Interaction of neuronal and network mechanisms on firing propagation in a feedforward network

Abstract

The mammalian brain has enormously complex neuronal diversity and a highly modular structure. The propagation of information in the modular brain network can be modeled by a feedforward network (FFN). Although studies in this area have yielded many important results, neuronal diversity has rarely been considered. In the current work, we investigate the complex interactions between the intrinsic properties of neurons and the FFN structure in the propagation of spiking activity. Here, four typical types of cortical neurons reproduced by the Izhikevich neuron model are introduced. A homogeneous FFN composed of a single type of excitatory neuron (regular spiking, mixed model, or tonic bursting) can propagate spiking activity. However, an FFN with fast spiking neurons does not propagate spiking activity. By modifying the network structure and synaptic weights, the spiking propagation of the homogeneous FFNs can vary from synchronous transmission (with a high firing rate) to asynchronous transmission (with a low firing rate). Among the homogeneous FFNs, both the firing rate and the synchrony of the FFN with tonic bursting neurons are the highest, but those of the FFN with regular spiking neurons is lowest, even when implementing the same FFN structure. For the FFN with mixed neuronal types, interestingly, the spiking propagation is very sensitive to the composition of the four types of neurons. By introducing fast spiking neurons into the homogeneous FFN composed of excitatory neurons, spiking propagation can be modified from synchronous to asynchronous. Similarly, changing the proportion of any of the types of neuron affects the spiking propagation, even for very small changes. The underlying mechanism of these observed results has also been discussed.

Introduction

Neurons convert a stimulus from the external environment or signals from upstream neurons into a train of action potentials (spikes), which are believed to be the fundamental process in the realization of brain function. The brain is the most complex organ in the human body, consisting of approximately 100 billion neurons. In the brain, neurons exhibit great diversity in their shapes and functions [1], [2]. Triggered with an external stimulus, different neurons respond with different firing patterns. Based on the firing patterns observed by intracellular recordings, neurons in the mammalian brain can be divided into various types [3], [4]. Regular spiking, tonic bursting and fast spiking neurons are typical neurons that were discovered in in vivo physiology and laminar network anatomy experiments [3], [5]. In the subiculum, different pyramidal neurons can generate regular spiking and bursting spiking patterns [6]. Neuronal diversity contributes to the emergent properties of neural networks and, consequently, plays an important role in information processing in the nervous system [7], [8], [9]. Therefore, studies about the activities of the neuronal network of a brain region should take into account intrinsic behavioral diversity both within and between neuron types.

Moreover, the nervous system is a highly modular structure [10]. Information processing in the nervous system is related to different functional groups of neurons, by which the information is transferred from one group to its downstream connected groups [11]. Thus, conditions under which spiking activity can be propagated among the neuronal groups are a crucial issue for information processing in the modular brain. In the last decade, a multilayer feedforward network (FFN) has been introduced to exploit the issue.

Multilayer FFNs, which provide important insights into the mechanisms of cortical computation, can mimic the properties of the propagation of spiking activity. Each layer of the network is related to a functional group of neurons, in which neurons in one group receive inputs from many neurons in the previous group, and thus the information is transmitted from one group to the next [11]. Recently, several computational studies have identified two dynamic activity modes that support the propagation of rate coding and temporal coding in the FFN [12], [13], [14]. When neurons in the first layer are subject to white noise, firing rate can be propagated in an FFN by the synchronized firings of Hodgkin–Huxley (HH) neurons [15], [16]. When weak signals are input to the neurons in layer 1, successive layers are able to propagate and amplify the signals while the neurons are subject to intrinsic or external noise above a certain level [17], [18]. Current computational studies also reveal that the propagation of spiking activities in an FFN depends on the interlayer connection probability, synaptic intensity, noise, and input signals [15], [16], [19], [20], [21]. Nevertheless, the diversity of neurons is rarely considered in studies of FFNs. Thus, it is not yet clear that the behaviors of spiking activity propagating within an FFN consider the intrinsic electrical properties of different neuronal types.

As mentioned earlier, neuronal diversity plays a crucial role in information processing in the nervous system. In the mammalian brain, the diversity of neuronal and nonneuronal cell types guarantees the execution of high-order cognitive, sensory, and motor behaviors [22]. Networks with diverse neuronal types offer superior decoding ability compared with homogeneous networks [23]. In a hybrid coupled neural network, it was shown that different types of population spiking patterns can emerge [24] and strongly synchronized population spiking events lead to complete activity cessation [25], by changing the types of connections in the network. It has also been found that increasing the diversity of intrinsic neuronal types can enhance the encoding performance of neuronal populations [26]. Andrew Bogaard et al. found that the response of a neuronal network is strongly affected by intrinsic neuronal properties and that the introduction of a small number of cells with different excitability properties can profoundly influence the spatiotemporal activity of the neural network [27].

Considering the importance of neuronal diversity in the nervous system, the effect of the intrinsic electrical properties of different neuronal types on the propagation of spiking activity in an FFN is addressed in the current work. This paper is organized as follows. In Section 2, the neuronal model, the FFN connections and the simulation methods are introduced. Simulation results are shown in Section 3. The conclusions and discussion are presented in Section 4.