Search PubMed⌕ Search
Home/Archive/PubMed

PubMed · 11032032

Spike-driven synaptic plasticity: theory, simulation, VLSI implementation.

S Fusi·M Annunziato·D Badoni·A Salamon·D J Amit
View original source ↗Export metadata ↓

Abstract

We present a model for spike-driven dynamics of a plastic synapse, suited for aVLSI implementation. The synaptic device behaves as a capacitor on short timescales and preserves the memory of two stable states (efficacies) on long timescales. The transitions (LTP/LTD) are stochastic because both the number and the distribution of neural spikes in any finite (stimulation) interval fluctuate, even at fixed pre- and postsynaptic spike rates. The dynamics of the single synapse is studied analytically by extending the solution to a classic problem in queuing theory (Takacs process). The model of the synapse is implemented in aVLSI and consists of only 18 transistors. It is also directly simulated. The simulations indicate that LTP/LTD probabilities versus rates are robust to fluctuations of the electronic parameters in a wide range of rates. The solutions for these probabilities are in very good agreement with both the simulations and measurements. Moreover, the probabilities are readily manipulable by variations of the chip's parameters, even in ranges where they are very small. The tests of the electronic device cover the range from spontaneous activity (3-4 Hz) to stimulus-driven rates (50 Hz). Low transition probabilities can be maintained in all ranges, even though the intrinsic time constants of the device are short (approximately 100 ms). Synaptic transitions are triggered by elevated presynaptic rates: for low presynaptic rates, there are essentially no transitions. The synaptic device can preserve its memory for years in the absence of stimulation. Stochasticity of learning is a result of the variability of interspike intervals; noise is a feature of the distributed dynamics of the network. The fact that the synapse is binary on long timescales solves the stability problem of synaptic efficacies in the absence of stimulation. Yet stochastic learning theory ensures that it does not affect the collective behavior of the network, if the transition probabilities are low and LTP is balanced against LTD.

Explore related subjects

Action PotentialsArtificial IntelligenceComputer SimulationComputersElectric ConductivityLong-Term PotentiationModels, NeurologicalNeural InhibitionNeural Networks, ComputerNeuronal PlasticityProbabilityStochastic ProcessesSynapsesJournal ArticleResearch Support, Non-U.S. Gov't

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

S Fusi, M Annunziato, D Badoni, A Salamon, D J Amit. 2000. Spike-driven synaptic plasticity: theory, simulation, VLSI implementation.. https://doi.org/10.1162/089976600300014917

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2001

Channel activity of deamidated isoforms of prion protein fragment 106-126 in planar lipid bilayers.

Using the lipid bilayer technique, we have found that age-related derivatives, PrP[106-126] (L-Asp108) and PrP[106-126] (L-iso-Asp108), of the prion protein fragment 106-126 (PrP[106-126] (Asn108)) form heterogeneous ion channels. The deamidated isoforms, PrP[106-126] (L-Asp108) and PrP[106-126] (L-iso-Asp108), showed no enhanced propensity to form heterogeneous channels compared with PrP[106-126] (Asn108). One of the PrP[106-126] (L-Asp108)- and PrP[106-126] (L-iso-Asp108)-formed channels had three kinetic modes. The current-voltage (I-V) relationship of this channel, which had a reversal potential, E(rev), between -40 and -10 mV close to the equilibrium potential for K+ (E(K)-35 mV), exhibited a sigmoidal shape. The value of the maximal slope conductance (g(max)) was 62.5 pS at positive potentials between 0 and 140 mV. The probability (P(o)) and the frequency (F(o)) of the channel being open had inverted and bell-shaped curves, respectively, with a peak at membrane potential (V(m)) between -80 and +80 mV. The mean open and closed times (T(o) and T(c)) had inverted bell-shaped curves. The biophysical properties of PrP[106-126] (L-Asp108)- and PrP[106-126] (L-iso-Asp108)-formed channels and their response to Cu(2+) were similar to those of channels formed with PrP[106-126] (Asn108). Cu(2+) shifted the kinetics of the channel from being in the open state to a "burst state" in which rapid channel activities were separated by long durations of inactivity. The action of Cu(2+) on the open channel activity was both time-dependent and voltage-dependent. The fact that Cu(2+) induced changes in the kinetics of this channel with no changes in the conductance of the channel indicated that Cu(2+) binds at the mouth of the channel. Consistently with the hydrophilic and structural properties of PrP[106-126], the Cu(2+)-induced changes in the kinetic parameters of this channel suggest that the Cu(2+) binding site could be located at M(109) and H(111) of this prion fragment.

Action Potentials↗
PubMed2001

A system for applying rapid warming or cooling stimuli to cells during patch clamp recording or ion imaging.

We describe a system for superfusing small groups of cells at a precisely controlled and rapidly adjustable local temperature. Before being applied to the cell or cells under study, solutions are heated or cooled in a chamber of small volume ( approximately 150 microl) and large surface area, sandwiched between four small Peltier elements. The current through the Peltier elements is controlled by a microprocessor using a PID (proportional-integral-derivative) feedback algorithm. The chamber can be heated to at least 60 degrees C and cooled to 0 degrees C, changing its temperature at a maximum rate of about 7 degrees C per second; temperature ramps can be followed under feedback control at up to 4 degrees C per second. Temperature commands can be applied from the digital-to-analogue converter of any laboratory interface or generated digitally by the microprocessor. The peak-to-peak noise contributed by the system does not exceed that contributed by a patch pipette, holder and headstage, making it suitable for single channel as well as whole cell recordings.

Action Potentials↗
PubMed2001

Region-specific developmental specialization of GABA-glycine cosynapses in laminas I-II of the rat spinal dorsal horn.

The spinal dorsal horn is the first level of the CNS in which nociceptive input from sensory afferents is integrated and transmitted. Although inhibitory control in this region has a crucial impact on pain transmission, the respective contribution of GABA and glycine to this inhibition remains elusive. We have previously documented co-release of GABA and glycine at the same inhibitory synapse in spinal laminas I-II of adult rats [older than postnatal day 30 (P30)]. However, despite this co-release, individual miniature inhibitory postsynaptic currents (mIPSCs) were mediated by either glycine receptors (GlyR) or GABA(A) receptors (GABA(A)R), yet never by the two together. In contrast, recent studies of ventral horn immature inhibitory synapses (</=P21) reported individual mIPSCs that were mediated by both GABA(A)Rs and GlyRs. This raises the question of whether mixed mIPSCs are present in immature lamina I-II neurons yet are lost through a maturation-dependent synaptic specialization. To test this, we recorded mIPSCs using patch-clamp techniques in lamina I-II neurons in spinal slices taken at different stages of development. We found that, in neurons younger than P23, both GlyR-only and GABA(A)R-only mIPSCs could be recorded, in addition to mixed GABA(A)R-GlyR mIPSCs. With maturation however, both lamina I-II neurons gradually discontinued exhibiting mixed mIPSCs, although with differing patterns of specialization. Yet, at all developmental stages, benzodiazepine administration could unmask mixed mIPSCs. Together, these findings indicate that, although GABA and glycine are continually co-released throughout development, junctional codetection ceases by adulthood. This indicates an age-dependent postsynaptic tuning of inhibitory synapses that occurs in a region-specific manner.

Action Potentials↗