Cosyne 2007 Workshops
February 26-27, 2007
The Canyons, Utah
Surya Ganguli
Oscillations, Plasticity and Memory Formation in a Solvable Recurrent Spiking Network
It is thought that the hippocampus is a fast plasticity system whose synapses rapidly encode episodic memories which are then, over longer time scales, consolidated and transferred to cortical areas. For such a transfer to occur, newly encoded hippocampal synaptic patterns must successfully compete against and induce changes in existing cortical synaptic patterns. We examine the conditions under which such changes are possible in an exactly solvable recurrent spiking network which faithfully takes into account 3 biophysical time scales: the EPSP time scale, the time scale of a spike-timing dependent plasticity (STDP) window, and the time scale of any ambient network oscillation. Through analytic calculations, we find a synergistic interaction between these three time scales that allows for the transfer of information embedded in hippocampal synapses to cortical synapses in a way that would not be possible in the absence of hippocampal oscillations. Mechanistically, this synergy arises because of two reasons: 1) The (potentiating or depotentiating) force on any given cortical synapse due to hippocampal synapses is given by the sum of individual forces exerted by each possible synaptic pathway from the hippocampus to the cortical synapse. 2) Ambient oscillations allow the resulting oscillatory forces exerted by these individual synaptic pathways to constructively interfere, in turn allowing a larger force to be exerted by the hippocampus on any given cortical synapse. The oscillation frequency controls the relative phase of the force exerted by each synaptic pathway, and hence controls the degree of constructive interference between all pathways. For physiological ranges of the EPSP and STDP time scales, we find analytically that oscillations in the gamma range maximize such interference, and hence maximize the transfer of information between hippocampal and cortical synaptic patterns.
