What mechanisms cause a cortical neuron to fire an action potential? We have shown previously that it is unlikely that cortical cells in vivo spike as a result of the temporal summation of small and random excitatory synaptic inputs. Their high firing irregularity could result from very strong and fast depolarizing events (such as dendritic spikes) or from extremely strong repolarizing currents during """"""""integration"""""""" (such as voltage-gated potassium currents or sub-ms membrane decay time constants) or from massive inhibition. We have shown in preliminary studies that pyramidal cells in slice fire very regularly, and that the spike-generating mechanism itself is not the source of the high variability. We wish to study how synaptic inputs under physiological conditions trigger spikes and how this process affects both the nonlinearity and the time-scale (sub-ms vs. tens of ms) of single neuron information computation. We propose different numerical/analytical studies to be performed on single-cell recordings supplied to us from the laboratory of R. Douglas and K. Martin, and compartmental-modeling simulations to reproduce simultaneously all of the characteristic extra- and intra- cellular responses of the recorded cells. One study will examine the fluctuations in the subthreshold intracellular voltage, as measured by Fourier power-spectrum analysis. Another study will quantify aspects of the cell's firing mechanism-such as after hyperpolarization and firing """"""""threshold""""""""-that may provide insight into the locations, time-scales, and strengths of the conductances. Other studies will examine the contribution of massive inhibition and slower conductances, and use algorithmic complexity and related measures to characterize the cell's firing behavior. We will also examine how synaptic input in the presence of unreliable synapses can cause cortical cells in the behaving monkey to modulate their firing rate reliably at the 2-10 ms level. All of these analyses will be compared to simulations of pyramidal and stellate cells (reconstructed from cat striate cortex) with passive and active dendritic conductance and in the presence of correlated synaptic input.
The aim i s to find models which can simultaneously reproduce all the cells' responses (including their variability and their ability to modulate precisely in time), and which can be reduced to conceptually simpler and less computationally demanding forms. We hope that this unique combination of intracellular recordings and theoretical expertise can Illuminate the time-scale and complexity of the process by which single cortical cells process their input.
