Fig. 1: Co-dependent synaptic plasticity model.
a, Co-dependent excitatory (top) and inhibitory (bottom) plasticity. Plasticity of a synapse (highlighted with black contour) depends on the activation of its neighboring excitatory (red) and inhibitory (blue) synapses, together with its synapse-specific presynaptic and postsynaptic activity—that is, spike times, indicated by PRE and POST, respectively. Variables E and I integrate NMDA and GABAergic currents (low-pass filters), respectively. b, Excitatory weight change, ΔwE, as a function of the time interval between postsynaptic and presynaptic spikes, Δt, and neighboring synaptic inputs, E and I. \(\Delta t={t}_{{{{\rm{post}}}}}-{t}_{{{{\rm{pre}}}}}\), where tpost,and \({t}_{{{{\rm{pre}}}}}\) are spike times of postsynaptic and presynaptic neurons, respectively, so that Δt > 0 for pre-before-post and Δt < 0 for post-before-pre spike patterns. c, Excitatory inputs, E, control Hebbian LTP (green line; Δt > 0) and heterosynaptic plasticity (orange line), which combined (gray line) create a common setpoint for the total excitatory input (red dot). d, Inhibitory inputs, I, gate excitatory plasticity (‘ON’ versus ‘OFF’). e, Inhibitory weight change, ΔwI, is a function of Δt and neighboring synaptic inputs (as in b). f,g, Synaptic changes in inhibitory synapses as a function of excitatory (f) and inhibitory (g) inputs.
