Figure 3.

Maximum storage capacity obtained in attractor neural networks with dynamic synapses with both depressing and facilitating mechanisms. (A) Behaviour of α_c as a function of U, τ_rec and τ_fac. The solid lines correspond the theoretical prediction from the mean field Equation (14) and symbols are obtained from Monte Carlo simulations. From top to bottom, it is shown α_c(U) for τ_rec = 2 and different values of τ_fac, α_c(τ_rec) for U = 0.2 and different values of τ_fac and α_c(τ_fac) for U = 0.2 and different values of τ_rec, respectively. The horizontal dotted lines correspond to the static synapses limit α_c ≈ 0.138. (B) Mean-field results from Equation (14) for the dependence of α_c for different combinations of relevant parameters. This corresponds—from top to bottom—to the surfaces α_c(U, τ_fac) for τ_rec = 2, α_c(U, τ_fac) for τ_rec = 50 and α_c(τ_rec, τ_fac) for U = 0.02. In all panels, τ_rec and τ_fac are given in MCS units that can be associated to a value of 5 ms if one assumes that a MCS corresponds to the duration of the refractory period in actual neurons.