*All authors contributed equally to this paper.
Can the retina alone detect the sign of defocus?
Article first published online: 12 MAY 2013
Ophthalmic & Physiological Optics © 2013 The College of Optometrists
Ophthalmic and Physiological Optics
Special Issue: Understanding & Controlling Myopia - Where We Are Now. A compilation to honour the research achievements and mark the passing of Josh Wallman
Volume 33, Issue 3, pages 362–367, May 2013
How to Cite
Point-counterpoint. Can the retina alone detect the sign of defocus?. Ophthalmic Physiol Opt 2013, 33: 362–367.& .
- Issue published online: 12 MAY 2013
- Article first published online: 12 MAY 2013
In the following point-counterpoint article, internationally-acclaimed myopia researchers were challenged to defend the two opposing sides of the topic defined by the title; their contributions, which appear in the order, Point followed by Counterpoint, were peer-reviewed by both the editorial team and an external reviewer. Independently of the invited authors, the named member of the editorial team provided an Introduction and Summary, both of which were reviewed by the other members of the editorial team. By their nature, views expressed in each section of the Point-Counterpoint article are those of the author concerned and may not reflect the views of all of the authors.
At the population level, ‘emmetropisation’ describes the shift in refractive errors towards emmetropia or low hyperopia seen during infancy, from widely dispersed starting points encompassing both high myopia and high hyperopia. At the level of the individual, emmetropisation describes the maintenance of an optimal refractive error during the early post-natal stages of eye maturation, or the transition from a sub-optimal refractive error at birth towards a more optimal refractive error during infancy, guided by visual feedback. This system of visually-guided refractive development has fascinated vision scientists. Firstly, it offers plausible explanations for why refractive errors are found in adults: e.g. ‘emmetropisation fails’ or, ‘the emmetropisation system purposefully directs the eye towards myopia driven by misleading visual cues from our 21st Century environment’. Secondly – and possibly independently – it offers the promise of effective ways of correcting or preventing refractive errors: e.g. ‘use nature's own system to guide refractive development away from myopia’. Thus, understanding how emmetropisation works is seen as an issue of central importance. Here, two pioneers in the field of myopia research, Frank Schaeffel and Christine Wildsoet, debate the key question, ‘Can the retina alone detect the sign of defocus?’
A fundamental question in emmetropisation is whether the retina by itself can perform the image processing necessary to derive the sign of defocus without feedback from the brain. It is well documented that optical degradation of the retinal image by lid suture or diffusers in front of animal eyes, or by pathological conditions in humans like early cataract, keratitits, ptosis or blepharospasms cause excessive axial eye growth and ‘deprivation myopia’. Since the retinal image is also degraded by defocus, no matter of which sign, one would expect that deprivation myopia would develop in both cases – but this is not true. Experiments in Josh Wallman's laboratory and in our own laboratory have shown that the retina can inhibit axial eye growth when the focal plane is on the vitreal side the photoreceptor layer. Therefore, there must be more than the mechanism of deprivation myopia, at least in the chicken. If this would be true also in primates, undercorrection of myopia should be no risk since the focal plane would also be in front of the retina, at least for far vision. After the classical findings by Wallman et al. that eye growth can be locally stimulated by local degradation of the retinal image, even after the optic nerve was cut,[2, 3] it was clear that the retina has at least the complete machinery to convert image features into growth signals. It would even be possible to emmetropise the eye just by quantifying deprivation over time since a hyperopic eye has less focus on average and should grow more than a myopic eye, assuming similar accommodation tonus – but it can be seen below that this is not what happens. Sigrid Diether also found that hemifield spectacle lenses can generate local changes in eye growth, affecting only the defocused half of the eye. More importantly, these responses were also bi-directional, depending on the sign of the lenses. Accommodation in chicks shifts the focal plane uniformly all over the visual field, like in most vertebrates, except fishes. This excludes that the local growth effects were mediated by rotational asymmetries of accommodation and suggests that the retina can determine the sign of defocus – although this could still be done by comparing the focus in the different areas of the visual field. Perhaps the most convincing experiment to show that the retina can extract the sign of defocus was done both by Josh Wallman's laboratory and by Sigrid Diether.[7, 8] Chicks individually placed in the centre of a large vertical drum which permitted only one viewing distance wore either strong positive or negative lenses, and, in addition, accommodation was paralyzed. The lens power was chosen so that the far point of the chicks was in front or behind the retina by about the same dioptric amount. Accordingly, the image was similarly low pass filtered and, if only the spatial frequency content would be important, the animals should all have become myopic. But, in fact, their eyes grew still less with positive lenses and more with negative lenses – difficult to explain without assuming that the retina can detect the sign of defocus. Later, Michaela Bitzer found that also the regulation of the transcription factor ZENK in the retinae of these chicks was controlled by the sign of defocus in the drum experiment, indicating that the directional signal was present at the molecular level as well. The findings are in line with the finding from the Wallman laboratory that there is a powerful inhibition of eye growth when the image plane is on the vitreal side of the retina.
It is unsatisfactory that the image processing algorithms used by the retina to extract the sign of defocus from the image remain unclear. Several potential candidate cues have been tested and were found not to be obligatory. At least chickens compensated for spectacle lenses of different sign correctly when cues from chromatic aberration were removed by raising them in monochromatic light, although very careful studies by Frances Rucker and Josh Wallman have identified subtle effects of manipulations of longitudinal chromatic aberration. Monochromatic aberrations, including astigmatism, would be able to provide information about the sign of defocus. This is possible because they generate slightly different images on the retina for defocus of the same amount but with different sign. But chickens have very good optics and very little astigmatism and therefore these cues should be very weak. A few other cues, like differences in image magnification have also been excluded (review in ref. ). A potentially important observation was made by Michaela Bitzer. If anesthetized animals are placed in the laboratory in a stationary holder with spectacle lenses in front of their eyes and lid retractors so that they have visual exposure to the lab environment, no eye growth changes are observed subsequently if they are placed in the dark after recovery from anaesthesia. Two different anaesthetics with different pharmacological targets (ketamine: a non-competitive NMDA receptor blocker, and a mixture of Medetomidin and Domitor, a selective and specific alpha-2-receptor agonist and an enhancer of GABA, respectively) were tested to exclude that a specific action in the retina might have blocked the changes in ZENK expression and eye growth. Also, no directional changes in the expression of the transcription factor ZENK were found. A possible explanation is that miniature fixational eye movements, causing a continuously jittering or moving images on the retina, are necessary for the retina to detect the sign of defocus. In fact, a study by Greschner et al. showed that such eye movements strongly improve feature detection by ganglion cells in an in vitro preparation of the turtle retina.
The factor that makes it difficult to predict how the eye will grow is accommodation. Even if one takes the view that the retina can determine the sign of defocus, the accommodation tonus modifies the focus of the retinal image and therefore the output of the retinal image processor. In chicks it was found that accommodation does not fully compensate for the power of spectacle lenses placed in front of their eyes. This was concluded from the increase in suprathreshold contrast sensitivity that was found in chicks that wore lenses of both signs for one hour. Therefore, a focus error signal was available for the retina. The issue becomes less clear if chicks are placed in either red or blue light. They immediately compensate for the longitudinal chromatic aberration by accommodating one dioptre more in the red than in the blue. Nevertheless, after 2 days, refractive state as measured in complete darkness had also shifted in the same direction and by about the same amount as the prior shift in accommodation which can only be explained if accommodation was insufficient to fully refocus the image.
In summary, work in chickens clearly suggests that the retina can extract the sign of imposed defocus. Not all observations on human refractive development can be explained based on this finding, for instance (1) if an image in front of the retina tells the retina to inhibit axial eye growth, undercorrection should represent an effective way to inhibit myopia progression, but the results are not clear. Perhaps accommodation becomes ‘lazy’ with undercorrection, causing the focal plane to shift more to the other side during near work. This question needs to be studied in the future. (2) The same should apply to uncorrected myopia – it should be self-limiting but it is obviously not. Accommodation in myopes has been extensively studied and was found to be less accurate than in emmetropes, perhaps because myopes have more tolerance to poorly focused images (i.e. ref. 17). Therefore it could also be too weak during reading, causing the retina to stimulate more axial eye growth despite that the eye is already myopic.
Interest in the ability of the retina to encode changes in the visual environment and specifically, to decode the nature of defocus inherent in retinal images, has grown significantly in response to accumulating data indicating that myopic progression is slowed by orthokeratology (ortho-k), with similar trends reported for some concentric multifocal (MF) soft contact lens designs. The possibility that these treatment effects reflect a response of the peripheral retina to imposed myopic defocus rests on the assumption that the retina is able to decode the sign of defocus. Indeed, much on-going research into new optical treatments for myopia is now based on the premise that eye growth is bidirectionally and locally regulated. Yet are there other interpretations of these data? Is it really the case that optical defocus can be decoded by the retina alone, without central input? Two pieces of evidence, one indirect and one direct, point to a role for accommodation and thus central influences, and two other pieces of evidence argue against the ability of the retina to decode the sign of defocus. All studies involve the chick model.
Citation of original literature in this article is limited to those published too recently to be included in the four review papers by Smith, Flitcroft, Wallman and Winawer, and Wildsoet.
The first piece of evidence involves an experiment in which binocular negative lenses (−5 D), were fitted to birds that had undergone ciliary nerve section (CNX) to one eye.21 Because the chicks appeared to feed normally, it is assumed that they accommodated near continuously using their unlesioned eye, which thus would have experienced minimal optical defocus, while their CNX eye would have experienced continuous defocus. Nonetheless, both eyes compensated at approximately the same rate, as if accommodation activity was somehow integrated into the growth signal for the unlesioned eye. Nonetheless, proponents of the local retinal control model might argue that residual defocus, no matter how small, drives eye growth in this case.
The second piece of evidence comes from two closely related experiments in young chicks in which competing defocus stimuli of opposite sign were presented simultaneously;[22, 23] (see Figure 1). In both cases, the effect of imposed myopic defocus dominated ocular growth in eyes with intact accommodation i.e. eyes slowed their eye growth, while the opposite was true for CNX eyes. These results suggest that accommodation is essential to decoding imposed defocus stimuli. The presence of residual accommodation in cyclopleged eyes is offered as a potential explanation for seemingly contradictory data from the Wallman and Schaeffel laboratories in which chicks responded appropriately to similar blur conditions generated by defocus of opposite sign under controlled conditions. While the nature of the ocular growth regulatory circuit through which accommodation activity might be integrated remains to be established, observations of altered defocus responses in chicks after selective lesioning of the autonomic innervation to the eye may be of relevance.
The third and fourth pieces of evidence come from another study from the Schaeffel laboratory, in which chicks experiencing optical defocus only under general anaesthesia showed no response to imposed optical defocus, irrespective of its sign, and isolated retinal preparations exposed to optical defocus did not show the predicted defocus-related changes in ZENK expression. Thus both results argue against the notion that the retina alone can decode the sign of defocus.
While proponents of the retinal model of ocular growth control include data from studies involving optic nerve section (ONX) as supporting evidence, a closer inspection of these data reveals a more complicated picture. As just three examples, (1) eyes undergoing ONX surgery alone tend to be more hyperopic than normal and likewise eyes recovering from form deprivation myopia overshoot emmetropia to become more hyperopic than normal (e.g. ref 25), (2) ONX eyes in which lens wear is delayed to allow some stabilization of the retinal architecture show hyperopic biases in both their baseline and final refractions relative to unlesioned eyes even though compensatory changes in refractions approximately match the imposed defocus, and (3) ONX eyes fully or under-corrected after being made myopic show much greater variability in their refractions than similar treated normal eyes. Such studies are of interest because severing of the optic nerve also breaks the accommodative feedback loop.
All of the preceding cited studies involve the chick model. There is one relevant lesioning study involving the guinea pig in which ONX eyes showed accelerated growth in response to positive as well as negative lenses, implying that they could not decode the sign of defocus. A much earlier study in monkeys by Raviola and Wiesel also addresses the possibility of higher, non-ocular control of emmetropisation, although the seemingly implausible species difference in the effect of intracranial ONX, which prevented the development of form deprivation myopia in the stump-tail monkey but not rhesus monkey, warrants verification.
If the retina alone is not able to decode the sign of defocus then how can one explain the local ocular growth changes elicited by regionally localized stimuli? A model similar to an early one proposed by the Schaeffel lab is offered. The working assumption is that spatial features associated with an accurately focused image, i.e. high contrast and spatial frequencies, are strongly inhibitory to eye growth. Another critical determinant of output is likely the temporal pattern of exposure to different levels of defocus. Thus the transition from induced hyperopia to induced myopia at around +50 D when increasingly high powered lenses are fitted to chicks, presumably reflects some weighted average of sharp vision during feeding, a preoccupation of young chicks, and blurred vision for more distant objects. A recent review by Flitcroft is essential reading for understanding the optical complexities of the natural visual environment.
Finally, any argument against local retinal decoding of defocus must offer an explanation for the observations of bidirectional gene expression changes in the retina and retinal pigment epithelium related to the direction of imposed defocus, e.g. ref. 29 and 30. Could centrifugal fibres projecting from higher centres via the optic nerve onto amacrine cells, provide the critical missing piece of the puzzle? Such fibres are well developed in chicks, are known to modulate the activity of a subset of amacrine cells, and could offer a central control mechanism for regional growth control. Importantly, they are eliminated in ONX, which also reduces the expression of retinal ZENK, one of the bidirectionally modulated genes. Could their elimination explain the altered set-point of emmetropisation after ONX? Interestingly, lesioning of the centrifugal system at the level of the isthmo-optic nucleus, also induces hyperopia, albeit transiently. Nonetheless, this explanation leaves many questions unanswered, foremost of which is why emmetropisation is altered by ONX in the guinea pig when this tract is less developed in mammals and primates?
Jeremy A Guggenheim
Laboratory of Experimental Optometry,Centre for Myopia Research,School of Optometry,The Hong Kong Polytechnic University,Hong Kong SAR,China
Points of agreement
- With the majority of studies having been carried out using the chick model, the generality of many findings await confirmation in mammals, particularly mammals possessing a fovea.
- At some stage, the sign of defocus becomes encoded at the molecular level within the retina (e.g. in the level of ZENK expression in specific amacrine cell types). Thus, even in the absence of agreement that the decoding process is confined to the retina, this does at least provide a putative site for an early stage of the emmetropisation signalling pathway.
- Localised, sign of defocus-dependent changes in eye shape resulting from wearing lenses affecting only part of the visual field are very difficult to explain without invoking a retinal decoding mechanism.
Issues to be resolved
- Interpreting chick lens wear studies is greatly complicated by the potential influence of an active accommodation system. This has led researchers to try to disable accommodation, either pharmacologically or by nerve lesioning. Crucially, the results of lens wear experiments differ markedly depending on which method of accommodation nullification is employed. Disentangling the primary effect of disabling accommodation from any secondary effects of the drugs or lesions will be an important step forward.
- Following on from the above, since a fully-functioning accommodation system is the norm, how does emmetropisation succeed in decoding defocus cues so well when accommodation would seemingly either (1) dampen much of the blur experienced by the retina, which might be a problem if the retina is solely responsible for the decoding feat, or (2) have to be integrated into the decoding equation, if accommodation assists the retina in the decoding process?
- Lastly, and most importantly, there is much we still have to learn about defocus detection in emmetropisation. If the retina alone decodes defocus, what cues does it use? If the retina receives assistance in the process, what inputs are used and how do they somehow switch a unidirectional ‘I see blur’ signal produced by the retina from a grow to a stop message, depending on the sign of defocus?
- 18Optical treatment strategies to slow myopia progression: effects of the visual extent of the optical treatment zone. Exp Eye Res 2013; doi: 10.1016/j.exer.2012.11.019..
- 23Competing defocus stimuli of opposite sign produce opposite effects in eyes with intact and sectioned optic nerves in the chick. Invest Ophthalmol Vis Sci 2000; 1: S738.& .
- 25Retinal control of refractive error and eye growth regulation in the chick. PhD thesis, University of Queensland, 1992..
- 27Mammalian eyes need an intact optic nerve to detect the sign of defocus during emmetropisation. ARVO Abstract 2009; 50: E-Abstract 1620.& .
- 32Regulation of ZENK mRNA levels in the chicken retina by duration of light exposure, by image quality and by inter-ocular coupling. ARVO Abstr 2006; 47: E-Abstract 1146., , & .
- 33Centrifugal visual system influences early refractive development. ARVO Abstr 2012; 53: E-Abstract 3432., & .