Computer simulations were carried out to evaluate the influence of var
ying the membrane resistance (R(m)) on the dendritic integration capac
ity of three classes of ganglion cells in the mudpuppy (Necturus macul
osus) retina. Three broadly different morphological classes of ganglio
n cells were selected for this study and represent the range of dendri
tic tree sizes found in the ganglion cell population of this species.
Simulations were conducted on anatomical data obtained from cells stai
ned with horseradish peroxidase; each cell was traced, using a compute
r as an entry device and later converted to a compartmental (electrica
l) representation of the cell. Computer-simulation analysis used a tim
e-variant conductance change which was similar in waveform to light-ac
tivated bipolar cell input. The simulated membrane resistance for each
cell varied between 5000 and 100,000 Omega cm(2), and conductance cha
nges were introduced into different regions of the soma-dendritic tree
to evaluate dendritic integration efficiency. When higher values of R
(m) are used, even the largest cells become electrotonically compact a
nd attenuation of voltage responses is minimized from distal to soma r
egions. Responses were less attenuated from proximal to distal regions
of the cell because of the favorable impedance matching, and because
less current is required to polarize small ''sealed'' dendritic termin
ations. Steady-state responses integrate more effectively than transie
nt responses, particularly when R(m) is high, since transient response
s were more attenuated by the membrane capacitance. The possibility th
at R(m) is a dynamic property of retinal ganglion cells is discussed i
n view of the functional organization of dendritic integration efficie
ncy as R(m) fluctuates from low to high values.