Spatiotemporal Response of Neurons to Time-Dependent Spectral Features

PABLO BALENZUELA; JORDI GARCIA OJALVO

Montevideo, Uruguay

Workshop; 7th International Neural Coding Workshop; 2007

In this work, we analyze the behavior of simulated neurons when they are stimulated by a synaptic train current with variable frequency. For fixed input frequency and similarly to the case of harmonic inputs, these neurons exhibit a resonant behavior. We interpret this phenomenon in terms of the subthreshold response of the neuron. In the presence of dynamical trains of input pulses whose frequency varies continuously in time, the receiving neuron synchronizes episodically to the input pulses, whenever the pulse frequency lies within the neuron’s locking range. We tested this mechanism constructing heterogeneous networks of neurons where different subsets respond to the stimulus depending on its frequency. Input signals received by most non-sensory neurons take the form of pulse trains, coming from the spiking activity of neighboring neurons. Therefore, in order to understand the mechanisms of information processing in neural systems, it is very important to characterize in detail the re- sponse of neurons to pulse trains. Furthermore, realistic pulse trains are intrinsically dynamical, with an instantaneous firing frequency that varies continuously in time. It is therefore necessary to assess the influence of this non-stationarity in the neuronal response. Most studies of driven neurons have been restricted so far to harmonic driving signals [1, 2, 3, 4, 5]. Many of these works have shown that for certain types of neurons, i.e. those exhibiting what is called type I I excitability, a resonant behavior arises with respect to the external driving frequency [6, 7, 8]. Excitability in those neurons is usually associated with an Andronov-Hopf bifurcation, which leads to the existence of subthreshold oscillations in the excitable regime. When the frequency of these oscillations equals that of the harmonic driving, a resonance arises. A similar resonant behavior exists for pulsed inputs. In that case, the same pulse train impinging on two different neurons can elicit a response on only one of them, i.e. on the one that is tuned to resonate with the incoming pulse frequency. Izhikevich and co-workers [9, 10] proposed this type of phenomenon as a mechanism for selective multiplexed communication, whereby a single neural channel can be used to transmit multiple signals, each of which is detected by distinct groups of neurons depending on their interspike frequency. Here we pursue this idea further, analyzing the resonant behavior in terms of the subthreshold response of the neuron. On the other hand, the question of how neurons response to changing input compared with stationary case was addressed in [11, 12, 13] when a pacemaker firing neuron was stimulated with a inhibitory synapses. In this work, we also study the response of a quiescent simulated neuron to a frequency dependent excitatory synaptic input. We show that this configuration leads to episodic synchronization between the neuron’s output and an input with dynamically varying firing rate [14].