transient lumanopia at high intensities
TRANSCRIPT
www.elsevier.com/locate/visres
Vision Research 44 (2004) 3203–3209
Transient lumanopia at high intensities
Adam Reeves *, Shuang Wu
Department of Psychology, Northeastern University, 125 NI, 360 Huntington Avenue, Boston, MA 02115, USA
Received 2 September 2003; received in revised form 21 May 2004
Abstract
Transient lumanopia is the loss of sensitivity to a flicker burst presented in early dark adaptation. We earlier reported that luman-
opia with 18Hz flicker was greatest after turning off 400td adapting fields and progressively disappeared as the adapting field was
intensified. We now report that lumanopia can occur with intense adapting fields, but only with faster flickers (e.g. 40Hz, 5000td
fields). Dimming the field (but not brightening it) can also produce lumanopia. The results illustrate frequency-dependent attenu-
ation by a temporal filter whose parameters are set by light adaptation and which change abruptly when the field is dimmed or
extinguished.
� 2004 Elsevier Ltd. All rights reserved.
1. Introduction
Early dark adaptation has been studied extensively
with detection thresholds––the least light needed for an
observer to detect a small, brief, test flash. Detection
thresholds drop far and fast when an adapting field is
dimmed or turned off; for example, thresholds fordetecting foveal light spots can fall over 100 times just
one-tenth of a second after turning off an intense field
(Ahn & MacLeod, 1993; Baker, 1963; Boynton & Kan-
del, 1957). This rapid initial recovery is clearly distinct
from the much slower processes which intervene to re-
store photopic vision after several minutes in the dark
(Baker, 1963; Hecht, Haig, & Chase, 1937). However,
vision does not always recover in early dark adaptation.Eisner (1989) found that red flicker thresholds can rise
following extinction of bleaching fields, and Reeves
and Wu (1997) found that achromatic flicker thresholds
can rise after extinction of much dimmer fields. We
called this rather surprising desensitization of the lumi-
nance pathway transient lumanopia, by analogy with
0042-6989/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.visres.2004.07.015
* Corresponding author. Tel.: +1 6173734708; fax: +1 6173738714.
E-mail address: [email protected] (A. Reeves).
transient tritanopia, a somewhat similar desensitization
of the yellow/blue hue pathway (Mollon & Polden,
1979).
Observers in Reeves and Wu (1997) adjusted the
intensity of a white, 560ms, 100% modulated, 5.3 0 arc,
foveal flicker burst until they could just report seeing
flicker. While detection thresholds (i.e. thresholds forjust seeing the stimulus, whether it appeared to flicker
or not) always fell at the start of dark adaptation, flicker
thresholds sometimes rose quite markedly. The effect
was frequency specific. For example, the threshold for
seeing 10 and 12Hz flicker thresholds always recovered,
no matter what the field intensity, but thresholds for 18
Hz flicker rose 60-fold after turning off a white 400td
adaptation field. These frequency effects were specificto early dark adaptation; flicker sensitivity was flat be-
tween 10 and 18Hz when the flickering spot was pre-
sented either after long-term dark adaptation or on
any of the steady light-adapting backgrounds.
Transient lumanopia may be an example of a broad
class of phenomena known as �flicker response suppres-sion�, in which the visual response to the modulatedcomponent of the flickering stimulus is suppressed rela-tive to the visual response to its time-averaged
luminance. Flicker response suppression has been
3204 A. Reeves, S. Wu / Vision Research 44 (2004) 3203–3209
previously obtained with simultaneously-presented an-
nuli (Eisner, 1995), however, and we have only em-
ployed uniform fields. Flicker response suppression is
inferred from flicker thresholds measured by raising
the time-averaged luminance until flicker just becomes
visible, while maintaining a fixed modulation depth(e.g. Eisner, 1995; Eisner, Shapiro, & Middleton, 1998;
Snippe, Poot, & van Hateren, 2000). We also kept the
modulation of the test spot fixed (at 100%); only the
mean level was varied to reach threshold. In a different
tradition in flicker research, the mean level is fixed and
the modulation is varied. To obtain a complete account
of transient lumanopia, all four variables (field intensity,
test mean level, test modulation depth, and test fre-quency) will have to be varied.
An additional surprising finding in the Reeves and
Wu (1997) study, other than just the existence of the
effect, was that transient lumanopia at 18Hz grew rap-
idly with field intensity up to 400td, but then disap-
peared with much brighter fields. Transient tritanopia
also disappears with brighter fields (Mollon & Polden,
1979), as does its analogue in the red/green hue pathway(Reeves, 1983). One might then expect that transient
lumanopia would always disappear at higher field levels,
if it is produced by a similar mechanism. However, at
high light levels the temporal modulation transfer func-
tion shifts to higher frequencies (Kelly, 1971; Snowden,
Hess, & Waugh, 1995), suggesting that transient luman-
opia might also scale with flicker rate. In this case
lumanopia might be evident with adapting intense fieldsusing higher flicker rates. We therefore extended the
range of flicker rates used in the current study. We also
checked whether transient lumanopia could be obtained
if the field intensity was dimmed or enhanced, not just if
the eye was plunged into darkness.
2. Method
2.1. Stimuli
Stimuli were presented by a three-channel Maxwel-
lian view system, using the same methods to generate
and measure flicker as described in Reeves and Wu
(1997) except that the flickering spot was enlarged from
5.43 0 arc to 0.88� and the range of flicker rates was in-creased from 10–18Hz to 8–50Hz. Briefly, the 500ms,
100% modulated, flickering test spot was centered on a
10.4� white adapting field in the �ON� condition, or itwas presented 0.2 s after the adapting field had been
turned off (the �OFF� condition). The field remainedoff for a further 0.4 s, and then returned for 5s to top-
up light adaptation before the next trial could start
(see insert in the top left panel of Fig. 1). The fieldwas fully extinguished within 1ms by advancing a vane
attached to a loudspeaker coil into a 2 mm diameter no-
dal point in the beam. To dim (intensify) the field in later
experiments, a neutral-density gelatin filter was ad-
vanced by the vane into (out of) the nodal point, with
the same timing as in the OFF condition.
Thresholds were measured with a brief flicker burst
and 5s between trials to limit adaptation to the flicker(Pantle, 1971), although some adaptation to the dc com-
ponent of the flicker may well have occurred. Using
brief, infrequent flicker bursts helps ensure that the state
of light adaptation is controlled by the steady adapta-
tion field, rather than by the burst. The intensity of
the test beam was ramped up by 0.3 log units over the
first 100ms of the 500ms exposure and ramped down
again over the last 100ms. This soft-shoulder envelopewas also used in Reeves and Wu (1997) to make it less
likely that flicker would be detected from transients at
the start and end of the burst (Smith, 1970). Reeves
and Wu (1997) showed that the flicker thresholds were
elevated and stable from 0.1 to 1s after field offset. (Such
stability is important in distinguishing lumanopia from
masking-by-flash, in which thresholds recover within
0.1s after turning off the field; Boynton & Kandel,1957; Geisler, 1983.) In the present study we employed
a fixed delay of 0.2 s from field offset to stimulus onset.
The waveform generated by the test flicker was
approximately sinusoidal, and had full amplitude from
8 to 60Hz, as seen on an oscilloscope driven by a pin
diode in the path of the light beam and triggered by
the rising phase of the wave. In Reeves and Wu (1997)
the waveform was square. The change from square tosinusoid was undertaken so that the range of flicker
rates could be extended down to 8Hz without concern
for higher harmonics. At higher rates the change will
only affect flicker sensitivity via the change in the ampli-
tude of the fundamental, as the higher harmonics in the
square (and the slight distortions in our sinusoid) are
invisible at threshold (De Lange, 1958).
On each trial the subject judged whether he or shecould see the test flicker, to obtain flicker thresholds,
or, in other trials, whether the test was visible or not inde-
pendently of whether it seemed to flicker (detection
thresholds). Tests flickering at intensities between the
detection and flicker thresholds were visible, but ap-
peared smooth. The test was intensified after a No re-
sponse and dimmed after a Yes response. The step size,
initially 0.3 log units, was halved if the response changedand was doubled (up to a maximum of 0.3 log units) if it
was the same for three successive responses, a useful trick
to avoid long runs of below threshold trials. The final va-
lue of the wedge was recorded when the step size reached
0.02 log units. After five final values were obtained in suc-
cession, the computer calculated their mean, which was
taken as the threshold. Results are the means of two
thresholds obtained on different days.The photopic absolute threshold for detection was
obtained after 3min dark adaptation at the start of each
ME 40 Hz extinction
log td in field
0
1
2
3
4
lumanopia
recovery
detection:OFF
flicker:OFF
flicker:ON
detection:ON
AR 40 Hz extinction
log td in Field
0
1
2
3
4
lumanopia recovery
flicker:OFF
flicker:ON
AR 14 Hz extinction
log td in field
0
1
2
3
4
lumanopia recovery
flicker:ON
flicker:OFF
ME 14 Hz extinction
0
1
2
3
4
recoverylumanopia
detection:OFF
detection:ON
flicker:OFF
flicker:ON
SW: 10 Hz extinction
log td in field
0
1
2
3
4
recovery
flicker:OFF
flicker:ON
AR: 10 Hz extinction
log td in field1 5
0
1
2
3
4Field OnField On
Flickering test: 500 msDelay 200 ms
Wait 400 ms
Field Off
recoveryflicker:ON
flicker:OFFLog
Thre
shol
d R
e Ab
sLo
g Th
resh
old
Re
Abs
Log
Thre
shol
d R
e Ab
s
2 3 4 1 52 3 4
1 52 3 4 1 52 3 4
1 52 3 41 52 3 4
Fig. 1. Log threshold of the test spot relative to absolute threshold (Abs), plotted against log field intensity in td; 10Hz flicker (top), 14Hz (middle),
40Hz (bottom). Observers AR (left panels) and SW or ME (right panels). Open symbols: thresholds in extinction; closed symbols, on the steady field.
Circles: flicker thresholds. Squares: detection thresholds (for ME only).
A. Reeves, S. Wu / Vision Research 44 (2004) 3203–3209 3205
session. The observer then adapted to one field intensity
on some days, or, on other days, to an increasing series
of field intensities. After an increase of intensity the ob-
server re-adapted for 2min to a moderate field or 3min
to a bright field. As it took a full hour-long session to
complete 3 or 4 light levels, complete threshold-versus-
field intensity (tvi) curves were stitched together across
sessions. Field intensity was measured at the start of
each session with a calibrated UDT-10 pin diode placed
at a nodal point conjugate with the pupil.
3206 A. Reeves, S. Wu / Vision Research 44 (2004) 3203–3209
3. Results
3.1. Threshold versus intensity curves
Fig. 1 plots log10 flicker thresholds, relative to abso-
lute photopic threshold, against log field intensity in tro-lands (tvi curves). The ON flicker thresholds (solid
circles) increase monotonically with field intensity and
approach the Weber law (a slope of 1.0). The approach
to the Weber law is perhaps not surprising because on
bright fields, 10 and 14Hz flickers are visible almost at
detection threshold, and detections obey the Weber
law. The OFF flicker thresholds (open circles) are more
interesting. They show recovery at 10Hz (top two pan-els, for AR and SW), but they rise above the ON flicker
thresholds at 14Hz after extinction of the dimmer fields
(middle two panels, for AR and ME), showing luman-
opia. These results with 10 and 14Hz flickers serve to
replicate Reeves and Wu (1997), but with an enlarged
(0.88�) test spot.The bottom panels of Fig. 1 show the new result,
namely, evidence of strong lumanopia at 40Hz afterextinction of a moderately bright 1000 td field. Flicker
thresholds do show recovery at 40Hz, but only after
extinction of fields above 4000td (AR) or 10,000 td
(ME). We conclude that lumanopia can indeed be dem-
onstrated on bright fields by using fast flickers.
In contrast to the flicker thresholds, detection thresh-
olds recover in the OFF condition, typically falling half-
way to absolute threshold within 200ms of fieldextinction (Krauskopf & Reeves, 1980) (data for the
authors, AR and SW, are in Reeves, Wu, & Schirillo,
1998). As detection thresholds have not been published
for ME, they are plotted in the lower right panels by
squares (solid for ON, open for OFF). They show the
expected recovery at the start of dark adaptation. They
also show for ME at 14Hz what was shown for AR and
SW in Reeves and Wu (1997), namely, that the flickerthresholds are very close to the detection thresholds in
the light. Only in the dark do these thresholds diverge.
The difference between flicker and detection thresholds
illustrates flicker response suppression, as defined in
the Introduction, if it is assumed that the detection re-
sponse is based on the time-averaged luminance of the
test––a reasonable assumption, given that flicker rate
had no systematic effect on detection thresholds.
3.2. Flicker sensitivity curves
We next fixed the field intensity in each session, and
varied flicker rate across trials by stepping the wave gen-
erator. Flicker rate was stepped from low to high or high
to low in different sessions; as direction had little effect it
was averaged over. Either the lower half or the upperhalf of the frequency range was tested in a session.
Fields were dim (42–54 td), moderate (340–437td), or
bright (5370–5500td), corresponding to about 1.6, 2.6,
and 3.6 log td, as shown in the top, middle, and bottom
panels of Fig. 2. Columns in this figure show data for
observers AR, SW, and AF (ME was not run).
Log flicker threshold is plotted downwards in Fig. 2, a
standard convention for exhibiting flicker sensitivity(e.g. Kelly, 1971; Watson, 1986). A �0� at the top of eachplot, if present, would represent the photopic absolute
threshold. The slower flickers (8–12Hz, on the far left
of each plot) show the usual recovery, the OFF thresh-
olds (open circles) now being plotted above the ON ones
(closed circles). Thresholds for the faster flickers show
lumanopia above 20–30Hz, in that the open circles
now plot below the closed ones, except for observerAF at 5370td (lower right hand corner).
Sensitivity generally declines with flicker rate, but the
data are less good than in typical flicker experiments in
which the field is steady. Moreover, an increase of 2 log
units in field intensity should nearly double the cff meas-
ured at the fovea (Tyler & Hamer, 1990); the sensitivities
of our observers do increase with field level in the ON
condition, but not to this extent. Finally, there are someaberrant data points and large individual differences.
Perhaps the largest of these was at 3.74 log td (5500td),
where AF showed recovery whereas AR and SW both
showed lumanopia (Fig. 2). This difference may just re-
flect light adaptability, as AR, for example, does recover
at 4.2 log td (Fig. 1). However, AF also showed less
lumanopia at lower light levels. Thus the data, though
indicative of transient lumanopia, are probably not ofsufficient quality to warrant precise modeling.
3.3. Dimming or brightening the field
One way of explaining these data is to imagine that
flicker is controlled by a temporal filter whose parame-
ters are under the control of the adapting field. When
light level increases, the filter moves to the right on thefrequency axis, such that sensitivity increases to higher
frequencies (the standard account of the Ferry–Porter
law). At the start of dark adaptation, or after the field
is abruptly dimmed, it might be that the filter abruptly
shifts to the left, so that fast flickers which had been vis-
ible in the light are attenuated. If, on the other hand, the
field level is temporarily raised, the filter should not shift
left. These ideas predict that lumanopia should occurwhen the field is dimmed but not when it is brightened.
We tested both of these predictions in observers AR,
SW, and ME, and we also tested the dimming prediction
in AF.
Observers AR and ME were run to obtain tvi curves
with the field dimmed (by 0.8 log units), and the results
were similar to those obtained in Fig. 1 with the magn-
itudes of recovery and lumanopia somewhat reduced.Observers AR and SW were also run with flicker rate
varied (as in Fig. 2), and their results in the dimming
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
Log
Sens
itivi
ty (-
Log
Thre
shol
d R
e Ab
s)Lo
g Se
nsiti
vity
(-Lo
g Th
resh
old
Re
Abs)
Log
Sens
itivi
ty (-
Log
Thre
shol
d R
e Ab
s)
Hz
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
AF: 5370 td
recovery
Hz
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
4.0
AR: 5500 td
recovery
lumanopia
Hz
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
4.0
recoverylumanopia
SW: 5500 td
10 20 30 40 50
0.5
1.0
1.5
2.0
2.5
3.0
3.5
AR: 437 td
recovery lumanopia
10 20 30 40 50
0.5
1.0
1.5
2.0
2.5
3.0
3.5
SW: 340 td
recovery
lumanopia
10 20 30 40 50
0.5
1.0
1.5
2.0
2.5
3.0
3.5
SW: 54 td
recovery
lumanopia
10 20 30 40 50
0.5
1.0
1.5
2.0
2.5
3.0
3.5
AF: 42 td
recovery
lumanopia ?
10 20 30 40 50
0.5
1.0
1.5
2.0
2.5
3.0
3.5
AR: 42 td
recovery
lumanopia
AF: 440 td
recovery
lumanopia
Fig. 2. Log sensitivity (�log threshold re Abs, or absolute threshold) for just detecting flicker, plotted against flicker rate for field intensities of42–64td (top), 340–440td (middle), 5370–5500td (bottom). Observers AR (left panels), SW (middle), and AF (right). Closed circles: on the steady
field. Open circles: in extinction.
A. Reeves, S. Wu / Vision Research 44 (2004) 3203–3209 3207
condition are shown in Fig. 3 (AR on the left, SW on the
right) at three field adaptation levels. Lumanopia oc-
curred at the faster rates for the middle field intensity
(central plots) for both observers, and at the highest field
intensity for AR.
Data obtained on AR and ME when the field level
was temporarily increased by the same amount (0.8 logunits) showed no effect of frequency. Flicker thresholds
simply increased to the same extent at every flicker rate
tested. In further tests on AR we found that raising the
field intensity by 0.3 log units and also by 1.2 log units
also had no frequency-selective effect (data not shown).
Thus we have seen no evidence of desensitization 200ms
after the field is brightened, which is consistent with the
well-known rapid dynamics of light adaptation (e.g. Wu,
Burns, Elsner, Eskew, & He, 1997).We regard the dimming results as useful because they
show that lumanopia can occur even when the field is
10 20 30 40 50
0.0
0.5
1.0
1.5
2.0
2.5
3.0
SW: 53 td
recovery
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
4.0
SW: 617 td
recoverylumanopia
Hz
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
4.0
SW: 5754 td
recovery
10 20 30 40 50
0.0
0.5
1.0
1.5
2.0
2.5
3.0
AR: 42 td
recovery lumanopia ?
Log
Sen
sitiv
ity
10 20 30 40 50
0.5
1.0
1.5
2.0
2.5
3.0
3.5
recovery
lumanopia
AR: 1550 td
Log
Sen
sitiv
ity
Hz
10 20 30 40 50
1.0
1.5
2.0
2.5
3.0
3.5
4.0
AR: 5500 td
recovery
lumanopia
Log
Sen
sitiv
ity
Fig. 3. Log Sensitivity (�log threshold re Abs) for just detecting flicker, plotted against flicker rate for field intensities of 42–53td (top), 517–1550td(middle), and 5500–5754td (bottom). Observers AR (left panels), SW (right). Closed circles: ON (steady field). Open circles: 200ms after the field has
been attenuated by 0.8 log units.
3208 A. Reeves, S. Wu / Vision Research 44 (2004) 3203–3209
clearly visible to the photopic luminance pathway
throughout the entire trial. It is not necessary to switch
to complete darkness to get the effect.
4. Discussion
The results show that transient lumanopia can be ob-
tained at high light levels using faster flickers, thereby
complementing the account given in Reeves and Wu
(1997) for lower light levels and slower flickers.
Although the physiological origin of lumanopia is not
known, we can assume that the neural visual system acts
as a low-pass filter, at least for flickers above 8Hz (be-
low this rate, other flicker-sensitive channels may inter-vene: Hammett & Smith, 1992; Snowden et al., 1995; but
see Mandler & Makous, 1984). To account for luman-
opia, we postulated that extinguishing an adaptation
field lowers the corner frequency of a low-pass neural fil-
ter (Reeves & Wu, 1997). Faster flickers will be attenu-
ated at the start of dark adaptation and show
lumanopia, whereas much slower flickers will remain in-
side the low-pass region even at field offset, and so will
not be attenuated.
In the earlier work, thresholds were flat across fre-
quency from 10 to 18Hz in the light, and so thresholdsin the OFF condition could be used to characterize
lumanopia directly. In the present work, thresholds var-
ied in the ON condition as well, so the OFF data are no
longer so easy to interpret. Moreover, individual differ-
ences were striking, the drop in corner frequency being
substantial AR, smaller for SW and minor for AF (as
is obvious in Fig. 2). Finally, the method of varying
the time-averaged luminance until flicker just becomesvisible, which is used in studies of flicker response sup-
pression (e.g. Eisner, 1995), does not reveal the effect
(if any) of varying test modulation. Thus we do not pre-
sent a formal model of these data.
A. Reeves, S. Wu / Vision Research 44 (2004) 3203–3209 3209
Boynton et al. (1961) studied the threshold of an im-
pulse delivered at various moments during each cycle of
a flicker burst. Although impulse thresholds track the
sinusoidal shape, in that the thresholds were nearly in
phase with the sinusoid in the 20–50 Hz range, the wave-
form is distorted and the mean threshold is elevated dur-ing the flicker. A non-linearity due to a contrast gain
control may explain the elevation (Wu et al., 1997).
Above threshold, both AR and SW show strong non-
linearities in the perception of flicker bursts presented
on steady fields (Wu, Burns, Reeves, & Elsner, 1996;
unfortunately, AF and ME were not available for testing
in this respect). Flicker response suppression may be ex-
plained by a form of saturating non-linearity (Eisneret al., 1998; Snippe et al., 2000). Measurements of these
other types of flicker responses in early dark adaptation
may be useful in developing a formal model for transient
lumanopia.
Finally, we remark on the generality of transient
desensitization at the start of dark adaptation. We have
found that the luminance pathway as assessed by flicker
can lose sensitivity (transient lumanopia), as was foundearlier for transient tritanopia in the yellow-blue path-
way (Mollon & Polden, 1979) and for an analogous
desensitization in the red/green hue pathway (Reeves,
1983). Even though the mechanisms must differ, in that
the hue pathway desensitization effects are spectrally
opponent (Mollon, 1982), there are some marked simi-
larities: both sensitivity losses are long-lasting (1s or
more), both depend strongly on field intensity, and bothcan be substantial (60-fold or more). The initial recovery
seen in classical dark adaptation curves may reflect the
use of low temporal frequency/long duration luminance
stimuli, rather than being a typical feature of vision in
early dark adaptation.
References
Ahn, S. J., & MacLeod, D. I. A. (1993). Link-specific adaptation in the
luminance and chromatic channels. Vision Research, 33,
2271–2286.
Baker, H. D. (1963). Initial stages of dark and light adaptation.
Journal of the Optical Society of America, A, 53, 839–844.
Boynton, R. M., & Kandel, G. (1957). On responses in the human
visual system as a function of adaptation level. Journal of the
Optical Society of America, 47, 275–286.
De Lange, H. (1958). Research into the dynamic nature of the human
fovea-cortex systems with intermittent and modulated light.
Journal of the Optical Society of America, 48, 777–784.
Eisner, A. (1989). Losses of foveal flicker sensitivity during dark
adaptation following extended bleaches. Vision Research, 29,
1401–1423.
Eisner, A. (1995). Suppression of flicker response with increasing test
illuminance: Roles of temporal waveform, modulation depth, and
frequency. Journal of the Optical Society of America, A, 12,
214–224.
Eisner, A., Shapiro, A. G., & Middleton, J. A. (1998). Equivalence
between temporal frequency and modulation depth for flicker
response suppression. Journal of the Optical Society of America, 15,
1987–2002.
Geisler, W. S. (1983). Mechanisms of visual sensitivity: Backgrounds
and early dark adaptation. Vision Research, 23, 1423–1432.
Hammett, S. T., & Smith, A. T. (1992). Two temporal channels or
three? A re-evaluation. Vision Research, 32, 285–291.
Hecht, S., Haig, C., & Chase, A. M. (1937). The influence of light-
adaptation on subsequent dark-adaptation of the eye. Journal of
General Physiology, 20, 831–850.
Kelly, D. H. (1971). Theory of flicker and transient responses:
Uniform fields. Journal of the Optical Society of America, 61,
537–546.
Krauskopf, J., & Reeves, A. (1980). Measurement of the effect of
photon noise on detection. Vision Research, 20, 193–196.
Mandler, M., & Makous, W. (1984). A three channel model of
temporal frequency perception. Vision Research, 24, 1881–
1887.
Mollon, J., & Polden, P. G. (1979). An anomoly in the response of eye
to light of short wavelengths. Proceedings of the Royal Society of
London, 278, 207–240.
Mollon, J. D. (1982). Color vision. Annual Review of Psychology, 33,
41–85.
Pantle, A. (1971). Flicker adaptation. I. Effect on visual sensitivity to
temporal fluctuations of light intensity. Vision Research, 11,
943–952.
Reeves, A. (1983). Distinguishing opponent and non-opponent detec-
tion pathways in early dark adaptation. Vision Research, 23,
647–654.
Reeves, A., & Wu, S. (1997). Transient lumanopia: The invisibility of
flicker in early dark adaptation. Journal of the Optical Society of
America, A, 2509, 2516.
Reeves, A., Wu, S., & Schirillo, J. (1998). The effect of photon noise on
the detection of white flashes. Vision Research, 38, 691–703.
Smith, R. A. Jr., (1970). Adaptation of visual contrast sensitivity to
specific temporal frequencies. Vision Research, 10, 275–279.
Snippe, H. P., Poot, L., & van Hateren, J. H. (2000). A temporal
model for early vision that explains detection thresholds for light
pulses on flickering backgrounds. Visual Neuroscience, 17,
449–462.
Snowden, R. J., Hess, R. F., & Waugh, S. J. (1995). The processing of
temporal modulation at different levels of retinal illuminance.
Vision Research, 35, 775–789.
Tyler, C. W., & Hamer, R. D. (1990). Analysis of visual modulation
sensitivity. Validity of the Ferry–Porter law. Journal of the Optical
Society of America A, 7, 743–758.
Watson, A. B. (1986). Temporal sensitivity. In K. R. Boff, L.
Kaufman, & J. P. Thomas (Eds.). Handbook of perception and
human performance (vol. 1). New York: Wiley.
Wu, S., Burns, S., Elsner, A., Eskew, R. T., & He, J. (1997). Rapid
sensitivity changes on flickering backgrounds: Tests of models of
light adaptation. Journal of the Optical Society of America A, 14,
2367–2378.
Wu, S., Burns, S., Reeves, A., & Elsner, A. (1996). Flicker brightness
enhancement and visual nonlinearity. Vision Research, 36,
1573–1583.