Open Access
Special Issue  |   February 2019
IMI – Report on Experimental Models of Emmetropization and Myopia
Author Affiliations & Notes
  • David Troilo
    SUNY College of Optometry, State University of New York, New York, New York, United States
  • Earl L. Smith, III
    College of Optometry, University of Houston, Houston, Texas, United States
  • Debora L. Nickla
    Biomedical Sciences and Disease, New England College of Optometry, Boston, Massachusetts, United States
  • Regan Ashby
    Health Research Institute, University of Canberra, Canberra, Australia
  • Andrei V. Tkatchenko
    Department of Ophthalmology, Department of Pathology and Cell Biology, Columbia University, New York, New York, United States
  • Lisa A. Ostrin
    College of Optometry, University of Houston, Houston, Texas, United States
  • Timothy J. Gawne
    School of Optometry, University of Alabama Birmingham, Birmingham, Alabama, United States
  • Machelle T. Pardue
    Biomedical Engineering, Georgia Tech College of Engineering, Atlanta, Georgia, United States31
  • Jody A. Summers
    College of Medicine, University of Oklahoma, Oklahoma City, Oklahoma, United States
  • Chea-su Kee
    School of Optometry, The Hong Kong Polytechnic University, Hong Kong, SAR, China
  • Falk Schroedl
    Departments of Ophthalmology and Anatomy, Paracelsus Medical University, Salzburg, Austria
  • Siegfried Wahl
    Institute for Ophthalmic Research, University of Tuebingen, Zeiss Vision Science Laboratory, Tuebingen, Germany
  • Lyndon Jones
    CORE, School of Optometry and Vision Science, University of Waterloo, Ontario, Canada
  • Correspondence: David Troilo, SUNY College of Optometry, State University of New York, 33 West 42nd Street, New York, NY 10036, USA; dtroilo@sunyopt.edu
Investigative Ophthalmology & Visual Science February 2019, Vol.60, M31-M88. doi:https://doi.org/10.1167/iovs.18-25967
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      David Troilo, Earl L. Smith, Debora L. Nickla, Regan Ashby, Andrei V. Tkatchenko, Lisa A. Ostrin, Timothy J. Gawne, Machelle T. Pardue, Jody A. Summers, Chea-su Kee, Falk Schroedl, Siegfried Wahl, Lyndon Jones; IMI – Report on Experimental Models of Emmetropization and Myopia. Invest. Ophthalmol. Vis. Sci. 2019;60(3):M31-M88. doi: https://doi.org/10.1167/iovs.18-25967.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

  The results of many studies in a variety of species have significantly advanced our understanding of the role of visual experience and the mechanisms of postnatal eye growth, and the development of myopia. This paper surveys and reviews the major contributions that experimental studies using animal models have made to our thinking about emmetropization and development of myopia. These studies established important concepts informing our knowledge of the visual regulation of eye growth and refractive development and have transformed treatment strategies for myopia. Several major findings have come from studies of experimental animal models. These include the eye's ability to detect the sign of retinal defocus and undergo compensatory growth, the local retinal control of eye growth, regulatory changes in choroidal thickness, and the identification of components in the biochemistry of eye growth leading to the characterization of signal cascades regulating eye growth and refractive state. Several of these findings provided the proofs of concepts that form the scientific basis of new and effective clinical treatments for controlling myopia progression in humans. Experimental animal models continue to provide new insights into the cellular and molecular mechanisms of eye growth control, including the identification of potential new targets for drug development and future treatments needed to stem the increasing prevalence of myopia and the vision-threatening conditions associated with this disease.

1. Introduction
Emmetropization refers to the developmental process that matches the eye's optical power to its axial length so that the unaccommodated eye is focused at distance. Investigations using animal models have informed our understanding of the role of vision in postnatal eye growth, the mechanisms and operating characteristics of emmetropization, and the development of refractive errors (myopia, where the eye is typically too long for its optical power; and hyperopia, where the eye is too short for its optical power). Animal models have established the existence of visual regulation of eye growth and refractive development as well as local retinal control of eye growth. They have also revealed biochemical signaling cascades that transduce visual stimuli related to the sign of defocus into cellular and biochemical changes in the retina, which, in turn, signal changes in the retinal pigment epithelium (RPE), choroid, and eventually sclera, leading to altered eye growth and changes in refractive state. These studies provide a framework for the development of optical and pharmacologic treatments that can be used to effectively reduce the prevalence and progression of myopia, which has become a major public health concern.1 
In this paper, the findings of investigations using experimental animal models to study emmetropization and myopia development are reviewed. The contributions that studies with experimental animal models have made to understanding the mechanisms of emmetropization, the development of myopia, and new treatments to reduce myopia progression are summarized. Current models of eye growth control, areas of investigation and major findings, and frameworks for the development of new and effective treatments for myopia are described. 
2. Animal Models Commonly Used in Studies of Emmetropization and Myopia
Experimental models of myopia and the visual regulation of eye growth have been demonstrated in a wide variety of species from primates to invertebrates, including macaque and marmoset monkeys, tree shrews, guinea pigs, mice, chickens, fish, and squid. All of these species (with the exception of squid) have been shown to develop myopia in response to visual form deprivation (see Section 3.2), compensate for optically imposed myopic or hyperopic defocus by regulating axial length (see Section 3.4), and recover from the induced refractive error when form deprivation or optical defocus is removed (see Section 3.3). Even though the squid model is the least well-characterized, squid eye growth responds to improve focus under imposed visual conditions.2 Considering that all these varied species possess visually guided eye growth despite differences in ecology, ocular anatomy, visual function, and visual acuity, these results suggest that visual regulation of eye growth is a fundamental property of the camera-type eye, that it may have evolved more than once, and the mechanisms in vertebrates are evolutionarily conserved. From an experimental perspective, each species provides unique advantages to study the mechanisms of visually guided eye growth and key signaling pathways that regulate refractive eye development across species; however, anatomical and physiological differences must be taken into account when interpreting and translating results to humans. 
General retinal cellular organization and neural signaling circuitry are highly conserved among vertebrate species3,4; however, there are significant variations between species. Diurnal primates, like humans, have a single fovea for high acuity, whereas other species may be multifoveal, or have an area centralis or visual streak, which are retinal areas with higher photoreceptor and ganglion cell density. The visual photopigment types underlying color vision also vary between species, as does retinal vascular anatomy. Table 1 summarizes structural similarities and differences between the retinas of the most commonly used experimental species. 
Table 1
 
Retinal Differences in Species Used for Myopia Models
Table 1
 
Retinal Differences in Species Used for Myopia Models
There are also significant species differences in the mechanisms and amount of accommodation, which regulates the dioptric power of the eye and may be indirectly involved in myopia development through its effects on retinal defocus. In many species, including human, accommodation is achieved by changing the power of the crystalline lens by contraction of the ciliary muscle, whereas in other species it is achieved by moving the lens.5 Changes in corneal power have also been observed in some species.68 
For another recent review of different species used for experimental studies of emmetropization and myopia, see Schaeffel and Feldkaemper.9 
2.1 Comparative Ocular Anatomy and Visual Physiology of Animal Models
2.1.1 Nonhuman Primates
Macaque monkeys were used in the original studies showing form-deprivation myopia (FDM) and visual influences on eye growth.10,11 Since then, both Old World (rhesus macaque – Macaca mulatta) and New World (common marmoset – Callithrix jacchus) monkeys have been used for myopia research. Both species have foveal retinas, eyes that are optically scaled down versions of human eyes, and visual physiology which is essentially identical to that of humans.1215 The rhesus monkey retina is most similar to the human. It is rod-dominated (rod to cone ratio ∼20:1) with a cone-dominated fovea and possesses three cone types, with short-, middle- and long-wavelength sensitivities, in addition to rods.16 The fovea provides visual acuity of approximately 44 cyc/deg.13,14 The marmoset retina is cone-dominated with a well-developed fovea.12,15 The marmoset retina contains rods as well as cones, which exhibit a polymorphism of visual pigments, in which three photopigments are in the middle- to long-wavelength range, with peak sensitivities at 543, 556, and 563 nm.17 With this polymorphism, some animals are dichromatic (males and some females) while others are trichromatic (females). Visual acuity in marmosets is approximately 30 cyc/deg.12,18 Both rhesus and marmoset monkeys have vascular inner retinas with a foveal avascular zone. In rhesus monkeys, the optic nerve head contains a collagenous lamina cribrosa, closely resembling that in humans. In marmosets, the optic nerve also has a collagenous lamina cribrosa with characteristic sieve-like structure.19 
The accommodative system in rhesus monkeys and marmosets is closely related to that in humans and other primates.20,21 The ciliary muscle and its pharmacology are similar to those of humans allowing cycloplegia (paralysis of accommodation) to be produced with muscarinic antagonists as in humans. Juvenile macaques and marmosets have an accommodative response of at least 20 diopters (D).22,23 In previous studies, accommodation was successfully stimulated in awake-behaving marmosets and measured with photorefraction, showing stimulus response slopes similar to humans.22 Additionally, rhesus monkeys have been shown to develop presbyopia at a similar rate as humans, once corrected for life span.20,24 
Low availability due to low reproduction rate in macaques is a challenge, and the eyes and visual systems in macaques develop more slowly than in other species commonly used for myopia research. Marmosets give birth to twins or triplets approximately twice a year and are sexually mature at approximately 18 to 24 months.25 
2.1.2 Tree Shrew
Tree shrews belong to the order of Scandentia, which are closely related to primates. They are among the first species shown to develop FDM26 and have since been used by several laboratories for myopia research. Tree shrews have a cone-dominated retina with rods comprising approximately 14% of the photoreceptor population.27 Tree shrews do not have foveas, but the retina has an area centralis,2729 which provides a visual acuity of approximately 2.4 cyc/deg.30,31 Tree shrews are dichromatic, with short- and long-wavelength sensitive cones.32 The tree shrew inner retina is vascular. The optic nerve contains a collagenous lamina cribrosa with radially oriented laminar beams.33 
Tree shrew eyes have relatively large crystalline lenses and relatively small vitreous chambers compared with primates. They do not appear to exhibit substantial accommodation31,34; however, when stimulated with carbachol, tree shrews can accommodate up to 8 D.35 
Tree shrews typically give birth to two small litters a year. 
2.1.3 Guinea Pig
Guinea pigs are diurnal rodents, which have been increasingly used as a model for myopia research. Guinea pigs develop FDM and can compensate appropriately for both imposed myopic and hyperopic defocus.36,37 Guinea pigs are dichromatic. In addition to rods, the retinas of guinea pigs include middle- and short-wavelength-sensitive cones, which occupy superior and inferior areas of the retina, respectively, while the transition zone contains both cone types and cells with both pigments.38 Guinea pigs do not have a fovea; however, the retinas have a visual streak,39 which provides a visual acuity of approximately 2.7 cyc/deg. The guinea pig retina is avascular, having the retinal blood supply provided solely by the choroidal circulation. Because retinal nutrients must diffuse from the choroid, the retina is typically thinner than in animals possessing inner retinal vasculature.40 The optic nerve contains a collagenous lamina cribrosa with connective tissue beams.41 
Guinea pig eyes have relatively large crystalline lenses and relatively small vitreous chambers compared with primates.42 Guinea pigs do not appear to have an active accommodative response43; however, approximately 5 D of accommodation can be elicited pharmacologically in juvenile animals.44 
Guinea pigs are able to breed year-round and grow rapidly, which allows large-scale studies. 
2.1.4 Mouse
Mice are nocturnal rodents, which have been increasingly used for myopia research in recent years.4550 Although mice are classified as nocturnal animals, they are also active during the day,5153 photopic visual input plays an important role in their refractive development,54 and behavioral and functional studies suggest that vision is critical for accurate spatial navigation.5558 Mice develop FDM and respond appropriately to imposed hyperopic and, to some extent, myopic defocus.46,59 Mouse myopia is axial in nature and has features of human myopia.46 Mice are dichromatic and the organization of the mouse retina is similar to that of other mammals.3,4,60 Similar to guinea pigs, the mouse retina includes middle- and short-wavelength–sensitive cones, which occupy superior and inferior areas of the retina, respectively, while high levels of both photopigments are expressed in the transition zone.6163 The mouse retina does not possess a fovea, but a visual streak has been located just temporal of the optic disc,3,64,65 which provides an upper limit for visual acuity of approximately 1.4 cyc/deg.66 The mouse eye possesses an inner retinal vasculature with radially oriented vessels. The optic nerve contains a lamina cribrosa composed of glial cells.67 
Mouse eyes have large crystalline lenses and relatively small vitreous chambers compared with primates.68,69 Mice are not known to possess lenticular accommodation.70,71 
Mice breed year-round, produce large litters, and grow rapidly. Because of the availability of a large number of inbred and gene-targeted strains and well-established techniques for genome manipulation, the mouse has become a popular model for advanced genetic and molecular genetics studies of gene–environment interaction in refractive development and myopia. 
2.1.5 Chicken
Studies with chicks were among the first to show that visual experience can modulate eye growth and refractive development.72 Since then, chicks have been used extensively because they are easy to obtain, are visually precocial, and develop rapidly. Most of the chick studies are performed on different strains of White Leghorn chicks. Breed and strain differences have been found, indicating genetic differences in the susceptibility to visual experience in eye growth control,73,74 which have been confirmed with selective breeding.75 Chicks develop FDM and rapidly compensate for both imposed myopic and hyperopic defocus (see Section 3 below). 
The chick retina contains rods, four single cone photoreceptors, and one double cone photoreceptor.76,77 The cones contain oil droplets, which act as long-wavelength pass filters cutting off shorter wavelengths.78 Chick photoreceptors are present in a 3:2 cone-to-rod ratio with the majority of rods located in the inferior region of the retina and the majority of blue and violet cones in the superior retina.76,77 Chick retinas do not have a fovea, but have a largely rod-free area centralis that provides a visual acuity of approximately 7 cyc/deg.7983 Optical coherence tomography (OCT) imaging shows that the retina is thickest in the region of the area centralis.84 The chick inner retina is avascular and is supplied with oxygen and nutrients by the pecten oculi, which is a vascular structure continuous with the choroid and projecting into the vitreous chamber. The pecten extends from the optic nerve head and oscillates in the vitreous with saccadic eye movements to facilitate ocular perfusion.85 The optic nerve possesses a poorly formed lamina cribrosa with sparse, longitudinally oriented connective tissue bundles.19 Chick retina, unlike mammalian retina, receives efferent input from the brain (centrifugal inputs) and unique axon-bearing amacrine cells not found in mammals.86 
Chick eyes have small crystalline lenses and relatively large vitreous chambers.87 The chick possesses an active accommodative system with approximately 25 D of amplitude.88 Accommodation is achieved through changes in both corneal and lens surface curvatures, with the cornea being responsible for roughly 40% and the lens for 60% of the dioptric change.6,8,89 The ciliary muscle is responsible for both corneal and lenticular changes during accommodation.8 Unlike mammals, the chick ciliary muscle is striated and contains nicotinic acetylcholine receptors.90 Therefore, cycloplegia in chicks requires nicotinic antagonists. 
Chicks, like other birds and most vertebrates (except most mammals), have a cartilaginous and fibrous sclera (see Section 3.5.4) with scleral ossicles associated with the cartilaginous sclera in the anterior segment of the eye.5 
The circadian regulatory system in chicks is highly developed and possesses a number of differences from that of mammals,9194 which may make refractive development of the chick eye more sensitive to changes in light cycle, such as constant light.9599 For more on light cycles and circadian rhythms see Section 4 below. 
2.1.6 Fish
Fish eyes grow throughout life, and have been shown in several species to be affected by changes in the visual environment.100104 Teleost fish develop FDM and compensate for imposed hyperopic and myopic defocus.101,103 Fish from several species also compensate for defocus due to chromatic aberrations and effectively recover from induced refractive errors when visual form deprivation or imposed defocus is discontinued.100,101,103,104 Methods for accurately measuring zebrafish eyes and vision have been developed102,105109 
Zebrafish have tretrachromatic vision, with UV, short-, middle-, and long-wavelength–sensitive cones.110 Retinal morphology and stratification are similar to the mammalian retina.111 Ganglion cell counts show a region of higher density at the area centralis, which provides a visual acuity of approximately 0.7 cyc/deg.108,109,112 The zebrafish eye possesses an inner retinal vasculature that branches from the optic artery.111,113 The optic nerve head is comprised of an astroglial lamina cribrosa.114 
In zebrafish, as in other aquatic animals, the relative refractive power of the lens is higher than that of terrestrial animals because corneal power is neutralized in water. The zebrafish crystalline lens is spherical and is not known to accommodate115; however, other teleost fish are known to accommodate by moving the lens.116,117 The zebrafish is a promising model for studies of visually guided eye growth because of its fast development and the availability of well-established protocols for genome manipulation and large repository of gene-targeted mutants.118120 
2.2 Schematic Eyes
Paraxial schematic eyes have been developed for the following species used for experimental myopia research: chick,81,87 mouse,69 guinea pig,42 tree shrew,31 marmoset,12 and macaque.121,122 For reviews of the comparative optics of eyes of vertebrates see several papers by Hughes.123125 For a recent human schematic eye, see Atchison and Thibos.126 
2.3 Relative Ocular Maturation Rates
In many respects, the emmetropization process is essentially completed in a relatively short period of time in all species (see Section 3.1). On average, marmosets and macaques exhibit relatively stable refractive errors at approximately 2 and 5 months of age, respectively. In tree shrews, guinea pigs, mice, and chicks, refractive state stabilizes after approximately a few weeks of visual experience. However, the vision-dependent mechanisms responsible for emmetropization remain active well into early adult life127130 and help to maintain the optimal refractive error and ensure that an animal remains isometropic. 
Because the time required to achieve the target refractive state for a given animal depends in part on the magnitude of its initial ametropia, the relative rates of ocular axial elongation provides a reasonable interspecies metric for comparing the time course of emmetropization and refractive development. The top plot in Figure 1 illustrates axial length plotted as a function of age for individual rhesus monkeys. The solid red line, which is the best-fitting five-parameter, double exponential function that rises to a maximum value, provides a reasonable description of ocular elongation for individual macaque eyes (thin lines). The vertical dashed line indicates the age at which the “normal” eye completed half of its total axial growth. 
Figure 1
 
Eye growth in experimental animal models. (A) Axial length plotted as a function of age for individual rhesus monkeys.14,342 The symbols represent cross-sectional data; the thin black lines represent longitudinal data for individual monkeys. The solid red line shows the best-fitting double exponential function. The horizontal and vertical dashed lines show half-maximum axial length and the age when it was obtained, respectively. (B) Relative axial length changes for different species. The same double exponential function was used to fit the data for each species (humans, black line; rhesus monkey14,342,468 red line; tree shrew31,129,467 green line; mouse59,69,742,811813 blue line; guinea pigs42,466,814,815 cyan line; chicks130 410,481,816,817 pink line; marmoset153,202,343 dark red line) and the functions were normalized to the total change in axial length that occurred from birth or eye opening and adulthood. For mice and tree shrews the abscissa represents days of visual experience.
Figure 1
 
Eye growth in experimental animal models. (A) Axial length plotted as a function of age for individual rhesus monkeys.14,342 The symbols represent cross-sectional data; the thin black lines represent longitudinal data for individual monkeys. The solid red line shows the best-fitting double exponential function. The horizontal and vertical dashed lines show half-maximum axial length and the age when it was obtained, respectively. (B) Relative axial length changes for different species. The same double exponential function was used to fit the data for each species (humans, black line; rhesus monkey14,342,468 red line; tree shrew31,129,467 green line; mouse59,69,742,811813 blue line; guinea pigs42,466,814,815 cyan line; chicks130 410,481,816,817 pink line; marmoset153,202,343 dark red line) and the functions were normalized to the total change in axial length that occurred from birth or eye opening and adulthood. For mice and tree shrews the abscissa represents days of visual experience.
The bottom plot in Figure 1 compares the time course for axial elongation between humans (black line) and the experimental species commonly used in refractive error research. The same double exponential functions were fit to the axial growth data for each species. The functions were normalized to indicate the relative change in axial elongation as a function of age. The age at which half the total axial growth (t0.5 values) is obtained encompasses the period of most rapid growth in most species (i.e., the period of rapid emmetropization) and appears to be a reasonable measure of the relative rates of ocular growth between species. Accordingly, tree shews (t0.5 = 17 days of visual experience, green line) and mice (t0.5 = 22 days, blue line) exhibit the fastest relative rates of axial elongation (note: the eyes of tree shrews and mice do not open until ∼20 and 14 days of age, respectively; consequently, for tree shrews the abscissa in Figure 1 represents days of visual experience). The t0.5 values for chicks (41 days, pink line) and marmosets (40 days, dark red line) are approximately twice as long as those for mice and tree shrews. The t0.5 values for guinea pigs (cyan line) and rhesus monkeys (red line) are approximately six times longer than mice and tree shrews and approximately one-third the rate calculated for humans. The similarity of the time constants for guinea pigs and rhesus monkeys is somewhat surprising and due in large part to the fact that guinea pig eyes continue to increase in axial length at a relatively fast rate well into adult life after a stable refractive state error has been achieved. 
3. Visual Regulation of Eye Growth
It was once thought that the normal growth of the eye and the development of refractive errors were largely regulated by genetics.131133 However, primarily as a result of research involving animal models, it is now widely accepted that both genetic and environmental (visual) factors are involved in refractive development and particularly in the genesis of common refractive errors, such as juvenile-onset myopia. Consequently, controlling the visual conditions that affect eye growth offers both noninvasive and economic means to reduce myopia progression. In this respect, probably the most fundamental discovery from animal studies is that ocular growth and refractive development are regulated by visual feedback associated with the eye's effective refractive state. In particular, experimental studies over more than 40 years, using a variety of animal models, including nonhuman primates, leave little doubt that retinal defocus carries specific visual information used to regulate the growth and refractive state of the eye. This idea is supported by the following four primary observations described below: (1) emmetropization, (2) the phenomenon of FDM, (3) the recovery from FDM, and (4) compensation for optically imposed defocus. 
3.1 Evidence for Visual Regulation of Eye Growth: Emmetropization
At birth, or at the onset of visual experience, the eyes of the majority of animals used in refractive error research exhibit significant refractive errors and substantial individual differences in refractive error. These refractive errors diminish during early postnatal development as both eyes of individual animals grow in a coordinated fashion toward what is presumed to be the ideal refractive state for a given species through a process called emmetropization. 
Emmetropization proceeds in a qualitatively similar manner in most of the commonly used laboratory species. For example, as illustrated in Figure 2, which shows data for rhesus monkeys (top row), marmosets, tree shews, guinea pigs, and chicks (bottom row), neonates typically, but not always, exhibit substantial hyperopic errors that exceed the potential measurement artifacts associated with small eyes (red lines)134 and over time these eyes grow in a manner that reduces the degree of hyperopia. The fact that some neonates are myopic and exhibit relative hyperopic shifts during emmetropization emphasizes that the observed refractive changes are not simply a consequence of changes in the magnitude of the small eye hyperopic artifact that takes place as the eye grows. 
Figure 2
 
Emmetropization in experimental animal models. The left column shows refractive errors for (A) rhesus monkeys,342,468 (B) marmosets,153,343 (C) tree shews,31,129,467 (D) guinea pigs,42,814,815 and (E) chicks175,410,481,816,818 plotted as a function of age (or days of visual experience for tree shrews). Longitudinal data from individual animals are shown as solid lines without symbols. Cross-sectional data for individual animals are represented by individual data points. Symbols connected by lines show mean data (typically cross-sectional) from a given study. The solid red lines represent the small eye artifact associated with common measurement techniques like retinoscopy. The middle and right columns contain refractive error frequency distributions obtained near birth/hatching and at ages when refractive development was relatively stable, respectively. The red lines in the histograms show the Gaussian distributions calculated using the mean and standard deviations of the data.
Figure 2
 
Emmetropization in experimental animal models. The left column shows refractive errors for (A) rhesus monkeys,342,468 (B) marmosets,153,343 (C) tree shews,31,129,467 (D) guinea pigs,42,814,815 and (E) chicks175,410,481,816,818 plotted as a function of age (or days of visual experience for tree shrews). Longitudinal data from individual animals are shown as solid lines without symbols. Cross-sectional data for individual animals are represented by individual data points. Symbols connected by lines show mean data (typically cross-sectional) from a given study. The solid red lines represent the small eye artifact associated with common measurement techniques like retinoscopy. The middle and right columns contain refractive error frequency distributions obtained near birth/hatching and at ages when refractive development was relatively stable, respectively. The red lines in the histograms show the Gaussian distributions calculated using the mean and standard deviations of the data.
A hallmark of emmetropization is the systematic reduction over time of the intersubject differences in refractive error.135 The histograms in the middle and right columns of Figure 2, show, respectively, the distributions of refractive errors obtained early in the emmetropization process and at ages when the average refractive errors have stabilized. For all five of the represented animal species, the average refractive errors obtained later in life were less hyperopic than those obtained early during the emmetropization process, and the standard deviations of the means were substantially smaller. The optimization of refractive errors and the decrease in the between-subject variability is evidence that early ocular growth is regulated by visual feedback in a way that eliminates these early refractive errors. The fact that the course of ocular growth and refractive development become unpredictable when animals are reared in the dark 24 hours a day indicates that vision is important in the regulation of normal refractive development.54,136138 
Emmetropization is often thought of as the visual regulation of eye growth and not necessarily growth toward emmetropia. The target refractive state, or set point, for emmetropization, varies between experimental species. Like in humans, the eyes of rhesus monkeys, tree shrews, and chicks grow toward low amounts of hyperopia. On the other hand, the eyes of marmosets and guinea pigs develop low amounts of myopia. While these differences may reflect interspecies differences in the operational properties of the emmetropization process, it is well known that domesticated animals often exhibit less hyperopic/more myopic ametropias than their feral counterparts.139 In this respect, the low degrees of myopia in marmosets and guinea pigs may reflect an adaptation to their caged environments. 
Mice also appear to undergo emmetropization, although the pattern appears to be different from that exhibited by the five species included in Figure 2. As shown in Figure 3, near the onset of visual experience C57BL mice, a strain commonly used in studies of eye development, are myopic or exhibit low to moderate degrees of hyperopia and become relatively more hyperopic until approximately 50 days of age. However, it should be noted that technical difficulties measuring refractive errors in the small eyes of juvenile mice just after eye opening (at 12–14 days of age) prevents direct comparisons with other species. 
Figure 3
 
The mouse model of FDM. (A) Mean (±SD) refractive errors plotted as a function of age for C57BL mice.50,54,543,555,742,811813 (B) standard deviations of the mean refractive errors from the left panel are plotted as a function of age. The dashed red line represents the best-fitting linear regression and its 95% CIs.
Figure 3
 
The mouse model of FDM. (A) Mean (±SD) refractive errors plotted as a function of age for C57BL mice.50,54,543,555,742,811813 (B) standard deviations of the mean refractive errors from the left panel are plotted as a function of age. The dashed red line represents the best-fitting linear regression and its 95% CIs.
The small size of mouse eyes makes determination of refractive state difficult. It is not certain how much the small eye hyperopic artifact contributes to the measured hyperopia. Using retinoscopy, Glicksten and Millodot134 estimated that the hyperopic error was on the order of +14 to +16 D. Calculations based on the focal length of paraxial schematic eye models suggest that the artifact could be over +30 D and that these estimates suggest that the artifact should become more hyperopic with age.69 On the other hand, comparisons of refractive errors obtained by retinoscopy in rodents to those obtained using cortical visual-evoked potentials suggest that the small eye artifact is much smaller or nonexistent,140 possibly because the primary retinal structures contributing to the light reflection are deeper in the retina than the vitreoretinal interface. Estimates of refractive error in the mouse eye are complicated by the large amount of high-order aberrations (particularly spherical aberration) and the mouse eye's large depth of focus.57 The estimated depth of focus of the mouse eye can vary between subjects in a given study (1.7–11 D57) and between studies, with estimates ranging to over 20 D.48 
Perhaps due to refractive error measurements starting later in development, mice do not seem to exhibit an obvious reduction in the intersubject variability in refractive errors from 20 to 100 days of age. In the right plot in Figure 3, the standard deviations of the average measures are plotted as a function of age. Linear regression analysis indicated that the intersubject variability was essentially constant during early development, perhaps reflecting the small diopter range during the emmetropization process in this development period. 
3.2 Evidence for Visual Regulation of Eye Growth: Form-Deprivation Myopia
During the course of their investigations of the effects of abnormal visual experience on brain development, Hubel et al.141143 observed that surgical eyelid closure, a procedure employed to deprive an animal of spatial vision, produced axial myopia in infant monkeys. This serendipitous, but fundamental, discovery led to the development of the first truly useful animal model of myopia.10,11,144,145 Subsequently, the phenomenon of FDM has been studied in a wide range of animal species, and investigations of FDM have helped establish the role of vision in refractive development, define the operating characteristics of the vision-dependent mechanisms that influence ocular growth, define the ocular anatomic changes associated with vision-induced changes in refractive state, and identify functional changes in the retina, choroid, and sclera leading to our current understanding and theories of the cellular and biochemical mechanisms of eye growth control. 
3.2.1 Form Deprivation Myopia: The Basic Phenomenon
In many respects, the phenomenon of FDM has been the most useful experimental animal model of myopia. Many studies have shown that depriving the retina of patterned visual stimulation by suturing the eyelids closed, or more recently by securing a translucent diffuser over the eye, consistently produces axial myopia relative to untreated eyes. These observations provided powerful scientific proof that alterations in vision can produce robust myopic changes. In this respect, the form-deprivation paradigm eliminated potentially confounding issues related to evolutionary pressures and self-selection that had limited many previous animal and human studies on the effects of vision on refractive development. In addition, the fact that monocular form deprivation produces axial myopic anisometropia, which demonstrated that the effects of vision are largely independent in the two eyes, provided an in-animal control for many other environmental factors and, most importantly, potentially confounding genetic factors that could mask the effects of vision on refractive development. 
FDM has been observed in several experimental models (see Fig. 4) as well as in humans.146149 It is primarily the result of increased axial elongation, mainly vitreous chamber, along with thinning of the choroid and the fibrous sclera.36,46,128,137,150159 Only a few studies have reported changes in corneal curvature (see Section 3.5.5) and lens thickness with form deprivation.159163 The diversity of species exhibiting FDM is impressive, ranging from fish,101,103 to birds, to mammals,36,46,164,165 and to primates,10,153,166168 including humans (for another recent review see Schaeffel and Feldkaemper9). There are differences between species in the magnitude of FDM produced and rate of axial elongation, which in large part, reflect species differences in eye size and relative maturation rates. However, it is difficult to directly compare the quantitative differences between individual studies and animal models because of the differences in experimental paradigms, duration of imposed deprivation, degree of image degradation (e.g., variable reductions in image contrast through diffusers), normal pattern of emmetropization, inherent ocular anatomic variations, and/or differences in susceptibility to environmental myopia. The small numbers of qualitative, between-study inconsistencies in the effects of form deprivation that exist in the literature appear to reflect unintended side effects of the treatment strategies that may have masked axial myopic changes. For example, eyelid closure and some continuous contact lens–wearing strategies have been shown to alter the shape and power of the cornea masking potential axial myopic changes. Nonetheless, the fact that FDM occurs in such a wide variety of animals suggests that the vision-dependent mechanisms responsible for FDM are fundamental from an evolutionary point of view and have been conserved across species. Consequently, insights into the mechanisms that mediate FDM obtained in one species are likely to apply to other species, at least qualitatively. 
Figure 4
 
Examples of FDM in animal models and humans. Refractive error frequency distributions for normal (open symbols) and form-deprived eyes (filled symbols) from chicks137 (A), rhesus monkeys166,168,819 (B), and humans146 (C). Form deprivation was produced in chicks using diffuser lenses; the data were obtained after either 28 or 42 days of age. Form deprivation was produced by surgical eyelid closure in monkeys; the data were obtained over a range of ages and durations of deprivations. Form deprivation in children occurred as a result of conditions (hemangioma and eyelid ptosis) that interfered with a clear retinal image.
Figure 4
 
Examples of FDM in animal models and humans. Refractive error frequency distributions for normal (open symbols) and form-deprived eyes (filled symbols) from chicks137 (A), rhesus monkeys166,168,819 (B), and humans146 (C). Form deprivation was produced in chicks using diffuser lenses; the data were obtained after either 28 or 42 days of age. Form deprivation was produced by surgical eyelid closure in monkeys; the data were obtained over a range of ages and durations of deprivations. Form deprivation in children occurred as a result of conditions (hemangioma and eyelid ptosis) that interfered with a clear retinal image.
With respect to the role of vision in the regulation of ocular growth and refractive development, as first proposed by Schaeffel et al.,169 form deprivation is an open-loop condition that prevents the vision feedback that normally coordinates ocular growth and emmetropization. In particular, form deprivation, especially that associated with strong diffusers or eye lid closure, virtually eliminates meaningful visual feedback regarding the eye's refractive status. When viewing through a strong diffuser, the eye cannot determine if it is emmetropic, myopic, or hyperopic and, consequently, the eye elongates in an unregulated or undamped manner. 
The diffusers that are typically employed in form-deprivation experiments produce dramatic reductions in retinal image contrast, alterations in vision that would rarely be encountered during normal development. However, it is important to note that FDM is a graded phenomenon and that the degree of axial myopia is positively correlated with the degree of image degradation.157,170,171 Even relatively mild diffusers that reduce vision by amounts equivalent to small degrees of optical defocus can produce FDM, albeit smaller in magnitude than that produced by stronger diffusers. As a consequence, it is possible that the mechanisms responsible for FDM come into play during normal viewing conditions. More importantly, these results emphasize that the potential for a clear, high-contrast, retinal image is essential for normal emmetropization. 
In a given species, the degree of FDM depends on both environmental and genetic factors. For example, it is well established that the magnitude of the changes in eye growth and myopia are also correlated with age of onset and the duration of the period of deprivation.168 In general, the degree of FDM is larger for earlier and longer periods of form deprivation. However, there are also substantial individual differences in the susceptibility to FDM. For example, Schaeffel and Howland169 showed that in response to equivalent periods of binocular form deprivation the between-subject differences in FDM were much larger than the interocular differences found in individual animals. Individual differences in susceptibility to environmental influences are also probably responsible for the large range of myopic anisometropias produced by form deprivation. As illustrated in Figure 5, equivalent periods of form deprivation produced by identical diffuser lenses can result in a substantial range of relative myopic errors in infant monkeys. 
Figure 5
 
The effects of form deprivation are variable. Interocular differences in refractive error (treated eye − fellow eye) plotted as a function of age for individual rhesus macaque monkeys reared with monocular diffuser lenses. The first symbol of each plot represents the onset of form deprivation. The shaded area in each plot represents ±2 SDs of the mean anisometropia for normal control monkeys (adapted from Hung et al.470).
Figure 5
 
The effects of form deprivation are variable. Interocular differences in refractive error (treated eye − fellow eye) plotted as a function of age for individual rhesus macaque monkeys reared with monocular diffuser lenses. The first symbol of each plot represents the onset of form deprivation. The shaded area in each plot represents ±2 SDs of the mean anisometropia for normal control monkeys (adapted from Hung et al.470).
Constant darkness also deprives the eye of form vision. In chicks, constant darkness results in eye enlargement as it does in form deprivation; however, refraction becomes hyperopic because of significant corneal flattening induced by the constant darkness.137,172,173 This corneal effect appears to be related to the loss of circadian cues because similar effects were observed in constant light rearing as well.95 Raising macaque monkeys in constant darkness prevented emmetropization, leaving the monkeys generally more hyperopic than age-matched controls.136 In tree shrews, however, dark-rearing produced significantly more myopia than in control animals.174 The difference in response is unexplained, but taken together the results from all species generally support the importance of visual experience in emmetropization. 
3.3 Evidence for Visual Regulation of Eye Growth: Recovery From Form-Deprivation Myopia
Although the phenomenon of FDM clearly demonstrates that visual experience can influence ocular growth and refractive development, form-deprivation paradigms provide little about the nature of the visual signals that influence early ocular growth or the process of emmetropization. One of the first clear indications that ocular growth and refractive development are regulated by signals associated with the eye's refractive state came from studies of recovery from FDM. In a variety of species, upon removing the diffuser lenses used to produce monocular form deprivation, young animals showed rapid and systematic reductions in the experimentally induced myopic anisometropias, principally due to a decrease in the myopia in the originally deprived eye.36,101,129,175177 While nonvisual mechanisms that are sensitive to the overall shape of the eye173 may contribute to recovery from FDM, the fact that correcting the myopia induced by form deprivation with negative lenses prevents recovery confirms that vision-dependent mechanisms related to the eye's refractive state regulates eye growth and emmetropization.178,179 
The recovery from FDM comes about primarily as a result of changes in vitreous chamber elongation rates. Removing the diffusers from a young animal with monocular FDM, results in myopic defocus in the treated eye and produces a dramatic reduction in the deprived eye's vitreous chamber growth rate. The abnormal axial elongation produced in FDM virtually comes to a halt while the fellow eye continues to grow at a more normal rate (see Fig. 6). At the same time, the cornea and crystalline lens continue to follow their normal developmental course and become flatter in both eyes (i.e., the normal reductions in corneal and lenticular refractive power are not altered by the recovery process). The concomitant increase in the eye's focal length results in a systematic reduction of the myopia in the formerly deprived eye. Once the vitreous chamber depth of the fellow control eye catches up to that of the formerly deprived eye, the refractive errors in the two eyes are reasonably matched. Subsequently, the formerly deprived eye begins to grow again and both eyes adopt similar vitreous chamber growth rates. The anatomic changes are in large respect qualitatively similar in all species, although it is likely that rapid choroidal thickness changes play a larger role in the early refractive error changes in chicks151 than in mammalian species (see Section 3.5.3).155,180 
Figure 6
 
Example of recovery from FDM in rhesus macaques. (A) Spherical-equivalent refractive error plotted as a function of age for the treated (red and cyan symbols) and fellow control eyes (black and white symbols). (B) Interocular differences in refractive error for the same animal plotted as a function of age. (C) Vitreous chamber depth plotted as a function of age for the treated (red and cyan symbols) and fellow control eyes (black and white symbols). The first symbols represent the onset of treatment. The red and black symbols indicate the treatment period. The large green symbols represent the onset of the recovery period. The open and cyan symbols indicate the recovery period. The solid black lines in the top and middle panels are data from untreated control monkeys.
Figure 6
 
Example of recovery from FDM in rhesus macaques. (A) Spherical-equivalent refractive error plotted as a function of age for the treated (red and cyan symbols) and fellow control eyes (black and white symbols). (B) Interocular differences in refractive error for the same animal plotted as a function of age. (C) Vitreous chamber depth plotted as a function of age for the treated (red and cyan symbols) and fellow control eyes (black and white symbols). The first symbols represent the onset of treatment. The red and black symbols indicate the treatment period. The large green symbols represent the onset of the recovery period. The open and cyan symbols indicate the recovery period. The solid black lines in the top and middle panels are data from untreated control monkeys.
Due to the manner in which recovery from experimentally induced myopia is achieved, the ability of a given animal to recover will greatly depend on the degree of myopia and the age at which unrestricted vision is restored.176,181 For example, it is not likely that an animal could recover fully from FDM if unrestricted vision was restored after the age at which the cornea and lens had stopped flattening, or if the initial degree of axial elongation exceeded normal adult eye lengths. Because it does not appear that vision-dependent mechanisms can result in a significant absolute reduction in axial length (at least in primates182) or in compensating corneal or lens growth, stopping abnormal axial elongation in an optically mature eye would only stabilize myopia if the eye's optical power could be decreased in some other way. Wallman183 suggested that this age-dependent limitation in the ability of the eye to recover from myopia may explain why common forms of myopia that develop in adolescent or adult humans persist. In children, corneal power reaches adult levels by 18 to 24 months of age, and after 8 to 10 years of age, when most myopia is typically diagnosed, the changes in lens power are small.184 Therefore, whereas human infants with myopia shortly after birth usually show some emmetropization, children who become myopic after their corneas and lenses become optically mature are unlikely to recover.185 
3.4 Evidence for Visual Regulation of Eye Growth: Compensation for Lens-Imposed Defocus
The most rigorous and clinically relevant test for the hypothesis that ocular growth and refractive development are actively regulated by defocus was provided by studies that employed lenses to alter the eye's effective refractive state. The original study by Schaeffel et al.186 was first to show that the eyes of young chicks wearing positive or negative spectacle lenses compensated appropriately for the imposed defocus, essentially emmetropizing through the defocus imposed by the lens treatment. Specifically, placing a negative lens in front of an emmetropic eye optically simulated hyperopia and to compensate for the lens (i.e., to re-establish emmetropia when viewing through the lens), the chick eye grew until it developed a degree of myopia equivalent to the power of the lens. On the other hand, a positive lens produced myopic defocus on the retina, which led to inhibition of eye growth, resulting in the eye becoming more hyperopic in order to re-establish an emmetropic refractive state through the lens. The fact that chicks exhibit appropriate compensating eye growth for equivalent degrees of hyperopic and myopic defocus, even when accommodation and other behavioral cues to the sign of the effective refractive error are excluded, demonstrates that the eye can detect the sign of defocus and alter its growth in the appropriate direction to eliminate both myopic and hyperopic defocus.187,188 
Although some early primate studies189192 that employed contact lens regimens that produced unwanted corneal alterations failed to confirm the original findings of Schaeffel et al.,186 compensation for lens-imposed defocus (commonly referred to as “lens compensation”) has been replicated many times in chicks,193196 and reported in many other species, including primates,191,197199 tree shrews,200,201 guinea pigs,37 and mice.46,47,50,59 As illustrated in Figure 7, the effective operating range of the compensation process differs between species. For instance, in chicks, complete compensation has been shown for a range of spectacle lens powers between −10 and +20 D.194 Based on the available data, the ranges of compensating responses for other species is variable, but all species that have been studied in a systematic fashion exhibit compensating refractive changes for both negative and positive lenses as follows: macaque, −2 to +8 D198; marmosets, −8 to +5 D199,202; tree shrew, −5 to +5 D200,201; guinea pig, −4 to +4 D37; mice, −30 to +5 D46,47,59; fish (Tilapia), −8 to +8 D.103 
Figure 7
 
Compensation for lens-imposed retinal defocus occurs in a variety of species (A) chicks,398 tree shrews,159 marmosets,191 rhesus macaques,198 and guinea pigs,37 and (B) mice.46,47,59 The mean ametropia obtained at the end of the lens-rearing period is plotted as a function of the power of the treatment lenses.
Figure 7
 
Compensation for lens-imposed retinal defocus occurs in a variety of species (A) chicks,398 tree shrews,159 marmosets,191 rhesus macaques,198 and guinea pigs,37 and (B) mice.46,47,59 The mean ametropia obtained at the end of the lens-rearing period is plotted as a function of the power of the treatment lenses.
The observed differences in the effective operating ranges of the emmetropization processes in these different species are likely to reflect several factors. In particular, when expressed in terms of diopters, shorter eyes and eyes with lower spatial frequency response properties would be expected to exhibit larger lens-compensation ranges.203 Interspecies differences in the average refractive errors found in normal neonates would influence the effective degree of defocus produced by a given powered lens and, thus, the effective lens compensation range. Natural and imposed differences in the set point target refractive error for emmetropization are also likely to influence the observed compensation ranges. For example, housing animals in cages that significantly restrict viewing distances may shift the compensation range in the myopic direction.204,205 In addition, behavioral issues are also likely to influence the lens-compensation range. For instance, it is reasonable to expect that animals with large accommodative amplitudes would exhibit greater ranges of compensation for negative than for positive lenses. However, in animals, such as primates, with well-developed binocular vision, issues relevant to accommodative convergence and efforts to maintain binocular vision at the expense of a clear retinal image, could mask this predicted asymmetry. Moreover, although the eye's refractive state is defined for distance viewing, animals with imposed myopia may simply prefer to fixate near objects, effectively eliminating the need to compensate for the imposed lens power.198 
It is interesting that at the limits of the operating range for lens compensation, high degrees of either natural or imposed hyperopic defocus do not produce myopia. As illustrated in Figure 7 for chicks, mice, and primates, increasing negative lens powers beyond a species-specific value results in less compensating myopia or little or no changes in refractive error. It is not a simple limitation on the ability of the eye to increase its axial length because form deprivation and rearing strategies in which defocusing lens powers are increased gradually over time have been shown to produce much larger myopic errors.198 Why imposed hyperopic defocus beyond the operating limits of lens compensation often fails to consistently produce myopia is unclear. One possible explanation is that the higher degrees of optical defocus, especially with monocular treatment regimens, cause other visual system changes (e.g., accommodative vergence interactions and possibly amblyopia), which somehow interfere with the effects of chronic defocus on ocular growth. This, however, is not a particularly satisfying explanation because monocular form deprivation, which produces profound sensory deficits in young monkeys, consistently results in exaggerated ocular growth and high degrees of myopic anisometropia.192 Moreover, monkeys with severe form deprivation–induced amblyopia consistently exhibit recovery from FDM.176 
Although there has, until recently, been a paucity of evidence for lens compensation in humans, when comparable optical conditions are produced in humans who successfully underwent emmetropization early in life, the resulting changes in refractive error are qualitatively similar to those in laboratory animals.206 Figure 8 shows the compensating refractive error changes produced by optically imposed anisometropia in monkeys and humans. In humans, the compensating change produced by an imposed anisometropia may be more apparent because regardless of viewing distance or which eye is used for fixation, the optical treatment consistently imposes an anisometropia. As illustrated in Figure 8, individual monkeys and humans consistently exhibit compensatory anisometropic changes that are in the appropriate direction to compensate for the imposed optical imbalance. In addition, recent human studies have documented small, short-term bidirectional changes in axial length and choroidal thickness in response to 1 to 2 hours of myopic and hyperopic defocus in young adult subjects,207210 which suggests that the human eye can also detect the sign of imposed optical defocus and undergo appropriate compensatory changes in axial length. 
Figure 8
 
Examples of anisometropic compensation in individual infant rhesus macaque monkeys ([A] adapted from Hung L-F, Arumugam B, She Z, Ostrin L, Smith EL III. Narrow-band, long-wavelength lighting promotes hyperopia and retards vision-induced myopia in infant rhesus monkeys. Exp Eye Res. 2018;176:147–160. Copyright © 2018 Elsevier Ltd.)470 and adolescent humans (age of onset 11 years) ([B] adapted from Phillips JR. Monovision slows juvenile myopia progression unilaterally. Br J Ophthalmol. 2005;89:1196–200. Copyright © 2005 British Journal of Ophthalmology).206 The first symbol in each plot represents the onset of treatment. The monkeys were reared with a −3 D lens in front of their treated eyes and a plano lens in front of their fellow eyes. The human subjects were corrected using a monovision contact lens strategy. The dominant eyes were corrected for distance; the fellow eyes were uncorrected by <2 D.
Figure 8
 
Examples of anisometropic compensation in individual infant rhesus macaque monkeys ([A] adapted from Hung L-F, Arumugam B, She Z, Ostrin L, Smith EL III. Narrow-band, long-wavelength lighting promotes hyperopia and retards vision-induced myopia in infant rhesus monkeys. Exp Eye Res. 2018;176:147–160. Copyright © 2018 Elsevier Ltd.)470 and adolescent humans (age of onset 11 years) ([B] adapted from Phillips JR. Monovision slows juvenile myopia progression unilaterally. Br J Ophthalmol. 2005;89:1196–200. Copyright © 2005 British Journal of Ophthalmology).206 The first symbol in each plot represents the onset of treatment. The monkeys were reared with a −3 D lens in front of their treated eyes and a plano lens in front of their fellow eyes. The human subjects were corrected using a monovision contact lens strategy. The dominant eyes were corrected for distance; the fellow eyes were uncorrected by <2 D.
There is still much to learn about the phenomenon of lens compensation, but the results from animal studies have clearly demonstrated that something as simple as a spectacle lens can predictably alter ocular growth. These results provide a solid scientific foundation for optical treatment strategies to reduce the progression of juvenile-onset myopia in children (see accompanying International Myopia Institute reports in this issue211213). 
3.5 Ocular Anatomic Changes Associated with Experimentally Induced Refractive Errors
Experimentally induced changes in refractive state are associated with several anatomic changes to ocular components related to changes in eye shape and size, principally in the depth and shape of the vitreous chamber. These vision-dependent alterations are associated with a number of changes in the retina, RPE, choroid, and sclera. Anterior segment changes have been observed in eyes with experimentally induced ametropias, but have not been found to be related directly to the visual regulation of refractive state.214216 
3.5.1 Retina
The retina is the primary tissue where information about optical defocus is converted into molecular signals, which are then transmitted through the RPE and choroid to the sclera and translated into the structural changes in the sclera underlying development of myopia (see Section 5). Both visual form deprivation and lens-imposed defocus have been shown to cause large-scale changes in gene expression in the retina (see Section 6). Changes in gene expression induced by visual form deprivation have also been shown to result in increased proliferation of the retinal progenitors at the retinal periphery of monkeys resulting in increased neurogenesis and increased growth of the retina.217 
3.5.2 Retinal Pigment Epithelium
The RPE also shows distinct morphologic changes during the development of myopia in humans and animals.218224 In animal models, enlargement of the eye during the development of experimental myopia is associated with an increase in the overall surface area of the RPE through the expansion of individual RPE cells across the entire epithelium,218220 although less pronounced in the temporal region.220 Such expansion may be due to either passive stretch or active growth of these cells. Like in many other ocular tissues, there also appears to be active changes in fluid dynamics within the RPE during periods of altered growth. In response to recovery from FDM, following diffuser removal, Liang et al.221 reported increased fluid retention and edema within the retina, RPE, and choroid, as well as ultrastructural reorganization of the RPE basal lamina. The authors hypothesized that this represented active changes in fluid movement across the RPE whose tight junctions act as a barrier that allows the regulated exchange of ions and water between the subretinal space and the choroid through modulation of its ionic channels. The role that any such fluid movement plays in the regulation of ocular growth is yet to be fully elucidated. Crewther et al.225227 suggested that such ionically driven fluid exchange across the RPE between the subretinal space and choroid may, in fact, underlie the significant choroidal thickness changes observed during periods of altered eye growth. Specifically, the authors suggest that accumulation of ions within the subretinal space during the development of FDM may inhibit fluid movement from the vitreous to choroid, leading to vitreous chamber swelling and thinning of the choroidal lacunae in chicks. In contrast, during periods of reduced ocular growth associated with diffuser removal, reverse changes in the ionic state within the subretinal space may induce fluid movement from the vitreous chamber across the RPE causing swelling of the choroidal lacunae. Supporting this hypothesis, ion levels within freeze-dried preparations of the retina, RPE, and choroid have been reported to be significantly modulated during the development and recovery from FDM,221,226 while potassium and phosphate levels are reported to be reduced, and chloride levels increased in the vitreous chambers of form-deprived chicks.228 Furthermore, pharmacologic inhibition of ion movement has been shown to disrupt the compensatory response to lens wear in chicks.227 Together, these findings support the possibility that the choroidal thickness changes observed during alterations in the rate of ocular growth could be associated with adjustments in ionic fluid movement across the RPE. However, choroidal swelling may also be explained by exchanges of fluid between the choroidal vasculature and the neighboring suprachoroidea. In support of this, Liang et al.221 noted that the concentration of Na+ and Cl ions in the choroidal lymphatics rises steeply over the first 72 hours of recovery from FDM, during which the choroid rapidly swells. The most likely source of these accumulating ions is the choroidal vasculature. 
3.5.3 Choroid
The choroid is a highly vascular layer of connective tissue positioned between the RPE and sclera. Together with the ciliary body and iris, the choroid forms the uveal tract. 
The past hundred years or so have yielded episodic but compelling pieces of evidence that the functions of the choroid are substantially more than supplying blood to the outer retina.229 For instance, work by van Alphen230 indicated that the choroid, and not the sclera, might be a major determinant of the size and shape of the eye, because when the sclera was removed from the posterior pole, and pressure corresponding to normal IOP applied, the exposed choroid did not balloon out, but maintained its curvature while being displaced posteriorly. Moreover, mysterious neurons currently known as intrinsic choroidal neurons were reported in human choroid as long ago as 1859,231 and their functions are still largely unknown.232 Nonvascular smooth muscle is located in the choroid, which has been verified in various species (birds,5,150,233235 primates,236238 rabbits239). Finally, the existence of large lacunae, possibly lymphatic vessels, in most species,5,235,240242 including humans,238 indicate diverse functions unrelated to blood flow. Today, largely because of the finding, first in birds,150,151 then extended to pri-mates,155,180 that the thickness of the choroid changed in response to retinal defocus, thus acting as a means of positioning the image plane on the retina, it is widely accepted that the choroid is “multifunctional” and involved in numerous aspects of ocular/visual health. 
The first evidence for the compensatory choroidal thickness changes in experimental myopia research came from observations of gross changes in the appearance/consistency of choroids from dissected myopic chick eyes, which led to the critical findings that myopic defocus caused large increases in choroidal thickness, and form deprivation or hyperopic defocus caused choroidal thinning.150,151 The subsequent use of higher-frequency ultrasound allowed finer resolution, and demonstrated that the choroidal responses were rapid (within hours), bidirectional, and highly precise. In chicks, the compensatory changes in choroidal thickness are symmetric and linear over a range of imposed defocus from approximately −15 to +15 D.151 The speed of this choroidal compensation is intermediate between that of (fast) lenticular accommodation and the (slow) changes in scleral extracellular matrix (ECM) synthesis that alter eye size, and so these choroidal responses may function as a mechanism to sustain focus on the retina until the eye length “catches up” to the front optics. Subsequently, the choroid returns to normal at a pace in concert with the changing size of the globe. This process of the scleral changes altering globe size together with the choroidal thickness changes altering the image plane create an association between faster-growing (large) eyes and thinner choroids, versus slower-growing (small) eyes and thicker choroids. This phenomenon has since been observed in all other species tested, including marmosets,180 rhesus macaques,155 guinea pigs,37 and humans.208 The responses in mammals are, however, much smaller in magnitude than those in birds. 
Whether the thickness of the choroid influences the rate of scleral growth, perhaps by the secretion of regulatory molecules (see Section 5.3.1), has been a question of interest for some time because of its translational implications. If there were a causal relationship, for instance, then perhaps choroidal thickness in humans might be a “risk factor” for the development of myopia, which would make it a potentially valuable tool in deciding on treatment therapies for “at-risk” children.243 
Two studies using the chick model have addressed the question of whether choroidal thickness is a predictor of ocular growth rate. The first was a heritability analysis on nearly 900, 4-day-old chicks,75,244 which showed approximately 50% of the variation in choroidal thickness was determined by genetics. Furthermore, initial choroidal thickness was not related to initial eye size nor to subsequent growth rates. In an extension of this study, a cohort of 500 chicks were deprived of form vision for 4 days to induce myopia, and initial choroidal thickness did not predict the growth response to the deprivation.243 A smaller study from a different lab, however, reported a significant association between initial choroidal thickness and subsequent growth rates such that eyes with thinner choroids grew faster than those with thicker ones, perhaps supporting the association of thicker choroids with greater secretion of a growth inhibitor.245 The discrepancy between these two studies may reflect differences in the age at the onset of the experiments, as the first study used younger chicks. The latter study also reported a negative correlation between initial choroidal thickness and subsequent changes in thickness; the thinner choroids of faster-growing eyes showed greater subsequent thickening. By contrast, initial choroidal thickness was not predictive of ocular growth rates in eyes wearing either positive lenses (slowing elongation), or negative lenses or diffusers (stimulating elongation). Neither were the magnitudes of choroidal thickness changes in response to defocus predictive of ocular growth rates. These differences between untreated eyes, in which thickness was predictive of growth, and experimentally manipulated eyes in which it was not predictive, might reflect a decoupling of the “choroidal system” from the “growth system” under experimental visual conditions. Together, these findings weaken the hypothesis that the magnitude of choroidal thickening is related to its “potency” for either a secreted signal molecule, or as a mechanical barrier to such a signal molecule, supporting separate mechanisms for the choroidal thickness and scleral responses. 
Several other lines of evidence support separate mechanisms for the choroidal thickness and scleral responses to visual signals. First, a detailed study of the temporal integration characteristics of the two responses reported dissociations between choroidal thickness and scleral responses. If eyes were exposed to brief and infrequent episodes of defocus (7 minutes/4 times per day), in the case of positive lenses, only inhibition of axial elongation was observed, and not choroid thickening, while in the case of negative lenses, only the choroid thinning response was found, and not stimulation of axial elongation.246 Second, eyes with lesions of both ocular parasympathetic pathways (ciliary and pterygopalatine ganglia), did not respond to form deprivation with the usual development of myopia, but instead exhibited reduced axial growth.247 Surprisingly, however, choroids of the form-deprived eyes thinned, showing the usual compensatory response to form deprivation. This thinning of the choroid in these aberrantly slow-growing, form-deprived, lesioned eyes suggests a pathological response to form deprivation, as suggested by electron microscopy showing abnormalities in the photoreceptor outer segments and RPE in form-deprived eyes.221 Finally, a study in chicks found that oxotremorine, a muscarinic acetylcholine agonist, stimulated ocular growth, and thinned the choroid; however, two other agonists that were ineffective at stimulating growth also caused choroidal thinning.248,249 These three distinct lines of investigation show that choroidal thickness changes can be dissociated from axial growth suggesting that the former is not a necessary precursor for, or indicator of, the latter. 
3.5.3.1 Mechanisms Underlying Changes in Choroidal Thickness
In chicks, the main anatomic changes accounting for the large increases in thickness in response to myopic defocus occurred in the choroidal stroma, where the presence of large, fluid-filled lacunae suggested potential underlying mechanisms.150 In addition, α-actin–positive nonvascular smooth muscle cells were identified in the stroma,232,234,250 and are also present in other species, including humans.238,239,251 
The potential underlying mechanisms can be broadly divided into two categories as follows: those related to fluid-flux changes, and those related to smooth muscle activity (as reviewed by Nickla and Wallman229). The fluid-flux hypothesis posits that the rapidity (within hours) and magnitude (up to quadrupling) of the thickness changes favor a redistribution in fluid compartments as the main mechanism. This is supported by several lines of study. First, thicker choroids from eyes responding to myopic defocus synthesized more proteoglycans than thinner ones,252 suggesting that these hydrophilic matrix molecules play a role in the changing thickness of the stroma. However, the relatively small differences in synthesis rates between the two extremes in thicknesses weaken this hypothesis. Second, the permeability of the choroidal capillaries may increase, allowing movement of proteins from the lumen to the stromal matrix and/or lymphatics, followed by passive fluid flux.242 Several findings support this latter hypothesis, as follows: (1) form-deprived chick eyes had fewer fenestrations in its choriocapillaris253; (2) the protein content in suprachoroidal fluid was higher than normal in experimentally thickening choroids, and lower in experimentally thinned ones254; and (3) thicker choroids had higher amounts of fluorescein-dextran than thin ones after intravenous dextran injection,254 and these also had higher amounts of albumen.255 Third, because the anterior uvea (iris and ciliary body) is physically connected to the choroid, changes in the amount of aqueous flowing via the uveoscleral pathway might play a role. Finally, increased fluid flowing from the retina across the RPE might account for an increased amount of fluid in the stromal lacunae.256 
It is possible that choroidal thickening and thinning occur via different mechanisms. The fluid-flux mechanism is probably too slow to account for the finding that choroids can thin by approximately 50 μm within an hour.257 A more likely possibility involves smooth muscle contraction. In fact, the choroidal stroma in birds and primates, including humans, contains actin-positive nonvascular smooth muscle cells that are innervated by the parasympathetic and sympathetic systems.232,234,238 The axon terminals contacting the smooth muscle are positive for Nicotinamide adenine dinucleotide phosphate–diaphorase, indicating the presence of nitric oxide (NO), and for vasoactive intestinal peptide (VIP), which are both parasympathetic transmitters. Notably, stimulation of ciliary axons innervating explant choroids caused a contraction of the tissue, which was blocked by atropine, suggesting muscarinic cholinergic parasympathetic innervation.235 Further support for a muscle contraction–mediated thinning is the finding that muscarinic agonists thin chick choroids both in the intact eye and in vitro.235 
In summary, the choroid is a multifunctional structure, containing various tissue/cell types whose functions are as yet unknown. Many lines of study in animal models conclude that it plays important roles in the visual regulation of ocular growth. The existence in human choroids of similar features and physiological responses suggest a conservation of function among species. Recent studies on choroidal thickness changes in various ocular pathologies, including myopia and glaucoma, will help uncover its potential impact for the development of treatment therapies for vision health. 
3.5.4 Sclera
The sclera is a dense connective tissue that forms the outer coating of the eye and defines the eye's size and shape. The anatomy of the sclera varies among vertebrates.5 In most vertebrates it is composed of two layers—an inner layer of hyaline cartilage and an outer layer of dense fibrous connective tissue (Fig. 9A). The two layers vary in their relative thicknesses in different regions of the ocular globe, the fibrous and cartilaginous layers are approximately equal in thickness at the posterior pole, but the fibrous layer progressively thins in equatorial and anterior ocular regions. Scleral ossicles, rings of bone in the sclera found in the anterior segment, are also found in all vertebrates except for eutherian mammals and crocodiles.5 The sclera in humans258 and other eutherian mammals (such as macaque monkeys, marmosets, tree shrews, guinea pigs, and mice) is composed of only a fibrous layer156,259,260 (see Figs. 9B, 9C), which is made primarily of collagen type I with smaller amounts of types III and V collagen, and held together with elastin and proteoglycans. However, ECM molecules previously believed unique to cartilage, such as aggrecan,261263 proline arginine-rich and leucine-rich repeat protein,264 and cartilage olimeric matrix protein,265 are also present in the mammalian fibrous sclera, suggesting that cartilaginous components have been retained in the sclera through evolution and serve important biochemical and biomechanical functions. 
Figure 9
 
(A) Chick scleral cross section. The cartilaginous part (cart. sclera) facing the choroid and the fibrous part (fibr. sclera) forming the outer shell can be easily distinguished in this Toluidine blue stained semithin section. (B, C) Electron micrographs of marmoset sclera. (B) Layers of collagen fibers with various orientation are detectable. White arrowhead indicates the cell body of a fibroblast embedded between ECM layers. (C) Higher magnification showing longitudinally (white arrow) and cross-sectional (black arrow) collagen fiber bundles.
Figure 9
 
(A) Chick scleral cross section. The cartilaginous part (cart. sclera) facing the choroid and the fibrous part (fibr. sclera) forming the outer shell can be easily distinguished in this Toluidine blue stained semithin section. (B, C) Electron micrographs of marmoset sclera. (B) Layers of collagen fibers with various orientation are detectable. White arrowhead indicates the cell body of a fibroblast embedded between ECM layers. (C) Higher magnification showing longitudinally (white arrow) and cross-sectional (black arrow) collagen fiber bundles.
Significant changes in scleral ECM synthesis, accumulation, and turnover are associated with visually induced changes in eye size and refraction in a variety of animal models.159,260,266268 Despite the differences in scleral anatomy, the fibrous sclera of mammals and the fibrous layer of the avian sclera appear to change in a similar manner in experimentally induced myopia. When ocular elongation accelerates during myopia development, the fibrous sclera thins in mammals259,260 and birds.154,269 Thinning of the fibrous sclera in chicks is similar to what is seen in the fibrous mammalian sclera.156,270,271 The cartilaginous sclera of birds, however, demonstrates increased growth as the eye elongates, which is accompanied by an increase in synthesis and accumulation of proteoglycans and an increased dry weight.266,272 All vertebrates appear to use similar signaling mechanisms to control the structure of the sclera and do so by controlling growth in the cartilage, where it is present, and by controlling remodeling in the fibrous sclera. 
The scleral changes in experimental myopia development in primates, tree shrews, guinea pigs, and mice are similar to those associated with high myopia in humans. There is a restructuring of the ECM, a loss of ECM and scleral thinning.259,271,273275 These alterations are associated with several changes in the mechanical properties of the sclera. Specifically, there are increases in the viscoelasticity and creep rate of the sclera,276,277 which make the tissue more extensible so that normal IOP may produce an enlargement of the vitreous chamber. A recent study also suggested that the crimp angle of tree shrew scleral collagen fibril bundles increases during the development of myopia, which could decrease the stiffness of the sclera. Decreases in crimp angle were observed during recovery from myopia.278 
In contrast, myopia development in chicks is associated with active scleral growth due to increased ECM synthesis and the accumulation of proteoglycans in the cartilaginous layers of the sclera.266,279 The biochemical changes in the sclera and control of scleral growth during eye growth and myopia development will be discussed in Section 5.3. 
In humans, mammals, and chicks, scleral changes associated with myopia development are most pronounced at the posterior pole.259,260,280 The preferential involvement of the posterior sclera in myopia may be related to regional differences in the growth states of the scleral cells, differences in scleral tensile stresses at the posterior pole, or it may reflect the distribution and density of retinal, choroidal thickness, and scleral components in the vision-dependent cascade that regulates ocular growth.281 
3.5.5 Corneal and Anterior Segment Changes
While most of the vision-induced changes in the refractive state of the eye observed in experimental models and common refractive errors in humans can be explained by changes in the axial growth of the eye, changes in corneal curvature and anterior chamber depth have also been observed in some animal studies. Figure 10 shows the changes in corneal curvature and anterior chamber depth that have been described in experimentally induced myopia in several species. Overall, the largest changes in corneal curvature and anterior chamber depth were found in chicks, where high amounts of induced myopia were associated with steeper corneas and deeper anterior chambers. Smaller, but significant changes were found in other species. Nearly all of the significant changes for both corneal power and anterior chamber depth were observed in form-deprived animals, possibly reflecting the generally larger myopic errors obtained with form deprivation. However, steeper corneas were also correlated with increasing myopia produced by either diffusers or negative lenses in monkeys.163 Note that studies employing a lid-sutured paradigm, despite its significant myopia-induction effects, were not included in Figure 10 because corneal flattening is often a side effect of surgical lid closure.34,153,282284 
Figure 10
 
Changes in corneal power and anterior chamber depth found in different animal models with experimentally induced myopia. The x- and y-axis parameters represent either interocular difference (treated eye − fellow control eye) or intergroup differences (treated group − normal group). The filled and open symbols represent statistically significant and insignificant changes, respectively. Gray symbols indicate studies that did not perform statistical tests. Numbers inside or near each symbol represent different studies. █ Chicks: (1) Wallman et al.,72 diffusers; (2) Gottlieb et al.,137 diffusers; (3) Hayes et al.,820 diffusers; (4) Irving et al.,194 lenses; (5) Troilo et al.,160 diffusers; (6) Napper et al.,162 diffusers; (7) Napper et al.,556 diffusers; ⧫ tree shrews: (8) Guggenheim et al.,802 diffusers; (9) Siegwart et al.,129 diffusers; (10) McBrien et al.,271 lid-suture; ▴ guinea pigs: (11) Howlett et al.,36 diffusers; ▾ marmosets: (12) Graham and Judge,202 negative lenses; (13) Troilo and Nickla,347 diffusers; • rhesus monkeys: (14) Smith and Hung,157 diffusers; (15) Qiao-Grider et al.,176 diffusers; and (16) Qiao-Grider et al.,163 diffusers and negative lenses, induced myopia was not available, myopic anisometropia of more than −1.0 D was used. For chicks, the corneal radius of curvature values was converted to corneal powers using a refractive index of n' = 1.369.821
Figure 10
 
Changes in corneal power and anterior chamber depth found in different animal models with experimentally induced myopia. The x- and y-axis parameters represent either interocular difference (treated eye − fellow control eye) or intergroup differences (treated group − normal group). The filled and open symbols represent statistically significant and insignificant changes, respectively. Gray symbols indicate studies that did not perform statistical tests. Numbers inside or near each symbol represent different studies. █ Chicks: (1) Wallman et al.,72 diffusers; (2) Gottlieb et al.,137 diffusers; (3) Hayes et al.,820 diffusers; (4) Irving et al.,194 lenses; (5) Troilo et al.,160 diffusers; (6) Napper et al.,162 diffusers; (7) Napper et al.,556 diffusers; ⧫ tree shrews: (8) Guggenheim et al.,802 diffusers; (9) Siegwart et al.,129 diffusers; (10) McBrien et al.,271 lid-suture; ▴ guinea pigs: (11) Howlett et al.,36 diffusers; ▾ marmosets: (12) Graham and Judge,202 negative lenses; (13) Troilo and Nickla,347 diffusers; • rhesus monkeys: (14) Smith and Hung,157 diffusers; (15) Qiao-Grider et al.,176 diffusers; and (16) Qiao-Grider et al.,163 diffusers and negative lenses, induced myopia was not available, myopic anisometropia of more than −1.0 D was used. For chicks, the corneal radius of curvature values was converted to corneal powers using a refractive index of n' = 1.369.821
Although it is not clear how vision-dependent mechanisms could alter corneal power and anterior chamber depth during refractive development, some data suggest that the anterior segment changes are an epiphenomenon or side effect associated with changes in the posterior segment of the eye. For example, in chicks reared with hemiretinal form deprivation (i.e., diffusers that affected half of the retina), the nature of corneal changes (the direction of astigmatism in particular) varied with the location of the imposed deprivation (e.g., superior hemiretina versus temporal hemiretina285). Similarly, in monkeys both negative and positive spherical lens-rearing strategies, which elicited, respectively, either compensating increases or decreases in vitreous chamber elongation, produced similar corneal astigmatic errors.286 However, substantial vision-induced changes in vitreous chamber depth and refractive error can be produced in monkeys without concomitant changes in the anterior segment, suggesting that the anterior and posterior segments of the eye are independently regulated.163 In this respect, several manipulations have been shown to decouple anterior and posterior chamber alterations. For example, administration of a variety of neurotoxins can produce contrasting anterior and posterior segment changes.97,287289 However, this effect might be specific to birds reared under constant light (see Section 4.1). Nevertheless, the evidence is strong that the growth of the cornea and anterior segment is largely programmed growth, while emmetropization acts through visually guided changes of scleral growth and vitreous chamber size and shape changes. 
3.6 Key Operating Characteristics of Experimental Emmetropization
Understanding the functional operating characteristics of the vision-dependent mechanisms that regulate eye growth and emmetropization is critical for translating the concepts developed through animal research to human refractive development. 
3.6.1 Local Retinal Mechanisms
Investigations into the neural circuits mediating emmetropization have employed reduction strategies in efforts to identify critical components in the process that transforms visual signals into molecular signals that influence eye growth. Most typically, diffusers or powered lenses have been employed to induce changes in refractive development in combination with manipulations that were intended to eliminate or isolate potential circuit components. It was assumed that if a visual system component was essential for emmetropization, then inactivating or eliminating that component should prevent or alter the effects of the optical treatment regimens on refractive development. This series of investigations, which involved multiple labs and several species of experimental animals, led to one of the most interesting discoveries related to emmetropization, specifically, that the dominant vision-dependent mechanisms that regulate eye growth and refractive development are located entirely within the eye and operate in a local, regionally selective manner. 
The following experimental manipulations failed to prevent visually mediated changes in refractive error: (1) bilateral surgical removal of the striate cortices in form-deprived monkeys (i.e., resulting in perceptual blindness and eliminating a potential role for the visual cortex),145 (2) surgical transection of the optic nerve in form-deprived monkeys,145 in form-deprived and negative and positive lens–reared chicks,151,158,173,290 and in chicks recovering from FDM,291 (3) pharmacologic blockade of action potentials in retinal ganglion cells by tetrodotoxin in form-deprived tree shrews,292 and (4) sensory deafferentation by sectioning the trigeminal nerve in form-deprived monkeys.145 While many of these surgical manipulations can directly alter refractive development in isolation (e.g., optic nerve section generally produces hyperopic shifts in control eyes158,293), when these side effects are taken into account, the vision-induced changes in refractive development are comparable to those observed in lens- and diffuser-reared control animals. Consequently, these finding demonstrate that the visual signals (and other potential sensory signals) associated with form deprivation or optical defocus do not have to reach the central visual system or leave the eye for vision-dependent growth regulating mechanisms to function. 
In addition, eliminating the primary parasympathetic and sympathetic neural inputs to the eye (and their associated physiological processes, such as accommodation) does not eliminate vision-induced changes in refractive development. In particular, surgically disrupting the ciliary nerves (chicks),151,293295 the ciliary ganglion (monkeys),145 the Edinger-Westphal nucleus (chicks),187 or the superior cervical ganglion (monkeys)145 does not prevent FDM, lens-induced myopia (LIM), or lens-induced hyperopia (LIH). In addition, pharmacologic paralysis of accommodation does not prevent FDM or lens compensation in chicks296 and pharmacologic stimulation of accommodation does not prevent FDM in monkeys.145 Together these observations demonstrate that the act of accommodating, specifically ciliary muscle activity, is not essential for the visual regulation of ocular growth. Double ocular parasympathectomy (lesions of the ciliary and the pterygopalatine ganglion) affects FDM but not compensation for lens-induced defocus. This is further evidence of the existence of different mechanisms for FDM and LIM. 
Overall, the results described above demonstrate that neural mechanisms in the retina can detect the presence of defocus and generate signals that alter axial growth in a manner that eliminates the optical errors. Another key feature associated with this emmetropizing process is that the underlying mechanisms operate in a local, regionally selective manner across the retina. The strongest evidence to support this idea comes from experiments in which form deprivation or optical defocus were imposed over only half of the visual field. The use of hemiretinal manipulations was pioneered by Wallman et al.,281 who first showed that hemiretinal form deprivation in chicks produced localized axial elongation and myopia that was restricted to the treated hemiretina. These findings were subsequently replicated in mammals, including primates,159,297 and extended to apply to lens compensation for imposed hemifield hyperopic and myopic defocus.188,298,299 Figure 11 shows magnetic resonance images (MRIs) of macaque eyes with full and partial visual field deprivation. Monocular full-field form deprivation increases vitreous chamber elongation in the treated eye. The increases are greatest near the optic axis and decrease with eccentricity along the horizontal meridian in a relatively symmetrical manner (i.e., the treated eye becomes more prolate in shape). As a result, the degree of FDM is greatest near the optic axis and decreases with eccentricity, again in a relatively symmetrical manner (i.e., the changes in eye shape produced central myopia and relative peripheral hyperopia). In contrast, with nasal-field form deprivation, vitreous chamber elongation is restricted to the temporal hemiretina. The horizontal MRIs reveal an obvious change in the curvature of the posterior globe of the treated eye at the border between the deprived and nondeprived hemiretinas. As with full-field form deprivation, the anterior segment of the eye was not affected by hemifield form deprivation. As a consequence, the nasal field diffusers produce myopia that was restricted to the nasal visual field. 
Figure 11
 
MRIs obtained in the horizontal plane for the treated (left) and control eyes (right) of rhesus macaque monkeys reared with monocular full-field form deprivation (A) and monocular nasal-field form deprivation (B) (adapted from Smith EL III. Prentice Award Lecture 2010: a case for peripheral optical treatment strategies for myopia. Optom Vis Sci. 2011;88:1029–44. Copyright © 2011 American Academy of Optometry).326 The nasal and temporal retinas are designated as N and T, respectively. In the middle panels, the outlines for the treated (red) and fellow eyes (blue) have been superimposed after rotating the fellow eye images around the optic axes so that the nasal retinas (N) are shown to the right for both eyes. The superimposed images were aligned using the lines that connected the equatorial poles of the crystalline lenses as a reference (the red lines shown in the treated- and fellow-eye images in the left and middle columns).
Figure 11
 
MRIs obtained in the horizontal plane for the treated (left) and control eyes (right) of rhesus macaque monkeys reared with monocular full-field form deprivation (A) and monocular nasal-field form deprivation (B) (adapted from Smith EL III. Prentice Award Lecture 2010: a case for peripheral optical treatment strategies for myopia. Optom Vis Sci. 2011;88:1029–44. Copyright © 2011 American Academy of Optometry).326 The nasal and temporal retinas are designated as N and T, respectively. In the middle panels, the outlines for the treated (red) and fellow eyes (blue) have been superimposed after rotating the fellow eye images around the optic axes so that the nasal retinas (N) are shown to the right for both eyes. The superimposed images were aligned using the lines that connected the equatorial poles of the crystalline lenses as a reference (the red lines shown in the treated- and fellow-eye images in the left and middle columns).
The fact that the vision-dependent mechanisms that dominate refractive development operate in a regionally selective fashion and can produce changes in eye shape suggests that it is unlikely that some central neural mechanisms play a primary role in refractive development. For example, it has often been speculated that the act of accommodation contributes to the development of myopia. It is difficult to imagine how the act of accommodation or mechanical changes associated with accommodation (e.g., a potential increase in IOP) could produce the regional changes in eye shape and refractive error shown in Figure 11
The local nature of the vision-dependent emmetropization mechanisms may have evolved to optimize the eye's refractive state across the visual field (i.e., to promote optimal panoramic vision). It is likely that these local retinal mechanisms are also responsible for the relative lower-field myopia observed in many species300 and for the eccentricity-dependent changes in the pattern of refractive errors produced by rearing conditions that restrict viewing distance in a selective direction.301 
3.6.2 Temporal Integration of Visual Signals
Navigating in a three-dimensional world requires the eyes to scan and fixate different points in the environment, and depending on accommodative state, the sign and magnitude of defocus that the eye experiences varies over time. How competing visual signals are integrated over time presumably determines the eye's visually regulated growth. In this regard, eye growth does not appear to be regulated by the simple time-averaged level of defocus. Instead, evidence suggests that the temporal integration properties of vision-dependent eye growth are nonlinear and normally reduce the likelihood that the eye will become myopic.302 
Several nonlinear aspects of temporal integration of the visual signals for eye growth control have been identified. First, visual signals that slow ocular growth appear to have a greater effect on refractive development than signals that normally result in excessive ocular growth. For example, chick eyes exposed to successive, equal duration periods of myopic and hyperopic defocus exhibit increases in choroidal thickness, reduced axial elongation, and hyperopic shifts in refractive error.246,303305 Even when the duration of exposure to imposed hyperopic defocus was substantially longer than that for the myopic defocus, the signals generated by myopic defocus still dominated refractive development.246,304 By themselves, short daily periods of imposed myopic defocus were also sufficient to produce hyperopic shifts in animals who experienced unrestricted vision most of the day.295 Similarly, in chicks,162,295 tree shews,306 and monkeys,307,308 brief daily periods of unrestricted vision greatly reduced the axial myopia produced by imposed hyperopic defocus or form deprivation that was maintained for the rest of the day. Interestingly, the quantitative relationship between the daily duration of unrestricted vision and the relative reduction in myopia was very similar in these three species with only 2 hours of unrestricted vision, reducing the degree of FDM or LIM by approximately 80%.308 
The overall effects of both hyperopic and myopic defocus signals on refractive development depend on both the frequency and duration of daily exposure and not the total duration of exposure in a given day.246,309,310 When chicks are exposed to multiple brief periods of defocus throughout the day, with dark intervals between exposures, the compensating refractive changes were greater than those produced by a single-exposure period of the same total duration.246 Moreover, shorter duration but more frequent exposures were more effective than longer, less frequent exposures, as long as the total exposure duration was the same and the duration of each individual exposure period exceeded a critical duration. 
It has been argued that these nonlinearities come about because the compensating signals produced by defocus have relatively rapid rise times to saturation levels and that these signals decay more slowly between exposures.181,302,311 In a detailed study of the rise and fall times for individual exposures, Zhu and Wallman311 found that both myopic and hyperopic defocus produced near maximal choroidal thickness changes with exposure durations on the order of 5 to 7 minutes. The decay times for the defocus-induced changes in choroidal thickness were slower than the rise times, with the time required for the signals to decay to 50% of the maximum response being approximately twice as long for myopic versus hyperopic defocus (6.7 vs. 3.2 hours). On the other hand, temporal dynamics for axial growth changes were very different for myopic and hyperopic defocus, with response decay to myopic defocus being dramatically slower and more enduring than those to hyperopic defocus. The decay for hyperopic defocus was approximately 50 times faster than that for myopic defocus. These results indicated the existence of different signals for hyperopic versus myopic defocus and for compensating axial elongation versus choroidal thickness, and the results provide an explanation for the dominating effects of myopic over hyperopic defocus. 
The nonlinear temporal integration characteristics of the emmetropization process have important implications for efforts to determine the role of visual experience in the genesis of common refractive errors, such as juvenile-onset myopia. Specifically, the observed nonlinearities complicate the assessment of visual activities that may increase ocular growth and these nonlinearities may contribute to the inconsistencies related to the potential impact of near work on myopia.312 For instance, it is well-established that chronic hyperopic defocus promotes axial myopia in animal models, and it has been hypothesized that hyperopic defocus associated with underaccommodation during near work may promote the development of myopia in children,313,314 but there has been some disagreement in studies with children.315 Commonly used metrics of average near work in human subjects, such as the “diopter hour” (dioptric demand multiplied by hours spent at the near task),316 are only weakly correlated with myopia in children. These weak correlations may reflect the fact that these metrics do not take into consideration the manner in which different types of visual experience are integrated over time. Figure 12 considers the effects of providing infant monkeys reared with 3 D of imposed hyperopic defocus with four, 15-minute periods of unrestricted vision each day. With continuous lens wear, the average refractive error in animals reared with −3 D lenses is approximately 3 D more myopic than normal monkeys. Four daily 15-minute periods of unrestricted vision virtually eliminated this predictable compensation such that at the end of the treatment period, the average ametropia was not different from normal (although clearly, the pattern of refractive development in this group was different from normal). Using the control animals as a reference (i.e., 0-D hours), the monkeys that wore the −3-D lenses continuously and the lens-reared monkeys that had a total of 1 hour of unrestricted vision over the 12-hour daily lights-on cycle experienced, respectively, 36- and 33 D-hr/d of viewing conditions that would promote myopic growth. Considering the different outcomes for these two experimental groups, it is clear that diopter-hour units did not capture the critical aspects of visual experience that contributed to myopia. This is also supported by the very consistent protective effects reported for time outdoors against myopia.317,318 It likely reflects the fact that vision-dependent mechanisms that regulate refractive development are more sensitive, or more responsive, to stimuli that normally slow axial growth, making it easier to detect their influence on refractive development. 
Figure 12
 
Longitudinal changes in spherical equivalent refractive errors for the right eyes of infant rhesus macaque monkeys reared with binocular −3 D lenses. The monkeys represented in panel A wore the lenses continuously throughout the daily 12-hour lights-on cycle. For the monkeys represented in panel B, the −3 D lenses were removed for four 15-minute periods during the daily 12-hour lights-on cycle. The black lines in the upper plots show data from normal infant monkeys. The schematic in the lower left (C) shows the times when these animals were allowed unrestricted vision. The lower right plot (D) compares plotted as mean end-of-treatment ametropias for normal monkeys and the two experimental groups of monkeys (adapted from Kee C-S, Hung L-F, Qiao-Grider Y, et al. Temporal constraints on experimental emmetropization in infant monkeys. Invest Ophthalmol Vis Sci. 2007;48:957–962. Copyright © 2007 The Association for Research in Vision and Ophthalmology, Inc.).307
Figure 12
 
Longitudinal changes in spherical equivalent refractive errors for the right eyes of infant rhesus macaque monkeys reared with binocular −3 D lenses. The monkeys represented in panel A wore the lenses continuously throughout the daily 12-hour lights-on cycle. For the monkeys represented in panel B, the −3 D lenses were removed for four 15-minute periods during the daily 12-hour lights-on cycle. The black lines in the upper plots show data from normal infant monkeys. The schematic in the lower left (C) shows the times when these animals were allowed unrestricted vision. The lower right plot (D) compares plotted as mean end-of-treatment ametropias for normal monkeys and the two experimental groups of monkeys (adapted from Kee C-S, Hung L-F, Qiao-Grider Y, et al. Temporal constraints on experimental emmetropization in infant monkeys. Invest Ophthalmol Vis Sci. 2007;48:957–962. Copyright © 2007 The Association for Research in Vision and Ophthalmology, Inc.).307
3.6.3 Effects of Simultaneous Competing Defocus Signals
Competing myopic and hyperopic defocus can occur simultaneously for superimposed objects. More importantly, virtually all current optical treatment strategies for myopia produce simultaneous competing defocus signals. In particular, multifocal lenses (especially contact lenses) and corneal reshaping therapy or orthokeratology frequently produce spatially superimposed, simultaneous competing image planes across all or a large proportion of the retina. How these visual signals, which compete to increase and decrease axial growth, are integrated determines the overall direction of refractive development and the effectiveness of any optical treatment strategy. To study the effects of simultaneous, competing defocus signals on emmetropization, chicks,319 guinea pigs,320 marmosets,321 and rhesus monkeys322,323 have been reared wearing lenses with concentric annular zones with alternating refracting powers. These dual-focus lenses established two competing image planes across the entire retina. 
In chicks and guinea pigs, as illustrated in Figure 13, the compensation mechanisms of dual-focus lenses appear to direct refractive development toward either the average imposed defocus or to a refractive state slightly more hyperopic than the average. These results suggest that the vision-dependent mechanisms that regulate refractive development identify the effective sign and magnitude of defocus associated with each focal plane. They either average these signals in a linear manner (shown in guinea pigs320) or preferentially weight the image plane associated with the more positive powered lens component (shown in chicks319). This strategy is somewhat counterintuitive because the highest effective image contrasts would occur at the two secondary focal points associated with the lenses' two power zones, not at the dioptric midpoint. In marmosets reared with dual-focus contact lenses (+5-/−5-D power zones), the treated eyes developed a degree of hyperopia equivalent to that produce by +5-D single-vision lenses although, the degree of hyperopia did not completely compensate for the imposed myopic defocus.321 When the eyes of infant macaques were presented with two, approximately equally distinct focal planes, refractive development was directed toward the more myopic/less hyperopic focal plane and completely compensated for the more anterior foci.322 In all four species, the observed changes in refractive error were also associated with alterations in vitreous chamber elongation rate. There were, however, a number of methodologic issues that could explain the apparent differences between chicks, guinea pigs, and primates.322 
Figure 13
 
Effects of multifocal lens rearing. (A) Comparisons of the effects of dual focus, Fresnel-like lenses (50:50 area ratios) on refractive error development in rhesus macaques,322 chicks,319 marmosets,321 and guinea pigs.320 The left scale indicates the relative percentage change in ametropias at the end of treatment. For binocularly treated animals (rhesus monkeys), the ametropias for the right eyes are represented relative to that for control animals. For monocularly treated animals (all other species), the ametropias for the treated eyes are expressed relative to that of the fellow eye. Values of 0% and 100% indicate complete compensation for the most hyperopic and myopic image planes, respectively. Values of 50% indicate that the animals compensated for the average power of the dual focus treatment lenses (adapted from Arumugam B, Hung L-F, To C-H, Holden B, Smith EL III. The effects of simultaneous dual focus lenses on refractive development in infant monkeys. Invest Ophthalmol Vis Sci. 2014;55:7423–7432. Copyright © 2014 The Association for Research in Vision and Ophthalmology, Inc.).322 (B) The average ametropias for infant rhesus monkeys reared with dual focus Fresnel lenses (either +3 D and plano or −3 D and plano) plotted as a function of the percentage of surface areas that was devoted to the powered portions of the treatment lenses. Control monkeys reared with unrestricted vision are represented at the 0 point on the abscissa. Control monkeys reared with −3 and +3 D single-vision lenses are represented at the “100% −3 D” and “100% +3 D” positions, respectively. The dual-focus groups are positioned according to the proportion of lens surface areas devoted to the −3 and +3 D power zones (adapted from Arumugam B, Hung L-F, To C-H, Sankaridurg P, Smith EL III. The effects of the relative strength of simultaneous competing defocus signals on emmetropization in infant rhesus monkeys. Invest Ophthalmol Vis Sci. 2016;57:3949–3960. Licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.).323
Figure 13
 
Effects of multifocal lens rearing. (A) Comparisons of the effects of dual focus, Fresnel-like lenses (50:50 area ratios) on refractive error development in rhesus macaques,322 chicks,319 marmosets,321 and guinea pigs.320 The left scale indicates the relative percentage change in ametropias at the end of treatment. For binocularly treated animals (rhesus monkeys), the ametropias for the right eyes are represented relative to that for control animals. For monocularly treated animals (all other species), the ametropias for the treated eyes are expressed relative to that of the fellow eye. Values of 0% and 100% indicate complete compensation for the most hyperopic and myopic image planes, respectively. Values of 50% indicate that the animals compensated for the average power of the dual focus treatment lenses (adapted from Arumugam B, Hung L-F, To C-H, Holden B, Smith EL III. The effects of simultaneous dual focus lenses on refractive development in infant monkeys. Invest Ophthalmol Vis Sci. 2014;55:7423–7432. Copyright © 2014 The Association for Research in Vision and Ophthalmology, Inc.).322 (B) The average ametropias for infant rhesus monkeys reared with dual focus Fresnel lenses (either +3 D and plano or −3 D and plano) plotted as a function of the percentage of surface areas that was devoted to the powered portions of the treatment lenses. Control monkeys reared with unrestricted vision are represented at the 0 point on the abscissa. Control monkeys reared with −3 and +3 D single-vision lenses are represented at the “100% −3 D” and “100% +3 D” positions, respectively. The dual-focus groups are positioned according to the proportion of lens surface areas devoted to the −3 and +3 D power zones (adapted from Arumugam B, Hung L-F, To C-H, Sankaridurg P, Smith EL III. The effects of the relative strength of simultaneous competing defocus signals on emmetropization in infant rhesus monkeys. Invest Ophthalmol Vis Sci. 2016;57:3949–3960. Licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.).323
Dual-focus lenses complicate refractive development because both convergent and divergent rays are associated with both focal planes (i.e., both positive and negative defocus signals bracket both focal planes and to a lesser degree the dioptric midpoint between the two focal planes). The fact that the emmetropization mechanisms target the more anterior focal plane (or a point in front of the dioptric midpoint), has value from an evolutionary prospective because it reduces the likelihood that the eye will become myopic. In this respect, the eye responds to simultaneous competing defocus signals in a manner that is qualitatively similar to its responses to sequential competing defocus signals and the two focal planes associated with astigmatism (see Section 3.7.2). Moreover, this pattern of results obtained with dual-focus lenses indicates that multifocal lenses or correction strategies that impose simultaneous relative myopic defocus over a large part of the retina would be effective in slowing axial growth and reducing myopia progression in children. 
In terms of managing myopia, multifocal treatment strategies have some disadvantages. In particular with dual-focus lenses both power zones typically cover a portion of the pupil producing chronic retinal image degradation. This is potentially significant because even mild degrees of image degradation can produce FDM (see Section 3.2.1). However, it is important to note that the results from all four of the animal species that have been reared with dual-focus lenses (rhesus macaque, common marmoset, guinea pig, chick) revealed no signs that the resulting reduction in image contrast produced axial growth or a myopic shift. Nevertheless, depending on a number of lens parameters, dual-focus lenses can reduce the best-corrected visual acuity relative to traditional single-vision lenses, although it may be possible to reduce the saliency of the imposed myopic defocus and, thus, the impact of the imposed defocus on vision without losing the ability to control axial growth. Manipulating the relative surface area devoted to the two power zones of a dual-focus lens can alter the relative saliency of the two-image planes without altering the dioptric interval between them. In chicks319 and guinea pigs,320 the ability of the more positive-powered component of a dual-focus lens to control refractive development is influenced by the relative surface areas of the treatment lenses that are devoted to the two power zones (i.e., the relative amount of light contributing to each image plane). Specifically, decreasing the surface area devoted to the more positive-powered lens component shifts the target for emmetropization in favor of the more negative-powered component. However, the degree of relative myopia was always less than that produced by single-vision lenses of the same negative power. In guinea pigs and marmosets, the relative effectiveness of the two power zones in controlling axial growth appeared to be linearly related to the relative surface area of the lens associated with each power component.320,324 That does not appear to be the case in rhesus monkeys,323 although there were several methodologic differences between the studies, including binocular treatment in the rhesus study, which has important implications for human treatment of myopia. 
As illustrated in Figure 13, in infant macaques the surface area of a dual-focus lens devoted to the more positive-powered component can be reduced to one-fifth of a dual-focus lens' surface area without decreasing the ability of the more positive-powered component to reduce axial growth and produce relative hyperopic ametropias. Even when the saliency of the more posterior focal plane was much greater than for the more anterior focal plane, refractive development was still dominated by the relatively more myopic focal plane. From a lens-design perspective these results suggest that it may be possible to control myopia progression by imposing myopic defocus through a relatively small area of multifocal lenses, which should result in an overall improvement in central vision. In addition, this pattern of results indicates that as long as the imposed myopic defocus reaches threshold strength, the full treatment effect will prevail. If the strength of the myopic defocus signal, which is likely to be dependent on the magnitude of defocus and the amount of the lens' surface area devoted to the positive-powered component, does not reach this critical threshold, then there will be little or no treatment effects. In other words, the treatment effects will not be graded on an individual basis, but will likely follow an all-or-none rule.323 
3.6.4 Spatial Integration of Visual Signals Across the Retina
The existence of vision-dependent growth-regulating mechanisms that function in a regionally selective manner has important implications for refractive development, especially in primates with a foveal retina specialized for central vision. Because the refractive state at the fovea depends on ocular changes at the posterior pole and in the periphery (e.g., tangential scleral expansion in the periphery promotes central axial elongation), peripheral visual signals could, depending on the relative ability of mechanisms across the retina to alter scleral elongation, influence eye shape and central refractive development in a manner that is independent of central vision.325,326 
Little is known about the spatial integration properties of local growth regulating mechanisms. It would be valuable to know the size and effective sensitivity of the summation areas of these mechanisms and whether these properties change with eccentricity. Because cone photoreceptor density and resolution acuity are highest at the fovea, the fovea is the part of the retina that is most sensitive to optical defocus, and visual signals from the fovea largely control accommodation, it has historically been assumed that visual signals from the fovea would dominate axial growth and refractive development.327 However, several lines of evidence contradict this assumption. 
If visual signals from the fovea dominated refractive development, then eliminating these signals should alter visually directed ocular growth. But this does not seem to be the case because eliminating visual signals from the fovea by laser ablation of the central 8° to 10° of the retina in infant monkeys does not alter the course of emmetropization, the development of FDM, the ability of the eye to recover from experimentally induced ametropias, or compensation for imposed hyperopic defocus.328,329 It is likely, however, that foveal signals normally influence ocular growth (possibly in proportion to the absolute number of critical cascade elements in the fovea), but these results indicate that foveal signals are not unique or essential for many aspects of vision-dependent ocular growth and that the periphery, in isolation, can detect the presence of a refractive error and alter eye growth to eliminate the error. 
Moreover, when there are competing visual signals in the central versus the peripheral retina, experiments in chicks,330,331 marmosets,324 and macaques297,329 demonstrate that peripheral signals can dominate axial ocular growth and central refractive development. Figure 14 illustrates the effects of peripheral form deprivation and peripheral optical defocus on central refractive development.297,329 In all three subject groups, animals were viewing through treatment lenses with central apertures that allowed unrestricted central vision, but produced either form deprivation or optical defocus in the periphery. When viewing through these lenses, the central retina received visual signals that should have supported normal emmetropization, while the periphery experienced signals that normally result in either axial myopia (Fig. 14, panels A and B) or axial hyperopia (Fig. 14, panel C). Both peripheral form deprivation and peripheral hyperopic defocus produced central axial myopia; the range and average myopic errors were similar to those produced by full-field treatment lenses. Imposed peripheral myopic defocus slowed axial elongation producing central hyperopia and, interestingly, the degree of hyperopia was larger than that produced by full-field positive lenses. Presumably these higher degrees of hyperopia came about because, when viewing through the aperture lenses, the central retina controlled accommodation, which overcame the central compensating hyperopia while maintaining the peripheral myopic defocus (i.e., the peripheral signal to slow growth did not decrease as the eye developed central hyperopia). In contrast, with full-field positive lenses, the degree of myopic defocus in both the central and peripheral retina decreased as the eye developed compensating hyperopia. The ability of peripheral visual signals to override signals from the central retina can probably be attributed to the greater potential for spatial summation in the periphery. As suggested by Wallman and Winawer,181 although the density of many retinal neurons is highest in the central retina, the absolute numbers of neurons are higher in the periphery, simply because the peripheral retina is very large in comparison to the fovea. In addition, because of the geometry of the globe, small tangential expansions of the peripheral sclera would have a large effect on the axial position of the posterior retina.325,326 
Figure 14
 
The effects of imposing defocus on the peripheral retina. (A, B) Spherical equivalent refractive corrections obtained at ages corresponding to the end of the lens-rearing period for control monkeys (open diamonds) and monkeys reared with either diffusers ([A] adapted from Smith EL III, Kee C-S, Ramamirtham R, Qiao-Grider Y, Hung L-F. Peripheral vision can influence eye growth and refractive development in infant monkeys. Invest Ophthalmol Vis Sci. 2005;46:3965–3972. Copyright © 2005 The Association for Research in Vision and Ophthalmology, Inc.)329 or −3 D lenses ([B] adapted from Smith EL III, Hung LF, Huang J. Relative peripheral hyperopic defocus alters central refractive development in infant monkeys. Vis Res. 2009;49:2386–2392. Copyright © 2009 Elsevier Ltd.).822 The solid green and red symbols represent monkeys that worn treatment lenses that had central apertures that provided unrestricted vision for the central 24° to 32°. For comparison purposes, the half-filled diamonds represent monkeys that were reared with intact diffusers or −3 D lenses that altered vision across the entire field. The horizontal dashed line represents the average refractive error for the control monkeys; the solid lines denote ±1 SD from the control mean. (C) Changes in refractive error produced by rearing chicks with +5 D treatment lenses that had varying diameter central apertures that allowed unrestricted central vision (adapted from Liu Y, Wildsoet C. The effect of two-zone concentric bifocal spectacle lenses on refractive error development and eye growth in young chicks. Invest Ophthalmol Vis Sci. 2011;52:1078–1086. Copyright © 2011 Association for Research in Vision and Ophthalmology).330
Figure 14
 
The effects of imposing defocus on the peripheral retina. (A, B) Spherical equivalent refractive corrections obtained at ages corresponding to the end of the lens-rearing period for control monkeys (open diamonds) and monkeys reared with either diffusers ([A] adapted from Smith EL III, Kee C-S, Ramamirtham R, Qiao-Grider Y, Hung L-F. Peripheral vision can influence eye growth and refractive development in infant monkeys. Invest Ophthalmol Vis Sci. 2005;46:3965–3972. Copyright © 2005 The Association for Research in Vision and Ophthalmology, Inc.)329 or −3 D lenses ([B] adapted from Smith EL III, Hung LF, Huang J. Relative peripheral hyperopic defocus alters central refractive development in infant monkeys. Vis Res. 2009;49:2386–2392. Copyright © 2009 Elsevier Ltd.).822 The solid green and red symbols represent monkeys that worn treatment lenses that had central apertures that provided unrestricted vision for the central 24° to 32°. For comparison purposes, the half-filled diamonds represent monkeys that were reared with intact diffusers or −3 D lenses that altered vision across the entire field. The horizontal dashed line represents the average refractive error for the control monkeys; the solid lines denote ±1 SD from the control mean. (C) Changes in refractive error produced by rearing chicks with +5 D treatment lenses that had varying diameter central apertures that allowed unrestricted central vision (adapted from Liu Y, Wildsoet C. The effect of two-zone concentric bifocal spectacle lenses on refractive error development and eye growth in young chicks. Invest Ophthalmol Vis Sci. 2011;52:1078–1086. Copyright © 2011 Association for Research in Vision and Ophthalmology).330
Understanding the effects of peripheral vision on central refractive development is important because the eye's refractive state varies with eccentricity332334 (i.e., the signal for ocular growth varies with eccentricity). It has been known for some time that myopic eyes, due to their relatively prolate shape, exhibit less myopia (relative peripheral hyperopia) in the periphery, but whether this is a cause or an effect of axial myopia is unclear. Studies of eye shape in form-deprived monkeys indicate that relative peripheral hyperopia can be a consequence of vision-induced axial myopia.335 Several recent studies have not found peripheral refractive state to be a useful predictor for either myopia onset or progression,336,337 suggesting it is not a major factor in myopia development. However, none of those studies have looked at refraction beyond 30° off-axis, and cannot rule out the possibility that integration of the defocus signals off-axis may be involved in the progression of myopia. The apparently weak predictive value of peripheral refraction inside of 30° for myopia onset does not exclude a role for peripheral defocus signals in the control of eye growth, which can be exploited as a treatment strategy. Experimental and clinical studies both support this approach.326,338 Whether or not peripheral refractive state is a factor in the onset, or progression, of myopia, the fact that imposed defocus in the retinal periphery can affect axial refractive state is useful for myopia control and an important consideration in optical correction strategies for myopia. 
3.6.5 Age-Related Changes in Susceptibility to Visual Experience and Sensitive Periods for Myopia
In cold-blooded vertebrates, such as teleost fish, the eye continues to grow throughout their lifespan,339 and myopia can also be induced experimentally throughout the lifetime in these species.100,101 However, in warm-blooded vertebrates, the ability of visual experience to alter ocular growth and refractive state declines with age. In this respect, emmetropization can be considered to proceed in two phases. The “initial infantile phase” occurs during infancy and is characterized by a reduction in refractive error and a decrease in the variability of refractive state. As described in Section 3.1 and illustrated in Figure 15, many young animals are born with refractive errors; the emmetropization process rapidly reduces the refractive error and moves the eye toward a near emmetropic refractive state. This has been observed in humans,340,341 rhesus monkeys,14,342 marmosets,153,343 tree shrews,31 guinea pigs,42 mice,46,344 and chicks.345 As shown in Figure 15, for example, tree shrews initially have variable, hyperopic refractions that become less hyperopic as the eye grows rapidly during the initial infantile phase of emmetropization. The refractive changes during this period of rapid eye growth are known to be due, in part, to the visual regulation of eye growth, and passive optical scaling of refractive error in growing eyes.134,175 Following the infantile phase of emmetropization, there is a much longer “juvenile phase” of emmetropization, where refractions are relatively stable at or near emmetropia, while the eye is still growing. Experimental studies with animals have demonstrated that the stability of refraction during this phase is achieved by visually guided feedback, and the eye remains able to respond to imposed defocus as shown in Figure 15. Form deprivation can also produce myopia in older chickens,130,175,346 and monkeys127,128,347 even when their eyes have reached adult or near-adult size. During the juvenile phase, the rate of response and the magnitude of the changes in refractive error produced by visual experience decline with age. In addition, due to the nature of the ocular component changes produced by visual experience, the ability to compensate for positive power lenses declines much more rapidly with age.200 It is unknown, however, whether the visual control of eye growth is ever fully lost. 
Figure 15
 
Emmetropization and experimentally altered refractive state in tree shrew. (Top) Refraction as a function of days after eye opening (days of visual experience, or DVE). Each line is for an individual animal. Data for untreated animals and red light are shown as the average of both eyes; data for −5 D lens animals are for the treated eye only. Untreated tree shrews were raised under fluorescent colony lighting (data from Gawne et al.467), red light animals were exposed to ambient narrow-band red light stating at 95 DVE,467 and −5 D lens animals wore a monocular −5 D lens over one eye.823 Binocular +5 D lenses were worn starting at either 11 or 24 DVE.200 (Bottom) Axial length of the eyes as a function of time for the normal animals shown in the top panel.
Figure 15
 
Emmetropization and experimentally altered refractive state in tree shrew. (Top) Refraction as a function of days after eye opening (days of visual experience, or DVE). Each line is for an individual animal. Data for untreated animals and red light are shown as the average of both eyes; data for −5 D lens animals are for the treated eye only. Untreated tree shrews were raised under fluorescent colony lighting (data from Gawne et al.467), red light animals were exposed to ambient narrow-band red light stating at 95 DVE,467 and −5 D lens animals wore a monocular −5 D lens over one eye.823 Binocular +5 D lenses were worn starting at either 11 or 24 DVE.200 (Bottom) Axial length of the eyes as a function of time for the normal animals shown in the top panel.
3.7 Sign of Defocus and Nature of the Optical Signal
Animal studies have demonstrated convincingly that the fine-tuning of postnatal ocular growth to achieve and maintain emmetropia is actively controlled by visual signals related to defocus (see Sections 3.2.2 and 3.2.3 above). This regulation is primarily performed locally at the level of the retina acting on adjacent regions of sclera without much (if any) direct contribution from the central nervous system.158 The discovery that appropriate compensating growth occurs for equivalent degrees of imposed hyperopic and myopic defocus even when accommodation and all obvious behavioral cues are excluded,348 and that local bidirectional compensation occurs when the defocus is imposed over only half of the retina,188 indicates that the local retinal emmetropizing mechanisms can correctly identify the sign of defocus (i.e., whether the defocus is myopic or hyperopic). From an operational perspective, signals encoding the sign of defocus are ideal for regulating emmetropization. 
However, it has been difficult to determine precisely what visual cues are used to determine the appropriate direction for the emmetropization response. In part, this is due to the fact that there are a surprisingly large number of visual cues that could potentially be used. Also, the emmetropization mechanism might use multiple visual cues, and integrate these cues in complex nonlinear ways. Understanding how the emmetropization process encodes the sign of defocus is critical for understanding the role of vision in the genesis of common refractive errors and for optimizing treatment strategies. 
3.7.1 Longitudinal Chromatic Aberration
Experimental data suggest that signals derived from longitudinal chromatic aberration (LCA) provide directional cues for accommodation,349353 and there is increasing evidence that the same is true for emmetropization. LCA occurs because refractive index varies inversely with the wavelength of light; short-wavelength blue light is refracted more strongly than long-wavelength red light. Consequently, in polychromatic lighting color fringes occur around retinal images that change with the eye's refractive state, providing a chromatic signal that can be used to identify whether defocus is hyperopic or myopic.354 Specifically, when the eye is hyperopically defocused the red components of the retinal image will be more blurred than the blue components. When the eye is myopically defocused, the blue components of the retinal image will be more blurred than the red components. LCA is robust and consistent across individuals and species, it is relatively constant as a function of eccentricity, and the magnitude of LCA in diopters is unaffected by changes in pupil size or accommodation, making it a useful signal for guiding emmetropization. 
Experiments in chicks have identified several strategies involving LCA that can be used by emmetropization.352355 These sign-detecting strategies are based on contrast signals and are potentially more robust than strategies based on simple comparisons of relative cone-excitation levels as contrast signals are independent of the color of the illuminant. Rucker and Wallman352 analyzed the impact of simulations of chromatic contrast signals on emmetropization, which were similar to those that have previously been shown to drive reflex accommodation in the appropriate direction.350,356,357 Chicks were exposed to grating patterns in which the spatial contrast of the red and blue components of a printed image of black and white sine-wave gratings (3 and 5 cyc/deg) were modified to simulate myopic and hyperopic defocus. The results showed that eyes exposed to these grating simulations produced the predicted sign-dependent growth responses. When the blue component of a black/white bar pattern was blurred, and the red component was clear, indicating the eye was too long, the rate of axial growth was reduced. Conversely, when the printed bar pattern had the red component blurred, and the blue component clear, the rate of axial growth increased. 
Rucker and Wallman353 also revealed that changes in the eye's focus over time produced differences in the pattern of luminance and color contrasts (providing a temporal signal). Specifically, they showed that when the degree of hyperopic defocus decreases over time (as would occur during emmetropization), luminance contrast increases in conjunction with increases in the contrast in the M- and L-cone mechanisms. However, depending on the level of defocus, the contrast signals in the S-cones will decrease (i.e., decreasing hyperopic defocus produces an increase in luminance contrast and in the balance of chromatic contrast for the S-cone versus the M- and/or L-cone contrast mechanisms). In the case of increasing myopic defocus over time, the L- and M-cone luminance contrast signals decrease, but now the reductions in contrast for the S-cones and M- and L-cones are similar (i.e., the balance of chromatic contrast between S-cone and M- and/or L-cone components does not change over time). This analysis showed that the eye could theoretically detect the sign of defocus by detecting the presence or absence of a temporal chromatic signal across cone channels. A key feature of this idea is that as the eye grows toward emmetropia, the temporal chromatic signal will diminish until a point is reached when the contrast for all three cone types is approximately equal. Growth beyond this point would result in diminished luminance contrast of the retinal image without change in color contrast. Most importantly, flickering stimuli that simulate these two different scenarios produce predictable changes in ocular growth in young chicks.353 Specifically, to test this hypothesis, chicks were exposed to light that was modulated to produce changes in color or luminance contrast.353 The red, green, and blue components of a light-emitting diode (LED) were modulated in-phase to produce changes in luminance and in counterphase to produce red/green or blue/yellow changes in color. This experiment was performed at 2 Hz, in the middle of the range of temporal sensitivity of the chick,358 and with close to 100% contrast. The results showed that after 3 days of exposure to these lighting conditions, luminance flicker produced hyperopic shifts in refraction, while color flicker produced myopic shifts in refraction. These results support the hypothesis that the eye can use temporal signals associated with LCA for emmetropization. 
Luminance contrast modulation alone could signal when the eye is in focus because in most natural scenes an in-focus image would produce high temporal frequency luminance modulations, while blurred retinal images would produce low-temporal frequency luminance modulations.359,360 To test this idea, Rucker et al.361 exposed chicks to LEDs that produced 80%, white-light luminance modulation. Eye growth was reduced at 10 compared with 0.2 Hz. The experiment was repeated in yellow light (to simulate a “warm white” indoor illuminant). In this case, chick eyes exposed to 5- and 10-Hz stimuli grew less, as in white light, while chick eyes that were exposed to lower temporal frequencies (0.2, 1, and 2 Hz) grew more. These results suggest that the eye responds to rapid changes in luminance contrast by slowing growth, regardless of the color of the light. High temporal frequency stimulation indicates that the eye is in focus, halting growth. The results also indicated that yellow light promotes increased eye growth at low frequency temporal stimulation, when sensitivity to luminance modulation is reduced. At low temporal frequencies the eye seems to be able to detect the myopic wavelength defocus of the blue component of a white light source, thus reducing eye growth. 
Supporting the hypothesis that contrast is a critical variable underlying the effects of luminance modulation on eye growth, Rucker et al.361 found that absolute temporal contrast had a significant influence on the ability of temporally modulated stimuli to alter eye growth. At high contrast (>70%), high-temporal frequency stimulation slowed eye growth, but at lower contrast levels eye growth increased regardless of the temporal frequency or color. In other words, high-temporal contrasts, arising from an in-focus retinal image, are necessary for luminance modulation to slow eye growth. Other viewing conditions that induce high-contrast stimulation of the retina, such as high frequency, high contrast, stroboscopic, or sinusoidal flicker reduce eye growth in the chick.137,310,353,359 In fact, experiments overwhelmingly show that high temporal137,310,359,361,362 and spatial contrasts157,170,363366 are required for the eye to slow its growth and prevent myopia. 
Nevertheless, there is currently no consensus as to exactly how LCA cues are used for emmetropization, and indeed experiments using different wavelengths of light in different species have reported different results that are difficult to reconcile (see Section 4.2). Still, it is clear from many experiments across several species that changing the visible wavelength content of the environment can have significant effects on eye growth and refractive state, and it seems likely that chromatic cues are important for emmetropization. 
3.7.2 Higher-Order Monochromatic Aberrations
While spherical optical power and astigmatism (see Section 3.7.3) dominate the optical characteristics of the eye, the optical quality of the retinal image is influenced by a number of higher-order monochromatic aberrations (HOAs), which are related to the shape and configuration of the eye's optical components. All eyes have HOAs; however, there are large interindividual differences in the magnitude and specific characteristics of HOAs. Spherical aberration, coma, and trefoil are the most commonly studied individual HOAs in visual optics. The overall effect of all HOAs taken together is often considered for an optical system. Animal studies have provided important insights into the changes in HOAs that take place during emmetropization and during or the development of vision-induced refractive errors, the potential role for vision in improving the eye's aberrations, and the potential role of HOAs in the genesis of myopia. 
During the rapid postnatal infantile phase of ocular growth and emmetropization, there are substantial changes in the eye's optical and axial components that could influence the type and magnitude of HOAs. In particular, changes in the curvature of the cornea and lens and in the refractive index and thickness of the lens not only influence the eye's refractive status, but also alter the characteristics and magnitude of HOAs. Because HOAs influence retinal image quality and thus potentially the set-point and efficacy of emmetropization, it is important to understand the developmental changes that take place in HOAs. Cross-sectional studies show that, on average, HOAs are 20% to 50% greater in children than in adults.367,368 Longitudinal studies in chicks,369372 marmosets,373 and rhesus monkeys374 have confirmed that HOAs are greater in neonates and decrease in magnitude in a monotonic fashion during emmetropization. Although eye growth models can account for age-dependent improvements, the observed improvements in HOAs appear to exceed predictions based on a geometric increase in the overall scale of the eye associated with the normal increases in axial length. In each of these species the resulting optical quality of adult eyes is nearly diffraction limited. In infant monkeys, age-dependent improvement in the modulation transfer function associated with this decrease in HOAs play a limited role in the improvement in the spatial contrast sensitivity of infant monkeys; HOAs have a much smaller impact on behavioral performance than spherical and astigmatic defocus.374 
When results for different species are calculated for constant numeric apertures, the magnitude of HOAs in young animals is similar in chicks, marmosets, and rhesus monkeys.374 However, the characteristics of HOAs in these species are different, presumably reflecting interspecies differences in eye shape. For example, whereas the majority of humans and infant rhesus monkeys exhibit positive spherical aberration, marmosets exhibit negative spherical aberration and young chicks exhibit little or no spherical aberration. 
Several observations suggest that there is a link between myopia and HOAs. Many,375378 but not all studies,379,380 have reported that myopic humans have higher HOAs than nonmyopes. Because emmetropization is a vision-dependent process, it has been hypothesized that HOAs could promote the development of myopia in several ways. First, the chronic blur associated with HOAs could degrade the retinal image sufficiently to produce FDM.375,377,378 It is well established that chronic retinal image degradation promotes myopia in a graded manner. Even though retinal image degradation due to HOAs is usually small, the magnitude of HOAs is relatively constant over time, which increases the likelihood that a myopiagenic stimulus could produce axial elongation. HOAs could, by interacting with the eye's spherical ametropia, also alter the effective end point of emmetropization,381 and by increasing the eye's depth of focus, HOAs could result in greater variability in refractive errors.382 
With respect to the relationship between refractive errors and HOAs, studies in chicks,369372 marmosets,373 and rhesus monkeys,383 have demonstrated that viewing conditions that promote myopic growth, both form deprivation and optically imposed hyperopic defocus, also promote the development of larger than normal amounts of HOAs. The pattern of HOAs in ametropic eyes varies some between species; whereas ametropic rhesus macaque eyes showed larger amounts of positive spherical aberration and chicks and marmosets showed more negative spherical aberration. The alterations in HOAs observed in rhesus monkeys374 with experimentally induced ametropia were comparable to those observed in myopic humans.376,377 
The ocular changes responsible for the elevated levels of HOAs in eyes with experimentally induced ametropias are not well understood. Priolo et al.384 found larger than normal amounts of spherical aberration in the isolated crystalline lenses from chick eyes with FDM, which were attributed to changes in refractive indices of the lens. However, many of the observed changes in HOAs probably reflect changes in the shapes and relative positions of the eye's optical components. While vision-induced spherical refractive errors are primarily the result of alterations in vitreous chamber elongation rate, the expansion of the globe is not symmetric in either humans or monkeys. In particular, nasotemporal asymmetries are common in myopic eyes and could affect the shape of the crystalline lens or its alignment with respect to the cornea.335 Changes in lens alignment and tilt could explain the alterations in coma and trefoil observed in ametropic monkeys.383 
Several observations in animals with experimentally induced refractive errors suggest that changes in HOAs are a consequence rather than a cause of myopia. In rhesus monkeys, increased HOAs were found in both myopic and hyperopic monkeys and the patterns of HOAs were similar to those described in human ametropias.383 Every monkey eye that had elevated HOAs also had significant spherical and/or astigmatic refractive errors and the amount of HOAs were positively correlated with degree of axial ametropia (both myopic and hyperopic). Elevated HOAs did not prevent recovery from experimentally induced refractive errors, indicating that higher levels of HOAs do not prevent the eye from responding to the defocus signal.176,198 This is probably not surprising because when expressed in terms of equivalent spherical defocus, the magnitudes of HOAs observed in ametropic monkeys represent increases in defocus of less than 0.17 D.383 
There is little support for the hypothesis that the age-dependent reductions in HOAs observed during postnatal emmetropization are mediated by vision-dependent mechanisms. In chicks and primates with experimentally induced refractive errors there were concomitant decreases in the total HOAs over the treatment period.370,372,373,383 Although these reductions in HOAs were smaller than those observed in untreated eyes, it is clear that a significant part of the early decrease in HOAs occurs passively and is independent of the visual experience. In this respect, there are also numerous examples of treated marmoset373 and rhesus macaque eyes383 that experienced substantial defocus or form deprivation, but showed HOA patterns that did not differ from controls. So, if there are vision-dependent mechanisms that optimize HOAs, they can function normally in the presence of highly degraded retinal images. It is more likely that the decrease in HOAs associated with emmetropization or in experimentally treated animals over time reflect passive changes associated with growth, such as those described by Artal et al.385 
3.7.3 Astigmatism
Astigmatism is a type of refractive error that results from irregular curvature of the cornea or lens, or from the way the optics of these elements are combined. In children, the prevalence and degree of astigmatism is high during early infancy and generally decreases to adult levels before school age.386388 However, astigmatism is frequently associated with spherical ametropias; both children and adults with high amounts of myopia or hyperopia also frequently exhibit high amounts of astigmatism.389392 While studies of astigmatism in animals have been somewhat limited (for a review see Kee393), the results do provide insight into the causes of astigmatism and the question of whether astigmatism interferes with emmetropization. 
As in humans, studies with chicks show the magnitude of astigmatism is higher at birth/hatching and decreases with age.394396 The magnitude and axis of astigmatism found in chicks varies between studies (possibly reflecting strain differences), with Schmid and Wildoset395 reporting the largest astigmatic errors of approximately 8 D at hatching. The amount of refractive astigmatism found in chicks and the observed decrease with age are correlated with changes in the direction and magnitude of corneal astigmatism.395,396 Significant astigmatism is much less prevalent in infant macaques but, as in chicks and humans, when it exists it is primarily due to corneal toricity.397 In chicks, the fact that visual manipulations that enhanced corneal growth resulted in less astigmatism, but those that inhibited corneal development produced more, suggests astigmatism early in life is linked to anterior chamber development.395 
Studies involving chicks and monkeys investigated the possibility that astigmatism is regulated in a vision-dependent manner like emmetropization. There is some evidence chicks can compensate for imposed astigmatic errors. Irving et al.398 and Chu and Kee399 found partial compensation for astigmatism in chicks reared with cylinder lenses. The magnitude of compensation varied with the axis of the cylinder lenses (and possibly the power); however, there was disagreement between these studies in terms of the axis of the cylinder lenses that produced the largest compensating changes. The compensating astigmatic errors were attributed, in part, to alterations in corneal toricity, but Chu and Kee399 also reported significant correlations with a variety of eye-shape parameters. On the other hand, Schmid and Wildsoet395 found no evidence that chicks were able to compensate for imposed astigmatic focusing errors. Rearing rhesus macaques with cylinder lenses can produce significant amounts of astigmatism that is corneal in nature. However, regardless of the axis of the imposed astigmatism, the axis of the ocular astigmatism in monkeys was always oblique and in most cases was not in the appropriate direction to compensate for the imposed error and in some cases actually compounded the imposed astigmatic errors.400 Thus, while visual experience can alter corneal shape and produce astigmatic errors in monkeys, there is little evidence for a visually guided mechanism that minimizes astigmatic errors. 
Can the presence of astigmatism influence emmetropization? It has been hypothesized that astigmatism could disrupt emmetropization in a manner analogous to form deprivation.391,401 Like form deprivation, astigmatism degrades the retinal image and cannot be eliminated by either changing viewing distances or via accommodation. Studies in both chicks and monkeys do indicate that astigmatism can alter the course of emmetropization; however, there is little evidence that astigmatism promotes myopia. In chicks reared with optically imposed astigmatism, spherical emmetropization appears to target either the circle of least confusion,398 a point slightly in front of the dioptric midpoint,402 or the more myopic principal meridian.395 In macaque monkeys reared with optically imposed astigmatism, regardless of the axis, emmetropization was not directed toward the circle of least confusion, but toward one of two focal planes associated with the astigmatic principal meridians, most commonly the more anterior focal plane (i.e., astigmatism usually promoted relative hyperopic shifts).403 The pattern of results in monkeys suggests that the emmetropization process is insensitive to stimulus orientation and was targeting the image planes that contained the maximum effective contrasts. 
There is evidence from animal models that visual manipulations that produce axial hyperopia or myopia also produce significant astigmatic errors. For example, in chicks, lens compensation to either positive or negative lenses is frequently accompanied by astigmatism.194,370,396 The astigmatism is due to changes in corneal toricity. The axis has been reported to be either predominately against-the-rule396 or oblique,370 and the magnitude of astigmatism is significantly correlated with the degree of spherical ametropia.396 In rhesus monkeys, astigmatism has also been observed to accompany FDM and both lens-induced hyperopia and myopia. These astigmatic errors, which were more frequently associated with large ametropias, especially high hyperopia, were corneal in nature, oblique in axis, bilaterally mirror-symmetric in binocularly lens-reared animals, and reversible.286 The results from both chicks and macaques suggest that these induced astigmatic errors are the passive consequence of altered axial growth, possibly as a result of vision-dependent changes in the shape of the globe that take place during axial elongation.393 The association of astigmatism with spherical ametropias observed in humans could reflect a similar process. 
4. Effects of Ocular Circadian Rhythms and Light on Eye Growth and Myopia
Experimental data suggest that the emmetropization process is influenced by the lighting parameters in which animals are reared. Specifically, the duration,95 rhythmicity,404 spectral composition,354 and intensity of the ambient lighting405407 can alter ocular growth and refractive development. Each of these areas have been comprehensively reviewed recently,354,404407 and some key points are addressed below. 
4.1 Diurnal Light Cycles and Ocular Circadian Rhythms
Raising chicks in constant light or constant darkness cause excessive ocular elongation and flattening of the cornea, which combine to alter refractive state.95,99,137,173,408