294 R. M. SHAPLEYAND C.

ENROTH-CUGELL

i ooo" ' caused by quantal fluctuations. One must suppose

| I ' \ ]

that this high frequency noise is immune to the

adaptational effects of the mean level o f
illumination. Perhaps the noise which limits
detection of high frequency responses is post-
D-R
retinal. In any case, a complete and adequate
explanation for Kelly's "linear range" requires
future research.

2OO
~oo ['-
<3
500 O.S c:pd
2 [ ] r I I i~
IC)~ a;" ~,~ I (d)
10 o
20
5or
100 ;
200
8 ¢pd D-R 12 cpd
~ T l , ] [
1 2 5 10 20 50
TEMPORAL FREQUENCY (Hz)
FIG. 20. Threshold illumination as a function of temporal
frequency at different spatial frequencies. The threshold
illumination AB was the amplitude of sine grating which
could just be seen on a mean illumination of B td. In each
of the four panels of the figure, the four different curves are
f r o m m e a s u r e m e n t s at the following m e a n retinal
illuminations: 36 td (arrowheads), 114 td (triangles), 360 td
(diamonds), and 1140 td (circles). The test target was a pale-
green CRT which subtended 7 deg; it was viewed monocularly
and fixated. The gratings were modulated in time with a
sinusoidal waveform; the temporal frequency o f the
modulation is plotted on the horizontal axis. Measurements
from four spatial frequencies are shown: 0.5 c deg -1,
2 c deg -1, 8 c deg -~, and 12 c deg-L Thresholds, which are
separated from those at other mean illuminations by a factor
which is equal to the ratio of the mean levels, conform to
Weber's Law, and have a W written next to the curve.
Thresholds which rise like the square root o f the mean level
are labeled D - R for the de Vries-Rose law, synonymous with
the square root law. From Kelly (1972).
be quantal noise and must be independent of mean
light level. This is not implausible. Most sources of
noise in the visual system, e.g. channel opening and
closing in neuronal membranes, or spontaneous
transmitter release, should have wide-band
components. These components may be relatively
larger at high temporal frequencies than at low,
compared to the light evoked neural shot noise
2 cpd lg70 o
3. GAIN AND CONTRAST GAIN IN
R E T I N A L G A N G L I O N CELLS
RECTILINEAR PATTERNS
Subj. TZ
At the outset of this section on the physiology
of retinal adaptation we concentrate on retinal
ganglion cells, the output stage of the retina. All
o
information which flows from the retina to the
__ ~>
brain about the visual appearance of the outside
° °° °° °o o world passes along the axons of these ganglion cells.
The evidence of retinal adaptation in the activity
of these neurons allows us to establish a link
O-R
i J I I r
between the visual, perceptual function of light
1 2 5 10 20 50
adaptation and the underlying retinal mechanisms.
We will further concentrate our attention on two
kinds of retinal ganglion cells in the cat's retina, the
X and Y cells (see Appendix 2), because most is
known about them. Comparison with the retinas
of other species and with human vision will be made
frequently. As in the Introduction, we stress the
importance of a hierarchy of gain control
mechanisms at different sites in the retina.
Furthermore, the role of retinal gain controls in
making the retina respond to contrast will be made
evident.
One can speak about the g a i n of retinal ganglion
cells because their impulse rate variation caused by
increments (or decrements) of illumination are
proportional to the magnitude of the increment (or
decrement) over a considerable range of response
amplitude. This is illustrated by Fig. 22 (Shapley
and Kaplan, unpublished). The stimuli were fine
gratings which stimulated the center of the receptive
field. (In this initial discussion we will be dealing
with the gain of the center only, but will consider
the gain of the surround below.)
The ratio of the change in impulse rate with
change in stimulus magnitude is the gain, G
G = dR/dI. (19a)