Bridging Biological and Artificial Neural Networks with Emerging Neuromorphic Devices: Fundamentals, Progress, and Challenges

Synaptic transmission is a dynamic process. Postsynaptic responses wax and wane as presynaptic activity evolves. This prominent characteristic of chemical synaptic transmission is a crucial determinant of the response properties of synapses and, in turn, of the stimulus properties selected by neural networks and of the patterns of activity generated by those networks. This review focuses on synaptic changes that result from prior activity in the synapse under study, and is restricted to short-term effects that last for at most a few minutes. Forms of synaptic enhancement, such as facilitation, augmentation, and post-tetanic potentiation, are usually attributed to effects of a residual elevation in presynaptic [Ca(2+)]i, acting on one or more molecular targets that appear to be distinct from the secretory trigger responsible for fast exocytosis and phasic release of transmitter to single action potentials. We discuss the evidence for this hypothesis, and the origins of the different kinetic phases of synaptic enhancement, as well as the interpretation of statistical changes in transmitter release and roles played by other factors such as alterations in presynaptic Ca(2+) influx or postsynaptic levels of [Ca(2+)]i. Synaptic depression dominates enhancement at many synapses. Depression is usually attributed to depletion of some pool of readily releasable vesicles, and various forms of the depletion model are discussed. Depression can also arise from feedback activation of presynaptic receptors and from postsynaptic processes such as receptor desensitization. In addition, glial-neuronal interactions can contribute to short-term synaptic plasticity. Finally, we summarize the recent literature on putative molecular players in synaptic plasticity and the effects of genetic manipulations and other modulatory influences.

A memristor is a two-terminal electronic device whose conductance can be precisely modulated by charge or flux through it. Here we experimentally demonstrate a nanoscale silicon-based memristor device and show that a hybrid system composed of complementary metal-oxide semiconductor neurons and memristor synapses can support important synaptic functions such as spike timing dependent plasticity. Using memristors as synapses in neuromorphic circuits can potentially offer both high connectivity and high density required for efficient computing.

Hebbian models of development and learning require both activity-dependent synaptic plasticity and a mechanism that induces competition between different synapses. One form of experimentally observed long-term synaptic plasticity, which we call spike-timing-dependent plasticity (STDP), depends on the relative timing of pre- and postsynaptic action potentials. In modeling studies, we find that this form of synaptic modification can automatically balance synaptic strengths to make postsynaptic firing irregular but more sensitive to presynaptic spike timing. It has been argued that neurons in vivo operate in such a balanced regime. Synapses modifiable by STDP compete for control of the timing of postsynaptic action potentials. Inputs that fire the postsynaptic neuron with short latency or that act in correlated groups are able to compete most successfully and develop strong synapses, while synapses of longer-latency or less-effective inputs are weakened.