Stimulus Requirements for the Decoding of Myopic and Hyperopic Defocus under Single and Competing Defocus Conditions in the Chicken

Abstract

purpose. The bidirectional nature of emmetropization, as observed in young chicks, implies that eyes are able to distinguish between myopic and hyperopic focusing errors. In the current study the spatial frequency and contrast dependence of this process were investigated in an experimental paradigm that allowed strict control over both parameters of the retinal image. Also investigated was the influence of accommodation.

methods. Defocusing stimuli were presented through lens-cone devices with attached targets. These devices were monocularly applied to 5-day-old chickens for 4 days. Defocus conditions included: (1) 7 D of myopic defocus, (2) 7 D of hyperopic defocus, and (3) a combination of the two. Two high contrast target designs, a spatially rich, striped Maltese cross (target 1) and a standard Maltese cross (target 2) were used, except in some experiments where target contrast or spatial frequency content was further manipulated. To test the role of accommodation, the treated eye of some chicks underwent ciliary nerve section before attachment of the device. Refractive error (RE) was measured by retinoscopy and axial ocular dimensions measured by A-scan ultrasonography, both in chicks under anesthesia.

results. With imposed myopic defocus and high contrast, target 1 elicited significantly better compensation than did target 2. With imposed hyperopic defocus, both targets elicited near normal compensatory responses. Reducing image contrast to 32% for target 2 and to 16% for target 1 precluded compensation for myopic defocus, inducing myopia instead. The low-pass–filtered target also induced myopia, irrespective of the sign of imposed defocus. With competing defocus and intact accommodation, target 1 induced a transient hyperopic growth response, whereas myopia was consistently observed with target 2. When accommodation was rendered inactive, both targets induced myopia under these competitive conditions.

conclusions. Compensation to myopic defocus is critically dependent on the inclusion of middle to high spatial frequencies in the stimulus and has a spatial frequency–dependent threshold contrast requirement. With competing myopic and hyperopic defocus, the former transiently dominates the latter as a determinant of ocular growth, provided that the stimulus conditions include sufficient middle to high spatial frequency information and that accommodation cues are available.

During postnatal development, the length of the vertebrate eye becomes matched to its focal length through a process referred to as emmetropization. Animal studies have provided convincing evidence that emmetropization is an active, visually guided process. For example, the chick eye rapidly compensates for refractive (focusing) errors (RE) imposed by spectacle lenses by appropriately adjusting its choroidal thickness ¹ and axial length. ^{1 2} Specifically, hyperopic defocus (imposed by negative lenses) induces choroidal thinning and accelerated eye elongation, whereas myopic defocus (imposed by positive lenses) induces choroidal thickening and inhibits eye elongation. This emmetropization process also exists in mammals, including monkeys, ^{3 4} although the range of compensation is reduced relative to that in chickens (rhesus monkeys, −3 to +6 D ⁵ ; chickens, −10 to +15 D ⁶ ). 

That eyes can detect and appropriately respond to both myopic and hyperopic defocus implies that they are able to distinguish the sign of the imposed defocus. However, we are still far from understanding how this is accomplished, even in terms of which features of the defocused retinal image are used by the eye to decode this sign information. Because of its relevance to myopia control—insights into this sign detection problem may allow control through manipulation of the visual environment—we targeted the stimulus requirements for emmetropization in the present study. 

Of relevant, already published studies, most relate to normal developmental emmetropization. Specifically, form deprivation experiments indicate that normal developmental emmetropization has both spatial frequency and contrast requirements. The devices used in such experiments (e.g., frosted, translucent diffusers), typically show low-pass filter characteristics, eliminating moderate to high spatial frequency information as well as reducing image contrast. The net result is the derailment of emmetropization, with increased axial elongation leading to myopia. ^{7 8} That eyes can recover from this induced myopia when normal vision is restored at a sufficiently early age represents a more direct example of emmetropization. Predictably, this recovery process can be prevented in chicks by low-pass filtering of the visual image. ⁹ In another relevant study, form-deprived chicks were exposed daily to brief periods of “normal vision” ¹⁰ ; manipulation of the spatial frequency information available during these exposures showed intermediate spatial frequencies (0.86 cyc/deg) to be more effective than either higher or lower frequencies in preventing the development of myopia. This spatial frequency dependence of emmetropization is similar to that reported for accommodation, another ocular focusing mechanism. ^{11 12}  

A limitation of the experimental paradigms used in the cited studies is the need to restrain the animal during visual manipulation. This imposes constraints on the duration of exposure. In the present study, we made use of a cone-shaped imaging system (lens-cone device) that allows strict and sustained control over the visual information presented as well as retinal image defocus. ^{13 14} We took advantage of its flexibility in allowing visual information to be presented in one or more planes, at different levels of defocus, with control over both spatial frequency and contrast. For some of these conditions, we also added ciliary nerve lesioning surgery by way of testing the influence of accommodation. 

The present study builds on the primary result of a recently published study by one of the authors indicating that compensation in response to defocused stimuli is directed by the imposed optical vergence in the absence of other cues to distance. ¹⁴ In the present study, we investigated the effect of manipulating the spatial frequency and contrast content of retinal images on the eye’s growth response to both single and competing defocus stimuli. We find that compensation for myopic defocus has both spatial frequency and contrast requirements. The inclusion of the competing defocus stimuli was intended to simulate better the conditions encountered in the natural environment and follows up on another study involving chicks wearing multifocal lenses that imposed defocus stimuli of the opposite sign; hyperopia was observed when accommodation was left intact, but myopia occurred when accommodation was prevented surgically (Wildsoet CF, et al. IOVS 2000;41:ARVO Abstract 3930). We report herein similar changes in the response bias under competing defocus conditions when accommodation was eliminated. 

Methods

Animals and Rearing

White-Leghorn chickens (Gallus domesticus) were used in this study, obtained as 1-day-old hatchlings from a commercial hatchery (Privett Hatchery, Portales, NM). They were reared under diurnal lighting conditions (12-hour light–dark cycle) with access to food and water ad libitum. The experimental treatments conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Animal Care and Use Committee of the University of California-Berkeley. 

Experimental Treatments

Table 1provides a summary of the specific details of each of the treatment conditions tested, including imposed defocus information, whether accommodation was left intact, and the number of birds assigned to each treatment. More details about the design of the cone devices and attached targets are provided in the following sections. In all cases, the devices were applied monocularly to 5-day-old chicks for 4 days. The untreated contralateral eyes served as controls. 

Lens Cone Devices.

The design of the lens-cone devices has been described in detail. ^{13 14} The cones were made from white, translucent polyethylene and were attached by Velcro ring supports. They were designed to provide a field of view of approximately 45°. In a pilot experiment comparing three different fields of view (approximately 30°, 45°, and 60°), the 45° field size was the best compromise between the need to avoid peripheral form deprivation on the one hand, and to provide uniform defocus across the target plane on the other. Accordingly, cones providing a field of view of approximately 45° were used in all subsequent experiments. Each cone was fitted with a positive lens (modified human PMMA contact lenses) at its proximal end and a target at its distal end. To avoid any visual experience beyond the target plane, each target was backed with an opaque white sheet that spanned the distal opening. The imposed defocus conditions were manipulated by varying either the lens power that determined the location of the focal plane and/or the target distance. In all, three defocus conditions were tested: 7 D of myopic defocus (Fig. 1A ; target located beyond the focal plane of the +40 D imaging lens), 7 D of hyperopic defocus (Fig. 1B ; target located within the focal plane of the +30 D lens), and a combination of myopic and hyperopic defocus (Fig. 1C , two targets located beyond and within the focal plane of a +40 D lens). We regularly monitored the chicks to ensure that the devices remained properly attached throughout each experiment. Also, to avoid unintended deprivation effects, we cleaned the imaging lenses daily. 

Targets.

We made use of two basic Maltese cross target designs that differed in their spatial frequency content (Fig. 2) . Target 1 (T1; striped target) was a black-and-white, 1.2 cyc/deg striped filler pattern, and target 2 (T2) was a black Maltese cross without any filler pattern. The spatial frequency selected for the filler pattern approximately matched the peak of the contrast sensitivity function of the chick. ¹⁵ Overall, target 1 was richer in high frequency information over the range detectible by the young chick eye (visual acuity, 6–8 cyc/deg). ¹⁵ Both of these target designs were prepared at five different spatial contrast (C) levels (100%, 65%, 32%, 16%, and 11%; C = [L _max − L _min]/[L _max + L _min]) to study contrast effects. We also manipulated the spatial frequency composition of the two basic target designs for some experiments. Specifically, we applied Gaussian filters, using the a commercial software program (Canvas 5, Deneba; ACD Systems International, Saanichton, British Columbia, Canada), to create two new digital versions of these two target designs, target 3 (T3), derived from target 1, and target 4 (T4), derived from target 2 (Fig. 2) . These target manipulations had the effect of significantly attenuating spatial frequency information above 1.3 cyc/deg for target 3 and above 0.2 cyc/deg for target 4. A high-resolution printer (Phaser 840 DP; Tektronix, Beaverton, OR) was used to generate hard-copy versions of the targets, which were presented on a white background, except when two targets were presented simultaneously. To ensure the visibility of both targets in the latter case, the nearer one was presented on a transparent background. We also narrowed the arms of the Maltese cross targets from 45° to 27° and offset one target with respect to the other by rotating it 45°. 

Surgery

In two of the groups exposed to competing defocus stimuli, the ciliary nerve of treated eyes was initially sectioned (CNX) to eliminate accommodative activity. This surgery was performed 1 day before the attachment of the cone devices, in chicks under halothane anesthesia (1.5% halothane in oxygen; for details of this procedure, see Ref. ¹⁶ ). 

Measurements

All measurements were performed with chicks under general anesthesia (1.5% halothane in oxygen). Refractive errors (REs) were measured by static retinoscopy. Data represent averages of readings obtained for the horizontal and the vertical meridians. Axial ocular dimensions were measured with high-frequency A-scan ultrasonography (for details, see Refs. ^{17 18} ). Axial length was defined as the axial distance from the front of the cornea to the inner surface of the sclera (i.e., corneal thickness + anterior chamber depth + lens thickness + vitreous chamber depth + retinal thickness + choroidal thickness). Baseline REs and ultrasonography data were collected before attachment of the cones, and these measurements were repeated on day 4 of the treatment period and, in the case of ultrasonography, also on treatment day 2. 

Data Analysis

Treatment effects are expressed as interocular differences. Data are presented graphically as either the mean (±SEM) of interocular differences (see Figs. 3 4 6 ) or the mean (±SEM) of derived changes in interocular difference over the 4-day treatment period (see Fig. 5 ). The stimulus-dependence of induced defocus responses was tested with two-way, repeated-measures ANOVA (Figs. 3 4 6)or two-way ANOVA (Fig. 5)for intergroup comparisons of interocular difference data. Post hoc testing (Tukey-Kramer honest significant difference [HSD]) was undertaken when appropriate. See also Tables 2 and 3 . 

Results

Spatial Frequency Effects on Compensation to Imposed Defocus

Response to Unfiltered Targets with Different Spatial Frequency Composition.

In this first set of experiments, either of two high-contrast targets, a striped Maltese cross (target 1), or a standard Maltese cross (target 2), was presented monocularly under either myopic or hyperopic defocus conditions. The now well-characterized responses of increased vitreous chamber elongation and choroidal thinning to hyperopic defocus and slowed vitreous elongation and choroidal thickening to myopic defocus were observed in both cases. Whereas all four groups showed significant increases in anterior chamber depth in the treated eyes (Fig. 3E) , there were differences related to the target design in the case of the response to myopic defocus (Fig. 3 , Table 2 ), as described in the following section. 

With imposed myopic defocus, the striped target (target 1), elicited significantly better compensation than the other (target 2). Notably, the inhibitory effect on vitreous chamber elongation was sustained over the 4-day treatment period for target 1, but was not maintained after day 2 with target 2. This difference was reflected in the progressive decrease in the vitreous chamber depth of treated eyes relative to that of fellow eyes for target 1, whereas for target 2 the vitreous chamber depth of treated eyes showed an early relative decrease, similar to that for target 1, followed by a slight regression. The choroidal thickness changes were also transient for target 2, with the early increases regressing after day 2. In contrast, interocular differences in choroidal thickness increased over the treatment period with target 1, more so over the first 2 days (Figs. 3A 3B) . Target-related differences in responses were statistically significant for both vitreous chamber depth and choroidal thickness (Table 2 ; two-way, repeated-measures ANOVA, choroidal thickness: significant target effect [P = 0.0047]), vitreous chamber depth: significant target × time interaction [P = 0.0086]). Increases in anterior chamber depth in treated eyes tended to neutralize the opposing effect on axial length of reduced scleral growth (and thus reduced vitreous chamber elongation). Thus, neither group showed a significant treatment effect on axial length (Fig. 3D) . These intergroup differences also translated into a significant hyperopic shift in refractive error for the target 1 group (Fig. 3C) , and minimal change in refractive error for the target 2 group, although there was also increased interanimal variability in the refraction data for target 2 compared with target 1. 

With imposed hyperopic defocus, both targets 1 and 2 elicited similar compensatory responses (Fig. 3 , Table 2 ). Both groups showed increases in vitreous chamber depth and decreases in choroidal thickness in treated relative to fellow control eyes, and these changes were coupled to myopic shifts in refraction in both cases (Figs. 3A 3B 3C) . In both groups, there were also significant increases in axial length (Fig. 3D) , implying that the sclera as well as the choroid contributed to the vitreous chamber elongation and thus to refractive error changes. 

Response to Low-Pass Filtered Targets.

To investigate the influence of high spatial frequency information on compensation for imposed defocus, the response to the standard Maltese cross target (target 2), was compared with the response to a low-pass–filtered version of the same (target 4). These targets were tested in both myopic and hyperopic defocus conditions (Fig. 4 , Table 2 ). 

Within the first 2 days of treatment, the unfiltered target 2 elicited near normal response patterns for both imposed myopic and hyperopic defocus, opposite in direction in accordance with the difference in sign. Intergroup differences in induced vitreous chamber depth, choroidal thickness, and refractive error changes achieved statistical significance (two-way, repeated-measures ANOVA: P = 0.0425, 0.0028, and 0.0193, respectively, Figs. 4A 4B 4C , open symbols; Table 2 ). In contrast, the response patterns recorded with target 4 do not show any sign dependence. Regardless of whether myopic or hyperopic defocus was imposed, treated eyes showed enhanced vitreous chamber elongation and, consequently, development of myopia (Fig. 4 , filled symbols; Table 2 ). Note that the increase in vitreous chamber depth elicited with hyperopic defocus and target 2 was smaller than the increases with both the hyperopic and myopic defocus conditions and target 4 (P = 0.0011 and 0.0173, respectively; Fig. 4A ). For target 4, both defocus groups also showed choroidal thinning in treated eyes within the first 2 days of treatment. Thus, interocular differences in choroidal thickness decreased in both groups over this time frame. With imposed myopic defocus, the interocular difference in choroidal thickness continued to decrease over the subsequent 2 days of treatment, whereas with hyperopic defocus, it returned to the pretreatment level (Fig. 4B) . Corresponding induced refractive error changes were nearly identical in the two defocus groups (Fig. 4C) . All four groups (each of the two targets combined with each of the two defocus conditions) showed treatment-induced increases in axial length and anterior chamber depth that were greater over the final 2 days compared with the first 2 days of the treatment period (Figs. 4D 4E) , and while axial length data hinted at sign-dependent differences, they did not reach statistical significance for either target (Fig. 4D) . 

Spatial Contrast Effects on Compensation to Defocus

In this experimental series, both targets 1 and 2 were presented at five different contrast steps, ranging from low (11%) to high (100%; Fig. 5 , Table 3 ). Because sufficiently low contrast targets are likely to elicit a form deprivation response that would be indistinguishable from an appropriate response to hyperopic defocus, only the myopic defocus condition was tested in this study. For the two highest levels of contrast, the usual effect of imposed myopic defocus on the vitreous chamber—that is, decreased elongation—occurred while lower-contrast targets typically elicited the opposite response (i.e., increased elongation, as in form deprivation; Fig. 5 , top). Likewise, the choroid of treated eyes thickened as expected with the higher-contrast targets but thinned with the low-contrast targets (Fig. 5 , middle). In an interesting observation, some of the chicks initially responded in the correct direction but then underwent a change in the direction of eye growth (data not shown). The net result in refractive terms of the various growth changes just described was a hyperopic shift in the refractive error of treated eyes for the high contrast targets and a myopic shift in refractive error for the lower-contrast targets (Fig. 5 , bottom). Note that this effect of contrast was also target dependent. For target 1, the striped Maltese cross, the 32% contrast version elicited minimal response over the treatment period, whereas the equivalent version of target 2 induced significant changes but in the direction opposite that required for compensation. The latter response pattern also was seen with the lower contrast (11%) versions of both targets. That the contrast requirement for compensation was lower for the spatially rich (striped) target than for the standard target was confirmed statistically (Tukey-Kramer post hoc test; see summary in Table 3 ). These results imply that the threshold contrast for compensation for imposed myopic defocus is dependent on the spatial frequency composition of the retinal image. 

Accommodation and Spatial Frequency Effects on Compensation to Competing Defocus Stimuli

In the preceding experiments, eyes were exposed to a single defocused stimulus. In the final series of experiments described herein, a second target was added to the cone imaging device, at a physically different location and thus a different defocus level. The response to such competing defocus stimuli was investigated for both targets 1 and 2. In all cases, pairs of targets were presented simultaneously in hyperopic and myopic defocus. This paradigm was tested on both normal eyes and eyes with sectioned ciliary nerves (CNX), to examine the influence of accommodation in these responses. To study further the importance of high spatial frequency information for emmetropization, we conducted a third experiment, in which we made use of the low-pass–filtered version of target 1 presented in myopic defocus, combined with the unfiltered version of the same presented in hyperopic defocus. The results of this series of experiments are summarized in Figure 6and Table 2 . 

In normal eyes with normal accommodation, treatment-induced eye growth changes were target dependent (Fig. 6 , left). Specifically, with the standard Maltese cross target (target 2) at both positions, there was a myopic shift in refractive error, a consequence of the steady treatment-related increase in vitreous chamber depth and axial length over the treatment period. Choroidal thinning contributed to, but did not fully account for, the vitreous chamber changes. This myopic bias was replaced by a small hyperopic bias when striped Maltese cross targets (target 1) were used instead. Vitreous chamber depth decreased and choroidal thickness increased over the first 2 days of treatment, with both effects decreasing to become negligible after a further 2 days of treatment. These target-design–related differences in response patterns were statistically significant in all cases (Table 2 : two-way, repeated-measures ANOVA: P = 0.0218, refractive error; P = 0.0109, choroidal thickness; P = 0.0011, vitreous chamber depth). 

When the striped target was replaced by a low-pass–filtered version of the same (target 3) at the myopic position (Fig. 6 , middle) and accommodation was left intact, the result was a myopic bias in refractive terms, with increases in vitreous chamber depth and axial length as well as choroidal thinning, more similar to the pattern obtained for the pair of unfiltered standard targets (target 2). Compared to the response pattern seen with the target 1 design at both defocus positions, there was a refractive error shift in the direction of myopia, thinning instead of thickening of the choroid, and an increase instead of a decrease in vitreous chamber depth. All differences were statistically significant (Table 2 , two-way, repeated-measures ANOVA: P = 0.0433, 0.0401, and 0.0207, respectively). 

When accommodation was paralyzed (CNX birds), both the striped and standard target designs (targets 1 and 2), elicited myopic growth patterns—that is, increased vitreous chamber growth and axial length (Fig. 6 , right)—although also in this case there was a significant effect of target design, with significantly more myopia developing with target 2 (Table 2 : two-way, repeated-measures ANOVA, P = 0.0115). Note that the responses of both CNX and normal chicks were in the same direction for target 2, although the responses of CNX chicks were significantly larger (ANOVA, P = 0.0002, vitreous chamber depth; P = 0.0011, axial length; Fig. 6 , right column versus left column). Consistent with the fact that the normal chicks but not the CNX chicks exhibited compensatory responses to the striped target, differences in the two related data sets also were statistically significant (ANOVA: P = 0.0002, vitreous chamber depth; P = 0.0048, axial length). 

Discussion

Why excessive near work might cause myopia in susceptible people and how the eye decodes the sign of defocus present two of the most studied but as yet unanswered questions in this field. In the present study, we sought to investigate further, under controlled conditions, the stimulus requirements for emmetropization as exemplified by the compensatory ocular growth responses in young eyes exposed to experimentally imposed defocus. Working from a standard Maltese cross target design, we generated stimuli that had more or less high spatial frequency information and also varied spatial contrast. We found significant stimulus-dependent differences in ocular growth responses in the presence of imposed defocus. In brief, we found that (1) compensation for myopic defocus was critically dependent on the inclusion of middle to high spatial frequencies (edges) in the stimulus; (2) compensatory responses to defocus have contrast thresholds below which responses resemble form deprivation myopia; (3) with competing defocus situations, myopic defocus transiently dominates over hyperopic defocus as a determinant of ocular growth, provided the stimulus conditions include sufficient middle to high spatial frequency information; and (4) decoding of, and thus compensation for, imposed defocus is impaired when accommodation is inactivated. In the following discussion, we attempt to put these results into context, through comparisons with related data from already published studies. We also address the significance of these results for emmetropization. 

That chicks are able to emmetropize to myopic defocus under restricted visual conditions, as demonstrated in the present study, complements the results of two closely related studies using a very different paradigm. ^{19 20} In both cases, young chicks were restrained in the center of a drum, the inside lining of which provided the only available visual information. The latter study included a wide range of spatial frequencies, although limited to one viewing distance as in the present study, and eyes showed compensatory responses to imposed myopic defocus as well as imposed hyperopic defocus, as was also true in the present study. Of relevance to later discussion is an additional finding from one of these studies ¹⁹ that pharmacologic inhibition of accommodation did not prevent compensation for myopic defocus. 

Also of relevance to the current discussion are experiments contained in another “cone” study by Schmid et al. ²¹ (see also Schmid KL, et al. IOVS 2002;43:ARVO E-Abstract 187) although their results are paradoxical. Specifically, myopia instead of hyperopia was consistently reported in response to imposed myopic defocus for a target similar in design to the standard target design used in the present study. In addressing why the outcomes of these two studies were different, it is important to review the differences between them. Potentially important features of the study by Schmid et al. ²¹ include (1) the wider field of view (60°) provided by their cone device, (2) the generally higher levels of imposed myopic defocus used (e.g., 13.5 and 20.5 D), and (3) the restrictions on measurements to the last day of the treatment period. Note that both lens power and cone length were manipulated to achieve different levels of defocus in this study. The use of wider cones cannot account for the discrepancy between the present study and that of Schmid et al., since wider cones do not reduce but rather increase the amount of imposed myopic defocus, averaged across the field (it increases with field eccentricity). Of other likely explanations for the different study outcomes, it is possible that the compensatory range of the chicks was exceeded. This argument rests on the assumption that the restricted visual conditions also reduced the range of myopic defocus for which there can be compensation. This assumption seems reasonable, given that under the lens-only conditions that elicit the best compensation, chicks are able to approach nearby objects, thereby reducing the defocus experienced. The apparent absence of any compensation for the lowest level of imposed defocus (∼5 D) may reflect the limitations imposed by their measurement protocol. By measuring eyes only at the end of the 4-day treatment period, Schmid et al. may have missed transient compensatory changes in eye growth that were a feature of the response to the standard Maltese cross target in the present study. 

In the present study, the more transient nature and high interanimal variability (high standard errors) of the responses to the standard Maltese cross presented under myopic defocus suggests that there may be another competing signal with a slower time constant that drives eye growth in the direction of myopia, away from that required for compensation. A similar model could explain why compensation for the imposed myopic defocus was always incomplete, irrespective of the target used. Treatment duration is not a limiting factor, as larger compensatory shifts in refraction are observed over the same time frame when spectacle lenses alone are used to impose defocus. ^{6 17} An alternative explanation is that the larger ranges of compensation observed in such spectacle lens experiments are a consequence of the chicks being able to move around freely and so alter their defocus experience by moving closer to objects. In the context of emmetropization, others have argued that the critical difference between imposed myopic and hyperopic defocus is the amount of blur experienced (the drive for increased growth), with animals experiencing less blur with myopic defocus when allowed free movement. Under the restricted conditions of the present study, only negative accommodation could have reduced the level of defocus experienced in the case of myopic defocus, and this is limited in capacity compared with positive accommodation (Troilo D, et al. IOVS 1993;34:ARVO Abstract 2990). Thus, the cone devices arguably increased the experience of blur per diopter unit of imposed myopic defocus. 

The demonstration of a spatial frequency dependence of the emmetropization “defocus sign detector” is an important result, as errors in this regard always resulted in myopia. Thus, when either the standard Maltese cross target or a low-pass–filtered version of either standard or striped targets was used, the ocular growth responses elicited by imposed myopic defocus resulted in more rather than less myopia. The possibility that brightness differences between the standard and striped targets may be a contributing factor was ruled out in a subsequent experiment involving a luminance-matched version of the striped target (two of the white arms shaded black), which induced the same eye growth changes as target 1 (data not reported). The simplest interpretation of these results is that spatial frequencies in the middle to high range provide cues to myopic defocus. The same conclusion concerning the importance of high spatial frequencies for emmetropization can be drawn from an earlier study in which the visual experience of young chicks was limited to low-pass–filtered video clips. ⁹ Specifically, restricting visual experience in this way prevented recovery from form deprivation myopia and induced myopia in normal chicks. 

It is not clear from the present study whether the spatial frequency dependence observed in the compensatory responses to imposed myopic defocus reflects a general property of emmetropization, because, for compensation for hyperopic defocus, the noncompensating response would be in the same direction as the response required for compensation, thus increasing myopia. Although low-pass filtering of the targets appeared to exaggerate the myopic growth in response to hyperopic defocus, this effect may be an artifact. Because the removal of high spatial frequency information increases the functional depth of focus, eyes presumably “overgrow” the emmetropization end point before reaching the point of detectable defocus once more. Using spatial noise to generate targets in a cone-based study, Schmid et al. noted a similar increase in myopia when the amplitude of the noise was limited to low spatial frequencies (Schmid KL, et al. IOVS 2001;42:ARVO Abstract 323). At odds with these results is a report that the addition of diffusers to negative defocusing lenses does not affect the response pattern ²⁰ ; such filters may also be expected to remove high spatial frequency information and so to have similar effects on the ocular depth of focus. The reason for this different outcome is unclear, although it may be related to differences in treatment duration and/or sex and strain of chicken used. ^{22 23 24}  

Our competing defocus paradigm served as another demonstration of the importance of higher spatial frequencies for emmetropization, and in particular, for compensation for myopic defocus. The latter response was generally more robust than compensation for imposed hyperopic defocus under competing defocus conditions. A similar robustness is evidenced in results from unrelated studies in which spectacle lenses of opposite sign were interposed successively. The imposed myopic defocus dominated the responses, even when hyperopic defocus was imposed for a much longer period. ^{25 26} For eyes to distinguish myopic defocus from hyperopic defocus when presented in competition under the restricted conditions used in the present study (with cone imaging devices), there appear to be two prerequisites: The retinal image must contain sufficient middle to high spatial frequency information, and accommodation cues must be available. If both prerequisites are fulfilled, myopic defocus effects appear to dominate over hyperopic defocus effects, at least in the short term. Thus, in the present study, growth changes were in the direction appropriate for that of imposed myopia within the first 2 days in response to the paired striped Maltese cross targets. However, by the end of the 4-day treatment period, these early changes had regressed and become almost negligible. The latter result is more in keeping with the results from a closely related experiment in the study by Schmid et al. ²¹ The explanation for this late regression is unclear, although it hints at a second competing growth signal with a slower time constant, as discussed earlier. 

What insight does the present study provide about the role of accommodation in emmetropization? That accommodation plays some role is implied by results of a study in which a similar experimental cone paradigm was used. ¹⁴ In the present study, under competing defocus conditions, eyes showed different response biases, depending on whether accommodation was functional. Specifically, after ciliary nerve section, eyes showed increased vitreous chamber elongation, more consistent with compensation for hyperopic defocus and opposite for the trend in normal chicks. This observation suggests that active accommodation is necessary for the decoding of myopic defocus in this competitive situation. Earlier, we discussed the possibility that negative accommodation may serve to reduce the amount of imposed defocus, bringing it into the response range of the emmetropization mechanism. However, the results of another study in which multifocal spectacle lenses were used to present competing defocus stimuli implies that the role of accommodation goes beyond this (Wildsoet CF, et al. IOVS 2000;41:ARVO Abstract 3930). In this study, chicks were allowed free movement and thus could reduce the amount of myopic defocus experienced by approaching nearby objects in their environment. Yet, in that study as well, the elimination of accommodation by CNX resulted in a shift from a hyperopic to a myopic response bias in the presence of competing defocus stimuli of opposite sign. Results of another study in the chick also indicate that constant defocus is not a prerequisite for compensation for hyperopic defocus. When bilateral negative spectacle lenses were combined with monocular CNX surgery, good compensation was observed, even though the unlesioned eye would have had to clear the imposed defocus on a near-continuous basis for the animal to see. ²⁷ The latter result implies that the emmetropization mechanism must somehow encode accommodative activity. Together, these various results also suggest a role for accommodation in the decoding of defocus during emmetropization, although the specific details of its role are as yet unresolved. 

Manipulation of target contrast in the present study convincingly demonstrated that there is a threshold contrast requirement for the decoding of myopic defocus and that reducing the middle to high spatial frequency content of the targets increased this threshold. Although the observation that the contrast threshold was lower for the spatially rich (striped) target than for the standard target seems counterintuitive, it is likely that the spatial frequency of the striped filler pattern, ∼1.2 cyc/deg, was sufficiently low to survive the degrading effect of the 7 D of defocus imposed (based on model calculations for the young chick eye and a 3 mm pupil). The spatial contrast sensitivity function of the chick, which peaks around 1.2 cyc/deg, favors the detection of this information and any attenuation of the imposed defocus through negative accommodation also would have improved the visibility of this spatial information. Nonetheless, changes in the direction of eye growth during the 4-day treatment period, away from that required for compensation, were observed in some of the chicks. These changes imply that there is a competing signal, as already alluded to elsewhere in this discussion, perhaps driven by low spatial contrast and/or spatial frequencies. 

The question of how the sign of the defocus is decoded in emmetropization is not fully resolved by the current work. If spatial contrast provides the signal, then feedback would have to be part of this process, as spatial contrast is reduced by both myopic and hyperopic defocus. Other possibilities are raised by research into accommodation, another ocular focusing mechanism that also must decode the sign of defocus. Monochromatic aberrations and the Stiles-Crawford effect are two such possibilities that warrant investigation in the context of emmetropization. ²⁸  

In summary, middle to high spatial frequencies appear to be critical to the decoding of imposed myopic defocus, allowing the associated retinal image blur to be distinguished from that resulting from imposed hyperopic defocus. The nature of the dependence, if any, of compensation for hyperopic defocus on the spatial frequency composition of the visual and thus retinal images remains unresolved. For the decoding of myopic defocus, there is also a threshold contrast requirement, and reductions in the high spatial frequency content of the retinal image increase this threshold. When there are competing defocus signals of opposite sign, the response shows a transient bias toward imposed myopic defocus, although this bias requires two prerequisites to be met: the target (and retinal image) must contain sufficient middle to high spatial frequency information, and accommodation cues must be available. The latter findings provide new insight into how emmetropization might operate in the normal visual environment, where competing defocus signals and variations in spatial frequencies and contrasts are the norm. 

 Supported by German Research Council (DFG) Grant DI 834/1-1 and National Eye Institute Grant EY012392.

 Submitted for publication October 8, 2004; revised March 8, 2005; accepted March 13, 2005.

 Disclosure: S. Diether, None; C.F. Wildsoet, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Summary of Treatment Groups

Imposed Defocus	Type of Target	Target Contrast (%)	CNX	Chicks (n)
Myopic defocus (+7D)	Target 1	100	No	18
		65	No	8
		32	No	10
		16	No	8
		11	No	5
	Target 2	100	No	19
		65	No	8
		32	No	10
		16	No	8
		11	No	5
	Target 4	100	No	8
Hyperopic defocus (−7D)	Target 1	100	No	13
	Target 2	100	No	13
	Target 4	100	No	7
Competing defocus (±7D)	Target 1 at both positions	100	No	14 + 9*
			Yes	9
	Target 2 at both positions	100	No	14
			Yes	10
	Target 3 (myopic defocus) and target 1 (hyperopic defocus)	100	No	9

 Data include the defocus imposed by the lens-cone device used, type of Maltese cross target used striped (target 1), standard (target 2), low-pass filtered striped (target 3), or low-pass filtered standard (target 4), target contrast, intact or paralyzed accommodation, and number of birds in each group.

*  This experiment was performed twice, as a direct comparison experiment (same batch and identical conditions) to two other competing defocus experiments, presenting (1) target 2 at both positions and (2) target 3 in myopic defocus and target 1 in hyperopic defocus.

Figure 1.

 

Diagram of the three lens cone devices used in the study. (A) Device imposing 7 D of myopic defocus: target located beyond the focal plane of the +40 D lens and consequently imaged in front of the retina (at ). (B) Device producing 7 D of hyperopic defocus: target located within the focal plane of the +30 D lens and thus imaged behind the retina. (C) Device imposing competing defocus: two targets located beyond and within the focal plane of a +40 D lens. The latter is imaged behind the retina, generating 7 D of hyperopic defocus; the former is imaged in front of the retina, generating 7 D of myopic defocus.

Figure 2.

 

Maltese cross targets attached to the lens cone devices. Targets with 45° arms were used in single defocus conditions. A design with the narrower arms was used in competing defocus conditions. Target 1 included a black-and-white, 1.2 cyc/deg, striped filler pattern; target 2 consisted of a solid black Maltese cross (no filler pattern). Both targets were presented at five different contrast steps. Two other versions of these two targets were generated with Gaussian filters. The resultant low-pass–filtered targets are shown on the right (targets 3 and 4). For competing defocus conditions, narrow-armed versions were used (e.g., top right).

Figure 3.

 

Effect of target design on compensation in response to imposed myopic and hyperopic defocus (+7 D or −7 D). Mean interocular differences (± SEM) measured on day 0 before treatment and after 2 and 4 days of treatment, are shown for the striped (target 1) and the standard (target 2) Maltese cross targets. With imposed myopic defocus, target 1 elicited significantly better compensation than target 2. With imposed hyperopic defocus, both targets yielded similar compensatory responses. Interocular differences in (A) vitreous chamber depth, (B) choroidal thickness, (C) refractive error, (D) axial length, and (E) anterior chamber depth are shown. Differences between target 1 vs. target 2 under myopic defocus significant (**P < 0.01).

Table 2.

 

Effect of Target Design on Compensation in Response to Imposed Defocus, Compared Statistically

Treatment	Interocular Differences
		VCD (mm)	CHT (mm)	RE (D)
Myopic defocus, T1 vs. T2	Sig. target × time interaction(P = 0.0086)	Sig. target effect(P = 0.0047)	Sig. target effect(P = 0.0077)
Hyperopic defocus, T1 vs. T2	NS	NS	NS
T2, myopic defocus vs. hyperopic defocus	Sig. defocus effect (P = 0.0425)	Sig. defocus effect (P = 0.0028)	Sig. defocus effect (P = 0.0193)
T4, myopic defocus vs. hyperopic defocus	NS	Sig. defocus effect (P = 0.0024)	NS
Competing defocus, T1 at both positions vs. T2 at both positions	Sig. target effect (P = 0.0011)	Sig. target effect (P = 0.0109)	Sig. target effect (P = 0.0218)
Competing defocus, T1 at both positions vs. T3 (myopic defocus) and T1 (hyperopic defocus)	Sig. target effect (P = 0.0207)	Sig. target effect (P = 0.0401)	Sig. target × time interaction (P = 0.0433)
Competing defocus, paralyzed accommodation, T1 at both positions vs. T2 at both positions	Sig. target effect (P = 0.0437)	NS	Sig. target effect (P = 0.0115)

 Defocus included hyperopic, −7 D; myopic, +7 D; and competing, ±7 D. Target designs: Striped Maltese cross (T1), standard Maltese cross (T2), low-pass filtered version of the striped Maltese cross (T3), low-pass filtered version of the standard Maltese cross (T4). Effects on vitreous chamber depth (VCD), choroidal thickness (CHT) and refractive error (RE) were compared using 2-way repeated measures ANOVAs. Data are shown in Figures 3 4 and 6 . Sig., significant; NS, not significant.

Figure 4.

 

Effect of target design on compensation for imposed myopic and hyperopic defocus (+7 D or −7 D). Mean interocular differences (± SEM) measured on day 0 before treatment and after 2 and 4 days of treatment are presented for the standard Maltese cross target (target 2) and the low-pass–filtered version of the same target (target 4). Sign-dependent differences were evident in the responses to target 2. In contrast, the responses recorded with target 4 did not exhibit any sign dependence, with myopia being observed, irrespective of the sign of the imposed defocus. Interocular differences in (A) vitreous chamber depth, (B) choroidal thickness, (C) refractive error, (D) axial length, and (E) anterior chamber depth are shown. Differences between myopic vs. hyperopic defocus significant for target 2 (*P < 0.05; **P < 0.01).

Figure 5.

 

Effect of target contrast on compensation for imposed myopic defocus (+7 D). Presented are the means (± SEM) of derived changes in interocular difference over the 4-day treatment period for the striped (target 1) and the standard (target 2) Maltese cross targets. Both targets were presented at five different contrast steps. There is a threshold contrast requirement for the decoding of myopic defocus below which eyes develop myopia instead of compensatory hyperopia. Note that this contrast threshold was lower for target 1. Change in interocular difference is shown in (top) vitreous chamber depth, (middle) choroidal thickness, and (bottom) refractive error. Differences between target 1 vs. target 2 significant (*P < 0.05, **P < 0.01).

Table 3.

 

Effects of Contrast Compared Statistically for the Striped (Target 1) and the Standard (Target 2) Maltese Cross Targets Presented in Myopic Defocus

Eye Dimension	Target 1 (% Contrast)	Target 2 (% Contrast)
VCD		100–32
	100–16	100–16
	100–11	100–11
	65–16	65–32
	65–11	65–16
		65–11
CHT	100–32	100–32
	100–16	100–16
	100–11	100–11
RE		100–32
	100–16
	100–11
		65–32
	65–16	65–16
		65–11

 For each target, the listed group data were significantly different (p < 0.05; 2-way ANOVA followed by Tukey-Kramer HSD post-hoc testing). Data shown in Figure 5 .

Figure 6.

 

Effect of target design and accommodation on compensation for competing defocus. Two targets were presented, one located beyond and the other one within the focal plane of the imaging lens. Mean interocular differences (± SEM) measured on day 0 before treatment and after 2 and 4 days of treatment are shown for striped Maltese cross target pairs (2 × target 1) and standard Maltese cross target pairs (2 × target 2) both for normal eyes (left column) and eyes with sectioned ciliary nerves (right column). With normal accommodation and target 1 at both defocus positions, choroidal thickening and decreased elongation of the vitreous chamber and axial length are evident on day 2, whereas with target 2 at both positions, the response bias was consistently in the opposite direction. When a low-pass–filtered version of target 1 (target 3) was presented in myopic defocus, combined with the unfiltered version of target 1 in hyperopic defocus (middle column), treated eyes exhibited a myopic response bias, as seen with target 2 at both positions. When accommodation was paralyzed, both target 1 and 2 designs elicited myopic response patterns, although significantly more myopia was evident with target 2. Compared differences significant (*P < 0.05; **P < 0.01; ***P < 0.001).