Proc. Nat. Acad. Sci.

USA
Vol. 68, No. 11, pp. 2802-2805, November 1971

Flicker Responses in Monkey Lateral Geniculate Nucleus and Human

Perception of Flicker
(de Lange characteristics/microelectrodes/macaques/single cells/critical flicker frequency)

H. SPEKREIJSE*, D. VAN NORREN, AND T. J. T. P. VAN DEN BERG*

Institute for Perception TNO, Kampweg 5, Soesterberg, The Netherlands
Communicated by Floyd Ratliff, September 2, 1971

ABSTRACT An analysis was made of the impulse dis- scales. Examples of such curves of flicker sensitivity versus
charge patterns evoked by sinusoidal luminance modula- frequency (de Lange curves) are given in Fig. 1. These psy-
tion-of single cells in the lateral geniculate nucleus of the
macaque monkey. The goal was to determine whether a chophysical data were obtained for a human subject under
correspondence could be observed between flicker detec- the same stimulus conditions as those with which the animal
tion by human subjects in psychophysical experiments experiments described in this paper were performed. The
and electrophysiological measurements of discharge pat- curves illustrate the extreme steepness of the fall-off in human
terns of single cells of the lateral geniculate nucleus. We flicker sensitivity for frequencies above 10 Hz.
found that the average discharge patterns of single cells
exhibited the following behavior when mean retinal With the particular aim of accounting for the extremely
illumination was changed: In the low-frequency region sharp attenuation of the de Lange curves in the high-frequency
(less than about 10 Hz) the response strength (impulses/ region, models have been proposed that are made up of as
sec) is independent of the mean luminance, in accordance many as ten cascaded temporal integration or RC-elements
with Weber's law; in the high-frequency region (above
about 10 Hz) the response depends on the absolute modu- (9-14). If these N-stage integrator models are applied to human
lation amplitude, in accordance with the Ferry-Porter flicker sensitivity, then the problem arises of deciding how these
law. Therefore the main features of human critical flicker elements are distributed along the visual pathway. In order
frequency data are already present in the cellular (lateral to answer this question, we studied in monkey the impulse
geniculate nucleus) response of the macaque monkey. discharge patterns evoked by sinusoidal luminance modulation
However, the steepness of the high frequency fall-off in
the response characteristics of these cells is much less than in single lateral geniculate cells. We chose macaques for our
the corresponding fall-off in the human critical-flicker- experiment since these monkeys are almost ideal animals for
frequency curves. investigations of the relations between psychophysical data
It was de Lange (1-3) who introduced the harmonic analysis
(in man) and single-unit activity. Behavioral experiments
have shown that not only their wavelength discrimination
approach that underlies present day understanding of much function, but also their photopic and scotopic spectral-
of the chaotic literature (4) about flicker perception. He (2) sensitivity functions, resemble those of man (15-17). More-
demonstrated that the high-frequency flicker threshold does over, recent electroretinographical (18) measurements in-
not depend on the waveform of the flickering light provided dicate an identity--except for a slight deviation for wave-
that the amplitude of the fundamental Fourier component lengths below 450 nm-between the photopic spectral sensi-
of this waveform was held constant. The fruitfulness of Fourier tivities of man and macaque.
analysis in the study of flicker perception under scotopic
conditions had already been indicated by Ives (5) in 1922, but
the full potential of harmonic analysis only became evident METHODS
after de Lange's extensive study at photopic levels of illu- 2-3 kg macaques (Macaca mulatta and Macaca speciosa) were
mination (6-8). maintained under light nembutal anaesthesia (10 mg/kg per
De Lange's contribution stems from his separation of the hr) throughout the experiments, which lasted 15-20 hr. The
effects of luminance change and mean luminance. He was monkeys were held in a stereotaxic instrument. The eyelids
able to separate these effects by using modulated light rather were retracted and the pupils dilated. The sclera of each eye
than the classical on/off square-wave stimulus. As an example was sutured to the stereotaxic instrument so as to prevent
of luminance modulation at a constant mean luminance, a spontaneous eye movements. The single-cell activity in the
sinusoidal modulating waveform is depicted in the insert of lateral geniculate nucleus (LGN) was recorded with 3 M
Fig. 1. The two important stimulus parameters are the average KCl-filled micropipettes with resistances of 5-20 Mg. The
luminance Lo, and the amplitude of the luminance modulation electrodes were lowered through a trephine hole in the skull.
A, which can be controlled independently of Lo. For a given In order to improve the signal-to-noise ratio, the shaped im-
average luminance, Lo, de Lange plotted the threshold modu- pulse responses were averaged by a computer of average
lation m (m = A/Lo) as a function of frequency on logarithmic transients (CAT). The number of summations triggered by
the periodic light stimulus was generally 50-600, depending
Abbreviation: LGN, lateral geniculate nucleus. upon the discharge rates. A final record of the impulse density
* Present address: Laboratory of Medical Physics, University of distribution was obtained by plotting the analogue output
Amsterdam, Herengracht 196, Amsterdam, The Netherlands. of the computer with an X-Y recorder.
2802
 Proc. Nat. Acad. Sci. USA 68 (1971) Monkey Responses to Flicker 2803

sp/sec

I- 20

75% . r>K. 0
ilo _time

E
0
0~~~~~~~~~~~~~~~ b 1:20

Q1 Q4 0 0 40 4
50%
440
1
o

100 .

C I- 20
* frequency [Hz]

FIG 1. Modulation depth for flicker fusion in man as a func- 25%

tion of frequency. White, sinusoidally modulated light is used 0
at three different average retinal illuminations: 30, 3, and 0.3
trolands. The Maxwellian field subtended 150 of visual angle.
Insert: The intensity of a sinusoidal light stimulus is depicted d
as a function of time. The unmodulated stimulus, of intensity -20
Lo, changes at a given moment into a sinusoidal modulated light 1-0
with amplitude A. The depth of modulation is conveniently ex- 10% , /
-
pressed by the parameter m = A/Lo. The stimulus frequency is
f(Hz). The mathematical representation of the waveform is FIG. 2. Average impulse responses of the excitatory compo-
X (t) = Lo(1 +msin27rft). nent of a green+-red- LGN cell are given as a function of the
modulation depth of a 8-Hz sinusoidally modulated light. For
relatively weak stimuli (c and d), the density distribution of the
The optical system consisted of three separate pathways impulses has roughly the same shape as the sinusoidal stimulus.
combined through a beam splitter. One channel contained in- For higher modulation depths (a and b), distortion becomes
terference filters, neutral-density filters, and an electrical evident. The number of summations is 100; the mean retinal
shutter (rise time, 2 msec). The second channel provided a illumination amounts to 480 trolands.
chromatic adaptation light with adjustable intensity. The
final beam, containing an aperture stop that could be varied end of the spectrum, and excitation to wavelengths at the
in size and position, was focused in the plane of the pupil of other end of the spectrum. These two cell types have been
either eye of the animal, so as to present the monkey with a 20
Maxwellian field that subtended up to 150 of visual angle.
This apparatus was used to classify the various LGN im-
lulse discharges on the basis of their color coding.
In order to facilitate comparisons of our results with pre-
vious work, and in particular with the extensive pioneering 15
studies of Russell L. DeValois, we followed the DeValois
method* (19) of data reduction.
Once a unit had been isolated and classified by DeValois'
procedure, the first beam was replaced by a third beam, which 10
allowed the unit to be stimulated with modulated waveforms -i
c
at a constant average luminance. The light source that 0
E
generated this third beam was a television projection tube 0
c
(Philips MW 6/2); its modulated output was monitored and 0 5
controlled by a photocell (20).
RESULTS E
On the basis of their responses to different stimulus wave-
0
lengths the LGN cells can be divided into two classes. These 0 10 25 50 75
are: (a) cells that show either excitation or inhibition of the modulation depth [Y] - .
spontaneous activity for all stimulus wavelengths, and (b) FIG. 3. This figure gives the amplitude of the fundamental
cells that show an inhibition to stimulus wavelengths at one component of the averaged discharge pattern of the excitatory
component of a green+-red-, LGN cell to sine-wave modulated
*
This method consisted of recording the responses to light flashes light of various frequencies. Only for extremely-low modulation
of one second in duration. The number of impulses in the one depths is the amplitude of the average response proportional to
second periods directly preceding, during and after the cessation the strength of the input sine wave. The mean retinal illumina-
of the light flash were counted. tion is about 480 trolands.
2804 Physiology: Spekreijse et al. Proc. Nat. Acad. Sci. USA 68 (1971)

0.1
0

._O.
CL
0
0
4
-o0
E
31
0
M, 4'A
2U, I0
C-

0
U

frequency CHz)

30

4-
c0
L-

IV
*0
E
0
0 a
0
o2ir .O-
0 0
0)
U,
0
0.
I
o 480 trolands
41+- a 37
o 4 H

0.4 1 4 10
0 frequency (Hz]
FIG. 4. Amplitude (above) and phase (below) characteristics FIG. 5. Absolute modulation amplitude, mLo, at flicker fusion
of a yellow--blue+ LGN cell determined with blue, sinusoidally- in a human psycho-physical experiment and at constant response
modulated light at three different retinal illuminations (480, of a monkey LGN cell (top), as a function of frequency. The
37, 4 troland). The amplitude characteristics (de Lange curves) top-half data are replotted from Fig. 4, the bottom-half data
are obtained by plotting the modulation depth required are obtained from Fig. 1. It should be noted that in contrast to
for each stimulus frequency to reach a constant response of 8.6 the flicker-fusion data, the values at the vertical axis of the
impulses/sec. monkey data largely depend on the arbitrarily chosen constant-
response criterion. The mean retinal illuminations of the sinu-
soidal stimuli are given in the figures.
called by DeValois (19) non-opponent and opponent cells.
The action spectrum of opponent cells has only one neutral We have derived the de Lange curves from plots of the
point; i.e., only one wavelength can be found that does not fundamental component in the average responses versus
produce any change in firing rate. Depending on the loca- modulation depth. Fig. 3 gives such plots for a range of fre-
tion of this neutral point, the opponent cells can be classified quencies of the constant-luminance sine-wave stimulus. The
(21) as red-green and yellow-blue cells. tangents (straight lines in Fig. 3) were drawn through the
In our experiments, we first determined the color coding of origin of these curves, and the intersections of these tangents
a particular LGN cell. Sinusoidally modulated light with a with some convenient constant threshold criterion (13 spikes/
color coinciding with one of the peaks in the action spectrum see in Fig. 3) were determined. Next, the modulation depths
was then used as a tool to investigate the impulse discharges required to reach this criterion were plotted as a function of
as a function of frequency and modulation depth. Examples stimulus frequencyt. We used de Lange's convention of
of average responses to an 8-Hz sinusoidally modulated light plotting the ordinate with increasing modulation downwards,
are illustrated as a function of modulation depth in Fig. 2. so as to emphasize similarities between these curves and the
This figure shows that for relatively weak stimuli (Fig. 2 attenuation chracteristics of electronic systems.
c and d), the density distribution of the nerve impulses has The top of Fig. 4 shows a set of de Lange curves deter-
roughly the same shape as the sinusoidal stimulus. However,
at higher modulation depths the average responses exhibit t It should be noted that the choice of the threshold criterion is of
such nonlinear behavior as saturation on the one hand and no influence upon the shape of the de Lange characteristics.
clipping (zero spike frequency) on the other hand (Fig. 2 a The threshold criterion sets the amplitude characteristic along
and b). the vertical axis; the tangent lines determine the shape.
 Proc. Nat. A cad. Sci. USA 68 (1971) Monkey Responses to Flicker 2805

mined for a yellow--blue+ cell, stimulated with a blue light along this envelope, the high-frequency threshold depends on
at three different retinal illuminations (480 trol, 37 trol, and the absolute and not on the relative modulation amplitude,
4 troll). Two conclusions emerge directly from this figure: in accordance with the well-known Ferry-Porter law. The
(a) At the low-frequency end, the flicker threshold is indepen- major difference between the two sets of data is the steepness
dent of the mean luminance and solely determined by the of the high-frequency fall-off. Whereas this fall-off in psycho-
modulation depth. This shows the applicability of the Weber physical data may reach values as high as 60 dB/octave (1),
law in this frequency range. (b) For frequencies above about we never observed such high values in our LGN data; in the
10 Hz, the LGN cells become increasingly more sensitive as case shown, the attenuation amounts only to 12 dB/ octave.
luminance is raised. Since this is the only relevant difference between the two sets
The bottom half of Fig. 4 gives the phase characteristics of data, our experimental findings indicate directly that the
that belong to the de Lange curves just described. These main features of the de Lange characteristics are already
characteristics demonstrate an increased phase lag between present in the lateral geniculate response. Therefore, assuming
the stimulating sinusoid and the fundamental component in that the behavioral de Lange characteristics in monkey prove
the average response as stimulus frequency increases. This to be comparable to those of man, it may be that further stages
phase lag seems compatible with the shapes of the electro- do no more than add to the high-frequency fall-off.
physiological de Lange curves.
Ten other cells, both opponent and non-opponent, recorded We are grateful to P. Padmos, M.Sc. for allowing us to use his
by us in this investigation, show in general the same features equipment, and for giving us advice necessary to carry out the
as depicted for the particular cell in Fig. 4. However, the experiments. We are greatly indebted to Mrs. N. H. Blokland
for her technical assistance.
behavior at the low-frequency end of the de Lange charac-
teristic is not always as independent of the mean luminance 1. de Lange, H., Ph.D. thesis, Technological University, Delft
as for the cell presented. But even for cells dissimilar to that (1954).
shown in Fig. 4, two regions can be distinguished: a low- 2. de Lange, H., J. Opt. Soc. Amer., 44, 380 (1954).
frequency one in which the threshold is mainly determined 3. de Lange, H., J. Opt. Soc. Amer., 48, 777, and 784 (1958).
by the modulation depth, and a high-frequency one in which 4. Landis, C., An Annotated Bibliography of Flicker Fusion
the sensitivity increases with luminance. Phenomena (Armed Forces-National Research Council,
Univ. Michigan, 1953).
DISCUSSION 5. Ives, H. E., J. Opt. Soc. Amer. Rev. Sci. Instrum., 6, 254
(1922).
Kelly (22) and also Levinson and Harmon (7) have proposed 6. Levinson, J., Science, 130, 919 (1959).
that an envelope can be obtained for the de Lange curves 7. Levinson, J., and L. D. Harmon, Kybernetik, 1, 107 (1961).
by plotting the absolute threshold amplitude, mL0, rather than 8. Kelly, D. H., Doc. Ophthalmol., 18, 16 (1964).
the critical modulation depth, m, as a function of frequency. 9. De Voe, R. D., Doc. Ophthalmol., 18, 128 (1964).
Fig. 5 shows two sets of modulation sensitivity curves, re- 10. Campbell, F. W., and J. G. Robson, Doc. Ophthalmol., 18,83
plotted in luminance amplutude only, without regard to the (1964).
adaptation level. In the top half of this figure, the data of 11. Fuortes, M. G. F., and A. L. Hodgkin, J. Physiol. (London),
172, 239 (1964).
Fig. 4 are replotted; the bottom half gives the replotted data 12. Sperling, G., Doc. Ophthalmol., 18, 3 (1964).
of Fig. 1. The data of Fig. 1 were obtained for a human sub- 13. Marimont, R. B., J. Physiol. (London), 179, 489 (1965).
ject under the same stimulus conditions as those of the animal 14. Pinter, R. B., J. Gen. Physiol., 49, 565 (1966).
experiment. The similarity between the two sets of data is 15. DeValois, R. L., and J. H. Jacobs, Science, 162, 533 (1968).
evident. In both conditions the high-frequency data join- 16. Schrier, A. M., and D. S. Blough, J. Comp. Physiol. Psychol.,
within limits§-together in a common envelope. Anywhere 62, 457 (1966).
17. Sidley, M. A., and G. H. Sperling, J. Opt. Soc. Amer., 57, 816
(1967).
I It should be noted that the same troland value, because of the 18. van Norren, D., Vision Res., in press.
smaller nodal distance of the monkey eye, represents a higher 19. DeValois, R. L., I. Abramov, and J. H. Jacobs, J. Opt. Soc.
retinal illumination in the monkey eye than in the human eye. Amer., 56, 966 (1966).
§ As Kelly (8) has pointed out, a common envelope holds only 20. Denier van der Gon, J. J., J. Strackee, and L. H. van der
within a restricted-but still large-range of retinal illumina- Tweel, Phy8. Med. Biol., 3-2, 164 (1958).
tions. For retinal illuminations less than about 0.1 troland or 21. DeValois, R. L., J. H. Jacobs, and I. Abramov, Science, 146,
higher than about 9000 trolands, the replotted de Lange curves 1184 (1964).
deviate from a common envelope. 22. Kelly, D. H., J. Opt. Soc. Amer., 49, 730 (1961).