Title : Frustrated synaptic state for a memory with vast storage
Abstract:
A population of neurons modeled by a frustrated array of Josephson junctions has memory. The array can be thought of as a spin chain, say two dimensional. The dynamics tries to minimize the energy, and this requires the spins to point in the same direction. Frustration, however, is a strong local boundary condition inhibiting the system to acquire such a state. This gives birth to an energy landscape holding the memory as a specific collection of local minimum states in a vast configuration space. Now, to bring the picture closer to a neuron model, the synapse is going to define the adjacency matrix; that is, the coupling strength, in the junction model. In the Hebbian synapse both pre-synapse and post-synapse are active together, and the synapse gets stronger if in this correlation the pre-synapse comes first. This is incorporated in a Learning Rule known as Spike Timing-Dependent Plasticity (STDP), providing us with the dynamics. In this dynamics we look for synchronous states. We stimulate the array of neurons by a pacemaker neuron. Frustration and plasticity, together, lead to hysteresis. Along with the potential for an immense number of memory states, as the landscape holds the memory, we also discuss how our model develops a Lisman switch needed for learning and memory.
