Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE credits
Azadeh Hassannejad Nazir
STDP between a pair of recurrently connected neurons with asynchronous and synchronous stimulation
A mammalian nervous system receives a lot of information from the environment which it interacts with. Therefore, a mammalian brain needs to be dynamic to learn and store both short and long term information without being saturated. Learning and memory occurs through the formation of cell assemblies and the dynamics of their synapses. ?This thesis focuses on modeling the dynamics of long term memory in a network of two Izhikevich spiking neurons with reciprocal synapses and Spike Timing Dependent Plasticity (STDP).
After preliminary work showed that additive STDP was incapable of forming a recurrent network and a stable structure, the STDP learning rule was modified to be multiplicative and to include a non-Hebbian component. In addition, asynchronous, synchronous, and polysynchronous thalamic stimulation was included in the model. The introduction of different kinds of stimulation, with the contribution of STDP, created the conditions for stable bidirectional synaptic weight growth, stable unidirectional synaptic weight growth and absence of weight growth in either connection.
It was found that the firing frequency of the neurons, connection conduction delays and input stimuli are all factors affecting synaptic plasticity and memory storage. Synapses adaptable by STDP regulated the timing of the postsynaptic firings. With asynchronous stimulation, STDP contributed to the controlled drift of firing time of the connected neurons.? With synchronous stimulation, weight stability was achieved for both feed-forward and recurrent connections, dependent on frequency and connection delays.? With polysynchronous stimulation, the neuron stimulated at the lower frequency increases its firing rate to the higher frequency. Considering these results, the emergence of strongly connected neurons have a tendency to fire together in a specific temporal window. This may indicate how more complex cell assemblies form in a larger network of neurons.