VISUAL ADAPTATION AND RETINAL GAIN CONTROLS 283

10.0
2 C/DEG
used by the retina, in an area o f 1 deg 2, over a time
of 0.1 s, then the total available quanta from such
8.0 a background are 1250. If quantal fluctuations are
the limiting factor, then the standard deviation of
>..
t-,- 6,0 the quantal count with this mean number of 1250
must be the limiting source of noise. The standard
4.0 o
o
deviation of the quantal noise must be the square
o o

• 0.5 ClDEG
root of 1250, which is about 35. In most detection
1/1 2.0 experiments the s i g n a l - noise ratio is taken to be
about 4, so that the threshold amount of quanta
z~ 0.0 would have to be about 4-35 = 140. This is 140
t.)
I 1 L I l I I I neurally triggered quantal events; we must multiply
-2.50 -2.00 -I.50 -I.00 -0.50 0.00 0.50
M E A N ILLUMINATION (LOG TROLANDS)
by 40 (the reciprocal of 1/40 which is the product
of the quantal efficiency of 1/4 times the integration
FIG. 11. The dependence of the Rayleigh contrast sensitivity time of 0.1 s) to obtain the stimulus illumination,
on mean level. Results for two spatial frequencies of test
target are shown: 0.5 c deg-' as empty circles, and 2 c deg-1 measured at the cornea, to produce this many
as filled circles. An artificial pupil of 1 mm diameter was neural events. Thus we get that the stimulus
used. The test targets were sinusoidal grating patterns illumination must be equivalent to 5.6.103
generated on the face of a cathode ray tube (CRT) display
which subtended 8 deg by 3 deg, centered 12 deg from the quanta(deg 2 s)-'. The best quantum-limited (Weber)
fovea. The phosphor of the CRT was P31(yellow-green) contrast sensitivity one could expect would
but a Wratten 65 filter was interposed between subject and therefore be (Ia/MT) = (5" 104)/(5.6 • 103) or about
screen to shift the dominant wavelength of the phosphor
towards the peak of the rod spectral sensitivity function. The 9, which is about as high as the human (Weber)
target was presented for 200 ms every second. Control contrast sensitivity reaches in the scotopic range.
experiments established that the rods were the photoreceptors It is interesting that Aguilar and Stiles (1954) went
which determined threshold up to 0 log td. Plotted from
Daitch and Green (1969). through an analogous calculation in the Discussion
of their famous paper, and concluded that they had
discredited the idea that quantal fluctuations limited
part in limiting contrast sensitivity in the scotopic
detection of contrast. However, in their calculation
range. Suppose we make the conservative
they made the assumption that, because they used
assumption that no less than one quarter of the
a disk 9 deg in diameter as a stimulus, all the quanta
quanta incident on the cornea may be used for
in the 64 deg 2 area of the disk were available for
contrast detection. Further we assume that the
the detection task. We now know this assumption
integration time over which quanta may be counted
is unreasonable. Presently available knowledge
is about 0.1 s. Finally, and most crucially, we
about the distribution of receptive field sizes
assume that the integration area for the neurons
supports Rose's (1948) hypothesis about the
which are involved in the detection task is at most
importance of quantum fluctuations in scotopic
1 deg 2. This last assumption is based on the
vision.
measurements of receptive field center sizes of
retinal ganglion cells in cats and monkeys at The great improvement in contrast sensitivity
comparable retinal loci to the ones used in the associated with the shift from rod to cone pathways
human psychophysical experiments (Hubel and is complicated by the effects of retinal inhomo-
Wiesel, 1960; de Monasterio, 1978; Cleland et al., geneity studied by Koenderink et al. (1978). They
1979), and on the notions of the Channel measured contrast sensitivity functions with sine
Hypothesis expanded in Section 2.1.3. Given these gratings which subtended a fixed area of 4 deg by
assumptions, one may calculate what the highest 4 deg, at several locations in the visual field.
contrast sensitivity might be in the scotopic range, Figure 12 shows their finding that in the far
at the highest background in this range which is periphery there is no increase of contrast sensitivity
about 10-' td. This background retinal illumination between scotopic and photopic background levels,
corresponds to 5" 104 quanta(deg ~ s) -' incident on and in the near periphery there is only an increase
the cornea. If one quarter of these quanta may be of about a factor of two in the photopic contrast