Extended Data Fig. 8: Estimating Isyn with different combinations of model parameters.
a–c, Relationship between PSD area and Isyn, estimated using four different parameter combinations in the compartmental model. The white circles and dashed lines show Isyn estimated from the default model used throughout this paper (as in Fig. 3f); for clarity, data points are shifted slightly horizontally. a, b, Estimates of Isyn were remarkably robust to changes in Ri and Rm. Parameters were tuned such that the model cell reproduced our somatically recorded current–voltage responses. Thus, changing Ri required Rm to be adjusted in the opposite direction. This appeared to have little net effect on the transfer impedance, which was then used to estimate Isyn. c, Lowering Cm resulted in lower estimates of Isyn. We chose a value for Cm of 1 μF cm−2 because of its widespread use in previous L2/3 models; we note, however, that Cm values below 2 μF cm−2 resulted in unrealistically short membrane time constants when compared with our in vitro recordings from the neuron that was later reconstructed for the model cell (data not shown). Models with Cm values of 2 μF cm−2 reproduce our experimental current–voltage responses markedly better and thus provide more realistic fits. Thus, the plot does not resemble an accurate estimate of Isyn, but rather a minimum bound for Isyn as a function of parameter tuning. d, Table summarizing model parameter combinations and non-parametric Spearman correlation coefficients (r), P values of correlations, and slopes of linear regressions for the four resulting PSD–Isyn relationships; varied parameters are indicated with bold font. The specific combinations of model parameters were chosen because they conformed with experimental measurements or were used in previous models of L2/3 neurons (see Methods). Note that all tested models resulted in statistically significant PSD–Isyn relationships.
