12 higher-order and subjective aspects of perception

8/12/2019 12 Higher-Order and Subjective Aspects of Perception

1/16

12Higher-Order and Subjective Aspects

of Perception: How Are Low-LevelStimulus Properties Transformed intoHigh-Level Percept Qualities?

The Dynamic Core Hypothesis . . . . . . . 3733 What Kinds of Observers Can Produce

Conscious Perceptual Experiences? . . . . . . 374Arguments Based onGeneral Principles . . . . . . . . . . . . . . . . . 374Arguments Based on

Empirical Validation . . . . . . . . . . . . . . . . 374Validation Based on Correlationwith Actions . . . . . . . . . . . . . . . . . . . . 374Validation Based on aTuring Test . . . . . . . . . . . . . . . . . . . . . 375Validation Based on Correlationwith Neural Processing . . . . . . . . . . . 375

4 An Overall Conceptual Schemeof Perception . . . . . . . . . . . . . . . . . . . . . . . 376Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 376Selected Reading List . . . . . . . . . . . . . . . . . 377

362

1 Neural Correlates of Percept Qualities . . . 363Early Processing Primarily ReflectsLow-Level Image Properties . . . . . . . . . 363Neural Correlates ofIllusory Percepts . . . . . . . . . . . . . . . . . . . 364Influences of Attention and Memory . . . 364

Neural Activity Correlatedwith Bistable Percepts . . . . . . . . . . . . . . 365Activation of Visual Processing AreasDuring Imagery . . . . . . . . . . . . . . . . . . . 366

2 Neural Correlates of ConsciousPerceptual Experiences . . . . . . . . . . . . . . . 367

Classic Studies of Penfield . . . . . . . . . . . 367Motion Detector Single Units asBridge Loci . . . . . . . . . . . . . . . . . . . . . . . 369Blindsight as Evidence of a SpecialRole for V1 . . . . . . . . . . . . . . . . . . . . . . . 369

Questions

After reading Chapter 12, you should be able toanswer the following questions:

1. Give some examples of perceptual qualitiesthat are likely to be derived primarily from theobserver instead of the stimulus.

2. Describe a relatively low-level representation of

sensory information that can be measured inthe dLGN. Compare this with representationsseen at later stages of cortical processing.

3. What role do illusions play in allowing scien-tists to study neural correlates of percepts?

4. Summarize some of the empirical evidenceregarding where along the neural pathwayscorrelates can be found for higher-orderprocesses such as attention and memory.

5. Describe some neural correlates of bistablepercepts.

6. What brain areas are activated during visualimagery?

7. Summarize the empirical results of studies

that utilized electrical stimulation of the brainsof humans and monkeys, and the implicationsof those results for theories trying to explainthe neural correlates of conscious perceptualexperiences.


2/16

Neural Correlates of Percept Qualities 363

8. What is blindsight, and what relevance doesit have to issues of neural correlates ofconsciousness?

9. What is the dynamic core hypothesis regardinga neural correlate of conscious perceptual expe-riences, and how does this hypothesis differfrom bridge locus hypotheses?

10. Summarize the arguments based on generalprinciples about whether or not monkeys ormachines can have perceptual experiences.

11. Describe some methods based on empiricalvalidation that might be used to decide whetheror not a monkey or a machine has perceptualexperiences.

12. Summarize the overall scheme of perceptionthat has been presented in this book.

As signals travel from the receptors through thevarious stages of neural processing, a transforma-tion takes place in the information being coded.Initially, neural signals reflect primarily low-levelproperties of the proximal stimulus. However, atlater stages of processing, signals begin to take onmore and more high-level qualities of percepts.This idea was introduced in the section on Infor-mation Processing Theories in Chapter 1 (Figure1.12). Subsequent chapters have given repeatedempirical examples of the emergence of perceptual

qualities during processing:

Absorbed photons of various wavelength aretransformed into color.

Discontinuities in luminance across space aretransformed into perceived shape.

Discontinuities in luminance across space andtime are transformed into perceived motion.

Disparities in the images in the two eyes aretransformed into a single binocular percept of athree-dimensional world.

These examples suggest that perceptual processinghas the effect of adding something to the perceptover and above what was provided by the sensoryinformation available at the sense organ.

Another set of examples leading to the samesuggestion involves illusions, cases in which thepercept differs in some substantial way from thestimulus. These illusory qualities of perceptsreflect characteristics of the observer rather thanof the stimulus. A third example involves thesubjective conscious experiences that humans

frequently describe as being an integral part oftheir percepts. This final chapter focuses on thesevalue-added, higher-order qualities that aresometimes conferred onto percepts during percep-tual processing.

1 Neural Correlates ofPercept Qualities

Early stages of processing in the visual system pri-marily reflect low-level properties of the retinalimage. However, as neural signals travel along thevisual streams of processing, an edited representa-tion of the visual world begins to emerge. Informa-tion deemed irrelevant to the observer is filteredout; Information deemed important or associatedwith the observers current focus of attention areaccentuated; Information about objects whose pres-ence is remembered or inferred is added in.

Early Processing Primarily Reflects

Low-Level Image Properties

Examples of neural representation related closely tolow-level image properties can be found in theretina and dLGN. Figure 12.1 illustrates an examplein which the mean firing rates of individualneurons in the dLGN appear to code a simple rep-resentation of the relative amounts of light presentat each location of the retinal image.

The visual scene shown in the left panel wasimaged onto the retina of an anesthetized monkey

while physiological activity was recorded fromneurons in the dLGN. The right panel shows agrayscale map in which portions of the image thatare represented by a high rate of spikes are shown

FIGURE 12.1. Comparison of intensity values of theretinal image (left panel) and the rate of neural activityin the dLGN of a monkey (right panel). Adapted from

R. Srinivasan, Reading a neural code: How images are en-coded by neurons of the macaque monkeys visual system.Doctoral dissertation Graduate Program in Neuroscienceat Emory University, Atlanta, Georgia, USA, 2000, by per-mission of R. Srinivasan and J. Wilson.


3/16

364 12. Higher-Order and Subjective Aspects of Perception

in white and those with a low rate in black. Com-parisons between intensities in the image andamounts of neural activity in the topographic mapreveal a close correspondence.

When this experiment was repeated in corticalarea V1, the relationship was much less obvious,revealing that as early as V1, information is beingreencoded in forms less closely related to imageintensities. The following sections discuss several

examples in which the neural representations seenin striate and extrastriate cortical areas are relatedmore closely to the percept than to the stimulus.

Neural Correlates of Illusory Percepts

In cases in which perception is veridical, qualitiesof the percept resemble properties of the stimulus,making it difficult for scientists to disentanglewhether properties of the neural representation

derive from the stimulus or from the observer.For this reason, scientists interested in studyingsensory neural coding often study conditionsthat are known to cause illusory perception.Although these conditions may not be particularlyrelevant to understanding how perception operatesin ecologically valid conditions, they have theadvantage that neural representations of the stimu-lus and the percept are easier to distinguish fromone another.

The term illusion refers to dramatic and easily

demonstrable examples in which an observerspercept of a stimulus is distorted or properties areperceived that are not present in the stimulus. Anexample of an illusory perceptual phenomenon issubjective visual contours, discussed in Chapter 8.

A stimulus that gives rise to an illusory subjectivecontour is shown in the lower left corner ofFigure 12.2.

While viewing this stimulus, human subjectsreport seeing a vertical contour running down themiddle of the figure. However, close examinationof the stimulus reveals that the physical contourspresent in the stimulus are all oriented horizontally;there is no vertical contour present. Thus, the ver-

tical contour that is perceived is illusory.Figure 12.2 also shows the responses of a neuron

in extrastriate cortical area V2 of a monkey to thissubjective contour. The top panel reveals that thisneuron responds vigorously to a vertically orientedline swept across its receptive field. Additionaltesting of this neuron, not represented in this figure,demonstrated that this neuron has strong orienta-tion tuning and does not respond to a horizontallyoriented line. The bottom panel shows that theresponse to the subjective vertical contour is quali-

tatively similar to that elicited by a physical verti-cal contour. In other words, the physiologicalresponse of this neuron is more closely related tothe subjective quality of the percept (verticalcontour) than to the physical property of the retinalimage (horizontal contours).

Influences of Attention and Memory

Visual attention is used to direct limited process-

ing resources toward information about the aspectsof the environment that seem most relevant to theindividual observer and away from informationthat seems less relevant. There are several kinds ofvisual attention. One major type, called directed

FIGURE 12.2. A neuron in V2 respondsto an illusory percept of a vertical lineas well as to a physical vertical line.Adapted from R. von der Heydt and E.Peterhans, Mechanisms of contour per-ception in monkey visual cortex: I. Lines

of pattern discontinuity. J. Neurosci.9:17311748, 1989, by permission ofSociety for Neuroscience.


4/16

Neural Correlates of Percept Qualities 365

attention, is involved in shifting attention from oneobject to another. This can be reflexive and underthe control of bottom-up processes, as when a newobject appears in the scene, as discussed in Chapter4. However, attention can also be under voluntarycontrol of top-down processes, as when searchingfor an object having certain characteristics in a clut-tered environment consisting of many differentobjects.

A neural correlate of directed attention is foundin neurons that respond more strongly to a stimu-lus when it is being attended than to the identicalstimulus when it is not being attended to. Studiesin humans and monkeys have discovered neuronswith these properties. Directed attention can resultin small modulations of neural activity as earlyas V1 and in stronger modulations of activity inother areas of the extrastriate cortex, including V2and V4.

Another kind of directed attention is involvedin continuous tracking of more than one event.An example is the visual processing involved inplaying a team sport like basketball. A player mustcontinuously monitor the locations of opponentsand teammates. This aspect of attention has beenstudied in the laboratory by having humansperform on a multiple-tracking task in which anumber of randomly moving bouncing balls are

displayed on a screen. The individual balls are iden-tical in color and shape and cannot be distinguishedfrom one another except by their position. While theobserver is viewing the display, three or four indi-vidual balls are tagged by changing their color fora short period. Then the balls revert to being all thesame color, and the subject must continue trackingthe positions of the tagged balls.

Brain imaging during this task revealed nomodulations in activity in V1 during this form ofdirected attention. However, moderate, but signifi-

cant, modulations of neural activity were found inthe human brain areas associated with the extras-triate motion-processing stream. Greater effectswere seen in a number of parietal areas and in thefrontal cortex.

Attention can also be maintained on visualstimuli that disappear from sight temporarily, as,for example, when a stimulus being attended topasses behind another object or when the lights ina room are extinguished for a short period. Thesehigher-level aspects of perceptual attention are

often referred to as reflecting object permanence.An observer usually does not perceive that anobject goes out of existence when it disappears

briefly and then magically come back into existencewhen it reappears. Instead, a percept is formed of a

permanent object that remains present when it goesout of view temporarily. Neural correlates of theseactivities have been reported in extrastriate visualareas and in higher-order portions of the How-perception stream.

Maunsell and colleagues studied the responsesof neurons in the brains of monkeys performingon tasks in which they had to attend to a targetthat disappeared behind an occluding object. Theydiscovered many neurons in extrastriate corticalareas such as MST that not only responded tothe moving stimulus while it was visible butalso remained active while the unseen stimuluswas moving behind an occluding object. How-ever, this occurred only if the monkey was per-forming on a task that required paying attention tothe stimulus. Otherwise the neuron stoppedresponding to the stimulus whenever it disap-peared from view.

A similar finding has been obtained by MichaelGraziano and colleagues from neurons in laterstages of the How-perception stream in premotorcortex. Some neurons in these areas respond toseen objects that are within grasping distance.If the monkey is attending to an object that iscausing one of these neurons to fire, the neuronwill continue to respond in the dark if theroom lights are turned off as if the object were still

visible.

Neural Activity Correlated withBistable Percepts

Binocular rivalry can be induced in humans byhaving them view stimuli under dichoptic viewingcorditions arranged such that they are seen to movein one direction by one eye and in the oppositedirection by the other eye. Observers viewing these

stimuli report that during some periods motion isperceived in the direction of the stimulus seen withthe left eye and during other periods in the direc-tion of the stimulus seen with the right eye. Becauseconscious perception changes over time while thestimuli remain constant, this paradigm offers a wayto distinguish between neural activity related to thephysical stimulus and neural activity related to con-scious experience.

Nikos Logothetis and colleagues took advantageof this perceptual phenomenon to examine neural

correlates of percepts. Monkeys were trained toreport the direction of perceived motion underthese viewing conditions, and simultaneously elec-trical activity was recorded from directionally selec-tive units in a number of visual processing areas of


5/16


the brain. Most neurons in V1 respond to stimulibased on properties of retinal stimulation, irrespec-tive of the direction of perceived motion reported

by the monkey. However, in later stages ofprocessing in the visual motion-stream and in othervisual areas dealing with coding of higher-orderproperties, the activity of many neurons reflectedthe monkeys reported perception of motiondirection. An example of the response of a neuronin the extrastriate motion-stream is shown inFigure 12.3.

The top panel shows the response to a nonrival-rous stimulus in which the motion is in the samedirection for both eyes. The bottom panel is fora rivalrous stimulus in which the motion seen

by the left eye is upward and that seen by the righteye downward. The left panels show responses

recorded during periods when the animal reporteda percept of upward motion and the right panelsthose recorded during reports of downwardmotion. The histogram below each stimulus condi-tion indicates the average number of spikes gener-ated during presentation of the stimulus.

Examination of the histograms in the top panelsdemonstrates that this neuron gives a strongerresponse for upward than for downward motionwhen measured under nonrivalrous conditions.The bottom panel illustrates that under rivalrousconditions in which the retinal stimuli are movingin both directions, there is a stronger responseduring periods when the reported percept is ofupward motion.

Measurements of brain waves of human subjectsduring binocular rivalry have also reportedevidence of a neural correlate of the perceptrather than of the stimulus. For example, GiulioTononi and colleagues had human subjects viewa binocularly rivalrous stimulus in the form of a

vertical grating presented to one eye and ahorizontal grating presented to the other. Thegratings flickered, but at different rates. Brainactivity was monitored, and it was determined thatresponses at the same temporal frequencies asthe gratings could be recorded over widely dis-tributed areas of the occipital, temporal, and

frontal lobes of the brain. The subjects reportedwhich grating, vertical or horizontal, was beingconsciously perceived, and the brain waves associ-ated with each percept were compared. Themodulation of the brain waves at the temporalfrequency of each grating was 30% to 60% higherduring periods when the grating was beingconsciously perceived than when it was not beingperceived.

Activation of Visual Processing AreasDuring Imagery

Brain imaging studies performed in human subjectshave found that striate, extrastriate, and higher-order processing areas in the parietal-occipital andtemporal-occipital lobes sometimes become acti-vated during visual imagery even when no retinalstimulation is present.

For example, Stephen Kosslyn and colleaguesperformed a study in which human subjects were

instructed to close their eyes and visualize an objectwhile brain activity was being measured withpositron emission tomography (PET). They foundincreased brain activity in V1 that was localizedsomewhat differently depending on whether the

FIGURE 12.3. A neuron in an extrastriate cortical area ofthe motion processing stream was recorded from while amonkey discriminated the direction of motion of stimuliduring normal viewing conditions (top panel), and duringbinocular rivalry (bottom panel). The neurons response isshown in the form of histogram showing the averagenumber of spikes during the first 250msec following stim-ulus onset. This neuron gives a stronger response to upwardthan downward motion during normal viewing (top panel).

During rivalry the response reflects the monkeys reporteddirection. Adapted from N. Logothetis and J.D. Schall,Neuronal correlates of subjective visual perception.Science 245:761763, 1989, by permission of theAmerican Association for the Advancement of Service.


6/16

Neural Correlates of Conscious Perceptual Experiences 367

subject was visualizing a small, medium, or largeobject, as illustrated in Figure 12.4.

There was activation towards the posterior polewhile subjects visualized small objects, and thisactivation spread anteriorly when they visualizedlarger objects. The interpretation of this finding isthat the foveal representation of V1 is near the pos-terior pole, and visualizing a small object as straightahead would produce retinal stimulation near thefovea. Visualizing a larger object would producestimulation over a larger region of retina around thefovea and would be expected to activate more ante-rior portions of V1 due to its topographic mapping.

2 Neural Correlates of ConsciousPerceptual Experiences

Bridge locus theories make the assertion that thereare specific anatomical loci where neural activitiesgive rise to conscious perceptual experiences. Asdiscussed in Chapter 1, these ideas have a longhistory, going back at least to Descartes, whothought the bridge locus was located in the pinealgland. Similar ideas have continued to influence

our thinking to the present day. For example, theconcept of cortical streams of processing is oftentreated, at least implicitly, by modern neurosciencetexts as though these form bridge loci. For example,the concept of a motion-stream is often not only

used to refer to brain regions that process motioninformation but also assumed to give rise to motionexperiences, and similarly for streams associatedwith color, form, and depth and for the higher-order streams associated with What- and How-perception. This underlying belief in bridge lociprovides the rationale for studies of patientswith selective damage to these areas of the brain.It is assumed that these patients should have pre-dictable impairments related to specific kinds ofconscious perceptual experiences. Previous chap-ters have provided several examples of thisapproach. This chapter gives attention to moregeneral questions about whether certain anatomicalstructures are privileged, in the sense of beingnecessary and sufficient for producing consciousperceptual experiences.

Classic Studies of Penfield

One attempt to directly manipulate and studycausal relationships between brain activity and con-scious experiences was carried out by the neuro-surgeon Wilder Penfield and colleagues duringand prior to the 1960s. They studied a series ofpatients who were about to undergo removal of

brain tissue to eliminate neurological problems

such as epileptic seizures. The patients were awakeand sat in the operating room with the headnumbed by a local anesthetic while the brain wasstimulated electrically with a small electrode.

When primary sensory cortical areas were stim-ulated, patients reported simple sensations. Forexample, electrical stimulation of the primary visualcortex sometimes evoked a sensation of a pointof light in a specific region of space, and stimu-lation of the primary auditory cortex evoked asensation of a tone. Stimulation of other cortical

sites elicited more complex experiences, as illus-trated in Box 12.1.A summary of the stimulated sites that gave rise toexperiential responses in this group of patients isshown by the dots in Figure 12.5.

Reports of simple visual sensations were elicitedby stimulation of sites in the occipital lobe. Themajority of the sites giving rise to complex experi-ences were in the temporal lobe.

These findings are controversial and difficult tointerpret, because it is not exactly clear whether the

patients were reporting perceptual experiences,experiences of memories, hallucinatory ordreamlike experiences, or some combination ofthese. However, what is clear is that duringelectrical stimulation some patients reported an

FIGURE 12.4. Summary of PET results of human subjectswhile visualizing small (circle), medium (square), and large(triangle) stimuli. The activation while small images were

visualized was restricted to the most posterior region. Acti-vation for medium and large images spread anteriorly.Reproduced from S.M. Kosslyn, et al, Topographical repre-sentations of mental images in primary visual cortex.Nature 378:496498, 1995, by permission of MacmillanMagazines, Ltd.


7/16


Box 12.1Electrical Stimulation of the Human BrainCauses Perceptual Experiences

Patient RB received stimulation repeatedly atthe locations marked 5 and 7 on the tempo-ral lobe (Figure 1). A transcript of his responsesduring this stimulation is given below.

5. Patient did not reply.5. Repeated. Something.5. Patient did not reply.5. Repeated. Something.5. Repeated again. Peoples voices talking.

When asked, he said he could not tell whatthey were saying. They seemed to be faraway.

5. Stimulation without warning. He said, NowI hear them. Then he added, A little like ina dream.

7. Like footsteps walking on the radio.7. Repeated. Like company in the room.7. Repeated. He explained it was like being

in a dance hall, like standing in the doorway in a gymnasium like at the KenwoodHigh School. He added, If I wanted to gothere it would be similar to what I heard justnow.

7. Repeated. Patient said. Yes. Yes, yes. Afterwithdrawal of the stimulus, he said it was

like a lady was talking to a child. It seemedlike it was in a room, but it seemed as thoughit was by the ocean at the seashore.

7. Repeated. I tried to think. When askedwhether he saw something or heard some-thing, he said, I saw and heard. It seemedfamiliar, as though I had been there.

5. Repeated (20 minutes after last stimulation at5). Peoples voices. When asked, he said,Relatives, my mother. When asked if it wasover, he said, I do not know. When asked ifhe also realized he was in the operating room,he said Yes. He explained it seemed like adream.

5. Repeated. Patient said, I am trying. Afterwithdrawal of the electrode he said. Itseemed as if my niece and nephew were vis-iting at my home. It happened like that manytimes. They were getting ready to go home,putting their things on their coats and hats.When asked where, he said, In the diningroom the front room they were movingabout. There were three of them and mymother was talking to them. She was rushed in a hurry. I could not see them clearly orhear them clearly.

(Reproduced from page 614 of W. Penfieldand P. Perot, the Brains Record of auditory andvisual experience. Brain 86:595696, 1963, bypermission of Oxford University Press.

FIGURE 1. The brain regions marked 5 and 7 werestimulated repeatedly in patient R.B. Reproduced fromW. Penfield and P. Perot, The brains record of auditoryand visual experience. Brain 86:595696, 1963, bypermission of Oxford University Press.


8/16


experiential component that was described in per-ceptual terms such as I see such and such or Ihear so and so. Any theory of the neural correlatesof conscious perceptual experiences will have toaccount for these facts.

Motion Detector Single Units asBridge Loci

Bridge locus theories, when stated in their strongestform, assert that a conscious perceptual experienceof a complex object such as ones grandmother

can be caused by the activity of a single neuron,a grandmother cell. The strong version concept of a

bridge locus neuron was formulated as a theoreti-cal idea, but recent neuroscience studies supportthe possibility that hypotheses regarding bridgeloci are amenable to empirical tests.

The essential characteristics that would make aneuron a bridge locus are as follows: 1) The neuronshould fire in response to retinal stimulation fromwhatever object in the environment is related to theexperience. 2) If the object is present in the envi-

ronment but does not lead to firing of the neuron,the experience should not occur. 3) If the neuron iscaused to fire although the object is not present inthe environment, the experience should neverthe-less occur.

No studies performed to date have been able tomeet all of these criteria to establish a neural corre-late for a higher-order percept as complicated asgrandmother. However, the studies reported byNewsome and colleagues, discussed in Chapter 10,appear to meet these rudimentary criteria for estab-

lishing the existence of a bridge locus for a mid-level percept: an experience of coherent motionassociated with a particular part of the visual field.These studies have established that outputsrecorded from neurons in the motion-processing

extrastriate area V5, are both necessary and suffi-cient for producing a report, albeit by a monkey,that a percept of coherent motion in a particulardirection is present. Any theory regarding neuralcorrelates of conscious perceptual experiences, par-ticularly any theory that denies the existence of

bridge locus neurons, is going to have to accountfor these empirical results.

Blindsight as Evidence of a SpecialRole for V1

Blindsight is a neurological disorder in which indi-viduals with damage to V1 have implicit knowl-edge about certain explicit facts regarding what isout there even though they are not aware theypossess that knowledge, and in fact deny it whenasked. Some have used the clinical phenomenon of

blindsight as evidence to argue that V1 is a neces-sary anatomical structure for producing consciousawareness of visual stimuli in humans.

Blindsight was initially documented by ErnstPoppel and colleagues in 1973 and labeled with the

term blindsight by Lawrence Weiskrantz and col-leagues in a published case study of patient DB in1974. This patient began to experience headaches atthe age of 14. The headaches were so severe that DBwas often confined to bed and unable to carry outany activities except sleeping for periods up to 48hours. When DB was in his twenties, the frequencyof these episodes increased to about once every 3weeks, severely hampering his ability to live anormal life. Brain imaging revealed a malformationin the right occipital lobe. Surgery was performed

to remove the malformation, and the headachesstopped. The malformation was localized in thestriate cortex, and the surgery had the effect ofremoving the right striate cortex while leavingthe remainder of the occipital lobe intact. This is a

FIGURE 12.5. Summary of all the points on the lateral sur-faces of the two hemispheres of the brain that, upon stim-ulation, resulted in experiential responses. Reproduced

from W. Penfield and P. Perot, The brains record of audi-tory and visual experience. Brain 86:39696, 1963, bypermission of Oxford University Press.


9/16


very unusual occurrence. Typically, patients whoundergo this type of surgeries have damage tomultiple cortical areas.

Following surgery, patient DB was blind in theleft hemifield, as illustrated in Figure 12.6.

This diagram shows a standard visual field plot

obtained in a clinical setting. The patient fixates atthe center of a screen. Small spots of light are pre-sented, one at a time, at various locations on thescreen, and the patient reports whether each spot isseen. The plot on the left shows the results whileDB was viewing with the left eye, that on the rightthe results for the right eye. For spots of light pre-sented to the right of midline, DB had essentiallynormal fields in both eyes. However, DB was notable to see spots of light presented in most of theleft hemifield. The only exception was within some

small islands in the periphery of the upper hemi-field. This is accounted for by the fact that a smallportion of the striate cortex having a topographicprojection from this part of the field was spared bythe surgery.

The visual fields shown for patient DB are asexpected for a patient with extensive damage to theright occipital cortex. These results have beendemonstrated countless times in the neurology liter-ature, and if Weiskrantz and colleagues had notdecided to do additional testing, nothing remarkable

would have been noted when testing this patient.However, Weiskrantz and colleagues went on to

do some nontraditional assessments of patient DBusing psychophysical methods based on a combi-nation of Class A and Class B observations, as

defined in Chapter 2. In one example, DB sat infront of a screen on which X or O could beflashed in the left visual field. When asked to reportwhether he could see the flashed stimuli, DBreported no, as expected. However, when forcedto choose between the two alternatives in a forced-

choice task, DB was correct almost all of the time.In other words, when asked to describe his percep-tions with Class B observations, DB appeared to be

blind, but Class A observations collected with aforced-choice procedure demonstrated that he wasable to use his eyes to extract considerable infor-mation from the environment. This paradoxicalobservation is the essence of blindsight.

Subject DB was tested for acuity by presenting astimulus that was either a grating or a homoge-neous field of the same mean luminance. DB was

simply asked to guess on each trial whether thestimulus was a grating. His acuity was about 15cy/deg (~20/40 Snellen), which is a factor of twoworse than normal (see Chapter 8), but nowherenear being blind. In fact, patient DB could qualifyfor a drivers license in most places based on thislevel of acuity.

A similar set of methods was used to demonstrateresidual How-perception in addition to an uncon-scious awareness of some aspects of What-percep-tion. Subject DB was asked to point towards the

location of a stimulus projected onto the blindfield. Once again, DB claimed that he could notperform this task, but when he was forced to guessand reach out with his arm, the results were asshown in Figure 12.7.

FIGURE 12.6. The visual fields for each eye of DB, apatient with blindsight, measured with standard perimetrymethods in the clinic. The dark portions of the field demar-cate the blind portions of the fields. The left panel showsthe visual field while DB was viewing through the left eye

and the right panel the field viewed with the right eye.Adapted from L. Weiskrantz, et al, Visual capacity in thehemianopic field following a restricted occipital ablation.Brain 97:709728, 1974, by permission of Oxford Uni-versity Press.


10/16


The reaches into the blind hemifield were notquite as accurate as those into the normal hemi-field but nevertheless remarkably accurate basedon the fact that DB claimed to be simply guessing.

An initial concern of the investigators of this

study was that DB was malingering. They con-ducted several control experiments and concluded,. . . D.B. throughout convinced us of his reliabil-ity. One control study took advantage of the factthat the surgery had spared a small island of visionin the periphery of the upper left visual field.Weiskrantz et al. comment: If the projected stimu-lus happened to fall on [a portion of his visual fieldthat fell within this island] he reported thispromptly. Another way in which DB convincedthe experimenters that he was not malingering was

his reaction when he was told about or shown hisown results: [DB] expressed surprise and insistedseveral times that he thought he was just guessing.When he was shown a video film of his [perfor-mance] he was openly astonished. . . . There are

now published reports of several additionalpatients with striate cortex damage who exhibit

blindsight.Attempts have recently been made to establish

that the syndrome of blindsight also occurs in

monkeys in which the striate cortex has beenremoved surgically on one side of the brain. Oneparadigm that has been used to search for aphenomenon analogous to human blindsight inmonkeys is illustrated in Figure 12.8.

This figure depicts the test situations for amonkey with the left striate cortex removed, result-ing in a blind right hemifield and a normal lefthemifield. Two paradigms have been used to testvision in this monkey. In paradigm 1, shown in theleft panel, the monkey is trained to fixate at location

F. Then a single target is presented at one of fourlocations on the test screen and the monkey trainedto reach out and touch the target. This is a forced-choice task. Monkeys in which the striate cortex has

been removed on one side of the brain perform the

FIGURE 12.7. Accuracy of reaching to smallspots of light in the blind (left) andnormal (right) hemifields. Adapted from L.Weiskrantz, et al, Visual capacity in the hemi-anopic field following a restricted occipitalablation. Brain 97:709728, 1974, by per-

mission of Oxford University Press.


11/16


same on this task as normal monkeys and respondappropriately to stimuli in both the blind and thenormal hemifield.

In paradigm 2, shown in the right panel, themonkey is again trained to fixate at F, and whileit is fixating, a target is presented at one of fivefixed locations within the normal field on sometrials but not on other trials. The monkeys task isto reach for the target on trials in which one is pre-sented but to reach towards the location of a box inthe upper part of the normal hemifield on thosein which no target is presented. In other words,reaches to the box are used to respond to a blank

trial on which no stimulus was presented. When themonkey has learned this task, a single probe stim-ulus is presented in the blind hemifield duringfixation. When these probe trials are included fora normal monkey that has been trained on thistask, the monkey immediately reaches towards theprobe. However, monkeys with lesions to the striatecortex reach for the box, indicating a blank trial,whenever the target is presented to the blindhemifield.

These results are interpreted as support for the

hypothesis that monkeys lacking the striate cortex,like humans with blindsight, have no consciousawareness of a stimulus presented in the blindfield. When forced to choose, the monkeys respondcorrectly, but when allowed to respond with anoption designating whether or not a stimulus ispresent, the monkeys report that nothing is seen.

The phenomenon of blindsight has led to con-siderable speculation about the role of the striatecortex in shaping the properties of percepts. Somehave interpreted blindsight as evidence that only

neural signals passing through the striate cortex canproduce percepts that convey conscious awareness.The deficits in blindsight patients reflect either theabsence of processing in the striate cortex or lack ofprocessing in higher-order visual processing areas

that depend on input from the striate cortex. Theresidual visual capacities of patients with blind-sight must reflect neural processing in portions ofthe brain that still function following removal ofthe striate cortex.

Two neural pathways have been proposed as theones providing the residual capacities of patientswith blindsight. The first involves projections fromthe eyes that pass through midbrain structuressuch as the superior colliculus. In species with moreprimitive brains, such as frogs, these other struc-tures are the main visual processing areas of the

brain. It is only in mammals, and particularly in pri-

mates, that the massive geniculostriate projectionhas evolved. Observers in which no neural pro-cessing takes place in the striate cortex, be they pri-mates with damage to the striate cortex or speciesin which a striate cortex never evolved, may havecertain similarities in the properties of their per-cepts. Consider the percept formed when a fly

buzzes in front of a frogs eye. The visual system ofthe frog provides it with information about theenvironment that allows its tongue to shoot fromits mouth and obtain lunch. However, lacking a

geniculostriate system, the frog may not have anyconscious perceptual experience that a fly waspresent.

A second hypothesis about the neural processingareas of the brain that are responsible for residualcapacities in blindsight involves extrastriate corticalareas that receive input from the retina that is notderived from V1. Recall from Chapter 5 that mostvisual processing in primates passes through V1,which acts as a bottleneck, before fanning outto extrastriate cortical areas. However, there are

exceptions to this general principle. Some neuralsignals that pass through brainstem structuresare relayed through the pulvinar to reach severalextrastriate areas, including V5. There is also asmall anatomical projection from the dLGN that

FIGURE 12.8. Experimental paradigms for testing blindsight in monkeys. See text. Adapted from A. Cowey and P. Stoerig,Blindsight in monkeys. Nature 373:247249, 1995, by permission of Macmillan Magazines, Ltd.


12/16


bypasses the striate cortex and projects directlyto the extrastriate cortex in primates, includinghumans.

The two hypotheses about what anatomical areasare responsible for the residual capacities seen in

blindsight have somewhat different implicationsregarding the role of the striate cortex in consciousawareness. If only midbrain areas are responsiblefor the residual capacities seen in blindsight, thenthe lack of conscious awareness could be the resultof either missing striate cortex or missing inputsto extrastriate areas. However, if extrastriate areasare responsible for the residual capabilities of

blindsight, that implies that neural processingin the striate cortex in particular is needed forconscious awareness of visual percepts. The impli-cation is that the extrastriate areas responsiblefor processing specific kinds of information aboutthe environment, such as motion, color, shape,and depth, that still function during blindsight toprovide residual visual capabilities are not suffi-cient to provide conscious awareness. These issuesare currently unresolved. Any theories of neuralcorrelates of conscious perceptual awareness willhave to take the phenomenon of blindsight intoaccount.

The Dynamic Core Hypothesis

Tononi and colleagues have argued for a neural cor-relate of conscious perceptual experience called adynamic core, which they characterize in terms of

brain activity rather than as an anatomical location.They point out that conscious perceptual experi-ences have the property of being highly integratedand unified at any moment but also highly differenti-ated over time.

Stating that a momentary conscious state is

highly integrated and unified means that it cannotbe broken down into components. Consider, forexample, the perceptual experience that resultsfrom viewing ambiguous figures, such as theNecker cube in Figure 8.9. While viewing theNecker cube stimulus, normal human observerscan experience two incongruous percepts sequen-tially over time but are unable to experience the twopercepts simultaneously. Another example illus-trating the fact that a momentary conscious stateis highly integrated and unified is the fact that

humans cannot make more than one consciousdecision within a short interval of a few hundredmilliseconds.

Stating that conscious states are highly differen-tiated over time emphasizes that when one partic-

ular state rather than another occurs at a givenpoint in time, this fact is informative to an observerin the sense described by Shannon, as discussed inChapter 6. This is one criterion that can be usedto differentiate conscious perceptual experiencesoccurring in a human brain from states existingin simpler kinds of hardware. Consider why thedifferentiation between light and dark made by ahuman is associated with conscious experience,while that made by a photodiode is not. One criti-cal difference may be the amount of informationgenerated in the two cases. The discrimination

by the photodiode of dark from light conveys aminimal amount of information, in the sense ofthe degree of reduction of uncertainty. In a humanobserver, the experiences of lightness or darknessare highly informative since they reduce uncer-tainty regarding an enormous repertoire of possibleper-ceptual experiences.

Tononi and colleagues have argued that aneural process responsible for conscious experienceshould exhibit these same two properties: Theneural process should be both highly integratedand capable of exceptionally informative differenti-ation. They have argued that this could be achievedin the form of a unified neural process whosedynamic operation is distributed over variousportions of the brain rather than in a single place.

This neural process is what they refer to as thedynamic core.

A dynamic core comes into existence rapidly,within no more than a few hundred milliseconds,

by integrating, or binding, synchronous correlatedresponses from distributed neuronal groups.

For example, suppose someone is experiencing aconscious percept of a freshly baked chocolate chipcookie as he picks it up and brings it towards hismouth. Somatosensory portions of his brain areresponding to the touch of the cookie. Olfactory

parts of his brain are responding to its smell. Withinthe visual system, signals regarding the cookiesshape and color are reverberating through theWhat-perception stream. Signals guiding the armmuscles that are bringing the cookie towards themouth are reverberating through the How-perception stream. Simultaneously, higher-order,top-down, attentional processes are riveted on thecookie! These neural signals will be distributedover large portions of the brain, but their dynamicactivities are all correlated with one another. Over

a period of no more than a few hundred milli-seconds, a dynamic core of neural activity evolves,

based on reentrant long-range extrinsic connectionsamong brain areas. This dynamic core serves to

bind all of the correlated signals from all over the


13/16


brain into a unified activity that can be described inwords as a percept of a cookie.

3 What Kinds of Observers

Can Produce ConsciousPerceptual Experiences?

In addition to information that can be used forWhat-perception and How-perception, perceptsconvey to humans perceptual experiences of whichthey are consciously aware. It is worth consideringwhether the perceptual systems of other classesof observers, such as robotic machines, might alsooutput conscious perceptual experiences. Some

argue that this issue can be decided based ongeneral principles. Others argue that it can bedecided empirically.

Arguments Based on General Principles

The arguments about what kinds of physicalsystems can support perceptual experiences basedon general principles cover an immense range. Atone extreme is the liberal position that, in princi-ple, any physical system can support perceptualexperiences. This position was argued by thephilosopher David Hartley, writing in the 18thcentury:

. . . Matter, if it could be embued with the most simpleKinds of Sensations, might also arrive at all that Intelli-gence of which the human Mind is possessed. . . .

A modern version of the liberal position, calledstrong artificial intelligence, was described inChapter 6. Strong AI makes the radical claim thatany physical system can generate conscious experi-

ences just by virtue of implementing the properalgorithm. It is not supposed to make any differencewhat physical substance is used to implement thealgorithm.

The opposite extreme conservative position isthat the causal power to generate full-fledged con-scious experiences belongs only to living human

brains. A historical example of this position wouldbe that of Descartes as described in Chapter 1. Thegeneral idea that humans are special and that thecapacity to have conscious perceptual experiences

does not extend beyond humans remains pervasiveto the current day among many philosophers andscientists who study perception.

An intermediate conservative position is thatconscious perception extends beyond humans, but

only to biological organisms with brains similar toour own. This intermediate position is groundedin the ancient philosophical position of vitalism,which asserts that the forces involved in living

bodies are different from those in the inorganicworld.

In a recent debate about these issues, the philoso-pher John Searle adopted a modern variationon the intermediate position, arguing that anyphysical system capable of causing consciousnesswill have to have causal powers equivalent tothose of biological brains. Philosophers PatriciaChurchland and Paul Churchland respondedwith a rejoinder, arguing that Searles argument

begs the question until we know what thoseequivalent causal powers are. Furthermore, theyargued, it seems highly unlikely that all causalpowers present in biological brains would benecessary in order to achieve consciousness, forexample, the causal power to produce a bad smellwhen rotting.

Arguments Based on Empirical Validation

Since there is currently no consensus about whetheror not, in principle, physical systems other thanhuman brains can generate conscious perceptual

experiences, others have argued for an empiricalapproach. Such an approach tries to validate claimsfor conscious perceptual experiences in an observer

based on some observable aspect of behavior. Ageneral premise of all of these strategies is thatphenomena in the form of verbal reports of adulthuman observers serve as the gold standard forwhat is meant by the term conscious perceptualexperiences. The performance of other observers isevaluated by measuring it in some way against thisgold standard reference.

Validation Based on Correlation with Actions

One strategy is to find a secondary behavioralresponse in the form of an action that appears to behighly correlated with reports of a particular per-ceptual experience in verbal humans. An examplewould be the stereotypical reaction to a loomingobject that was described in the section on mea-suring How-perception in Chapter 2. In adulthumans, whenever this stereotypical action is

elicited, there is a corresponding experience of alooming object. Thus, it has been argued that ifa nonverbal observer exhibits the same action, itcan be inferred that the observer is experiencing alooming object.


14/16

What Kinds of Observers Can Produce Conscious Perceptual Experiences? 375

Willingness to accept results obtained by thisstrategy as evidence for perceptual experiencesis heavily influenced by the characteristics of theobserver. For example, many who accept such evi-dence as demonstrating a perceptual experience oflooming when applied to human infants showstrong resistance to accepting the same argumentfor a robot programmed to make similar stereo-typical movements to a looming object.

Validation Based on a Turing Test

A potential strategy that might be helpful for devis-ing an operational definition of perceptual experi-ences in machines comes from a solution proposedto solve a similar problem addressed in the artifi-cial intelligence literature:

How would one decide whether a machine is intelligent inthe same sense in which the term intelligence is applied tohumans?

Alan Turing, a pioneering scientist working in thefield of artificial intelligence, devised an operationaldefinition of intelligence, now commonly called theTuring test, to address this question. The issuesinvolved in evaluations of intelligence and of per-ceptual experience are similar enough to warrant alook at Turings test.

Turing had an important insight. Any judgetrying to decide whether an agent is intelligent hasaccess to two kinds of information: informationabout the agent being evaluated and informationabout functions the agent performs. It is difficultto disentangle these two sources of information.For example, many human judges will be biasedagainst the notion that a machine can be intelligent.Consequently, they will be willing to accept evi-dence obtained from a human agent as demon-strating intelligence but will reject the exact same

evidence when informed it came from a machine.There is a similar bias against attributing consciousperceptual experiences to a machine.

Turing tried to solve this problem with respect toquestions about intelligence by keeping the judgemasked as to the identity of the agent. The judgeholds a conversation with two agents, one humanand the other machine. Then the judge must decidewhich agent is the human. Turing argued that if the

judge attributes intelligence to the human andcould not distinguish between these two agents,

then the judge should also be compelled to attributeintelligence to the machine.

In fact, Turing even biased his test in such a waythat if the two agents were really of equal intelli-gence, the judge would more likely pick the human.

He did this by having both agents claim to behumans. Thus, the human agent being interrogated

just has to be truthful. However, the machine has topull off an act of deception that makes the judge

believe it is a human.A similar rationale can be used to propose the fol-

lowing Turing perception test as an operationaldefinition for perceptual experience. A judge isallowed to interrogate and/or watch the behaviorof two observers, only one of which is a human,while each claims to be having a particular type ofperceptual experience. The judges task is to iden-tify the human observer. The test must be arrangedin such a way that the judge cannot tell by outwardappearances which observer is the human. Theexact details of how this is accomplished are notimportant as long as the judge has access to therelevant performance of the two observers. If the

judge cannot reliably pick out which observer is thehuman, then the judge will be forced to attribute tothe nonhuman observer the same level of percep-tual experience as to the human. In other words,if the judge accepts that the human has consciousperceptual experiences based on an evaluation ofthe performance, then the judge will also have toattribute conscious perceptual experiences to theother observer.

Only nonlanguage forms of this proposed Turing

perception test are applicable to human infants orto animals. However, there is nothing, in principle,that precludes a machine with sufficient language(symbolic communication) capabilities from passinga language version of this test as well at some datein the future.

Validation Based on Correlation withNeural Processing

Another approach that can be used when evaluat-

ing animals with brains similar to human brains isto define perceptual experiences in a reductionistmanner in terms of patterns of neuronal activitythat are correlated with conscious perceptual expe-rience in adult humans. This book includes severalexamples in which noninvasive brain imagingmethods allow one to monitor simultaneously theaction of relevant neuronal populations and behav-ior. For example, brain activity can be measuredwhile a human views a particular visual stimulusand simultaneously gives a verbal report describ-

ing the associated perceptual experience. Supposea particular part of the brain reacts in a characteris-tic manner if and only if a human reports a certainperceptual experience. Brain activity can then bemeasured in animals with similar brains, such as


15/16


monkeys, under the same viewing conditions tolook for this same pattern of neuronal activity.

4 An Overall Conceptual Schemeof Perception

As this text comes to a close, it is worthwhile to con-sider one last time the overall conceptual scheme ofhuman perception that has been adopted, as char-acterized in Figure 12.9.

Objects and events in the environment (distalstimuli) cause activation of receptors in the retina(proximal stimuli). Signals generated by these recep-tors carry information that is analyzed as it passesthrough a number of parallel streams of neural pro-cessing. Within each stream, processing is organized

into stages, although extensive feedback circuits (notshown) have the effect that information beingprocessed at most stages is heavily influenced bywhat is happening simultaneously at other stages.Processing passes through the M-, P-, and K-streamsand then through various extrastriate processingstreams, and finally fans out to the What- and How-perception streams. At the end of perceptual pro-cessing, percepts are formed that convey to theobserver knowledge about what is present in theimmediate surrounding environment, knowledge

about how to interact with this environment, andperceptual experiences related to the environmen-tal stimulation. In most cases the percepts associatedwith What-perception are veridical, by which ismeant that the observers perceptual knowledgeabout what is out there corresponds to empirical

information derived from instruments. In somecases, What-perception can be shown to be in error,or illusory. Similarly, motor responses guided byHow-perception are usually appropriate and helphumans accomplish the goals of finding food, avoid-ing danger, and so on. In some cases performance isdeficient; for example, an outfielder fails to catch afly ball. Perceptual errors seldom occur in the eco-logical visual environment in which our speciesevolved, presumably because the individuals withperceptual systems that led to frequent errors did notsurvive. However, when perceptual errors arestudied in artificial laboratory environments, theyprovide rich information about the forms ofprocessing that operate in transforming low-levelproperties of sensory stimulation into high-levelqualities of percepts.

Summary

Physical stimulation of the visual sense organ pro-duces signals that are processed along the genicu-lostriate pathway, then the extrastriate corticalstreams, and finally the What-perception and How-perception streams. As information travels alongthese stages of processing, it is transformed so thatinstead of representing primarily low-level physi-

cal properties of sensory stimulation, it representshigher-order qualities of percepts. This transforma-tion has been demonstrated in numerous studiesthat compare properties of the neural activity atvarious stages of processing with properties of thesensory stimulation and qualities of the percepts.

FIGURE 12.9. Overview of the conceptualscheme for perception adopted in this book.


16/16

Selected Reading List 377

Studies at relatively early stages of processing,such as those that take place in the dLGN, havefound evidence for representations of relativelylow-level stimulus properties, such as the intensityat each point in the image. Studies at later stages, incortical areas, have found evidence for neural rep-resentations that relate more closely to perceptualqualities, including illusions and bistable percepts,and reflect the influences of higher-order processes,such as attention and memory, on percepts.

Bridge locus theories about the neural correlatesof conscious perceptual experiences assert that aconscious experience is caused at a specific anatom-ical locus. Bridge locus theories can be weak, assert-ing that large regions of the brain are involved, orstrong, asserting in the limiting case that a singleneuron is responsible for a single perceptual expe-rience. Accumulating evidence is consistent with

both weak and strong forms of bridge locus theo-ries. Theories that argue against the idea of a bridgelocus will have to find other ways of accounting forthese empirical findings.

An approach to neural correlates of consciousperceptual experiences that is similar in some waysto bridge locus theories but does not propose aspecific anatomical location is the dynamic corehypothesis, which asserts that the neural correlatetakes the form of distributed dynamic neural activ-

ity. This neural dynamic core binds synchronousneural responses from widely distributed regions ofthe brain to form a conscious percept.

It is an unresolved question whether observersother than humans can have conscious perceptualexperiences. Many scientists are willing to acceptthat animals with humanlike brains, such asmonkeys, have conscious perceptual experiencessimilar to those of humans. However, much skepti-cism remains about attributing conscious percep-tual experiences to other types of observers, such as

robotic machines. Some argue that issues aboutwhat kinds of observers can have conscious per-ceptual experiences should be decided based ongeneral principles. Others argue that these issuescan be validated empirically.

Selected Reading List

Boring, E.G. 1933. Dimensions of Consciousness. New York:Appleton-Century Crofts, Inc.

Bourassa, C.M. 1986. Models for sensation and perception:A selective history.Human Neurobiol. 5:2336.

Churchland, P.M., and Churchland, P.S. 1990. Could amachine think? Sci. Am. 262:3237.

Cowey, A., and Stoerig, P. 1991. The neurobiology of blind-sight. Trends Neurosci. 4:140145.

Culham, J.C., Brandt, S.A., Cavanagh, P., Kanwisher, N.G.,Dale, A.M., and Tootell, R.B.H. 1998. Cortical fMRI acti-

vation produced by attentive tracking of movingtargets.J. Neurophysiol. 80:26572670.

Desimone, R., and Duncan, J. 1995. Neural mechanismsof selective visual attention. Annu. Rev. Neurosci.18:193222.

Graziano, M.S.A., Hu, X.T., and Gross, C.G. 1997.Coding the locations of objects in the dark. Science 277:239241.

Heinze, H.J., Mangun, G.R., Burchert, W., Hinrichs, H.,Sholz, M., Munte, T.F., Gos, A., Scherg, M., Johannes, S.,Hundeshagen, H., Gazzaniga, M.S., and Hillyard, S.A.1994. Combined spatial and temporal imaging of brain

activity during selective attention in humans. Nature372:543546.Kosslyn, S.M., Thompson, W.L., Kim, I.J., and Alpert,

N.M. 1995. Topographical representations of mentalimages in primary visual cortex. Nature 378:496498.

Logothetis, N.K., and Schall, J.D. 1989. Neuronalcorrelates of subjective visual perception. Science245:761763.

Maunsell, J.H.R. 1995. The brains visual world: Repre-sentation of visual targets in cerebral cortex. Science270:764769.

Moore, T., Rodman, H.R., and Gross, C.G. 1998.Man, monkey, and blindsight. The Neuroscientist 4:227230.

Moran, J., and Desimone, R. 1985. Selective attention gatesvisual processing in the extrastriate cortex. Science229:782784.

Motter, B.C. 1993. Focal attention produces spatially selec-tive processing in visual cortical areas V1, V2, and V4in the presence of competing stimuli. J. Neurophysiol.70:909919.

Penfield, W., and Perot, P. 1963. The brains record ofauditory and visual experience. Brain 86:595696.

Poppel, E., Held, R., and Frost, D. 1973. Residual visual

function after brain wounds involving the central visualpathways in man. Nature 234:295296.

Tononi, G., and Edelman, G.M. 1998. Consciousness andcomplexity. Science 282:18461851.

van Voorhis, S.T., and Hillyard, S.A. 1977. Visual evokedpotentials and selective attention to points in space.Percept. Psychophys. 22:5462.

von der Heydt, R., and Peterhans, E. 1989. Mechanismsof contour perception in monkey visual cortex: I.Lines of pattern discontinuity. J. Neurosci. 9:17311748.

Weiskrantz, L., Warrington, E., Sanders, M., and Marshall,

J. 1974. Visual capacity in the hemianopic field follow-ing a restricted occipital ablation. Brain 97:709728.

12 higher-order and subjective aspects of perception

Documents