Effect of Blur Adaptation on Human Parafoveal Vision

Abstract

Purpose.: This study was conducted to investigate whether neural compensation for induced defocus can alter visual resolution in other areas of the human retina beyond the fovea. In certain circumstances, the blur adaptation response may be influenced by refractive status.

Methods.: The effect of blur adaptation on the central 10° of the retina was investigated in 20 normally sighted observers (10 emmetropes and 10 myopes; median age, 21 years). Visual acuity (VA) was measured at the fovea and at five locations of the parafoveal nasal visual field (2°, 4°, 6°, 8°, and 10°) with best corrected distance vision. Myopic defocus of 1 D was introduced, and the same measurements were repeated immediately before and after a 30-minute adaptation.

Results.: VA declined with increasing eccentricity in the clear, blurred, and blur-adapted viewing conditions. The rate of decline was quantified by the parameter E ₂, which represents the amount of eccentricity dependence of the acuity task. Foveal and parafoveal VA decreased with the introduction of optical defocus and improved significantly after a period of blur adaptation. The consistent value of E ₂ in each condition indicated that these changes in VA were not eccentricity dependent. Changes in VA under blurred and blur-adapted conditions were of similar magnitudes in myopic and emmetropic observers.

Conclusions.: Neural adaptation to blur improves VA under defocused conditions in the parafovea as well as the fovea, indicating that the underlying compensatory mechanism acts across a range of spatial scales and independently of retinal eccentricity. Foveal and parafoveal blur adaptation does not vary with refractive error.

The human visual system is able to compensate for optical defocus through a process of neural adaptation, producing an improvement in visual acuity (VA) without a change in ocular refraction. ^1,2 The introduction of defocus attenuates the high and medium spatial frequency content of an image. Mon-Williams et al. ¹ hypothesized that the visual system strives to recover the attenuated signal to improve resolution by recalibrating the spatial-frequency–processing channels. This adaptation could be achieved by an increase in the gain of high-frequency–selective channels coupled with a decrease in low-frequency–selective channel gain in an attempt to restore the image's preblur amplitude spectrum. The recalibration is continuous, allowing the visual system to adapt to an ever-changing visual diet. ^3,4  

Findings in studies have suggested that blur adaptation effects are influenced by ametropia, with myopes displaying a greater degree of adaptation. ^5,6 Myopes demonstrate an increased tolerance to blur compared with emmetropes, evidenced by a reduced response to blur-induced accommodation ^{7 –9} and a larger depth of focus found in young adult myopes. ¹⁰ These factors indicate a possible underlying physiological difference in the interpretation of a blurred signal, which may be caused by or contribute to the progression of myopia. ¹¹ Blur adaptation elevates the threshold for blur detection more significantly in early-onset myopia and improves low-contrast grating VA in myopic individuals more than in other refraction groups, ^5,6 but improvements in high-contrast VA are equivalent between myopes and emmetropes. ^5,6,12  

To date, the blur adaptation effect has been examined only at the fovea. Resolution acuity decreases with retinal eccentricity, limited by neural receptor density. In the parafoveal area—the central 10° of the retina—cone arrangement becomes less regular and cone diameter increases, ¹³ causing a reduction in VA ^14,15 and reduced sensitivity to high spatial frequencies. ¹⁶ The region remains responsive to induced defocus, although thresholds for both blur detection and discrimination are elevated in comparison to the fovea. ^{17 –19} From a clinical perspective, the study of parafoveal regions of the visual field in respect of responsiveness and adaptation to blur is timely, because of recent increases in interest in the potential role of the parafovea in the development of myopia. ²⁰ Evidence from animal models has shown that retinal image degradation away from the fovea could induce axial myopic progression. This finding has spurred the development of peripheral optical modifications to traditional vision correction modalities. ^21,22 Manipulation of peripheral refraction is one of several strategies for reducing the rate of myopia progression in human eyes. ^23,24 With this in mind, further study of the peripheral retina from a functional perspective and in particular its ability to adapt to blur, is necessary. 

We wanted to investigate the blur adaptation effect in parafoveal vision to establish whether this process can occur in an area with a reduced spatial frequency range. In light of past differences observed in the blur adaptation response between emmetropes and myopes, we will investigate both these refractive groups. 

Methods

Subjects

Twenty observers took part in the study. The median age of the group was 21 years (range, 19–39). The participants were recruited according to refractive error groups: 10 emmetropes (median age, 21 years; range 19–35) and 10 myopes (median age, 21 years; range, 20–39). Emmetropes were defined as those with a spherical equivalent refraction (SER) between −0.75 and +0.75 DS and with astigmatism not exceeding 0.50 DC (mean SER, +0.11 ± 0.39 D). Participants with an SER worse than −0.75 DS were classified as myopes (mean SER, −3.21 ± 2.03 D; range, −0.78 to −6.88 DS). Astigmatic error in myopic participants was no greater than 0.75 DC. Central refractive errors were corrected with a spherical soft contact lens (1-Day Acuvue Moist; Johnson & Johnson Medical, Ltd., Livingston, UK). The level of individual astigmatic error was low and as such would not interfere with the observer's adaptation to induced spherical defocus. ²⁵ Acuity measurements were taken on the right eye only, and parafoveal VA was measured in the nasal visual field, corresponding to the temporal retina. All observers were free from ocular disease and gave informed consent to participate in the study, which was conducted in accordance with the Declaration of Helsinki. Institutional approval of the protocol's ethics was obtained. 

Acuity Measurement

VA was measured using the Freiburg Visual Acuity and Contrast Test (FrACT) version 3.5.3. ²⁶ The program uses the best PEST (parameter estimation by sequential testing) algorithm ²⁷ to give fast, accurate measures of VA and has been validated elsewhere. ^26,28  

The stimulus presented was a Landolt-C optotype at 100% contrast, with the observer indicating the position of the gap within the letter C from four possible orientations. An eight-alternative, forced choice (AFC) is advocated to minimize the guessing rate ²⁸ ; however, in a pilot study, this approach was found to be too confusing for the participants. The number of stimulus presentations in one set of trials was 24. ^26,29 All free, so-called motivational trials, presenting a large optotype, were eliminated. A randomized pattern appeared on-screen for 200 ms between each stimulus presentation to eliminate visual after-effects. 

Hardware included a laptop computer (model S2410-504; Toshiba Europe, GmbH, Regensburg, Germany), and the stimulus was displayed on a 15-inch cathode ray tube (CRT) monitor (mean luminance, 88.6 cd/m²; resolution, 1024 × 768 pixels; Chuntex Electronics Co., Taipei, Taiwan). 

As a means of comparison and validation for the VA results obtained with FrACT under conditions of blur and blur adaptation, foveal VA was also measured in all three viewing conditions with a standard Early Treatment Diabetic Retinopathy Study (ETDRS) chart. ³⁰ VA measured with similar charts has been used widely in blur-adaptation work. ^6,12,31 Monocular VA was measured at 4 m. The observer was encouraged to read each letter on the chart, starting from the top row and stopping when three or more letters in a row were missed. All three ETDRS charts were used in random order to prevent letter memorization. 

Procedures

The stimulus was presented on a CRT monitor situated 4 m away from the observer's right eye; the left eye was occluded. To the right of the monitor was a panel with five LEDs, which served as a fixation target when parafoveal acuity was tested. The panel was positioned so that the center of the CRT monitor represented central fixation: The first LED was at 2° and the fifth at 10° retinal eccentricity. The panel did not obscure any part of the stimulus presented on screen (Fig. 1). The subject's fixation was monitored with a video camera. 

Figure 1.

 

Schematic representation of experimental setup.

Each observer was given instruction and training to become familiar with the procedure and software before commencement of the experiment. The experimenter randomized the sequence of fixation targets presented to overcome any effect of learning. Cycloplegia was not used in this experiment, as there is no demonstrated benefit in the measurement of parafoveal visual function. ¹⁸ Five trials were completed at each eccentricity in each of the following viewing conditions:

1.  
   Clear: best corrected distance vision (at 4 m).
2.  
   Blur, no adaptation: The observer wore a trial frame with a +1-DS full-aperture trial lens over the best refractive error correction. The lens was removed briefly after every trial to prevent adaptation to blur at this stage.
3.  
   Blur, with adaptation: The experiment was repeated immediately after 30 minutes' adaptation to blur. During this session, the blur-inducing trial lens was not removed between trials.

During the adaptation period, the observer watched a film on DVD shown on a television screen at a distance of 4 meters. During this time, the observer had binocular vision, with both eyes blurred by +1 DS. This level of myopic defocus has been shown to produce consistent levels of blur adaptation. ¹² To avoid fatigue, the experiment was conducted over two sessions: clear and blur with no adaptation VA was measured in one session, and blur-adapted VA was measured in the other trial. Each session lasted a maximum of 90 minutes, and the order of the trials was randomized. The trials were separated by a minimum of 48 hours. 

Data Analysis

Data were analyzed according to the expected relationship between eccentricity and VA ^32,33 :  

 

where MAR₀ is the minimum angle of resolution at the fovea, E is the target eccentricity and E ₂ is a parameter that determines the extent of eccentricity dependence of the acuity task. E ₂ is the eccentricity at which foveal MAR doubles, as evidenced by assigning E = E ₂ in the above formula. 

The effect of blur was modeled by using an amended version of equation 1, to include an added blur parameter σ_blur (minutes of arc):  

 

 

The extent of recovery after adaptation was evaluated by using an amended version of the previous equations that includes a blur recovery parameter, σ_adapt. The equation becomes:  

 

 

Results

Our baseline clear data (Fig. 2, open circles) conformed extremely well to the expected eccentricity dependence of VA (r ² = 0.999). The value of E ₂ for all observers is 2.545°. 

Figure 2.

 

Foveal and parafoveal VA for all participants in clear (open circles), blurred (open squares), and blur-adapted (black squares) conditions fitted with the models defined in equations 1 through 3. Error bars, 1 SD.

Equation 2 was used to analyze the preadaptation blurred data (Fig. 2, open squares). It is important to emphasize that these curve fits involve just a single free parameter (σ_blur). Other parameters (MAR₀ and E ₂) were not free to float, but were fixed at the respective values obtained from the initial curve fits to the clear data (Fig. 2, open circles). Despite this constraint, equation 2 provided a very good fit to the data for the whole cohort (r ² = 0.987). Examination of the blur parameter σ_blur allowed us to quantify the extent of added optical blur, which was 1.979 ± 0.065 minutes of arc for the group. 

Equation 3 was fitted to the blur-adapted data (Fig. 2, black squares) and, as before, this curve fit contains just a single free parameter, σ_adapt. Values for other parameters are assigned to those found from earlier applications of equations 1 and 2. Again, the quality of the fit was excellent (r ² = 0.994). The extent of recovery to blur through neural adaptation, quantified by the free parameter σ_adapt, was 0.682 ± 0.038 minutes of arc for all participants. 

When the equations are applied separately to emmetropes (Fig. 3a) and myopes (Fig. 3b), once again the models fit the data very well. The value of E ₂ for the emmetropes is 2.06°, which is similar to the group E ₂ value, although in myopes it is slightly larger (3.16°). This variance is statistically significant (paired t-test, t ₍₁₀₎ = 1.0; P < 0.005) and reflects the poorer foveal MAR in the latter group. 

Figure 3.

 

Foveal and parafoveal VA for (a) emmetropes and (b) myopes in clear (open circles), blurred (open squares), and blur-adapted (black squares) conditions fitted with the models defined in equations 1 through 3. Error bars, 1 SD.

The amount of optical blur added (equation 2) was similar in magnitude for both refractive groups: 2.03 ± 0.12 and 1.90 ± 0.05 minutes of arc for emmetropes and myopes, respectively. The difference between refractive groups in this viewing condition was not statistically significant (paired t-test, t ₍₁₀₎ = 1.0; P > 0.1). Likewise, the blur-adapted data fit the model well (r ² = 0.965 for the emmetropic group and 0.991 for myopic group). The σ_adapt parameter did not differ greatly between emmetropes and myopes: 0.71 ± 0.04 and 0.60 ± 0.08 minutes of arc and was not statistically significant (paired t-test, t ₍₁₀₎ = 1.59; P > 0.1). 

In a comparison of FrACT VA values to ETDRS VA, we found that FrACT VA was generally lower than ETDRS VA in all three viewing conditions for all participants (Table 1). The difference in VA recorded with the two methods fell within ±2 SD of the mean—that is, within the 95% limits of agreement. ³⁴ Figure 4 shows Bland-Altman plots for ETDRS and FrACT clear, preblur, and postblur adaptation data. For all viewing conditions, the Spearman rank correlation between the absolute difference and the mean was nonsignificant (P > 0.05). 

Table 1.

 

Comparison of Mean Foveal VA in Emmetropes and Myopes Measured with an ETDRS Chart and the FrACT Program

	ETDRS Chart		FrACT
Viewing Condition	Emmetropes	Myopes	Emmetropes	Myopes
Clear	−0.07 ± 0.12	−0.04 ± 0.08	−0.10 ± 0.13	0.00 ± 0.10
Preadaptation blur	0.38 ± 0.33	0.38 ± 0.12	0.43 ± 0.25	0.45 ± 0.11
Postadaptation blur	0.26 ± 0.19	0.26 ± 0.15	0.33 ± 0.20	0.31 ± 0.14

 Data are expressed as the mean VA in logMAR ± 1 SD.

Figure 4.

 

Bland-Altman plots comparing agreement between ETDRS and FrACT VA in (a) clear (open symbols), (b) preblur-adapted (black squares), and (c) blur-adapted (black diamonds) viewing conditions.

Discussion

Blur Adaptation in the Parafovea

This is the first investigation of parafoveal blur adaptation. We found that the human visual system is able to adapt to conditions of optical defocus to improve visual resolution in both the fovea and parafoveal area. The mean improvement in defocused foveal VA after adaptation to blur observed in this study is consistent with previous findings from this and other laboratories. ^1,6,12 The results point to two features of blur adaptation: (1) neural recalibration of spatial frequency channel output occurs across a range of frequencies; and (2) within the central 10° of the visual field, the adaptation process is independent of retinal location. 

The proposed mechanism for blur adaptation should be reconsidered in light of these findings: does the sensitivity of lower spatial frequency channels decrease to unmask higher spatial frequency information? This process may explain foveal adaptation, but if the VA increase were the result of an amplified high-spatial-frequency signal only, the adaptive effect would diminish in the region away from the fovea, where peak spatial frequency sensitivity is reduced and the range of detectable spatial frequencies is lower than that of the fovea. ³⁵ Previous studies analyzing the effects of blur adaptation on the contrast sensitivity function and on grating acuity have produced conflicting results. ^1,5,36 Our results indicate that a wider range of channels is involved in the recalibration process, suggesting that the mechanism may be independent of specific spatial frequencies. 

Blur adaptation alters the quality of the percept by retuning spatial frequency channels to preserve the overall spatial structure of the image. ⁴ This process is a requirement of processing across the entire visual field, albeit within the resolution limitations of each retinal zone. Hence, the activation of a process to compensate for imposed defocus occurs at both central and peripheral retinal locations, an effect that has been demonstrated with other perceptual adaptations, such as contrast constancy and orientation selectivity. ^3,35 The absence of high-frequency–resolving capability renders the parafoveal image blurry under normal viewing conditions. This study has shown that the parafovea is sensitive to additional optical defocus and has the ability to actively adapt and partially recover resolution in a manner similar to that of the fovea. Furthermore the application of our model using the E ₂ parameter demonstrates that the level of adaptation remains equal across the entire parafoveal zone. In the context of the debate about the role of both central and peripheral blur in myopia progression, this finding raises a question that may merit future investigation: Is there an interaction between the function of blur as a stimulus to growth and as a stimulus for neural adaptation? 

Effect of Ametropia on Foveal and Parafoveal Acuity

In our study we compared emmetropic and myopic VA at the fovea and in the parafoveal region under clear, blurred, and blur-adapted viewing conditions. Our nonblurred data for all observers correspond well with that from previous studies, and our value for E ₂ is consistent with values found elsewhere in the literature. ^{37 –39}  

We found a difference of 0.10 logMAR in baseline foveal VA between emmetropes and myopes. Some disparity between emmetropic and myopic VA levels may be expected, ^40,41 but the method of VA measurement could have also contributed to the variance. With the ETDRS chart, foveal VA was higher in the emmetropic participants, but the intergroup difference was only 0.03 logMAR. Our results (Fig. 4) support previous validations of the FrACT program with the ETDRS chart. ^28,42,43 However, the two methods are not directly comparable and differences in the testing paradigm, threshold calculation, task familiarity, and examiner dependence are significant factors that could explain higher ETDRS VA scores (Table 1). An interesting finding is that, under blurred and blur-adapted conditions, the FrACT scores were not as disparate between the refractive groups as they were under optimal viewing conditions, and they showed reasonable agreement with the ETDRS results, suggesting perhaps that this task is more suited to evaluation of reduced vision. 

In contrast to other studies ^10,44 we did not observe any difference between myopes and emmetropes in the level of sensitivity to induced blur. In myopes the change between blurred and nonblurred VA was smaller than in emmetropes, but in the present study this change was more attributable to their poorer level of nonblurred foveal VA, than to other factors, such as an enhanced depth of focus or increased aberrations. In our emmetropic participants the preadaptation blurred VA was less consistent with the model than for the myopes (Fig. 3). Changes in parafoveal astigmatic blur are as low as 0.28 D, ^45,46 and it is therefore more likely that the noisy appearance of the emmetropes' data represents intersubject variance. Nevertheless, both groups achieved similar VA scores under blurred conditions, evidenced further by the fact that the respective values for σ_blur are almost equal, indicating that parafoveal blur sensitivity thresholds are comparable in emmetropes and myopes. 

Summary

Neural adaptation to blur shows improvements in VA under defocus in both myopes and emmetropes in the parafovea similar to those at the fovea. These results demonstrate that the underlying neural recalibration process is active across spatial frequency channels and that the magnitude of adaptation is not dependent on retinal location. 

Footnotes

 Supported by a Doctoral Training Account grant from the Engineering and Physical Sciences Research Council (AM).

Footnotes

 Disclosure: A. Mankowska, None; K. Aziz, None; M.P. Cufflin, None; D. Whitaker, None; E.A.H. Mallen, None