Comparison of mechanisms for contrast-invariance of orientation selectivity in simple cells.

The simple cells of the visual cortex respond over a narrow range of stimulus orientations, and this tuning is invariant to the contrast at which the stimulus is presented. The inputs to a single cell derive from a population of thalamic cells that provide a bell-shaped tuning width and offset that increases with stimulus contrast. Synaptic depression, noise and inhibition have been proposed as feedforward mechanisms to explain why these increases do not appear in simple cells.

Influence of asymmetric attenuation of single and paired dendritic inputs on summation of synaptic potentials and initiation of action potentials.

Previous studies revealed mechanisms of dendritic inputs leading to action potential initiation at the axon initial segment and backpropagation into the dendritic tree. This interest has recently expanded toward the communication between different parts of the dendritic tree which could preprocess information before reaching the soma. This study tested for effects of asymmetric voltage attenuation between different sites in the dendritic tree on summation of synaptic inputs and action potential initiation using the NEURON simulation environment.

Effects of electrical coupling among layer 4 inhibitory interneurons on contrast-invariant orientation tuning.

Simulations of orientation selectivity in visual cortex have shown that layer 4 complex cells lacking orientation tuning are ideal for providing global inhibition that scales with contrast in order to produce simple cells with contrast-invariant orientation tuning (Lauritzen and Miller in J Neurosci 23:10201-10213, 2003). Inhibitory cortical cells have been shown to be electrically coupled by gap junctions (Fukuda and Kosaka in J Neurosci 120:5-20, 2003). Such coupling promotes, among other effects, spike synchronization and coordination of postsynaptic IPSPs (Beierlein et al.

Detecting and estimating rectification of gap junction conductance based on simulations of dual-cell recordings from a pair and a network of coupled cells.

Gap junctions can exhibit rectification of conductance. Some reports use inequality of coupling coefficients as the first sign of the possible existence of rectification (Devor and Yarom, 2002; Fan et al., 2005; Levavi-Sivan et al.

Estimating conductances of dual-recorded neurons within a network of coupled cells.

Simultaneous pre- and postsynaptic cell recordings are used to calculate gap junction conductance based on an equivalent electrical circuit of an electrically coupled pair of cells. This calculation is imprecise when recording from a cell pair that is coupled to neighboring cells providing indirect conductance paths between the recorded cells. Despite this imprecision, junctional conductance has been calculated for coupled cell networks during the past 40 years since a more accurate method was lacking.

Supervised learning in a recurrent network of rate-model neurons exhibiting frequency adaptation.

For gradient descent learning to yield connectivity consistent with real biological networks, the simulated neurons would have to include more realistic intrinsic properties such as frequency adaptation. However, gradient descent learning cannot be used straightforwardly with adapting rate-model neurons because the derivative of the activation function depends on the activation history. The objectives of this study were to (1) develop a simple computational approach to reproduce mathematical gradient descent and (2) use this computational approach to provide supervised learning in a network formed of rate-model neurons that exhibit frequency adaptation.