The Effect of Neural Noise on Spike Time Precision in a Detailed CA3 Neuron Model

Figure 3

The effect of synaptic noise of various synaptic types. (a) Visualization of spike train groups analyzed under synaptic noise. The ordinate represents the AP rank of APs fired at times shown on abscissa. The braided lines in the main graph represent different spike trains of individual trials and demonstrate an interesting preference of the spiking mechanism preserving AP times well aligned despite different numbers of fired APs in trials. The inset focuses on the region of an AP cluster, demonstrating the cluster method used to analyze APs: the dotted blue rectangle in the inset represents the 10 ms time window moving along aligned spike trains and collecting only those AP clusters having within the 10 ms window number of spikes larger than 70% the value of 𝑁_Tr. (b) Resulting ⟨𝑆_AP⟩ estimated using 3 s stimulation duration with different types of synapses (mean ISI = 70 ms). For NMDA channel the 𝐶_𝑉=0.65, for other channels the 𝐶_𝑉=0.42. At the top of the bar the ± STD of 𝑆_AP is shown by black bar size. (c) ⟨𝑆_AP⟩ for different synaptic 𝐶_𝑉 and synapse type in question was calculated for values of 𝑇_AP=10, locking the ISI at 10 ms. Higher values of ISI would yield larger 𝑆_AP as is the case in panel (b) in which the mean ISI = 70 ms. (d) Correlations curves 𝑟(𝑇_AP)(𝑉_AP,𝑆_AP) for neurons exhibiting channel noise only and postsynaptic NMDA noise only. Each correlation curve is delimited by 0.95 confidence interval. (e) The mean ⟨𝑉_AP⟩ curve obtained by averaging ⟨𝑉_AP⟩ over all AP clusters, delimited by the ⟨𝑉_AP⟩ ± standard deviation curves STD illustrating the range of ⟨𝑉_AP⟩ fluctuations over AP clusters. (f) Correlation curves 𝑟(𝑇_AP)(𝑉_AP,𝑆_AP) calculated by taking the time derivative of ⟨𝑉_AP⟩ are drawn here for channel noise only and postsynaptic NMDA noise only separately. Each correlation curve is delimited by 0.95 confidence interval. (g) Effect of synaptic weights. The effect of different synaptic types and their location on spiking variability in deterministic model (simulated without any membrane noise) is demonstrated by top-down graphs to the right, containing horizontal bars of various widths and colors. The color is mapped by the color bar to show the number of spike trains 𝑁_ST held in each spike train group, the widths in milliseconds represent 𝑆_AP of AP clusters in the group. (Compared to the channel noise effect analysis, the larger threshold Th_STG = 0.3 separating trains into different group was used here. The number 𝑁 of all independently acting presynaptic neurons was 1100 with 110 neurons driving each individual synaptic subgroup—see Supplementary Table 1). In panels (b), (c), (d), (f), and (g) the red lines or vertical bars represent the 95% confidence interval.