Fig. 6

(a) The heat capacity increases linearly for the LNL cascade, the EIF and the DG models. The heat capacity of the LNL cascade increases at a slightly greater rate than that of the EIF and DG models, which overlap. The heat capacity for the PME model saturates at a population of approximately \(N=30\) neurons. (b) Comparing the \(L(c)\) vs. the \(\tilde{L}( f(c))\) functions for the DG and LNL models, respectively (here input correlations \(\lambda=0.17\) for DG model). The two functions largely agree over about 2 standard deviations of the Gaussian pdf \(\phi_{\mathrm{DG}}(c)\) (shaded)