Early Experience and Visual Information Processing in Perceptual and Reading Disorders: Proceedings of a Conference Held October 27-30, 1968, at Lake Mohonk, New York, in Association With the Committee on Brain Sciences, Division of Medical Sciences, National Research Council. Edited (1970)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

FRANCIS A. YOUNG Development of Optical Characteristics for Seeing Although it is not yet possible to describe completely the development of the optical characteristics of the human eye, it is possible, by relating results of published and unpublished studies with some amount of con- jecture, to put together a likely description. Much of the available information concerning the effects of heredity and environment on the development of the optical characteristics of the eye is based on studies of subhuman primates, primarily chimpanzees and monkeys. However, a recent study of the Eskimo population at Barrow, Alaska, which supports the results of studies of subhuman pri- mates, suggests strongly that the early visual environment and early visual experience play an important role in developing and modifying the opti- cal characteristics of the eye, that the reaction of the eye to its visual environment plays a determining role in the development of the optical characteristics necessary for seeing and reading, that the mechanism of this role should be investigated in all children as they approach the read- ing age, and that the mechanism cannot be effectively assessed by deter- mining the Snellen visual acuity at 20 ft or at 20 in. REFRACTIVE CHARACTERISTICS When a person has 20/20 Snellen acuity, it may be said that he is able to resolve letters that subtend 1 min of visual angle; but little can be said 35

FRANCIS A. YOUNG about how he accomplishes this resolution. Hirsch6 has shown that, if one correlates the refractive error determined by retinoscopy with the Snellen visual acuity, one can demonstrate a considerable correlation (+0.95) in persons who show some degree of myopia. I found a correla- tion of+0.84 between the same measures.27 Little or no correlation was found in persons who show emmetropia or some degree of hyperopia, provided the hyperopic persons are not old enough to be also afflicted with presbyopia. For practical purposes, the primary value of the Snellen acuity at either far or near distance would be to demonstrate whether a person has some type of visual refractive error large enough to prevent compensating for it through the use of accommodation, head tilt, squint- ing, or other techniques. The refractive characteristics of the eye may be readily determined by retinoscopy, either with or without a cycloplegic drug, such as atropine or Cyclogyl (cyclopentolate). This technique provides an adequate de- scription of the gross refractive characteristics of the eye and indicates whether the person has any degree of astigmatism in conjunction with a measurable degree of myopia or hyperopia. This approach, which re- quires some clinical skill, permits the clinician to determine which lens or combination of lenses will bring about the neutralization of his ret- inoscopic shadow in a given eye under a given set of conditions. If the eye is under complete cycloplegia (which may be produced by adminis- tration three times daily of 1% atropine for 4 or 5 days), the clinician may state the minimum refractive ability of the subject's eye. Because he is examining under a state of maximally relaxed accommodation, not an ordinary state, he cannot accurately predict what will happen when the eye is functioning without the effect of the cycloplegic. If he per- forms retinoscopy without the use of a cycloplegic drug, but attempts to induce relaxed accommodation through the use of a plus lens, he can describe more accurately what the subject can accomplish visually under more nearly normal conditions. In that case, the clinician may be de- ceived effectively by the subject, and conclude that he has achieved a rather basic measure of refractive characteristics, whereas his results may actually be affected by a considerable degree of accommodation within the eye; he is not on much safer ground if he depends on a subjective refraction, calling for the patient's cooperation when he places lens com- binations in front of the patient's eye until the level of visual perfor- mance is satisfactory. In any determination of the refractive characteristics of the eye, one 36

Development of Optical Characteristics for Seeing can discuss the findings in terms of the types of lenses required to neu- tralize the movement of the retinoscope shadow or to achieve maximal acuity. If the person requires a minus, diverging lens to neutralize the movement or to achieve maximal acuity, he may be said to have myopia. If he requires a plus, converging lens to accomplish the same ends, he may be said to have hyperopia. If he requires no lens, he may be said to have emmetropia. If he requires a plus or minus cylinder to correct astig- matism, he may be said to have hyperopic or myopic astigmatism. The rest of this discussion will deal with the lens characteristics required to neutralize the movement of the retinoscope shadow or to achieve the best subjective acuity. An eye that requires a minus lens will be a "myopic eye"; no lens, an "emmetropic eye"; and a plus lens, a "hyperopic eye." The distinction between myopia and nonmyopia is arbitrarily taken as the use of a minus lens to neutralize the movement of the retinoscope shadow or to achieve best visual acuity subjectively. OPTICAL CHARACTERISTICS The determination of the refractive characteristics of the eye, although a great improvement over the determination of simple visual acuity, does not adequately describe the optical characteristics of the eye. The refrac- tive characteristics depend on the eye taken as a whole. The optical characteristics—the characteristics that contribute to the total refractive capacity of the eye—may change greatly with respect to one another without producing or affecting a refractive error. Thus, an emmetropic person may show a considerable change in axial length, which is com- pensated by a change in corneal curvature, and still have no refractive error. Retinoscopy or subjective refraction is insensitive to changes in the optical components, unless the changes themselves result in an im- balance of the optical components. Consequently, retinoscopy must be supplemented with some techniques that will provide a more accurate measure of the optical characteristics of the eye. Only by combining such techniques with the measurement of refractive error is it possible to describe what is taking place over time and then to determine the re- lative importance of changes that occur with growth and with the devel- opment of visual ability. The three techniques most commonly used to supplement retinoscopy and other measures of refractive characteristics are the determination of 37

FRANCIS A. YOUNG the corneal curvature by means of keratometry; the measurement of distances within the eye, such as the depth of the anterior chamber, the thickness of the lens, the depth of the vitreous body, and the overall axial length of the eye (by ultrasonography); and the determination of the curvature of the front and rear surfaces of the lens by ophthalmo- phacometry. Keratometry permits a highly accurate measurement of the surface curvature of a ball bearing but a somewhat less accurate measurement of the curvature of the cornea, which is not completely uniform but has more than one radius of curvature. However, if the same instrument is used consistently on the same eye, it is possible to develop reasonably accurate measures of changes within the cornea. Ophthalmophacometry is based on the demonstration of Purkinje-Sanson images, or the reflec- tion of light from the surfaces of the cornea and lens. Essentially, it in- volves directing two points of light into the eye and measuring the sepa- ration of their reflections from the back surface of the cornea, the front surface of the lens, and the back surface of the lens (the second, third, and fourth Purkinje-Sanson images). The combination of measures of refractive error, corneal curvature, depth of anterior chamber, front lens surface curvature, thickness of lens, rear lens surface curvature, depth of vitreous, and total axial length provides an accurate description of the optical characteristics of the eye; anything less than this combination does not. Although we have added considerably to our knowledge of the basic characteristics of the eye, we are still unable to describe exactly what will happen when the eye operates normally, because some of these measurements must be made while the eye is under cycloplegia. Without such a description and ac- curate measurements from birth onward in the same persons, it is not possible to characterize completely the development of the optical char- acteristics of the eye. Fortunately, techniques are being developed that will permit the description of ocular performance under dynamic condi- tions, and it may be hoped that a complete description of the eye, in- cluding its static and dynamic characteristics, may be developed within the next decade. Size There is no direct evidence dealing with the size of the eye in a living human at birth or during the first 3 years of life. The reasons for the 38

Development of Optical Characteristics for Seeing lack of information are related to the difficulty of applying the methods outlined to neonates and very young children. Consequently, most of the information available concerning the overall diameters of the eye has been obtained by measuring eyes post mortem. However, as soon as blood pressure drops, the intraocular pressure drops and the eye becomes quite flaccid; furthermore, most in vitro measurements, even when the eye is perfused to reinstate a probably normal intraocular pressure, can vary widely from the measurements that would have been obtained in vivo. Sorsby and Sheridan17 have provided possibly the best measurements available on the sagittal or anterior-posterior axial length of the eye of the newborn and of children 1-6 days old. There is no significant dif- ference between these two groups, and the mean sagittal diameter is ap- proximately 17.8 mm for both boys and girls. The sagittal diameters are smaller in premature infants, and follow closely the body weight at birth. In the full-term baby of some 3.4 kg, the sagittal diameter is about 17.5- 18.5 mm. Therefore, the sagittal diameter increases by around 5-7 mm during growth, if the adult value is taken to be 23-25 mm. It is likely that the usual increase is about 6 mm. There is apparently little growth during the first 2 weeks of life. Growth The most complete information available on the growth of the eye in the living human has been provided by Sorsby and co-workers4'14-18 in a series of studies; they used refraction techniques, photographic ophthalmophacometry, x-ray, and ultrasonic measurements to study the growth and development of the human eye. A series of investigations by van Alphen21 supplied some of the missing links in our understanding of factors that contribute to changes in ocular size. When the results of these studies are combined with those obtained by me and my co- workers, 23-44 on humans and other primates, a more complete descrip- tion of the growth characteristics of the optical components is possible. The sclera, choroid, and retina of the eye of a primate are closely ad- herent layers with various degrees of elasticity that enclose more or less viscid liquids to form a nearly spherical globe. This globe continues to increase in size after birth. According to van Alphen,21 the human eye at birth is three fourths of its adult size, and all the ocular structures are probably still growing at birth. The adult size of the cornea is reached 39

FRANCIS A. YOUNG between the first and second years, at which time the eye has not yet attained its adult size. Whether the sclera continues to grow after the cornea reaches its adult size is unknown, but intraocular pressure must be important in stretching the sclera. The adult size of the eye is related to the genetic growth component, the elasticity of the sclera, the intraocular pressure, and a number of other variables yet to be described. In cases of congenital glaucoma, large eyes with large, flat corneas develop as a result of high intraocular pressure and scleral elasticity. In cases of experimentally induced low intraocular pressure, the eye remains small (microphthalmia). But even in these extremes, as well as in most cases between them, the eye remains nearly spherical. Shape In the hyperopic eye, a comparison of the transverse, vertical, and axial dimensions based on radiographic measurements of 11 male adult eyes by Deller et a/.4 showed no differences between the transverse and vertical dimensions, but did show an axial length significantly longer than either at the 1% level of confidence. In the myopic eye, also, we find that the only significant deviation from sphericity occurs in the axial diameter. Furthermore, if the same significance level is used, there is no difference in diameter between the hyperopic and emmetropic eyes, and the myopic eye is significantly longer than the emmetropic eye in the axial diameter and significantly longer in every diameter than the hyperopic eye. These comparisons suggest that the eye is normally a sphere and that the shape is determined by the variables of genetics, scleral elasticity, and intraocular pressure. A developing eye that is characterized by a low intraocular pressure and a relatively high scleral rigidity may remain small in diameter even if the genetic component is directed toward greater size. Although one cannot evaluate the genetic component directly, it is possible to measure scleral rigidity and intraocular pressure independently and to estimate the contribution of the genetic compo- nent from those measurements. Sorsby et al.15 conclude that the scanty data in the literature on the dimensions of the eye at birth and in childhood suggest that dimensions almost equal to those of the adult are reached by the age of 2 years. Moreover, it is likely that the cornea has reached its adult size by the 40

Development of Optical Characteristics for Seeing end of the first year. Inasmuch as the globe is about 18 mm long at birth and some 5 mm longer by the age of 3 years and there is no drastic change in the refraction of the eye in the first 3 years, compensatory reduction in the powers of the cornea and the lens by as much as 20 diopters must occur during that period. The development of the myopic eye, with its exaggerated axial length, probably depends on the operation of other variables, although the three mentioned—genetics, scleral elasticity, and intraocular pressure—may also play a contributory role. Sorsby et al.1S also found a second growth period, the juvenile phase, during which the eye grows at a lower but measurable rate to reach its maximum growth at about 13-14 years of age. The hyperopic eye apparently never experiences this later growth, but remains arrested at the infantile phase. The emmetropic eye should also be included in this category, even though it is "compensated," be- cause it does not undergo the later growth changes. These growth changes are based on changes in axial length, which can be measured with phacometry and ultrasonography. It is not possible to say whether the total size of the eye or only the axial length is increased during the later growth period. Comparisons based on the x-ray measure- ment of adult eyes suggest strongly that only the anterior-posterior axial length or sagittal diameter increases during this later growth period; it thus causes a change in shape, as well as size, of the globe, which is probably not determined by genetic aspects of the eye itself. The fact that the dimensions of the eye remain relatively stable between the ages of 3 and 11 or 12 supports the concept that new factors contribute to the growth of the eye in the juvenile phase. MYOPIA AND NEAR-WORK One of the earliest suggestions as to the nature of the additional variables that may influence eye growth dates back to the early Chinese, who in- vented lenses and found that the minus lens seemed to assist scholars to see more clearly. In 1813, James Ware22 presented a paper to the Royal Society of London in which he described his investigation on nearsighted- ness. He found, for example, that among 10,000 footguards in the British Service not even a half-dozen men were known to be nearsighted. He pursued his inquiry at a military school at Chelsea where there were 1,300 boys; he found that the complaint of nearsightedness had never 41

FRANCIS A. YOUNG been made among them until he mentioned it, and even then only three experienced any inconvenience from it. He then inquired at several col- leges in Oxford and Cambridge and, although there was great diversity in the number of students who used glasses in the various colleges, glasses were used by a considerable portion of the total number of stu- dents in both universities. In one college in Oxford, he accumulated a list of names of no fewer than 32 of 127 students who used either a hand glass or spectacles, between 1803 and 1807. Ware described the effects of fitting concave or minus lenses to nearsighted persons as follows: It should be remembered, that, for common purposes, every near sighted eye can see with nearly equal accuracy through two glasses, one of which is one number deeper than the other; and though the sight be in a slight degree more assisted by the deepest of these than by the other, yet on its being first used, the deepest num- ber always occasions an uneasy sensation, as if the eye was strained. If, therefore, the glass that is most concave be at first employed, the eye, in a little time, will be accommodated to it, and then a glass one number deeper may be used with similar advantage to the sight; and if the wish for enjoying the most perfect vision be in- dulged, this glass may soon be changed for one that is a number still deeper, and so in succession, until at length it will be difficult to obtain a glass sufficiently concave to afford the assistance that the eye requires. In an appendix to Ware's paper, Sir Charles Blagden2 gave the follow- ing comments. Mr. Ware states in his Paper, that near sightedness comes on most frequently at an early age; that it is more common in the higher than in the lower ranks of life; and that particularly at the universities, and various colleges, a large proportion of the students make use of concave glasses. All this is exactly true, and to be accounted for by one single circumstance; namely, the habit of looking at near objects. Chil- dren born with eyes which are capable of adjusting themselves to the most distant objects, gradually lose that power soon after they begin to read and write; those who are most addicted to study become near sighted more rapidly; and, if no means are used to counteract the habit, their eyes at length lose irrecoverably the faculty of being brought to the adjustment for parallel rays. The statements appear to be as valid as they were in 1813, and this concept, that the use of the eyes for near-work is responsible for the development of myopia, has a long history in ophthalmology and op- tometry, but there is inadequate evidence for supporting or rejecting it. Myopia usually develops between 10 and 14 years of age and usually tends to increase with time but to stabilize around 18 years of age. How- 42

Development of Optical Characteristics for Seeing ever, there appear to be some persons who do not develop myopia until after age 18; they tend to stabilize around the age of 24. Approximately 8% of U.S. grade school children are myopic, 10-15% of junior high school children, 15-25% of high school children, 25-50% of college stu- dents, and 40-60% of graduate students. In a study made at Washington State University, 44% of 400 college freshmen were myopic.29 By the time this group reached the junior year, 56% had dropped out of the university, and the proportion who were myopic had increased to 50%. Among 148 men in the honors college, 57% were myopic, and among 226 women in the same college, 60% were myopic. Education versus Intelligence If this consistent finding—that the proportion of myopic persons in- creases with years of schooling—is accepted and combined with the known relationship between intelligence level and years of schooling (which parallels that for the development of myopia), it might be con- cluded that the myopic person who predominates at the higher educa- tional levels is also more intelligent than the nonmyopic. It should even be possible to estimate intelligence by determining the refractive char- acteristics of the eye. Most of the studies that have attempted to dem- onstrate a relationship between intelligence and refractive error have found none, except when intelligence was measured by written tests.7'30 There is a positive relationship between performance on such a test and refractive error: myopic persons tend to score higher than nonmyopic. However, when reading ability is statistically adjusted for, the correla- tion of refractive error and intelligence approximates zero. The myopic person is a substantially better reader than the nonmyopic. Myopia and Personality Studies of the personality characteristics of myopic and nonmyopic per- sons indicate that there are consistent personality characteristics asso- ciated with myopia.12'20'29 In general, the myopic person tends to be introverted, and the nonmyopic, extroverted. Several investigators have found that myopic persons on the average make significantly better grades in college than emmetropic or hyperopic students, tend to be more introverted in thinking and in social behavior than emmetropes, and are more emotionally inhibited and less inclined to motor activity than nonmyopes.20'29 It may be asked whether the myopic personality 43

FRANCIS A. YOUNG characteristics are present before the myopia develops and lead to the development of myopia, or whether the myopia causes the development of the personality characteristics. Apparently, the personality character- istics precede the development of myopia, inasmuch as the characteristics may be distinguished as early as the kindergarten and first-grade years, whereas the myopia usually does not develop until the fifth- and sixth- grade years. In spite of the consistent pattern that has been described, it is not pos- sible to conclude without more definite investigations that reading leads to myopia. It is clear that most children learn to read in this culture at approximately 6-7 years of age, whereas myopia ordinarily does not develop before the age of 11 or 12. Similarly, the development of sec- ondary sexual characteristics does not begin until 12-14 years of age. Many have argued that myopia is a delayed hereditary phenomenon that does not develop until puberty. Sorsby et al.15 specifically considered the possible relationship of height, weight, general growth rate, such traits as color of iris, hair, and skin, and puberty to myopia. They con- cluded that no correlation could be found. There was nothing to suggest a spurt of ocular growth at puberty, nor did variations in the age of on- set of menstruation influence ocular growth or ocular coordination. Others had reported similar findings—that there is no relationship be- tween physical characteristics and the development of myopia or be- tween nutritional characteristics and myopia (provided the nutrition is adequate).24 Myopia and Heredity Although there is general agreement that myopia is related to hereditary or environmental factors, there is no agreement as to their relative con- tributions. The disagreement is due to the inability of investigators to evaluate these contributions experimentally. Only the experimental ap- proach provides control over variables, which is essential to such an evaluation. An experimental design that permits control over all variables would be desirable, but one that permits control over either the hereditary or the environmental factor is essential. The use of identical twins in the typical co-twin control study holds hereditary factors constant, and com- plete control of the environment would permit holding environmental variables constant. 44

Development of Optical Characteristics for Seeing With human subjects, it is easier to hold heredity constant through the use of identical twins than it is to control environmental factors. The effects of near-work on the development of myopia could be tested by having one twin engage in little or no near-work while the other twin did a great deal of near-work. This situation could be replicated over a num- ber of pairs of twins, and the amount of near-work and the age at which it is done could be obtained. Unfortunately, this study has not been made, and it is not likely to be made, because the obstacles faced in ob- taining enough twins and exercising the necessary degree of control over their behavior are virtually insurmountable. The identical-twin approach is not the only adequate experimental test of the near-work hypothesis. The converse of the co-twin study may be used. Subjects of different hereditary constitutions may be exposed to the same near-work conditions. If all subjects develop the same amount of myopia at the same rate, heredity can be assumed to play no role in the development of myopia. If none of the subjects develops myopia at a rate different from that shown by a control group not exposed to near- work situations, environment can be assumed to play no role. Finally, if the experimental subjects do not develop myopia at the same rate, but the rate is significantly greater in the near-work than in the control situa- tion, it can be assumed that there is an interaction between heredity and environment. The problems faced in pursuing this type of experimental approach with human subjects are comparable with those confronted in the co-twin control approach. The availability of animals whose visual characteristics are similar to man's makes the second type of approach feasible because of the degree of control that can be exercised over animals. EXPERIMENTAL FINDINGS A series of studies has demonstrated that subhuman primates, particu- larly monkeys and chimpanzees, develop myopia under experimental conditions that restrict visual space to a distance of less than 20 in. from the eye.2s,26,as-as About three fourths of all adult monkeys placed in the restrictive space show an increase in minus refractive error, and more than half the animals show approximately one-half diopter of myopia or more within 3 months after being placed in it. If young animals (1 or 2 years old) are placed in this situation, it requires a longer period—4-5 45

FRANCIS A. YOUNG months—before any myopic changes are shown, but once they begin, they proceed much more rapidly than in the adult animals. If adolescent animals (2!/2-4 years old) are placed in this situation, the onset of myopia occurs in 2 or 3 months, and the total degree of myopia developed is greater than in the adult animals and less than in the younger animals. This suggests the possibility that the younger animals—and, by analogy, the younger humans—are able to withstand the stresses and strains of the near-work situation for a longer period than the older animals, but that, once they start to respond, they are capable of making a greater response than the older animals. In all groups of animals, approximately 65-75% show myopic changes under these experimental conditions. When a group of newly captured rhesus monkeys were examined, only 12 of 600 eyes were found to have any myopia.34 Among wild and laboratory monkeys, the mean and median refractive errors were found to be significantly more hyperopic in the newly captured rhesus mon- keys, the younger monkeys, and wild monkeys in general. More myopia was found among pig-tailed monkeys, older monkeys, and laboratory monkeys. When the monkey population is separated into wild and labora- tory animals and these are compared with the Pullman, Washington, population of human subjects (characterized by a high proportion of readers) and the nonreading Washington, D.C., subjects studied by Kempf et al.,9 there is good agreement between the laboratory animals and the reading human population and between the wild animals and the non- reading human population. This suggests a similarity between the effect of the laboratory environment on the refractive characteristics of the monkey eye and the effect of the undefined characteristic that causes an intellectually oriented population to have a higher incidence of myopia than a nonintellectually oriented population. Steiger19 and some of his followers have suggested that myopic per- sons gravitate to professions in which myopia is an aid, rather than a hindrance, and consequently end up in an intellectual profession. Thus, it may be that myopic monkeys are particularly well suited for labora- tory work and that they choose this profession. But if that is the case, why are there so few myopes among the wild monkeys destined to be- come laboratory animals? The relationship between wild and laboratory monkeys has been in- vestigated more intensively by comparing animals that have been matched for sex and species, because there are sex and species differences in re- fractive characteristics among monkeys that parallel the sex differences

Development of Optical Characteristics for Seeing found in humans.34 When 299 wild monkeys were compared with 323 laboratory monkeys, the wild monkeys were found to be significantly more hyperopic than the laboratory monkeys. When 143 wild monkeys were matched against the same number of laboratory monkeys on the basis of species, age, sex, and weight, the same results were obtained. Furthermore, when 50 inside-cage animals were matched against 50 outside-pen animals on the basis of species, age, sex, weight, diet, time in captivity, and time spent in cage or pen, the caged animals were sig- nificantly more myopic than the outside-pen animals. Because any heredi- tary contributions were confounded by the random factors involved in capture and placement of the monkeys, and the influences of age, sex, and diet were reduced by the matching procedures used, the conclusion that restriction of visual environment has an effect on the refractive characteristics seems to be supported. Control animals placed in chairs similar to those used in the visually restricted space but without the visual-restriction hoods showed an aver- age change of one eighth of a diopter over a 1-year period, while the ex- perimental animals were showing changes greater than one diopter over the same period. An attempt was made to evaluate the effect of different levels of illu- mination on the development of myopia in the visually restricted space.35 Three groups of animals were kept in the space for 7 months: three ani- mals at a level of illumination of 25 foot-candles (fc), six animals at 4 fc, and four animals at 0.02 fc. The animals at 4 fc developed an average of three fourths of a diopter of myopia, whereas the animals in the other two groups developed one fourth of a diopter of myopia in the same period. This difference is significant at the 1% level and suggests that the level of illumination plays a role in the development of myopia, with lower levels having a greater influence on the development of myopia than higher levels (possibly because of variations in the amount of ac- commodation exerted), except for the extremely low levels, which had little effect. In other words, the effect of illumination on myopia in- creases with an increase in illumination up to 4-5 fc. As the level of illu- mination is further increased, the effect on myopia decreases. If near-work has an effect on the development of myopia, it should operate with some relationship to the changes that occur when one looks from a far to a near object. The two major changes are in accommoda- tion and in convergence. To investigate the effect of changes in accom- modation, a group of monkeys were placed in the near-work situation 47

FRANCIS A. YOUNG until they started to show the changes toward myopia38; the animals then continued in the situation but were given three drops of 1% atropine each day for 2 months. During that period, the myopia decreased by about one-half diopter and remained constant for the duration of the study. This suggests that accommodation plays a role in the near-work effect. Several studies by clinical investigators have shown that children placed on cycloplegics show little or no change toward myopia while the cycloplegic is in effect.1'5 Sato" found that Japanese children taking a cycloplegic daily showed a regression in the measured amount of myopia of approximately one-half to three-fourths diopter as long as they were kept on the cycloplegic; shortly after they stopped taking it, they showed a further increase in myopia. Current studies on monkeys kept in the visually restricted space show no changes in refractive character- istics until a spasm of accommodation develops. (Ultrasound measure- ments and corneal measurements are combined with refractive measure- ments in these studies.) Under these conditions, the eye accommodates and maintains the accommodation for some period without relaxation. If this spasm is maintained for a month or more, it is followed by an in- crease in axial length, which apparently continues as long as the spasm continues. Sato reduced the spasm of accommodation through the pro- longed use of cycloplegics; but as soon as the subjects were taken off cycloplegics, they returned to their near-work environment and again established a spasm of accommodation, which resulted in further in- creases in myopia. Studies by van Alphen21 clearly support the finding that tension on the choroid is increased during the act of accommodation. If accommo- dation were prolonged, the tension on the choroid against the vitreous body would tend to reduce the blood flow through the choroid and re- sult in ischemia of the choroid and the retina, which depends on the choroid for its blood supply. This prolonged state of lowered blood supply could result in a gradual weakening of the retina, choroid, and sclera and a stretching of the layers at the posterior pole of the eye, which would increase the anterior-posterior axial length. Subjects who do not have this prolonged spasm of accommodation would probably not develop ischemia or the weakening of the eye itself. All these findings have been derived from studies on animals that can be placed in a situation of controlled visual environment. However, we are concerned primarily with human subjects, and it is important to demonstrate that the optical characteristics of animals are similar to 48

Development of Optical Characteristics for Seeing those of humans. Leary and I have made such a comparison between the development of myopia in chimpanzees at Holloman Air Force Base and human subjects studied in London. This comparison shows that the an- nual rate of change in vertical ocular refraction is the same in humans and chimpanzees, but with more variability in the latter. The only major difference between the human and the chimpanzee in terms of the opti- cal characteristics that accompany myopia is that the vertical corneal power change decreases in the human but increases in the chimpanzee. The changes in lens power, the power of the eye, and the reduced axial length are comparable in chimpanzees and humans, and in general, the course of development of myopia in humans and chimpanzees is the same except for the previously mentioned variability and the changes in vertical corneal power. As a result of this basic similarity, it should be possible to study the influence of various factors, both environmental and genetic, on the development of myopia in chimpanzees and, by ex- tension, on other primates and to generalize the findings to the humans. If it may be said that the visually restrictive environment leads to myopia in subhuman primates, what effect, if any, does heredity have on the development of myopia in these primates? I found no relation- ship between offspring and parents in refractive characteristics.26 Simi- larly, when intersibling correlations were determined on human subjects, no relationship was found between one sib and the other in refractive characteristics, although there was a significant correlation in height, weight, and IQ.23 Thus, environment appears to have a more important role than heredity in the development of myopia. The monkey studies suggest that there is little or no myopia without restriction of visual space and that, if one's visual space is restricted, myopia will develop and will do so more rapidly under low levels of illumination than under quite high levels of illumination. STUDIES ON ESKIMOS Recently, we studied the vision of an Eskimo population at Barrow, Alaska. Several investigators3'8'13 have found that primitive peoples who do not read or engage in substantial near-work have virtually no myopia. Cass states that "myopia is unknown among the pure-blooded adult Eskimo. The majority have negligible refractive errors and a small num- ber have low hypermetropia."3 Skeller has indicated that in the Ang- 49

FRANCIS A. YOUNG magssalik Eskimo population the incidence of myopia is not more than 2%.13 The Eskimos at Barrow appeared to be a desirable population for study, in that three generations were available within the village. Further- more, the number of children-per family is high, with an average of eight. Only the present generation has had required schooling compa- rable with that in the older states. Thus, there was the possibility of both a genetic study and a study of the effects of schooling and of reading under low illumination on refractive characteristics. In Barrow, houses are not equipped with light meters, and the city utility charges its customers on the basis of a unit consisting of the con- sumption of a 40-W bulb per month. Because the winter months are com- pletely without daylight, students must read either under the adequate fluorescent lighting in the school buildings or in the inadequate incan- descent light at home. This is a situation somewhat comparable with that of the monkey studies in which the animals were kept in a visually restricted space under low illumination. The study was done on 508 volunteer family members on whom com- plete data were obtained. Because volunteer subjects may be biased in favor of visual problems, a follow-up study was made of all the Eskimo schoolchildren in Barrow under 17 years of age. The proportions of various refractive errors are similar in both groups. Figure 1, based on data from the original group, presents the propor- tion of persons requiring a minus lens for correction, by age groupings from 6 to 88 years of age. No myopia exists in the oldest generation— those over 50. This is in line with the reports of little or no myopia among Eskimos. In the second generation, from 26 to 50, there is more myopia among the younger persons: fewer than 5% of the 41- to 50- year-olds have myopia, approximately 23% of the 31- to 40-year-olds, and 44% of the 26- to 30-year-olds. Overall, about 21% of the second generation are myopic. An amazing 88% of the 21- to 25-year-olds have myopia, about 58% of the 16- to 20-year-olds, and 52% of the 11- to 15-year-olds, for an overall 62% of the 11- to 25-year-olds. As indicated earlier, myopia usually does not develop until approximately 11-12 years of age; consequently, we would not expect to see a great degree of myopia below this age level. If the 6- to 10-year-olds are included, the overall percentage for the third generation (6-25 years old) is 43%. We rarely find such high proportions of myopia among white subjects living in the United States or Europe. Figure 2 is a plot of the mean refractive characteristics of the same

Development of Optical Characteristics for Seeing 100 6-10 11-15 16-20 21-2526-30 31-3536-40 41-45 46-50 51-55 56-6061-6566-70 71-88 AGE, YEARS FIGURE 1 Percentage of myopes in different age groups of Barrow Eskimos (number of per- sons measured in parentheses). age groups. It can be seen again that most of the older persons are hy- peropic, but those between 11 and 25 years old are myopic. The average refractive error of the 21- to 25-year-olds is approximately -2 diopters. Most of the members of the oldest generation live the typical Eskimo life, which involves no reading and much outdoor activity, inasmuch as they depend on hunting and fishing to provide their food. At the time of the Second World War, the armed forces began to gather the second- generation men as employees, and these men moved their families into the village of Barrow. These Eskimos learned to read, and the data show that among them the proportion of myopia begins to increase. The per- sons under 25 years old have had compulsory schooling comparable with that required of children in the older states. It is difficult to see how such a great increase in myopia could be ac- complished by hereditary changes within only three generations. Thus, it appears again that the visual environment plays a large role in the

FRANCIS A. YOUNG -2.00 6-10 11-15 16-2021-25 26-30 31-35 36-4041-4546-50 51-5556-60 61-65 66-70 71-88 AGE,YEARS FIGURE 2 Mean refractive error in different age groups of Barrow Eskimos. Same subjects as in Figure 1. development of myopia and that the suggestions of Ware22 and Blagden2 are applicable to this Eskimo population. In addition to reading with poor lighting for long periods, many of these children are overcorrected, which increases the amount of accommodation required for them to read. All these conditions are conducive to the development of myopia if myopia results from an exertion of continuous accommodation. As has been pointed out by Price,10 there has been a great change in the diet of the Eskimos during the period in question. That adds another variable that may be related to the development of myopia. Cass3 ex- plains the absence of myopia among adult Eskimos, and its appearance among children (along with dental caries), as due to change from a tra- ditional high-protein diet to a high-carbohydrate diet. Two facts should be considered in evaluating the possible effects of diet on the develop- ment of myopia: the diet is essentially a U.S. diet, but the incidence of 52

Development of Optical Characteristics for Seeing myopia far exceeds that in the United States; and diet was not a factor in the development of myopia in monkeys under similar environmental conditions. OPTICAL CHARACTERISTICS AND READING Because our purpose here is to determine the role of various factors in the failure to learn to read, we should turn from those who have become efficient readers (and, apparently as a result, have become myopic) to those who do not become efficient readers and usually do not become myopic. As has been stated many times, there are few myopes among school dropouts, and most school dropouts have good distance vision. It is conceivable that this good distance vision may be part of the reason they have never learned to read well. The hyperopic person is capable of seeing quite well at far distance with the exertion of a small amount of accommodation, but must exert an even greater amount of accommoda- tion when he attempts to read. These great amounts of accommodation, which permit him to read, are likely to result in serious visual symptoms that appear to be unrelated to the eye itself. He may develop double vision, inability to concentrate, blurred images, nausea, and general mal- aise. If the hyperopic person avoids reading, he readily avoids all these symptoms. He will be loath to spend any time in reading, because he finds it distasteful. Without practice, one cannot become a good reader. Therefore, he finds it difficult to master his schoolwork at the level at which reading is essential and may eventually become a dropout. If such a person is examined with a Snellen chart for near vision, he is likely to show normal visual acuity for short periods, and his visual problems are likely to be ignored. The emmetropic person should be able to read effectively as well as to see clearly at long distance and should not have to exert any more accommodation to read than a corrected myope would have to exert. Thus, he should be as comfortable as the myope in reading. However, the myope apparently can adjust satisfactorily to reading, so much so that he does a great deal of it, whereas the emmetrope may or may not be equipped to become an efficient reader, and whether he can become one can be determined only by the skillful application of visual measur- ing techniques, and not simply by measuring visual acuity at long or short distances. Because the myope has exerted a self-selection process, 53

FRANCIS A. YOUNG in that he has tested himself for reading and has found that he can ac- complish this task efficiently, we could probably conclude that the average myope is better equipped to read than the average emmetrope. Furthermore, inasmuch as the myope shows a decrement in visual per- formance on the Snellen distance acuity chart, he would be referred to a vision specialist—either an ophthalmologist or an optometrist. The specialist would subject him to a clinical examination and determine whether he had any other problems, such as suppressions, anisometropia, or phorias, that could affect his reading performance. Thus, he would have demonstrated a general ability to read and would be properly fitted with glasses to take care of a visual deficiency. The emmetrope, how- ever, would successfully pass the Snellen test, would not be referred to a specialist, and, consequently, could have any number of unrecognized visual problems that would affect his ability to read. For example, the emmetrope could have astigmatism of a magnitude that would affect his reading performance and yet be able to pass a Snellen test successfully. He could have accommodative problems that would make it difficult for him to get equally clear images in both eyes, so that his accommodation would be in a continual state of activity, seeking to clear different images on each retina. This activity might cause a good deal of asthenopia or visual discomfort and make reading diffi- cult. The emmetrope could become an alternate suppressor because of the difficulty of handling these different images, and he could pass the Snellen test using either eye or both eyes together, and yet be essentially one-eyed as far as reading is concerned, with a resulting decrease in read- ing efficiency. He could have convergence problems and overconverge or not be able to converge sufficiently. It is possible to have too much or too little convergence for the amount of accommodation exerted. But the interrelationship between accommodation and convergence cannot be detected by a simple test of acuity or a standard clinical refraction. It should be stressed that the hyperope and the myope may have similar problems, but only the myope is likely to be corrected for these prob- lems, because of the ease of detecting myopia in the school situation. Any child who has reading difficulties should be given a thorough examination by a qualified orthoptist as well as a qualified optometrist or ophthalmologist. The accommodation-convergence relationship should be carefully investigated; all refractive errors should be corrected and deficiencies in accommodation and convergence should be determined. Only with properly coordinated convergence and accommodation and 54

Development of Optical Characteristics for Seeing properly corrected refractive errors can a person be visually equipped to read effectively. This type of evaluation ideally involves a refraction with and without cycloplegia, as well as the thorough evaluation of the accom- modation-convergence relationship. Such an examination is usually not given by either the optometrist or the ophthalmologist in his ordinary clinical procedures. Because the person who is reading effectively has demonstrated that he does not have serious problems along these lines, only the person with definite reading problems would have to be ex- amined this thoroughly. Until such visual examinations are made, it is not possible to rule out visual defects as contributory factors in reading disabilities. The work reported here was supported in part by U.S. Public Health Service research grant NB 05459 from the National Institute of Neurological Diseases and Blindness; the 6571st Aeromedical Research Laboratory, Holloman Air Force Base; the Yerkes Regional Primate Research Center, Emory University, Atlanta, Georgia; the Wisconsin Regional Primate Research Center, Madison, Wisconsin; the National Center for Primate Biology, University of California at Davis; the Oregon Regional Primate Research Center, Beaverton, Oregon; and the University of Washington Regional Primate Research Center, Seattle, Washington. REFERENCES 1. Bedrossian, R. H. The effect of atropine on myopia, pp. 1-8. [Lecture 8.] In First International Conference on Myopia, Vancouver, Washington, Sept. 10-13, 1964. New York: Myopia Research Foundation, 1964. 2. Blagden, C. Appendix [to James Ware's paper, Observations relative to the near and distant sight of different persons]. Phil. Trans. Roy. Soc. Part 1:110-113, 1813. 3. Cass, E. Ocular conditions amongst the Canadian western Arctic Eskimos, pp. 1041-1053. In International Congress Series Number 146. Proceedings of the XX International Congress of Ophthalmology. Amsterdam, London, and New York: Excerpta Medica Foundation, 1966. 4. Deller, J. F. P., A. D. O'Connor, and A. Sorsby. X-ray measurement of the diameters of the living eye. Proc. Roy. Soc. 1348:456-467, 1947. 5. Gostin, S. B. Prophylactic management of progressive myopia. Guildcraft 37:5- 15, 1963. 6. Hirsch, M. J. Relation of visual acuity to myopia. Arch. Ophthal. 34:418-421, 1945. 7. Hirsch, M. J. The relationship between refractive state of the eye and intelli- gence test scores. Amer. J. Optom. 36:12-21, 1959. 8. Holm, S. Les dtats de la refraction oculaire chez les pale'ne'grides au Gabon, Afrique Equatoriale Frangaise; dtude de race pour e'clairer la gene'se de la reTrac- tion. Acta Ophthal. (Suppl. 13): 1-299, 1937. 55

FRANCIS A. YOUNG 9. Kempf, G. A., S. D. Collins, and B. L. Jarman. Refractive errors in the eyes of children as determined by retinoscopic examination with a cycloplegic. Results of eye examinations of 1,860 white school children in Washington, D.C. In Pub- lic Health Bulletin Number 182. Washington, D.C.: U.S. Government Printing Office, 1928. 56 pp. 10. Price, W. A. Nutrition and physical degeneration: A comparison of primitive and modern diets and their effects. Redlands, Calif.: The author, 1945. 11. Sato, T. The Causes and Prevention of Acquired Myopia. Tokyo: Kanehara Shuppan Co., 1944. 184 pp. 12. Schapero, M., and M. J. Hirsch. The relationship of refractive error and Guilford- Martin temperament test scores. Amer. J. Optom. 29:32-36, 1952. 13. Skeller, E. Anthropological and ophthalmological studies on the Angmagssalik Eskimos. Copenhagen: Reitzels, 1954. 231 pp. 14. Sorsby, A., B. Benjamin, J. B. Davey, M. Sheridan, and J. M. Tanner. Emmetropia and its aberrations; a study in the correlation of the optical components of the eye. In Special Report Series Medical Research Council Number 293. London: Her Majesty's Stationery Office, 1957. 69 pp. 15. Sorsby, A., B. Benjamin, M. Sheridan, J. Stone, and G. A. Leary. Refraction and its components during the growth of the eye from the age of three. In Special Report Series Medical Research Council Number 301. London: Her Majesty's Stationery Office, 1961. 67 pp. 16. Sorsby, A., G. A. Leary, and G. R. Fraser. Family studies on ocular refraction and its components. J. Med. Genet. 3:269-273, 1966. 17. Sorsby, A., and M. Sheridan. The eye at birth: measurement of the principal diameters in forty-eight cadavers. J. Anat. 94:192-197, 1960. 18. Sorsby, A., M. Sheridan, and G. A. Leary. Refraction and its components in twins. In Special Report Series Medical Research Council Number 303. London: Her Majesty's Stationery Office, 1962. 43 pp. 19. Steiger, A. Die Entstehung der spharischen Refraktionen des menschlichen Auges. Berlin: S. Karger, 1913. 567 pp. 20. Stevens, D. A., and H. H. Wolff. The relationship of myopia to performance on a test of leveling-sharpening. Percept. Motor Skills 21:399-403, 1965. 21. van Alphen, G. W. H. M. On emmetropia and ametropia. Ophthalmologica (Suppl. 142): 1-92, 1961. 22. Ware, J. Observations relative to the near and distant sight of different persons. Phil. Trans. Roy. Soc. Part 1:31-50, 1813. 23. Young, F. A. An estimate of the hereditary component of myopia. Amer. J. Optom. 35:337-345, 1958. 24. Young, F. A. An evaluation of the biological and nearwork concepts of myopia development. Amer. J. Optom. 32:354-366, 1955. 25. Young, F. A. Development and retention of myopia by monkeys. Amer. J. Optom. 38:545-555, 1961. 26. Young, F. A. Heredity and myopia in monkeys. Optom. Weekly 57:44-49, 1966. 27. Young, F. A. Interrelations of visual measures. Amer. J. Optom. 36:576- 585,1959. 28. Young, F. A. Myopes versus nonmyopes—a comparison. Amer. J. Optom. 32: 180-191, 1955. 29. Young, F. A. Myopia and personality. Amer. J. Optom. 44:192-201, 1967. 56

Development of Optical Characteristics for Seeing 30. Young, F. A. Reading, measures of intelligence and refractive errors. Amer. J. Optom. 40:257-264, 1963. 31. Young, F. A. Refraction of the monkey eye under general anesthesia. Vision Res. 3:331-339, 1963. 32. Young, F. A. The aetiology of myopia. Optom. Weekly 56:17-24, 1965. 33. Young, F. A. The distribution of refractive errors in monkeys. Exp. Eye Res. 3:230-238, 1964. 34. Young, F. A. The effect of atropine on the development of myopia in monkeys. Amer. J. Optom. 42:439-449, 1965. 35. Young, F. A. The effect of nearwork illumination level on monkey refraction. Amer. J. Optom. 39:60-67, 1962. 36. Young, F. A. The effect of restricted visual space on the primate eye. Amer. J. Ophthal. 52:799-806, 1961. 37. Young, F. A. The effect of restricted visual space on the refractive error of the young monkey eye. Invest. Ophthal. 2:571-577, 1963. 38. Young, F. A. Visual refractive errors of wild and laboratory monkeys. E.E.N.T. Digest 27:55-70, 1965. 39. Young, F. A., R. J. Beattie, F. J. Newby, and M. T. Swindal. The Pullman study: Part I. A visual survey of Pullman school children. Amer. J. Optom. 31:111-121, 1954. 40. Young, F. A., R. J. Beattie, F. J. Newby, and M. T. Swindal. The Pullman study: Part II. A visual survey of Pullman school children. Amer. J. Optom. 31:192- 203, 1954. 41. Young, F. A., and D. N. Farrer. Refractive characteristics of chimpanzees. Amer. J. Optom. 41:81-91, 1964. 42. Young, F. A., and G. A. Leary. A comparison of the optical characteristics of the human, ape, and monkey eye. Amer. Psychol. 525, 1967 (abstract). 43. Young, F. A., and G. A. Leary. Mechanisms underlying the development of myopia. Amer. J. Optom. (in press) 44. Young, F. A., G. A. Leary, and D. N. Farrer. Ultrasound and phakometry mea- surements of the primate eye. Amer. J. Optom. 43:370-386, 1966. DISCUSSION DR.ALPERN: Did you find a sex difference in myopia in children? DR. YOUNG: Yes. In my review of early studies, I found reports that girls develop myopia earlier than boys, but girls read much more effectively than boys. DR. ALPERN: There is an even more impressive sex relationship. Some workers in Ann Arbor have medically examined most of the population of Tecumseh, Michi- gan; one of the things they looked at was refractive errors (Francis and Epstein, 57

FRANCIS A. YOUNG in International Conference on Comparability in Epidemiological Studies, Mil- bank Memorial Fund Quarterly 43, No. 2,1965). They found that the age of onset of the increase in myopia is impressively related to the onset of puberty. As far as I know, no one has looked at this aspect of the problem. DR. YOUNG: We have looked at it in monkeys, and there is no relationship between the development of myopia and puberty in these animals. We do not believe that puberty is an essential aspect. Although a child starts to read in school at 6 years of age, he does not really get into substantial amounts of reading until the sixth or seventh grade. Some intensive reading may occur before the sixth grade, but most children are not required to do any substantial amounts of reading until they get to junior high school. DR. KEOGH : It is possible that the early experiences of boys and girls differ and that girls have been involved in the kind of things in school that would involve them in near-work from an earlier age than boys. DR. YOUNG: I think that this is really one of the main differences: that girls may be culturally exposed earlier to near-work situations and develop a much better visual performance, whereas boys find it more difficult to develop an effective near-work performance pattern later in life. If children are exposed to near-work situations earlier, they also run the risk of becoming more myopic, inasmuch as the degree of myopia is related to the age of onset. Perhaps reading glasses or drugs could be used to reduce accommodation and control myopia. DR. LUDLAM: We are involved in the same type of work as Dr. Young. Myopia, as we have discussed it here, is a developmental type of myopia. There is another type that, luckily, is not very common. There may be 2% of myopes born with a very high degree of error, perhaps something like 10 or 20 diopters of myopia, and it is very interesting in that generally it does not worsen. They may be born with 20 diopters of myopia, and they may reach the age of 20 with the same degree of myopia; or they may even experience a decrease, and the myopia may reduce to 8 diopters over the years. And I am talking about good readers. The type of myope that Dr. Young has been discussing is different. Ordinarily, the schoolchild has normal vision until the age of 7-10 years and then becomes progressively myopic through the school years, sometimes leveling off at the end of high school and sometimes continuing right on through college. I have patients who are engineers and have this myopic condition increasing right up to their 50's. There is still some question, but the relationship holds pretty well, that progressive myopes in general are good readers. These would be the uncompli- cated school myopes. There are also people who read early and do a lot of reading, and still do not develop myopia. You mention the refractive errors under the effect of a cycloplegic, a drug that paralyzes the muscle inside the eye that operate the lens (the ocular accom- modative mechanism). This suggests the finding of R. E. Bannon (Amer. J. Optom. 58

Development of Optical Characteristics for Seeing 24:513-568, 1947) that the effect of the drug on refractive error varies. There is no systematic shift in what is called the cycloplegic error from the manifest error. Some people show more change in error in this shift than others, and the usual error from this source is not small. I think we must remember that these effects may result from the use of cycloplegics. What monkeys do and when they do it does not seem to be completely rele- vant to humans. These shifts are due to endogenous factors. What solid evidence do you have? DR. YOUNG: The solid evidence is that I can show the very same effects in mon- keys. When you do not put them into the experimental situation, they do not show an increase in myopia. How relevant this is to the problem in humans, I do not know. The basic refractive characteristics of the eye are determined under a condition of relaxed accommodation, because this is a reproducible condition in which the eye is adjusted for vision at a distance of 20 ft or greater. The state of relaxed accommodation is usually obtained in one of two ways: through the use of atropine or some other type of cycloplegic drug, which acts to block the nerve impulses to the ciliary muscle, or through the use of plus lenses, which cause the retinal image of a distant object to appear blurred, which in turn leads to a reflex relaxation of accommodation. The measurement of refractive error or the optical characteristics of the eye under the cycloplegic condition would determine the cycloplegic error, and measurement under the plus lens, the manifest error. If the drug and the plus lens are equally effective, the refractive errors should be the same under both conditions except for errors of measurement. Bannon found that the errors are not the same under the two conditions, and, although the drug condition usually shows more hyperopia than the plus lens condition, by about 0.5 diopter, this is not always the case. The present concept of the mechanism of accommodation requires that accommodation be relaxed if there is no stimulation of the ciliary muscle. The "relaxed" state of accommoda- tion may not be as great as possible under all conditions; there may be variations in the tonic state of the ciliary muscle. It is generally thought that the use of cycloplegic drugs over such a period as several days to a week will induce the maximal degree of lenticular relaxation. A recent study of ours (Amer. J. Optom., in press) repeated the Bannon study but used Eskimos as subjects, rather than Caucasians. Those subjects with no re- fractive errors or with hyperopic errors demonstrated similar effects in both studies. When Bannon's 594 hyperopic eyes are compared with the 513 similar eyes in our study, 69% of the Bannon eyes and 78% of the eyes in our study show an average increase in hyperopia of about 0.5 diopter under the cycloplegic con- dition. Bannon had no change in 22%, and we had no change in only 10%. Bannon found 9% and we found 12% showing less hyperopia under the cycloplegic condi- tion. These eyes may represent instances of unreliability of the measurement of refractive error; there is no reason to believe that the lens condition would be 59

FRANCIS A. YOUNG superior to the drug condition in inducing relaxation of accommodation. When persons with myopic refractive errors are compared under the two re- laxation conditions, major differences are found. The myopic subjects in our study showed the same results as those demonstrated by the hyperopic subjects: 75% had more hyperopia, 15% no change, and 10%lesshyperopia. In contrast, among Bannon's myopic eyes, only 38% had more hyperopia, 36% no change, and 26% less hyperopia under the drug condition. The high proportion with less hyperopia under the drug condition is not easy to explain, if one uses the 9% found with hyperopic cases as an estimate of the unreliability of measurement. DR. SCHUBERT: A well-known specialist has recommended "learning glasses" that would be plus spherical lenses. What do you think about that in connection with what you have stated? Do you know of any cases in which that has been done? DR. YOUNG: I would like it. It would be a logical conclusion in terms of accommo- dation in this myopia problem. I know of a number of cases in which clinicians have used bifocal lenses, which are virtually the same thing. There is some confu- sion in the literature as to whether bifocals have therapeutic effect. I would say that most of the studies so far have not been well controlled. In cooperation with an ophthalmologist, we have been carrying on one study with several hundred children for the last 8 years; the results look very suggestive. If bifocals are prop- erly fitted and if the plus-lens segment is fitted high enough, the myopia will not increase at the same rate as in children who are not fitted with bifocals. DR. BOYNTON: You mentioned that the accommodative act causes the retina to move forward. This seems inconsistent with the fact that the axial length of the eye, according to your thesis, becomes longer as a result of accommodation. DR. YOUNG: Your question is how the elongation process can occur, if, in accom- modation, the back part of the eye is tightened and moved forward around the vitreous body. Our studies indicate that, if the subject (monkey or human) de- velops a continuous level of accommodation or spasm, the increase in axial length tends to follow within a variable period, being shorter for more adult animals and longer for young animals. It is our belief that this continuously exerted ten- sion interferes with the nutrition and metabolism of the retina, choroid, and sclera and results in the weakening of these structures, so that they begin to stretch. As they stretch, the space is filled with aqueous fluid, so that the process is more or less continuous as long as the spasm of accommodation continues. If the newly developed myope were not fitted with glasses for distance, this process should theoretically reach a stabilization point that would leave him fairly well adjusted for his most common nearpoint distance. However, if he is fitted with a distance correction, the process would be restarted, because he would again be trying to overcome the effect of the distance correction while effectively spending most of his time at a nearpoint distance. We have no reasonable idea as to the basic mechanism that results in the weakening and stretching of the tunics of the eye. Bill (Exp. Eye Res. 5:45-54, 60

Development of Optical Characteristics for Seeing 55-57,1966) has shown that, during the period of ciliary muscle contraction, the movement of the aqueous from the anterior chamber back through the posterior parts of the eye just beneath the sclera is blocked. This path represents one of the important avenues for the drainage of aqueous humor but may also serve some type of a nutritive function in this part of the eye. If that is the case, then perhaps the blocking of the aqueous plays some role in this process, which results in the weakening and stretching of the retina, choroid, and sclera. 61

DELWYN G. SCHUBERT Induced Refractive Errors in Human Subjects This presentation is based on two published studies that Walton and I conducted at the Los Angeles College of Optometry—one on induced myopia12 and the other on induced astigmatism.10 A third study, to measure the effect of induced hyperopia, is being planned. INDUCED MYOPIA AND FAR-POINT PERCEPTION Many teachers and reading specialists use tachistoscopes, controlled readers, films, and film strips as parts of reading improvement programs. Most teachers on occasion use chalkboards, flannel boards, and flash cards for instructional purposes. Teachers expect children to respond quickly to materials presented by these methods. Rarely, however, do teachers give attention to the sensory or motor skills required for effec- tive far-point achievement. In clinical refraction,4 visual acuity is determined by having the sub- ject identify letters of various sizes at optical infinity (20 ft). No precise time limit is applied; thus, the subject has an opportunity to study each letter. Under these conditions, Hirsch8 has determined the approximate visual acuities for various degrees of myopia. Weymouth14 and Weston13 emphasized the importance of considering the time factor in testing 62

Induced Refractive Errors in Human Subjects acuity. Hartley2 systematically explored the relationship of time and distance to visual acuity by showing a single fine line randomly in eight different positions. As the distance was increased, the probability of correct response decreased, and as the exposure time was reduced, the probability of correct response decreased. The purpose of our studies was to determine the effects of artificially induced nearsightedness on far-point tachistoscopic perception. Twenty-four college seniors were subjects in our study. They were trained observers with corrected or uncorrected 20/20 vision in each eye and were asked to report on words flashed for 1/25 sec from a tachisto- scope under constant light intensity at variable exposure distances: letter size expressed as Snellen fraction 8.7mm 20/20 10.9mm 20/25 17.4mm 20/40 34.8 mm (approximately the size of 20/80 chalkboard handwriting) Artificial myopia was induced by convex spherical lenses in increments of 0.25 diopter until the subjects began to suffer stress from the refrac- tive error. The results are shown in Tables 1-3. It is interesting to note that, al- though all subjects had 20/20 visual acuity, approximately half the 20/20-size words were missed and three fourths of the responses fell below the 80% level of accuracy. Because a short exposure reduces con- trast, we were actually measuring the subjects' contrast sensitivity. As shown in Tables 1 and 2, this factor influences the achievement for 20/25 and 20/40 words, but to a lesser extent than for the 20/20 words, be- cause the larger words subtend a greater angle at the nodal point of the eye and thus are more easily resolved. It is significant that even 0.50 diopter of induced myopia results in a decrement of performance with every letter size. Table 3 shows an even greater percentage loss when myopia is induced by 0.75 diopter. These findings indicate the need for optimal visual acuity in far-point tachisto- scopic training. It is also evident that students with maximal visual acuity have an advantage in classroom situations demanding rapid and accurate interpretation of material at a distance. This advantage is also applicable to distance seeing outside the classroom. 63

DELWYN G. SCHUBERT TABLE 1 Correct Responses for 10 Exposures (expressed in averages) Word Size Diopters of Induced Myopia (as Snellen fraction) Piano +0.25 +0.50 +0.75 +1.00 +1.25 +1.50 20/20 20/25 20/40 20/80 5.18 7.58 5.42 4.92 9.62 9.25 8.33 10.00 9.75 9.46 3.17 6.00 8.96 4.46 8.54 7.17 5.50 TABLE 2 Percentage of Responses Falling below 80 Percent Accuracy Level Word Size (as Snellen Diopters of Induced Myopia fraction) Piano +0.25 +0.50 +0.75 +1.00 +1.25 +1.50 20/20 20/25 20/40 20/80 72.7 50 70.5 75 4.2 8.3 20.8 0 0 4.2 95.8 50 62.5 12.5 41.6 TABLE 3 Percentage Loss with +0.75 Diopter of Induced Myopia 12.5 62.5 Word Size (as Snellen Induced Myopia fraction) Piano +0.75 Difference % Loss 20/25 20/40 20/80 7.58 3.17 9.62 6 10 8.96 -4.41 -3.62 -1.04 58 37.6 10.4 It is apparent that larger letters permit greater accuracy in tachisto- scopic training. The smallest letter for far-point training should be 34.8 mm (20/80); smaller letters result in a decrement of performance, even if the subject has 20/20 vision. SUBJECTIVE EFFECTS OF INDUCED ASTIGMATISM Is astigmatism detrimental to reading efficiency? Does it produce symp- toms? If so, what are they? Betts3 found astigmatism associated with many of his cases of severe 64

Induced Refractive Errors in Human Subjects reading disability and felt that it was one of the causes of the disability. Eames,6 however, found a greater incidence of astigmatism among good readers than among unselected ones. Many researchers7'11'15 report an inability to differentiate groups of good and poor readers on the basis of astigmatism. Several specialists have voiced the opinion that severe astig- matism might prove detrimental to efficient reading in individual cases. Romine9 thought that "it would seem most important to correct any marked degree of astigmatism," and Cleland,5 sharing that view, stated that "in severe cases of astigmatism it was found to be closely allied with reading failure." Several of the foregoing studies involved a comparison of the visual characteristics of students who were successful and unsuccessful in read- ing, and the investigators did not state whether the refractive errors were corrected or uncorrected at the time of testing. If the subjects were fully corrected, they would be emmetropic, and comparisons would therefore have involved reading ability with normal vision. This idiosyncrasy, along with the conflicting opinions of the investigators, piqued our curi- osity. If a group of readers were subjected to induced astigmatism, would they experience adverse visual and psychophysiologic effects? Astigmatism is a refractive condition in which a variation of refractive power exists in the different meridians of the eye. Generally, one meri- dian exhibits the greatest power and one the least, and these are known as the principal meridians. The cause is almost always a difference in curvature of the refractive surfaces of the ocular media. Most astigmatism is believed to result from unequal curvature of the cornea.4 Astigmatism is the most prevalent refractive anomaly. It is classified into the following corneal types: With the rule (direct)—the curvature of the greatest power lies verti- cally Against the rule (inverse or perverse)—the meridian of greatest curva- ture lies horizontally Oblique—the meridian of greatest curvature lies between the vertical and horizontal Bannon and Walsh,1 in a study of 2,000 patients with refractive prob- lems, found that five sixths of them had astigmatic errors of refraction; of the five sixths, about 40% had astigmatism with the rule, about 25% against the rule, and about 35% oblique. 65

DELWYN G. SCHUBERT Cavara, quoted by Borish,4 calculated the distribution of different degrees of astigmatism (Table 4). It is apparent that the greatest inci- dence of astigmatism is between 0.50 and 1.00 diopter, and it is most frequently with the rule. Therefore, for our study, we induced 1.00 diopter of astigmatism with the rule while our subjects, 35 seniors from the Los Angeles College of Optometry, from 22 to 47 years old, per- formed an intelligence test. At the conclusion of the test, each student recorded his own subjective reactions to the induced astigmatism. As shown in Table 5, well over half the students (63%) experienced TABLE 4 Distribution of Degrees of Astigmatism" Amount of Astigmatism, Diopters Distribution, % 0.50 22.94 0.50-1.00 42.44 1.00-1.50 16.18 1.50-2.00 9.21 2.00-3.00 6.39 3.00+ 2.84 "Derived from Borish.4 Number of subjects = 5,241. TABLE 5 Subjective Effects of Induced 1 -Diopter With-the-Rule Astigmatism in 35 Subjects Reported Not Reported No. Reporting Specific Effects Effect No. % No. % Visual 22 63 13 37 Blur 19 Diplopia 0 Distortion 9 Psychologic 24 69 11 31 Weariness 13 Exhaustion 10 Retreat 16 Headaches 24 69 11 31 Generalized 6 Intraocular 7 Frontal 14 Temporal 4 Unilateral 0 Occipital 2

Induced Refractive Errors in Human Subjects visual difficulties—blur and distortion, but no diplopia. Of the 35 sub- jects, 69% reported adverse psychologic effects from the induced astig- matism. The desire to retreat from the test situation appeared with greatest frequency; weariness and exhaustion were also experienced. Headaches were reported by 69% of the students. Frontal headaches were reported most frequently, but there were also reports of intraocu- lar, generalized, temporal, and occipital headaches. No one reported unilateral head pains. Astigmatism is generally believed to be one of the major causes of ocular asthenopia, and that belief was supported in this study. The fact that children and adults with uncorrected astigmatism fre- quently confuse similar letters (m and n, o and c) and similar words (flip, flap, flop) can be explained on the basis of the visual blur and dis- tortion encountered. It has been found that the symptoms of blur, dis- tortion, and headache become limiting factors in reading efficiency. A person plagued with these symptoms is markedly handicapped as a stu- dent because he cannot sustain concentrated reading for long periods. REFERENCES 1. Bannon, R. E., and R. Walsh. On astigmatism. Amer. J. Optom. 22:263-277, 1945. 2. Bartley, S. H. Some relations between optical resolution and response. Amer. J. Optom. 27:333-344, 1950. 3. Betts, E. A. The Prevention and Correction of Reading Difficulties, p. 156. Evanston, 111.: Row, Peterson and Co., 1936. 402 pp. 4. Borish, I. M. Clinical Refraction, pp. 45-46 (2nd ed.). Chicago: Professional Press, 1954.576pp. 5. Cleland, D. L. Seeing and reading. Amer. J. Optom. 30:467-481, 1953. 6. Eames, T. H. A frequency study of physical handicaps in reading disability and unselected groups. J. Educ. Res. 29:1-5, 1935. 7. Eames, T. H. Comparison of eye conditions among 1,000 reading failures, 500 ophthalmic patients, and 150 unselected children. Amer. J. Ophthal. 31:713- 717, 1948. 8. Hirsch, M. J. Relation of visual acuity to myopia. Arch. Ophthal. 34:418-421, 1945. 9. Romine, H. Reading difficulties and eye defects. Sightsav. Rev. 19:98-99, 1949. 10. Schubert, D. G., and H. N. Walton. Effects of induced astigmatism. The Reading Teacher 21:547-551, 1968. 11. Swanson, D. E., and J. Tiffin. Betts' physiological approach to the analysis of reading disabilities as applied to the college level. J. Educ. Res. 29:433-448, 1936. 67

DELWYN G. SCHUBERT 12. Walton, H. N., and D. G. Schubert. Induced myopia and far-point perception, pp. 276-278. In Improvement of Reading through Classroom Practice. J. A. Figurel, Ed. International Reading Association Conference Proceedings, Vol. 9. Newark, Del.: International Reading Association, 1964. 331 pp. 13. Weston, H. C. Sight, Light and Efficiency. London: H. K. Lewis Co., 1949. 308 pp. 14. Weymouth, F. W. Vision of the Aging Patient: an Optometric Symposium, pp. 45-46. M. J. Hirsch and R. E. Wick, Eds. Philadelphia: Chilton Co., 1960. 328 pp. 15. Witty, P., and D. Kopel. Factors associated with the etiology of reading disability. J. Educ. Res. 29:449-459, 1936. DISCUSSION DR. YOUNG : Dr. Schubert, did you have a control group who went through this with blank lenses? DR. SCHUBERT: No, we did not have a control group. DR. SILVER: How many hours did you allow the subjects taking the test? How much time was allowed for accommodation? DR. SCHUBERT: That was a weakness in the experiment. The subjects did not have time to adjust to induced astigmatism. DR. DOTY : I think one of the weaknesses was that they knew precisely what they were going to be subjected to. DR. SCHUBERT: They were aware of the facts, yes, but they did not know whether their responses were correct or incorrect. They tried their best, and we found out, as mentioned in the results for myopia, that their accuracy level did drop. DR. INGRAM: If you tell a group of patients or subjects that they have a tempera- ture of 99F, you will find that they have symptoms. If you tell a similar group of patients who have temperatures of 99F that they have normal temperatures, they will have fewer symptoms. I think you must have a control set of blank lenses to use in the same situation. And I think you might well have had someone who took the same test saying that he could not go on because he had a headache—while he was wearing blank lenses. DR. SCHUBERT: Would the subjects not perceive almost immediately that they were wearing blank lenses? DR. INGRAM : I think it is very easy to deceive them this way. If they are myopic, they are going to have symptoms; and even if they are not myopic, all you need to do is tell them what symptoms to expect, for instance, that eyestrain is asso- ciated with headaches. 68

Induced Refractive Errors in Human Subjects DR. ROBINSON: I have a number of patients on whom we put blank lenses. These were children who complained of vision problems, even though no refractive errors were found by cooperating specialists, who said that the lenses would help. Psychologically, the results were apparently very good, in the sense that the headaches disappeared and the reading improved. The second point I would like to make was already raised, but let us reiterate it. Your subjects were adults who had not become accustomed to this kind of correction, whereas children who have never seen any other way may not be aware of the correction. I am wondering, therefore, whether the results can be carried over from adults to children. DR. SCHUBERT: Probably not, although I think it gives us some insight, for ex- ample, into astigmatism. It appears that persons who see printing in a blurred fashion might develop a headache; I am sure there are headaches that develop under these circumstances. We are aware of the need to improve these studies and can try to improve them in the future. We plan to introduce another group to minus lenses in an attempt to induce hyperopia. At that time, we can use a control group that would be exposed to plain lenses and plus lenses. DR. YOUNG: I would suggest using, as well, a crossover group, a blind study. DR. LUDLAM : Yes, a blind study using plus errors. And the experimenters doing the test should not know who has the lenses and what they are, and the subjects should not know anything about the lenses. Then correlate your results, and I believe these observations will be accurate. Perhaps these students from the Col- lege of Optometry, who are trained observers, are not the best subjects. DR.MASLAND: I suggest that you take a group of subjects who have a refractive error and test them immediately after correcting that error, without letting them know. Give them a set of lenses that exactly correct their error. You might take a group of these subjects and give them intelligence tests on the very day that you fit them with bifocals. DR.ULLMANN: I would like to suggest one more variation, to determine whether women are more responsive than men to the experience of stress that you have introduced by the test procedure. The commandant at West Point once became very much concerned by the number of cadets (it seemed to be exceptionally large) who wore glasses at a football game. He was concerned whether the re- quirements for admission to West Point should be changed. This raises the ques- tion of how much variation is permissible under stress—how much stress is toler- able. It is possible to straighten out this point with two different subject groups to see whether they accept stress in two different ways or react similarly. DR. SILVER: The discussion has been in terms of similar errors in both eyes. What about children who have a large error in one eye and normal vision in the other eye? What relationship would this have to reading? DR. SCHUBERT: I do not know the answer to that, but there are vision specialists here who probably can respond to your question. 69

DAVID G. COGAN / JERRY B. WURSTER Normal and Abnormal Ocular Movements Ocular movements are customarily divided, according to the psychosen- sory stimulus that evokes them, into vestibular, regard, pursuit, and command categories. Their nature and development may be analyzed by objective observation of the normal infant and by study of congenital abnormalities. In this presentation, we will be concerned first with the definitions and neurologic bases of normal ocular movements and then with some of the abnormal conditions of developmental origin. NORMAL OCULAR MOVEMENTS Although the subject of ocular movements divides itself into categoric functions, as though each were a separate anatomic entity, all portions of the brain potentially participate in all movements. But the evidence indicates that discrete areas of the brain have a degree of autonomy in effecting movements in response to particular stimuli, and it is conve- nient from a pedagogic point of view to emphasize this separateness. Vestibular Movements Each labyrinth exerts a net tonic innervation tending to turn and rotate the eyes conjugately to the opposite side. The otolith organs are respon- 70

Normal and Abnormal Ocular Movements sible for the static tonus, and the semicircular canals respond to accelera- tion and deceleration in such a way as to maintain the inertia of the eye position. Thus, acceleration tends to turn the eyes to the side opposite the direction of gross movement of the head, and deceleration tends to turn the eyes toward the same side as the direction of movement of the head. This vestibulogenic movement is normally present at birth, even in the premature infant, and may be elicited simply by rotating the infant in one's arms. The otolithic influence on the horizontal and vertical displacement of the eyes is normally masked by the voluntary and optic sources of ocu- lar movement, and its influence on rotary displacement, an almost ex- clusively vestibulogenic function, is difficult to detect grossly. The in- fluence of the semicircular canals is, however, easily seen in the contra- versive displacement of the eyes in response to rotation of the head or irrigation of the ears with warm water (conversely, cold water causes a deviation of the eyes to the side that it is applied to). Normally, the vestibulogenic movement resulting from displacement of the endolymph in the semicircular canals is relatively slow and con- tinually corrected by quick movements bringing the eyes back toward the primary position. The cycles of slow vestibulogenic phase and fast corrective phase constitute a form of jerk nystagmus that can usually be elicited in the full-term neonate. In premature infants, however, the slow phase alone may be present for days or weeks, so that on rotation of the head the eyes maintain a conjugate deviation to one side, instead of de- veloping a nystagmus. This is called a "doll's-head," or "doll's-eye," movement. The fast or corrective phase in man is mediated through the cerebrum and does not depend on the labyrinth. It is served by pathways that are identical with, or closely allied with, the volitional movements. Thus, persistent doll's-head deviations are characteristic of some supra- nuclear lesions. Regard Movements The movements of regard are those elicited by objects attracting one's attention to an eccentric portion of the field. They are movements of attention and, although they are ordinarily evoked by visual stimuli, similar movements may result from auditory stimuli. The ocular movements to fixate an object of attention are quick and simulate command movements—with which they may, in fact, be iden- 71

DAVID G. COGAN / JERRY B. WURSTER tical. Little is known of the efferent pathways for movements of regard, but they are deficient in animals with bifrontal lesions4 and in human beings with Parkinson's disease. The infant shows few movements of regard during the first few weeks of life. Objects that alert his attention, as evidenced by a startle reaction or retraction of the lids, are met with a straightforward stare. Oculomotor apraxia is one condition in which regard movements (and pursuit and command movements) are never fully developed. Pursuit Movements Ocular movements in following a moving object are called pursuit or optokinetic movements. Unlike movements of regard (or command), they may be slow and they are only partially under voluntary control. Thus, if the entire environment moves relative to the observer, as in looking out of a train window or in an ideal optokinetic test, the eyes are compelled to follow the moving object. When the eyes have reached a comfortable limit of excursion they are brought back toward the pri- mary position by a quick movement, and again follow the moving objects. The repetitive cycles constitute optokinetic nystagmus, and are a con- venient test for the integrity of the pursuit movement. The standard optokinetic drum has black and white stripes that are rotated at a suitable speed in front of the subject's eyes. The afferent arc of the optokinetic response consists of the visual pathways, although those in only one hemisphere are sufficient to produce a response, whereas the efferent arc is mediated through the parietal lobes of the cortex. Further details of the pathways are somewhat obscure, but in- tegrity of neither the frontal nor the occipital lobes is necessary for the optokinetic response. In the infant, an optokinetic response may be evoked almost imme- diately after birth, provided the drum's stripes are large enough to be within the discriminative acuity of the neonate and that the moving ob- jects subtend practically the entire visual field.3'5 Previous reports that the optokinetic response did not develop until a month or more after birth were based on artifacts of testing. 72

Normal and Abnormal Ocular Movements Command Movemen ts Command movements are those elicited on command. They are often equated with volitional movements. Command movements are probably mediated through the frontal lobe. That is where one would expect the volitional centers to be represented, and experimental stimulation of a discrete area in the posterior portion of the second frontal convolution causes conjugate turning of the eyes. Lesions in this area, however, cause surprisingly little disturbance of volitional control, unless they involve both frontal centers. The inference is that the volitional control of ocular movements has a large measure of bilateral representation. The infant appears to have poor volitional control of his eye move- ments at birth and to be unable to turn his eyes at will or in response to an object of regard for the first few weeks of life. With congenital ocu- lomotor apraxia, there is a permanent defect in volitional control of horizontal movements. ABNORMAL OCULAR MOVEMENTS The most common examples of abnormal oculomotor development are congenital nystagmus, strabismus, palsies of conjugate gaze, and oculo- motor apraxia. Because these do not necessarily have any relationship with one another, they will be described separately. Congenital Nystagmus Nystagmus consists of rhythmic oscillations of the eyes. The congenital or developmental variety is almost always horizontal and conjugate (that is, manifest equally in the two eyes) but varies for different directions of gaze. There are three major types of congenital nystagmus; they will be described here only cursorily, because they have been treated in de- tail elsewhere.2 Congenital sensory nystagmus consists of predominantly pendular oscillations and is secondary to poor vision that dates from early life. If vision is lost after 4-6 years of age, nystagmus either does not develop or is abortive. The reason for this critical period is unknown, but it has a counterpart in other fields of neurology, such as speech development and somesthetic deprivation. Some authors have assumed that the ocu- 73

DAVID G. COGAN / JERRY B. WURSTER lar oscillations serve a useful function by providing a scan that partially compensates for the lack of central vision; but that assumption is con- tradicted by evidence that arresting the eyes improves vision. The most reasonable explanation for this type of nystagmus is to regard it as an ataxia of eye movements. Macular vision serves the eye muscles as the position sense that other skeletal muscles derive from proprioceptive end-organs. If macular vision fails to develop, the eyes cannot hold fixa- tion and thus develop an ataxic nystagmus. We have called this a sensory type of nystagmus because, in contrast to other types, the primary defect is perceptive. The magnitude or coarseness of the nystagmus will vary not only with the age at which vision is lost, but with the amount of visual loss. When the visual defect is not severe (e.g., Snellen acuity of 20/50 in the better eye), the nystagmus will consist of fine horizontal oscillations; with visual loss to 20/200 or greater, the nystagmic excur- sions will be progressively grosser. The so-called "searching movements of the blind" are an extreme of this sensory type of nystagmus. Congenital motor nystagmus consists of a jerking nystagmus with a fast component to the right on gaze to the right and fast component to the left on gaze to the left. The neutral point is the intermediate posi- tion at which the eyes are approximately stationary and the vision is approximately normal. When, as is often the case, the neutral point is eccentric, the patient stabilizes his eyes by turning his head. Habitual head-turn is thus a common presenting sign of this type of nystagmus. The motor type of congenital nystagmus is of genetic origin and is espe- cially common in males. It simulates a mild paresis of gaze wherein the patient is unable to maintain fixation to either side. Because the horizon- tal optokinetic response is also characteristically abnormal in these pa- tients, one may postulate a defect, possibly a failure of myelination, in the pathways from the parietal lobes that mediate the optokinetic reflex. Despite this presumed organic basis, the motor type of nystagmus is not characteristically associated with other neurologic abnormalities. Latent congenital nystagmus is so called because it is elicited by cover- ing one eye. Customarily, the eyes are stationary, but covering either eye evokes a conjugate jerk nystagmus with a fast component toward the side of the covered eye. The etiology of this type of nystagmus is obscure. It may be related to congenital motor nystagmus, inasmuch as it occurs significantly often with this type. It is not related to the func- tion of simultaneous binocular vision; it occurs commonly in patients with strabismus, and simple disruption of binocular vision does not bring 74

Normal and Abnormal Ocular Movements it out. It may be related to a phototonic balance between the two eyes, in that progressive darkening of one eye will cause a proportionately coarse nystagmus. Strabismus The complex subject of strabismus may be simplified by considering first the two common forms of functional (nonparalytic) strabismus. These are called "alternating strabismus" when the patient uses either eye for fixation and "monocular strabismus" or "concomitant strabis- mus" when one eye is used for fixation and the other eye habitually deviates. The basis for alternating strabismus is unknown, but it is as though the patient lacked sufficient stimulus for binocular vision. The vision is usually normal and equal in the two eyes. Monocular strabismus, however, is often associated with hyperopia and requires an excessive accommodation for clear distance vision. This in turn produces an ex- cessive stimulation of the correlated convergence mechanism, with con- sequent turning-in of one eye. The eye with the greater refractive error is almost invariably the one that turns. If, during the early years of life, the deviant eye is not forced to fixate, through patching of the good eye, the vision fails and the eye is said to have amblyopia ex anopsia. The basis for this amblyopia (popularly called "lazy eye") is obscure, but it seems to be age-dependent. Amblyopia occurs characteristically only in the first 6-7 years of life, during which it can be reversed by forced use of the amblyopic eye. After that age, and certainly after the first decade, amblyopia does not develop and, if already present, cannot ordinarily be reversed. Paralytic strabismus is the other major type. It is an esotropia (inward turn of one eye) when the lateral rectus muscle is paralyzed, an exotro- pia (outward turn) when the medial rectus is paralyzed, and a hyper- tropia when one of the vertically acting muscles is paralyzed. Paralytic strabismus does not differ appreciably, as a developmental abnormality, from that acquired later in life, but the diplopia that is so incapacitating in adult-onset cases is not present in infantile or childhood cases. The young patient readily suppresses the false image. Contrary to popular belief, strabismus itself causes little impairment of a child's visual functions. The absence of binocular vision and stereop- sis causes only minor problems in everyday life, and has no bearing on dyslexia. But it is important to prevent amblyopia before the vision in 75

DAVID G. COGAN / JERRY B. WURSTER the strabismic eye is irretrievably lost. The cosmetic effect of strabismus is, of course, also a major consideration. Palsies of Conjugate Gaze Lesions in the brain stem cause palsies of conjugate gaze, those in the pons affecting horizontal movements and those in the anterior midbrain affecting vertical movements. In addition to these acquired palsies, paralyses of gaze occur occasionally as congenital or developmental abnormalities. Best known is the Mobius syndrome, in which paralysis of conjugate lateral gaze to either side is associated with facial diplegia. Convergence is unaffected and, except for the compensatory head move- ments and the expressionless facies, the defect causes no functional handicap. Congenital Oculomotor Apraxia This is a condition in which, despite full random and vestibulogenic movements, a person is unable to move his eyes efficiently at will or in following a moving object to either side. Vertical movements are unaf- fected. Because the fast phase of vestibular nystagmus is apparently served by the same neural arc as that of voluntary movements, rotation of such a patient about a vertical axis causes a contraversive deviation of the eyes instead of a nystagmus. To fix an object to either side, the per- son turns his head instead of his eyes, but the contraversive deviation of the eyes necessitates an overshoot of the head for fixation. The result is a characteristic head thrust unlike that seen with other types of conju- gate palsies of gaze (see Figure 1). Congenital oculomotor apraxia is sometimes familial. It is not typi- cally associated with other neurologic abnormalities. Static visual func- tions are normal, but children with oculomotor apraxia are invariably slow readers and, despite adequate intelligence, do poorly in school. They have a true oculomotor dyslexia. Although head thrusts become progressively less conspicuous throughout the first decades of life, the defect is never fully outgrown. The person with oculomotor apraxia con- tinues to have some difficulty in rapid voluntary or following move- ments of his eyes to either side throughout life, and never becomes a facile reader. 76

Normal and Abnormal Ocular Movements The work reported here was supported by U.S. Public Health Service Center grant NB 05691 from the National Institute of Neurological Diseases and Blindness. NORMAL APRACT 1C FIGURE 1 Comparison of head-eye move- ments in the normal and apractic child on gaze to the left. Whereas the eyes precede the head on eccentric gaze in the normal person, the head precedes the eyes in the child with congenital apraxia. In the latter case, the eyes manifest a contraversive devia- tion, necessitating an overshoot of the head on fixation of a target. (The blink is usually, but not always, present in the normal person and usually, but not always, absent in the apraxic patient.) (Reprinted with permission fromCogan.1) 77

DAVID G. COGAN / JERRY B. WURSTER REFERENCES 1. Cogan, D. G. A type of congenital ocular motor apraxia presenting jerky head movements. Trans. Amer. Acad. Ophth. Otolaryng. 56:853-862, 1952. 2. Cogan, D. G. Congenital nystagmus. Canad. J. Ophthal. 2:4-10, 1967. 3. Gorman, J. J., D. G. Cogan, and S. S. Gellis. An apparatus for grading the visual acuity of infants on the basis of opticokinetic nystagmus. Pediatrics 19:1088- 1092, 1957. 4. Kennard, M. A. Alteration in response to visual stimuli following lesions of frontal lobe in monkeys. Arch. Neurol. Psychiat. 41:1153-1165, 1939. 5. McGinnis, J. M. Eye-movements and optic nystagmus in early infancy. Genet. Psychol. Monogr. 8:321-430, 1930. 78

KENNETH R. GAARDER Eye Movements and Perception A fundamental topic in understanding reading disability is the physiol- ogy of visual information processing. The understanding of that, in turn, depends on a recognition of the role of eye movements in perception. Older work on eye movements in reading (reviewed by Tinker16), com- bined with the ideas I shall present, shows that one way of viewing read- ing disabilities and perceptual disorders is in terms of ineffective pro- gramming of visual input related to faulty functioning of eye movement mechanisms. To understand this fully, it is necessary to grasp the extent to which the perceptual process depends on eye movements. Most of what I shall say is an examination of the mechanisms whereby eye move- ments mediate perception. Naturally, we will note that eye movements and perception during reading represent only special cases of eye move- ments and perception in a wider context. By first considering eye move- ments and perception in general, we hope to be in a position to under- stand them better in reading and in disordered perception. I shall attempt to establish two main points: first, that the input of visual information is discontinuous (packaged, sampled, gated, chopped, intermittent, incremental, or step-functioned), with the discontinuities mediated by jumping eye movements; and second, that we may usefully conceive of a hierarchic structuring of intrinsic units of visual percep- 79

KENNETH R.GAARDER tion, wherein eye movements determine the nature of the units at one level. The second point may best be understood by drawing analogies to other information-bearing systems. We shall arbitrarily choose printed language as an example of another information-bearing system, partly because it will lead back to further consideration of one focus of this meeting—reading. DISCONTINUITY OF VISUAL INPUT There are two major reasons why scientists have not recognized the dis- continuity of visual input before.9 The first is that vision is subjectively experienced as continuous over time and that our conceptual construct of the real external material world is overwhelmingly one of temporal continuity. As we experience the material world, we are aware of no "breaks" in the time during which our eyes move about, nor does this world seem made up in any way of "pieces." This is in sharp contrast with the facts of the input process. The second reason for not appreciat- ing discontinuity is that, until the arrival of the computer age, the distinc- tions between "continuous" and "discontinuous" processes were not so concrete as they have been since we have begun to use these problem- solving machines, which are either analog (continuous) or digital (discon- tinuous). What is at issue is the difference between an information- processing system that takes in information continuously and one that moves in steps, or incrementally, so as to process information in chunks or pieces. A few examples of continuous and discontinuous processes make the distinction clearer. Continuous processes are exemplified by the "coded" groove of a phonograph record and the modulations of radio waves; the discrete, tapped-out letters of a typewriter and the suc- cessive frames of a motion picture are discontinuous. Although mathe- maticians have long been aware of these distinctions, it is only now, with so many of us using computers, that they have become common, ex- perientially understood technologic tools. Time Course ofSaccades The reason for laboring this point is to bring to your consideration the idea that visual perceptual input is not continuous, as it seems to be, but discontinuous, very much like the successive frames of a motion picture. 80

Eye Movements and Perception But what kind of event in the visual system would represent the chang- ing of the frames? Let us consider some elementary facts about eye move- ments, overstating to some extent for simplification. The most impor- tant fact is that, in moving about to see the environment, the eye moves in virtually only one way—by abrupt, rapid, discrete jumps (exceptions, such as tracking movements, are well discussed in standard texts).1 Figure 1 shows that we are dealing with a discontinuous process—what engineers refer to as a "step-function" and the technical literature, "sac- cades." During reading, there are about four of these jumps per second. During other times, while the eyes are open, there are usually at least two jumps per second. The jumps continue during visual fixation, while the eyes fixate a target. The figure shows a typical recording of fixation eye movements showing the size of jump of about 10 min of arc (1/6 of a degree) and a rate of one or two per second. Before describing evidence of discontinuity of visual information in- put and processing, I would like to illustrate the manner in which eye movements affect the time course of the perceptual process. Figure 2 shows the effects of gross jumping of the eye about a simple scene. In the upper left is the scene, on which are superimposed five numbered dots connected by lines. The dots represent five successive hypothetical fixations of the fovea (the center of the retina) within the scene, and the lines show the track of the eye over the scene as it jumps. We may as- sume that the scene was briefly flashed (tachistoscopically) for several seconds on a screen and that during that time the viewer made the five 1 sec FIGURE 1 Tracings of eye movements during visual fixation, showing several rapid jumping eye move- ments. H marks the horizontal component of the movement and V the vertical component, recorded by reflecting onto photographic paper beams of light from mirrors mounted on contact lenses. 81

KENNETH R. GAARDER 1-2-3-4-5 1-2-5-3-4 1-2-5-4-3 1-2-3-5-4 1-2-4-3-5 1-2-4-5-3 FIGURE 2 Simulation of eye movements in viewing a scene. At upper left is the scene with five numbered dots and connecting lines superimposed. The dots represent five successive fixa- tions and the lines represent the track of the eye jumps between these fixations. The upper right shows arbitrary central retinal areas around each of these fixations. The row of circles simulates the time sequence of presentations to the brain of the chain of five successive central retinal views as the eye views the scene. The tracing below the circles shows the horizontal component of the successive fixations, with L and R representing the left and right directions. The bottom of the figure shows alternative tracks after the first two fixations, illustrating the stochastic na- ture of the process. 82

Eye Movements and Perception fixations along the indicated track. On the upper right is the same scene with five superimposed circles representing arbitrary equal central retinal areas as they would be on the retina or anywhere back of the retina. Be- low is a row of circles forming a chain, which represents the sequence in time of these successive central retinal areas as they would be on the ret- ina or anywhere back of the retina. Note that each circle represents a chunk or package of information. The bottom of the figure illustrates two less-important issues to be mentioned in passing. The horizontal tracing simulates a recording of the horizontal component of the eye movements as they occur; at the bottom of the figure, we assume that the same five fixation points were chosen but vary the sequence after fixations 1 and 2. Edge Visual Images Figure 3 simulates the effects on the retina of the eye jumps that occur during visual fixation on the same scene. It shows packaging of informa- tion, but of a slightly different sort. Because these fixation eye jumps are much smaller than in the previous example, they do not have the effect of causing the same massive transformation of the central retinal area as in the gross viewing eye jumps. Instead, they result in changes at the edges of objects on the retinal image, imitated here by a photographic technique. If positive and negative transparencies of the same scene are superimposed with a slight displacement, the resulting print shows by lightening or darkening of a particular edge the change that would take place on the retina as the result of a small eye jump. The only place where change occurs is at edges. The small arrows represent the vectors (size and direction measures) of the hypothetical jumps that would cause the changes shown. Note that the sets of edges generated are unique to the vector and that the set of vector-generated edges taken as a whole again implies the usefulness of stochastic models. The same thing along a single small arbitrary segment of edge on the retina is shown schemati- cally between A and B in Figure 4 (top). The center of Figure 4 simu- lates the position of the edge before (tj) and after (t2) small jumps of the eye to the left or right, and the bottom shows the net change of these jumps, which would result in "off or "on" firing of retinal ele- ments, inasmuch as what has happened are "off or "on" changes. Figure 5 is a simulation of central retinal areas during reading. The first lines show a sentence of text; the second lines, hypothetical fixa- 83

KENNETH R. GAARDER c<• N FIGURE 3 Simulation of retinal image edge generation by small fixation eye jumps. This figure shows the positive (A) and negative (B) prints of the same scene at the top. If transparen- cies of the two are fitted together with slight offset, the discrete edges of C through H result. This simulates the change of the retinal image produced by small eye jumps (indicated by the arrows). For photographic reproducibility, the displacements and the arrows are larger than the jumps that occur during fixation. At 38 cm, 1 deg of arc is about 6 mm. A typical fixation eye jump might result in an apparent displacement of 0.2-1 mm at that distance. 84

Eye Movements and Perception Saccadic Vector Retinal Image Displacement Net Change FIGURE 4 Diagrammatic representation of segment of edge on the retina. Top, segment of edge between points A and B (boundary lines are necessary for pictorial purposes). Center, dis- placement of the edge by a left or right eye-jump vector between time tj and time t2. Bottom, net change of edge produced by the jump. tions on the text; the third lines, arbitrary central retinal areas around these fixations; and the fourth lines, the time sequence of presentation to the retina and brain of the contents of these successive central retinal areas in a particular chain. It is difficult to accept these processes as rep- resenting anything other than discontinuous input mechanisms: each jump of the eye presents the brain with a discrete new package to be processed, and these events occur several times each second. 85

KENNETH R. GAARDER We are considering the idea that visual-input is discontinuous. B We are cojBdering the idfc that visuapinput is dJBontinuous. We are considering the id* (that visuafcnput is ditto ntinuous. We are consider is discontinuous. We are considering discontinuous. FIGURE S Simulation of the effect of eye jumps during reading. A, a short sample of text; B, simulation of a set of four eye fixations (dots) and the intervening eye-jump tracks (arrows) during reading of the text; C, simulation of arbitrary central retinal areas about each fixation; I >, simulation of the time sequence of presentation to the brain of the chain of four successive central retinal views. Note "overlap" or repeti- tion of words during successive fixations. Chunking or Packaging of Visual Input With this picture of the physical facts of the eye-movement system, we can now examine some of the other evidence for and against the idea of visual perception as a discontinuous process mediated by eye jumps. I shall refer to five sets of experiments and their implications. 86

Eye Movements and Perception 1. The oldest experimental evidence is the phenomenon of flicker fusion, from which it can be argued that, if, at some particular flicker rate, flicker is not perceived, these chunks of intermittently presented information are subjectively smoothed in the same way as the chunks mediated by eye movements. 2. Conversely, if there were a means to artificially prevent packaging of visual input, it could be predicted that perception would cease, as happens when eye jumps are automatically canceled in stopped-retinal- image experiments.s'14 3. Another argument holds that, if perceptual input is intermittent, there must be inhibition of vision during the periods when input is not being processed, i.e., during eye jumps. This is found to be the case dur- ing jumps: visual thresholds are raised and inhibitory neurons are acti- vated in the lateral geniculate nucleus.4'18 4. Another line of reasoning holds that, if eye jumps establish pack- ages of information, they should be followed by cortical activity marking the arrival of the packages. This is indeed the case: the eye jump triggers occipital activity, recorded as a typical averaged response.6'11 That the eye jumps are correlated with alpha rhythm is also relevant here, because it shows a relationship between packaging due to eye jumps and more general cortical packaging processes.10 5. Less-direct evidence that eye jumps establish discontinuity is pro- vided by the finding of changed fixation eye-jump vectors as a result of changes in visual stimulus. Here, the argument is that, if the form of visual input is controlled by a feedback output of the visual system, changing the stimulus would change the output that controls the input. Acknowledging these points requires that one conceive of perceptual input as discontinuous, because a discontinuous event (the eye jump) controls it.7 Evidence that might be taken to show that eye jumps have no role in perception includes the fact that visual acuity is as good during a flash that is too brief to allow eye movement as during prolonged viewing. One way of interpreting this is to say that if you can see as well during a flash too brief for the eye to jump, then you do not need eye jumps to see.2 This line of reasoning does not take into account the fact that it is not the eye jump itself that is important, but that the jump causes abrupt incremental change, which is also what the flash causes. In other words, abrupt incremental change of the stimulus is caused by flashing the stimulus or by jumps of the eye and allows vision to occur. 87

KENNETH R. GAARDER Smoothing of Input In considering discontinuity of perceptual input, a final point must be made for the sake of logic and completeness: If input is discontinuous, there must be "sampling" periods and "nonsampling" periods.17 Infor- mation that arrives during nonsampling periods must be either lost or held in some sort of short-term buffer memory. Inasmuch as a high per- centage of brief light flashes are seen routinely, there must be a short- term buffer memory in the visual system between input (sampling) mo- ments. The direct analogy to time-shared computer technology,12 which uses short-term buffer memory storage, should be noted. To recapitulate briefly what has been shown, the input of visual in- formation during perception is not continuous, but is interrupted several times each second by eye jumps, which naturally divide the input into chunks or packages; these packages are reassembled by the brain into a spatiotemporally continuous visual world including, for example, the continuous line of text read from a page; finally, the packages represent natural physiologic units—a step in the direction of reducing phenomena to units whose measurement reflects their intrinsic nature. ANALOGIES TO WRITTEN LANGUAGE We shall next explore several neglected areas, with the goal of better grasping some additional aspects of perception without which reading cannot be understood. By developing analogies between visual percep- tion and printed language, we place both the visual system and written language squarely in the generic category of information-bearing systems. We are exploring but several of the issues of interest while passing by others of equal promise (see Polanyi13). The first of these issues is the continuing quest of science for units that are intrinsic to natural phenomena rather than arbitrary. Chemistry and physics made great strides when protons, neutrons, electrons, atoms, and molecules replaced the arbitrary mass-space-time units of grams, meters, and seconds. In neurophysiology, progress is not so easy, but it would be conceded that the nerve spike must represent an aspect of such intrinsic units of nervous activity. Another issue now being perceived is that we are moving about within the domain of information rather than solely within the domain of energy.3 This means that units of energy 88

Eye Movements and Perception and space-time-mass measurement, whether intrinsic or arbitrary, al- though necessary, are inadequate, and that ultimately perception must be dealt with in informational units instead. A final issue, which also comes from information theory and general systematics, is that our models must be able to encompass the concepts of hierarchy and struc- ture. We will attempt simple definitions of these terms mainly by illus- tration. Hierarchy The three main elements of the analogy to printed language are the hier- archy of levels, intrinsic units, and formation of chains. The first level in the hierarchy of language is a set of letters drawn from the alphabet of that language (Table 1). Each letter is not only an element of the set, but also a natural intrinsic unit of the language, and the units are all of equal size. The next level of a language is a set of words, which make a larger set represented by the dictionary of the language. Words may be con- sidered natural units just as letters are, but they are not of equal size, because they are made up of different numbers of the basic units of let- ters. The next level of the hierarchy of language is a set of sentences, which may be considered as another type of unit forming a still larger set. TABLE 1 The Hierarchy of Sets in a Printed Language Description Examples Set of letters (alphabet) Set of words ("dictionary") Set of sentences (ruled by grammar) Set of text (ruled by style) a,b,c,d,.. .,x,y,z and, bird, came, doors, top Jack rolled the ball. Don't eat mushrooms. all articles, all books, all manuals Structure All these units are combined in various kinds of chains. A given text con- sists of chains of sentences, which consist of chains of words, which con- sist of chains of letters. We have seen that all the units in chains are discontinuous, rather than continuous, in the sense considered earlier and, furthermore, that they are units intrinsic to the nature of printed language, rather than arbitrary. Finally, because of the nature of infor- 89

KENNETH R. GAARDER mation and the limitations of language, each level of the hierarchy is able to contain only some types of information; the higher one goes in the hierarchy, the greater the degree of complexity that can be conveyed and the more complex the rules for this conveyance. What we are saying is that structure exists, and that it constrains all the possibilities and re- sults in the susceptibility of chains to probabilistic or stochastic models. (A simple example of constraint is that the letters making up a word must be put in the correct order for the word to exist; the structure of the word is a constraint on the set of all possible combinations of the letters in the word, e.g., CAT, CTA, ACT, ATC, TAG, TCA.) We will now look at how analogy to written language helps us to understand visual perception. The package of visual input mediated by eye jumps is analogous in several respects to the intermediate level of the language hierarchy. They share the properties of being discontinuous, of being intrinsic and natural to the function of the system, of being composed in some way of the smallest units (letters in printed language and nerve spikes in visual perception), and of being made up of various numbers of the smallest units. They also form chains to make up larger units (Figures 2 and 5). It is convenient to consider these packages as analogous to words if we bear in mind that we do not yet know enough about the visual system hierarchy to know whether there are other levels between nerve spikes and eye-jump packages, even though the hypothetically analogous letters and words are on adjacent levels. A great deal more could be said about the ramifications of each of these points of similarity, but, for the sake of clarity, only the three elements men- tioned will be established. Comparison First, a hierarchy is natural to a language as an information-bearing sys- tem; the visual system, as an analogous information-bearing system, must have a hierarchy. In other words, our analogy proposes that complex information-bearing systems are intrinsically and necessarily hierarchi- cally organized. Second, as to the question of units, we have cited two major characteristics of the analogy: the discontinuity of units at all levels (an interesting question is whether it is possible to carry informa- tion continuously at higher levels in a system whose lowest level is made up of discontinuous units—the nerve spike and the letter—and so con- strains the higher levels to be discontinuous) and the intrinsic natural re- 90

Eye Movements and Perception lationship of the units to the information-bearing system of which they are a part. Third, having noted the formation of chains in both visual perception and printed language, we can sense the importance of eye- jump packages in forming higher units of visual perception—"sentences," so to speak. For example, as you look at an object, the chain of your eye-jump packages will constitute a sentence that is completed when you glance at the next object and begin a new sentence in organizing a percept of it. We can be specific at two levels as to the nature of the visual system units: the lowest level of vision is a nerve spike, analogous to the lowest level of a written language, the alphabetic letter; and a higher level of visual perception is the package of information (mediated by the eye jump), which is analogous to one of the higher levels of a written language. THE READING MODEL We have shown that reading is a process that is divided into its natural units by the jumps of the eye. These units are somehow combined to create both a continuous visual spatiotemporal world and, in reading, a perceptual and cognitive continuity of the textual material. These eye- jump units have a rate of occurrence, with optimal and high and low rates. Thus, there is a framework on which to build a model of reading that involves programming much like that of a computer, with the same kind of vulnerability to faulty microsequences (for example, the various sequences of the letters in CAT) and interference from other sense mo- dalities or cognitive and motor spheres. This model is derived from what we have described earlier, inasmuch as programming is the arrangement of hierarchic units with better and worse alternative sequences and with alternative sets of units from different sense modalities and different spheres, which may or may not be included in the chains. The vulner- abilities referred to can briefly be considered further. The concept of faulty sequences can be amplified by analogy to our present knowledge of computer programming, from which we gain respect for the impor- tance of carrying out a series of operations in exactly the right order. From computer programming, we have learned that, even though there is more than one way to skin a cat, there is a still larger set of ways that will not work at all. In the older literature on eye movements in reading 91

KENNETH R.GAARDER one ineffective microsequence that was studied extensively was the use of regressive eye movements, that is, eye jumps that went back to a part of the text already covered. Another disorder of microsequences in- volves carrying them out too rapidly or too slowly.15 Too rapid eye jumps are undoubtedly associated with hyperaroused (overly alerted) states,8 whereas both too high and too low rates would lead to inter- ference from other spheres. Our model of visual perception has strong implications for a model of sensory processing and behavior in general that can help us to under- stand these interferences. It is apparent that, if the visual system is using its own particular coding for its own particular language, each of the other sensory systems is doing likewise, and the same thing is occurring in the cognitive and motor spheres. This can be illustrated most vividly by thinking of ourselves as individual towers of Babel or multilingual United Nations meetings—our eyes might speak German, our ears Arabic, and our stomach French, and the central processor must translate these all into English. If we accept the applicability of these analogies, we are in the useful position of being forced to make choices between time- sharing (i.e., serial processing) models and simultaneous (i.e., parallel processing) models to account for the processing between these differ- ent sense modalities. What emerges lucidly for our present concern, how- ever, is the desirability of inhibiting or "turning off other sense modali- ties, such as hearing, so as to reduce the interference with carrying out a specific function, such as reading. We can logically and theoretically characterize one class of reading disability as that mediated by interfer- ence from other sense modalities or cognitive and motor spheres. (We are not implying that this hypothetical class is uncontaminated by other classes of disability.) Also, it appears that hyperarousal may often char- acterize this type of disability, but that is another subject. REFERENCES 1. Alpern, M. Types of movement, pp. 63-151. In The Eye, Vol. 3, Muscular Mechanisms. H. Davson, Ed. New York: Academic Press, 1962. 151 pp. 2. Armington, J. C. Vision. Ann. Rev. Physiol. 27:162-182, 1965. 3. Ashby, W. R. An Introduction to Cybernetics (with answers to exercises). New York: John Wiley & Sons, 1963. 295 pp. 4. Bizzi, E. Discharge patterns of single geniculate neurons during the rapid eye movements of sleep. J. Neurophysiol. 29:1087-1095, 1966. 92

Eye Movements and Perception 5. Ditchburn, R. W., and B. L. Ginsborg. Vision with a stabilized retinal image. Nature 170:36-37, 1952. 6. Gaarder, K. Interpretive study of evoked responses elicited by gross saccadic eye movements. Percept. Motor Skills Monogr. Suppl. 2-27:683-703, 1968. 7. Gaarder, K. Mechanisms in fixation saccadic eye movements. Brit. J. Physiol. Opt. 24:28-44, 1967. 8. Gaarder, K. Some patterns of fixation saccadic eye movements. Psychon. Sci. 7:145-146, 1967. 9. Gaarder, K. Transmission of edge information in the human visual system. Nature 212:321-323, 1966. 10. Gaarder, K., R. Koresko, and W. Kropfl. The phasic relation of a component of alpha rhythm to fixation saccadic eye movements. Electroenceph. Clin. Neuro- physiol. 21:544-551, 1966. 11. Gaarder, K., J. Krauskopf, V. Graf, W. Kropfl, and J. C. Armington. Averaged brain activity following saccadic eye movement. Science 146:1481-1483, 1964. 12. Kristofferson, A. B. A time constant involved in attention and neural informa- tion processing. NASA Contractor Report CR-427. Abstract III. NASA, Wash- ington, D.C. Washington: Bolt Beranek and Newman Co., 1966. Ames Research Center, Cambridge, Mass. 39 pp. 13. Polanyi, M. Life's irreducible structure. Live mechanisms and information in DNA are boundary conditions with a sequence of boundaries above them. Science 160:1308-1312, 1968. 14. Riggs, L. A., F. Ratliff, J. C. Cornsweet, and T. N. Cornsweet. The disappear- ance of steadily fixated visual test objects. J. Opt. Soc. Amer. 43:495-501, 1953. 15. Silverman, J., and K. Gaarder. Rates of saccadic eye movement and size judg- ments of normals and schizophrenics. Percept. Motor Skills 25:661-667, 1967. 16. Tinker, M. A. Recent studies of eye movements in reading. Psychol. Bull. 55:215-231, 1958. 17. Young, L. R., and L. Stark. Variable feedback experiments testing a sampled data model for eye tracking movements. IEEE Trans. HFE-4:38-51, 1963. 18. Zuber, B. L., and L. Stark. Saccadic suppression: evaluation of visual threshold associated with saccadic eye movements. Exp. Neurol. 16:65-79, 1966. DISCUSSION DR. MASON : You have suggested that there is no way to get from one level to another. That is true of many experiences, and I think that this is the same sort of problem that has stymied psychologists for a long time: How do you get from one level to the next—from the level, say, of letters to words to sentences? DR. GAARDER: That is the crucial question, and to me it means that you cannot merely know everything about cell physiology, add it up, and make psychology 93

KENNETH R. GAARDER out of it. It will not work. It is a question of enriching pragmatically. It is a ques- tion of enriching one level by considering another level, and understanding more about one level by studying it and by referring to higher and lower levels. DR. BOYNTON: Would you guess that one could establish procedures for the transition from any one level to another? DR. GAARDER: Yes, in any domain that you care to pick—motor learning, per- ception, what have you. I think that Dr. Chall had a point about the question of whether children learn written language by coding or phonetics. She suggested that they may need to learn to code before proceeding to the phonetic parts. That may be a very rapid learning process once it is mastered. DR.LUDLAM: I would like to ask you to do an experiment. Suppose we carefully recorded the series of fixations and saccadic eye movements in the course of someone's reading and then processed them exactly the same way in the same sequence and with precisely the same timing. Suppose, for a second case, that we include the periphery of the visual fields, as well as the central area. 1 think it is almost certain that we would find that the compensation of the peripherals in artificial conditions would be very poor indeed, and it might be worthwhile to ask why. DR. GAARDER: The difference between those two situations is that an eye move- ment that is involved in the first case is absent in the second. The eye movement is under the control of the subject; what the eye is going to do next in the read- ing situation or in the more general perceptual experience is determined during the 100 msec or more of the fixational pause. I think this is exceedingly impor- tant as a problem in visual perception. In the normal situation, clearly, the "com- puter" knows where the eyes are going to go next and, as a consequence, is able to get a good deal more out of the visual input from successive fixations than could otherwise be possible. DR. ALPERN : Does this add to the relevance of eye movements for reading? DR. GAARDER: Yes, I think what the eye is doing in the previous 100 msec is very important in the problem of poor reading, and it is an aspect of the prob- lem that people have not paid very much attention to. DR. HIRSH : It seems to me you are causing yourself a great deal of difficulty by trying to extrapolate movements of the eye from the stimulus pattern alone. I think there are some rather special differences between eye movements during reading and eye movements around arbitrarily depicted shapes of the type that you show. DR. GAARDER: I suppose that one of the ways to predict eye movements involves an analogy to the way the language is structured and how well one knows that structure. At least, that appears to be the case. 94

ROBERT M. BOYNTON Retinal Contrast Mechanisms The verb "contrast," according to Webster's New Collegiate Dictionary, means "to exhibit noticeable differences when compared or set side by side." Implicit is the idea that small differences will become less notice- able or unnoticeable if the items being compared are separated. That is true for human vision.15 If two half-circles of light, each homogeneous, are very carefully butted against one another, they form a bipartite field. If their lumi- nances are equal, they will form a homogeneous disk without a discern- ible border between the two half-fields that make up the disk. Suppose that we can adjust the luminance of one half-field independently of the other. The luminance difference necessary for a border to be just per- ceived is about 0.5%.* But if the fields are moved barely apart—just *This value obtains for optimal conditions of viewing, when the just-noticeable difference is based on the standard deviation of many settings, and the subject attempts again and again to set the two half-fields exactly equal in brightness. Here and in experiments to be reviewed later, different conditions and experimental methods will inevitably yield different results. Further- more, there are significant differences among normal subjects, and in pathologic cases the values obtained may be different by more than one order of magnitude. It is not possible to introduce qualifying statements everywhere in this paper; the reader should accept values given as representative of typical subjects under optimal viewing conditions, unless otherwise stated. "Contrast" is commonly used to specify a physically measurable difference, as well as to 95

ROBERT M. BOYNTON enough to introduce a very thin black line between them—this value will increase to about 1%. Further separation will increase the value even more. Why is this? The answer is much more complicated than might be expected. Indeed, the results of even this simple experiment cannot be fully explained on the basis of our present knowledge, although a general understanding is possible. One purpose of this presentation is to review some of what is known about the retinal contrast mechanisms that underlie this and other observations. Another purpose is to relate this knowledge to the perception of small dark details against a brighter background, which is characteristic of the typical reading task. RETINAL-IMAGE CONTRAST: GRADIENTS We must begin by considering what sort of image is formed by an ex- ternal stimulus on the retina of the eye—the retinal image is by no means a perfect replica of what is outside. In any image-forming system, the image of a point is not a point, but rather an optical-spread function (see Figure 1). Diffraction provides an ultimate limit in any optical sys- tem. In the eye, aberration, light scatter, and accommodative errors broaden the function further. Its width also depends on pupil size, being minimal (and thus best) when the pupil is about 2.5-3 mm in diameter. The spread function does not depend on light intensity. At very low light levels, the function describes the probability of arrival of photons, at each spatial position, within a test period. Assuming the point-spread function as measured by Westheimer and Campbell,21 an edge between a bright field and one that is completely describe subjective experience. In this presentation, "contrast" is used to describe what can be measured with a photometer, and modifiers will be used to refer to the effects produced- optical, physiologic, or subjective-by the physical contrast stimulus. For small details seen against a large background, contrast will be defined as (/^-/i'^H />'_,,!, after Blackwell.3 Here, «, is the luminance of the small detail, or target; /;',., is the luminance of the background. Here the limit of negative contrast is 1.00 (100%), and positive physical con- trasts may assume any value. The justification for this specification is that the visibilities of dark targets (negative physical contrast) and bright targets (positive physical contrast), when seen against the same background, are approximately equal. In other situations, particularly when bipartite or striped fields are used, there can be no clear distinction between test and background, nor between positive and negative physical con- trast. Therefore, physical contrast in these cases is defined as (Bl-B2)l(Bl+B2). Here fii is the positive physical contrast and B2 the negative physical contrast; the limit of physical contrast is between zero and ±100%. 96

Retinal Contrast Mechanisms 2 0 Retinal Distance (Min. of Arc) FIGURE 1 Line- and point-spread functions on the human retina, as determined from the experimental data of Westheimer and Campbell21 by direct physical measurement on the human eye. (Courtesy of G. Westheimer.) dark produces on the retina the light distribution shown in Figure 2. We conclude that there is a gradient of intensity between the two fields, rather than an abrupt change. It is this sort of gradual variation in illumi- nation that the retina works with, and never the abrupt changes that are so easy to provide outside the eye.* RETINAL-IMAGE CONTRAST: BLACK LINES Let us now analyze what happens if the eye is confronted with a dark line, seen against a homogeneous white background. Suppose that the *For linear and homogeneous systems, a line-spread function such as that shown in Figure 2 can be converted into a modulation-transfer function. The latter shows the percentage of con- trast transferred through an optical system, plotted as a function of the frequency of sinusoids of spatial luminance variation. Although this procedure has some advantages, and there have been many experiments in which the response of the eye was examined with spatial sinusoids as the stimulus, the approach is difficult to apply when considering letters of print on a page. For that reason, I do not use the modulation-transfer approach in this discussion. 97

ROBERT M. BOYNTON 14 12 10 8 42024 Retinal Distance (Win. of Arc) 10 14 12 FIGURE 2 Light distribution in the retinal image of an edge in the human eye, best focus, 6-mm pupil, based on the measurements of Westheimer and Campbell.21 line is completely black. We can produce this, in the laboratory, by hold- ing a pair of white surfaces—say, 3 X 5-in. filing cards—in front of some kind of light trap, illuminating them diffusely from the front. If the edges of the cards are very sharp and precisely parallel, a thin black line can be produced by bringing the cards very close together. What will the retinal image look like in this situation? We can find the answer by adding together the two edge gradients, discussed earlier, as shown in Figure 3. Here it will be seen that the closer together the cards and therefore the narrower the line, the lower the contrast on the retina between the illumination at the center of the retinal image of the line and that of the uniform areas flanking it. The retinal contrast produced by lines of various widths is shown in Figure 4. We can now see that, when we are concerned with the vision of fine lines of high physical contrast, we are, nevertheless, dealing with low ret- inal contrast. The same is true for more complex forms, such as letters on a page, that are built up from fine lines. There are two basic ways to increase the retinal contrast produced by

Retinal Contrast Mechanisms looking at a line target. The first is to increase the angle subtended by the line at the eye. This can be accomplished by making the line wider, or by moving a line of fixed width closer to the eye. Increasing the width of the line produces an effect that is easily seen in the use of boldface type. Boldface looks blacker because retinal contrast is higher, although the objective contrast is the same as for regular type (produced with the same ink). The second way to increase retinal contrast is to in- crease the inherent contrast of the line with respect to its background. This can be done by making the page whiter or the ink blacker. But most inks are black enough so that, even if they could be caused to have zero reflectance, the contrast gain would be rather small. A much greater gain can be had by making the line wider. Nevertheless, the inherently low contrast of a cheap paperback book—caused by small type, low- reflectance paper, and poor control of the width and reflectance of the letters—will result in a noticeable loss of retinal contrast and consequent Retinal Distance (2-min. steps) FIGURE 3 Retinal illuminance produced by completely black lines of the widths indicated, seen against a bright background (dotted curves). The curve for a 2-min line has been derived by adding the distributions of the two edges shown by the solid line. The other dotted curves were similarly constructed by moving the edge gradients closer together or farther apart than the distance shown. 99

ROBERT M. BOYNTON 100 s 10 -L 0.1 1 2 Visual Angle Subtended by Dark Line (Mm.) FIGURE 4 Log percent retinal contrast measured as a function of log visual angle (in minutes of arc) from data of DeMott7 and Westheimer and Campbell.21 Data are from the steer eye (open circles) and human eye (dots). Replotted from Boynton.5 reduction of visibility. The bizarre inks and backgrounds now used for artistic purposes in some popular magazines sometimes produce very low physical and retinal contrasts and can thus be very difficult to read. Despite the low retinal contrasts that fine lines produce, our ability to resolve such lines is remarkable. Under optimal conditions, a good observer can detect a line that subtends only 1/2 sec of arc,10 which corresponds to seeing a wire only 1/16 in. in diameter at a distance of 1/2 mile! Extrapolating from Figure 4, it can be estimated that that cor- responds to a retinal contrast of less than 0.01%, and that is based on an illuminance distribution with gradual contours. It is probable that the contrasts plotted in Figure 4 are too low, because of difficulties of ex- perimental measurement in both experiments and the double-traverse of 100

Retinal Contrast Mechanisms the light through the eye in the human measurements. But even if the retinal contrasts were as much as 10 times as high as this figure shows, the retinal contrast at the threshold of detection is still very low. A PRIORI EXPECTA TIONS CONCERNING RETINAL MECHANISMS From the foregoing, it should be clear that one of the most critical prob- lems that the visual system meets, and somehow solves, is the detection of very low contrasts on the retina. If each retinal receptor had a private pathway to the brain, then very small differences in the initial signals produced in adjacent or nearby receptors would require preservation all the way to the brain for the difference to be discriminated there. Be- cause noise is inevitably introduced in each stage of any information- transmission system, including a biologic one, a small difference would undoubtedly be obscured by noise by the time the original activity ex- pressed itself in the brain. For this reason, we would expect a priori to find neural mechanisms to detect and augment the differences in signal strengths near the receptors. Furthermore, it would seem helpful to in- volve a large population of receptors to increase the statistical reliability of the difference. That the detectability of a line of fixed width has been found to be critically dependent on its length suggests that this is so. CONTRAST DETECTION AND LIGHT INTENSITY Although retinal contrast does not depend on light intensity, the thresh- old of detectable contrast does. We are all familiar with this from every- day experience. The physical contrast provided by, say, a newspaper depends only on the reflectances of the paper and ink used, and not on the level of illumination of the newspaper; nevertheless, we know per- fectly well that it is difficult to read the fine print in dim light, and under still less favorable conditions—for example, under the light of the moon—only the largest headlines can be resolved. The most extensive data related to this matter have been concerned with the detection of circular patches of light against a uniform back- ground of variable luminance. When the test spot is very small, the task has much in common with what is involved in the resolution of fine de- tail. From extensive data collected by Blackwell and McCready,4 I have 101

ROBERT M. BOYNTON selected the values shown in Figure 5. A test spot subtending only 1 min of visual angle and lasting for only 1 msec was used. The function la- beled AL/I shows that the contrast required just to see the test spot drops from more than 1000% at the lowest luminance used to slightly more than 1% at the highest. The same data can be plotted in another way. The curve labeled AL shows the just-visible (threshold) increment of luminance provided by the test spot. At very low levels, this threshold increment is independent of background luminance over a range of luminances that, although very dim, allow distinct visibility. As the log of background luminance (in footlamberts) is increased to more than -2, the threshold rises—slowly at first, accelerating gradually, and finally approaching a unit slope (cor- responding to a AL/Z, slope of nearly zero). Many other detectors of contrast, such as photographic plates and television pickup tubes, behave somewhat similarly. Like the eye, they -3 -2 -1 Lag L (Footlamberts) FIGURE 5 Contrast thresholds (A / // ) as a function of log background luminance, for a 1-min spot exposed for 1 msec (data from Blackwell and McCready4). Also shown is the corresponding curve for the increment threshold, A/ _ For this curve, the values on the ordinate are in log fl. The dashed curve shows the predicted behavior of an ideal detector. 102

Retinal Contrast Mechanisms perform better when there is more light. Ideal detectors also behave this way: as the numbers of photons increase, so does the statistical evidence available to discriminate an increment from its background. The dotted line in Figure 5 shows what is predicted for the AZ, curve of the ideal detector. This line has a slope of exactly 0.5 and may be extended in- definitely in both directions. This comparison shows that the eye is not an ideal detector. For the conditions shown in Figure 5, in which a very small and brief test spot was used, optimal performance is achieved at a log background luminance of about 0.5. At luminances both below and above that, more light is required in the increment spot than the ideal-detector model would pre- dict. At low levels, this has most often been explained by postulating an intrinsic noise of the retina. It causes a minimal value of AZ, to be reached (about 7 fl in this case) at low background levels; the value would be the same in total darkness. In other words, this much luminance is required for the test spot to produce a signal in the visual system that can be dis- criminated from random background activity that exists in total dark- ness. At high background levels, the eye is also responding less efficiently than it might, probably because of the influence of various adaptive mechanisms whose function is to prevent saturation of the signals being transmitted through the visual system.* RETINAL CONTRAST VERSUS SUBJECTIVE CONTRAST One further psychophysical observation bears reporting before we turn to physiologic mechanisms: the highly nonlinear relationship between retinal contrast and its visual effects. This relationship is most easily examined in a large bipartite field, where physical contrast and retinal contrast are essentially the same. The nonlinearity first expresses itself in the fact that, for very low objec- tive contrasts, no border is seen. With further increases in contrast, the border will become visible, and subjective contrast will then increase very rapidly at first, and then more slowly. For an objective contrast of more than 20% or 30%, the border between the half-fields has already become so vivid and distinct that subjective contrast will not increase Further information on the effect of light intensity on contrast detection can be found in papers by Jones,12'13 Barlow,1 and Rushton,20 which also contain additional references. 103

ROBERT M. BOYNTON much further; further increases in physical contrast produce only very small additional increases in subjective contrast. We have attempted to measure this in the laboratory, using psychologic scaling methods. It is most difficult, because the observer cannot ignore brightness differences that are needed to produce, and therefore are correlated with, physical contrast. Figure 6 is a schematic representation of the relationship be- tween objective and subjective contrast. A more objective way to demonstrate this relationship is to make some measurements of visual performance as a function of contrast. Some years ago, we measured the ability of observers to seek out and recognize complex critical targets presented against a background of pseudo-targets. Figure 7 shows that the main improvement in visual per- formance is associated with the lower range of physical contrasts. MAX o o 50 100 OBJECTIVE CONTRAST (PCT.) FIGURE 6 Schematic representation of the relationship between objective and subjective contrast. 104

Retinal Contrast Mechanisms 100 - 20 40 60 80 100 PERCENT CONTRAST FIGURE 7 Percentage correct target identification by five subjects, in a complex search task, as a function of the contrast between the stimuli searched and the background against which they were presented. (Reprinted from Boynton and Bush.6) RETINAL ANATOMY We will now review some of the evidence that clearly tells us that the outputs from the retinal receptors become highly interconnected as in- formation is processed in the retina. Consider first the anatomy of the retina. In the human eye, there are about 125 times as many receptors (rods and cones) as optic nerve fibers. Such convergence constitutes one basic form of retinal neural interaction, and it is a principal determinant of the summative receptive field. If many receptors feed information to the same optic nerve fiber, the re- sults of feeble excitation in the individual receptors (each too weak to evoke sensation) can summate, vastly improving the probability that in- formation will be delivered to the optic nerve fiber and thence to the 105

ROBERT M. BOYNTON brain. That is accomplished, however, at a sacrifice of visual resolving power. Convergence is very nonuniform and depends greatly on the part of the retina stimulated. In the peripheral retina, many thousands of receptors deliver their outputs, via the intermediary bipolar cells, to a single ganglion cell. The central retina is very differently organized. When we fixate on a stimulus, we move our eyes to place its image in a highly specialized region of the retina, the fovea centralis. Only cone receptors are found in this region, and it is here that our visual acuity is by far the best. The convergence ratio in the fovea is about unity. This has sometimes been interpreted to mean that each foveal receptor has a private pathway to the brain. That is a mistaken conception; on the basis of the arguments presented above, we would not expect such pri- vate pathways, and the direct evidence now to be reviewed shows that we do not find them. Anatomic evidence shows many interconnections among the retinal pathways, including those which begin in the fovea. On the basis of light microscopy, it has long been known that, in addition to convergence where it occurs, the human retina is richly supplied with cells that do not appear to be involved in the direct transfer of information from receptor to bipolar to ganglion cell (the classical visual pathway in the retina), but that seem to exist specifically to provide lateral interconnec- tions between the neurons of the basic pathways. An excellent notion of this may be gleaned by examination of Figure 8, a schematic diagram of the retinal connections, based heavily on recent evidence provided by electron microscopy. Note that there are two classes of cells that pro- vide lateral interconnections. Horizontal cells (H) interconnect the rod and cone receptors; they are believed to receive information from and feed it back into the receptors. Deeper in the retina, amacrine cells (A) interconnect bipolar and ganglion cells; they are also interconnected with one another and thus have the potential to carry information over very long lateral distances in the retina. The receptors themselves are also in intimate contact, with so-called tight junctions between rods and cones and between cones and cones. The details of retinal anatomy vary from one species to another, but all eyes that have been studied so far have sufficient lateral connections to produce substantial lateral interaction effects. The list includes the lowly horseshoe crab, Limulus, long a favorite specimen for visual in- vestigation. This animal has a faceted, compound eye; each ommatidium 106

Retinal Contrast Mechanisms FIGURE 8 Summary diagram of the interconnections among the neural cells in the retina. A, amacrine; H, horizontal; C, cone; R, rod; MB, midget bipolar; RB, rod bipolar; FB, flat bipolar; MG, midget ganglion; DG, diffuse ganglion. (Reprinted with permission from Dowling and Boycott.8) 107

ROBERT M. BOYNTON connects primarily to a single optic nerve fiber. Hartline first chose it, many years ago, because he believed it to be free of complicating lateral interconnections. But he and Ratliff9 have since shown that a lateral plexus of fibers exist that interconnect the pathways, and furthermore that these connections have an exclusively inhibitory function. This work, described in detail in Ratliff s fine book, Much Bands,11 provides some of the clearest available illustrations demonstrating that these inter- connections provide a neural mechanism for contrast enhancement. INHIBITOR Y MECHANISM IN LIMULUS An experimental arrangement of Hartline and Ratliff,9 using the eye of Limulus, is shown in Figure 9. Light A can stimulate only ommatidium A. In the absence of other stimulation, this produces a particular frequency of firing in the optic nerve fiber A. Light B stimulates only ommatidium B and produces a particular rate of discharge in fiber B. Suppose that, with light A turned on at its original intensity, a light stimulus to B is provided also. At a very low intensity of stimulation of B, the result will be no firing of fiber B, and no effect on the firing rate of fiber A. But, as the intensity of stimulus B is gradually increased, fiber B begins to fire. Light Light Lateral plexus Receptors Optic nerve To Amplifier A To Amplifier B FIGURE 9 Experimental arrange- ment of an experiment by Hartline and Ratliff (1957), designed to show lateral inhibition in the eye of Limulus. (Adapted from Ratliff.17) 108

Retinal Contrast Mechanisms 4.0 3.0 2.0 1.0 ii 0 5.0- 4.0- 3.0- 2.0- 1.0- 10 20 Fiber A 30 40 10 20 30 Fiber B Frequency (impulses per sec.) 40 FIGURE 10 Top, inhibition in fiber B produced by activity in fiber A. Bot- tom, converse inhibitory action. (Re- printed with permission from Hartline and Ratliff.9) As shown in the lower graph of Figure 10, this activity in fiber B is asso- ciated with a decrease in the frequency of firing in fiber A, which is still being directly activated only by the original light intensity delivered to ommatidium A. It turns out that, once some threshold frequency is reached, the inhibitory effect of B (as measured by the decrease in fre- quency of firing in fiber A) is linearly related to the frequency of firing of fiber B. It is therefore concluded that the B system is doing the in- hibiting, with the strength of the inhibition depending on the rate of activity in the B fiber. The experiment can also be done the other way around, and it is found that activity in fiber A produces a reciprocal in- hibitory effect on the normal firing of fiber B (upper graph of Figure 10). The nature of the inhibitory effect depends also on the distance be- tween the ommatidia being stimulated. In Figure 11, inhibitory effects are measured in turn from fibers B and C, in response to variable intensi- ties of illumination of ommatidium A, the latter leading to a variable frequency of response in fiber A, as shown on the abscissa. During re- cording from fiber B, no inhibition occurs until there are about five im- 109

ROBERT M. BOYNTON FIGURE 11 Experiment to show inhibition of activity in fiber A, produced by nearby stimulation at B and more remote stimulation at C. (Reprinted with permission fromRatliff.17) 10 15 Fiber A Frequency (impulses per sec) 20 25 pulses per second in fiber A; inhibition then rises rather steeply with increasing frequencies of A. During recording from fiber C, which is more remote from A than is B, the threshold is reached at much higher firing levels in A (nearly 20 impulses per second), and inhibition increases more gradually than for fiber B. It is to be emphasized that the inhibitory action associated with a particular system, such as the A system of the previous example, de- pends only on the frequency of response of that system, without regard to how the frequency was produced. An interesting example of this is an experiment on disinhibition shown in Figure 12. Here, three stimulus fields—A, B, and C—are used, with recordings taken from fibers A and B. Stimulation of A alone produces an intermediate discharge rate, shown on the left side of the bottom record. Field C, at the intensity used, is too far from A to have any effect, as shown in the middle part of the lower record: when C is turned on, the frequency of discharge in A con- tinues much as before. The left side of the upper record shows what happens when ommatidium B is stimulated along with ommatidium A. The stimulus to B is strong enough to produce a high frequency of dis- charge in ommatidium B, which in turn is associated with a marked in- hibition of the firing rate in A. If field C is turned on, it produces a marked inhibition of fiber B, as reflected in the middle part of the upper- most record. This reduction in the response rate of fiber B releases in turn some of the original inhibitory effect of B on A, as revealed by the fact that, while field C is turned on, the response to A increases. Note that the intensity of stimulation of B has not been varied in this 110

Retinal Contrast Mechanisms example. The response to a fixed intensity of stimulation of B has been modified by the action of C. The same effect could have been produced simply by reducing the intensity of stimulation of B. This example indi- cates that it is the response to B that is the simplest variable to consider in predicting the inhibitory action of B on A, rather than the intensity of the stimulus to B. As stated previously, these inhibitory interactions are reciprocal. If A inhibits B, then B also inhibits A. But the inhibitory action of B on A is less, because of its inhibition by A, than it would be otherwise—and so on. Consequently, inhibitory relationships that are relatively easy to understand in terms of response rate become difficult to calculate on the basis of stimulus intensity patterns alone. The equations required to do this have been worked out by Ratliff, and their predictions have been tested in direct experiments. In one such experiment, Hartline and Ratliff9 were able to show clearly that neural gradients are sharpened and enhanced by the inhibitory process. Moreover, the subjective phe- nomena of Mach bands, found in human observers, can be accounted for quantitatively in terms of the inhibitory relationships worked out in Limulus (see Figure 13). This is very important, because it strongly sug- A • B ( C © 5 mm A ' alone FIGURE 12 Oscillographic records of electrical activity of two optic nerve fibers (A and B), showing disinhibition. (Reprinted with permission from Ratliff.17)

ROBERT M. BOYNTON II o ^ 7.0 6.0 > £ 5.0 O 0 ,G O <LI •"-' in -£ £ 4.0 3.0 2.0 1.0 0.0 - 1.0 O 0. g E 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0 »- <D S. I Q. 13 ^ O 0.5 mm at the eye FIGURE 13 The upper record was obtained when the stimulating light was allowed to illumi- nate only one ommatidium from whose optic nerve fiber the records were taken. The stimulus was a lamp, having a light distribution as shown in the inset, moved laterally across the omma- tidium (X-axis). When all the ommatidia were uncovered and the stimulus was moved laterally as before, the lower record was obtained. Note the evidence of overshoot and undershoot near the top and bottom of the curve. This corresponds roughly to the appearance of Mach bands by human observers when they view such a stimulus. (Reprinted with permission from Ratliff and Hartline.18) gests that the kinds of inhibitory effects found in Limulus also occur in the human retina. RECEPTIVE FIELDS AND TRIGGER FEATURES In addition to the purely summative types of receptive fields already mentioned, which are found in the dark-adapted eye, much more com- plex arrangements are typical in warm-blooded vertebrates. Figure 14 shows one example. A microelectrode is plunged into the cat retina through the front of the eye and is in contact with a single retinal ganglion cell whose axon forms an optic nerve fiber. The experimental arrangement allows light spots to be flashed on and off in various parts 112

Retinal Contrast Mechanisms of the visual field. A central area is found where the onset of a light spot produces an increased discharge in the ganglion cell (an ON re- sponse), whereas the turning off of the light yields no response. In the horizontally shaded area shown surrounding it, both ON and OFF re- sponses are obtained. In the outer field (diagonally shaded), only OFF discharges can be recorded. This arrangement of an excitatory center field and inhibitory sur- rounding field has been found in many other experiments. (The excita- tion-inhibition relationship is often reversed.) In the warm-blooded vertebrate, unlike Limulus, it is found that the retinal ganglion cells re- spond at a modest rate, even in the absence of light stimulation. Thus, inhibition can also reveal itself as a reduction in the resting rate of re- sponse during a period of prolonged stimulation. In the frog, Lettvin et al.16 were the first to show that a single gan- glion cell responds to remarkably specific trigger features of the stimu- lus, depending on the cell recorded from. In some cases, for example, a unit will respond well only to convex dark spots moving in very particu- lar ways, and to no other stimulus. As Lettvin et al. remark, it is tempt- ing to call such a specialized unit a "bug detector." More recently, the name "trigger feature" has become associated with the peculiar, com- 1 mm FIGURE 14 Receptive field of a cat ganglion cell, obtained by Kuffler.14 This unit has an ON center and an OFF surround. (Reprinted with permission from Riggs.19) 113

ROBERT M. BOYNTON plex aspects of a visual stimulus that are often found necessary to trig- ger activity in a single unit in the retina or higher visual center. Trigger features have been examined in detail in the primate visual cortex, especially by Hubel and Wiesel in a long series of experiments (see Hubel11). Although consideration of the cortex is outside the boundaries of this presentation, it is significant to note that the center- surround relationships that are found when recording from primate cor- tical cells are nearly rectangular, rather than circularly organized. These cells are often most sensitive to moving lines, and typically are direc- tionally sensitive. In the optic nerve fibers of spider monkeys, only the concentric ON-OFF, center-surround arrangement was found at this level of the primate visual system. Directionally sensitive units have been found in the retina of the rabbit by Barlow and Levick.2 Some of the results from one of their experiments are shown in Figure 15. A black edge is caused to move across an illuminated slit. When the slit width subtends 34 min, the directional effect is very clear. When the slit is closing in a preferred Position 45°S 5°P in visual field Background Prefer Unit 4-21 8' 0.5 sec FIGURE 15 Responses to the motion of black edges across slits of various widths. (Reprinted with permission from Barlow and Levick.2) 114

Retinal Contrast Mechanisms direction (from top to bottom, as shown on the left side of the figure), the retinal ganglion cell under investigation responds with a vigorous ON-burst. But when the experiment is repeated exactly, except that the slit closes from bottom to top, only a very weak response is recorded. When the slit width is only 8 min, no significant directional effect is ob- tained. This suggests that the directionally sensitive mechanism must be contained in a larger retinal region. These investigators have done a good deal of speculation about the retinal interconnections responsible for such behavior; the horizontal cells of the retina have an important role to play in their schema. It should be emphasized that the temporal char- acteristics of retinal responses are critical in such a system. These have been investigated in some detail in further work of Barlow and Levick, as well as in Limulus by Ratliff. The electrophysiologic work carried out to date has consistently in- volved the use of very-high-contrast visual stimulation. It is not easy to relate the psychophysical results described in the first part of this pre- sentation to the electrophysiologic data discussed in the last. The work of Hartline and Ratliff is clearly the most relevant in a formal sense, in that it deals explicitly with contours and shows how the visual nervous system can operate to enhance them. Although it is a long way from Limulus to man, there is strongly suggestive evidence that similar mech- anisms operate in the human eye. It has perhaps not been emphasized enough that these inhibitory mechanisms appear, on the basis of both behavioral and electrophysio- logic evidence in vertebrates, to be specific to the light-adapted eye. In the dark-adapted state, in which the task of the eye is to gather as many photons as possible, receptive fields are purely excitatory. When more light is available, the inhibitory mechanisms come into play, and they are clearly implicated as mechanisms of contrast enhancement that help in the perception of fine detail. I wish to thank Thomas R. Corwin for his critical reading of this manuscript. REFERENCES 1. Barlow, H. B. Optic nerve impulses and Weber's law. Sympos. Quant. Biol. 30:539-546, 1965. 2. Barlow, H. B., and W. R. Levick. The mechanism of directionally selective units in rabbit's retina. J. Physiol. 178:477-504, 1965. i 115

ROBERT M. BOYNTON 3. Blackwell, H. R. Contrast thresholds of the human eye. J. Opt. Soc. Amer. 36:624-643, 1946. 4. Blackwell, H. R., and D. W. McCready, Jr. Foveal contrast thresholds for various durations of single pulses. University of Michigan Eng. Research Inst. Rept. 2455-13-F. Ann Arbor: University of Michigan, 1958. 5. Boynton, R. M. Progress in physiological optics. Appl. Optics 6:1283-1293, 1967. 6. Boynton, R. M., and W. R. Bush. Laboratory studies pertaining to visual air reconnaissance. WADC Technical Report 55-304. ASTIA Document No. AD 118250. Dayton, Ohio: Wright-Patterson Air Force Base, 1957. 48 pp. 7. DeMott, D. W. Direct measures of the retinal image. J. Opt. Soc. Amer. 49:571- 579, 1959. 8. Dowling, J. E., and B. B. Boycott. Organization of the primate retina: electron microscopy. Proc. Roy. Soc. 1668:80-111, 1966. 9. Hartline, H. K., and F. Ratliff. Inhibitory interaction of receptor units in the eye of Limulus. J. Gen. Physiol. 40:357-376, 1957. 10. Hecht, S., S. Ross, and C. G. Mueller. The visibility of lines and squares at high brightnesses. J. Opt. Soc. Amer. 37:500-507, 1947. 11. Hubel, D. H. The visual cortex of the brain. Sci. Amer. 209(5):54-62, 1963. 12. Jones, R. C. How images are detected. Sci. Amer. 219(3):111-117, 1968. 13. Jones, R. C. Quantum efficiency of human vision. J. Opt. Soc. Amer. 49:645- 653, 1959. 14. Kuffler, S. W. Discharge patterns and functional organization of mammalian retina. J. Neurophysiol. 16:37-68, 1953. 15. LeGrand, Y. Light, Colour and Vision, p. 266 (tr. R. W. G. Hunt and others). New York: John Wiley & Sons, 1957. 512 pp. 16. Lettvin, J. Y., M. R. Marurana, W. S. McCulloch, and W. H. Pitts. What the frog's eye tells the frog's brain. Proc. IRE 47:1940-1951, 1959. 17. Ratliff, F. Mach Bands: Quantitative Studies on Neural Networks in the Retina. San Francisco: Holden-Day, Inc., 1965. 365 pp. 18. Ratliff, F., and H. K. Hartline. The responses of Limulus optic nerve fibers to patterns of illumination on the receptor mosaic. J. Gen. Physiol. 42:1241- 1255, 1959. 19. Riggs, L. A. Electrophysiology of vision, pp. 81-131. In C. H. Graham, Ed. Vision and Visual Perception. New York: John Wiley & Sons, 1965. 637 pp. 20. Rushton, W. A. H. The Ferrier Lecture, 1962. Visual adaptation. Proc. Roy. Soc. 162B:20-46, 1965. 21. Westheimer, G., and F. W. Campbell. Light distribution in the image formed by the living human eye. J. Opt. Soc. Amer. 52:1040-1045, 1962. DISCUSSION DR. BOYNTON: I hope that this story has left the correct impression that the business of dissecting the retinal image, generating the neurologic code, and then 116

Retinal Contrast Mechanisms transmitting this information to the brain is not a set of passive mechanisms. The code is related to what has adaptive significance to the organism, and at many points along the way a very subtle difficulty with the pathways involved could conceivably result in improper functioning of these mechanisms. DR. LiNDSLEY: I am glad that Dr. Boynton brought up the spatial relationship of interaction within the retina. There is another important complexity: the tem- poral relationship of the stimuli. D. N. Robinson (Science 156:1263-1264, 1967) recently noted that the response of the eye to a second flash was masked, pre- sumably through lateral inhibition, by what was contained in the first flash. He noted that a third flash, like your spatial elements in a section, would mask the second one in the same way that the second was masking the first. Thus, on the temporal side, there are phenomena somewhat similar to those on the spatial side. This adds another element of complexity. Where reading difficulties are concerned, it might be better to focus on temporal than on spatial contrast. You can, in fact, combine them; that is, you can give stimuli both in sequence and separated in space. On the basis of what we know about lateral inhibition of the retina, there is a reason for what occurs be- tween 60 and 90 days of age in an infant, as Robert Fantz shows (see p. 351) using a series of concentric circles or checkerboards. Why should concentric circles have a specific effect? DR. BOYNTON: As far as I know, there is no explanation. DR. SCHUBERT: In connection with perception of printed symbols, contrast would depend on illumination, and a number of figures are given by authorities as to how much light you need on the printed page. There is no consensus in this regard. How many footcandles are needed? Can any of the participants offer data concerning the contrast that is desirable in enhancing legibility? DR. BOYNTON: There is a problem of definition that might be worth clearing up. When I used the term "contrast," I referred specifically to the physical, objective definition, the difference in luminance between two areas divided by the lumi- nance of the larger of these two areas. Physical contrast so defined is indepen- dent of the illumination on a reflecting target, such as a letter on a page. The subjective "contrast" associated with this, however, is critically dependent on illumination level, and it increases with increasing illumination. There are prob- ably dozens of mechanisms involved in this distinction. As to what constitutes a proper level of illumination, I feel that, in spite of all the arguments and all the research, we do not know exactly what does con- stitute a proper illumination level. Obviously, it depends on the inherent con- trast and size of the material being viewed, what is being looked at, and why. Blackwell (J. Opt. Soc. Amer. 36:624-643,1946), on the basis of an extensive series of investigations at Ohio State University, has come up with a set of figures that have generated a good deal of controversy. We keep learning new things. I was very impressed with what Dr. Young said 117

ROBERT M. BOYNTON concerning the effect of low levels of illumination on the development of my- opia, which is ascribed to a great deal of activity in the accommodative mech- anisms. If the illumination level is high, the depth of field of the eye is increased because the pupil is reduced. For this and probably other reasons, less accommo- dation is required at high illumination levels. Thus, there is now some evidence that tends to support the admonition, "Don't read in dim light or you'll ruin your eyes"—something my mother used to say to me. Possibly, she was correct in her assumption. But this is the first evidence that I know of to support this old wives' tale. DR. BUSER: Concerning the degree of retinal inhibition: to effect that, the retina is active as well as passive. I think that there may be some very strong evidence of this. Take, for example, the directional-sensitivity cells, which are activated preferentially by a moving light. According to Barlow and Levick, special struc- tures are very numerous in the retina of the rabbit, but the cat has very few; I do not know about primates. DR. BOYNTON: I strongly concur that this is true. I tried to point out that there are species differences. I picked Limulus for a detailed illustration largely in deference to historical values: it was the first experimental animal in which clearly defined retinal interaction mechanisms were demonstrated, although many workers had felt for many years that they must be present in the human retina. Only within the last 15 years or so have they been found physiologically in any animal, let alone man. Your point is very well taken. 118

MATHEW ALPERN The Pupillary Light Reflex and Binocular Interaction I am going to discuss some recent experiments on one aspect of the cross-talk between the two eyes in what I am constantly reminded is thought to be the simplest of all reflexes, the pupillary reflex to light. I want to speak about binocular cross-talk in this "simple" reflex, as well as about some findings on a disorder of perception—something not associated with the reading problem in any proper sense of the term, but an interesting perceptual disorder nonetheless. If all other conditions are equal, the pupil is always smaller when both retinas are illuminated than when one or the other is in the dark. Figure 1 shows some measurements of the size of the pupil in a junior medical student when one eye was in the dark and when both eyes were illuminated equally. Note that in the binocular case the pupil is always a bit smaller. By making a simple downward displacement of the monocular curves, one can get a fair prediction of the binocular data. This is not the place to document the fact that the obvious sorts of artifacts that might account for this result—fusional movements, ac- commodative movements, and so on—do not play any role in the results of such experiments. Nor do I need to describe the experiments that show that it does not matter which retina is illuminated; both pupils always go together. 119

MATHEW ALPERN R.W. 12° Field N = 400 0 Monocular • Binocular -1 LOG 10 012345 RETINAL ILLUMINANCE (TROLANDS) FIGURE 1 Diameter of the pupil of the left eye when both retinas are equally il- luminated (dots) and when the left eye is in the dark (circles). The abscissa scale in each instance is the ret- inal illuminance (log tro- lands) of the right eye. (Re- printed with permission from ten Doesschate and Alpern.2) Figure 2 shows the nerve pathways involved. There are two places in the central nervous system that separately or together would allow added activity from each eye to pool in such a way that the binocular process produces a smaller pupil than the monocular. The places are at the level of the midbrain, and presumably nothing higher than the mid- brain is involved. The experiments illustrated in Figure 3 allow us to exclude two very simple ways in which the pooling might occur. We might assume that the nervous system responds to the light that goes to the left eye and adds it, as though the same light had been given to the right eye. In this figure, the dashed line is the predicted binocular pupil size ac- cording to this hypothesis. Clearly, it does not agree at all with the measured values (circles). Furthermore, the nervous system does not add the amount of contractions of the pupillary muscle. The solid line in the figure shows the expected result according to that idea. Simple addition, either of lights or of contractions, is an inadequate description of the results. Although one can write the equations for this effect, the physiology cannot yet be said to be well understood. In the process of striving to build a reasonable model, Prof. J. ten Doesschate of Utrecht and I2 stumbled onto something that might be of interest. We borrowed from 120

The Pupillary Light Reflex and Binocular Interaction a paper by Cooper et al.l Suppose you had an excitation pool (Figure 4) in the midbrain, and that the output of all the cells was determining the size of the pupil. Suppose that everything in the right circle were driven by the right retina and everything in the left circle by the left retina. (In binocular vision, of course, both systems respond.) This sort of scheme in a rough, qualitative way will account for the results we obtained. This is purely speculative. Can we find any experimental evidence for this view? Perhaps we have a clue in an experimental finding of Hubel and Wiesel3 on single nerve cells in visual cortex of kittens. Cutting the eye muscles of a newborn kitten results in an alternating divergent stra- bismus. After this strabismus had developed, Hubel and Wiesel mea- sured the percentage of cells in the visual cortex that were binocularly driven. In the normal kitten—one with only a sham surgical procedure— TEMPORAL RETINA RIGHT EYE PRETECTAL_i NUCLEUS I EDINGER I WESTPHAL NUCLEUS I LEFT III NERVE TO LEFT CILIARY GANGLION THEN TO LEFT IRIS SPHINCTURE MIDBRAIN RIGHT III NERVE TO RIGHT CILIARY GANGLION ETC. FIGURE 2 Black-box sketch of the nerve pathways for the pupillary light reflex. 121

MATHEW ALPERN -1012345 LOG 10 RETINAL ILLUMINANCE (TROLANOS) FIGURE 3 Change in diameter of the pupil of the left eye for equal binocular retinal illuminance. The circles repre- sent empirical observations; lines are theoretical predictions based on the monocular measurements. The solid line is the change in size predicted if each monocular response is added lin- early; the dashed line is the change predicted if the lights are added lin- early. Neither fit is satisfactory. (Re- printed with permission from ten Doesschate and Alpern.2) FROM THE LEFT RETINA FROM THE RIGHT RETINA TO BOTH PUPILS FIGURE 4 Scheme of midbrain pooling that could account for the fact that the pupil is always smaller when the retinas are equally illuminated, compared with the case when one retina is in the dark. The output of the pool drives both pupils, equally larger contractions being produced by having more cells responding. The cells contained in the right circle are excited by the right retina; those in the left circle are excited by the left retina. 722

The Pupillary Light Reflex and Binocular Interaction a vast majority of the cells are binocularly driven. However, in the test kittens with strabismus, the matter is quite different. The vast majority of cells in their visual cortex are only monocularly driven. This looks like a paradigm for testing binocular pupil additivity. A person with alternating strabismus might be expected to show a much greater binoc- ular additivity than in the normal eye (because fewer cells would be bin- ocularly driven). As it happens, I have an alternating strabismus, so I put myself into this apparatus and did the experiment. It turned out to be nonsense: we could not get the expected results at all. The result was interesting, however. I would like you to think of the problem (a per- ceptual problem) of someone with strabismus whose eyes aim simultan- eously at different parts of the visual world. Such a person's way of viewing the visual world binocularly is remarkably different from the way of a person with normal binocular vision. To avoid double vision with an alternating strabismus, the retinal activity of one of the two eyes must be turned down (if not off), In the clinical literature, this phenomenon is referred to as "suppression"; there is little concrete understanding of its physiology. Figure 5 shows some of the measurements of the pupil (of my eye) in alternating strabismus. I was fixating with my left eye, although my right eye is the dominant one. Instead of getting the predicted super- additivity, there was no additivity at all. In binocular viewing, the pupils were the same size as when the right eye was in the dark. The nonfixat- ing eye made no contribution to the binocular response. That was the first interesting aspect of our results. The second was that, when the fixating eye was in the dark, the pupils were much wider than when the nonfixating eye was in the dark. In the normal subject, the eyes are equally effective in making the pupil smaller, irrespective of which one is fixating. But in strabismus, when the fixating eye (the left in Figure 5) is stimulated, the pupil is much smaller than when the nonfixating eye is stimulated. That is not a pe- culiarity of the left eye, but characteristic of whichever eye happens to be fixating at the time. Figure 6 substantiates the last statement. When the right retina is fixating, it produces the smaller pupil. When the left retina is fixating, it produces the smaller pupil. It depends, not on which is the so-called better eye, as far as visual acuity or dominance is concerned, but on which of the two eyes happens to be fixating at the moment. Having found that, I wanted to see whether there was a paradigm for 123

MATHEW ALPERN 1/1 ce 1 6 _. 3 M.A. N • 150 Right eye only 0 Left eye only • Binocular -1012345 LOG1Q RETINAL ILLUMINANCE (TROLANDS) FIGURE 5 Size of the pupil of the left eye in observer M. A. (with alternating strabismus) as a function of the intensity of retinal illuminance. Left eye is always fixating. The dots show the results when the eyes were equally illuminated; the small circles, the results when the fixating eye was illuminated; and the large circles, the results when the squinting eye was illuminated. (Reprinted with permission from ten Doesschate and Alpern.2) a similar effect in normal eyes by making measurements that involve the classical experiment of retinal rivalry. Briefly, one puts overlapping con- tours onto the two retinas (for example, vertical black and white stripes seen by the left, horizontal stripes by the right). The result is that these conflicting borders cannot be unified into a perfectly consistent whole. Normal observers alternately hold clear first, say, the vertical lines, and then the horizontal lines, oscillating from one to the other. I thought that perhaps the phenomenon of turning down or turning off the retinal input to the pupil, which occurs in strabismus, would also show up on normal eyes, if they were forced into this rivalry situation. In fact, it looks as though it does. The experiment is tricky, because one must remember that in the normal eye each retina is driving both pupils to the same extent. If the subject is presented with the overlapping contours, shown in Figure 7, and the illumination on the two eyes is the same, it does not matter 124

The Pupillary Light Reflex and Binocular Interaction 4 E c i 3 O- i ' 1*1 S 6 A. LEFT EYE FIXATING B. RIGHT EYE FIXATING o RIGHT RETINA ONLY ILLUMINATED • LEFT RETINA ONLY ILLUMINATED 012345 LOG1Q RETINAL ILLUMINANCE (TROLANDS) FIGURE 6 The amount of pupil contraction evoked in alternating strabismus depends on which retina is fixating. Illuminating the fixating eye always produces the smaller pupil. 725

MATHEW ALPERN £ 8 5 7 o R.E. DOMINANT • LE. DOMINANT 10 TROLANDS 538 nm I, -VARIABLE "WHITE" 012345 LOG10 INTENSITY RETINAL ILLUMINANCE LEFT EYE(TROLANDS) FIGURE 7 The pupil size (of the left eye) evoked by different amounts of retinal illumina- tion in the normal eye during a retinal-rivalry experiment. The abscissa shows the intensity of retinal illuminance of the left retina. The circles represent measurements made while the right eye dominated, the dots, those made while the left eye dominated. Upper, results when the two retinas were always illuminated the same amount. No differences in the responses were mea- sured when the data were obtained in phase and out of phase with the illumination of the dom- inant retina. Lower, results when the illumination to the right retina was held fixed (green light of 10 trolands) while that to the left retina was varied. The pupil is smaller when the more in- tensely illuminated retina is dominant than when the more weakly illuminated retina is dominant. 126

The Pupillary Light Reflex and Binocular Interaction which retina dominates; either will give the same result. Thus, the top graph in Figure 7 shows the same pupil size, regardless of whether the pupil is photographed in or out of phase with the rivalry. No difference would be expected, because the illumination on each retina was always the same. To bring out the influence of rivalry, one must introduce a difference in the retinal illumination in the two eyes. The experiment is straightforward, and the results are illustrated in the bottom graph of Figure 7. We hold the retinal illuminance of the right eye fixed and vary the intensity of illuminance of the left retina only. The right eye views green light of 10 trolands that is not very dif- ferent from the intensity of illuminance of the low levels used to stim- ulate the left retina, and the result—as we have already seen—is that there is no difference in the size of the pupil in the left eye, whether the left or the right retina dominates. However, as we increase the in- tensity of illuminance to the left retina, so that it is appreciably greater than that of the right, it matters a good deal which retina is dominant. At any abscissa setting, the intensities of illuminance to the two retinas do not change; but the pupil is much wider when the retina receiving the weaker light is dominant than when it is not. In general, the phenomenon that we are trying to understand in strabismus seems to have a counterpart in the normal eyes' viewing a rivalry target. I do not fully understand the implications of what I have described. This is, in many ways, a very surprising result: where vision is sup- pressed by the dominance of one retina, photopupillary motion is also suppressed. One suggestion is that whatever is turning down this visual impres- sion is turning it down at the cortex. If so, it must also send separate signals down to the midbrain level to turn down the photopupillary re- sponse. Alternatively, perhaps the turning down is going on at the retina itself, in which case a separate turning down for the pupil is not neces- sary, because vision and photopupillary motion are probably mediated by the same nerve pathways at the level of the retina. Either suggestion is rather disturbing, and the evidence for each is not very impressive. What is the relationship of these ideas to the problems of reading? In the process of reading, one makes eye movements of very high velocity, saccades; in making a saccade, the visual system also undergoes a mo- mentary turning down of activity (it takes more light to produce a threshold than when the eye is immobile). It has been speculated that 727

MATHEW ALPERN the turning down of the sensitivity of vision during saccadic eye move- ments resembles the turning down of vision in retinal rivalry, and the same sort of turning down that is found in strabismus. I find it difficult to conceive a good experiment that might accumulate evidence to sup- port such ideas. I am indebted to Dr. Keith Burnes and Joel Sugar for their able technical assistance in some of the experiments discussed here. REFERENCES 1. Cooper, S., D. E. Denny-Brown, and C. Sherrington. Interaction between ipsi- lateral spinal reflexes acting on the flexor muscles of the hind-limb. Proc. Roy. Soc. Med. 1018:262-303, 1927. 2. Doesschate, J. ten, and M. Alpern. Effect of photoexcitation of the two retinas on pupil size. J. Neurophysiol. 30:562-576, 1967. 3. Hubel, D. H., and T. N. Wiesel. Binocular interaction in striate cortex of kittens reared with artificial squint. J. Neurophysiol. 28:1041-1059, 1965. DISCUSSION DR. WADE MARSHALL: Curt Richter (Johns Hopkins Med. J. 122:218-223,1968) recently reported an interesting study of alternating strabismus on 24-hr cycles. This is such a curious phenomenon that I wrote to him, and he replied that he had checked 30 cases in which it operated like clockwork, the disturbance tend- ing to improve as the patient got older. Psychiatrists tell me that they sometimes see 24-hr cycles in manic-depressive patients. DR. ALPERN: The phenomenon of periodic strabismus, in which the patient is normal some of the time and has strabismus some of the time, is very well known. I did not know that the alternation can occur with such beautiful regularity. The mechanism of turning down vision during a saccade is not a simple mat- ter. There is a large literature concerning whether saccadic suppression occurs at all. Evidence is accumulating that the output from the retina is not completely turned off during the saccade. However, the activity is clearly reduced (because of the extremely high velocity of eye movements), and there is also some reduc- 725

The Pupillary Light Reflex and Binocular Interaction tion in visibility. The best data I know of are those of Riggs and his students, Amy Schick and Francis Volkmann (J. Opt. Soc. Amer. 58:562-569, 1968). There does seem to be a genuine reduction of sensitivity that is not due to any of a number of sources of artifact. Zuber etal. (Exp. Neurol. 14:351-370, 1966) have found evidence of the suppression of pupillary light response during the saccade, but there are still some problems in understanding the phenomenon. DR. INGRAM: Is there any evidence concerning persons with constant nystagmus while reading? How do they manage to read, to get a retinal outline down? DR.ALPERN: I have no evidence on this except the subjective reports of a profes- sor of mathematics at the University of Michigan who has very poor visual acu- ity and a constant nystagmus. He reported that, when he was observing the ro- tating turntable of his phonograph, which had a speed of rotation in phase with his nystagmus eye movements, the phonograph record on the turntable appeared stationary. That is the only evidence I know of. DR. SILVER: Did you notice a difference in position or shape of the pupil? Some- times on a flash of light, the pupil becomes eccentric, and I wondered whether you had seen that. DR. ALPERN: These studies were not done with flashing lights, but with an optical system presenting a binocular Maxwellian view. The pupil was photographed in the steady state under infrared light with infrared film. There may be such ef- fects, but I would not have seen them in these experiments. DR. LINDSLEY: You referred to retinal suppression in Riggs's experiment. DR. ALPERN: I did not mean to. There is a saccadic suppression, which Riggs does not for a minute believe is in the retina. If I were to infer what Riggs thinks, sac- cadic suppression is in the cortex, and I have no way of knowing whether sup- pression is in the cortex in this other system responsible for the observations I have presented. It is a bit awkward, because multiple connections are needed, not only between the cortex and retina for vision, but through the pretectum for photopupillary motion. It is not obvious why anyone would build a railroad that way. However, if saccadic suppression were retinal or if this suppression in rivalry were retinal, then it would not be at all surprising for both vision and photopupillary response to be suppressed. That would be the logical conse- quence of the neuroanatomy, inasmuch as in every way that we can make the tests it is evident that, within the retina, the nerve pathways for vision and photopupillary motion are identical. DR. LINDSLEY: There is already some evidence. Dr. Buser and his co-workers (C. R. Soc. Biol. Paris 154:38-42,1960; J. Neurophysiol. 26:677-691, 1963) have shown that there are pathways to and from not only the tectum, but also the lenticular formation. DR. ALPERN: But why would anyone want to develop a special pathway from cortex to midbrain just to turn down the pupillary response to a binocular rivalry stimulus? 129

MITCHELL GLICKSTEIN Neural Organization in Vision I would like to review some aspects of the retina, the lateral geniculate body, and the visual cortex and the pathways that connect them. The basic descriptions of the structure of the visual system found in classical writings have guided the efforts of physiologists and psychologists in planning and interpreting experiments on visual function. I will com- ment on some of these classical teachings in the light of recent experi- ments in our own and other laboratories. Let us first look at an example of a typical vertebrate eye, that of a rhesus monkey. We know from psychophysical studies, such as those of Blough and Schrier,1'12 that monkey vision is quite similar to our own; hence, we should be able to understand a good deal about the human eye from studies on monkeys. Figure 1 is a low-power picture of the monkey eye, showing the typical results of histologic procedures for fixation, embedding, and staining of eyes. The figure also illustrates one of the problems en- countered by retinal histologists: the retina often is detached in fixa- tion or embedding, most often at the fovea. Figure 2 is a higher-power view of the monkey retina, centered at the fovea. Note that in the fovea the outer cellular layers of the retina are displaced away from the dense array of thin central cones. A short distance away from the fovea, the 130

Neural Organization in Vision cones become thicker, and they are less numerous and less densely packed. MORPHOLOGIC DIFFERENCES Monkeys and man have a mixed retina, in which there are both rod and cone receptors. Many animals have almost exclusively one or the other type of receptor. I would like to discuss some aspects qf comparative anatomy of the retina of mammals—both because it is interesting in it- self and because comparative study can show in a simple way something about our human retina. Figure 3 is a photograph of the receptors in the retina of a tree shrew, Tupaia glis. This relatively simple retina has a single row of cones arranged in a homogeneous mosaic at the back of the eye.11 There does not appear to be a fovea or any obvious center of specialization within the eye. Such a pure cone retina is a form of spe- FIGURE 1 Low-power photomicrograph of monkey (Macaco mulatto) eye. Susa fixation; embedded in low-viscosity nitrocellulose; Cason's Mallory stain; 10-JU section. 131

MITCHELL GLICKSTEIN FIGURE 2 Higher-power view of monkey fovea; same section as in Figure 1. (about X 150) cialization that is found in several species of diurnal mammal. Another aspect of specialization for living in daylight illumination is the dense black pigment that surrounds the outer segments of the cones. The pig- ment absorbs light that does not strike the outer segments of receptors. Figure 4 shows the receptor layer of a kinkajou, Potos flavus, quite a different kind of animal, which is active largely at night. Zookeepers place it in dim red illumination and often reverse the normal light-dark cycle, because in nature these animals behave very little in the daytime. Their receptors are nearly all rods, with layer after layer of rod nuclei packed below the inner segments of the rods. If one were to package these two retinas together, with rods scattered among the cones and rod nuclei arranged beneath cone nuclei, one might expect the organization seen in Figure 5, the retina of a leopard, Pantherus pardus. This kind of receptor and outer nuclear layer is typical of the basically nocturnal animals that are also capable of vision in daylight. 132

Neural Organization in Vision Another major difference in structure associated with nocturnal versus diurnal vision is the ratio of receptor nuclei to different cell types in each of the cellular layers of the retina. If we count the number of receptor nuclei and the number of cells in the inner nu- clear and ganglion-cell layers of nocturnal mammals, we see that each successive layer contains fewer cells. There is summation from many receptors onto a smaller number of cells in the inner nuclear layer, and summation in turn from inner nuclear cells onto ganglion cells. By contrast, Figure 6 illustrates the ratios in a tree shrew. There is a single line of cones, and a roughly equivalent number of ganglion cells. In the inner nuclear layer, which lies between the receptors and ganglion cells, there is a much greater number of cells than in either of the other two cellular layers. As we noted, many animals capable of vision both at night and in the FIGURE 3 Oil-immersion photomicrograph of tree shrew (Tupaia glis) retina. Bouin's fixative; em- bedding, sectioning, and staining as in Figure 1. Note dark pigment layer, part of inner segments of cones, and mottled cone nuclei, (about X 1,540) (Reprinted with permission from Glickstein.6) 133

MITCHELL GLICKSTEIN FIGURE 4 Oil-immersion photomicrograph of kinkajou (Potos flavus) retina. Histologic techniques as in Figure 1. daytime have a mixed retina. Figure 6 shows a common arrangement of such retinas: a single line of cone nuclei just adjacent to the outer limit- ing membrane, below which are many layers of rod nuclei. Figure 7 shows the same general arrangement in the monkey retina a few degrees away from the center of gaze. Note the single line of cone nuclei, and the deeper layer of rod nuclei. This figure shows several identifying characteristics of cone nuclei. Cone nuclei are larger, may stain differ- ently from the nuclei of rods, have a more diffuse distribution of chro- matin within the nucleus, and lie closer to the outer limiting membrane. FUNCTIONAL COMPARISONS I would like to draw a lesson from some of these structural considera- tions. The presence of two types of nucleus gives a laminar appearance 134

Neural Organization in Vision to the outer nuclear layer. We know that rods and cones operate under vastly different conditions of illumination; hence, there is an easy inter- pretation of the laminar pattern. The nuclei of cells that function under different lighting conditions are grouped into distinguishable sublayers. Lamination is present in other visual structures, but is not as well un- derstood: the optic tectum of birds and the lateral geniculate nucleus and the cortex of mammals all have a layered appearance. Figure 8 shows the lateral geniculate nucleus of a squirrel monkey. If one eye is removed and a sufficient amount of time elapses, a covert lamination is revealed in the geniculate.4 Atrophied cells are smaller than those seen in the lateral geniculate nucleus of a normal animal. The pattern of lamination is such that layers 1, 4, and 6 connect to the contralateral eye, and layers 2, 3, and 5 to the ipsilateral eye. Such an arrangement is present in many old and new world primates, as well as in man. FIGURE 5 Oil-immersion photomicrograph of leopard (Pantherus pardus) retina. Histologic techniques as in Figure 1. 135

MITCHELL GLICKSTEIN FIGURE 6 Photomicrograph of tree shrew (Tupaia glis) retina. Histologic techniques as in Figure 1. Although geniculate lamination may be obvious, functional interpre- tation of lamination is not. The visual fields are mapped and remapped six times in the lateral geniculate nucleus of the monkey, and at least three times in the cat. Although a beginning has been made in analysis of receptive-field differences of cells in individual layers of the genicu- late,2'17 there is no clear answer to the question of what is being segregated. PROJECTION TO CORTEX I would like now to consider the projections from the lateral geniculate nucleus to the cortex. Classical teachings would hold that there is only one representation of the visual fields in the cortex. The cortex is said to be topologically organized and unique, with neighboring points on 136

Neural Organization in Vision the retina projected onto neighboring points on the cortex, and the en- tire retina represented once and only once. The physiologic study of the problem of visual cortical projection had its origin in Wade Marshall's laboratory. Talbot and Marshall14 first began to study systematically the potentials evoked in the cat's brain by flashes of light. On the basis of their observations and those of later workers, we know that gross evoked potentials to flash can be recorded not only in area 17, the striate cortex, but also in area 18: Indeed, Doty3 showed that the evoked potentials in area 18 are of short latency and actually of higher amplitude than those in area 17. Evoked-potential studies reveal that area 18 also maps the visual fields in an orderly way; hence, Talbot13 names this region of cortex "visual II." Until recently, it was usually assumed that activity in visual II is due to indirect activa- tion via a relay from primary striate cortex. However, anatomic studies FIGURE 7 Oil-immersion photomicrograph of monkey (Macaco mulatto) eye; same section as in Figure 1. 137

MITCHELL GLICKSTEIN /;'t*i ••.•/•• • k*y•';v N •_• 2 B5&a\?JWj3SSR. ..^fe^? ' •*-".• r:'w: -^i - , ..^w—^ •," C. * •*,.•> i«s2n •i ' frvjam • I*^ is? - v^'.5*l FIGURE 8 Low-power photomicrograph of lateral geniculate nucleus of squirrel monkey (Saimiri sciureus) 1 year after enucleation of ipsilateral eye. Note transneuronal atrophy of layers 2, 3, and 5. (about X 13) 138

Neural Organization in Vision reveal that visual II of the cat receives a dense and heavy input of nerve fibers from the geniculate.7 In the cat, there appear to be at least two ordered projections from lateral geniculate to cortex, each of which maps the visual fields in parallel. INTERPRETATION What functional interpretation might there be for parallel projection from the lateral geniculate to two independent regions on the cortex? I would like to raise this question first with relation to theories of the visual function. Classical neurologic thinking is influenced heavily by the concept of a unique cortical projection of the lateral geniculate to area 17. Visual recall is thought to be a function of connections from primary visual cortex to association cortex nearby. Current theories of receptive-field organization,8'9 for example, suggest that cells in area 18 derive their complex receptive fields from simpler receptive fields of cells in area 17. However, the input from the geniculate to area 18 is a major one in the cat. This projection is both in parallel and in series with area 17. We might try to think of some aspect of vision that might be mediated by area 18. Gordon Walls16 suggested, and I think it is a good argument, that area 18 of the cat may be a visual center regulat- ing visual reflex actions—reactions of the head and neck, reactions of the eyes—that in the cat serve to maintain objects in constant view. I do not believe that there is good enough evidence that there are two independent projections of the lateral geniculate in man. I also must confess that I feel that the answer to the problems of the struc- tural basis of developmental dyslexia must be found outside the classi- cal pathways of the visual system. Animals and man can tolerate a sur- prisingly large loss of the visual cortex without major symptoms of blindness. Lashley10 showed that rats deprived of striate cortex, al- though visually impaired, were capable of solving problems based on form if 1/50 of the cells of the visual cortex remained. I confirmed that fully for the monkey. There is also the evidence of Galambos et a/.,5 who cut more than 85% of the optic tracts in cats, leaving only a tiny fraction of the visual system, and yet could train form discrimina- tion to a high level. One might argue that functional use of residual visual cortex occurs only in animals, but there are human cases that show the same thing. 139

MITCHELL GLICKSTEIN Teuber et al.15 discuss a man who had a massive peripheral scotoma with only a small region of central vision remaining after brain injury. His condition escaped detection during hospitalization; he went to work as a mail sorter, and his disability was not discovered until he thought he needed glasses several years later. Although the visual pathways are fascinating, I am not sure of thek relevance to an understanding of disabilities in reading. The evidence suggests to me that the deficit in reading disorders lies elsewhere. REFERENCES 1. Blough, D. S., and A. M. Schrier. Scotopic spectral sensitivity in the monkey. Science 139:493-494, 1963. 2. De Valois, R. L., and A. E. Jones. Single-cell analysis of the organization of the primate color-vision system, pp. 178-191. In R. Jung, and H. Kornhuber, Eds. Neurophysiologie und Psychophysik des Visuellen Systems. Berlin: Springer- Verlag, 1961. 524pp. 3. Doty, R. W. Potentials evoked in cat cerebral cortex by diffuse and by puncti- form photic stimuli. J. Neurophysiol. 21:437-464, 1958. 4. Doty, R. W., M. Glickstein, and W. H. Galvin. Lamination of the lateral genicu- late nucleus in the squirrel monkey, Saimiri sciureus. J. Comp. Neurol. 127:335- 340, 1966. 5. Galambos, R., T. T. Norton, and G. P. Frommer. Optic tract lesions sparing pattern vision in cats. Exp. Neurol. 18:8-25, 1967. 6. Glickstein, M. Organization of the visual pathways. Science 164:917-926, 1969. 7. Glickstein, M., R. King, J. Miller, and M. Berkley. Cortical projections from the dorsal lateral geniculate nucleus of cats. J. Comp. Neurol. 130:55-75, 1967. 8. Hubel, D. H., and T. N. Wiesel. Receptive fields and functional architecture in two nonstriate visual areas (18 and 19) of the cat. J. Neurophysiol. 28:229- 289, 1965. 9. Hubel, D. H., and T. N. Wiesel. Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. J. Physiol. 160:106-154, 1962. 10. Lashley, K. S. Brain Mechanisms and Intelligence: A Quantitative Study of Injuries to the Brain. New York: Dover Publications, Inc., 1963. 186 pp. 11. Samorajski, T., J. M. Ordy, and J. R. Keefe. Structural organization of the retina in the tree shrew (Tupaia glis). J. Cell Biol. 28:489-504, 1966. 12. Schrier, A. M., and D. S. Blough. Photopic spectral sensitivity of macaque monkeys. J. Comp. Physiol. Psychol. 62:457-458, 1966. 13. Talbot, S. A. A lateral localization in cat's visual cortex. Fed. Proc. 1:84, 1942. 14. Talbot, S. A., and W. H. Marshall. Physiological studies on neural mechanisms of visual localization and discrimination. Amer. J. Ophthal. 24:1255-1264, 1941. 15. Teuber, H-L., W. S. Battersby, and M. B. Bender. Visual Field Defects after Penetrating Missile Wounds of the Brain. Cambridge: Harvard University Press, 1960. 143pp. 140

Neural Organization in Vision 16. Walls, G. L. The Lateral Geniculate Nucleus and Visual Histophysiology. Berkeley: University of California Press, 1953. 100 pp. 17. Wiesel, T. N., and D. H. Hubel. Spatial and chromatic interactions in the lateral geniculate body of the rhesus monkey. J. Neurophysiol. 29:1115-1156, 1966. DISCUSSION DR. ALPERN: It is necessary to raise a word of caution regarding the concepts of interaction of rods with cones. The inhibition of rods by cones is frequently proposed as the "explanation" for a variety of psychophysical phenomena, for example, the failure of the rods to play an important role in daylight vision or to determine threshold in the early moments in the dark after a full rhodopsin bleach. Such suggestions are usually made glibly without good evidence, and we now know that the "explanation" is usually wrong. When one looks hard, the almost invariable finding is that rods, far from being inhibited by cones, are inhibited by other rods, and by other rods alone (Alpern, J. Physiol. 176:462- 472,1965). There are, however, three phenomena known to me in which hypoth- esized inhibition of rods by cones cannot yet be dismissed: (1) in the pupil reflex to light (ten Doesschate and Alpern, J. Neurophysiol. 30:571,1967, Figure 7); (2) in the color matches of the extrafoveal retina (Clark, Optica Acta 7:355-384, 1960); and (3) in the occlusion of action potential spikes of the ganglion cells in the monkey retina (Gouras and Link, J. Physiol. 184:499, 1966). Because we can functionally measure differences in the contributions of rods and cones, at the least, by their different spectral sensitivities, directional sensi- tivities, and kinetics of their visual pigments, it is not unreasonable to expect proposed cone inhibition of rods to be documented in at least these three ways. This has never been done. DR. GLICKSTEIN: Cone signals can pre-empt the ganglion cell. DR. MARSHALL: You question the retrograde connections of area 18 to the lateral geniculate nucleus. DR. DOTY: I did not find any degeneration after taking out this high-amplitude strip. There was a slight amount left that might have damaged fibers going into area 17. The amount of retrograde degeneration in man involves degeneration from area 18, occurring in a little nuclear group abutting the lateral geniculate nucleus on its median dorsal edge. The cat has a little extra area on the genicu- late—that is, on area 17, far anterior on the marginal gyrus. 141

MITCHELL GLICKSTEIN DR. GLICKSTEIN: There is a recent paper by Carey and Powell [Proc. Roy. Soc. (Biol.) 169:107-126, 1967] in which retrograde degeneration was found in the lateral geniculate after lesions were placed in area 18, so I think the retrograde studies are going to confirm the antegrade studies. DR. DOTY: I can confirm that. I have found that area 18 was just as full of de- generating material as area 17 was. DR. GLICKSTEIN: There isa little dot in the corner of one of Dr. Marshall's illustrations (J. Neurophysiol. 6:1-15, 1943). It has a short-latency evoked potential to flash similar to that in 17 or 18. This region also receives a direct lateral geniculate projection. DR. BERING: Are you referring to efferent fibers to the retina from the cortex? DR. GLICKSTEIN: Efferents to the retina from cortex do not exist. However, in birds, there is a demonstrated efferent projection to the retina from the isthmo- optic nucleus [Cowan and Powell, Proc. Roy. Soc. (Biol.) 158:232-252, 1963]. It is a small nucleus just medial and deep to the tectum, which seems to send a definite efferent projection out of the brain. Brindley and Hamasaki (J. Physiol. 184:444-449, 1966) presented histologic evidence against the view that the optic nerve of the cat contains efferent fibers. DR. MARSHALL: MacLean and associates (J. Neurophysiol. 31:870-883, 1968; R. Hassler and H. Stephan, Eds., Evolution of the Forebrain, 1966, pp. 443- 453) found that Myers's temporal loop projects visual signals into the posterior hippocampal gyrus, and certainly gives a direct entrance into a system having an influence on the emotional functions, i.e., the limbic system. The evidence of that is microelectrode findings of photically responsive units in the posterior parahippocampal cortex and the results of neuroanatomic studies, using a recent modification of the Nauta-Gygax technique for demonstrating fine cortical fibers. 142

ROBERT W. DOTY Modulation of Visual Input by Brain-Stem Systems SOME DIFFICULTIES WITH THE TOPOLOGIC MATRIX The brain creates the world of visual experience from 2 million unit- pulsed fibers in the optic nerves and does it by processing this mosaic of digital input into higher order abstractions that are smoothly contin- uous in space and time. The manner in which this is accomplished is still so far from adequate scientific explanation that the description "miraculous" is appropriate. Efforts to comprehend this process have emphasized that the retinal projection maps the world on the cortex. Recent elegant elaborations of this approach have demonstrated feature extraction and a hierarchic organization within this topologic matrix (e.g., see Hubel and Wiesel14). Although these phenomena are unques- tionably of great importance and relevance, a number of recent obser- vations cannot be incorporated readily into this basically topologic approach to the explanation of visual phenomena. The most disturbing observation is that the topologic system in the neocortex of cats (areas 17 and 18) can be fully removed without apparent detriment to such complex phenomena as pattern vision and visual estimation of distance, provided the removal takes place in the neonatal period.6 The same appears to be true in tree shrews, even as adults,17 and possibly to a lesser degree in adult cats.19 Thus, the ex- quisitely refined neural circuitry of the topologic matrix has no neces- sary relevance to pattern vision, although it would be difficult to be- 143

ROBERT W. DOTY lieve that the systems of areas 17 and 18, when present, do not function in such phenomena. It is also unlikely that higher primates could per- form pattern and distance analyses under any circumstances in the ab- sence of area 17; but that does not lessen the problem of defining the necessary attributes of a system that can function well in cats in the absence of the cortical topologic system. There is now evidence that the input can be scrambled by random destruction of 95% of the optic tracts in cats, and yet maintain some degree of pattern discrimination and a normal distribution of photically evoked potentials.12'13 That suggests that the punctate information optically focused at the retina is widely elaborated, both at the retina and in the visual cortex. Further electrophysiologic evidence of such elaboration is seen with localized electrical stimulation of the retina: weak stimuli applied even to the nasal retina can evoke potentials in the ipsilateral visual cortex that somewhat resemble photically elicited potentials.8 In addition to this extensive elaboration of the signal with- in the visual system far beyond the confines of point-to-point projec- tion, there is wide distribution of photically elicited potentials within the neocortex in both primates1 and cats (see the contribution of Buser, p. 157). In cats, many of these areas of neocortex remain responsive to photic stimulation even after total extirpation of visual cortex and de- generation of the lateral geniculate nucleus, pars dorsalis.7 Perhaps equally disturbing to any simplistic concept that the visual system operates via a mere topologic hierarchically organized matrix is the fact that visual information can be drastically modified at the first and subsequent central relays by action of the centrencephalic system. In primates, this gating function seems to be exerted predominantly at the lateral geniculate nucleus. This influence is so powerful as to sug- gest that it constitutes the raison d'etre for this thalamic relay nucleus. Because some features of the electrophysiology of the visual system of primates differ importantly from those in cats,10 I will present a brief summary of some of them before proceeding with discussion of the centrencephalic influences. 144

Modulation of Visual Input by Brain-Stem Systems GENERAL FEATURES OF THE ELECTROPHYSIOLOGY OF THE PRIMATE VISUAL SYSTEM Whereas in the cat it is relatively easy to elicit potentials throughout areas 17 and 18 by stimulation at a single electrode placement in the optic tract, it is essentially impossible in squirrel monkeys and ma- caques. Apparently, the glial investiture of fiber bundles in the primate optic tract has such a high electrical impedance that effective current spread is severely limited, and only a small group of fibers can be ex- cited from any one placement. Surgical levels of anesthesia severely de- press synaptic transmission at the lateral geniculate nucleus (LGN ), and even one-tenth the anesthetic dose of Nembutal lengthens the recovery time of synaptic transmission from 15 msec to 50-100 msec. These facts impose a number of technical difficulties in exploring the electro- physiology of the central visual system in primates. In the primate LGN, there is a clear separation, of unknown impor- tance, into magnocellular and parvocellular laminae. The large cells in the LGN are innervated by the fast fibers from the retina, and photically elicited potentials occur about 5 msec earlier in magnocellular than in parvocellular laminae.5 Conduction velocities of the fast and slow sys- tems are 15 and 6 m/sec, respectively in the optic tract and at least twice as high in the optic radiation. Paradoxically, the parvocellular ele- ments recover synaptic transmission after an excitatory volley slightly faster than do the magnocellular elements—about 8 msec versus 12 msec to full recovery in the unanesthetized macaque with chronically implanted electrodes. The first cortical synapses recover even faster. The cortical response to a single volley ascending the optic radiation is similar to that in the cat, except that there are apparently twice as many waves because of the separation in time of arrival at the cortex of the magnocellular and the parvocellular components. Electrophysio- logic and anatomic evidence suggests that the magnocellular group does not project to the foveal representation in the area striata, but this needs further investigation. One of the more gratifying things in working with macaques with permanently implanted electrodes in their visual systems is the realiza- tion that evoked potentials are commonplace and are not just a creation of the artificial conditions of electrophysiologic experiments. As the animal looks about in a normally lighted room, potentials, complete with high-frequency oscillations,9 are constantly being evoked in optic tract and striate cortex by changes in direction of gaze. 145

ROBERT W. DOTY CONTROL OF EXCITABILITY IN THE CENTRAL VISUAL SYSTEM IN PRIM A TES In most squirrel monkeys and macaques, the excitability of the striate cortex to a volley ascending the optic radiation is greatly reduced if the unanesthetized animal is in the dark. Thus, some background activity from the retina has a very important role in controlling cortical excita- bility. For unknown reasons, the effect of this activity (the Chang ef- fect) in some animals is minimal, and in most it does not change trans- mission at the LGN. Bilateral enucleation, however, has very dramatic effects at both the geniculate and the cortical levels. The time course of the changes after enucleation has not been carefully studied, but the changes are well developed within a few hours, and over the course of 2 to 3 days they reach a maximum that is maintained indefinitely. In one animal, a "world record" evoked potential was obtained: the re- sponses in area 17 to stimulation of the optic tract changed from 100 fjiV peak-to-peak preoperatively to as much as 9 mV after enucleation. This great change in excitability to afferent excitation is accompanied by the development of a very bizarre, convulsive type of electroenceph- alogram (EEG) in area striata, with 0.5- to 2-sec runs of high-voltage irregular spikes, punctuated by approximately equal periods of almost complete silence. Similar bizarre patterns have been recorded in the EEG of the human blind.4'15 It is thus apparent that, in addition to the modulation of central excitability exerted by the retina in light, com- pared with dark, there is some powerful control of the retina itself over background activity of the central system. In the normal macaque sitting in the dark while potentials are evoked at different points in area striata several millimeters apart for stimula- tion of the optic radiation, great independent variation is seen in the excitability of the several striate loci. One gets the impression that each point of the cortical mosaic is subject to a large degree of localized con- trol in the dark. Still more dramatic changes of a more global nature oc- cur with fluctuation in the attentive state of the animal in the dark. Indeed, when the monkey is relaxed and probably dozing, transmission through the LGN almost ceases for single volleys coming over the optic tract, and the cortical response is correspondingly reduced. The reduced cortical response, however, belies the true state of the cortex, in that stimulation of the optic radiation at such times produces a severalfold increase in the cortical response. Thus, as attention lags, the LGN is 146

Modulation of Visual Input by Brain-Stem Systems "shut off and the cortex "runs loose." Other data from studies of evoked potentials, as well as data from enucleation studies, similarly suggest that activity in the LGN somehow has a tonically inhibitory in- fluence on area striata. When the animal is alerted, the foregoing comparison of excitability at the cortex and the LGN is immediately reversed. The fluctuation in excitability at the LGN can be very rapid. For example, in one squirrel monkey, as the optic tract was tetanized at 30 pulses/sec, it could be seen that the response in the optic radiation often varied severalfold from one pulse to the next. In acute experiments with light barbiturate anesthesia, it is readily shown that the focus of this system that modulates LGN excitability is in the mesencephalic reticular formation (MRF). A single pulse applied to the MRF shuts off transmission through the LGN within about 8 msec and keeps it suppressed for 25-30 msec. Recovery is complete by about 50 msec, and it is usually followed by great augmentation of the re- sponse for about 100 msec. Studies show that the inhibition so ob- tained is presynaptic.16 However, the facilitory effect is prepotent, and in unanesthetized monkeys the inhibitory effect disappears (Wilson, Pecci-Saavedra, and Doty, unpublished data). Also, in some anesthe- tized preparations it is difficult to obtain the inhibitory effect, whereas the facilitory effect may always be obtained unless the animal is already at a peak of alertness (as is often true with unanesthetized macaques). When the inhibitory effect is present, it is overwhelmed by facilitation if a short train of three to six pulses at 330/sec, rather than a single pulse, is applied to the MRF. Similar facilitation can be obtained by pulse trains applied to the superior colliculus, vestibular nuclei, locus ceruleus, and other areas; but it is never as great, and the threshold is always considerably higher than for stimulation of the MRF. Facilita- tion of the LGN is unaffected by enucleation or removal of the area striata. Unlike similar effects in the cat, facilitation can be observed in photically elicited responses, as well as in responses to an electrically elicited volley. The pathway by which these facilitory influences reach the LGN from the MRF is still obscure and is probably diffuse. It does not seem to follow the brachium of the superior colliculus, and it survives ex- tensive brain-stem lesions in the area between the two structures. The significance of this control of visual input is equally obscure. P. O. Bishop (personal communication) has made the ingenious sugges- 147

ROBERT W. DOTY tion that it might be related to control of inputs lying within and out- side the horopter. However, at least some aspects of the control are probably related to eye movements. This is suggested by the extremely rapid fluctuations in LGN excitability in normal squirrel monkeys, as noted, and by the occurrence of potentials in the parvocellular portions of the LGN 50-70 msec after the occurrence of eye movements.11 Facil- itation does not occur after stimulation of the oculomotor nucleus; hence, if it is associated with eye movements, it does not arise as a di- rect feedback. The MRF stimulation that facilitates transmission through the LGN is not consistently linked to loci controlling eye movements, and the potentials that appear in LGN in response to eye movements also occur in response to tactile or auditory stimulation.11 In cats, po- tentials in the LGN also occur during eye movements and are correlated with discharge in the pons,3 discharges of neurons in the visual cortex,18 and presynaptic inhibition of the LGN.2 To gain some insight into the meaning of the modulation of LGN ex- citability by the MRF, John Bartlett in my laboratory has been studying the effect of MRF stimulation on responses of single units in the area striata of painlessly immobilized, unanesthetized squirrel monkeys. A natural stimulus, such as a moving line, is presented, to which the unit responds; 50 msec before every other presentation, the MRF is stimu- lated with six pulses, which in the nonimmobilized animal produces a mild alerting. The average of 10-15 presentations with and without the MRF stimulation shows that the unit responds faster when the visual in- put is preceded by the alerting stimulus. So far in these still preliminary experiments, this effect holds for all classes of units (e.g., those respond- ing to intensity of diffuse illumination, to movement, and to various combinations of color and movement), and seems merely to mimic the change in response obtained by increasing the intensity of the natural stimulus. Perhaps with further analysis, some difference in pattern of discharge may be discernible for change in intensity versus change in alertness, but it is not yet apparent. A major problem thus arises concerning the ambiguity of the in- formation passed on by the LGN, which may not be an accurate reflec- tion of the event in the real world. The ambiguity is, of course, com- pounded by complex interactions among abstracted qualities of the stimulus. For example, some of the units found by Bartlett display di- rectional sensitivity that is color-dependent. Thus, with white light, a unit may respond best when a line is moved toward, say, 1 o'clock, but 148

Modulation of Visual Input by Brain-Stem Systems respond still more vigorously to a red line whose intensity is about 1 log unit lower moved in the same direction. When the line is green, how- ever, the direction of movement giving the same maximal response changes to 3 o'clock. Thus, discharge of this particular unit may not distinguish between red and green lines moving in different directions, these lines moving in nonoptimal directions but preceded by MRF stimulation, or a white line moving in the optimal direction after MRF stimulation. In other words, discharge of this unit is ambiguously com- pounded from color, direction, velocity, intensity, and state of alert- ness. How, from such a melange, the nervous system can form a repre- sentation of reality is obviously still elusively miraculous. However, from such complications it can at least be inferred that the process significantly transcends a mere duplication and hierarchic extraction of the retinal image in the topologic matrix of the cerebral cortex. The work reported here was supported by U.S. Public Health Service grant NB 03606 from the National Institute of Neurological Diseases and Blindness and National Science Foundation grant GB7522X. REFERENCES 1. Bignall, K. E., and P. Singer. Auditory, somatic and visual input to association and motor cortex of the squirrel monkey. Exp. Neurol. 18:300-312, 1967. 2. Bizzi, E. Changes in the orthodromic and antidromic response of optic tract during the eye movements of sleep. Physiologist 8:113, 1965. 3. Brooks, D. C. Waves associated with eye movement in the awake and sleeping cat. Electroenceph. Clin. Neurophysiol. 24:532-541, 1968. 4. Cohen, J., L. D. Boshes, and R. S. Snider. Electroencephalographic changes following retrolental fibroplasia. Electroenceph. Clin. Neurophysiol. 13:914- 922, 1961. 5. Doty, R. W. Characteristics of central visual pathways in Macaques. Physiolo- gist 8:154, 1965. 6. Doty, R. W. Functional significance of the topographical aspects of the retino- cortical projection, pp. 228-245. In R. Jung and H. Kornhuber, Eds. Neuro- physiologie und Psychophysik des visuellen Systems. Berlin-Gdttingen- Heidelberg: Springer-Verlag, 1961. 524 pp. 7. Doty, R. W. Potentials evoked in cat cerebral cortex by diffuse and by puncti- form photic stimuli. J. Neurophysiol. 21:437-464, 1958.. 8. Doty, R. W., and F. R. Grimm. Cortical responses to local electrical stimula- tion of retina. Exp. Neurol. 5:319-334, 1962. 149

ROBERT W. DOTY 9. Doty, R. W., and D. S. Kimura. Oscillatory potentials in the visual system of cats and monkeys. J. Physiol. 168:205-218, 1963. 10. Doty, R. W., D. S. Kimura, and G. J. Mogenson. Photically and electrically elicited responses in the central visual system of the squirrel monkey. Exp. Neurol. 10:19-51, 1964. 11. Feldman, M., and B. Cohen. Electrical activity in the lateral geniculate body of the alert monkey associated with eye movements. J. Neurophysiol. 31:455-466, 1968. 12. Frommer, G. P., R. Galambos, and T. T. Norton. Visual evoked responses in cats with optic tract lesions. Exp. Neurol. 21:346-363, 1968. 13. Galambos, R., T. T. Norton, and G. P. Frommer. Optic tract lesions sparing pattern vision in cats. Exp. Neurol. 18:8-25, 1967. 14. Hubel, D. H., and T. N. Wiesel. Receptive fields and functional architecture in two nonstriate visual areas (18 and 19) of the cat. J. Neurophysiol. 28:229- 289, 1965. 15. Novikova, L. A. Effect of visual afferent impulses on formation of cortical rhythms, pp. 200-212. In Current Problems in Electrophysiology of the Cen- tral Nervous System. Moscow: Science Press, 1967. 16. Pecci-Saavedra, J., P. D. Wilson, and R. W. Doty. Presynaptic inhibition in primate lateral geniculate nucleus. Nature 210:740-742, 1966. 17. Snyder, M., W. C. Hall, and I. T. Diamond. Vision in tree shrews (Tupaia glis) after removal of striate cortex. Psychonomic Sci. 6:243-244, 1966. 18. Valleala, P. The temporal relation of unit discharge in visual cortex and activity of the extraocular muscles during sleep. Arch. Itel. Biol. 105:1-14, 1967. 19. Winans, S. S. Visual form discrimination after removal of the visual cortex in cats. Science 158:944-946, 1967. 150

EL WIN MARG A Neurologic Approach to Perceptual Problems Dr. Boynton has suggested that we might learn something more about perceptual problems by exploring the receptive-field organization of visual neurons in the human brain. This has been our goal for almost a decade, and we have developed methods for doing it in man. These methods are based on the use of microelectrodes developed for im- plantation in the brains of patients with intractable temporal lobe epilepsy.7'9 The patients in our study were in a group studied by Dr. John E. Adams of the University of California Medical Center in San Francisco. They were to undergo diagnostic and therapeutic brain sur- gery for relief of their seizures and consented to having the fine micro- electrodes added to the usual gross ones.8'10 Briefly, the method involves implantation of flexible bundles of eight microelectrodes (Figure 1) in the cortex. Each electrode is made from a 50-ju straight tungsten wire etched to a l-n tip and coated with multi- ple layers of Isonel 31. They may be included in an indwelling micro- drive that can move them from one neural unit to another in the cortex, or they may be left in a fixed cortical locus, in which case electrical pickup of single units is likely because of the large number of active neurons at the tip (Figure 2).s The electrodes are introduced through a burrhole 2 cm to one side of the inion. All we can say in identifying 757

ELWIN MARG FIGURE 1 Microelectrode bundle consisting of eight microtips loosely held together by a small segment of plastic tubing. A separate "ground" lead is deflected to one side. In this model, the tungsten wires are welded to insulated stainless-steel leads for greater flexibility and length. The splice is within the Silastic mass, which is held firmly in the burrhole by the application of additional Silastic, which forms a plug. 5 msec FIGURE 2 Oscillogram recorded from a single unit in the human visual cortex. 152

A Neurologic Approach to Perceptual Problems the cytoarchitectonic areas is that they are in the visual cortex. It is impossible to distinguish between areas 17, 18, and 19 without histo- logic confirmation, which we have never had. Dr. Richard Jung and co-workers6 first recorded single units in the visual cortex of experimental animals; Hubel and Wiesel5 and others later demonstrated the receptive-field organization of these cells. The human cortical receptive fields resemble, with some important differ- ences,8'10 those found in the monkey.4 In a series of 15 patients, we observed many units in the visual cor- tex that did not seem to respond to any stimulus we could provide, whether visual or otherwise. Their "bursty," spontaneous activity appeared independent of external influences. Other units showed a response superimposed on the spontaneous activity when targets were brought within the visual field. This electrical response was amplified until it could be heard over a loudspeaker. With this type of response, we were able to plot nine receptive fields, five in response to disks and the others to bars or lines (Figure 3). The patient fixated a mark on a large sheet of cardboard 1 meter from his eyes. We then moved bars and disks of various sizes and colors and mounted on stiff wire wands within I— 230 - 1 z j> ' * E 0 1 20 - S — \ ( } a. •y 5^ 10 .t'fzm?*. J ,. H 0 D C^-^j'.A . ^~\ F J. f A < IT'..' ( )- X ^1 /tK v_y G( ' ' } 5 10 O ^ — * cc. IL 220 2 01 0 fc30 B _i o ^ 40 f_ - a: LU 50 - . , , , , 1 50 40 30 20 10 0 10 20 30 Left Field Right Field HORIZONTAL CM FROM FIXATION POINT FIGURE 3 Receptive fields recorded from single units in the human visual cortex. See text for explanation. 153

ELWIN MARG his field of vision and listened for a response. The receptive fields were outlined in pencil on the cardboard for later measurement. Because plotting was rapid and could be repeated rapidly, and be- cause the edges of most of the receptive fields were sharp, any wander- ing of fixation could be detected and thus did not affect the size or position of the plot. Generally, the patients were very cooperative and maintained visual fixation well. Monocular and binocular fields were plotted, and we detected in these patients the various degrees of domi- nance that have been reported in laboratory animals. All the receptive fields that we plotted had some characteristics in common. The responses to black on a white background, white on a black or red background, and red, yellow, green, or blue on any con- trasting background were equal; the cells were, so to speak, all color-blind. None of the units or their plotted receptive fields could be influ- enced by a patient's efforts to change them. For example, we in- creased the audio gain until the patient could hear the pulses of a unit firing in his cortex and then asked: "Can you hear that? Can you do anything to influence it? Can you increase or decrease it, or affect it in any way?" No matter how much the patient tried to in- fluence the response, we could detect no changes. We also brought the target into the receptive field and asked: "Did you hear that sound when the target was brought here? Now, the target is withdrawn. Ima- gine it is there and try to make the same sound come from the loud- speaker." No one succeeded in doing that. None of the units appeared to be influenced by stimuli to other sensory modalities. All the plotted fields came from excitatory or "on" units, the re- sponse being superimposed on the irregular, "bursty" spontaneous rhythm. If there are any inhibitory or "off" units, they appear to be uncommon. A single unit was usually recordable for the length of a 1- to 2-hr session. At times, a unit would be recordable from one day to the next over the same microelectrode. The maximal receptive-field response occurred when the target stimulus matched the size and shape of the field. This simple method of target presentation would not be expected to elicit the response of a weak inhibitory surround, and, in fact, we could find no evidence of such a response. 154

A Neurologic Approach to Perceptual Problems One aspect of our work has direct bearing on the plasticity of the brain, a subject basic to the interests of this conference. Some units showed a progressive attenuation3 or habituation2 in their response to repeated stimulation in their receptive fields, which lasted longer than the minute or two reported for laboratory animals. The phenomenon was cortical, in that habituation of a binocular unit by the stimulation of the receptive field of one eye caused a decrement of response in the receptive field of the other eye. A new, nonmonotonous stimulus, either to vision or to another sensory modality, did not restore the response— i.e., did not produce a dishabituation.1 The receptive fields are plotted in Figure 3, in which "p" is the fixa- tion point 1 meter from the eyes. "A" is a receptive field (width, 17 min of arc) of a unit in the ipsilateral or left cortex; the left eye was dominant. All other receptive fields were contralateral to the cortex where their units lay. "B" is a monocular field, right eye, as is "C." "D" is a complex receptive field. Here, a horizontal bar would give a response anywhere within the extent of the field delineated by the horizontal dashed lines. This receptive field also showed marked habit- uation. "E" is a disk-shaped and "F" is a bar-shaped binocularly equal receptive field; "F" showed marked habituation. "G," "H," and "J" are binocularly equal receptive fields; only "G" showed strong habituation. Spatial plasticity was also observed in some receptive fields. There appears to be a systematic change with fixation distance of some re- ceptive fields of the angular diameter and the angular position relative to the fixation point. It may involve a size constancy or scaling mechanism.11'12 If we are going to investigate plasticity and other subtle functions of the brain, we should, for several reasons, do it in man. First, there appear to be species differences, even between man and monkeys, in receptive-field organization. Second, man not only cooperates with prolonged fixation and specific directed eye movements, but can also be directed to make mental efforts and to describe perceptual re- sponses to stimulation. In this way, the relationship, in terms of unit activity, between simple and complex perceptual tasks, such as read- ing and its neural basis, can be investigated. Future developments in neurosurgery may increase the number of potential volunteers for these perceptual-neurophysiologic studies by making unit recording valuable in prognosis and diagnosis of postoper- 755

ELWIN MARG ative patients with evacuated hematoma or traumatic encephalopathy. Indwelling microelectrodes that probe the neural organization of the brain should shed more light on the neurophysiologic basis of percep- tual disorders. REFERENCES 1. Horn, G. Neuronal mechanisms of habituation. Nature 215:707-711, 1967. 2. Horn, G., and R. M. Hill. Responsiveness to sensory stimulation of units in the superior colliculus and subjacent tectotegmental regions of the rabbit. Exp. Neurol. 14:199-223, 1966. 3. Hubel, D. H., and T. N. Wiesel. Receptive fields and functional architecture in two nonstriate visual areas (18 and 19) of the cat. J. Neurophysiol. 28:229- 289, 1965. 4. Hubel, D. H., and T. N. Wiesel. Receptive fields and functional architecture of monkey striate cortex. J. Physiol. 195:215-243, 1968. 5. Hubel, D. H., and T. N. Wiesel. Receptive fields of single neurons in the cat's striate cortex. J. Physiol. 148:574-591, 1959. 6. Jung, R., R. von Baumgarten, and G. Baumgartner. Mikroableitungen von einzelnen Nervenzellen im optischen Cortex der Katze: die lichtaktivierten B- Neurone. Arch. Psychiat. 189:521-539, 1952. 7. Marg, E. A rugged, reliable and sterilizable microelectrode for recording single units from the brain. Nature 202:601-603, 1964. 8. Marg, E. The jigsaw puzzle of visual neurophysiology. First International Con- ference on Visual Science. Indiana University, Bloomington, 2-4 April 1968. (to be published) 9. Marg, E., and J. E. Adams. Indwelling multiple microelectrodes in the brain. Electroenceph. Clin. Neurophysiol. 23:277-280, 1967. 10. Marg, E., J. E. Adams, and B. Rutkin. Receptive fields of cells in the human visual cortex. Experientia 24:348-350, 1968. 11. Richards, W. Apparent modifiability of receptive fields during accommodation and convergence and a model for size constancy. Neuropsychologia 5:63-72, 1967. 12. Richards, W. Spatial remapping in the primate visual system. Kybernetik 4:146-156, 1968. 756

PIERRE B USER Nonspecific Visual Projections In an interdisciplinary discussion of dyslexia, two categories of electro- physiologic data on the visual system of animal and man may be of interest. One category concerns what we consider the primary visual system, from the retina to the cortical receptive area. There is no doubt that increased knowledge of the organization and functioning of this pathway, considered at the retinal, thalamic, cortical, and collicular levels, is essential to clarify some of the major problems posed by dys- lexia (see the reports by Glickstein, p. 130, and Doty, p. 143). The other category includes data related, in one way or another, to other visual projections in the brain. The study of the primary system does not encompass all that we know about the spread of visual information through the cortex or subcortical structures. Animal experiments have clearly indicated for more than 20 years that visual projections exist in many structures outside the primary pathway. We may call these "non- primary" or, to use a more common but somewhat misleading term, "nonspecific." It is precisely when considering higher integrative processes, such as reading, that one must bear in mind the existence of nonspecific pro- jections, especially because many of the "higher functions" (pattern recognition, memorization, association processes, and so on) may de- 757

PIERRE BUSER pend on these nonspecific projections at cortical or even subcortical levels. Therefore, it seems relevant to summarize very briefly various observations in this field. Although it is necessary to concentrate on functional data, I shall discuss, first, observations on the "topographic" extent of nonspecific projections and, second, results indicating their possible role in higher integrative processes. EXTENT OF VISUAL PROJECTIONS FROM TOPOGRAPHIC DA TA Electroanatomic studies in the cat, and a few in the monkey, have indi- cated the existence of visual projections to various stations outside the primary pathway—mesencephalic, diencephalic, and cortical. It is true, however, that the extent of such projections is still a matter of discus- sion. In fact, their apparent extent depends heavily on experimental conditions, such as physiologic characteristics (e.g., level of attention or wakefulness) when free animals are considered and type of prepara- tion (for instance, anesthetic) when "acute" investigations are performed. It is beyond the scope of this brief review to consider all these topo- graphic data and discuss the anatomic problems that they raise. But it may be of interest to point out some aspects as an introduction to functional considerations. At the reticular level of the mesencephalon, the data show visual inputs, with longer latencies and far greater variability than in the pri- mary pathway, including the superior colliculus.9'19 Among extrageniculate thalamic nuclei that respond to stimulation of the retina, some belong to the "associative" group, as defined ana- tomically, such as the lateralis posterior and pulvinar. However, visual responses have also been characterized within another group of tha- lamic structures, namely those belonging to the group of nonspecific, diffusely projecting nuclei, as defined through physiologic methods by Dempsey and Morison20 and later by Jasper31 and others. Anatomically, these nuclei include intralaminar nuclei, midline nuclei, and the reticu- lar nucleus. Here again, the electrical visual activities recorded are much more variable and display longer latencies than those at stations in the primary system. At the cortical level, systematic studies in cats have revealed a variety of projections to areas outside the primary receptive field, following 158

Nonspecific Visual Projections diffuse or focal illumination of the retina. Some can be recorded from the associative cortices: in the cat, the suprasylvian gyrus, the anterior lateral area, and some areas on the medial aspect9; and in the monkey, areas in the frontal lobe5 and superior temporal gyrus.27'39 Such re- sponses exist also in the motor cortex, a fact that explains why light stimuli can elicit pyramidal discharges in some experimental conditions. Although there are indications that some of these cortical inputs origi- nate from associative nuclei, most of the cortical extraprimary projec- tions seem to depend on the nonspecific system—reticular and thalamic. The exact nature of this dependence is unknown. (See Buser and Bignall9 for discussion.) Pathways from nonspecific subcortical centers also project on the primary visual system itself at the geniculate level, from the reticular formation,2'10 and at the cortical level, from various mesencephalic or thalamic sources.8'16'18 Finally, almost all nonspecific structures throughout the brain— whether reticular, thalamic, or cortical "associative" and motor area— are multisensory, i.e., they respond almost equally well to visual, acoustic, and somatic sensory inputs.1'4 FUNCTION A L DA TA In general, the functional meaning of the nonspecific projections of visual input is largely hypothetical. Nonspecific visual projections exist in subcortical structures that have already been shown to be essential for regulating states of alertness (wakefulness and sleep). That is espe- cially true of the reticular formation of the brain stem. It thus follows that visual inputs may act on the level of wakefulness and selective at- tentiveness through these structures. It is also possible that the general level of activity of the nonspecific system may control information transfers in the primary pathway. This may occur through a facilita- tion of the visual cortex or, electrophysiologically, by addition of a multisynaptic late visual input to the earlier input transmitted through the primary channel. The importance of these various processes, which are not mutually exclusive but complementary, remains to be determined. Some nonspecific projections have their end-stations in cortical areas that have been suggested or shown to be essential to cognitive integra- 159

PIERRE BUSER tion processes. It is hypothesized that visual nonspecific information contributes to such processes of visual association and elaboration. From older theoretical schemes on complex brain operations (as ini- tially suggested by Flechsig24), one would expect associative thalamic nuclei to play the major role in conveying visual input to associative cortex. But that does not seem to be the case, inasmuch as the major input comes from nonspecific nuclei. The fact that most of the non- specific areas are basically multimodal may be important, because in- termodality associations constitute one of the prerequisites for cogni- tive or decision mechanisms. Let us now consider some facts relevant to these functional problems. Modulatory Influence A large group of experiments were undertaken to demonstrate the "modulatory" influence of projections from the nonspecific system (as elicited by electrical stimulation of its central components). Vari- ous macroelectrode7'11'23 and microelectrode16"18'33'34 studies have shown that stimulation of the reticular formation or of thalamic intra- laminar nuclei can modify the pattern of cortical visual responsiveness. Depending on the experimental procedure, these modulatory influ- ences act in various ways, although they often are facilitating rather than inhibiting. This is the case with the visual evoked potential re- corded through macroelectrodes and at the single-cell level; thus the temporal discrimination of cells in the visual cortex of cat has been shown to increase during stimulation of the nonspecific system.34 Participation in Input Another category of data emphasizes the participation of the nonspe- cific system in visual input to the visual cortex. Many studies, per- formed on animal and human subjects (in the latter case with averaging autocorrelation techniques), have suggested, or even established, that the gross evoked potential due to diffuse illumination of the retina com- prises components that are of "nonspecific" origin.9'12-15'35 These com- ponents usually appear slower and later than the others, which represent the activity of the primary input to the cortex or of the cortical neu- rons that are directly activated. It is also remarkable that these late, slow components (whatever their number and shape) are far more sensi- tive to various external or internal pharmacologic or physiologic factors, 160

Nonspecific Visual Projections and that (especially in human studies) their cortical localization goes beyond the limits of the visual area. Correlation with Attention A third group of data are related to possible correlations between the amplitude of the cortical evoked response to visual stimulation and some psychologic factors, especially variations in selective atten- tion between visual and nonvisual stimuli. Many earlier data were somewhat controversial, mainly because the experimental conditions were not defined with sufficient precision as to the direction and selec- tivity of the attentive state; recent experiments are far clearer in their conclusions.6'21'22'25'28-30'32'37 To summarize these data, which were usually (but not always) obtained from scalp recordings in human sub- jects, it can be noted that gross evoked responses to light (1) tend to decrease when stimuli are monotonously repeated, i.e., when a decline in visual attentiveness and habituation to the stimulus occurs; (2) de- crease when a subject is presented with nonvisual stimuli, i.e., when his attention shifts toward other sensory modalities; (3) are correlated with reaction time to visual stimuli, faster reactions being associated with larger responses; and (4) increase when a subject is visually attentive— not only visually searching, but also perceiving significant stimuli. Using a vigilance task, Haider et a/.28 could force subjects into one or another type of attention, visual or auditory. It appeared that, when a subject's attention was directed toward visual and not auditory stim- uli, visual responses increased, whereas the sound-evoked potential in the temporal lobe not only declined, but also developed a longer la- tency. The most conspicuous changes in these experiments involved the late components (peak latencies, 160-300 msec), i.e., those which originate from or depend for their amplitude on the nonspecific sys- tem. Such results clearly emphasize the essential role of the nonspecific system in higher elaboration and cognition of visual information at sub- cortical or, more probably, cortical stations of the visual primary system. FUNCTIONAL CORRELATIONS The last group of investigations to be dealt with herein should concern the functional role of visual projections to nonvisual cortex—i.e., to 161

PIERRE BUSER motor or associative areas or even areas that are primary for other modalities. Actually, there are very few data, if any, regarding this sub- ject. In spite of evidence, from ablation studies in animals or clinical in- vestigations in man, of impairment in visual discrimination tasks or visual cognition after lesions of some associative areas, no consistent electrophysiologic findings seem to do more than show the existence of such projections. Except for the fact that evoked responses recorded from the associative or motor cortices in animals display larger varia- tions than those recorded from primary areas under various psychologic conditions (arousal, habituation, and so on), true functional correlations remain to be established.9 In the monkey, for example, visual responses have been identified, as mentioned earlier, in the parietal-temporal- occipital area, where lesions produce major impairment of visual tasks.36 Tentative studies of electrophysiologic correlations in this field, during visual discrimination,26 have failed. Thus, more data are required if we are to understand the function of these associative projections. REFERENCES \. Albe-Fessard, D., and A. Fessard. Thalamic integrations and their consequences at the telencephalic level. Progr. Brain Res. 1:115-148, 1963. 2. Arden, G. B., and U. Sdderberg. The transfer of optic information through the lateral geniculate body of the rabbit, pp. 521-544. In W. A. Rosenblith, Ed. Sensory Communication. Cambridge, Mass.: M.I.T. Press; and New York: John Wiley and Sons, 1961. 844 pp. 3. Armengol, V., W. Lifschitz, and M. Palestini. Inhibitory influences on primary and secondary cortical photic potentials originating in the lower brain stem. J. Physiol. 159:451-460, 1961. 4. Bell, C., G. Sierra, N. Buendia, and J. P. Segundo. Sensory properties of neurons in the mesencephalic reticular formation. J. Neurophysiol. 27:961-987, 1964. 5. liitin.il!, K. E., and M. Imbert. Polysensory and cortico-cortical projections to frontal lobe of squirrel and rhesus monkeys. Electroenceph. Clin. Neurophysiol. 26:206-215, 1969. 6. Bogacz, J., A. Vanzulli, P. Handler, and E. Garcia-Austt. Evoked responses in man. II. Habituation of visual evoked response. Acta Neurol. Lat. Amer. 6:353- 362, 1960. 7. Bremer, F., and N. Stoupel. Discussion du mecanisme de la facilitation reticulaire des potentiels evoques corticaux. J. Physiol. 51:420-429, 1959. 8. Bruner, J. Afferences visuelles non-primaires vers le cortex cerebral chez le Chat. J. Physiol. 57:Suppl. 12:1-120, 1965. 9. Buser, P., and K. E. Bignall. Non-primary sensory projections on the cat neo- cortex. Int. Rev. Neurobiol. 10:111-165, 1967. 162

Nonspecific Visual Projections 10. Buser, P., and J. Segundo. Influences reticulaires somesthesiques et corticales au niveau du corps genouille lateral du thalamus chez le Chat. C. R. Acad. Sci. 249:571-573, 1959. 11. Chi, C. C., and J. P. Flynn. The effects of hypothalamic and reticular stimula- tion on evoked responses in the visual system of the cat. Electroenceph. Clin. Neurophysiol. 24:343-356, 1968. 12. Ciganek, L. Evoked potentials in man: interaction of sound and light. Electro- enceph. Clin. Neurophysiol. 21:28-33, 1966. 13. Cobb, W. A., and G. D. Dawson. The latency and form in man of the occipital potentials evoked by bright flashes. J. Physiol. 152:108-121, 1960. 14. Contamin, F., and H. P. Cathala. Reponses electro-corticales de Phomme normal e'veille a des eclairs lumineux. Resultats obtenus a partir d'enregistre- ments sur le cuir chevelu, a 1'aide d'un dispositif d'integration. Electroenceph. Clin. Neurophysiol. 13:674-694, 1961. 15. Cooper, R., W. G. Walter, and A. L. Winter. Responses to visual, auditory and tactile stimuli recorded from scalp and intracerebral electrodes with electronic averaging. Electroenceph. Clin. Neurophysiol. 14:296P, 1962. 16. Creutzfeldt, O., and H. Akimoto. Konvergenz und gegenseitige Beeinfliissung von Impulsen aus der Retina und den unspezifischen Thalamuskernen an ein- zelenen Neuronen des optischen Cortex. Arch. Psychiat. Z. Neurol. 196:520- 538, 1958. 17. Creutzfeldt, O., and O. J. Griisser. Beeinflussung der Flimmerreaktion einzelner corticaler Neurone durch elektrische Reize unspezifischer Thalamus Kerne. Proc. Inst. Congr. Neurol. Sci. 3:349-355, 1959. 18. Creutzfeldt, O., R. Spehlman, and D. Lehmann. Veranderung der Neuronak- tivitat des visuellen Cortex durch Reizung der Substantia reticularis mesencephali, pp. 351-363. In R. Jung and H. Kornhuber, Eds. Neurophysiologie und Psycho- physik des visuellen Systems. Berlin-Gottingen-Heidelberg: Springer-Verlag, 1961. 524pp. 19. Dell, P. Correlations entre le systeme vegetatif et le systeme de la vie de relation. Mesencephale, diencephale et cortex cerebral. J. Physiol. 44:471-557, 1952. 20. Dempsey, E. W., and R. S. Morison. The production of rhythmically recurrent cortical potentials after localized thalamic stimulation. Amer. J. Physiol. 135: 293-300, 1942. 21. Donchin E., and L. Cohen. Averaged evoked potentials and intramodality selective attention. Electroenceph. Clin. Neurophysiol. 22:537-546, 1967. 22. Donchin, E., and D. B. Lindsley. Average evoked potentials and reaction times to visual stimuli. Electroenceph. Clin. Neurophysiol. 20:217-223, 1966. 23. Dumont, S., and P. Dell. Facilitation reticulaire des mecanismes visuels corti- caux. Electroenceph. Clin. Neurophysiol. 12:769-796, 1960. 24. Flechsig, P. E. Anatomie des menschlichen Gehirns und Ruckenmarks auf myelogenetischer Grundlage. Leipzig: Thieme, 1920. 25. Garcia-Austt, E. Influence of the states of awareness upon sensory evoked potentials. Electroenceph. Clin. Neurophysiol. Suppl. 24:76-89, 1963. 26. Gerstein, G. L., C. G. Gross, and M. Weinstein. Inferotemporal evoked potentials during visual discrimination performance by monkeys. J. Comp. Physiol. Psychol. 65:526-528, 1968. 27. Gross, C. G., P. H. Schiller, C. Wells, and G. L. Gerstein. Single-unit activity in temporal association cortex of the monkey. J. Neurophysiol. 30:833-843, 1967. 163

PIERRE BUSER 28. Haider, M., P. Spong, and D. B. Lindsley. Attention, vigilance, and cortical evoked- potentials in humans. Science 145:180-182, 1964. 29. Hernandez-Peon, R., C. Guzman-Flores, M. Alcaraz, and A. Fernandez-Guardiola. Sensory transmission in visual pathway during "attention" in unanesthetized cats. Acta Neurol. Lat. Amer. 3:1-8, 1957. 30. Horn, G. Electrical activity of the cerebral cortex of the unanesthetized cat during attentive behavior. Brain 83:57-76, 1960. 31. Jasper, H. H. Unspecific thalamocortical relations, pp. 1307-1321. In J. Field, H. W. Magoun, and V. E. Hall, Eds. Handbook of Physiology. Section I. Volume II. Neurophysiology. Washington, D.C.: American Physiological Society, 1960. 1439 pp. 32. Jouvet, M., and J. Courjon. Variations des reponses visuelles sous-corticales au cours de 1'attention chez 1'homme. Rev. Neurol. 99:177-178, 1958. 33. Jung, R. Psychische Funktionen und vegetatives Nervensystem "a" Der schlaf, pp. 650-684. In M. Monnier, Ed. Physiologic und Pathophysiologie des Vegeta- tiven Nervensystems. Band II. Pathophysiologie. Stuttgart: Hippokrates-Verlag, 1963.960pp. 34. Komhuber, H. H. Zur Bedentung multisensorischerintegration im Nervensystem. Deutsch. Z. Nervenheilk. 187:478-484, 1965. 35. Levonian, E. Evoked potential in relation to subsequent alpha frequency. Sci- ence 152:1280-1282, 1966. 36. Mishkin, M. Visual mechanisms beyond the striate cortex, pp. 93-119. In R. W. Russell, Ed. Frontiers in Physiological Psychology. New York: Academic Press, 1966. 261 pp. 37. Spong, P., M. Haider, and D. B. Lindsley. Selective attentiveness and cortical evoked responses to visual and auditory stimuli. Science 148:395-397, 1965. 38. Steriade, M., and M. Demetrescu. Reticular facilitation of responses to acoustic stimuli. Electroenceph. Clin. Neurophysiol. 14:21-36, 1962. 39. Vaughan, H. G., Jr., and C. G. Gross. Observations on visual evoked responses to unanesthetized monkeys. Electroenceph. Clin. Neurophysiol. 21:405P- 406P, 1966. DISCUSSION DR. ALPERN: In regard to that flash-evoked response in the experimental eye fields, what is wrong about the view that this represents projection along the primary pathways to the occipital cortex and then to some associated occipital pathway? Would that not be the same way of looking at that? DR. BUSER: Yes, of course, but as long as that has not been demonstrated, my personal bias would favor a parallel system more than a system in series with the visual cortex. That is based on my experience with cats. 164

Nonspecific Visual Projections Now, can we jump from a cat to man? I do not know. That is why we have to be cautious; I am not ready to jump to that conclusion. There may be another ex- planation of this in view of Bignall's findings. DR. LINDSLEY: What about the effect of blinking on these responses? Does this effect happen the moment you get the pathway? DR. BUSER: No, this is not blinking, I am sure. Blinking would be a micromove- ment, and in that case it would not be visual information. DR.GAARDER: If you were to push your finding to the extreme, you might say that an eye jump is also an evoked response. You might say that the cortex is like a giant screen, on every part of which there is some manifestation of a re- sponse or an increment of visual input. Then, to speculate further, perhaps the input from one's sensory system is manifested in some way on almost every part of his cortex. This is going beyond anything we can see. DR. BUSER: I do not think I will go that far, because to me there is a great differ- ence between input and activity. We had some cortices that seemed to behave electrophysiologically like associative cortex, and primary cortices that behaved in quite a different way, even electrophysiologically. The difference between input and activity mentioned earlier is chiefly, I believe, in the nonprimary area. As soon as you enter the classical domain—a primary system—things change com- pletely. Even evoked responses of single units in visual areas—for example, to acoustic stimulation—are different. I do not want to say that the whole cortex has equal potential. DR. LINDSLEY: Every single muscle is connected by pathways, if you are willing to pursue this deep enough and take the connection to its logical extreme. DR. BUSER: Yes, but you forget about the long connection pathways. We do not know what these connections do when they arrive upstairs: they are projecting something down. We do not know much, if anything, about the input-output function at the corticocortical level. I am aware that this kind of study is com- pletely artificial: First, we are eliminating the entire retina, and that is probably not the best way to study visual functions; and second, I did not show macro- recording activity, because in that case I would be showing mostly older material. We are intending to look further for the maximal possible spread of informa- tion, and I ended by saying that, according to our data, some of the spread must be modulated by primary areas. DR. RiESEN: What you have said sounds a great deal like the classical views of Karl Lashley. I wonder whether you have any comment on recent reports by MacLean et al. (J. Neurophysiol. 31:870-883,1968) that he is getting visually evoked activities in monkeys. DR. BUSER: Our work was in the cat and involved only the lateral convexity on the medial wall. It is perfectly clear that we have the same sort of projections as his. 165

PIERRE BUSER DR. INGRAM: In what experimental animal situation do you get spread of evoked potentials into the so-called associative areas? DR. BUSER: I showed the mostly artificial conditions required for this kind of preparation, but I think some would agree that you can also get this result in normal animals. Some functional variations correlate with the potentials' amplitude, such as the state of wakefulness of the animal, as has been de- scribed in the older literature. The general idea is that the nonprimary responses are much more like primary responses in this area and much less sensitive to behavioral conditions like sleep, wakefulness, and so on. By comparing the single-unit response in the nonprimary central medianum nucleus of the thalamus with that in the lateral geniculate, we can show that the state of sleep or wakefulness of an animal is changing all the time. Of course, when the animal is strongly aroused, it is very difficult to record something outside the primary fields. 166

ROGER W. SPERR Y Cerebral Dominance in Perception This presentation will be concerned largely with a review of some recent evidence obtained by Dr. Ronald Saul and me4 on the effects on visual perception of congenital absence of the corpus callosum in man. The behavioral symptoms seen with congenital absence of the callosum will be compared with those produced by surgical elimina- tion of the callosum and other cerebral commissures. In the latter case, of course, the two hemispheres, which have functioned together for years with the channels for cross-communication intact, must suddenly get along without the accustomed direct lines for cross-talk. With con- genital failure, the hemispheres must get along from the very beginning without the normal cross-communication. The increased functional compensation that is achieved in the congenital situation, compared with that after surgery in the fully developed system, will give us some indication of the degree of functional plasticity that exists in the grow- ing and developing brain, beyond that seen in the fully developed brain. We have been fortunate during the last year in having available for testing and study a patient (S.K.) recently diagnosed from x-ray studies to have complete agenesis of the corpus callosum. We had seen others in the past, but this particular patient, first seen by Dr. William Wright at the Los Angeles County General Hospital, has one of those very rare 767

ROGER W. SPERRY "asymptomatic" cases; no signs of abnormality were discovered until the age of 19, when headaches developed after an acute attack of hydro- cephalus. The patient recovered quickly with treatment and returned to college, where she is currently a sophomore with an average scholastic record (C's and B's). It was our first thought that, even though no functional symptoms had been evident in ordinary behavior, such symptoms associated with loss of the corpus callosum could probably be demonstrated if we could get her into the laboratory and put her through some of the series of tests for interhemispheric integration with which we had been success- ful in recent years in demonstrating symptoms in surgical patients with cerebral commissurotomy (Figure 1). Normal subjects perform these tests without difficulty, but a group of patients of Vogel and Bogen who have undergone surgical section of the corpus callosum and ante- rior commissure either fail completely or show gross impairment.6'8 Our patient went through every test without hesitation, performing easily and apparently at normal efficiency task after task that had stopped the surgical patients. I will run through a few examples to illus- trate the kinds of functions involved, with emphasis on test perfor- mances that involve vision and language. These will help to give an idea ot the functional reorganization and compensation of the cerebral mechanisms underlying vision and language that are possible in the still developing and growing brain but not in the fully developed brain. This patient exemplifies the functional plasticity of neural maturation7 that is presumed basic to many phenomena in which early experience is criti- cal in the shaping of adult behavior. Patients deprived of the corpus callosum by surgery are unable to describe in speech or writing things that they see in the left half-field of vision. Whereas they have no trouble with items in the right half-field, they consistently report that they see nothing when stimuli are pre- sented on the left side of the vertical meridian. In these tests, the visual stimuli are flashed at 1 /10 sec or less to prevent the use of rapid eye movements to get the stimuli into the other half-field. With further testing, however, it becomes evident that these commissurotomy pa- tients are able to speak about their inner experiences from one of their hemispheres only—specifically, the left hemisphere, generally dominant in right-handed persons. Other kinds of tests show that, when the major hemisphere reports that it did not see a left-field stimulus, it speaks for itself alone, and that the stimulus was indeed seen and often well com- 168

Cerebral Dominance in Perception FIGURE 1 Drawing of experimental setup used to demonstrate subject's ability to compre- hend a stimulus confined to one visual field. prehended by the nontalking, the mute, or minor hemisphere. The mi- nor hemisphere's comprehension is expressed in nonverbal tests in which the subject selects the correct name of the stimulus or a match- ing picture or object by pointing. Ability of the subject to retrieve by touch alone objects pictured in the left half-field and emotional re- sponses to left-field stimuli also show that the left-field stimuli, about which the subject verbally disclaims any knowledge, are actually seen and recognized in the minor hemisphere. Figure 2 shows some of the relationships diagrammatically. The inner visual world of these subjects has been inferred from such evidence to be double, rather than single, with a separate conscious 169

ROGER W. SPERRY FIGURE 2 Schematic diagram of visual fields, optic tracts, and associated brain areas, show- ing left and right lateralization in man. 770

Cerebral Dominance in Perception visual awareness in each hemisphere. The right and left inner visual spheres lack their normal conscious connection. A profound absence of awareness in each hemisphere of the mental experiences of the other is consistently evident in the test data. Along with the immediate visual perception, visual memories and all kinds of mental associations of vi- sion with language, with calculation, and with other sensory modes— including touch, hearing, and olfaction—are all confined to the same hemisphere. No evidence of a similar separation and doubling of inner experience, visual or otherwise, was found in our patient with congenital absence of the callosum. She gave verbal reports from either visual half-field with no hesitation. She was able to read words and numbers across the verti- cal meridian with no sign that right and left halves were perceived sep- arately. Unlike the surgical patients, she could retrieve with either hand objects seen in either visual half-field. She could add and multiply pairs of numbers shown one in the left and one in the right half-field. In tests involving stereognosis and auditory and olfactory input, she also dis- played seemingly normal right-left cross integration. The extent to which functional compensation has been achieved in this patient's vision is perhaps best illustrated by tests that involved the rapid reading of words presented tachistoscopically, part of the word to the left and part to the right of the vertical midline. For example (see Figure 3), the letters "a b" might fall in the left field for projection to the right hemisphere, and the letters "o v e" in the right field for pro- jection to the left hemisphere. In the same list may be other words like "stout," "only," and "rely," so that the subject cannot use the left or LYE ABOVE ST OUT ON LY RE LY AL IGN FIGURE 3 List of words used to test RE IGN subject's ability to integrate stimuli coming from left and right visual ALLY fields. 171

ROGER W. SPERRY right part of the word to cue in the whole word. Even the pronuncia- tion of a syllable or two seen on one side of the vertical meridian often cannot be inferred without consideration of the rest of the word, on the other side. Both parts of the word must thus be taken into account and integrated into a proper whole. S.K. was able to read these words with the mixed right-left input promptly and as well as with the unified right or left input. We can see the extent to which callosal compensation had been achieved, but we cannot yet explain it satisfactorily. The radical dif- ference in the functional symptoms produced by congenital and by surgical separation of the hemispheres appears to be a direct reflection of the greater plasticity of the developing nervous system, compared with the fully developed system. The underlying factors responsible are very likely basic to many of the more general phenomena that illustrate the functional plasticity of neural maturation. Any insight into the underlying neural factors in this or any other situation would have im- portance for the whole field of developmental psychobiology, with wide implications extending into ethology, psychiatry, pediatrics, and other disciplines concerned with the effects of early experience on adult behavior. To account for the compensation achieved in patient S.K., we had best start by reaffirming the absence of any readily apparent explana- tion. Her normal or near-normal performance on the tests mentioned remains puzzling and difficult to account for in terms of the anatomy and physiology of known neural pathways. Although the anterior com- missure often is also absent in such cases, it appears to be present in this person, judging by her x-rays, and to be slightly enlarged, as is not un- common among cases of agenesis of the corpus callosum. The extra fibers in the anterior commissure might thus be a contributing factor. However, these extra fibers probably total less than 2% of the missing callosal system and have only indirect cross-connections for many of the functions tested. Accordingly, we must look much further for a full explanation. The hippocampal, posterior, and other cerebral commis- sures apparently are not subject to hypertrophy in callosal agenesis.3 In an asymptomatic case examined microscopically by Slager et al.,s the two hemispheres were found to exhibit an essentially typical cytoarchi- tecture, except for the missing commissure fibers. The histologic exam- inations described to date, however, have generally been rough, and they do not rule out the presence of an enrichment and elaboration of 772

Cerebral Dominance in Perception commissures and decussations at midbrain and lower levels. To account for the observed degree of functional compensation in S.K., it would seem necessary to postulate at least a functional elabora- tion of brain-stem and perhaps lower cross-connection systems. The thinness of the cerebral aqueduct and the ease with which it becomes blocked make one wonder about the presence of an atypical hyper- trophy among the midbrain centers. In addition to a purely functional reinforcement of whatever connection possibilities exist at these lower levels, there might also be purely embryonic reactions associated with the agenesis of the neocortical system that would make for an en- hanced development of the older brain-stem systems that handled higher visual, auditory, somatesthetic, and other functions before the neocortex evolved. Hypertrophy or functional reinforcement of the normally weak ipsilateral sensory projection system would go far to account for the observed compensation. The behavioral results of early, compared with late, hemispherectomy illustrate the capacity for such development in the somatesthetic system. The ipsilateral auditory and kinesthetic com- ponents are already highly developed, and their enrichment would seem to offer no problem. To attain an adequate ipsilateral function in the visual mode would pose the greatest problem, and for present purposes we can focus on the observed visual cross-integration and possible ex- planatory factors. The observed ability of S.K. to rapidly read words that fall partly in one half-field and partly in the other seems to imply that the ipsilateral half-field had become projected into the same hemisphere as the oppo- site half-field. It follows that both half-fields must be closely integrated with speech, also .in the same hemisphere. Possible anatomic pathways for this are not easy to see. One remotely possible pathway for such cross-integration is the anterior commissure. This commissure cross-connects the temporal lobes that are known to be involved in vision. The route is indirect, however, and it is unlikely that the requisite sensory information could be transmitted in sufficient detail to permit one hemisphere to read letters projected from the other hemisphere through the anterior commissure. Better possibilities probably can be found in cross-connections at midbrain levels associated with visual function in the superior collicular, pretectal, and pulvinar systems.9 Before evolutionary development of 775

ROGER W. SPERRY the neocortex, the midbrain systems carried out visual integration at the highest levels. The upper levels of midbrain vision in present mam- mals are difficult to assess because of close interaction with the neo- cortex and dependence on cortical connections. In any case, it is im- portant that visual deficits produced by neonatal removal of occipital lobes are much smaller than those produced by adult removal.10 We are speaking here mainly of the focal identifying type of vision, rather than the orientational sort more characteristic of the midbrain in the cortically intact mammal. Assuming that a latent potential for high-level focal vision in the mid- brain may be evoked by agenesis of the callosum, as well as by early cor- tical damage, there would still remain the problem of getting the refined pattern information for reading small letters across the midline, up to the cortex, and integrated with the contralateral information for a verbal readout. In our latest tests for visual cross-integration, I have used only two- and three-letter words, in an effort to avoid the variables introduced by peripheral vision. The initial scores of S.K. under these conditions show an encouraging difference between the lateral unified input and the mixed or combined right-left input, indicating that in these near-thresh- old performances that might separate midbrain from direct cortical channels she handles the left-field input better than either the right-field or the combined right-left input. The question arises of whether speech is bilateralized in this patient. Conclusive evidence is lacking. She is ambidextrous to a high degree, as is often the case in patients with agenesis of the callosum. For example, she writes mainly with her left hand, but she uses scissors better with her right hand. Some preliminary evoked-potential records taken during visual performance suggest that only her left hemisphere is active in vision. It is conceivable that speech, somatesthesis, audition, and vision are all handled in a single dominant left hemisphere. This fits with the findings on near-threshold reading of words from combined right- and left-field input. Regardless of whether speech is bilateralized or is developed only in a dominant hemisphere, there are indications that in S.K. speech has been developed at the expense of other mental faculties, such as spatial perception. After we had established the lack of functional deficits in the regular series of tests used to demonstrate cross-integrational symp- toms after commissurotomy, we started to administer other types of 174

Cerebral Dominance in Perception tests—more generalized tests aimed at the upper limits of various mental and sensorimotor faculties, regardless of lateralization, following the approach of Jeeves.1 The results to date are only suggestive, but they begin to point to subnormal function in a number of nonverbal capac- ities. S.K. fairly consistently does better on verbal tasks than on per- formance or perceptual tasks. She also draws poorly and has difficulty with geography, block design arrangements, and matching patterns—all specialties of the minor hemisphere in typical right-handed persons. At this stage, our evidence suggests a distinction between two some- what different types of cross-integrational functions mediated by the corpus callosum: those which can be compensated for in congenital absence of the corpus callosum and those for which compensation is more difficult or impossible. The kinds of functions for which com- pensation is achieved involve the more direct sensory and motor cross- integrations that were carried out at subcortical levels before evolution of the neocortex. When the neocortical system for vision, normal stere- ognosis, and other functions evolved, their cross-integrational mechan- isms also had to be moved upstairs. The kinds of cross-integrational functions for which compensation is not so easily achieved are those associated with cerebral dominance and the lateral differentiation of higher mental faculties that is peculiar to the human brain. Particularly affected are performances that depend on the mental faculties spe- cialized in the minor hemisphere. If S.K. has double speech—that is, bilateralized development of speech in both hemispheres—or if speech in her dominant hemisphere has no direct cross-communication with the other hemisphere, the results are much the same. In either case, there is a handicap in that the verbal activities cannot be so well reinforced by functions for which the minor hemisphere is normally specialized, owing to lack of cross- talk in the former case and to intrahemispheric competition in the lat- ter. These functions of the minor hemisphere seem to include spatial and orientational activities, abstract thinking, and creative mathemati- cal and geometric abilities, all of which normally would cooperate with and embellish the verbal hemisphere through the corpus callosum. It is pertinent that, according to the literature, even the least symptomatic subjects with agenesis of the corpus callosum have not been brilliant or even above normal in intellect. The current view is that they attain me- diocre intelligence at the most, although they may be highly verbal and even multilingual. 775

ROGER W. SPERRY The interpretation that loss of the corpus callosum prevents reinforce- ment by minor hemisphere functions fits also with results of some re- cent work.2 We have considerable evidence that the functions of the minor hemisphere are sufficiently different in kind from those of the major hemisphere that the two tend to conflict and interfere with each other, making it a real advantage to put the two types of activity in separate hemispheres. The minor hemisphere seems to be a specialist at configurational, spatial, synthetic, and geometric activity, whereas the major hemisphere is specialized for sequential, verbal, logical, and analytic activity. The two functions do more than compete for brain space in evolution; the basic difference in the nature of their organiza- tion means that excellence in one tends to interfere with top-level per- formance in the other. On the basis of evidence collected from patients with congenital and surgical absence of the corpus callosum, as well as from the literature, this fundamental antagonism in the nature of these modes of brain functions might be a causal factor behind the evolution of cerebral dominance and lateral specialization in the human brain. What meaning this may have for problems of dyslexia remains to be seen. One wonders whether a possible factor in dyslexia is an overly strong or extensive, perhaps bilateral, development of the verbal, major- hemisphere type of organization that tends to interfere with an ade- quate development of spatial gnosis in the minor hemisphere. The facts that general verbal capacity tends to be good in dyslexics and that the frequency of dyslexia is higher among left-handed persons would fit such an interpretation. Extra training in spatial gnosis with special ref- erence to alphabet patterns and the troublesome letters and words sub- ject to directional reversals would seem a natural approach to these problems. Original work reported here was supported by U.S. Public Health Service grant MH 3372 from the National Institute of Mental Health and by the F. P. Hixon Fund of the California Institute of Technology. REFERENCES 1. Jeeves, M. A. Agenesis of the corpus callosum—physiopathological and clinical aspects. Proc. Aust. Assoc. Neurol. 3:41-48, 1965. 2. Levy-Agresti, J., and R. W. Sperry. Differential perceptual capacities in major and minor hemispheres. Proc. Nat. Acad. Sci. USA 61:1151, 1968. 176

Cerebral Dominance in Perception 3. Loeser, J. D., and E. C. Alvord, Jr. Agenesis of the corpus callosum. Brain 91: 553-570, 1968. 4. Saul, R. E., and R. W. Sperry. Absence of commissurotomy symptoms with agenesis of the corpus callosum. Neurology 18:307, 1968. (abstract) 5. Slager, U. T., A. B. Kelley, and J. A. Wagner. Congenital absence of the corpus callosum. New Eng. J. Med. 256:1171-1176, 1957. 6. Sperry, R. W. Mental unity following surgical disconnection of the cerebral hemispheres, pp. 293-323. In The Harvey Lectures. Series 62 (1966-67). Harvey Society of New York. New York: Academic Press, 1968. 364 pp. 7. Sperry, R. W. Plasticity of neural maturation, pp. 306-327. In M. Locke, Ed. The Emergence of Order in Developing Systems. 27th Symposium of the Society for Developmental Biology. Supplement 2. New York: Academic Press, 1968.350pp. 8. Sperry, R. W., M. S. Gazzaniga, and J. E. Bogen. Function of neocortical com- missures: syndrome of hemisphere deconnection. In P. J. Vinken and G. W. Bruyn, Eds. Handbook of Clinical Neurology. Amsterdam: North-Holland, 1969. 9. Trevarthen, C. B. Two mechanisms of vision in primates. Psychol. Forsch. 31:299-348, 1968. 10. Wetzel, A. B., V. E. Thompson, J. A. Horel, and P. M. Meyer. Some conse- quences of perinatal lesions of the visual cortex in the cat. Psychon. Sci. 3:381-382, 1965. DISCUSSION DR. INGRAM: I suggest that the dyslexic individual is faced with several problems in addition to the spatial relationships that he cannot visualize. He also requires some teaching in terms of auditory concepts, because this function is in the hemisphere that is functioning best and that should be concentrated on. DR. MASLAND: Dr. Ingram has suggested that there is a very important additional element in the problem of reading. It is not merely a matter of recognizing the shape of an object, but the establishment of an association between the shape of an object and a verbal sound information element. That function has already been established in the left hemisphere of the average child, and it seems to me that the fundamental problem of the dyslexic child, particularly the child who is having letter reversals, is the necessity of establishing a relationship between a visual spatial function, which is most likely mediated in the right hemisphere, and a language function, which has already been established in the left hemi- sphere. The problem is to develop techniques whereby spatial functions are dis- sociated from the language functions of the left side. 777

ROGER W. SPERRY DR. INGRAM: There is a possibility that this dyslexic schoolchild would be taught by the so-called "look and say" method. I think that would be disastrous. This child has to be taught to relate the visual symbol, as Dr. Masland says, to the spoken syllable. I think by the stage of learning to read the child is probably able to recognize the visual symbol, but not to relate it to the auditory symbol. Therefore, you have to work with a phonic approach, and establish the phonic relationships. I am trying to point out that it is difficult to short-circuit a func- tion that you think is not there. DR. SPERRY: I was referring specifically to the perception of spatial relations during early learning of reading and writing, when letters and words tend to be reversed, and did not mean to imply a general application to reading aloud and to all forms of dyslexia. DR. MASLAND: Maybe it is unwise to generalize, but for a person to learn to read, obviously he has to have the ability to see, he has to be able to analyze and rec- ognize the material being seen, and he must be able to associate that object with an auditory symbol. DR. BERING: The auditory counterpart is not necessary. It has been brought up here that people can learn to read without it if the hemispheres are intact. There are auditory and spoken relationships with reading, but neither is absolutely necessary. 178

ATTENTIONAL AND PERCEPTUAL MECHANISMS