Neural mechanisms of spatiotemporal signal processing

We have studied the synaptic, dendritic, and network mechanisms of spatiotemporal signal processing underlying the computation of visual motion in the avian tectum. Such mechanisms are critical for information processing in all vertebrates, but have been difficult to elucidate in mammals because of anatomical limitations. We have therefore developed a chick tectal slice preparation, which has features that help us circumvent these limitations. Using single-electrode multi-pulse synaptic stimulation experiments we found that the SGC-I cell responds to synaptic stimulation in a binary manner and its response is phasic in a time dependent probabilistic manner over large time scales. Synaptic inputs at two locations typically interact in a mutually exclusive manner when delivered within the "interaction time" of approximately 30 ms. Then we constructed a model of SGC-I cell and the retinal inputs to examine the role of the observed non-linear cellular properties in shaping the response of SGC-I neurons to assumed retinal representations of dynamic spatiotemporal visual stimuli. We found that by these properties, SGC-I cells can classify different stimuli. Especially without the phasic synaptic signal transfer the model SGC-I cell fails to distinguish between the static stationary stimuli and dynamic spatiotemporal stimuli. Based on one-site synaptic response probability and the assumption of independent neighboring dendritic endings we predicted the response probability of SGC-I cells to multiple synaptic inputs. We tested this independence-based model prediction and found that the independency assumption is not valid. The measured SGC-I response probability to multiple synaptic inputs does not increase with the number of synaptic inputs. The presence of GABAergic horizontal cells in layer 5 suggest an inhibitory effect of these cells on the SGC-I retino-tectal synaptic responses. In our experiment we found that the measured SGC-I response probability to multiple synaptic inputs is reduced when inhibitory tectal circuits are blocked. By predicting the SGC-I response to multiple synaptic inputs based on blocked inhibitory circuitry we found that the response probability is closer to independent situation but not exactly. So there is more than just inhibitory mechanism involved.;To characterize the dependency between two neighboring synapses we used 2-site stimulation experiments and measured the effect of one stimulation on a spatially separate synapse. To determine whether this inhibitory mechanism is pre-synaptic or post-synaptic we used chloride channel blocker intracellularly. We saw an increase in response probability when post-synaptic chloride channels are blocked.;Finally we found a good agreement between our prediction and experimental results for Poisson spike trains which may be considered more natural stimuli. Only the early stage of SGC-I response is carrying most of the information.;Analyzing the SGC-I spike timing and the accuracy of latency is the last part of the thesis.