Figure 6.

Comparing the sensitivity of synaptic changes with common-input correlations and firing rate changes. A, Illustration of two mutually coupled integrate-and-fire neurons that are driven by a barrage of independent (gray) and common fluctuating inputs (black). The fraction of common input, c, is varied and sets the correlation between the two neurons. The cross-correlation function can be determined mathematically (right; see Materials and Methods) (Ostojic et al., 2009). B, Increase in synaptic strength as function of the correlation coefficient between both neurons. The simulation results (circles) are well approximated by a linear increase, the slope of which can be determined mathematically (colored lines; see Materials and Methods). The slope depends on the baseline firing rate (two examples shown for the calcium- and the triplet-based models). C–E, Change in synaptic strength as a function of the firing rate of uncorrelated, intermediate and strongly correlated integrate-and-fire neurons, for the calcium- (with linear calcium dynamics, C), the triplet- (D), and the pair-based (E) models. Both neurons receive (1) no common input (ρ = 0, black), (2) weak common input yielding ρ = 0.2 (light red), and (3) strong common input yielding ρ = 0.4 (dark red). F–H, Sensitivity of synaptic plasticity to correlations (Eq. 25) as a function of the firing rate, for the calcium- (E), the triplet- (G), and the pair-based (H) models. The correlation coefficients are ρ = 0.2 (light red) and ρ = 0.4 (dark red). I, J, Sensitivity of synaptic plasticity to firing rate changes (Eq. 26) for the calcium- (I) and triplet-based (J) models. Color code illustrates the synaptic change for a given baseline firing rate (x-axis) and increase in firing rate (y-axis). Light and dark red lines indicate the firing rate increase evoking the same synaptic change as correlations from common input with ρ = 0.2 and ρ = 0.4, respectively, for the respective baseline rate.