Documente Academic
Documente Profesional
Documente Cultură
PI' 51 1-520
0 l T O M E T R Y AND VISION SCIENCE
Gopynght 0 1997 Amencan Academy o f O p ~ o m e t i ~
I
ORIGINAL ARTICLE
Centrefor Eye Research, School of Optometry, Queensland University of Technology, Brisbane, Australia (DAA); Department of Optometry
and Vision Sciences, UMIST, Manchester, United Kingdom (WNC);and Department of Vision Sciences, Glaspw Caledonian University,
Glasgow, Scotland (RLW)
ACT: An experiment is described in which the subjective depth-of-focus (DOF) of the eye, defined as the range
of focusing errors for which the image of the target appears t o have the same clarity, contrast, and form as the optimal
in-focus image, was measured as a function of the size of high contrast (99%) Snellen Es for 5 trained subjects under
cydoplegia. Mean DOF increased by approximately 60% as the size of the letter detail increased from -0.2 t o 0.87
log min arc (Snellen equivalent: 6 / 3 3 to 6/45), although there were considerable intersubject variations. DOF declined
with increasing pupil diameter, the mean total DOFs being 0.86, 0.59, and 0.55 D for 2-, 4-, and 6-mm pupils,
respectively. I n a second experiment, use of low (21 %) contrast letters with a 4-mm pupil and 4 subjects marginally
increased the DOF (by 0.08 0.05 D); refraction also shifted i n a myopic direction by a mean of 0.15 0.06 D
compared with the high contrast letters. A third experiment with four less-experienced subjects demonstrated the
importance of instruction and training i n any measurement involving judgment of just-perceptible defocus blur. The
clinical implications of the results for measurements of refraction and amplitude of accommodation are discussed.
(Optom Vis Sci 1997;74:511-520)
epth-of-focus (DOF) is a widely used concept in visual for which the image of the target appears to have the same clarity,
science which is of great clinical importance in relation to contrast, and form as the optimal in-focus image.l5>
refractive procedures and corrections. Cline et al.' distin- The first two criteria are objective criteria, the third is a subjec-
guish between "depth-of-focus" and "depth-of-field", the former tive criterion related to measuringvisual performance, the fourth is
being an image quantity, whereas the latter is an object quantity. a subjective criterion relating to the through-focus characteristics
Smith and Atchison2 make no distinction between the two quan- of the retinal image of a dynamic speckle pattern formed in coher-
tities, a practice which we shall follow in this paper, defining them ent light, and the last two are subjective criteria related to the
as the greatest range of dioptric focusing error which does not perception of "blur." Whichever criterion is used, the DOF results
result in objectionable deterioration in the retinal image quality. depend upon a number of target factors including luminance,
Various criteria have been used to specify what constitutes an color, shape, and size, as well as factors relating to the subject, such
"objectionable deterioration in the retinal image quality."3~ These as pupil size and ocular aberration.
include, but are not limited to, the range of focusing errors: (1) for The last of these criteria has widespread relevance to subjective
which the accommodation response does not change (the "accom- refraction and to estimates of the amplitude of accommodation or
modative dead space")5'6-this is most appropriately measured near additions as determined in clinical practice. Reliable, accurate
with an infrared optometer; (2) for which changes in the objec- refractions and amplitude measurements depend on the precision
tively measured line spread function do not exceed some chosen of the patient's judgments about the distinctness of high contrast
tolerance7; (3) for which the visual acuity or contrast sensitivity letters or other targets. A large DOF will reduce the reliability and
does not decrease by more than a certain amount or below a certain accuracy of the refraction and increase the subjective amplitude of
limit, e.g., 616 visual acuity3**' 9; (4) for which detection of the accommodation. There is thus considerable interest in how the last
direction of speckle motion in a laser optometer is not possi- criterion (6) influences DOF and hence the related clinical measures.
ble'', 'I; ( 5 ) for which changes in contrast are not detected for a There have been relatively few systematic investigations of the
target dioptrically modulated in a sinusoidal manneri2 I*; and (6) focusing range for which a target appears clear to a subject. Miles17
found a value of the order of 20.19 D in a clinical population, but et al.,23who found when using the push-up method that the mea-
did not control for pupil size. Campbell15 determined the subjec- sured amplitude was smaller when the target detail was close to the
tive DOF by placing black spots (10 min arc in subtense) on three subject's acuity limit; a very similar result was obtained by Rosen-
glass plates. The fixed, middle plate was 50 cm from the subject. field and ohe en.^^
The subject, with free accommodation, focused on its black spots It is worth noting the difference between legibility (or recogni-
and the front and back plates were moved toward the middle plate tion, e.g,. visual acuity) and DOF. From the point of view of
until the spots on them were just judged to be in sharp focus. The reading performance, larger letters can be read with greater
D O F was found to be approximately inversely proportional to amounts of defocus because only a few cycles per letter are required
pupil size when retinal illuminance was kept constant. For 3-mnl to produce legibility.25Thus even though a large, defocused letter
pupils under photopic conditions, the D O F in 7 subjects varied may be readable it may also appear blurred due to the subthreshold
from 20.33 to 2 0 . 5 4 D, with a mean of 20.44 D. It is possible contrast of the high spatial frequency components in its image.
that this study somewhat overestimated DOF, because with free Hence, measures of the effects of defocus on DO12 and legibility are
accommodation and changes in fixation, the subjects might have expected to differ.
altered their accommodation to focus midway between the central The aim of the present study was to explore further the possible
and each extreme target in turn, rather than strictly maintaining dependence of DOF on target detail size, using letters covering a
focus on the central target at all times. range of sizes. Pupil diameter and target contrast were also in-
Oshima,18 following Von Bahr,19 viewed an edge target cluded as variables, because these have previously been found to
through two sides of an optical system (the bottom half through affect DOF.15-
the left-hand side and the top half through the right-hand side) and
made judgments as to when the bottom half of the target appeared
blurred when the left-hand channel was lengthened or shortened.
His results show extremely small D O F values (approximately
2 0 . 0 5 D); Ogle and Schwartz3 suggested that the published values The right eyes of nine subjects in good ocular health and with
might be too low by a factor of 10 and it may be that Oshima was modest refractive error were used. Table 1 gives
- details of the
in fact detecting slight lateral movements of one image and hence subjects. Cycloplegia was attained either by using 1% cyclopento-
that he was making vernier rather than blur judgments. late for the younger subjects or 0.5% cyclopentolate for the older
Jacobs et al.,16 using their "observer" method, found blur subjects. All subjects gave informed consent to participating in the
thresholds for high contrast Landolt Cs (>O.O log min arc letter study. Corrected acuities were recorded at high (95%) and low
detail) which corresponded to a D O F of 20.18 D for a 4.2-mm (1 5%) contrast using Australian vision chartsz6with a background
pupil. However, observers were found to be more sensitive to de- luminance which approximately matched that used in the DOF
focus when smaller Landolt targets were viewed, the D O F falling experiment (85 cd/m2); contrast is defined here as
to approximately 20.1 0 D for a -0.2 log rnin arc target detail
which was at the acuity limit of the subjects.
This last finding is particularly interesting, because it has often where Ln,,., is the luminance of the background and Lmii1is that of
been considered on somewhat anecdotal grounds that D O F in- the letter
creases with the size of the target detail. Thus, as a subjective
refraction proceeds, the patient's attention is drawn to smaller and
eriment 1: Effect of letter size an il size on
smaller letters as visual acuity improves, in the hope that this will
DOE
enhance the reliability of the procedure. Similarly, small detail is
often recommended for clinical measurements of the subjective Five experienced subjects were used in this experiment. All were
amplitude of accommodation.2022This is supported by Atchison given preliminary training runs before the formal experiment.
'TABLE 1 .
Details of subjects.
DAA -2.00
WNC -0.25
RLW +0.50
MLG Plano
NLC -2.751-0.25 X 90
JMW -0.50 X 180
PGS -0.75
CLW -0.371-0.25 X 75
ARB Piano
With both contrast levels (Fig. 5), again there is a small but
significant effect ofletter size (p = 0.004). Under the conditions of
the experiment any differences in D O F at the two contrast levels
are very small. D O F at low contrast, averaged over all subjects and
letter sizes, is only slightly greater than that for high contrast by
far
DAA 0.08 k 0.05 D, which is of marginal statistical significance (p =
near 4mm high contrast 0.06). The effect of contrast is due to one subject (RLW), because
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 when the analysis is repeated without this subject, there is no
Target detail (log minutes of arc) significant effect (p = 0.14), whereas the effect of letter size re-
mains the same (p = 0.004). This same subject did not follow the
general trend in experiment 1, again suggesting that he used a
different criterion from the other subjects.
It is interesting to note that there is a systematic myopic shift in
the refraction (i.e., dioptric centerpoint of the range of clear focus)
when the contrast is low for some subjects (p = O.OI), the effect
being largest for the smallest, -0.2 log min arc letter detail (Fig. 6).
The mean value of this shift across subjects is 0.15 ? 0.06 D.
Again, we suspect this may be due to varying amounts of under-
corrected spherical aberration in the different subjects, the lower
spatial frequency content being relatively more important with the
low contrast images, for which visual acuity is lower (see Table 1).
Comparison of the high contrast results of this experiment with
those obtained by the same subjects in experiment 1 showed a good
level of agreement (Fig. 7), confirming that the subjects were ap-
plying consistent criteria over a period of the order of 1 week (no
et detail (log minutes of arc) significant effect of session, p = 0.24).
Typical data for the change in far and near settings with letter size in
experiment 'I. Pupil size was 4 mm. Results for two subjects are shown
(DAA and WNC). The zero in the ordinate scale is arbitrary but increasing
negative target vergences indicate an increasing myopic position of focus. Once more the overall results of this experiment confirmed the
Error bars indicate 21 SD. small but significant effect of letter size on D O F (p < 0.001). The
DA A
WNC
RLW
MG
NC
9 \ \ mean
1
0.0 T - -
-0.4 - 0 . 2
- v - r - - " - ~ - - -
0.0 0.2 0.4 0.6 0.8
1
1.0
0.0 - ~ - - T - - - T - ~ - - T - - - T T ~ - - -
-0.4 -0.2 0.0 0.2 0.4 0.6
minutes of
0.8 1.0
arc)
Target detail (log minutes of arc)
Mean results for total D O F for all pupil sizes in experiment 1 . Error bars
indicate 2 1 SD.
high contrast
low contrast
1.O - - DAA
3 WNC
0
0 RLW
- ,% MG
*
0 NC
0.5 mean
0.0
0.4 -0.2 0.0
--
0.2 0.4 0.6 0.8 1.0 1
0.0 - - - , ' - - - l - - Y - - - - l - ~ "
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
minutes of arc)
minutes of arc)
Mean DOF for both low contrast letters and high contrast letters as a
function of letter size for the four subjects in experiment 2. Pupil size was
4 mm. Error bars indicate SD of group results.
DAA
of the other three subjects appeared to reduce after the more de-
WNC tailed instructions and this effect appeared most marked for the
RLW
MG largest letters (Fig. 8, B to D). The apparent reduction in D O F
NC after detailed instruction was not statistically significant (p =
mean
0.1 I), although this may be a function of the small sample size.
This emphasizes the importance of instruction in studies of the
present type where, due to the experimental design, the subject's
criterion is not controlled. Practice may also play a role in reducing
0.0
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 thresholds and decreasing D O F values.15
Target detail (log minutes of arc)
Results for total DOF as a function of letter size for (A) Zmm, (B) 4-mm,
and (C) 6-mm pupils in experiment 1 . In each case, the thin lines give the
The dominant finding common to all three experiments is the
data for individual subjects the filled circlc?sand thici< lines the small depes~denceof D O F on letter ske within the range of sizes
means for all subjects. studied. Despite the large between-subject variance, ANOVA re-
vealed sienificant within-subiect effects. There was a significant
effect ofboth pupil size and letter size on the measured D O F in
two repeated sets of DOF measurements with less experienced experinlent 1. 'Tl~etrend for increasing D O F with increasing letter
subjects showed that although negligible changes occurred be- size was significant for all three pupil sizes (p < 0.001, p = 0.004,
tween measurements after the simple and detailed in~truction~ for p < 0.001 in experiments 1, 2, and 3, respectively).
the oldest, relatively most experienced subject (Fig. 8A), the D O F This dependence of D O F on letter size may be due to a change
FIGURE 6.
-3 -0.6- 4
- 0 , 4 -0.2 0.0
.%
.
0.2
1
0.4
Target detail (log minutes of arc)
.-..
. ..--
+
0.6
JMW
mean
0.8 1.0
-
subjective DOF must depend in some way on the changes, with
focus, in the spatial frequency content of the retinal image.30This RLW 4mm Run 1
involves two important elements: the change in the modulation Run2
transfer function (MTF) of the eye with defocus and the spatial
frequency spectrum of the target under observation.
As an example of the changes in modulation transfer, Fig. 9A
shows the focus dependence of the MTF for an aberration-free
model eye at a wavelength of 500 nm with a pupil diameter of 2
mm. Real eyes will, of course, suffer from aberrations but there is
typically little monochromatic aberration with a 2-mm pupi131and
longitudinal chromatic aberration only has minor effects with
small pupils.32Note in particular that the relative changes in mod-
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
ulation transfer with defocus occur at spatial frequencies approxi- Target detail (log minutes of arc)
mately halfway to the optical cut-off frequency of 70 cpd under
these conditions (i.e., at about 20 to 50 cpd).' Also shown in Fig. FIGURE 7.
9A is the typical contrast threshold at the retinalbrain level." It can DOF for high contrast letters as a function of letter size, with 4-mm pupil
diameter. Error bars indicate SD. Results are shown for each of three
be seen that this rises relatively slowly from a value of about 0.005 subjects who repeated the measurements after an interval of approxi-
at 5 cpd to about 0.1 at 60 cpd. mately 1 week. Note that the data are quite reproducible, suggesting that
The two-dimensional spatial frequency spectrum of the "E" the criteria for judging blur were used consistently.
targets used in the experiment is relatively complex. For simplicity,
Fig. 9B shows one-dimensional spectra for targets consisting of
three long, parallel dark bars separated by light bars ofequal width, ingly, broad peaks exist in the continuous 3-bar spectra which can
analogous to the three limbs of the letter E. The spectra shown are be related to the discrete fundamental and harmonics of the famil-
for targets where the bar widths correspond to the -0.2,0.1, and iar square-wave spectrum. The important feature of these spectra is
0.87 log min arc letter detail used in the experiments. Not surpris- that, within the spatial frequency band of interest, the "fundamen-
simple
0.0L
0.4
-7-----"-r-
-0.2 0.0 0.2
Target detail (log minutes of arc)
0.4
detailed
AB 4mm
simple
detailed 7
,
o . o ~ - - , 8
DOF for high contrast letters as a function of letter size after simple instruction and then after detailed
instruction i n experiment 3. Pupil size was 4 mm. Results are shown for each of four subjects, with error
bars indicating SD
tal" peak of the target moves to progressively lower spatial frequen- Lastly, as the letter size increases further, for example to +0.87
cies and its relative amplitude increases as the target gets larger. For log min arc, the fundamental peak of the spectrum now lies at low
example, the fundamental peak for the smallest target (bar width spatial frequencies where relative changes in modulation transfer
-0.2 log min arc) is located at approximately 48 cpd, whereas that with defocus are small. The small harmonic peaks at higher spatial
for the largest
- (-0.87 log- min arc) is at about 4 cpd. frequencies will be much more affected by focus changes. These
Fig. 9 gives useful insights into the way in which an observer's harmonics influence the retinal illuminance gradients at the edges
criterion ofdefocus might change over the range ofletter sizes used. of the bars, so that the defocus criterion may shift to being one
For the smallest, -0.2 log min arc letter detail target, the funda- based on edge sharpness rather than bar contrast.
mental occurs at approximately 48 cpd, a frequency at which mod- As noted earlier, aberration and -pupil - diameter will influence
ulation transfer is always low and retinalbrain threshold relatively the details of the through-focus MTF changes, whereas the three-
high. Thus, most subjects are unlikely to detect this basic period- bar spectrum is a simplification of the true two-dimensional E
icity, even at optimal focus. This means that detection of focus spectrum. Furthermore, the detection of image change is more
change is likely to be based on lower spatial frequency information likely to be related to detectable contrast changes13 than to the
which primarily affects the overall contrast of the dark patch con- absolute thresholds illustrated in Fig. 9.Nevertheless, the sugges-
stituting the retinal image of the letter. tion that observers are likely to use slightly different defocus criteria
As the letter detail size increases to, for example, 0.1 log- mm arc, with different target sizes within the range studied would appear to
the fundamental peak of the pattern (approximately 24 cpd) moves be generally valid.
into the spatial frequency band where modulation transfer is most In general the present data confirm that, with proper instruc-
sensitive to focus change, whereas the "harmonic" peaks will be less tions, subjects can detect quite small changes in the focus of letter
easily detected. Thus, it is reasonable to expect a focus criterion targets, sensitivity to defocus improving as the pupil diameter is
which is largely based on the apparent contrast otthe dark and light increased. The overall mean DOFs and standard deviations of the
bars of the target. group data for all subjects and letter sizes in experiment 1 are
1.o
k -
*
k
.
0
(0
0.8 diameter pupil
.
gg
0-4-
Present study
<= 0.6
0 u 0.3 -
h- 22
0.4 O 2
e
.-
0 ;! 0.2-
2
&-
is 5
0.2 0.1 -
3 cobs, Smith and Chan 4.2mm
0
5 0.0 0 . 0 - 1 , 8 , 8 . , . L , , . , . I -
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
Target detail (log minutes of arc)
0 10 20 30 40 50 60 70 FIGURE 10.
Spatial frequency (cyclesldegree) Comparison of our results for the dependence of the error of focus for
just-detectable blur of our high contrast Snellen E objects and a 4-mm
pupil diameter, with the results of Jacobs et al.lf3for high contrast Landolt
'C" objects and a 4.2-mm diameter pupil. Although the magnitude of the
DOFs are different, due to the different psychophysical procedures and
other conditions in the two studies, both studies show an increase in DOF
2 with letter size over the range studied. Error bars indicate SD of subject
1 groups.
0
-1
-2
-3 merit), although influenced by changes in subject criterion, was
-4 - -0.2 log min arc
-5 chosen to reflect clinical measurement procedures. Hence, it is not
-6 - 0.1 log min arc surprising that our measured D O F values, which are larger than
-7 . ,
.................... 0.87 log min arc with a less criterion-dependent method, were closer to values re-
ported in clinical evaluations and are broadly similar to the levels of
reliability found in current subjective methods of clinical refrac-
t i ~ n . ~ ~
It is also interesting to compare the results with the tolerance to
Spatial frequency (cyclesldegree) defocus found by Burton and Haigh35 for an effectively aberration-
FIGURE 9. free eye in monochromatic light. They found that they could dis-
A: MTFs for a diffraction-limited eye with a pupil diameter of 2 mm criminate a defocused picture from an in-focus picture with 90%
working at a wavelength of 500 nm. Each MTF is labeled with the certainty when the defocus wavefront aberration was about 0.29
corresponding error of focus in diopters. Also shown is the contrast wavelengths for a 2-mm pupil (approximately 20.32 D), quite
threshold function at retinalbrain level as measured by W i l l i a m ~using
~~ close to our value of 2 0 . 4 3 D. Burton and Haigh suggested that
interference fringes. B: Spatial frequency spectra for targets consisting of
three long, parallel bars with widths and interbar spacing of -0.2,O.I and
for larger pupil sizes there will be a rough balance between the gain
0.87 log min arc. The vertical arrows indicate the spatial frequencies due to diffraction and the loss due to aberration, so that D O F will
("fundamentals") corresponding to the reciprocal of the center-center remain at a roughly constant dioptric level. The present data sup-
spacing of adjacent dark bars in each target. port earlier authors in suggesting that Burton and Haigh may be a
little pessimistic about the degrading effects of aberration with
0.86 ? 0.28, 0.59 2 0.20, and 0.55 2 0.20 D for the 2-, 4-, and larger pupils and that dioptric D O F continues to reduce for pupils
6-mm pupils, respectively. These compare quite well with the larger than 2 mm, although not to as great a degree as would be
value of 0.88 D found by Campbell15 for a 3-mm pupil using a expected from diffraction considerations alone.
similar method of adjustment but are rather greater than the equiv- Somewhat surprisingly, the results with low contrast letters in
alent mean value of 0.36 D found by Jacobs et al.16 for a 4.2-mm experiment 2 show, at most, only minor increases in D O F (Fig. 5),
pupil using their forced-choice observer method. Jacobs et al.'s which were largely attributable to one subject. This disagrees with
results, which were for the change of focus from its optimal value the results of Campbell15and Oshima,18 each ofwhom found that
that was required to produce threshold blur (i.e., a value equal to for a 3-mm pupil decreasing the contrast from 99% to 21% in-
one-half the D O F as defined in our paper), are compared with creased D O F by a factor of the order of 2. We are unable to explain
those of our study as a function of letter size in Fig. 10. For the the discrepancy, although experimental techniques and procedures
purposes of comparison our own D O F values have been halved to were different. Larger differences might have been observed with a
make them compatible with Jacobs et al.'s original data. Note that lower contrast. Our results imply that accurate refractive correc-
although our values are higher, which is probably due to differ- tion is as important for the observation of detail at low contrast as
ences in the task, trends are similar. Increase in D O F is slower with at high contrast.
increase in letter size to approximately 0.0 log min arc in our study, It is, however, interesting to note that, with a 4-mm pupil,
but is more rapid at larger sizes. Our experimental method (adjust- refraction may shift in the myopic direction for some subjects (Fig.
G), due presumably to the influence of undercorrected spherical rion with larger letters, as observed with some of our less experi-
aberration. Such a shift may be of marginal clinical significance for enced subjects in experiment 3 (Fig. 8) and in the amplitude
some patients when using low contrast test objects. The effects on measurements of Rosenfield and ohe en.^*
refraction and visual acuity were not tested in our study.
The limited study with less-experienced subjects emphasizes the
importance of instruction set, training, and experience in subjec-
tive judgments involving DOF. Even so, it appears that the sub- The present data support the view that ocular DOF as measured
ject's criterion is flexible, changing with letter size. In some subjects by the range over which no change in the quality in the image of a
DOFs were reduced by a factor of 2 after further, more detailed letter can be detected is, for many subjects, affected by the subtense
instruction, although the effects for different subjects were idio- of the letter. Greater sensitivity to defocus is found with letters
syncratic (Fig. 8). Our subjects had an optometry background, and which are close to the acuity limit. This finding favors the hypoth-
it is likely that the magnitude of effects of instruction would have esis that performance is determined by the spatial frequency spec-
been greater with subjects who were more naive regarding visual trum of the target. Changes in subject criterion with letter size are
judgments. Layton et a l . have previously shown that a simple implied by our results, with significant intersubject differences
instruction set can lead to the use of very different blur criteria by apparent. It also lends support to the approach of using small
naive subjects. Similarly, Campbell15suggested that measured sub- letters to judge optimal focus during refractive procedures, and to
jective DOFs may reduce with practice and familiarity with the assess amplitudes of accommodation. With trained subjects, using
apparatus used. stringent blur criteria, a 4-mm pupil, and high contrast, the mag-
As stressed already, the mean results shown in Figs. 3 and 4 nitude ofthe DOF found ranges from approximately 20.22 D for
support the reduction in D O F found by Jacobs et al.16 for targets -0.2 log min arc letter detail to 20.39 D for 0.87 log min arc
lying close to the acuity limit. Similar effects are found in the other letter detail.
experiments. In the present data for a 4-mm pupil, DOF increases An interesting finding is that, when low contrast letters are used,
by a factor of approximately 1.8 between -0.2 and 0.87 log min subjects may show a small myopic shift in their refraction, even
arc letter derail. This is the same as was found by Jacobs et al. under though the reduction in contrast has only minor effects on DOF.
broadly similar conditions, although the exact level and size depen- The shift in refraction is attributed to the combination of under-
dence of the D O F is different (Fig. 10). Exact agreement cannot be corrected ocular spherical aberration with the relatively greater
expected, due to the different psychophysical methods (staircase in importance of low spatial frequencies in the images of low contrast
Jacobs et al. vs. method of adjustment in this study), different letters. Because optimal focus is spatial frequency-dependent in the
defocus (i.e., by experimenter vs. by subject), different presenta- presence of spherical aberration, with optimal focus for low spatial
tion methods (static vs. dynamic), and different targets (Landolt C frequencies lying closer to the marginal focus, it follows that refrac-
vs. Snellen E) used. tion for low contrast letters corresponds more closely to the mar-
It seems probable that the considerable intersubject differences ginal focus than that for high contrast letters, where higher spatial
shown in Fig. 3 are attributable to criterion differences rather than frequencies are above threshold.
to differences in acuity because experiment 3 demonstrates that
subjects may alter their decision criterion. There is also the possi-
bility that some subjects are better able to detect blur, although this
seems unrelated to visual acuity. We thank Arthur Bradley and Larry Thibos for commenting on the manu-
The differences in D O F with letter size found in the present script, the Borish Center for Ophthalmic Research for supporting a visit by
study (ranging from 0.25 D for -0.2 log min arc letter detail to DavidAtchison to the School of Optometry at Indiana University, and Queen-
stand University of Technology for supporting Neil Charman as a Visiting
0.44 D for 0.87 log min arc letter detail) are too small to explain the Research Fellow to the Centrefor Eye Research.
much larger differences (approximately 1 to 2 D, depending upon Received November 1, 1996; revision received April 18, 1997.
age) found in clinical measures of amplitude of accommodation by
Atchison et al.23when using different sizes of letter target to judge
blur. Several author^^--^^ have found that the accommodation
response to gratings of high spatial frequency is less accurate than 1 Cline D, Hofstetter HW, Griffin JR. Dictionary of Visual Science.
that to intermediate spatial frequencies around the peak of the 3rd ed. Radnor, PA: Chilton Book Company, 1980.
contrast sensitivity function (approximately 5 cpd). Hence, ampli- 2. Smith G , Atchison DA. The Eye and Visual Optical Instruments.
tudes of accommodation measured when observing high spatial New York: Cambridge University Press, 1997.
frequency detail (small letters) could perhaps be smaller than those 3. Ogle KN, Schwai'tz JT. Depth of focus of the human eye. J Opt Soc
for detail of predominantly intermediate spatial frequency. Al- Am 1959;49:273-80.
though the contrast at each spatial frequency varies with letter size 4 Lcgge GE, Mullen KT, Woo GC, Campbell FW. Tolerance to visual
dcfocus. J Opt Soc Am A 1987;4:851-63.
( e . Fig. 9B), this seems unlikely to be the explanation of the
5. Ludlam WM, Wittenberg S, Giglio EJ, Rosenberg R. Accommoda-
results of Atchison et aLZ3because their accommodation stimuli
tive responses to small changes in dioptric stimulation. Am J Optom
always contained a broad band of spatial frequencies including Physiol Opt 1968;45:483-506.
many lower spatial frequencies due to the nature of the near read- 6. Kotulak JC, Schor CM. The accommodation response to subthresh-
ing chart (i.e., lines ofwords). It seems more likely that the changes old blur and to fading during the Troxler phenomenon. Perception
in accommodative amplitudes reported by Atchison et al.23 re- 1986;15:7-15.
sulted from the subjects adopting a much more relaxed blur crite- 7. Charman WN, Jennings JAM. The optical quality of the monochro-