Enhanced Contrast Sensitivity Confirms Active Compensation in Blur Adaptation

Abstract

Purpose.: To determine the effects of defocus-induced blur adaptation on human contrast sensitivity (CS) function.

Methods.: Defocused (+2 D) CS was measured for spatial frequencies between 0.5 and 12 cycles per degree (cpd) before and after adaptation to +2 D blur in six subjects with normal vision. During the 30-minute adaptation period with +2 D lens, subjects were exposed to a succession of static calibrated natural images that were also used to “top-up” adaptation between postadaptation trials.

Results.: After 30 minutes of blur adaptation, CS was found to be significantly reduced at 0.5 cpd (P = 0.023), though it was enhanced at 8 cpd (P = 0.007) and 12 cpd (P = 0.005). The average sensitivity reduction at 0.5 cpd was 0.20 log₁₀ units, whereas enhancements were 0.09 and 0.16 log₁₀ units at 8 and 12 cpd, respectively.

Conclusions.: The present study demonstrates a novel finding that 30 minutes of defocused viewing results in enhanced high spatial frequency CS. The concurrent observation of low spatial frequency CS reduction suggests that the changes are not caused by simple learning effects but are likely caused by neural adaptation.

People with myopia often report that their unaided vision is worse immediately after spectacle removal in comparison with their vision after a prolonged period without spectacle wear. Pesudovs and Brennan ¹ first investigated this phenomenon and reported improvement in high-contrast letter acuity when their experimental subjects with myopia had not worn their spectacle correction for 90 minutes. Subsequent research among people with emmetropia and myopia has consistently shown substantial improvements in high-contrast letter acuity when short-term exposure to blur was introduced through optical defocus for 30 to 180 minutes (Rosenfield M, et al. IOVS 2003;44:E-Abstract 4315; Rosenfield M, et al. IOVS 2002;43:E-Abstract 1902). ^2–6  

Mon-Williams et al. ⁴ suggested that the adaptation process occurs at central binocular sites in the visual cortex by observing that blurring one eye results in improvements to letter acuity measured in the fellow eye. Furthermore, studies have shown that the improved letter acuity is not associated with any significant change in refractive status (Rosenfield M, et al. IOVS 2002;43:E-Abstract 1902), ^1,2,4 crystalline lens thickness, ¹ pupil size, ⁶ or accommodation ^6,7 though Vera-Diaz et al. ⁸ have shown increased accommodation to near targets after 3 minutes of viewing through diffusing filters. These observations imply that blur adaptation is caused by neural, not optical, compensation. 

In addition to the reports of changes in letter acuity after defocus-induced blur exposure, there is evidence of CS reduction after sustained optical defocus. ^4,9 Mon-Williams et al. ⁴ reported this reduction for spatial frequencies between 5 and 25 cpd but little or no change at spatial frequencies above 25 cpd or below 5 cpd after 30 minutes of exposure to a +2 D defocus (n = 6). In contrast, Webster ⁹ noted consistent reduction in CS between 0.5 and 4 cpd but little or no change at either 8 or 16 cpd when measurements were made after 5 minutes of adaptation through either positive defocus lenses or exposure to artificially blurred images that simulated equivalent defocus conditions (n = 5). Subramanian and Mutti (Subramanian V, et al. IOVS 2005;46:E-Abstract 5604) also provide some data regarding blur adaptation-induced changes in CS, but definite conclusions cannot be drawn because of the lack of sufficient details in their brief report. 

Similarly, there is evidence of changes in grating acuity after exposure to optical defocus. ^2,3 After 120 minutes of blur induction with +2.5 D lenses, people with myopia showed significant improvements in grating acuity for contrast levels between 2.5% and 16%. ³ According to the authors, ³ the improved grating acuity corresponded to spatial frequencies between 10 and 20 cpd. Although grating acuity and CS measures are not necessarily analogous, the authors suggested, based on their observations, that blur adaptation produces an increase in sensitivity for selected spatial frequency channels rather than for all channels, a result not found in any of the CS studies to date. Authors from the same laboratory ² reported improvements in grating acuity for contrast levels up to 40% in subjects with uncorrected myopia after 30 minutes of blur adaptation. 

In summary, previous studies have consistently demonstrated beneficial blur adaptation (i.e., improved high-contrast letter acuity and improved grating acuity at low contrast levels) after subjects' exposure to defocus-induced blur. Although direct comparisons between these CS studies ^4,9 are difficult because of methodological differences, two issues motivated us to investigate CS measures further. First, improvement but not reduction in CS, at least at some mid to high spatial frequency range, might have been expected, particularly given that letter and grating acuity showed improvements. Second, although both studies reported reductions in CS for certain spatial frequencies, the affected frequency range was different in each. Therefore, we investigated the effects of blur adaptation on CS measurements. 

Materials and Methods

Subjects

Six subjects with normal or corrected-to-normal acuity participated in this study. The age range of subjects was 22 to 32 years, with a mean of 25.5 years. Only the right eye was tested throughout the experiment. The refractive status of all subjects was determined objectively using an autorefractor (SRW-5000; Shin-Nippon, Tokyo, Japan)—an infrared open view refractor validated by Mallen et al. ¹⁰ and refined subjectively using a trial frame (data shown in Fig. 1). RN and AK were experienced psychophysical observers. Except for RN, who is one of the authors, all subjects were naive to the purpose of the experiment. All subjects attended a screening visit to rule out any ocular abnormality and were familiarized with psychophysical procedures before baseline measures were collected. The subjects with myopia were required to wear their refractive correction for at least 60 minutes continuously before the start of the experiment. The research followed the tenets of the Declaration of Helsinki and was approved by the University of Melbourne Health Sciences Ethics Committee. Informed consent was obtained from subjects after they received a verbal and written explanation of the nature and possible consequences of the study. 

Apparatus

All the stimuli were presented on a 19-inch CRT monitor (Diamond Plus 93SB [Mitsubishi, Tokyo, Japan]; spatial resolution, 1024 × 768 pixels; temporal resolution, 85 Hz). Mean luminance was set to 46 cd/m². Screen luminance was linearized with a colorimeter (SpectraScan 650; Photo Research, Chatsworth, CA). The graphics card (ATI Radeon 9000; AMD, Markham, ON, Canada), installed in a computer (Power Mac G4; Apple, Cupertino, CA), was controlled by successive versions of EXPO software (courtesy of Peter Lennie; available at http://corevision.cns.nyu.edu/expo/wiki) to draw the experimental stimuli on the monitor and record responses during the experiment. 

Procedures and Stimuli

CS was measured at a distance of 5 m for six different spatial frequencies (0.5, 1, 2, 4, 8, and 12 cpd) using an interleaved staircase method; spatial frequencies greater than 12 cpd were not used because pilot studies established that these were barely visible even at maximum contrast with +2 D defocus. All measurements were performed monocularly before and after 30 minutes of blur adaptation under natural pupil conditions with a +2 D lens (over the habitual refractive correction in myopia). Throughout the duration of blur adaptation and CS measurements, subjects were required to maintain distance fixation, and occlusion of the contralateral (left) eye was achieved using a frosted glass occluder to minimize retinal rivalry or Ganzfeld blankout. ¹¹  

To determine preadaptation CS, subjects were asked to perform a two-interval, two-alternative forced choice (2-IFC) task. Horizontal Gabor patches of the aforementioned spatial frequencies were used; the diameter of the circular Gaussian window (±3.3 σ) corresponded to 2.64° of visual angle. The individual trial sequence for preadaptation measurements was as follows: each trial began with a small fixation target on a uniform background lasting 250 ms. After a gap of 500 ms, 2-IFC stimuli were accompanied by an audible tone in each interval while the stimulus appeared randomly in one of the intervals with equal probability. Each stimulus interval lasted 500 ms, separated by a 250-ms interstimulus interval. Responses were self-paced, and immediate feedback was provided by a different audible tone whenever the responses were incorrect. A new trial began 2400 ms after a correct response or the feedback sound that followed an incorrect response. The preadaptation recording session consisted of at least 40 trials presented at each of the six spatial frequencies and took approximately 22 minutes to complete. Although a (blurred) fixation target was briefly visible between trials, the visual diet during this phase of testing consisted predominantly of a blank gray screen so that blur adaptation was unlikely to have occurred. A series of pilot studies expressly addressing this issue confirmed that this was the case (data not shown). 

During 30 minutes of blur adaptation with a +2 D lens in front of the right eye, subjects were exposed to a succession of calibrated natural images. Eight different images, consisting only of outdoor scenes (trees and leaves) obtained from in and around the Melbourne University's campus or from the free databases of image collection from Google were used. Subjects were familiarized with those images before the start of the experiment. Each 512 × 512 pixel image spanning 2.12° of visual angle was presented for 300 ms and was replaced immediately by another image chosen at random. Ideas about the choice of images originated from the report of Webster and Miyahara. ¹² All the images were adjusted to have the same mean luminance, which was equal to the luminance of the CRT monitor's gray background. Unlike Webster, ⁹ who used images chosen from a very large database, we used a total of only eight familiar images throughout the adaptation period in order that subjects recognize and actively engage with the images even when they appeared significantly blurred. 

To determine postadaptation CS, the procedures were similar to preadaptation measurements except that instead of having a blank gray screen of 2400 ms during the intertrial interval, natural image sequences as described (8 images, 300 ms each) were used to top-up adaptation before each trial. 

Pupil size, with the defocus lens in situ, was dynamically assessed during CS measurements before and after blur adaptation using a digital camera (Handycam DCR-HC96; Sony; Tokyo, Japan) with an internal infrared light source. 

Psychophysical Paradigm

A modified two-down, one-up staircase procedure, described by Levitt, ¹³ was used to determine threshold contrasts. Briefly, the two-down, one-up design waits for two correct responses in a row before there is contrast reduction, whereas every incorrect response leads to an increase in contrast, so that stimuli tend to be presented with contrasts near the 71% correct response point of the psychometric function. It is a modified procedure because the initial step size was kept high (0.2 log₁₀ units) to allow faster convergence onto threshold. After every two reversals, step size was halved; the least step size was no less than 0.05 log₁₀ units. Initial contrast was always chosen to be suprathreshold. 

Statistical Analysis

Psychometric functions were fit for all the raw data of each spatial frequency by using a base-2 Weibull function often referred to as the Quick formulation, ¹⁴ named after Quick, ¹⁵ to determine CS. To obtain the best fit, a nonlinear least-square method was used to minimize the difference between the raw data and the fitted curve. All functions were fit using custom-written scripts in numerical computing software (MATLAB, version 7.0.1; The MathWorks, Inc., Natick, MA). The 95% confidence intervals on CS estimates were obtained by bootstrapping the original data sets 1000 times for each spatial frequency. ¹⁶ For all statistical analyses, we computed two-tailed, paired t-tests with α set at 0.05 using commercial software (SPSS version 16.0; SPSS, Inc., Chicago, IL). 

Results

The results of CS functions for each of the six subjects between preadaptation and postadaptation conditions are shown in Figure 1. Throughout this report, Log CS refers to −Log₁₀ threshold contrast, where maximum contrast is 1. For high spatial frequency stimuli, each of the six observers showed enhanced sensitivity, though it was modest for subject ME. For the lowest tested spatial frequency, there was consistent reduction in sensitivity. These differences were statistically significant at the 95% level for 0.5, 8, and 12 cpd (0.5 cpd, P = 0.023; 8 cpd, P = 0.007; 12 cpd, P = 0.005). The data also show a general reduction in average sensitivity at 1 cpd, but this difference does not have a high degree of statistical significance (P = 0.105). No clear patterns of sensitivity change emerged for spatial frequencies of 2 and 4 cpd. To investigate the repeatability of these observations, CS was remeasured over two or three visits in three subjects (RN, RS, ME). The results plotted in Figure 1 are, for these subjects, derived by fitting psychometric functions to data from 80 to 120 trials at each spatial frequency. For all three subjects, similar changes between preadaptation and postadaptation CS measurements were present in the repeat data for 0.5, 8, and 12 cpd; midspatial frequency results were less robust. 

Figure 1.

 

CS results for the blur-adapted eye of each of the six subjects measured before (dotted line) and after (continuous line) blur adaptation with +2 D defocus. Error bars indicate 95% CI derived by bootstrapping the original data set 1000 times.

The size of the average sensitivity change induced by blur adaptation, though small in log units, represents a significant change in percentage terms. The average sensitivity changes for 0.5, 8, and 12 cpd were −66% (−0.20 log₁₀ contrast), 18% (0.09 log₁₀ contrast), and 29% (0.16 log₁₀ contrast), respectively. The average sensitivity differences are shown in Figure 2. 

Figure 2.

 

Average CS change between pre- and post-blur adaptation with +2 D defocus (n = 6). Positive values (upper) represent increased sensitivity and vice versa. Error bars indicate ±1 SEM.

No significant differences in pupil size were noted between preadaptation and postadaptation conditions (average difference, 0.00 ± 0.07 mm, P = 0.99). Pupil size of one subject (AK) was approximately 8 mm; it ranged between 3.5 and 4.5 mm in the remaining subjects. Analysis of the data revealed that pupil size had no bearing on our results. 

Discussion

This study demonstrates that 30 minutes of defocus-induced blur adaptation resulted in a reduction of CS at the lowest tested spatial frequency and enhancements at high spatial frequencies. Although the magnitude differed among the subjects, the demonstrated CS changes in six subjects suggested that the effects can be generalized to majority of the population with normal vision. ¹⁷  

The observation of consistent enhancement in CS in the high-frequency range after blur adaptation is interesting and is in accordance with the observation of improved letter and grating acuity measures found in other studies (Rosenfield M, et al. IOVS 2003;44:E-Abstract 4315; Rosenfield M, et al. IOVS 2002;43:E-Abstract 1902). ^{1, 2–7} This enhancement effect eluded detection in other contrast sensitivity studies, ^4,9 possibly because of critical differences in methodology introduced in this study. Although Mon-Williams et al. ⁴ reported improvements in letter acuity after 30 minutes of blur adaptation, potential CS improvements might not have been captured because top-up images were not used to avoid the decay of adaptation during lengthy CS measurements, a common but not universal practice. ^9,12,18,19 Indeed, when we measured CS without blur adaptation top-up in our laboratory, enhancements were very difficult to prove (see Fig. 3 and the legend for details). This idea is further supported by the concurrent finding of improved letter acuity, ⁴ which is very quick to measure so that adaptation decay effects do not arise. Although Webster ⁹ used calibrated natural images to top-up adaptation during CS measurements in his study, subjects were exposed to blur only for 5 minutes, which might not have been sufficient to accrue detectable high spatial frequency enhancement effects. The observation of significant improvements in letter acuity by Cufflin et al. ⁶ after 30 minutes, but not 10 or 20 minutes, of blur exposure with +1 D and +3 D defocus lenses adds further weight to this hypothesis. 

Figure 3.

 

Without using top-up images, average pre- and post-blur adaptation CS with +2 D defocus show no enhancement effects. Error bars indicate ±1 SEM. These pilot data, measured without using top-up images during the postadaptation measurement period, were collected before the start of this study. Adaptation procedures used were as those described by Mon-Williams et al. ⁴ These data represent the average of 11 subjects, six of whom also participated in this study. Careful reevaluation of individual results also confirmed that no CS enhancements were evident when no top-up images were used. Note that 16 (not 12) cpd was used earlier.

In addition to the beneficial CS effects shown here resulting specifically from adaptation to blur, there have been reports of facilitation or enhancement of grating detection ^18,19 and orientation discrimination ^19,20 after specific adaptation to given spatial frequencies and orientations. Performance improvements shown in these studies were noted only when adapting and test targets were distinct along the dimensions of spatial frequency or orientation. For example, De Valois ¹⁸ noted a reduction in CS around the adapted frequency and enhanced CS approximately 3 octaves away from the adapted frequency. The effect fell to zero by approximately 1 octave from the adapted frequency, and there was little or no effect of CS between 1 and 2 octaves away from the adapted frequency. De Valois ¹⁸ attributed these effects to a general “inhibition plus fatigue” model, which would seem to predict the results found here. However, this study is different from older studies reporting benefits of adaptation since, unlike gratings of a particular frequency or orientation, natural images used here during adaptation contain a broad range of frequencies and orientations with an inverse spatial frequency amplitude spectrum fall-off. ²¹ Even though the defocus lens used in this study would further reduce the relative amplitude of high spatial frequency detail, the adapting stimuli nevertheless remain broadband. Furthermore, recent results from physiological experiments in awake primates have shown that the responses of cells in the primary visual cortex to broadband stimuli are not simply predictable from responses to individual gratings. ²² Nevertheless, adaptive spatial frequency selective enhancements of neural responses, at least for low spatial frequencies, have recently been shown using different types of broadband stimuli (white noise and natural images) in simple cells of the feline visual cortex. ²³  

Although it is tempting to speculate about differential effects between subjects of varying refractive status, the primary aim of this article is to document the presence of enhancement effects after blur adaptation regardless of refractive condition. A number of previous studies ^3,5,6 demonstrated improvements in letter acuity after blur adaptation but failed to find differential effects between people with myopia and emmetropia. If improvements in letter acuity reflect enhanced CS at high frequencies, one might also expect no postadaptation differences in CS between refractive conditions. A recent study ²⁴ that reported positive contrast-adaptation effects (as measured by interocular contrast matching of a suprathreshold 3.22 cpd grating) also failed to find significant differences between people with myopia and emmetropia. An interesting related finding of that study is the suggestion that contrast-adaptation effects were induced only by positive but not by negative defocus. Although, in contrast to our study, measurements were made without defocus lenses, no significant changes in detection threshold were found at 3.22 cpd (the only tested spatial frequency) for either people with myopia or emmetropia, ²⁴ thus concurring with our own study. On the other hand, people with myopia (not emmetropia) displayed a greater improvement in grating acuity at low contrast levels after blur adaptation. ^2,3 In a different study, ⁵ subjects with early-onset myopia displayed greater blur thresholds (a mean difference of ∼0.25 D) after blur adaptation than did subjects with emmetropia or late-onset myopia. In another blur sensitivity study not specifically addressing blur adaption, ²⁵ people with myopia displayed greater (mean difference, 0.08 D) blur detection threshold than people with emmetropia. Exactly how differential blur sensitivity might affect post-blur adaptation CS is unknown, but the magnitude of defocus-induced blur in this study (+2 D) is significantly greater than the differences in subjective blur sensitivity noted between refractive groups, and we have no reason to expect different CS enhancements among people with myopia and emmetropia. Nevertheless, especially in light of the grating acuity differences between people with myopia and emmetropia, ^2,3 it remains an intriguing idea to test this more fully in a future study. 

It is interesting to consider our results alongside what is known about blur adaptation. As stated earlier, previous studies (Rosenfield M, et al. IOVS 2002;43:E-Abstract 1902) ^1,2,4,6,7 have ruled out the role of optical contributions toward improved letter or grating acuity after blur adaptation. Pupil size might be an important parameter regarding adaptation effects, but an analysis of our results failed to identify this as a significant factor. Similarly, perceptual learning has been implicitly ruled out by other investigators, ^1,4 who found no improvements when blurred tests were performed after exposure to nonblurred adapting scenes, a result repeated in our own laboratory (data not shown). The observation here of a CS reduction, not improvement, at 0.5 cpd further adds weight that the results are not caused by simple learning effects. Having ruled out optical and learning contributions, the only remaining plausible explanation is that neural mechanisms are responsible for the phenomenon of blur adaptation; this is supported by the observation of interocular transfer of blur adaptation. ⁴ This study's findings of reduced CS at the lowest tested frequency and enhancements in the high-frequency range confirm that neural mechanisms play an active compensatory role in blur adaptation. 

Footnotes

 Supported by an Endeavour International Postgraduate Research Scholarship (NR) and a Science Faculty Scholarship from the University of Melbourne (NR), and a student allowance from the Department of Optometry and Vision Sciences of the University of Melbourne.

Footnotes

 Disclosure: N. Rajeev, None; A. Metha, None

The authors thank Phillip Bedggood for his constructive comments on a previous version of this manuscript.