Section: New Results

From the microscopic to the mesoscopic scale

Participants: Axel Hutt, Laure Buhry, Meysam Hashemi, Pedro Garcia Rodriguez, Peter beim Graben, James Wallace Sleigh.

Several previous studies focus on the derivation of neural population models. However most of these studies do not consider explicitly the microscopic properties of neurons, such as synaptic receptor dynamics or ion-channel distributions, although they may be implicit. The resulting models in some previous studies are poorly tractable analytically due to their complexity. Moreover, the complexity of previous models makes it difficult to discover those elements in the model that induce certain dynamical features as observed in experiments. Essentially most of previous studies do not consider the spatial interactions of neurons and, importantly, neglect delays present in biological networks. We aim to improve some previous models and a first step to a new statistical approach has been developed [11] , [18] , [17] , [22] to bridge the scales between the network activity of coupled spiking single neurons and statistical quantities of populations, e.g., the mean membrane potential in the network and the networks population firing rate. Our work considers the specific effect of anaesthetics and takes into account the physiological effects of extra-synaptic GABA${}_A$-receptors at single neurons, which are highly sensitive to anaesthetic drugs, such as propofol. We find numerically by simulation of a spiking neural network that propofol on single neuron level diminishes the network oscillation power in the $\alpha -$frequency band and affects strongly the spike coherence in the population. Such effects have been shown in previous experimental data obtained during propofol anaesthesia demonstrating the importance of extra-synaptic receptor dynamics in the understanding of experimental phenomena in anaesthesia.

The neural origin of generation and planning of motor action in humans is still unknown. In this context, psychophysical experiments and the neural modeling of the gained results may lead to further insight. We have participated in an experimental and theoretical study [8] to reveal the effect of temporal attention on non-conscious prime processing. Our stochastic accumulator model improves extensively the standard accumulator model for reaction time by involving additional stochastic neural accumulators, which permits an almost perfect fit to experimental data. The model indicates that motor action, which is generated on a population level, obeys a stochastic accumulation of activity of single neuron activity.