Most of the discussion about neural coding focuses on the question whether neurons use a rate code or a time code. But it is rarely considered that neurons could carry information not only with action potentials but also with graded sub-threshold fluctuations of their membrane potentials. In fact, a variety of cells is known not to produce action potentials at all. E.g. cells in the retina and in the olfactory bulb of vertebrates and many invertebrate neurons transmit information via synapses with graded transmitter release that is regulated up or down depending on the membrane potential of the presynaptic cell (review: Roberts & Bush, "Neurones without Impulses", Cambridge University Press, 1981).

From the perspective of information transmission a graded signal can always transmit more information in shorter times than a binary signal like an all or none spike train. Hence, it can be suspected that many cell types should use graded information transmission for local interactions in addition to spikes which are better suitable to transmit information over long distances. Finding of dendro-dendritic synapses could be a hint that not spike-mediated synaptic transmission is a general mechanism for neuronal information processing. The goal of my studies is to compare spike-mediated and graded neuronal information processing.

I investigate this question in two systems and with two different approaches:

• In a computational project based on experimental data from the fly visual system synaptic transmission is analyzed with a conductance based computer model simulation. This project is done in Prof. Sejnowski's group at the Salk Institute in colaboration with Prof. Egelhaaf's group at Bielefeld University in Germany.

• In an experimental project I perform intracellular double recordings from neurons in the leech nervous system to compare graded and spike-mediated information transmission. The experiments are done in Prof. Kristan's lab at UCSD.

Graded and Spike-triggered Synapses in the Visual System of the Fly

In the visual system of the fly a population of large field motion sensitive neurons consists of three types of neurons: spiking, graded and ``mixed''. Even though all of these neurons are well characterized, the properties of their input cells remain unknown. We developed a computational model for synaptic transmission that can simulate the continuum between spiking, mixed and non-spiking transmission. The kind of transmission is charactirized by only two parameters of a sigmoidal transfer function, the slope and the midpoint. Applying this model to experimental data from the fly, we tried to predict which kind of synaptic transmission the input elements of the fly's motion-sensitive neurons are using. The transfer function which reproduced experimental data best has a steep slope like it is typical for spike-triggered synapses but it is centered around the presynaptic resting potential like in the case of graded transmission.

Graded and Spike-triggered Synaptic Transmission in the Leech Nervous System

The leech nervous system is a very good system to compare graded and spike-triggered information transmission, because the same neuron can provide continuous and spike-triggered synaptic output. For example the interneuron 115, which is located on the dorsal side of each ganglion and is involved in the generation of swimming, local bending and shortening, contacts motorneuron in the same ganglion as well as in at least two posterior ganglia. The connections within the same ganglion do not depend on spikes, while for the long-range information transfer to the next ganglion spikes are essential.

To compare the synaptic transmission from interneuron 115 to the motorneuron 4 within one ganglion and between ganglia I perform simultaneous intracellular recordings. Preliminary data suggests that the synaptic strength changes activity-dependent. I plan to further analyze this activity-dependence and the reliability of the transmission in both modes with model simulations based on the experimental data.

Cell fills of Motorneuron 4 and Interneuron 115 in one ganglion.