Connectivity reflects coding: a model of voltage-based STDP with homeostasis

Abstract

Electrophysiological connectivity patterns in cortex often show a few strong connections, sometimes bidirectional, in a sea of weak connections. In order to explain these connectivity patterns, we use a model of Spike--Timing--Dependent Plasticity where synaptic changes depend on presynaptic spike arrival and the postsynaptic membrane potential, filtered with two different time constants. The model describes several nonlinear effects in STDP experiments, as well as the voltage dependence of plasticity. We show that in a simulated recurrent network of spiking neurons our plasticity rule leads not only to development of localized receptive fields, but also to connectivity patterns that reflect the neural code: for temporal coding paradigms with spatio--temporal input correlations, strong connections are predominantly unidirectional, whereas they are bidirectional under rate coded input with spatial correlations only. Thus variable connectivity patterns in the brain could reflect different coding principles across brain areas; moreover simulations suggest that plasticity is surprisingly fast.

How feedback inhibition shapes spike-timing-dependent plasticity and its implications for recent Schizophrenia models

“…Generally this is either created specifically, or is derived from rather computational principles, for example, the "spike response" model (Jolivet et al, 2003). Recently, this classical model has been improved by taking into account the postsynaptic potential (Clopath et al, 2010). In our case the STDP curve emerges by integrating the learning rule (Eq.…”

“…Typically pre-and then postsynaptic stimulation causes long term potentiation (LTP) while post-and then presynaptic stimulation causes long term depression (LTD). It has been shown that the STDP curve is not constant but changes its shape when the pre-and postsynaptic potentials change Tamosiunaite et al, 2006;Voegtlin, 2009;Clopath et al, 2010). For example, backpropagating spikes can change the shape of the STDP curve and distal dendrites have their own STDP curves because the dynamics of the postsynaptic potentials is much slower far away from the soma (Tamosiunaite et al, 2007a).…”

Networks that learn the precise timing of event sequences

“…Variables symbol description I j external stimulus for excitatory population j u j non-dimensional firing rate of excitatory population j (maximum u j = 1) v non-dimensional firing rate of global inhibitory population (maximum v = 1) p j level of facilitation of synapses from population j (baseline p j = 1) w jk , w strength of excitation from population k to excitatory population j T j , T duration of stimulus Similar assumptions have been used in previous rate-based models of LTP/LTD (von der Malsburg, 1973;Bienenstock et al, 1982;Oja, 1982;Miller, 1994), and it has been shown that calciumbased (Graupner and Brunel, 2012) and spike-time dependent (Clopath et al, 2010;Gjorgjieva et al, 2011) plasticity rules can be reduced to such ratebased rules (Pfister and Gerstner, 2006). Furthermore, the fact that pre-synaptic activity is necessary to initiate either LTP or LTD is supported by experimental observations that plasticity depends on calcium influx through NMDA receptors (Malenka and Bear, 2004).…”

Use of multi-electrode array recordings in studies of network synaptic plasticity in both time and space

“…Their primary designs have been reported by Egert et al [15] and Oka et al [16] . and pathological conditions [27,28] . In this respect, our lab investigated the possible existence of spatial plasticity in the hippocampal formation (HF) induced by persistent peripheral nociceptive stimuli in a rodent model of bee venom (BV)-evoked persistent inflammatory pain [29] .…”