322 R. M . SHAPLEY AND C.
ENROTH-CUGELL
the right as would follow from gain re-adjustment difference would not be surprising because the
by illumination of the surround (see Section 1.2.2.). mudpuppy and cat have evolved quite differently
Control experiments established that this was not with widely different visual capacities. The details
a result of light scatter onto the center (Werblin of exactly how the retinal network is connected
1974). This is evidence that, in the mudpuppy spatially to regulate gain might well differ between
retina, the gain of the bipolar cell is set by signals two such distantly related animals.
coming from its surround, which in this species is Ashmore and Falk (1980) have demonstrated that
believed to be mediated by horizontal cells (Werblin the gain of bipolar cells, in the almost all-rod retina
and Dowling, 1969; Thibos and Werblin, 1978a). of the dogfish, begins to drop at extremely low
Such results suggest that, in the mudpuppy, backgrounds because of saturation in the bipolar
horizontal cells act as a gain control on bipolar cells. cell itself. In the dogfish retina, there is a very high
Two points of comparison with previously amplification of rod signals at the rod-bipolar
presented psychophysical and physiological results synapse, and as a result the bipolar cells approach
are needed here, to prevent the (probably erroneous) their response ceilings at very low backgrounds.
inference that this conclusion is generally applicable There is no evidence for a gain control, and
to all vertebrates. t h e r e f o r e true light a d a p t a t i o n , in these
The first comparison of Fig. 47 is with the experiments. However, the results of Werblin (1974)
sensitization p h e n o m e n o n in human vision and Naka et al. (1979) illustrate how an automatic
discovered by Westheimer (1965). The results in gain control, acting on signals from photoreceptors
Fig. 47 are the opposite o f W e s t h e i m e r to bipolar cells, staves off saturation in mudpuppy
sensitization. Illumination in the periphery of the and catfish bipolars. In Fig. 46 for example, the
mudpuppy bipolar's receptive field desensitizes the catfish bipolar's gain begins to drop at a lower mean
center in Werblin's (1974) and Thibos' and level than the horizontal cell's gain, presumably due
Werblin's (1978a) experiments. There is, however,
a puzzling and unresolved contradiction with -17-
Burkhardt's (1974) report of sensitization in E
mudpuppy bipolar cells. Perhaps it has to do with -6
"- --i8-
different receptor input to the bipolar cells in the 4-'
two sets of experiments. Thibos and Werblin o
(1978a) were working at low backgrounds at which -19-
rods were the predominant input while Burkhardt
E
was working with a highly light adapted mudpuppy
retina in which it is probable that cones were the --20 ,,~ T I i" I I ~ I I
-7 --6 5 4
predominant photoreceptor inputs to the proximal
retinal neurons. If this is the explanation for the
FIG. 47. The effect of a steady illumination in the periphery
opposite results, it would be an interesting and of the receptive field on the stimulus - response function of
unusual example of a reversal in functional mudpuppy bipolar cells; intracellular recording. The stimulus
characteristics because of the transition from rod was a flashing spot centered on the receptive field of the
bipolar cell. The spot's diameter, 0.4 m m , was chosen to
to cone pathways. stimulate the center optimally. Stimulus duration was 1 s.
A second comparison worth pursuing is that The response measure was the steady state plateau of the 1 s
between the strong effect on the gain of the center response. Filled circles are for no illumination in the periphery
o f the receptive field; empty circles are response magnitudes
exerted by the surround of mudpuppy bipolar cells when a - 4 log unit steadily illuminated annulus was placed
(Fig. 47) vis-a-vis negligible or, at most, weak effect in the periphery of the field. A modified N a k a - R u s h t o n
of the surround on the gain of the center in X and function was fitted to the data in the condition of no
peripheral illumination, and then translated laterally to fit
Y cat retinal ganglion cells (Figs 3 8 - 4 0 ) . This the results obtained with the annulus. The light source was
difference between results on the spatial extent of a tungsten lamp, attenuated by neutral filters. The
"adaptation pools" suggests that quite different unattenuated retinal illumination was calculated to be about
10'3 quanta(522 nm) (cm ~ s)-t. Illuminations are given as log
mechanisms are involved in the control of gain in attenuation relative to this value. From Thibos and Werblin
the mudpuppy and cat retinas. In a sense such a (1978a).
