The synapses that horizontal 
cells make with receptors are 
likewise unusual, lacking the 
electron-microscopic features 
that would normally tell us 
which way the information is 
conveyed. It is clear that 
receptors feed information to 
horizontal cells through 
excitatory synapses because 
horizontal cells, like receptors, 
are in most cases 
hyperpolarized, or turned off, 
by light. It is less clear where 
the horizontal cell sends its 
output: in some species such as 
turtles we know that they feed 
information back to receptors; 
in other species they make 
synapses with the dendrites of 
bipolar cells and doubtless feed 
into them; in primates we do 
not have either type of 
information. In summary, 
horizontal cells get their input 
from receptors; their output is 
still uncertain, but is either 
back to receptors, or to bipolar 
cells, or to both.

    The relatively wide retinal 
area over which receptors 
supply horizontal cells suggests 
that the receptive fields of 
horizontal cells should be 
large, and they are. They are 
about the size of the entire 
receptive fields of bipolar cells 
or ganglion cells, center plus 
surround. They are uniform, 
giving hyperpolarization 
wherever you stimulate, and 
more hyperpolarization the 
larger the spot. Much evidence 
points to the horizontal cells as 
being responsible for the 
receptive-field surrounds of the 
bipolar cells--indeed they are 
the only plausible candidates, 
being the only cells that 
connect to receptors over so 
wide an expanse. When 
horizontal cells connect 
directly to bipolars, the 
synapses to on-bipolars would 
have to be excitatory (for the 
effect of light in the surround 
to be inhibitory) and those to 
off-bipolars, inhibitory. If the 
influence is by way of the 
receptors, the synapses would 
have to be inhibitory.

    To sum this up: Bipolar cells 
have center-surround receptive 
fields. The center is supplied by 
direct input from a small group 
of receptors; the surround 
arises from an indirect path 
stemming from a wider expanse 
of receptors that feed into 
horizontal cells, which 
probably feed into the bipolars. 
The indirect path could also be 
the result of the horizontal 
cells feeding back and 
inhibiting the receptors.