Controlling spike timing and synchrony in oscillatory neurons

Abstract

Many processes in the brain, both normal and patholo-gical, involve oscillations of neuronal populations. The ability to enhance or disrupt neuronal synchrony has been clinically demonstrated to have remarkable effects in a number of neurological disorders including: Parkin-son's disease, Epilepsy, and Depression [1,2]. We have developed an algorithm to control the spike timing of periodically firing neuron using patch clamp and real-time dynamic clamp techniques [3]. Furthermore, this algorithm was expanded to control the relative spike timing between two oscillating neurons. These present the first steps towards more precise population control schemes. The single cell controller uses the neurons phase response curve and the relationship between the spike advance and the current injection at a given stimulus phase to create a control function [4]. The two cell con-troller uses the same premise as the single cell control-ler, but incorporates additional logic to determine the direction and number of periods required to achieve the target phase offset. We tested our controller using a real-time model neu-ron and CA1 pyramidal neurons [5]. Noise was added to the model neuron to replicate that seen in biological neurons. In single cell control experiments, the control-ler could account for ~99% and ~87% of the neurons variance for the model and CA1 pyramidal neuron, respectively. In two cell experiments, we tested the con-troller using two noisy model neurons and we per-formed hybrid testing using a model neuron as the leader and a pyramidal neuron as the follower. In the hybrid case, the controller accuracy was moderate, with a normalized vector correlation of 0.69. In the two model neuron case, the controller accuracy was high, with a normalized vector correlation of 0.98 [6]. Using the two neuron model case, we tested the controller for robustness across ISI mismatching and target phase off-set. The controller was robust to ISI mismatch as long as it was less than the extremes of the single pulse spike advance, outside that range the accuracy of the control-ler dropped sharply. The controller accuracy was inde-pendent of the desired phase offset between two neurons. In both cases the level of noise injected into the neuron was the primary modulator of controller accuracy.