jonke_synapse – Synapse type for spike-timing dependent plasticity with additional additive factors.

Description

jonke_synapse is a connector to create synapses with spike time dependent plasticity. Unlike stdp_synapse, we use the update equations:

\[\begin{split}\Delta w &= \lambda \cdot w_{max} \cdot (K_+(w) \cdot F_+(t) - \beta) & \quad if t - t_j^(k) > 0 \\ \Delta w &= \lambda \cdot w_{max} \cdot (-alpha \cdot K_-(w) \cdot F_-(t) - \beta) & \quad else\end{split}\]

where

\[\begin{split}K_+(w) &= \exp(\nu_+ w) \\ K_-(w) &= \exp(\nu_- w)\end{split}\]

and

\[\begin{split}F_+(t) &= \exp((t - t_j^(k))/\tau_+) \\ F_-(t) &= \exp((t - t_j^(k))/\tau_-)\end{split}\]

This makes it possible to implement update rules which approximate the rule stated in [1], and for examples, the rules given in [2] and [3].

Warning

This synaptic plasticity rule does not take precise spike timing into account. When calculating the weight update, the precise spike time part of the timestamp is ignored.

Parameters

lambda	double	Step size
Wmax	double	Maximum allowed weight, note that this scales each weight update
alpha	double	Determine shape of depression term
mu_plus	double	Set weight dependency of facilitating update
mu_minus	double	Set weight dependency of depressing update
tau_plus	double	Time constant of STDP window, potentiation in ms
beta	double	Set negative offset for both updates

(tau_minus is defined in the postsynaptic neuron.)

Transmits

SpikeEvent

References

[1]

Nessler, Bernhard, et al. “Bayesian computation emerges in generic cortical microcircuits through spike-timing-dependent plasticity.” PLoS computational biology 9.4 (2013): e1003037.

[2]

Legenstein, Robert, et al. “Assembly pointers for variable binding in networks of spiking neurons.” arXiv preprint arXiv:1611.03698 (2016).

[3]

Jonke, Zeno, et al. “Feedback inhibition shapes emergent computational properties of cortical microcircuit motifs.” arXiv preprint arXiv:1705.07614 (2017).

See also

Synapse, Spike-Timing-Dependent Plasticity

Examples using this model

None