Figure 1 | Scientific Reports

Figure 1

From: E-I balance emerges naturally from continuous Hebbian learning in autonomous neural networks

Figure 1

Left. In red the nonlinear transfer function relating membrane potentials and rates in the neural model (1). The typical activity rates enforced by the intrinsic plasticity rule (y t  = 0.2) result in the neuron operating at the foot of the non-linearity, where it is practically indistinguishable (m.s.e. = 0.027 for x [−5:0]) from a threshold-powerlaw with exponent n = 2.5 (in blue), typically considered a suitable model for experimental findings in cortical neurons33,34,35. Right. The effective synaptic strength multiplier φ(t)u(t) of the Tsodyks-Markram model (4). Here β = α = 0.01 and U max  = 4 was used. The red/violet curves correspond to the values as measured respectively for excitatory synapses in the medial prefrontal cortex of ferrets43 and for inhibitory layer 2–4 neurons of the somatosensory cortex layer of Wistar rats44. The presynaptic neuron is active for t [100, 300] (ms), and inactive otherwise.

Back to article page