [n such a task, called the "subjective visual vertical" the in vestigator could produce a vestibular sti mulus by tilting the s ubject. That s ubject would be provided with a visible line in otherwise dark s urrotmd.ings and with the ability to rota te the line to the perceived Earth-vertical. Alternative!); haptic sensation, which you \vill learn n10re about in Ompter l3r could be utilized ins tead of vision. For this technique, s ubjects might be asked to use their sense of limb p osition to align a bar tha t they h old- but cannot see-with perceived ver tical. FURTHER DISCUSSION of similar matching in connection with the can be found in Olapter 13 on pages 417-418.

sense o f touch

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE VESTIBULAR SYSTIEM AND OUR SENSE OF EQUILIBRIUM

FIGURE 12. 16 The re d line on this graph s hows the angular velocity of a person at rest, with gyes closed, w ho was suddenly rotated at a coostant speecl for 60 seconds and then was abruptt>,r return9d to r.eet. Initially, th9 subject's estimate of angular velocity was accurate (purpl9 line at time 0). but then the percaiwd vefoc· ity dropped to near 0 after about 20 S€1Conds. stopped. the subject perrOOved an abrupt w loclty In the opposite directioo, which mimic k.9::1 the initial P9f'C9ption but in the opposite diroctlon. For comparison , WQ shOIN a reprBSentatlon o f the canal rooponse in bll1e, slmllar to that ::.tiown in Flgise 12. 12. No te thefast9r return to for the canal response. (Aftef Young, 1984 .)

/Actu al ,·elocity

"'SemicircuJ,1r can al respon se

10

'1l

3IJ

40

50

367

70

I'll

Tune (s)

Rotation Perception U yo u are spun o n a barstool in the dark a t a nearly co nstant velocity, you w ill initially perceive an angula r velocity tha t is roughly the same as the actua l rot.."ltion. H owever, if the constant-velocity rota tio n lasts more than a second or two, you wlll p erceive tha t you are s lowing d own (Figure 12.16). If the consta nt-ve locity rotation continues for more than 60 seconds o r so, you w ill perceive that you are no longer rota ting. This descrip tion may rem ind you of the way the response of the semicircular canals d ecays during consta nt-velocity rotation (see Fig u re 12.12), a nd it is another example of how the vestibular sys te m is a ttuned to changes in mo tion. Interestingly, th ough, the time course of the perceprual d ecay is m ore gradual than th.a t of the semicircular-canal s ignal sent to the br(1m. TIUs effect is so1netimes ca ll ed ;'velocity storage" because the perception of rotati on persists after the a fferent signa l from. the semicircular c.:1 nals h as dissipated . The veloci ty storage phenomen on is interesting and importan t because it shows that the brain has improved on the incoming sensory informati on to yield a rotn tion perception that-while far from perfect- is closer to the actual rotation than if the pe rception s im p ly fo llowed th e time course of the semicircular-canal afferent signal (Berto lini et al., 2011). Later, if you are abruptly brought to a stop fo llowing ex tended rotation, you ,.... ill perceive an an gular velocity opposite the one you experien ced w hile rotatin g (see Figure 12.16). This rotation illusion is one that m any of us p layed with as children wh en we would spin ourselves for a w hile and then s uddenly stop spinning and try to stand or walk. (A nd som e amusement park jtmkies still play with this illusion as you 'll see at the end of th is d1apterl) The dizziness and Imbalance th at we experien ced when. we stopped rotating were due to an illusion o f self-rotation caused by the se1nicircular-cana l response. One can develop an intuitive tmders tanding of the reason fo r th is illusion by considering the analogy of riding in a car. When you 're riding at a constant velocity, you and the car are moving together. But when the car suddenly stops, you are throvvn forv•.rru-d beca use you have momenhun and keep m ov ing even thou gh the car has s topped. When you're rotating at a constant velocity, there is little o r no hair cell deflec tion, because the endolymph and cupula are moving together. \ Vhen the rotation is sudd enly h alted, h owever, the cu.pula s tops moving quickly but the endoly mph has m omenhun and tend s to keep rnovi.ng. The hair cells are therefore d eflected, the direction of the hai r cell resp onse is opposite the one measured when the cons tant-velocity rotation bega n. H ow sensitive we to rotation? Direction recognition threshold s for ya\.v rot..1tion have been m easu red for rotational motion frequencies ranging from 0.05

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

dizziness A commonly used lay term that rK:flspedfically ln:llcates any form of pEfceti.red spatial dlsorlenta· tlon, with or wlt11out lnstablllty. Imbalance Lack of balance; urstead lness; rearly faHlrr;i over.

368

CHAPTER 12

FIGURE 12.17 Mean v elocity threshold as a function of frequency for seven subjects. Threshold wlocity was the peak vo:iloclty a::::hieved during a single cycle of slnusddal accel9t"ation at which subjiects correctly recognized the direction o f rno tbn most o f th.e time. The plotted cu rv e shows a rnod'111 frt to the data. The inset shows the angular acoel9ratbn, ang ular w locity, and angular displacement for a single cycle of slm.JSOidaJ acceleration at 0.5 1-tz. For the QXafOpte sho wn, the peak magnituOO of the aci:geration is 1.57 degrees/s2, the p::iak velocity is 1 degreek., and the peak dispacerrient Is 1 degree. t;After Grabherr et al., 2008.)

1 !

-2 Velocity

4

31

j

0.5

0.02

005

0.1

0.2

1.5

0.5

Frequency (Hz)

Hz {one back-and-forth cycle of y m. . ·accele rntion in 20 second s) to 5 Hz (five back-and-forth cycles of acceleration in l second ). For frequencies above 1 Hz {one back-and-forth accelera tion cycle in 1 second), direction recognition thresholds are rou ghly constant; your head has to be m ovin g at a sp eed o f just a little below 1 degree pel' second , w hich a t 5 Hz corresponds to a head an gular di splacement o f just 0. 1 degree or ju.st 0.02 of a minute on a dockface! C learly, we are very se ns itive to rotation. Por freq uencies below 0.5 Hz (one back-and-forth oscilla tion in 2 seconds), thresholds increase wi th d ecreasin g frequ ency (Ftgure 12.17). The important poin t here is to recognize that rotation thresholds vary as the freq ue ncy of the angulM acceleration stimulus varies.

(Recall tl1at hearing tluesholds changed wi tl1 frequency too [see Figure 9.22] tho ugh the cause of the c hanges is different for th e tvvo modalities.}

Translation Perception

mathematlcal integration Computing an Integral-me of the two main operations In calculus (the other, the Inverse operntion, Is differentiation). Veklclty Is the Integral of acceleration . Change of position Is the integral of velocity.

When subjects a re p assively translated sh or t d istances while seated in a chair in the dark and then asked, w hile s till seated in the chair, to u se a joystick to actively move the di.air to reproduce the distance that they had been passively h'anslated, they do so accurately. But even though n ot C\Sked to do so, they also reprod uce the 7.'elocity of the passive-motion trajectory (Be rthoz et al., 1995). The unrequested replication of velocity s uggests that the brain remembers and replicates the velocity trajectory. E..1.rlier we sai d that the o to lith org.:'l n.s tra nsduce linear acceleration, which is the change in linea r velocity. Therefore, replication of the velocity trajectory m eans tha t the brain also see ms to mathematically Integrate the acceleration signal provided by the otolith organs to y ield a p erception of linear velocity. Th.is apparen t calcula ti on suggests that w hile otolith organs .sense linear acceleration , our brains tum this i.nfom,ation int o a perception of linear velocity. As for yaw rota tion, directi on recognition thresholds for s ide- to-side (ynxis) transla tion h ave been measured for m oti on freque ncies ranging fro m

0.05 Hz fo•)e b..ldc-ond-forth cyde of

in 20 seconds) to S Hz (fi ve

back-anCl-forth cycles of ac le rafion m 1 second). For frequencies above 1 H z (on e back-a nd-for th cycle in 1 second), direction recogn ition thresholds are ro ughlyconstnnt. To sense tmnsl1tion direction youl' head has to be

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE VESTIBULAR SYSTIEM AND OUR SENSE OF EQUILIBRIUM

369

moving at least a t a sp eed of 5.0 mm pe r second, \·vhich a t 5 H z corresponds to a head an gular d isplaceme nt of jus t 0.5 mm! Clearly, we are also very s ens itive to trans btion, As for yaw rotation, for frequencies below 05 H z {on e back-and-forth oscillation in 2 second s), thresholds increase \vi th d ecreasing freque ncy (Va lko et a l., 2012).

77/t Perception How wel l d o we perceive OlU' tilt w hen we are s la nted away from tn1 e Earthvertical? For tilt angles less than 90 degrees--thatis, body orien tations ben.-.·een standing up (0 degrees) and ly ing d0\¥n (90 degrees)- we .:'lre pretty good according to a va riety of magnitude esti rnation techniques. Observers p rodu ce relia ble a nd consistent answers if they indica te their perceived tilt verbally or ii they align a handheld probe with perceived vertical (Figure 12.18). We are not perfec t, thoug h. Some con sistent e rrors appear, especially in the s ubjective visual vertical task described nea r the beginning o f the ''Spatial Orientation Perception " section. Back in 1861, He rmann Rud olf Aubert fm md that w h en he roll-tilted his head to the left or right whiJe loo king at a ve rtical streak of light, the vertical Line appeared to tilt in the direction opposite his head tilt. For tilt angles less than 90 d egrees, the illusory tilt e rror is typically a bout 10 d egrees, but the app aren t tilt of the vis ua l line cc1 n be as large as 45 d egrees w hen the hea d is tilted 135 d egrees. You can investigate this i!Jusion in a completely d..1rk room by lea vin g th e d oor open jus t a crack-jus t

IE-OL Hapticallyindicated tilt

(11") Pitch Backw,ud

,;.....

100

0

] ..-. 50

j

0

- lOO

Q

• ,,

-50

Back- - 100 ward

50

100 Forward

(c) Y.lw

Right

-50

0

50

Actual tilt (degrees)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

Left

FIG UR E 12.18 Subjects ar9 genorally pretty good at Indicating how much they are tilted. Data (opM c lrc lf'). and yaw (c) tilts. P€1rc9v9d tilt roughty" reflects the actual tilt (solid lines) for aH three tilt directions. (Aft.er Bor· tolaml ot al.. 2006.)

370 CHAPTER 12 enough to see a line of light in the doonvay without illuminating the room at all-while you hold your head in a tilted position. Does the line appear to tilt w hen your head is tilted? Does it move in the direction of your tilted head or in the opposite direction? Titresh olds for recogniz ing the direction of tilt show that normal s ubjects correctly re port the direction of a static tilt w he n tilted about 1 d egree off vertica l in the dark. This sensitivity serves our ability to s tand upright1 because the farther we are tilted from upright, the m ore difficult standing up is. For example, the amount of muscular effort required to maintain posture roughly d oubles if we a.re tilted 2 degrees in.stead of 1 degree, so just imagine the effort of trying to stand tilted 20 d egrees away fro n1 upright[

Sensory Integration 1he senses d o n o t opera te independently. Ins tead , the brain combines signals from differen t sensory sys te ms via neural pr0t."e.Sses of sensory Integration. Fo r example, visua l cues influen ce sound localiza tio n-an effect u sed by \o"entriloquists. Vestibula r signals combine wi th information from nWTi erous sensory syste ms to p rovide us with a n understandin g of the p osi tion a nd movem ents o f the a nd body. FU RT HER DISa.JSSION of a remarkable example of sensory integra· tion-the McGurk effect on speech perception-can be fourd in Chapter 11 en page 336.

Visual-Vestibular Integration

sensory integration The process of comblnlr.g different serroory signals. Twlcaly, combining several signals

yields more accurate and/or more pre-

cise Information than can be obtained Iran lrdMdual sensory signals. This Is oot the mathematical process of Integration learned ln calculus (e.g., the Integral of acceleration Is VelocltY). vection

An Illusory srose of self-

motion caused by moving visual cues when one is rot, In fact. actualty

moving .

Most of us have experien ced illusions of self-motion ca used by moving vis ual 1..-ues. Perhaps you 've perceived self-motion w h.ilewatchin gan IM...l\X movie. Or perh ap s you've felt as if you were moving backward w he n you were stationary but the car (or train or bus) next to you began to 1nove forward. Or perhaps you felt unsteady w hen standing on a bridge looking at the water fl owing beneath. A ll of these situa tions le."td to pen:"eptions of illusoiy self-motion called vec. tlon. \'ection can be very compe!Ling. For example, drivers stopped in traffic often press harder on the brake pedal w he n they perceive that they and their stationary car are moving, even though it is the cars around them that are moving. To cons ider hO\·V vection contributes to spatial orienta tion , imagine a person s tanding upright w hile vievving the ins ide o f a sph ere rotating about an Earth-horizontal axis (Figure 12.19a). At first, subjective perceptions match reality; human s initially p erceive that th ey a re s tationary and that the s phere is rotating. But if they continue to observe the rotating visual d isplay for 10 .secon ds or so, they us ually begi n to perreive that they're rotating in the direction opp osite the sph ere rotation (Figure 12.19b). lhis illusory ro tational vection demonstrates the crucial contributi ons of vision to our sense of self-rotation . ln fact, signals rela ted to vision con verge with th e se1nicircular-canal signals in the vestibula r nuclei, w hich is the firs t place in the brain tha t vestibular infor mation reaches. (Web Essay 12.2: Canal·Otollth Integration discusses a s inUlar interactio n of s igna ls from the semicircular cetn.:'1ls and otolith organs.) Paradoxically, subjects experiencing such ro tational vection almost never report that they are tumbling head ove r heels as they wo ul d if they truly were rotating to the extent that they perceive. In fact, s ubjects typically experience a simulta.neo tts illusory sensation of tilt that graduaUy builds up to a relatively cons tant level (Rgure 12_.1 9c). Theseperceptions--experienced as a sensation of motion \11i--ithou t getting O:Ul}""--Vh ere--are rontrad ictory,since we canno t be rotating

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE \IESTIBULAR SYSTEM AND OUR SENSE OF EQUILIBRIUM

371

(a)

relative to gravity (as suggested by the visual a.1es) while also maintaining a constant orientation with respect to gravity (as indicated by the otolith organs}. ln addition to exemplifying sensory integraticu, this sensation of motion without getting anywhere demonstrates that our sense of spatial orientation is not constrained to combinations of motion and orientation that are physically possible. In this example, the role played by the vestibular system is to put the brakes on visually induced vection. Patients suffering from severe vestibular damage generally report greater vection than do normal s ubjects.Astronauts experiencing rotational vection in space-in the absence of gravitational signals--report a head-ovet'-'h eels twnbling sensation that is a bsent on Earth. These findings are explained by the fact tha t neither the patients n or the astronauts receive normal gravitational cues fr om the o to lith organs to contradict their visual

rotational cu es. (Web Essay 12.3: Space Motion Sickness discusses another aspect of spaceflight that many astronauts experience.) Since illusory motion is greater when there are no o tolith cues to contradict the visual cues, '\o\"e can infer tha t, under nonnal circumstances, infonnation from the vestibular system is combined w ith visual information to }';eld a "consensus" about our sense of spatial orientation.

FIGURE 12.19 Rotational vectlon . A subject views a \Jlsual dlspla'f rotati ng In roll. For demonstration purposes, the visual display Is shoW1 hQrg as trans· parent. '8) lnitlaHy, the subjoct correctty senses tha.t she Is stationary and the visual display Is rotating. '1:;i) The subj9ct begins to peroetve vectlon, a sense o f self·rotation in the dlrectlon opposite the rotation of the visual dlspla'f. (c) Roll V9Ctlon ls often accompanied by an Hlusion of roll tilt that is induC8d by the peroetved roll rotation. The d!rectlcn o f the percelVQd tilt Is consistent 'With the tilt direction that 1NOuld occur if thQ subject Wf!f9 truly rotating In roll. CAfter Young, Sh&lhamer, and Mod&stino,

1986.)

Active Sensing Our sensory systems are simultaneously activa ted

as the result of our ow n actiol\s and changes in the external world. Indeed, m ost of our sensory experiences are gained by active exploration of the world resulting from locomotion, eye m ovements, tou ching, and other interactive activities. Our brain's ability to d istinguish sensory events that are self-generated , sometimes called sensory reafference, from those that arise externally, sometimes called sensory exafference, is essential for perceptual stability and accurate m o to r control. For example, w hen our eyes move, the im.age of the world m oves across our retinas; yet we do n ot pe.rcei ve the image of the world as m oving. In order to avoid responding to sensory inputs that arise from self-generated actions, the sensory system needs to know what the motor system has d one.

Based on their observations, Von Holst and Mittelstaedt (1950) proposed the

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

sensory reafference Olange In afference caused by self-generated activity. For the vestibular system, ves tibular afference evoked by an active head motion wruld yield smsory reaffermce.

sensory exatterence O'lange In afferenoe ca.used by external strnull. Fe< the vestlt:>tJar system, vestibular afference evoked by pas sive head motion would yield sensory

exafferenoe.

372

CHAPTER 12

(a)

Vestibular

Estimated

Estimated

.Ufe1l:'nce - vestibular = veslibufar

Estimated vestibul.u reafference

reufference exafferenoe

(c) Active

Ve&tibular Estimat.ed. E'sti.nlate head . ln Figure 12.21b, the eyes counterrotate in the head, which helps stabilize the visual field on the retina. If the eyes do not counterrotate in the head, the retinal image tends to blur during head rotation. (Remember how your fingers blurred when you rapidly moved them back and forth?) The COWlterrotation of the eye in the head helps reduce this blur by reducing the motion of the image of an object across the retina. Note that because of eye counterrotation, the actual rotation of the eyes with respect to the external world is much less than that of the head. The most direct neural path for the VO Rs consists of an arc of three neurons that yields reflexive eye responses with a latency of less than 10 milliseconds

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

THE VESTIBULAR SYSTIEM AND OUR SENSE OF EQUILIBRIUM

.

Left eye

'

Right eye

Supeno'---., .

""""'

Lateral

""'"'' -

/

Superior

""tus Llteral

Vestibular nuclei:

Superior Medi.al L1teral Inferio r -

Pons

I

Vestibular organs

(ms) b etween the start of head motion and the eye movement. (That is really

fast. Try doing anything else in less than 10 ins!) The first ne urons in the arc are the afferent n eurons {bottom right in Figure 12.22); these ne urons transrnit

information from the vestibular periphe ry to the vestibular nuclei. the afferent neurons synapse on intemeurons. These intemeurons synapse oneffer-

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

375

FIG URE 12.22 Neural pathways for the thr.ee-neuron arc of the angular VOR. Eac h type of neuron Is using a different ooto r. Eac h of the vestibular afferent neurons (btue) - one branch for each of th9 three canals projects to one of thrge vestibular nuc lei: superior, mediaJ . or lateral. Ntlllrons that oonn9Ct aff91"'9f"lt to Q'ff9rient nrurons. caJled intern;:o.urons (green), project from the vestibular nuclei to the thrQQ ocular m otor nuc lei: abdUCGns, trochloor, and oculornotor. (Only the nuc leus, which is O M of the three ocular n10tor nuclei, is sho'M'I.) The six eff9rent oculomotor nieurons (red) project frcrn these thr99 ocular motor nuclei to the six oculotnotor musd.as.; lnferbr oblique, superior inftiflor rec tus, superior rgc tus , lateral rectus, and m9dial roc.tus. Recal l that you previousty W9re introduc9d t o the orutomotor muscles in Agure 8.16 In C haptEM" 8.

376

CHAPTER 12

(a) O.Dl Hz

(b) 0.05 Hz

Tnne(s)

FIG URE 1 2.23

Tnne(s)

VO R responses in the dark at thr99

(c) 1.0 H,

Time{s)

Thie s timulus

(sinusoidal head rotation; top row} is shown for (a) 0.01, {b) 0.05, and {c) 1.0 Hz. S inusoidal rnotio n evokQs an oscillatory VOR (bottom row'), \Nhlch o ppos9S the hoBad rotation suc h that head rotations to the right are acoompanied by eye rotation s to the left. If the VO R weKe perfec t, It would have the same pMk am plitude (labeled •A" in the figure) as the head rotation , m ean ing that thg, eye rota ted OJ

decreases. For ccmparison, the ncnnalizOO frequency rooponse of semicircular-canal neurons that we sav-1 in Figure 12. 13 is indicat.ed by th9 gr9811 c urv9. Sln09 the

0.2

Qreoen curve rep-esmts the inforrnatton pro\lided to the brain and the black line represents the VOR respon se at different frequencigs, the d ltferenc.e between the black and grgen curves rspresents nrural compMsation performed b y the brain. 0.05

0. 1

0. 2

0.5

1.0

2.0

Frequency (Hz)

\¥e also h.n:e a translational VOR that is evoked when the o tolith organs especia ll y h igh-fre quency h ead translation . Th is se nse head translational VOR helps us keep our eyes pointed a t an object w hen the head tra n.sl'-'l tes in one direction or the o ther.

Vestibulo-Autonomic Responses The vestibu.lar .system also m akes contributions to responses o f the autonomic nervous system. Perhaps the most vivid of these responses is v1otion Si.¥e l a vestibu.lo-autonom.ic ordeal that m.any of us wish we had never experienced. Severe sy mptoms of motion sickness i.ndude nausea and vomiting. Motion sickness typical.ly results when there is a disagreem ent between the motion and orienta tion signals p rovided by the semicircular canals, otoHth orgt.ms, and vision (Oman , 1990; Reason and Brand, 1975). For example, if you nre below deck on a boat, your vestibular system w i.11 200

379

FIGURE 12.26 The contributions of the vestibular $'.{Stern to balance are demonstrated by the comparison of postural responses of subjects with normal Vo9stibukat" func tbn to r€!1Spo nses of patients suffGiring swere bilateral vestibular loss. Subjects We1rning system tha t tell s u s '"'·hen some tlU.ng might be in ternally wrong o r w he n an external s timulus might be dangerous,

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

klnesthesia Perception of the position and movement of our limbs In space. proprtocepllon Perceptlai me:Jlated by kinesthetic am Internal receptors.

somatosensation conectlvely, sensory signals from the skin, rnusdes. tendons. Joints. and Internal roceptors.

390

CHAPTER 13

FIG URE 13.1 CQf"amicJar. Qing period, O"iina, mid-eighteenth c entury.

enabling us to defend our bodies as quickly as p ossible {e.g., by rapidly moving away from tbe noxious stimulus). Ternperature sensa tions enable us to seek or create-a thermallysafeen vi ronrnent. Mechanical sensa tions play an important role in our intimate sexual a nd reproductive activities, and they p rovide a p owerful means of communicating om thoughts and e motions nonverbally. On a m ore fund amental level, touch is important because we can use it to identi fy and manipulate objects tha t cannot be seen or hea rd. Blindfold yoursel f for a t least 10 minutes and try d oin g some routine tasks, like makin g a sandwich, ge tting dressed, or taking a sho\'\:-er. The first thing you v.ill notice while doing this exercise is just how much our species n ormally relies on vision to inform us about the world arm.md us. But you s hould also discover that touch ca n substitute for vision to a surprisin g d egree: You probably won't have as 1nuch tro uble distinguishing the peanut butter jar fr om the je lly jar as you rn.ightthink. And if yo u pO&..timub ORou/#>•timuli

700 'forget present

(slope =-1)

&. 500

Target absent (>lop• =9)

I Number of fingers stimulated

(c)

FIGURE 13.21 An experiment investigating whett1er touch supports preatt€SNs, s plit the epitheHa of our right and left nostri ls. H owever, current°'researm in rod (Li th less ed u ca tion (Boesveldt e t al., 2011 ). TI1e correlation is presumably due to an associa ti on betv-.·een education and naming abili ty or verb al fluency- the more ed u m tion you have, the better your vocabulary and verba l fluency-and this buffers the odor identification declines of aging. Notably, age-rela ted declines in ability are not nearly as pronounced for odor detection and discri rni.nrition as they are for identification, because they rely mud 1 lesson sen1antic processing (Hedner et al.,2010). It s hould be mentioned, thou gh , tha t just as wi th hearing and vision, there is con siderable indi vidual variability in age-rela ted declines in odor perception, and there are m an y oc togenarians \·vi th excellent olfactory ability. TI1e connectio n behveen namin g a nd olfactory identification turns out to be very important in identifying certain ne urologica l diseases a t their earliest stages, in particular Alzheimer' s d isease. This is because AD is a disease characterized by loss of memory, especia lly sem a ntic rne mory (mem ory for the names of things and factual infonn ation), and s ince, as alrea dy explained, odor naming (corn pared to naming any thing else)is most v ulnerab le to s light cognitive p erturbations. Testing for s uch a deficit is crucial, because the sooner peop le are correctly diagnosed wi th Alz heimer's disease and treated, the better their quality and duration of life w ill be. (See "Sensation & Perception in Everyday Lile: Anosmia" on pages 432-433.)

Adaptation Have you ever had the experience of noticing coworkers or classmates w ho seem to be po ming a bo ttle of co logne over their head every m onUng. leaving you choki ng on the overpowerin g nroma? Can't they smell anything? O r perhaps you've n oticed tl1nt, after having a bottle of perfume fo r a few months, you can't sm ell the fragrance in it anymore. Or you've gone away on vaca tion and retunied home to find that your house seei11s to have a "funny" smell that

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

OLFACTION

it didn' t have when you left. \Vhat's going on ? The answer has two parts: the first involves the nose, and the second involves the mind. The sense o f smell is a change d etection sys tem. When a new chemical con1.es along, yo ur olfactory receptors fire in response to it, and yo u perceive a scent. For ex.ample, when you first enter a bakery, you n otice the mouthwatering arom as of the cakes, cookies, pies, and brea d s. But if you stand in.side inspecting the s\oV-eet baked goods for a. while, you may find that by the time you've p icked out what cake you want for d essel't, you can n o lon ger s mell it.\ Vhat has h appened is that the odorant m olecule.s that make up the bakery aroma have bound to the corresponding olfacto ry sen.sory ne urons in your n ose. When this h appens the O Rs retreat into the cell body {Firestein, 2001 ). The receptors are the refore no longer physically available to respond to the bakery scent m olecules. This response is a process in "receptor recycling.1' Specifically, od orant binding to an OR 01 uses the OR to be internalized into its cell body, where it becornes unbound from the odorant and is then recyded through theceUand emerges again in a number o f minutes. Receptor recycling is a mechanism common to all receptors in the class to which ORs be lo n g: G protein-coupled receptors (GPCRs). FURTH ER DISCUSSION of parallels to visual adaptation can be found in

Chapter 2 s tarting o n page 4 7.

This process i.s ca lled receptor adaptation. The precise len gth of time req uired for ad apt.'l tion varies as a function of both the indi vidua l (Da lton, 2002) and the o do rant (Pierce et al_, 1996). On average it takes about 15-20 minutes o f continual expostue to an odorant for the m olec ules to s top eliciting an olfacto ry response, but a daptation ca n also occ ur in less tha n a rninute. Recep tor adaptation can also be und one relatively quickly. Stepping outside the bakery for a few nUnutes g ives unbound olfactory recep tors a chance to acc umulate on tbe cell surfuceagain, so when you wa lk back in, you can enjoy the appetizing scents once rnore. The magnitude of adaptation is also affected b y od or intensity {Kado hisa and Wilson, 2006). As the concentra ti on of a n odorant increases, the degree o f adaptation d ecreases. For example, it takes lon ger to adapt to the aroma \¥afting fronl an app le pie baking in the oven than to the scen t emana ting from a cool slice of pie on the kitchen counte r. Th is is beca use m ore rnolec ules of apple pie aroma become volatile and a re thus available to acti va te ORs when the pie is hot than when it is cold . One W-eoupled receptor (GPCR) Any of the class of receptors that are present m the surface of cifactory sensory neurms. All GPCRs are charactertzed by a common structural feature of seven membranespanning a-he Ces. receptor adaptaUon The bk>chemical pheronenm, occurrlrg after contlnuaJ exposure to an odorant. whereby receptors stop responding to an odaant am detection ceases.

452

CHAPTER 14

cross-adaptation

Th3 reduction In detection of one cdcrant followlrQ

expC6Ure to another odorant. Cross·

adaptation Is presumed to occur because the ocmp:nents of the odors (cr odcrants) in questbn sh•re one

cr more otfactor1 receptors for their tranOOuction, but the order In which odorants are presented also plays a

role.

tha t a factory mus t be e mitting r'da ngerous che micals" frequently complain that they can smell a 1:nalod or corning from the fac ility, rega rdless of w hat the factory p rod uces or whether it is even ope.rating. New findings shed light on why psychological factors often sup ersede physical reality in odor perception. Recent research (Krusemark et al.., 2013) has s hmvn that when we are anxious, initially ne utral odors becom e perceived as tmpleasant and these now negative odors corresponding ly elici t a ug mented responses in highe r-order olfactory a nd e moti onal processing cente rs in the brain (orbitofrontal cortex and pregenua l anterior d ng ula te cortex). M oreover, anxiety stren g thens the connection between olfactory processin g in the orbito frontal cortex and e motion .al processingln the:an,lygdala This-s uggests that w hen we are worried tha t an odor may harm us, we perceive it as more unpleasant, and it elicits more-intense emotional processing and potentiates forthe r negati ve e motiona l response to it. For example, if you smell an odor co ming from a factory that you believe is producing dangerous chemicals, tha t odor will smell bad to you even though that same odor could be one you like in a different context. Em o tional hypersensitivity can also create olfactory illusions, and you may believe that you are smell in g the 1'bad " odor, merely w hen seeing the factory, even w h en no odor a t all IB being emi tted (En gen ,

1972; Herz and Von Clef, 2001). O ne of the benefits of o lfactory adaptation is th ..1t it en.."lbles us to filter o ut stable background odors, and th is filtering a bility can be enhanced through active s n iffing-ta king d e libera te, quick inha lations (Kepecs, Uchida, a nd Mainen, 2007). Sniffing makes OR neu rons less resp onsive to st'ave sleep (S\VS; deep sleep) or REM sleep, they did not awa ken or show any electroencephalograrn (EEG) sleep pattern changes. More recently, it \"·as fo und tha t expos ing s leepe rs to a rtificial sm oke during all stages of s leep had no affect on arousal frequen cy o r EEG activity (Heiser et al. , 2012). These findings underscore the .serious ness of using auditory sm oke detectors, and why o ur sense of s mell cannot p rotect us from s moke inhalation and consequent catastrophes w hile we s leep . In s pite o f these findings, a fon cti onal magn etic resonance imaging (f11RI) s h1dy rep orted that p resenting a n odor during S\.VS that had previously been present during an awake learning tas k produced hippocampal d1an ge.s con sistent with those observe d in memory consolidation (B. et al , 2007). Ho\·vever, the odor had no effect if it \"'as presen ted during o ther sleep s tages, including REf\.'I s leep. This finding .suggests that od ors may be detected by the brain during S\VS, but this d etection rnay no t re.suit in overt arousal It is also known from ne uroim agin g research on awake subjects that durin g smell tasks, paying attention to odors alters brain activ ity and odor d etection ability (Plallly et al., 2008; Zelan oet al., 2005). \ Ve use more of o ur brain and can sme11 more acutely w hen we conscious ly focus on s melling than w hen \.Ve d on' t. It may be that,, because attention is.effectively cut off during sleep, so is o ur abili ty to respond to od ors. A further tantalizing piece of eviden ce supporting the difference beh'\o·een olfaction during w ake ver sus s leep states comes fr om the recent find ing in mice that g ranul e cell activity in the olfactory bulb during s leep is s low, consistent,, and tuned to the brea th ing cycle, whereas during \-vaking it exhibi ts much stronger, wilder, and spontaneous firing, Tha t is, gra nule cells, w hid 1 may be the ha.sis o f odornnt identifi cation operate in a consistent fa shion independent of the od or env ironment during s leep, but by contrast appear to respond to specifi c o d or an ts as they enter the environment while we are (Caz..'lkoff et al., 2014).

Olfactory Hedonics The m ost immedi1Jte and basic response we have to an odor is whether \>Ye it or not. Such affec ti ve e valuations are knom1 n.s odor hedonlcs. Jn tests of odor hedonic evalu ation, people are typically asked to rate ho""' pleasa nt, fru1liliar, and intense a gi ven odor is. TI1ese measures are then used to d etermine the h edonic value of a specific s me ll. It is obvious that perceived pleasantness is related to our liking for an cxior. But how are familiari ty and intensity related?

Familiarity and Intensity As with ma ny other facets of life, \ •le tend to like odors tha t we've s melled many times before. That is,, we tend to like familiar od ors better than unfamiliar odors. Moreover, we often perceive pleasant odors as familiar, even if we ha ven 't smelled th em before (Moskowitz, Dravnieks, a nd Klarman, 1976; Sulmont, lssa nch ou, and Koster, 2002). Thus, ratings of odor pleasanh1ess and fami liarity s how a linea r relationship with od or liking. Intensity has a m ore complex relati onsh ip to od o r likin g th a t is often represented by an inverted-U function, but this depends on the odorant. A rose scent1 like pheny lethyl alcohol, may be evaluated as more posi tive with increas ing intens ity, up to a p oint; then the fun ction reverses, and as the scent becom.es stron ger, it is judged to be m ore disagreeable (Figure 14.16a). By contrast, a fi sh y od or, like trimethy lan1ine, may be acceptable at low concentr1Jtions, but as intensity increases, its pe rception becomes s teadil y rnOl'e n egative (Figure 14.1 Gb )_ Note a lso that in di vi dual differences in the nwn be r and type of receptors exp ressed may influen ce on e's sensitiv ity (i ntens ity

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

OLFACTION

FIGURE 14.16 Pleasantness ratings o f odorants plott9d against (a) The relationship between odor intet'"1$1ty and pleasantness is often described by an lnwrted-U func tion, if the odorant Is considtfed pleas.ant, as the synthetc rose S09nt (ph9n)'iethyl alcohoQ usually is. (b) For an od orant that is initial ty c cnsidered t ol€fable (but not necessatily pleasant), such as fishy smelling trimethylamine, the relationship may more aptly bQ d9SCtib9d by a linear graph like this.

455

(a) Phenylethyl ak:ohol

perception) and hence the predisposition to experien ce specific odors along a pleasantness continuum.

d

Nature or Nurture?

;1;

11!!] Unplm•nt

A long-st.mding debate in olfaction centers around whether hedonicresponses to odors are innate or learned . Researdlers on the inn.ate side of the d ebate l!!J Low --- - - - - -•H'ogh daim that we are bo.m \vi th a predisp osition to like or dislike various s m e lls. Intensity In o ther wo rds, rose is inherently a good s me ll and skunk is inherently a bad (b) smell, the way bitter is inherently Lmpleasant to us and sweet inherently Trimethylamine pleasa nt (see Chapte·r 15). ln contras t1 researchers taking the Ier rnt¥i vJew ho ld that we are born merely \vith a predisposi tion to learn to like or diSlike sm ells, that w hether a sm ell is liked or not is d etennined by the emobon..1 1 va lue (good or bad) of the experiences that have been associated with it. Tha t is, if we like rose and dislike sk unk, the reason is that we have a good and a bad associa tion, respectively, with these hvo scents. \Ve need n ot have di rect c: contact with a skunk to form s uch an associa ti on1 thou gh, because cultural lea mi ng provides mea ning to many unen countered s timuli. If asked to take a position you rself, on the sole basis of your ov..'11 per.sonal experiences, it's pretty like ly you '"''ou ld come d own on the ilm.a te side o f the deba te. After a ll, wh o could like the s mell of sktmk1 and who wouldn' t like the s me ll of a rose? [n fact1 however, a g reat deal of evi d en ce s uggests that l!!J odor hedonics are cllmost exclusively lea rned. A good place to sta rt lookin g for Intensity such e vidence is wi th infa.nts. If odor preferences a re innate1 then newborns should display them. However, researchers have repeatedly found that infants and children o ften display very different preferenc-e:s from those of adults. Fo r in.st.mce, infants d o n ot find the s me lls of sweat and feces unpleasant (Engen, 1982; M. Ste in, O ttenberg, and Roulet, 1958), and toddlers often do not hed onically differen tiate between odorants that adults find ei ther very unp lea.s.a nt (e.g., butyric acid, wh ich s mells like dirty socks) or pleasant {e.g., amyl acetate1 which smells like bana na). 01e difficulty witl1 these typwn to intensify sweet. At first this did not seem to be practical because the effects vi1ere so small. H01i\'ever. a. serendipitous finding during a tomato (botanically, a fruit) experiment offered a different point of view. The tomato experiment was aimed at a problem focd shoppers are well aware of: it's hard to find a good tom ato in a supennarket. Recently, heirioom tomatoes have been finding their way into market s. Why would these tomatoes taste better? The answer is t hat intensive

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

469

470

CHAPTER 15

b reed ing to giJe tom atoes charac teristic s thought to b e desirable (uni-

form ripening time, fruit Sze, and so on) have in some cases led to a d eterioration in flavor. Heirloom to matoes are genetically m ore diverse b e::ause they cane from a time pri or to the intensive breedir"Q. Thus some of them have tt-e flavors rem embered 1rom those earlier days. Collaboration between plant bldogists and psycho log ists at the University of Florida 100 to an experiment utilizing eighty heirloom to matoes g rovvn on university fatmlard. After harvest ing , half went t o a

chemistry lab where the sug._qrs, acids, and vci atlles v.iere measured :

8230336 Amna A

tast e bud A globul..- cluster of eels that has the ft.netlon of aeatlng neural sk;Jnals c onveyed to tl1e brain by the

taste naves. Some of the cells In a

taste bud have specla.lized sites on their apical projections that Interact w ith taste stimuli. Some of the cells form S')'Tlapses with taste nerve fibers.

half went to a psychophysic s lab where taste . flavor (I.e .. retro nasal olfac tlo n), and preference W ff'e measured. Regression analyses ldentlfie:J the comp:::>nents responsib le fo r the sensory properties of the tomatoes as well as how much they were liked . The solutio n to the problem turned out to be sim ple. Some tomato oonst ituents correlat ed positiJely with liking; that is. the more of that const ituent that was in the tomato , the mo re it was likOO . Some did the reverse and some did not m atter. To make a better to mato , Increase the c onstituents oontritutlng to liking and decrease t hose contributing to dlel iklng. Knowing what to aim for. cross-b reed ing \vi ii give us better t omatces. The serendipitous result cam e from a mathematic al anatysls of the to mato dataset. Multiple regression is used widely in the social sciences to examine vartcus sources for a given effect. Fo r exam ple, an investigator mig ht \o'tant t o look at contributions to 10 from a variety of sourt::es (age, health , lnoome , educaticn , and so o n). Multip le regression was appl ied to the tom ato data, to see If any oonstituents other than the sugars were oontributing to sweetness. The resu lt w as startling. A considerable amo unt of sweetness was c oming from the volatiles. It was sLddenly apparent that smal l effects from individ ual volatiles were adding up such that a oonslderable am ount of the sweetness of a given tomat o was produced by the volatiles (perc eived retronasally). For exampl e, increasing the conc entrations of those volatiles could double the s1,veetness of a tomato . The Imp lic ations for sugar reduct ion are dear: adding the correct volarnes can red uce the amount of sugar needed to sweeten foods and b everages. However. the potential goes ei1en further. There are ot her volatiles that can enhance salty and still m ore that can supp-ess t.:Ctter. The d iscoveiy that plants use volatiles In this way is very new. Are these enacts in the brain ? Are they acquired somehow from experience? The original thinking about volatiles that oould enliance sweet was that these would be limited to fruity flavors because we so oft en experience fruit and sweet together. However. one of the tomato volatiles that enhanced sweet was lsovalerlc acid, which smells like S\Neaty seeks. Sweaty socks and sweet d o not seem to be a oombinatio n that would b e exp!rienced to gether very often! Muc h is yet to be learned. However. better to matoes are surely right around the comer. alo ng with safe new ways to sweeten or salt foods and reduce unw anted bitter (e. g .. bitter ve;ietablas , m edications) to follow.

Anatomy and Physiology of the Gustatory System Ta ste percep tion co nsists of th e follow ing seq uen ce of events (the stm ctu res involv·ed are illus trate d in Figure 15.2): Che \vi.ng b reaks d o vvn food s ubsta nces into m o lec ules, w hich are disso lved in saliva. The sa li va-b o rne food m o lecu les flow .into a taste pore tha t lead s to the tas te buds h oused in structures

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

TASTE

471

FIG URE 15.2 Th9 locatbns of ea.ch type o1 taste papilla ar.:. Identified in the of the t Cf"lg ue at lo wef right. N9Ural signals from the t aste buds in those papilla€! are transmitted via cranial nerves Vll, IX, and X to the bra.in.

T.tSterec-eptor.s

V

tb_.

.o..,

,

CircumvaJl,1te

papillae

Faliate ._..,., pillae

C!ol'oSopbaryngeaJ nen -e(cranial nen•eIX)

cal led papillae (singulm- papifla) that are loca ted mostly on the ed ges of the ton gu e in a rough ov al (ii the olfactory epithelium is the retina of the nose, the tongue is the retina of the m o uth). Taste buds, in tum, multiple taste receptor cells,each of which resp onds toa limited number of molecule types. When it comes in contact with one o f its p referred molecules, the taste recep tor cell prod uces action p o tentials that send informa tion along one of the crania l nerves to the brain. See Web Activity 15.2: Gustatory Anatomy for a n interactive overv iew of the sys te m, w hich is described in gre.:1 ter de tail in the sections that foll ow.

Papillae Papillae give the tongue its bumpy appearance a come in four major varietie.s: filiforrn, hmgifom1, foliate, and d rcumvallate. The last three of these contain taste buds. Flllform papillae, th e ones w ithout nny taste hmc ti on, are Joc,1 ted on the anterior p ortion of the ton gue (the pa rt we s tick o ut when giv ing someone n raspberry) and come i.n different shapes in diffe rent species. In ca ts, they are shaped like tiny sp oons with sh.c'1 rp edges. The filifonn papillae on our tong ues d o no t have these sharp ed ges, which is why you will find lapping milk from a bowl considerably more difficult than your cat does. Fungiform papillae, so named because they resemble tiny bu tton mus h rooms, are also located on the ante rior part of the tong ue. They are visible to the na ked eye, but blue food colorin g sv•rnbbed onto the tong ue makes the m particularly easy to see (blue food coloring sta ins filliorm papillae mu ch be tter than fun.g iform papillae, so the fungifonn papillae appear as light circles again.st a darker blu e background )_ Ftmgiform papillae in di.a.meter, but the maxi mum is about l millimeter (mm). On average, about six taste buds are buried in the surfare o f each fungiform papilla. If we s tain the tongues

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

papilla Arr/ of multiple structures that gll/e the tongue Its bumpy appearance. From smallest to largest the papilla types that contain taste buds are follate, and clrcumvallate; flllform papillae, which do not contain taste buds, are the smallest and

most numerrus.

taste receptor cell A cell v.;thln the taste bud that contains sites on Its apical projection that can Interact with taste stimuli. These sttes fall Into two major categories: those lnteractlrg with charged particles (e.g., sodk.rn and hydrogen lms), and those Interacting v.;th specific chemical structures. tiltforrn papillae Small structures m the tmgue that prO'Ade most of the bumpy appearance. Fillform papillae have no taste functlm.

fungiforrn papillae Mushroomshaped strwtures (maximum diameter 1 millimeter) tl1at are dlstnbuted moot densety on the edges of the tongue, espedally the tlp. Taste buds (an average of six per papilla) are burled In tt'e surface.

472

CHAPTER 15

FIGURE 15.3 EM"llYlples showing typical varlabit ity in the density of fungiform papil!M from one indMdual to the next. 1h9 circles show the 6-mm template area u sed for counting fungiform papllaeon tongues stained with blue food coloring. (a) Tile tongue o f an averagQ taster has 16 fungifon-n papillae. In extrQrne cases. normal Individuals may have as few as 5 fungiform papitlaG in that area or as marry as 00. (b) A supertaster's tongue (supertasters are discuss9d latier in the chapti9f) has 60 funglfotm papiUM in that area.

foliate papillae Fc>ds of tissue

containing taste buds. Foliate papll lae are looated on the rear of the tmgue lateral to the drcumvallate papillae, where the tongue attachoo to the rnouth. clrcumvallate papillae Circular structures that form an inverted Von the rear of the tongue (three to I ve m each side. with the largest In the 001ter). O rcumvallate papillae are moundllke structures, each surrounded by a trench Qike a moat). Thooe papillae are much larger than funglform papillae.

microvilli Slender projections of the cell membrane on tre tips of some taste bud cells trat extend Into the

taste pae.

of many individuals, we see a large amount of variation (Figure 15.3). Son'!e tongues have so few hmgifom1 papillae that their stained tongues appear to have p olka d ots on the m. Other tong ues ha ve so many that the p olka dots are wa ll-to-wall. Foliate papillae are l.ocated on the sid es of the tongue r1t the point where tl"l fu ngue 4rfgnifica tion, they look like a series of folds. Taste buds are buried in the folds. Finally, clrcumvalate papllaeare relatively lu-ge,circular slmctures fom1ing an inverted Von the rear o f the tongue. 1l1ese papillae look like tiny islands su rrounded by moJts. The tls te buds are buried in the sides of the moats. Although most people don' t real ize this, there are also taste buds on th e roof of the mo uth where the hard and soft palates meet. To d em ons trate these, wet your finger and dip it into salt crystals. Touch the roof of your mouth and move your finger back until you feel the bone end (the margin be tw"een the hard and soft palates). You will experien ce a flash of sa ltiness as you move the salt crystal s onto the taste buds arrayed on tha t margin. In stun , the taste buds are distributed ina line acros.s the roof of the mouth and in papillae distributed in an oval on the tong ue. Fungiform papillae make up the front of the oval, a nd foliate and circum valla te papillae make up its :re.., r. Note that we have no subjective awareness of this distribution of tas te buds. (For m ore on this topic, see Web Essay 15.1: Scientific Urban Legend- The Bogus Tongue Map.)

Taste Buds and Taste Receptor Cells Tas te neurons are pseud ounipolar: a sing le process exi ts the cel l body and then splits into peripheral and central limbs. The periphe ral axons make up the n erves that project into the tongue (e.g., ch o rda tympani and glossophdryngea l ne rves). The central axons project to the brain. Ec1ch taste bud is a clus te r of elonga ted cells, organized much Hke the segments of an orange (Figure 15.4a). Th e tips of som e of the cells end in s lende r mlcrovilll (singular microvi1lr1s) containing sites that bind to taste s ubstances.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

473

TASTE

FIGURE 15.4 Tas1"' buds. (a) A taste bud buri€1d in the tissue o f a fungitorm papllla. (/J) Stimulation of 1h9 three taste receptor cell types results in responses in th9 taste nerves. T)ll)Q IH ceUs have Neurotransmitters {e.g., ATP (adenosine t1iphosphateD may be secreted by cells 'Without TuesQ neurotransmltte:irs may act on adjet.oent as we 1as taste nEf'le fibers, to ultimatel'j produce nQUral signals that \llr'ill trav91 to the brain. (Aft'1t'" Ropgr and Olaudhati, 2009.)

In an ear lier era, these microvilli were mistakenly thought to be tiny hairs; we n ow know that microvilli are extensions of the cell membran e. Som e years ago, t.:'lste receptor cells \'\-"ere th ought to have sites sti mulated by taste stimuli at one end and synapses with taste axons at the other end. But rnore recent work has identified taste receptor cells that do not synapse with taste nerve fibers. H ow is information from these receptor cells conveyed to the brain? Details of how this happens are begi.Jming to emerge

(Hemess et al., 2005; Roper, 2006, 2013). Taste bud cells a:munun.imte with each other within the taste bud. An.atomically, taste bud ce lls fall into three different groups with different functions (Figure 15.4b): Type I ceUs appea r to be ho usekeepin g cells th.at excrete potassium through the taste pore allowing the other ce ll types to maintain their resting membrane potential, but they may also play a role in sal t taste. Type il ceUs are true receptor cells that tend to respond to only one of s\-1,reet, bitter, or lllTh:'Uni s timuli. These cells have G protein""'oupled receptors (GPCRs) on the microvi lli but do not have synapses. Type III cells are nm ...· called "presynaptk" cells: they have S}1'apses. Presynnptic cells may also play a role in sour taste. Many deta.ils of the ce ll-to-cell comnnm.ication are yet to be determined, but the overall m essage is that different tastants exci te different tas te bud ce lls utilizing different transmitters. The contributions of this within-taste-bud processin g to our taste experiences are not yet knO\Yn (Roper, 2013 ). ln fun giform papillae, the taste ne rve fibe rs that ente r the taste buds branch, so an individual cell can be innervate d by mo re than o ne taste fiber, and a n individual taste fiber can irmervate more than on e cell. Taste receptors haven limited life spa n. After a matter of days they die and are replaced by new ce lls. This ronshmt renewal en ables th e taste syste m to recover fron1 a variety of sou rces of damage, and it e.xplains why our i.:'lste systerns remain robust even into o ld age. Recordings from taste nerve fibers show that different receptor cells contacted by branches of a single fiber sho'"'. simila r specificities to taste sti muli. In other words, it appears that the ne rve fibers are somehow able to select the cells 1 I

Typi> II

Sdty?

Swn""f, Nttcr,

Sour?

l

um:.1mi

l

•

30

tastant tasted.

Any s timulus that can be

474

CHAPTER 15

causes nn action p otential to be sent to the brain. GPCRs a lso contribute to what is called the ''umami" taste. More abo ut this later. FURTHER DISCUSSION of Gl'CRs in the context of olfaotion can be found in Chapta- 14 on page 451.

Taste Processing in the Central Nervous Sx.stem insular cortex

The primary cortical

processing area fcr taste-the part of the cortex that first receives taste lnfor· rnatlon. Aloo call:ld lnsula or i;,tJStatl)()'

con ex.

After le..'l'\.;ng the taste buds through the crania l ne es, g wtatory info rmi\tion tra vels through W'-'Y s tntions in the medulla an d thala mus before reaching the insular cortex (also refe rred to as the gustatory cortex) (Figure 15.5) {Pritchard and Norgren, 2004). 111e functional sep:lration of the taste quali ties s uggested to early investiga tors tha t there woul d likely be a gus toto p k ma p in the cortex, but evidence for this h;)s been elusive. However, two recent contributions have s hed n ew light on earlier findings. One study (in mice) using a calci um imag ing technique identified clus ters o f cells showing neural resp onsiveness to bitter, sweet, and sodium compo unds. In a dditio n, some

cells we !'e s timul ated by m on o po tassium g lutama te (the use of p o tassium was intend ed to evoke g lutama te and no t salty in o rde r to locali ze urnami taste) (Che n, 2011 ). Oddly, this p rocedure d id no t e voke responses to acids, but the authors suggested that a sour ch.1ster might be outside the area they

FIG URE 15.5 Taste info rmatio n projects from the tongue to the m€ldulla (Via OQl'V9S VII , IX, and X), then to the thalamus (shown in cross sQCf:lon 1 of the brain) , then to the insula (cross section 2), and finally to the oratofrontal cortex (cross sectbn 3').

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

TASTE

sampled. The separation of th ose clusters is consistent with a g ustotopic map. Even m ore exci ting, a meticulous new lesion-mapping strategy ha s localized ta ste function ally by utiBzing conditioned taste ave rs ions {Schier et al., 2014). The areas that underlie conditi oned tas te aversions fall into a "subregion encmnpassing the posterior gustatory cortex and the s urrounding ins ular cortex cons idered to be visceroceptive." Earl ier contradictions across s tudies may reflect less p recise lesion mapping. The orbitofrontal cortex receives projections from the insular cortex. Some they resp ond to temperature, orbitofrontal neurons are multhnodal; that touch, and s mell, as ' "'·ell as to taste, s uggesting that the orbitofr ontal cortex may be an integrati on area. Inhibition plays an important role in the processing of taste in.fonnation in the brain On.e of the functions of th.is inhibition m ay be to protect our wholemo uth perception of tas te in the fuce of injuries to the taste system . Our br(lins receive taste input from several nerves (see Figtue 15.5). Damage to one of them diminishes its contribution to the whole; however1 that da mage also re.leases the inhibition that is normally produced by the damaged nerve. 1l1e result is that whole-m outh taste intens ities are relatively tmch•.-mged. Unforhmately, this p reserved who le- mouth perception comes at a cost in some cases. Localized taste damage is often accompanied by " phantom. taste 11 .sensations {recall the phantom limbs experienced by many limb amputees, described in Chapter 13), as if the release o f inhibition permits eveJ1 noise in the nervous system to be perceived as a taste. Descending inhibition from the taste cortex to a variety of o ther s tructures may also serve o ther ftmctions, For exa mple, mouth injuries th.:1 t lead to oral pain make it harder to ei.lt. 11-'te inhibition o f s uch pain perceptions by tasteprocessing parts of the brai n would m ake ea ting easier a nd thus increase the likelihood of s urvh'al (beca use no mrttter h ow mu ch the mouth hurts1 we still ha ve to eat). Consisten t with this P-atients with a serio us oral pain d1sorder (burning mo uth syndrome) sh wt\ to hnve.Joccilized ta ste d amage as \'\-ell (Gn1shka and Bartoshuk, 2CKX.l). Purthem1ore, women w h o h ave tas te damage are m ore like ly to suffer from severe nausea and vomiting during pregnancy (Sipior"1 et al., 2000); and can cer patien ts, whose chemo thera py and radfatio n the.rnp y is known to damage the taste sys te m, are m ore li ke ly to experience coughing, gagging, hiccups, and pain. In all these cases, inhibitory signals from the taste cortex that n ormally help prevent ea ting-disrnptive sy mptom s (oral pain, vorn.iting, hiccuping, and so on) may have been turned off beca use of the d amage to the taste systeJ:n.

The Four Basic Tastes We learned in the Chapte r 14 that we are able to dis tinguish 1na ny different odorants. Already in this d'!apter, h owever, '"1e have seen that when olfaction is taken out of the equation 1 much o f the complexity of the sensa tio ns evoked by foods vanis hes. Tirns we are led to believe that the number of basic taste qualities is qui te s mall. In fact1 the current universally accepted lis t includes only the four basic tastes previously mentioned in the "Taste Buds and Taste Receptor Cells 11 section o f this cha pter: salty, sour, bitter, and sweet. As '""·e di scuss these in the sections that follow, n ote that one of the mos t important features of these basic tastes is that om liking (or d isliking) for them is hardw ired in the brain; that is, we are essentially born liking o r disliking these tastes. This is very different from the way we lea rn to like or dis like odors. Incidentally, the num ber of basic tastes h as been the s ubject of a rgmnent that goes back even before Aristotle (see &1.rtoshuk, 1978). Various investigators

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

475

orbitofrontal cortex (OFC) The part of the frontal lobe of the cortex that Ilea behind the bona (orbit) con taining lhe eyes. The OFC Is responsible fa tha ccnsclous experience of dfactlon, as mll as the integration of pleasure and displeasure from food; It has be€fl referrecl to as the secondary olfactory eol'rf of the four taste qualltles that are generally agreed to describe hlman laste expeneme: sweet, salty, sour, bitter. salty One of the four basic tastes; the taste quaftty produced by the cations of salts (e.g .. the sodium In sodium chlcrlde prOOuces the salty taste). Some cations also produce other taste qualities (e. g., potassium tastes bitter as well as salty). lhl purest salty taste Is proouood by sodium chlalde (Naa), commm table salt.

sour One of tlie four basic tastes; the taste quality produced by the hydrogen Ion In acids.

bitter Ole of the frur basic tastes; the taste quality, generally considered unpleasant, produced by substances Ilka qulnlre or caffeine.

sweet One of the four basic tastes; the taste quality produced by some

sugars, sum as glucose. fructose,

and sucrose. These three sugars are particularly useful to us,

and our sweet receptors are tuned to them. Sane ot11er cornpoun:::Js (e.g., saccharin, aspartame) are also sweet.

476

CHAPTER 15

FIGURE 15.6 Diagram of a taste receptor cell, Illustrating the diffi9f''9nt receptor mechanisms for ionic stirnuli (salty and sour}. {a) SaJty taste Is produced by the cation in a salt. In Naa. the sodium cation is admitted to tl"la receptor cell by sodium (ENaC). (/)} Sour taste is produced by hydrogen io ns (H•), which can errter tl"le csll in tw'O ways tile t9Xt for C haudhari and Roper. details).

2010.)

(•) NaCl

(b) HO

Outside cell J;NaC

Outside cell H + ion-selective ch.UUlel H"'

H• .............. H•Ac lnsi.de cell

blSid€'cell

have su ggested adding to the Ii.st of basic tastes. The two m os t l'e cent contenders (umami and fatty) are discussed later because o f their special as.sociati.o n \vi th leaJlle.d preferences. Since there is no de finition o f ' 1basic taste,'' it is easy to call particular oral sens.:i tio ns ''basic.'' Hmvever, th e authors o f this text prefer to reserve '1basic taste" fo r those with hardwired affect, since that is arguably the characteristic of taste that m os t distinguishes it from the other senses. \Vill huther d etails about a gusto topic map in the cortex settle the argtmwnt over what we s ho uld ca ll "bask tastes" ? Sadly, no they \.\.' ill not. Consider the new d ata on dlli"iers of cells in the cortex that responded to bitter, sweet, salty, and t11na mi. If tas te quality is coded bya gustotopic m ap, then different taste qualities must create different dusters.

Salty Salts are made up of tv.·o charged particles: a cation (posi tively charged) and an a n.ion (neg..1. tively charged). For example, common table salt is NaO; th e sodi mn is the ca tion (Na+), and the chloride is the anion (Cl -). The source of a the s.a lty taste is the cation (Figure 15.6a). Alth ough all sa lts taste at little salty to humans, pure NaCl is the sa ltiest-tasting sa lt arotmd . Sodium mu st be available in relatively large quantities in the body to m::iintain nerve and m usc le fun ction, and loss of body sodium leads to a swift death. O u.r ability to perceive sa ltiness is n ot static. Gary Bea uch amp a nd his colleagu es (Bertino, Beauchamp, and Engelman, 1982) s howed that di et CL\l\ affect the perception of saltiness. Fortunately for those on low-sodium die ts, reduced sodium intake increases the intensity of saltiness overtime. lndividuals who are initially su ccessful in reducing their sodium intake will find that foods they used to love may now taste too sa lt): This adjustment in perception helps them keep their soditm1 intake d own. Our liking for sa ltiness is n ot static either. Early experiences can mod ify salt preference. In 1978 a nd 1979,several hundred infants were fed soy fomrnlas that were accidenta lly deficient in chloride beca use of an error in formulati on . Chloride de ficiency has effects on human physiology that mim ic the effects of_ sodiu111 d fi cie.ncy. the infants who \Vere dtloride-deficient offered an important way to study odium deficiency in humans. The Centers for Disease Control and Prevention (COC) in Atlanta monitored these infants, and a of s tudies were done to assess a ny potential damage. On e of the consequences was that the salt preference of the child.ren increased (l. J.Stein et al., 1996). Experiences during gestatio n can also affect salt preference. For example, college studen ts whose moth ers had experienced rnod era te to severe morn.mg sickn ess during pregnancy sh owed an increased preference for salty s nacks (Cryst.al and Bern.stein, 1995). The exact mechanism s by which these abnormal me tabolic events enh .."1.nce salt pre fe rence are still not unders tood.

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

TASTE

477

Sour As you may reme mber from high school chemis try, a solution containing hydrogen ions (H+) and hydroxide ions (OH-) in equal proportions produces water (HOH, or As the relative proportion of H+ increases (decrea sing the pH level), the solution becom es m ore acidic. Why do you need to be renUnded of all this? Beca.usesou1· taste is produced by hydrogen ions. Hydrogen ions enter the receptor cell throuohJon c;harmels; an ad ditional mechanism for sour allows undissociated aad molecU.les (infuct molecules that have not split into h.vo charged pmtides) to enter as well. The undissociated

acid molecules dissociate inside the cell. Ultirn1.1tely, the s timu.Jus that triggers sour taste is the hydrogen concentration inside the rece ptor ce ll (Ftgure 15.6b} (DeSim one et al., 2011). Some illdiv;duals like the sourness of acids in relatively low concentrations. fvtany adults enjoy pickles and sa ue rkraut, both of w hich get their sour tastes from acids . In addition, many children in particular like sour candies (Liem and Me1mella, 2003). At high con cen trations , h owever,acids "'ill damage both externaJ a nd internal body tiss ues.

Bitter The Human Genome Project revealed a mtJtigene family responsible for about 25 different bitter receptors. The HUGO Gene Nomenclature Committee has established the rules for naming genes. (HUGO stands for Hmnan Genome Organisation, an international organization of scientists involved in human genetic and genomic research.) The bitter gene family is TAS2R ("TAS" stands for tas te, with the "2" indicating bitter); numbers following the R indicate the particular gene that is a member of that farnily. The25 bitter genes are loca ted on three differentduomosomes: 5, 7, and 12. TI1e receptors these genes express are designated without using it.alics (e.g., TAS2RHor T2R#). The bittet receptors expressed by these genes face a formidable task. Wolfgang f\·1 eyerhof, one of the world '.s experts on bitter genes, estimates that there are thousands of bitter m olecules (many coming from plants that protect from predators by tasting bitter). How can only 25 bitter receptors handle the job? Part of the ansvver is tha t som e of the T2 receptors only to specific compounds (e.g., th e propylthiourncil [PROP] receptor [see Figure 15.lOb] responds to a s mall group of chemically related compo unds), but o thers (bitter "genern.lists" ) respond to many compounds (Figure 15.7 ). Th is also means that some bitter s timuli s timulate many bitter receptors {e.g., qui.nine stim ulates 9 o f the 25 bitter receptors). Tonic wa ter (which cm1tains quinine) was originally fornmlated as a tre..1t1ne nt for malaria; now, however, we know that tonk water does not contain enough quinine for that purpose. H owever, tonk water does contain enough quinin e to taste very bitter, a nd for this reason lots of s ugar was added to make the tonic water palatable. This approach works because sweet and bitter tastes inhibit on e another; tonic water tastes much less bitter than the quinine FIGURE 15 .7 BittGr l"o9CQptors ar9 designated byTAS2Rit, where# Is the nurnbo9r o f tho9 r9ceptor. Th" rQOBP· tor numbers arg shown at the top o f the tabl9. Exampl9S of cornrnon bitte.r compounds are list{ld at th9 l9ft. A fill9d box indlcat9s a ri909Ptor that responds to the bitter compound on the left. The total numbQr of stimulat9d by a gtvQi compound is shown on the right. (After MQYQffiof et 31 .. 2010.)

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

4 78

CHAPTER 15

content alone wouldr and it also tastes much less sweet th.:"U1 the s ugar content

alone would Tonic water actually contains about the same amount of s ugar that' sodas have. Although a great rnany different compounds taste we genera ll y do n ot d isting uish between the tastes of these comp mmds; we s imply avoid them a ll. ll-1e diversi ty of receptors for bitterness enables species or even indiv iduals in a given species to ha\'e va rying responses to an array of bitte r comp ounds. One of the most fam ou s of these is 11 taste blindness'1 to phenylthiocarbamide (PTC) found in human s--a phe nomenon we \.Yill revis it la ter in this chapter_ A.I though bitte r taste usuall y s ignals p oison , some bitter stimuli are actually good for us. For bitter com poun ds in some vegetables he lp pro tec t against cancer. We would like to be ab le to "turn off'' these bitte r sensati ons to make it easier for people to ea t their vegetables. Jn pursuit of this goal, Robert Margolskee, a p ioneer in s tudies of bitter tran sduction, used his understanding of the bitter system to identify a s ubstance that can inhibit .sorne bitter sensations: adenosine mono phosphate (Afl.1P) (Ming, Ninomiya, and Margolskee, 1999). Al\llP may actua lly hmction as a natura l bitter inhibitor in mo ther 's milk. A number of compounds in m.iLk, su ch as casein (milk protein ) a nd calci um sal ts, taste bitter; and, as we will see aversions to bitter tastes are present a t birth. The presence of AMP in mo ther 's milk may s uppress those bitte r taste.s enough to a llow milk to be pa latable to babies \.vho are particul arly respons ive to the m . Bitte r perception is also affected by hormone levels jn women. Sens itivity to bitterness inten sifies during a nd diminis hes a fte r men opause (Duffy et al., 1998). These d ifferences ma ke sense in the context o f the foncti on of bitte rness as a p oison d etection med1anism. Intensifyin g the pe rcepti on of bitter ea rly in p regnan cy, when toxins exert thei r maximum e ffects, has clear biological value. Consistent with this coi-relation, some o f the aversions during pregnan cy occur w ith foods o r bevera ges tha t h.ave bitter (e.g., coffee). One of the m os t interesting recen t develop ments oon ceming bitter is the discovery th.a t bitte r receptors CE\ll be found on locati ons o ther th.an the tongue. For exam pier bitter receptors in the gut appear to s low absorption Qeon et al ., 2008). flliusf f tQxlns get µasl l)ie-n101)th, tl>e g ut can still preven t p oisoning.

Sweet

FIGURE 15.8 The molecular structure of sucrose, commo n table sugar. This disaccharide is formed from a combination o f a glucose molecule and a fru ctoSiQ moh;icule. G lucose, which is easily extracted fro m sucrose by the digestive system. is the maJn fuel that powers ahrost every biologic al engln9 (including the human brain c urrently reading this b ook).

Sweebless is evoked by s ugars, simple carbohydrates that generally con foml to the chemical fomm la (CH 20),,, w here n isa number between 3 and 7. G lucose, o ne o f the S\'\-·eetest-tasting sug.1rs, is the principal source o f e nergy in humans (as wel l as nearly every o ther livi ng thing o n Earth). Common tab le s ugar, su crose (Figure 1 5.B)--w hich is a oombin ation of glucose and yet aneven sweete r. oth er su ga r, The bio logical fonction of sweet is diffe re nt from that o f bitter, a nd the way taste receptors are h.med suppo rts th at bi ological diffe re n ce. Many different molecules taste bitter. Our bio logical task is not to distinguish among c.c-1 ,2 Glrcosidic linkage

vf-":"1, :6)--4'"'"'"" +

H

OH

a-0-Glucose

OH

,...-->----., H

H GI OMH

H

t

2

H OH

M

O

0

CHj:)H

H

F11.lc tose

a:-o-G\uoost.>

Fruc tose S ucrose

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

TASTE

Outside cell

479

FIGURE 15.9 Structure of the T1R2T1 R3 heterodltner sweet receptor , showing binding sites for both large and small SVV99t rnol9CUIQS. (Afto9r T ernussl, 2007 .)

the m but ra ther to avoid them a ll. Thus, we have multiple bitter receptors to encompass the chemical diversity of poisons, but they alJ feed into comm on lines le1ding to rejection. With regard tosvi,•eet, sorne bio logica lly llseless s ugars have structures vel'y similar to those of glucose, fmctose, and sucrose. [n this case, then, the task of the taste system is to tune recep tors so specificall y that the biologically important sugars stimulate sweet taste but the others do not. Consis tent w ith the biological purp ose, only h'lo G protein -co upled recep tors are involved wi th sweet taste (Tl R2 and TI R3) , These two proteins combine to form a singl.e recep tor ca lled a heterodlmer (Figure 15.9). ll1.e oute r p or tions of the two proteins resemble the shape of Venus fl y traps; large sweetene r molecules can enter the fly traps to stimula te the receptor. A variety of othe r binding sites accommoda te s maller molec ules like su ga rs, sacchari n, a nd aspartame. Lnitia lly, the TI R2-Tl R3 he terodimer was thou g ht to be responsib le fo r a ll 5'"/eetness, but it introduced a new puzzle: No matter how the heterodfo1er is stimulated , the receptOI' produces on ly one signal. Th erefore, we \vould expect a ll sweeteners--sugars and artifidal sweeten ers alike-to produce the same sweetness. However, artificial sweete ne rs like sacch ari n and aspartame d o no t taste exactly like su gar; if they did, there would be no need to continually search for better artificial sweeteners. Some claim tha t arti ficial swee teners Produce additio na l tastes tha t -account for the difference. For example, saccharin tastes bitter as well as S\Veet to many. But some of us (the au th ors included. ) do not taste the bitterness of saccharin a t all; \-V"e are quite ronvin ced tha t it is the na hu-e of th e sweetness that differs. Geneticsh1dies may offer some help he re . TI1e receptor TlR3 appears to be able to fun ction alone to respond only to high of sucrose (Zhao et al., 2003). In addition, Margolskee and colleagu es (Yee et al., 201 1) have identified yet an other mechanism in addition to the h eterodi mer that can mediate sweetness. Thus we can exp lain w hy m.."l. ny of us perceive diffe rences between th e sweetness of s ucrose and those of artificial sweetene rs. Sin ce our taste system can produce sweet receptors so precisely tuned to the biologically usehtl su gars, \vha t is going on wi th artificial sweeten ers? Are the.se nonsu gar molecules sweet because they accid enta lly stimula te the S\\' eet heterodimer? We don ' t know. Are artificial sweetene rs medically useful? Certainly they en able diabetics to enjoy a swee t taste wi thout the d an gers of sugar, but the ea rly hopes tha t artificia l sweeteners wo uld be a p an acea fo r weight loss now appear VvTon g, Saccharin w,'l.S discovered in 1879 w hen the chem.ist [ra Re msen, working on coal tar d eriva tives, failed to wash up before dinne r and subseciuently no ti ced tha t the ta r residue on h is han ds tasted sweet. Another artificial sweetene r was

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

heterod mer A chain of t1MJ molecules that are different Iran each other.

480

CHAPTER 15

nontaster (of PTC/PROP) An lrdlvldual born "1th t1M> recessive alle>is for the TAS2R38 gene and unable to taste the ccmpounds phffiYlthlocarbarnlde and prcp)1thlruracll. taster (Of PTC/PROP) An Individual bcrn with one a OOth dominant alleles for the TAS2R38 gene and able to taste the ccmpounds phffiYlthlocarbamlde and prop)1thlouracll. PTC/ PRO P tasters also hao;e a high denslrf of funglform papillae are PROP

wro

supertai::.ters.

(a) Pheny lthiocarh1n-Ud.e H

H 1

H

II

I N - C- NH,

II

I

-

s

I H

(b) Pmpy lthiourocil H

"c,,. .

t

I

CJ.L,CH 1 CH3

I 111n

II FIG UR E 15.1 o The chemical structur€1.1een taste nerves and between taste arid other oral sensations. Recentl}'i indi viduals with his tories of o titis media (usually in infancy ordllldhood), tonsillec tom y, o r head injury were fo und to gain weight (Bartoshuk, 2012; BartoshtL.k., et al. 2013). Sensory tes ting revealed a likely scena_ri o. The localized taste damage released inhibition on und ..unaged nerves s uch that whole- mouth taste, perception of tou ch (fats in foods), and retronasal olfaction all increased. T11ese sensory changes were associated with enhanced palatability o f high-fut foods, which could potentially lea d to weight gain.

Wisdom of the Body: How Do We Solve the "Omnivore's Dilemma" ? Some species h ave few food ch oices to ma ke; for exarnple, koala bears eat primarily leaves from the eucalyptus tree . However; humans (and rats1 rue omnivores; we are confronted with an rtJTa.y of choices, and we must foods and avoid poisons to s urvive. Paul Rozin coinevired basic tastes and learned responses to food flavors.

The Special Case of Umami Umaml arose as a candidate for a fifth basic taste as part of advertising claims by manufacturers of monosodium glutamate (MSG), the sodium salt o f glutamicacid. Identified by Japanese chem ists in the early 1900s, MSG was initially marketed as a flavor enhancer, said to suppress unpleasant tastes and enhance pleasant ones. Tas te experts expressed skepticis m, MSG manufacturers went on to claim tha t MSG was a fifth basic taste, speculating that it signaled protein and thus played an important role in nutrition. Unfortunately, although having special receptors for proteins might be nutrition.ally h elpful, protein m olecules are too large to s timulate taste or olfaction . Glutam..nd to the sugars that are biologically useful to us: s ucrose, glucose, and fructose. S...nocular dlspanty alma. Named after the one-eyed Cyclops of Homer's Ooyssey. 161 cytochrome oxldase (CO) An enzyme used to reveal tl1e regular array of "CO bbbs. " which are spaced abcul 0.5 rnllllmeter apart In the pnrnary visual cortex. 13] D decay The part of a oound during which amplitude decreases (oflSeU. 11 0] decibel (dB) A unit of measure for the physical Intensity of scund. Decibels define the dltterenoo between two sounds as the ratio betlNeen two souM pressures. Each 10:1 sound pressure ratio equals 20 dB. and a 100:1 ratio equals 40 dB. 191 dermis

The Inner o f two major lay -

ers of skin, consisting of nutritive and oonnectrve tissues, within which He the mechanoreooptors. 11 31 deuteranope An Individual who suffers from color blindness that Is due to the absence of M-coneG. 151 dlchopUc Referring to the presentaticn of two different stimuli, cne to each eye. Dltterent from binocular presentation, which oo..jd lrNdve both eyes k:oklng at a single stimulus. 161 diffuse bipolar cell A bipolar retinal

cell

processes are spread out to

reoolve Input from multiple oones. 121 diopter (0) A unit of measur.,nent of the optic power of a lens. It Is equal to tl1'l reciprocal of the focal length. In meters. A 2-dlopter lens w111 bring paral· lei rays of light Into focus at Y, meter (50 crn). 121 dlplopla Double vision. If >Asble in both eyes, stimuli falling outside of Panum's fuslonal area will appear dlploplc. [6] direction Tl1e line one moves along or faces. with reference to the pdnt or region one Is m0\1ng toward or facing. 11 21 directional transfer tunctlon (DTF) A measure th9:t describes how the pinna. ear canal. head. and torso the Intensity of sounds with

dllferent frequencies that arrive at each ear from different locations In space (azimuth and eJevatlon). 1101 distractor In v1sual search1 arr;1 stimulus other than t11e target. 171 divergence The ablllty of thl two eyes to turn outward . often used In crder to place the two Images of a fea· ture In the workJ on correspond ing locations In the two retinal Images on tl1e fovea of each eye). DIVergence reduoos the disparity of that feature to zero (or nearly zero). 161 dizziness A commonly usOO lay term that nonspeclfioally Indicates any forrn of peroolved spatial disonentatlon. with cr without Instability. 11 2] doctrine of specific nerve energies A doctrine. fonnulated by Johannes MOiier. stating that the nature of a eensatk:fl depands on which sensay fibers are stimulated. not on hoW fi bers are stimulated. 111 dorsal column-medial lemnlsc al (DCML) pathway The route from the spinal cord to the brain that carrlee sQnals fran skin, muscles. tendons, and Joints. 11 3) dorsal horn A region at the rear of the spinal cord that reoowes Inputs from receptors In the skin. 11 31 double dissociation

The phenom·

enrn in which one of tvvo functions.-

such as hearing andsl;Jht, or first- and

second· order motion - can be dam· aged wtthoot 11arm to the ot11er. and vice versa. 14, BJ double-opponent cell A cell type. found ln tl1e visual cortex, In which one region is excited by one cone type . comblnatk:.n of cones. or color and inhibited by the opponent cones or color (e.g .. R+.G-J. Another adjacent region would be Inhibited by thl fi rst Input and excited by the sooond In this example. R-/G+). 151 dualism The Idea that the mlr-.:t has an e.xlstenoe separate frorn the mat€flal """Id of the body. 111 duplex In reference to the retina, consisting of two parts: thl rods and cones, which operate under different conditions. 121 E ear canal The canal that corducts sound vibrations from the pinna to the tympanlc membrane and prevents damage to the 1)1npanlc mernbrar1'l. 191 eccentricity The distance between thl retinal Image and the fovea. 121

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

499

efferent commands lnforma· tlon flowing outward from the central nervcus system to thl penphery. A ocmrnon example is motor cornrnands that regLdate muscle contraction. The ccpy of such motcr commands Is often callee! an "efferent copy." See alsoeffer· ent fiber. 11 21 efferent fiber A neuron that carries Information frcm the central nervous system to the periphery. Compare afferent fiber. 19, 12) egocenter The center of a referen:::e frame used to represent locat1ons rela· twe to the body. [1 3] electroencephalography (EEG)

At•ctJil;Jue that provide most of the bumpy appearance. Flllfonn papillae have no taste !Unction. (1 5) filter

An acoustic. electrical. elec-

tronic, or optic device. lnstrurnent 1 computer prc.Qram, or neuron that allows

the passage of some range of parame-

Integers that Increase wlt11 relatr1e freq...,ncy. (11)

Fourier analysls A mathematical procedure by which any signal can be

separated into cx:-:rnp;::nent slne wtwes at different frequencies. Combining these sine waves will reproduce the original signal. (1 , 12) fovea A smaJ I pit, nearthe centEf of the macula, that rontalns the highest ccncentratlon of ccnes. and no rods. It Is the portion of the retina that produces the highest visual acuity and serves as the point of flxatlm. 121

frame of reference The COCfdlnate systEfn used to define locatlms In spaoe. (1 3]

tree fusion The technique o1converg Ing (crossing) 0< d werglng the eyes In order to view a stereogram w1tl10ut a stereoooope.

161

frequency Fa sound, the number of times per sooond that a pattern of pressure chan;;ie repeats. Freciuency is peroelved as pitch. (9) functional magnetic resonance Imaging (IMRQ A variant of magnetic resonarce !magln;J that makes It possible to measure localized patterns of actwlty In the brain. Activated

neurons provoke increased blood ftow, which oan be quantified by measuring changes in the response o1 oxygenated and deoXl'!Jenated blood to strong magnetic fields. (1)

fundamental frequency The lowest-frequency canr.:onent of a ocmplex

ters (e. g., crlentatlons, frequencies) and

peI of thought. 141

Gestalt

In German. literal tt "form."

In reference to peroeptlm, a schcd of thought stressing that the perceptual w1101e oould be greeter than the apparent sum of the parts. (41 gestation Fetal development during pregnancy. (141

glabrous In reference to skin, hair. (1 31 global superiority effect The fimlng In various e:xpenments that the prcpertles of the 1.vhole object take preoedenoe over the properties of parts of 1he ct>ject. (4 I glomerulus

Any or 1he spherlC stating that two elements will tend to group together If they seem to lie on the sarre ocruour. (4] graded potential An electnc al potential that can vary continuously In amplltude. (2] granule cells

Like mltral cells, gran-

ule oells are at the deepest level of the dfactorybulb. Theyoompnse an

gustatlon

The sense of taste. (141

H hair cell Any c-dlmenslonaJ projection of that world. 161 proprto ceptlon Perception mediated by kinesthetic and Internal recep tors. 1131

An Inability to rec-

ccnstant frequency at lm reasing lnt'en· sltles. 191 r eactlon11 rno (RT) A rneasure o f the time !rem the onset of a s!lrrulus to a response. l7J realism A phliosq:ltlcal position arguing that there Is a real wor1d to sense. 16] receive r o perati ng charac t erlsUc (ROC) c urve In studies of signal detection, the graphb al p lot of the hit

rate as a function of the false·alarm rate. If these are the sarre, pd nts fall on the diagonal, Indicating that the

observer canrot tell the dlff€fence

protanope An indivk:l ual win suffers from color b lindness that Is due to the absence of L-cones. 151 proximity A Gestalt grouping rule stating that the ta1dency of two fea-

tures to group togeth€f w111 Increase as the distance between t hem decreases. 141

psychoacoustics

The study of the

psychological correlates of the physical

dlrnensJcns of acoustics: a branch of p syd>0physbs. 191

psychophysics

rate-lntens tty func tion A graph p lo ttin g the firtng rate of an auditory

nerve fiber In response to a soum of

projective geometry

prosopag nosla ogn ize faoas. 141

and tunher stimulation ts Incapable of lncr-ng the firing rate. [91

The science of

defining quantltatr..m otolith agans. A saclike structure that con -

tains the saooutar macula AJso called I121 salience The vlvK1noos of a stimulus relative tQ 'ts neig hbors. [7] salty of the fOLr basic tastes; t11e taste quality produced by the cations of sallts (e.g., the sodium In oodlurn chloride produoes the ealfy taste). Some cations also proooce other taste qualities (e.g., potassium tastes bltt€f as wen as ealty). The purest salty taste is prod uced by sodium chlonde (NaCl), common table salt. I151 scatter To disperse something such as light-In an Irregular fashion. 12] S8Cct//U8 '

scene-based guidance lnfcrmatlon In our urderstandlrg of scenes that helJOS us find speci e objects In scenes (e.g .. objects do not float In air. fauoets are near sinks). [7]

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

507

scotoma Tile blind spot In the middle of the visual l eld. 121 scotoplc Referring to light Intensities that are bright enough to stimulate the rod receptors but too dim to stimulate the oone receptcrs. 151 second- order motion The motion of an object tllat Is defined by changes In contrast or texture, but not by luminance. 181 sele ctive attention The form of attention Involved when processing Is restricted to a subset of the possible stimuli. 17] semi cl rcular canal A11y of three torddal tubes In the vootlbular system tllat sense angular motion. 11 21 sensaUon The ability to detect a stimulus and. perhaps, to turn that detection Into a private experlenoe. [1 J sense of angular motion The perceptual modality that senses rotation. 11 21 sense of linear motion The perooptual modality that senses transtat k::n . 11 2] sense of tilt The perceptual modality that senses head lrcllnatlon with respect to gravity. 11 21 sensitivity 1. The ability to respond to transmitted slgras. 2. In signal detectloo thecry, a value th:tt defines the ease 1._th whk::h an observer can tell the difference between the presence and absence of a stimulus or the difference betV\fSen stimulus 1 and stimulus 2. 11 , 21 sensorineural hearing loss Hearing loss due to defects In t11e cochlea cr auditory nerve. 191 sensory exafference Change In afference caused by external stlmull. For the vestibular system, vestlWar afference evoked by passive head motion would )'leld sensory exafference. Compare senscvy reefference. 11 21 sensory Integration The process of combining different sensory Typically. com bining several signals yietls more accurate and/or more precise Information than can be obtained from individual sensory signals. This is rlOi the mat11ematlcal proooss of Integration learned In calculus (e .g. 1 t11e Integral of aocel8'atlon Is velocity). 1121 sensory reafference Change In afferenoe caused by self-generated actMty. For the vestibular system. vestibular afferenoe evcked by an active -generated head motion would )'leld

508

GLOSSARY

sensory reafferenoo. Compare sensory

exafference. [12]

serial self-termlnaUng search A search trom Item to Item, ending

when a target Is found. [7] set size The number of !terns In a Visual display. 171 shape.-pattem theory lhe current dcmlnant biochemical theory for how chemicals ccme to be perceived as sped! c odors. Shape-pattern theory conten::Js that different scents- as a functic:n of the flt between odorarit shape to OR shape- activate different arrays of dfactory receptcrs In the olfactory eplthelia. ll'1ese varklus arrays produce spedl c !nng patterns of neurons In the olfactory bulb, which then determine the partlrular scent we parceve. [141 sharper tuning

An effect of attention

en the response of a neuron In wtilch

the neurcn responding to an attended stlmulus responds mcra predsety. For

example, a neuron that responds to lines VY!th or1entatlons from -20 degrees to +20 degrees might come to respord to ± 10-degree llnea. [7] signal detection theory A psychcphyslcal theory that quantrnea the reSfXAISe of an observer to the presentation of a signal In the presence of noise. Measures obtained from a series of presentations are sensitivity Id') and criterion of the observer. [1] similarity A Gestalt grouping rule statln;i that the tendency of two features to group together will Increase as the be™""1 them Increases. [4] simple cell A cortlcal neuron wtose receptive field has defined excitatory and Inhibitory regions. [3] simultagnosla An Inability to perceive more than one object at a time. Slmultagrl061a Is a cor-.sequence of bilateral damage to the parietal lcbes (Balint syndrcme). [7] sine wave A simple, smoothly changing ooclllatlcn that repeats across space. Hlgl1erlfe)'.1uemyslflllw'""'s have more osc:llatbns and lo.vef trequendes haVe f.......- oscillations CNer a given distance. 1. In hearing, a waveform for which as a fLnCtlon of time Is a sine 1Ll'JC11on. Also caled pure tone. 2. In \lislon, a pattern for v..tllch vanatlrn In a property like brightness CT color as a function of space Is a sine function. [11

The wavefam for wl1icl1 varlatlcn as a tunctlcn of time Is a sine f]..!nctlon. [9] sine wave grating A grating with a sinusoidal rumlnarce profile as shown in Figure 3.4b. [3] single-opponent cell Another way to refer to OCfle-opp::nent cells. In order to differentiate them frcm double-opponent cells. [5] sinusoidal Referring to any ooclllatlon. SL.Ch as a so.Jrd wwe CT rotaticnal motion. v.ih::t>e waveform Is that of a sine curve. The period of a sinusoidal osci llation Is the time that rt takes for one fUll back-and-forth cycle of the motion to occur. The frequency of a sinusoidal osclllatl:::n Is de! ned as the numeral 1 dM::fed by the period. [1 2] smooth pursuit A of volu ntaty movement in which the eyes move smoothly to follow a moving obJeot. [8] sine wave or pure tone

somatosensation

Cdlectlvely,

sensory signals from the skin. muscles. tendons, joints, and Internal receptas. [1 3] somatosensory area 1 (51) The primary receVlng area for touch in the cortex. [1 3]

somatosensory area 2 (52) The secondary receiving area for tou::::h in the cortex. [13] somatotoplc Spatially mapped in the somatoeensory cortex ln correspondenoe to spatial events on the slbility across the adult life span Aging Neuropsycho1

Cogn l 300-311

Aubert, H. (1861).

bedeutende Drehung von Objekten bei Neigwlg des Kopf es nach rechts od er links. Vird1ows Ard1. 20:

381-393. Aubelt, H. (1886). Die Bewegungs-

Arclr Ges Plr!f5iol 39, Ayabe-Kanamura, Schick.er, I., Laska1 M., Hudson, R., Distel H . Kobayakawa, T., and Saito, (1W8).

s:

Diffe1·ences in pet'ception of everyday odors: A JapaneS(!-German c1oss-cultural studr Ow11Senses23:

31-38.

514

REFERENCES

Babadi, B., Casti, A., Xiao, Y., Kaplan, E., and Panin.ski, L (2010). A generalized linear model of the impact of direct and indirect inputs to the lateral ge-

niculate nucleus. JVis 10: 22.

Bahill, A. T. and Stru·k, L (1979). Tu. trajectories of saccadic eye movements. Sci Am 240(1 )o 1OS-11 7. Baloh, R. and Halmagyi, G. M . (Eds.). (1996). Disorders of t!ie Vestibular System. Oxfo1tl, UK Oxfo1d Universitv Press.

S., Shamtna, S. A, and Kanold, P. 0. (2010). Dichotomy of functional organization in the mouse auditory cortex. Nat Neurosci 13;

361-368. Banks, M S., Aslin, R. N .. and Letsot\ R. 0. (1975). Sensitive period for the de-

velopment of human binocul.aa· visi011. Science 19J: 675-677. Barclay, C. D., Cutting. J. E., and Kozlowski, LT. (1978). Temporal and spatial factors in gait that influence gl?flder recognition. Perctpf

Psyclwphys 230 145-152. Baring a, M. (2002). How the brain's clock gets daily enlighteruu ent. Scimce

2950 955-957. Barlo\.,', H B. (1972). Single units and sensation: A neuron doctrine for pe1a::!ptual psychology. Percrption 1:

371-394. Barlow, H. B. (1995). The newon doc-

ability to taste the bitter substance 6-n-p1opylthiouracil (PROP). Science

2050 934-93S. Ba!toshuk, L M. (1991). Taste,smell and pleasure. In R C. Bolles (Ed.t Tile Hedonicso/Ta.ste (pp. 15-28). Hillsdalo, NJ: Erlbatun. Baitoshuk, L M .,. Cartalanotto, j. A., Hoffman, H.J.. Logan, H. L, and Snyder, D.J. (2012). Tastedamago (ot!tis media, tonsillectom.y and head and neck cancer) can intenify oral :wnsations . Physiol BeltliV 107: 516-526.

Bartoshuk, L. f\t, Fast, K., and D. (2005). Differences in our senso1y wol'lds: Invalid comparisons with labeled scales. Ct1rr Dir Psucho/ Sci 14:

122-125.

, Battoshuk, L lvl and Kloe, H.J. (2013). Bette1· fmits and vegetables tluough sensory analysis. Crm Biol 23: R374R378.

Bartoshuk, L M.,. Mad.no, S., D. J., and Stamps, J. (2013). Head tratuua, taste damage and weight gain. Chrm Smses 3B: 626. Baitoshuk, L. M. and Wolfo,j. M (1990). Conditioned taste aversions in hum.ans: Are- they olfactory avet'sions?

Oiem Se11Srs 15: 551. Bash!onl,j. A. and R. M. (19.s7). Multiple phonemic restorations follow the tttles for audito1y induction.

Pem.'Pt PsycJr::ipJ1ys 42: 114-121.

tt'ine in pen:eption. In M. S. ·Gazzaniga {Ed.), The Cognitive (pp. 415-435). Camb1idge, MAo MIT Pt'ess.

&lsson, .M. D., Bartoshuk, L M .. Dichello, S. Z , Weiffonbach, )., and Duffy, V. B. (2003) . Colon cance1· and genetic variation in taste. Oiem Senses 28: 109.

Barlow, H B.,. Blakemore, C., and Pettigrew, J. D. (1967). The now·al mocha-

Bates, LA., Sayialel, K. N.r Njiraini, N. W., Moss, C. J., Poole, ). H ., and Byrne, R. w_ '1007. Elephants classify htunan ethnic groups by odor and garment oolor. C11rr Biol 17, 1938-1 942. Baldauf, D. and Desimone, R (2014).

Biulow, H. B. and Levicl

1484-1525.

Carskado n. M and Herz, R S. (2004). Minimal olfacto1y perception during

sleep: Wl1y odo1· alarms .,'Ii.I I not work for htu11ans. Sleep 'Zl: 402-405.

Carskadon, M. A., Wyatt,). , Etgon, C., and Rosekind, M. R (1989). Nonvisual sen..."01')' experiences in dream s of college students. Sfeep Res l S: 1S9.

Casagrande, V. A., Yazar, F., Jonos, K D.,

Chapuis, L Messaoucli, B., Ferreira, C., and Ravel, N. (2007). Importance of retronasal and o rtho nasal olfaction for odor aversion memory in rats. Bellav

Neurasci 121: 1383-1392.

and Ding. Y. (2007). The m o1phology of the koniocellulat· axon pathway in the macaqu e monkey. Cereb Corte.r 17:

Chaudhari, N., Pereira, E. , and Roper, S. D. (2009). Taste 1·eceptors for tunami.: the case for multiple receptors. Am JClin Nutr90:

Castelhano, M. S. -and Heave1\ C. (2010). lh? relative conhi.bution o f scene context and target features to visual search in scenes. Aum Percept Psychoplt!f' 72, 128'l-1297.

Chaudhari, N. and Rope1; S. D. (2010). The cell biology oi taste.J Cell Biol l'Xlo 285-296. Chen, X , Cabitto, M., Peng, Y., Ryba, N. J. P., and Zukor, C. S. (2011). A

2334-2345.

S., and Masson, C. S. Castet, E., (2001). "Saccadic supp1·ession" -no need fo1· an active extra-1't'tinal m echanism. Trends Nemo fd 24: 316-318. Castro , J. 8., Rarnana.than, A , and Chennubhotla, C. S. (2013). Catego1ical dimensions of hu.m,m odo1' descriptor space revealed by non-negative matrix factorization. PLoS ONES.: e73289. doio10.1371 / joumal.pone.0073259 Cavanagh, P., Hunt, A. R , Afraz, A., ai1Cl Rolfs, M. (2010). Visual stability based o n remapping of attention pointers.

Trends Cog11 Sci 14' 147-153. Cavll11'lgh, P. and Leclerc, Y. C. (1989). Shape from shadows. / Exp Psycliol Hum PerC't'pt Perfonll 15: 3-27. Cave, K. Rand Bichot, N. P. (1999). Vis uo-spatial attQntion: Beyond a spotlight model Psjptior1. Pilysiol Bel•w '71: 21'.l-228.

Doty, R. L.,Shrunan, P., Dami, M. (1984). Development of the Uni\'ersity of Pennsyln nia Smell Identi.£ication Test: A standardized m icroencapsula ted. test of olfactory fw1ction. Physiol

BflCe visual search. Scie.uce 247: 721-723. Enroth-Cugoll, C. and Robson, J. G. (1984). Functional dmracteristics -846. FellemaJl, D. J. and Van fuoen, D. C. (1991). Disttibuted hierrucllical proa;issing in the primate cerebral cortex. Cereb Conex 1: 1-47. Fernandez, C. and Goldb'1·g, J. (1971). Physiology of peripheral neu!'ons innervating semicircular canals of the squirrel monkey. II. Response to sinusoidal stimulation and dynamics of pet'i phe1\U 1;estibulat' system.. JNerirophysiol 34: 661-675. Fettiplace, R. and Hackney, C. M. (2006). The .se.nso1y and motor toles of auditory hair cells. Nntun 7: 19-29. Field, D. )., Hayes, A , and Hess, R. F. (1992). Contour integmtion by tlie human visual system: Evidena;i for a loca1 "association field." Vision Res 33: 173-193. Field, T. M., Schanberg, S. M., Srnfidi, F., Bauer, C. R, Vega-Lahr, N., Garcia, R., Nysttom,J .. and Kuhn, C. M. (1986). Tactile / kinesthetic stimulation effects on pretei.m neonates. Pf'diatrics 71: 65,µ;5s. Fielder, AR. and Moeeley, M. J. (1996). Does stereopsis matter in humans? Eye m 233-238. Fh'estein, S. (2001). How the olfactory system makes sense of scents. Nature 413: 211-218. Fischer, R. and Griffin, E (1964). Pharmacogenetic aspects of gustation. Drng Ro; 14: 673-686. Fisher, S. Kand Ciu!fi-247. Hayes, ). E., Bartoshuk, L M., Kidd,). R., and Duffy, V. B. (2008). Supettasting and PROP bittl?rness d epends on more than the Tas2.r38 gene. 01em

Se11::cts 33: 255-265. Hayward, W. G. and Williams, P. (2000). Viewpoint dependence and object

discriminability. Psydiol Sci lt 7-12 Hedger. S. C., Heald, S. L M., and Nu.sbatun, H . C. (2013). Absolute pitch may not be SQ absolute. JAcoust Soc Am 24: 1496-1502. Hedner, fi.·1., Larsson, M., Arnold, N., Zucco, G. M , and Hummel, T. (2010). Cognitive factors in odor detection, odor discrimination, and odot' identification tasks. JClin Exp Neuropsych

32, 1()12-1067. Heeger, D. (2006). Visual motion tion. Lecture notes.. New Yot'k University, Depai1ment of Psychology. pen::eption/ lecturenotes/motion/ motion.html E. R (1972). Univers als in color naming and m e:rn o1y. J Exµ Psycho/. 93:

10-20. Heise, G. A and Mlller, G. A (1951). An expm.'imental study of a t1dito1y pattems. Am JPsycllol 64: 6&-77.

C., Baja, j ., Lenz, F.

J. U., Hormann, K , Hen.RM_ wi.d

Stuck, B. A. (2012). Effects of an ai·tifidal smoke on arousals during hwnan sleep. CJiemosens Percept 5: Tl4-279. Held, R. T. and Hui, T. T. (2011). A g uide to stereoscopic 30 d.isplays in m edi-

cine. Aard Radiol 18, 1005-1048. Held, R. T., Coopr, E. A., and Banks, MS. (2012). Blur and disparity are

complenwntary cues to depth. Curr

Biol 2'.C 426-431.

Held, R., Ostrovsky, Y., deGelder, B., Gandhi, T., Ganesh. S., M_, and Sinha , P. (2011) Newly sight«! cannot ma tch seen with felt. Nnt Ne11rof330 309-319. Lam, R. S. and P. (2013). Odorant r12:Sponsiveness of embryonic mouse olfrtcto1y sensory neurons

the oclora.nt receptot·s Sl or

MOR23. Er1r J Nerirruci38o 2210-2217.

LaM-. R.H. and Srh1lvasart, M.A.

(19JI). Sutface microgrometryoTactile perception and neural encoding. In

O. Frol\Z4'f\ at1d ). Westn•"' (I.Ms.),

1"formntio11 Proceg;i11g ;n the SomntC>System (pp. 49-SS). London : Macmillan.

Lmd E. H. • nd Mc:Cann, J. J. (1971). Lightness and retinex theoty. JOpt Soc

A1116l: 1-11.

Larsson , M. and Willander,J. (2009) . Autobiographical odor memory. An11 NY Acad Sci 1170: 318-323. Laska, M., Koch, B., Heid, B.. 5: 157-170. Miller, C., Ti·btu', ). M., and Jordan, B. D. (2007). Ovulatory cycle efiects on tip ea.m.ings by lap dancers: Economic evide nce fo:t· human estrus? Eool Hwll Beh•v 28, 375-38L Miller, G. A. ai1d Hei.se, G. A. (1950).

The trill threshold. JAcottsf Soc A m 22.:

637-636. Miller, S. L. and Maner j. K (2010). Scent of a woman: Men's testostel'One responses to olfactory ovulation cues.

Psyd10/ Sci 21, 276-283.

receptor activation of gustdud n inhibits gustatory responses to bitter compounds. Proc. Natl Actrd Sci USA 9& 9903-9906. Miranda, M I. (2012). Taste and odor recognition memory: The emotion..11

flavor of life. Rev Neuroscience 23: 481-499

Miya,..•aki, Y., Uchida, H ., Yamashita, 0 ., Sato, M, .Morito, Y., Tanabe. H , Sadato, N., at'id Kamitani, Y. (2005). Visual image reconstruction fl'Om human brain activity using a combination of multiscale local image

de.rcoders. Neuron 60: 915-29.

Mombaerts, P., Wang, F., Dulac, C., Chao, S. K.. Nemes, A., Mendelsohn, M., Edmondson, Land Axel, R. (1996). Visualizing an olfactory sensory map. Ce/187, 675-t in mt insular cortex w ith impaitW ex pression of taste aversion learning Proc N,11 AmJ Sci USA doi: 10.1073/

pnas.1315624 lll . Schlld, D. and Restrepo, D. (1998). Transduction mechanism i.n \ 'e t'tE!brate olfactory cells. PJiysiol R£v 37: 369-375. P.H. and Sandell, ). H. (1983). Tnternctions bet\\'E!en visually and electrically eli:dt:W saccades before and after superio1· coUiculus and frontal eye field ablations in the rhesus monkey. Exp Brain RRs49:381-392.

Schleidt, M., Hold, B., and Attila, C. (1981). A aSS-O.tltttral study on the attitude towatds pel"sonal odors. I Oiem Ecol 7: 19--31. Sclmapf, J. L., Kraft, T. W., ruid Baylor, D. A. (1987). Spectml sensitivity of hu1na n oone photo1'0C'eptms. Natme 325: 439-441. Schoonevelt, C. P. and B. C. J. (1989). Comodulation masking release (CMR) as a function of m asker ba.nd,..·idth, modulator bandwidth, and signal duration./ Aco11st Soc Am

85:m- 2Sl . Seay, C. F. (1975). First 1710ugJ1ts 01111171eologyof M1,sicfrDm tile Psalter. Dallas, TX: Dallas Theological Semi.na1y. Seeba, F. and Klump, C. M. (2009). Stimulus farniliarity aHects perceptual restoration in the European starling f.Strm1Us v1ugaris). PU>S e5974.

doi'1 0.1371/ jot11nal.pone.OOOS974 Seidl, A. H. and C.-othe, B. (2005). Developmen t of sound localization mechanisms in the Mongolian gerbil is s haped by early acoustic: experi-

ence. / Nr11ropl1ysiol 94: 1021'>-1036. Selfridge, 0. G. (1959). Pandemonium: A

pal'adigm for k>arning . h1 0 . V, Blake and A. M. Uttley (Eds.), Procttdings of the Symposirrm 011 the of Tllorigllf Proc"5es (pp. 511-529). l.ondon: Her Majesty'..sStatione1yOffia:i. Serenc:es., J. T. and Yantis, S. (2006). Selective visual attention and perceptual coherence. Trends Cogn Sci 10: 38-45. Serre, T., Oliva, A., atid Poggio, T. (2007). A feedfo!"ward ar chitecture accounts

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

531

fo1· rapid categorization. Proc Natl Acad Sci USA 104: 6424-6429. Shapiro, K. L. (1994). Th• attentional

blink: The brain's eyeblink. Crnr Dir

Psycho/ Sci 3: 86-$9. Shapley, R. and Hawken, M. J. (2011). Color ln the rortex: Single- and double-oppone nt cells. Vision Res .51: 701- 717. Shruan, L., Rosenholtz, R. and Adelsot\ E. H. (2009). Material pot«>ptio1' What can you see in a brief glance? JVis 9: 784, "W>'/W.jomnalofvision . org / rontent/ 9/ 8/ 784, doU0.11 67 /9. 8.784.

Shepard, R. N. (1967). Recognition memot'y for \.,·ords. sentences., and

pictures. JVerL\11 Lenm VedMI Belum 6: 156-163. Sherl"ick C. and Cholewiak, R. (1986). Cutaneous sensitii,.·fty. ln K. R Boff, L. K.uimai\ and) . P. Thomas (Eds.),

Hfl11dbo'1lxm1 infants. In J. F. Bc&na (F.d..), Development hi the Fews and lllfant (pp. 254-278'). Washington, DC: U.S. Qwermnent P1'inting Office. Stel lman, S. D . and Carkinkel. L. (1988). Patterns of artHida.l sweetener use and weight change in an Amel"irnn Cancer Society Prospective Stud)' Appetiff 11 : SS-91 . Stevens, ). C. (1959). C ross-modality

validation of s ubjective scales for loudness, vibration,. and elect1ic

shock. JExp Psycho/ 57, 201-209. Stevens, ). C. and Cai1\ W. S. (1987).

Old-age deficits in the sense of smell as gauged by thresholds, magnitude

An evaluation version of novaPDF was used to create this PDF file. Purchase a license to generate PDF files without this notice.

McGraw-Hill. by electt'ical stimulation of monkey superior rolliculus. E.xp Bmin Res 23: Hll-112.

Sulmont, C., fosanchou, S., and Koster, E. P. (2002). Selection of odorants for memory tests on the basis of fomiliarHy, perceived complexity, pleasantness. similarity and identification. Chem Senses 27: 307-317. Su" Y-C., Zhao, Z.-Q., Meng, X.-L., Yin, ) ., Liu, X.-Y., and Chen, Z.-F. (2009). Cellular basis of itch sensation. Scimce 325, 1531-1534. Suzuki, Y and Takeshima, H. (2004).

Equal-loudness-level contours for pure tones.] Aco11st Soc Am 116: 918- 933. Svoetichin, C. ond Macnichol, E. F., Jr. (1959). Retinal mechanisms fo1· chro-

matic and achromatic vision. A1111 NY Acad Sci 74: 385-404 . Szma.jda. B. AFCriinert, U., and Martin, P. R. (2008) . Retinal ganglion a>ll

inputs to the koniocellular pathway.] Comp Ne11rol 510: 251-268 . Takeuclli, A.H. and Hulse, S. H . (1993). Absolute pit Vlad61/Shutterstock. 5.29bo© Marcel Mooij / Shutterstock. 5.29c: o5 Dynarnic Crnphics Croup/Creatas / Alamy.

Courtesy of Andrew D. Sinauer. 9.7 mo11key: e c.'\Sxsoonxas/istock. 9.1 l a: Courtesy of A J. Hudspeth. 9.26' Courtesy o£ Cochl"ar. CHAPTER 10

mond. Nei l Farrimond./ ShutterStock. CHAPTER 11 11.lSbo Photo by David Mcintyre, cow"tesy of Aru1ie Huyler, Pioneer Valloy Chinchillas. 11.16 mo11key: © GlobalP/ i.stock.11.16cow: nl / istock. 11.19: C0tutesy of Nanc:Y Asai and Joanne

Delphia. CHAPTER 12 12.300 ©Grega Roz.1c/

Alamy. CHAPTER 13 13.UoV&A lniages/ Alamy.13.ll oFrom Rainville et al.,

1997. CHAPTER 14 Ore1iero © Mylaphotography / Dreamstime. 14.6: © Afri.Pics. com/ Alamy. 14.8bo© Sorge Kozak / Corbis. 14.15' David Mclntyr