Synaptic plasticity: from biophysics to learning

• Applicant: Dr. Jonas Ranft
• Subject Area: Cognitive, Systems and Behavioural Neurobiology ; Statistical Physics, Nonlinear Dynamics, Complex Systems, Soft and Fluid Matter, Biological Physics
• Term: from 2014 to 2017
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 255001987

As a principal site of memory storage in the brain, synapses need to be able to lastingly modify their connection strength in response to neural activity, a phenomenon referred to as synaptic plasticity. Understanding synaptic plasticity is key to eventually explain the fascinating fact that synapses, which are highly dynamic structures and subject to significant stochasticity at the molecular level, can reliably store and maintain memories on the timescale of years. In this project, I aim to shed new light on three aspects of synaptic plasticity which together bridge the gap between biophysics and function: (1) Implementation at the molecular level: Synaptic plasticity is at least in part mediated by the number of postsynaptically available neurotransmitter receptors. Experimental data show that receptor clusters are dynamic, which was not or only rudimentary taken into account by previous theoretical approaches. Using stochastic simulations and stability analysis, I aim to quantitatively understand the dynamics and statistics of the clusters formed by receptor and scaffolding molecules, and I investigate ramifications for the implementation of synaptic plasticity.(2) Control at the synaptic level: Pre- and postsynaptic activity control plasticity via concentration transients of signaling molecules. I plan to develop a model of calcium- and NO-based spike-timing dependent plasticity of the central synapse in the cerebellum, which I further analyze in terms of stability of learned synaptic weights. I explore the hypothesis suggested by preliminary experimental data which states that synaptic bistability is caused by differential plasticity, i.e., plasticity that depends itself on current synaptic weight.(3) Function at the network level: It has not yet been investigated how such spike-timing dependent plasticity is put to use in biologically plausible learning scenarios. Using simulation and mathematical analysis, I plan to investigate the dynamics of spiking network models of the cerebellum that incorporate the model of synaptic plasticity developed in (2). Here, I focus on the plausibility of biological learning algorithms and the long-term stability of learned memories.I am convinced that given the current state of research it is timely to explore these aspects and that the cerebellum serves as an excellent model system to address questions at the functional level. Close collaboration with experimental groups for all three parts of the project will allow to test model predictions and to continuously inform model development by new, so far unchallenged experimental data.

• DFG Programme: Research Fellowships
• International Connection: France
• Host: Dr. Vincent Hakim