Abstract
Developed through evolution, brain neural system selforganizes into an economical and dynamic network structure with the modulation of repetitive neuronal firing activities through synaptic plasticity. These highly variable electric activities inevitably produce a spontaneous magnetic field, which also significantly modulates the dynamic neuronal behaviors in the brain. However, how this spontaneous electromagnetic induction affects the selforganization process and what is its role in the formation of an economical neuronal network still have not been reported. Here, we investigate the effects of spontaneous electromagnetic induction on the selforganization process and the topological properties of the selforganized neuronal network. We first find that spontaneous electromagnetic induction slows down the selforganization process of the neuronal network by decreasing the neuronal excitability. In addition, spontaneous electromagnetic induction can result in a more homogeneous directedweighted network structure with lower causal relationship and less modularity which supports weaker neuronal synchronization. Furthermore, we show that spontaneous electromagnetic induction can reconfigure synaptic connections to optimize the economical connectivity pattern of selforganized neuronal networks, endowing it with enhanced local and global efficiency from the perspective of graph theory. Our results reveal the critical role of spontaneous electromagnetic induction in the formation of an economical selforganized neuronal network and are also helpful for understanding the evolution of the brain neural system.
Introduction
Neurons in the nervous system embedded in neural networks are connected via synapses where the efficiency of the network’s connectivity pattern is crucial for effective communication of neural information^{1,2}. Developed through evolution, these neuronal networks are optimized to economically transfer as much neural information as possible with a low energy cost^{1,2,3,4,5,6}; such a network can be well described by a smallworld network model with a high efficiency of information communication on both local and global scales^{2,7,8}. Due to the synaptic plasticity, macroscopic brain functions, such as learning, memory, and similar cognitive processes, promote the selforganization of microscopic neuronal networks to dynamically match the functional demands^{9,10,11,12,13,14}. Such a selforganized structure possessing both smallworld and scalefree properties has been found to be a more realistic way to characterize the real neuronal network^{15} and to have a considerable effect on the collective dynamics of neurons^{16,17,18,19,20,21,22}. For example, a microscopic selforganized neuronal network has higher coherence resonance, stochastic resonance and efficiency of information transmission than a globally coupled network or a random network while they have the same mean synaptic weight^{16}. Therefore, exploring how the neural system selforganizes to optimize its network structure to possess economical topological properties is helpful for studying collective neuronal dynamics and is also important for a more realistic understanding of the brain evolution.
The selforganized structure is originated from the feedback coupling between global dynamics and local neuronal networks through the biologically seen spiketimingdependent plasticity (STDP) rule, which has been observed in various vivo and vitro experiments^{23,24,25,26,27,28,29,30}. This STDP rule updates synaptic weights between neurons according to the relative timing between pre and postsynaptic action potentials on a millisecond timescale^{29}. If the firing time for the presynaptic neuron is ahead of that for the postsynaptic neuron, the synaptic connection is enhanced; otherwise, it is weakened^{23}. The modulated synapses in turn affect the neural responses. This feedback coupling between global network dynamics and local neuronal circuits is an important operating mode of the brain^{31} and is closely related to the mechanisms of learning and memory^{32,33}. However, the neuronal electric activities generated by the transmembrane flow of ions not only modulate the synaptic connection but also inevitably produce a timevarying electric field as well as a magnetic field, according to Maxwell’s theory of electromagnetic induction^{34,35,36}. This spontaneous magnetic field around neurons may be the foundation of brain transferring sensory stimulus via complex electromagnetic flows to the cortex^{37} and has significant effects on the dynamical properties of neurons and neuronal networks^{34,38,39,40,41}; for example, it induces multiple firing modes of neurons^{42,43}, promotes the double coherence resonance, inhibits the stochastic resonance^{40} and modulates spatiotemporal patterns^{44}. Meanwhile, magnetic field interactions between neurons support a potential spatial channel for neural information transmission^{45,46} and can significantly modulate signal communications between neurons^{45,47}, induce firing synchronization^{48,49}, trigger complex mode transitions of electrical activities^{50,51,52,53}, and even offset the effect of a blocked potassium ion channel on the collective dynamics^{54}. Furthermore, this spontaneous electromagnetic induction can induce abundant chaotic dynamics in a previously steadystate neural network; this phenomenon has been experimentally proven with an equivalent electric circuit model of the neural network in PSpice^{53}. More importantly, spontaneous electromagnetic induction in neurons provides a reasonable and effective path to study the effect of external electromagnetic fields on neural system functions^{48,55,56,57,58}. For example, the electromagnetic radiation can induce diverse synchronizations^{55,59}, stochastic resonance^{60}, rhythm disorder^{59} and mode transitions of electrical activities^{56} in the neuronal network; in particular, an optimal electromagnetic radiation intensity can induce the occurrence of stochastic resonance and a maximal neuronal response to weak signals^{55}. Even the spontaneous electromagnetic induction (i.e., negative magnetic feedback on neurons and magnetic coupling between them) has great effects on the dynamics of neurons and neuronal networks, its complex modulation on the selforganization of the neuronal network is still lacking.
Here, we investigate the effects of spontaneous electromagnetic induction on the evolutionary process and topological properties of the selforganized neuronal network at the microscopic scale. We first study how the negative feedback of magnetic fields on neurons and the magnetic coupling between them affect the selforganization process in the neuronal network, and then we utilize graph theory to explore the topological properties of the network. We find that spontaneous electromagnetic induction has complex effects on the evolutionary process and topological structure of the selforganized neuronal network. In particular, negative the feedback from a magnetic field can induce the economical neuronal network structure by reconfiguring synaptic connections during the selforganization process, and the magnetic coupling can further promote the formation of economical selforganized network structures.
Results
In this paper, the EulerMaruyama algorithm is used to solve the differential equations with a time step of 0.005 ms and a total time of 200 ms. The 80 excitatory neurons and 20 inhibitory neurons considered in the network are initially globally coupled by chemical synapses, where the synaptic weight is set to g_{max}/2 for excitatory synapses and to 3 g_{max}/2 for inhibitory synapses^{16,17,18}. During the selforganization process, the weights of excitatory synapses are updated based on the STDP rule, but the weight remains constant for inhibitory synapses^{16,17,18}.
Negative feedback of spontaneous magnetic field inhibits the selforganization process
To investigate the selforganization process of the neuronal network structure as modulated by spontaneous electromagnetic induction, we first focus on the negative feedback of magnetic fields on neurons rather than the magnetic coupling between them, i.e., fixing the magnetic coupling strength D = 0 and varying the negative feedback strength k_{1}. Meanwhile, to simply investigate the evolutionary process of the selforganized neuronal network, we use P_{0} to denote the percentage of synapses with weights in the range [0, 0.1 g_{max}] (weak coupling), P_{1} to represent the percentage of synapses with weights in the range [0.9 g_{max}, g_{max}] (strong coupling), and P_{2} to stand for all other cases^{16,18}. Due to the competition between heterogeneous neurons, synaptic connections from highactive neurons to lowactive neurons are enhanced, but they are weakened from lowactive neurons to highactive ones^{16,17}. Thus, P_{0} and P_{1} increase during the selforganization process, whereas P_{2} decreases^{16,18} (Fig. 1a), reflecting the modulation of chemical synapses by neuronal electric activities and the formation of a sparse selforganized structure. As P_{0}, P_{1} and P_{2} gradually reach stable values, the neuronal network achieves a dynamically stable state (i.e., longer than 150 ms; see Fig. 1a) and selforganizes into a directedweighted network structure.
Despite the similar evolutionary processes of neuronal networks for different negative feedbacks, the spontaneous magnetic field significantly affects the selforganization speed. We use P_{1} to measure the evolutionary speed of the neuronal network due to its relatively small fluctuation in the dynamically stable state (see Fig. 1a). Assuming that \(\overline{{P}_{1}}\) is the mean value of P_{1} in the dynamically stable state from 150 ms to 200 ms and f is an artificial fluctuation coefficient of P_{1}, if P_{1} always fluctuates within a range of \([(1f)\overline{{P}_{1}},(1+f)\overline{{P}_{1}}]\) over a transition time T, we claim that the neuronal network has achieved a dynamically stable state on the time scale T. Thus, the neuronal network has a higher selforganization speed for a shorter time T. This measure is apparently dependent on the fluctuation coefficient f and reflects the relative evolutionary speed. To present more reliable results, we calculate the transition times T for different fluctuation coefficients (Fig. 2). As k_{1} increases, the transition time T tends to increases — in particular, neuronal networks with k_{1} > 0.0 have significantly higher transition times than that for k_{1} = 0.0 — and this tendency is robust for different fluctuation coefficients (Fig. 2). These results indicate that spontaneous coupling between the magnetic field and the membrane potential slows down the selforganization process of the neuronal network.
Negative feedback of spontaneous magnetic field induces a more homogeneous directed network structure
By affecting the selforganization process, spontaneous magnetic field also has an influence on the synaptic weights in the stable selforganized neuronal network. To analyze the synaptic weight, we calculate the mean values of P_{0}, P_{1} and P_{2} in the stable state from 150 ms to 200 ms with a sampling step of 0.05 ms. As k_{1} increases, the mean P_{0} first increases and then decreases, the mean P_{2} shows a completely opposite variational trend, and the mean P_{1} nearly always decreases (Fig. 3a), reflecting nonmonotonic and complex modulation on synapses. For moderate negative feedback (e.g., k_{1} = 0.5), neuronal networks have smaller P_{2} and larger P_{0} values than those for k_{1} = 0.0, reflecting that the negative feedback from the magnetic field promotes the modulation of synaptic connections, such that synapses with intermediate weights (i.e., g_{ij} = 0.05) are reconfigured into weak synapses (see Fig. 1b). For strong negative feedback (e.g., k_{1} = 1.0), neuronal networks have smaller P_{0} and P_{1} values but larger P_{2} values than those for k_{1} = 0.0 (Fig. 3a), indicating that in this case, the magnetic field instead weakens the modulation of synaptic connections, such that many synapses with intermediate weights are not modulated (see Fig. 1b for k_{1} = 1.0). Thus, spontaneous magnetic field first promotes the modulation of synaptic connections in the stable selforganized neuronal network and then inhibits the modulation.
During the STDP updating process, directed synaptic connections are modulated by the firing lag between the pre and postsynaptic neurons, which actually reflects a kind of causal relationship between them. Spontaneous magnetic field has a great influence on the modulation of synaptic connections and will consequently affect the causal relationships between neurons. We utilize a causal flow (defined in Eq. 8) to measure the causal relationship among neurons in selforganized neuronal networks. A neuron with a higher positive causal flow has a greater causal influence on others and is more likely to be a “causal source” in the neuronal network. By contrast, a neuron with a small negative causal flow is greatly affected by other neurons and is called a “causal sink” of the network^{61}. Previous studies have found that a neuron with higher excitability tends to have a larger outdegree and a smaller indegree^{16,17}, i.e., is more likely to be a causal source, and a neuron with lower excitability is more likely to be a causal sink. And the local excitation can promote the synchronous propagate of neural activities through the heterogeneous selforganized neuronal network^{62}. Due to the negative feedback exerted by magnetic fields on firing activities, it is reasonable to suspect that spontaneous magnetic field will enhance the synaptic inputs to highactive neurons and weaken their outputs. We find that as the negative feedback becomes stronger, the negative causal flow increases and the positive causal flow decreases (Fig. 3b), indicating that the negative feedback from a magnetic field weakens the causal relationship by decreasing the neuronal excitability. These results also reveal that spontaneous magnetic field induces a more homogeneous directed selforganized network structure with less heterogeneity between in and outsynaptic weights of neurons.
Despite the spontaneous magnetic field inhibits the synchronization of the neuronal network with undirected and constant synapses^{49}, it has quite complex effects on the synchronization of the directed selforganized network by modulating the feedback between global network dynamics and local neuronal circuits. As the negative feedback strength increases, the mean synaptic weight in the selforganized neuronal network first decreases to the minimum value (i.e. k_{1} = 0.3, Fig. 3c), accompanied by a increase of network synchronization (i.e. a higher factor of synchronization, see Fig. 3d). In fact, in the directed and heterogeneous selforganized network, synaptic connections from highactive neurons to lowactive neurons are significantly larger than those from lowactive neurons to highactive neurons^{16,17}. While the weak connections are very small, lowactive neurons are mainly driven by highactive neurons and thus the network can achieve high synchronization. At k_{1} = 0.3, spontaneous electromagnetic field further increases the percentage of weak synaptic connections from lowactive neurons (i.e. higher P_{0}, Fig. 3a) and consequently enhances the firing synchronization (Fig. 3d). With the negative feedback strength increasing, the percentage of weak synapses are decreased (i.e. lower P_{0}, Fig. 3a) and the mean synaptic weight increases (Fig. 3c), accompanied with more homogeneity between in and outsynaptic weights of neurons (Fig. 3b). In this case, highactive neurons would receive more synaptic inputs from lowactive neurons and their dominant roles are weakened. Thus, the network synchronization decreases (Fig. 3d). These results clearly reveal that due to the complex modulation on synaptic connections, spontaneous magnetic field significantly decreases the network synchronization by inducing a more homogeneous directedweighted selforganized neuronal network.
Negative feedback of spontaneous magnetic field induces the formation of economical network structure
We now investigate the effect of spontaneous magnetic field on the topological properties of selforganized neuronal networks from the perspective of graph theory. Modularity is the most obvious property of neuronal networks^{18,63} and has great effects on the network dynamics^{64}. During the selforganization process, neurons with similar firing properties are more likely to be organized into a module through competition, such that neurons in the same module have relatively higher synaptic weights than those in different modules. This phenomenon ensures more efficient communication of neural information within modules^{63,65}. The modularity (defined in Eq. 9) measures the division of a network into modules, where networks with high modularity have denser modules^{18,63}. We observe that as the feedback strength increases, the modularity decreases (Fig. 4a), indicating that there are sparse modules in the selforganized neuronal network. Thus, spontaneous magnetic field can decrease the modularity of the network structure by inducing the formation of a more homogeneous directed selforganized neuronal network.
However, synaptic weights between modules should also be enough strong to support the global integration of neural information, which will inevitably wipe away the modular structure^{63,65}. The most economical structure should have a moderate degree of modularity, corresponding to a relatively high efficiency of neural information flow within and between modules. We observe that as the negative feedback strength initially increases (e.g., k_{1} ≤ 0.3), the local efficiency (see Eq. 10) temporarily decreases (Fig. 4b), indicating that the selforganized neuronal network has less ability to transfer neural information in local circuits due to the increased percentage of weak synapses induced by the negative feedback from magnetic fields (i.e., increased P_{0}, Fig. 3a). As the negative feedback strength further increases, the mean synaptic weight increases and the percentage of modulated synapses remains at a high value (i.e. low P_{2}, see Fig. 3a), accompanied by an increase in local efficiency, reflecting the strengthening and relative optimization of synaptic connections such that neurons can more easily cluster together to transfer neural information. For higher negative feedback strengths, the percentage of modulated synapses decreases with increased P_{2} (Fig. 3a), and the local efficiency is again reduced to a small value, suggesting that less reconfigured synaptic connections limit the transmission of neural information in the local scale.
Considering the complex modulation of synaptic connections, the negative feedback from magnetic fields will first reduce the synaptic weights (especially for weak synapses) and then increase the number of synapses with less modulation as k_{1} increases. In a sparse selforganized network, the combination of the decreased synaptic weight and number of synapses will definitely result in fewer and weaker synaptic paths for the transmission of neural information, resulting in a temporary decrease in the transmission of neural information in the global scale (see Fig. 4c for k_{1} < 0.3). As the negative feedback strength increases, the percentage of modulated synapses decreases, accompanied by a larger number of synapses and higher mean synaptic weight that can provide more paths to transfer neural information. Specifically, the percentage of weak synapses decreases (i.e. decreased P_{0}, Fig. 3a), suggesting more paths from lowactive neurons to highactive neurons. Consequently, the global efficiency (see Eq. 11) begins to increase (Fig. 4c). Meanwhile, the synaptic weight begins to recover but is still smaller than that for k_{1} = 0.0, whereas the global efficiency is larger than that for k_{1} = 0.0. These results reflect that selforganized neuronal networks under spontaneous magnetic field may have higher ability to integrate neural information throughout the whole network.
In addition, it is important to notice the anomalous phenomenon that for k_{1} = 0.9–1.1, both the local efficiency and global efficiency are significantly larger than those for k_{1} = 0.0 (Fig. 4b,c), reflecting a more economical neuronal network structure from the perspective of graph theory. From the joint probability distribution of the local efficiency and global efficiency in dynamically selforganized networks, we can identify the most economical network structure, with the highest local and global efficiency (i.e., k_{1} = 1.1; see Fig. 5), for which the percentage of modulated synapses and the mean synaptic weight are significant smaller than those for k_{1} = 0.0. These results indicate that spontaneous magnetic field can optimally reconfigure synaptic connections to promote the formation of an economical neuronal network structure via the selforganization process and produce the most economical structure with a negative feedback strength of k_{1} = 1.1.
Spontaneous magnetic coupling enhances the selforganization process
A spontaneously formed magnetic field not only exerts negative feedback on its corresponding neuron but also can interact with the magnetic fields of other neurons, thereby physically providing a spatial channel for the transfer of neural information^{45,53}. Previous studies have found that the spatial coupling between magnetic fields has a significant influence on the collective dynamics of neurons^{45,53,56}. It is thus interesting to investigate how the spatial coupling between magnetic fields modulates the selforganization process and the topological properties of the neuronal network.
By fixing the negative feedback strength k_{1} = 1.1 and varying the coupling strength D, we first analyze how this magnetic coupling modulates the selforganization process. As D increases, the transition time T first increases and then decreases (Fig. 6a), and this phenomenon is robust for different artificial fluctuation coefficients f, reflecting the complex modulation exerted by magnetic coupling on the selforganization speed. By promoting the communication of neural information among neurons^{49,66}, the magnetic coupling, on the one hand, enhances the selforganization process, and on the other hand, induces higher negative feedback on neurons and thus contrarily slows down the selforganization process. These competing influences on the selforganization process result in the complex modulation of the selforganization process by the magnetic coupling (Fig. 6a). However, it is important to note that the transition time T under electromagnetic induction is significantly larger than that without magnetic feedback (7.995, 5.265 and 3.973 for f = 0.1, 0.15 and 0.2, respectively; see Fig. 2). Thus, although the magnetic coupling between neurons contributes to the selforganization process, the negative feedback exerted by magnetic fields on neurons significantly decreases the neuronal excitability and eventually inhibits the selforganization process. These results demonstrate that spontaneous electromagnetic induction slows down the selforganization process of the neuronal network.
Spontaneous magnetic coupling contributes to a homogeneous directed selforganized neuronal network
We now investigate how the spontaneous magnetic coupling affects the synaptic connections in the directed selforganized neuronal network. As the coupling strength D increases, the mean P_{0} first increases and then decreases (Fig. 6b), which results in a temporal decrease of mean synaptic weight (Fig. 6c). However, the mean P_{1} always increases, which would enhance the mean synaptic weight. Indeed, the magnetic coupling between neurons contributes to the transmission of neural information^{49,52,56,66,67} and thus can promote the modulation of synaptic connections between neurons (i.e. low P_{2}, see Fig. 6b), which decreases the mean synaptic weight. However, the enhanced negative feedback on neurons induced by magnetic coupling would also decrease the neuronal excitability, and consequently, the modulation is inhibited with increased P_{2} (Fig. 6b), which conversely increases the synaptic weights. This complex modulation on synapses by spontaneous magnetic coupling would result in a Ushaped curve describing the mean synaptic weight as the coupling strength D increases (Fig. 6c).
Despite the complex modulation of synaptic weights caused by magnetic fields, we observe a continuous decrease of causal flow as the coupling strength D increases (Fig. 7a), indicating that the negative feedback weakens the causal relationships between neurons, consistent with the previous results in Fig. 3b. Thus, the causal relationships between neurons are mainly related to neuronal excitability (i.e., controlled by negative magnetic feedback). Meanwhile, the decreased causal flow reflects the less difference between in and outsynaptic weights, which results in a lower network synchronization (Fig. 6d). Considering the increased percentage of strong synapses and the decreased percentage of weak synapses (Fig. 6b), it is thus reasonable to suspect that, by decreasing the neuronal excitability, spontaneous magnetic coupling may enhance the synaptic connections from lowactive neurons to highactive neurons such that the number of strong synapses is improved but the network synchronization decreases. These results indicate that spontaneous electromagnetic induction results in a more homogeneous directedweighted selforganized neuronal network structure with less causal relationship and weaker synchronization.
Spontaneous magnetic coupling promotes the formation of economical neuronal network structure
We now investigate how the spontaneous magnetic coupling affects the topological properties of neuronal networks as formed through the selforganization process. The weakening of the causal relationship between neurons results in a less heterogeneous directed network structure in terms of the synaptic weights, which leads to decrease in network modularity (Fig. 7b). As a result, the modularity of selforganized neuronal networks further decreases for stronger magnetic coupling and is significantly smaller than that for k_{1} = 0.0 (Fig. 7b). Meanwhile, the number of strong synapses are increasing as the coupling strength D is enhanced (i.e. increased P_{1}, Fig. 6b), which provides more strong paths to transfer neural information and results in a continuously increase in global efficiency in the directedweighted neuronal network (Fig. 7d). Furthermore, we observe that the local efficiency first decreases and then increases as the coupling strength D increases (Fig. 7c), similar to the mean synaptic weight (Fig. 6c). Due to the increased percentage of weak synaptic weights induced by spontaneous magnetic coupling (i.e. increased P_{0}, Fig. 6b), the selforganized neuronal network has less ability to transfer neural information in local circuits, which results in a temporary decrease in local efficiency (Fig. 7c), similar to the results in Fig. 6c. However, as the coupling strength further increases, the number of weak synapses decreases but the number of strong synapses increases, accompanied by a increases in local efficiency, indicating the reconfiguration of synaptic connections. Compared to the network efficiency without magnetic fields (i.e., k_{1} = 0.0), the magnetic coupling always produces a higher global efficiency (Fig. 7d) and can also induce a higher local efficiency at suitable coupling strengths. These results reveal that spontaneous magnetic coupling between neurons promotes the formation of an economical neuronal network structure by reconfiguring the synaptic connections.
Discussion
In this paper, we investigate how the brain neuronal network selforganizes into an economical structure under modulation by spontaneous electromagnetic induction. We first show that negative feedback from spontaneous magnetic fields slows down the selforganization process. Then, for suitable negative feedbacks, magnetic fields induce weaker firing synchronization by homogenizing the directed synaptic connections, where the connectivity patterns of selforganized neuronal networks have lower modularity, higher local and global efficiency. Finally, with suitable magnetic couplings, spontaneous magnetic fields can further homogenize the synaptic connections, decrease the modularity and enhance the local and glocal efficiency. Our results provide a comprehensive understanding of the significant effects of spontaneous electromagnetic induction on the selforganization process and the resulting modulation of the formation of an economical neuronal network.
The macroscopic dynamical behaviors in the neuronal network promote the formation of the selforganized structure due to the synaptic plasticity, which in turn produces different network dynamics. Thus, the topological properties of selforganized neuronal networks are closely related to the global dynamics, i.e. network synchronization. The modulation of chemical synapses by STDP is controlled by the firing time lag between the pre and postsynaptic neurons^{14}, i.e., the synchronization between neurons (see Eq. 7), and thus, the selforganization process will be faster with more rapid global synchronization. The negative feedback of the magnetic field inhibits the neuronal excitability so as to decrease the network synchronization^{47,68}, it is thus reasonable that spontaneous electromagnetic induction would slow down the selforganization process. Meanwhile, the magnetic coupling between neurons can contribute to the signal exchange between neurons and promote the global synchronization of the neuronal network with a constant structure^{45,47,51,52,66,69,70}, and thus could be another effective way for signal propagation in the network. But the increase in the magnetic coupling also produces higher negative feedback on neurons to inhibit the synchronization, ultimately inhibiting the selforganization process. Therefore, spontaneous electromagnetic induction slows down the selforganization process with weaker neuronal synchronization.
As a result of the slowly selforganization process under spontaneous electromagnetic induction, the neuronal network can contradictorily achieve high efficiency of information transmission on both the local and global scales from the perspective of graph theory. Indeed, the complex collective dynamics are formed from the connections between neurons and the topological structure of the neuronal network is crucial to effective communication of neural information^{16,71,72}. Spontaneous electromagnetic induction weakens the update of synapses by decreasing the neuronal excitation, which would create more paths for neural information transmission in selforganized neuronal networks and produce high local and global efficiency. This modulation also generates higher loop numbers among neurons, supporting weaker global synchronization^{73}. Note that the economical selforganized network provides a potentially structural ability to support the efficient information communication with low energy cost, and the structural economy is not the same to the economical cost of biophysical or biological energy consumed by neurons and synapses, but they have complex relationships^{36,74,75,76,77,78,79,80}. Our results show an economical selforganized neuronal network structure induced by spontaneous electromagnetic induction, which intrinsically supports the weaker neuronal synchronization to transfer less information and may provide a structural foundation for the economic energy consumption of neurons and synapses. Finally, it needs to note that the precise timing of pre and postsynaptic spikes determines the outcome of STDP rule, and the propagation delay of neural signals crucially affects the connectivity patterns and dynamics of selforganized neuronal networks^{21,22}. Thus, the propagation delay would affect the results presented in this paper, which needs to be further investigated.
Models and Methods
Neuronal network model
We used the FitzHughNagumo model to construct the neuronal network^{18,81,82}:
where V_{i} is the membrane potential of the i th neuron and W_{i} is the corresponding membrane recovery variable. ε, a and b_{i} are dimensionless parameters, and I_{ext} is the external applied current. \({{I}_{i}}^{syn}\) is the total chemical synaptic current received by the i th neuron and is described as:
Here, g_{ij} is a measure of the synaptic weight from neuron j to neuron i and is modulated by the neuronal electric activities through the STDP rule. The synaptic variable s_{j} is controlled by the synaptic recovery and decay processes, i.e., α(V_{j})(1 − s_{j}) and −βs_{j}, where α(V_{j}) is the synaptic recovery function and β is the decay rate. While the presynaptic membrane potential is below the threshold (i.e., V_{j} < 0 mV), the chemical synapse cannot be activated and subsequently decays with a rate of β, and thus, s_{j} can be reduced to ds_{j}/dt = −βs_{j}. Otherwise, the chemical synapse is strongly activated to act on the postsynaptic neuron, and s_{j} quickly jumps to 1. In addition, V_{syn} is the reversal potential of the chemical synapse and is related to the neurotransmitters and receptors; its value is set to 0 mV for excitatory synapses and to 2 mV for inhibitory synapses^{16,18}.
Model of spontaneous electromagnetic induction
The transmembrane transport of ions in neurons inevitably induces a temporally variable electromagnetic field^{75,83}, which conversely exerts negative feedback on the neuronal electric activities^{34,45}. A memristor model can be used to describe the physical correspondence between magnetic flux and electric charge and has recently been introduced to describe the coupling between a magnetic field and the membrane potential of a neuron^{34,43,69}. The feedback current \({{I}_{i}}^{mag}\) induced by the magnetic field on the membrane potential is described as^{34,69}:
where the negative sign reflects the inhibitory effect of the magnetic field on the neuronal membrane potential and k_{1} represents the feedback strength. ρ(ϕ_{i}) is the memory conductance of the memristor and is often chosen to be ρ(ϕ) = c + 3dϕ^{2}, with constant parameters c and d. ϕ_{i} is a state variable of the magnetic field and satisfies:
This equation illustrates that on the one hand, the magnetic field is driven by the electric activity of the neuron with driving strength k_{3}, and on the other hand, it decays with a rate of k_{2} in the assumed homogeneous medium. In the neural system, spontaneous electricmagnetic fields are interacted among neurons, which provide a hidden spatial channel for the communication of neural information^{45,47,52}. Considering the magnetic coupling between neurons, Eq. 4 can be rewritten as:
Here, D represents the strength of the magnetic field coupling between neurons and f(ϕ_{j}, ϕ_{i}) describes the form of the coupling^{70}. The magnetic coupling depends on many factors, such as the density of the medium and the spatial distance between neurons^{45,47}. Here, we focus on the collective dynamics in a local neuronal circuit and thus assume a simple linear diffusive magnetic field coupling between neurons, i.e., f(ϕ_{j}, ϕ_{i}) = ϕ_{j} − ϕ_{i}^{47}.
Spiketimingdependent plasticity
When the presynaptic neuron i fires at time t_{i} and the postsynaptic neuron j fires at t_{j}, the updating of the synaptic weight through the STDP rule is described as^{16,17}:
with the updating function:
where Δt = t_{i} − t_{j} is the firing time lag, τ_{+} and τ_{−} are the temporal windows for synaptic refinement, and A_{+} and A_{−} determine the maximum magnitude of the synaptic update. The firing time lag Δt is measured within the temporal windows, and the synaptic update is performed once the temporal windows have passed. In addition, the synaptic weights are restricted to the range [0, g_{max}] to ensure the convergence of the STDP algorithm.
The values of the remaining parameters utilized in the above models are ε = 0.08, I_{ext} = 0.1, a = 0.7, α_{0} = 2, β = 1, V_{shp} = 0.05, c = 0.1, d = 0.02, A_{+} = 0.05, A_{−} = 0.0525, τ_{+} = τ_{−} = 2, and g_{max} = 0.1. The parameter b decides the degree of excitability of neurons^{16}. A small value of b corresponds to high neuronal excitability; thus, these neurons are highly dominant in the selforganization process and have a high outsynaptic strength^{16}. To promote the competition among neurons, we define b to be uniformly distributed in [0.25, 0.95] to introduce heterogeneity in the neuronal network. The variable parameters k_{1}, k_{2} and k_{3} control the negative magnetic feedback and can induce significantly different and complex dynamic patterns in neurons, such as instabilities and multiple frequencies^{68}. Larger k_{1} and k_{3} values or smaller k_{2} values, in principle, all result in stronger negative feedback on neurons. Without considering the complex modulation of neuronal dynamics induced by spontaneous electromagnetic induction, we set k_{2} = 1 and k_{3} = 1 and vary k_{1} to control the negative feedback strength.
Graph theory measures
Causal flow
In a directed weighted network, the causal flow of a node measures its causal influence on another node^{61} and is defined as:
Here, \({{k}_{i}}^{out}={\sum }_{j\in N}\,{w}_{ij}\) is the outdegree of the node, and \({{k}_{i}}^{in}={\sum }_{j\in N}\,{w}_{ji}\) is the corresponding indegree, where N is the set of all nodes and w_{ij} is the connection weight from node i to node j.
Modularity
The modularity is a measure of the extent to which a network is divided into modules, such that the nodes within a module have dense connections but nodes in different modules have sparse connections. In a directed weighted network, the modularity is described as^{84}:
where \(l={\sum }_{i,j\in N}\,{w}_{ij}\) is the sum of all connection weights in the network, \({\delta }_{{m}_{i},{m}_{j}}\) is the Kronecker delta function, and m_{i} is the module containing node i. If nodes i and j are in the same module, \({\delta }_{{m}_{i},{m}_{j}}=1\); otherwise, \({\delta }_{{m}_{i},{m}_{j}}=0\).
Network efficiency
The local efficiency of a node is a measure of the information exchange among its neighbors; in a directed weighted network, this measure is defined as^{84}:
where n is the number of nodes and d_{jh}(N_{i}) is the shortest path from node j to node h that contains node i. The information exchange throughout the whole network is quantified by the global efficiency^{84}:
Here, all graph theory measures are calculated with the Brain Connectivity Toolbox (BCT, https://www.nitrc.org/projects/bct) in MATLAB, with no additional processing.
Neuronal synchronization
The network synchronization is measured by a statistical factor of synchronization which is defined as follows^{45,49}:
with
Here, N is the number of neurons, V_{i} is the time series for neuron i in Eq. 1 and < >is the average over time. The R is in the range [0, 1]. R = 1 corresponds to the completely synchronous state, and the lower value reflects weaker network synchronization.
References
 1.
Bassett, D. S. & Bullmore, E. T. Smallworld brain networks revisited. Neurosci. 23, 499–516 (2017).
 2.
Liao, X., Vasilakos, A. V. & He, Y. Smallworld human brain networks: perspectives and challenges. Neurosci. Biobehav. Rev. 77, 286–300 (2017).
 3.
Wang, R., Tsuda, I. & Zhang, Z. A new work mechanism on neuronal activity. Int. J. Neural Syst. 25, 1450037 (2015).
 4.
Laughlin, S. B. & Sejnowski, T. J. Communication in neuronal networks. Sci. 301, 1870–1874 (2003).
 5.
Wang, Y., Wang, R. & Zhu, Y. Optimal pathfinding through mental exploration based on neural energy field gradients. Cogn. Neurodyn. 11, 99–111 (2017).
 6.
Zhu, F., Wang, R., Pan, X. & Zhu, Z. Energy expenditure computation of a single bursting neuron. Cogn. Neurodyn. 13, 75–87 (2019).
 7.
Qu, J., Wang, R., Yan, C. & Du, Y. Spatiotemporal behavior of smallworld neuronal networks using a mapbased model. Neural Process. Lett. 45, 689–701 (2017).
 8.
Qu, J., Wang, R., Yan, C. & Du, Y. Oscillations and synchrony in a cortical neural network. Cogn. Neurodyn. 8, 157–166 (2014).
 9.
D’amour, J. A. & Froemke, R. C. Inhibitory and excitatory spiketimingdependent plasticity in the auditory cortex. Neuron 86, 514–528 (2015).
 10.
Carcea, I. & Froemke, R. C. Cortical plasticity, excitatory–inhibitory balance, and sensory perception. Prog. Brain Res. 207, 65–90 (2013).
 11.
Alvarez, V. A. & Sabatini, B. L. Anatomical and physiological plasticity of dendritic spines. Annu. Rev. Neurosci. 30, 79–97 (2007).
 12.
Bullmore, E. & Sporns, O. Complex brain networks: graph theoretical analysis of structural and functional systems. Nat. Rev. Neurosci. 10, 186–198 (2009).
 13.
Daoudal, G. & Debanne, D. Longterm plasticity of intrinsic excitability: learning rules and mechanisms. Learn. Mem. 10, 456–65 (2003).
 14.
Takeuchi, T., Duszkiewicz, A. J. & Morris, R. G. The synaptic plasticity and memory hypothesis: encoding, storage and persistence. Phil. Trans. R. Soc. B 369, 20130288 (2014).
 15.
Song, S., Miller, K. D. & Abbott, L. F. Competitive hebbian learning through spiketimingdependent synaptic plasticity. Nat. Neurosci. 3, 919–926 (2000).
 16.
Li, X., Zhang, J. & Small, M. Selforganization of a neural network with heterogeneous neurons enhances coherence and stochastic resonance. Chaos 19, 013126 (2009).
 17.
Li, X. & Small, M. Neuronal avalanches of a selforganized neural network with activeneurondominant structure. Chaos 22, 023104 (2012).
 18.
Wang, R., Wu, Y., Wang, L., Du, M. & Li, J. Structure and dynamics of selforganized neuronal network with an improved stdp rule. Nonlinear Dyn. 88, 1855–1868 (2017).
 19.
Prezioso, M., Merrikh Bayat, F., Hoskins, B., Likharev, K. & Strukov, D. Selfadaptive spiketimedependent plasticity of metaloxide memristors. Sci. Rep. 6, 21331 (2016).
 20.
Kim, S.Y. & Lim, W. Effect of spiketimingdependent plasticity on stochastic burst synchronization in a scalefree neuronal network. Cogn. Neurodyn. 12, 315–342 (2018).
 21.
Madadi Asl, M., Valizadeh, A. & Tass, P. A. Dendritic and axonal propagation delays determine emergent structures of neuronal networks with plastic synapses. Sci. Rep. 7, 39682 (2017).
 22.
Madadi Asl, M., Valizadeh, A. & Tass, P. A. Delayinduced multistability and loop formation in neuronal networks with spiketimingdependent plasticity. Sci. Rep. 8, 12068 (2018).
 23.
Markram, H. Regulation of synaptic efficacy by coincidence of postsynaptic aps and epsps. Sci. 275, 213–215 (1997).
 24.
Debanne, D., Gähwiler, B. H. & Thompson, S. M. Longterm synaptic plasticity between pairs of individual ca3 pyramidal cells in rat hippocampal slice cultures. J. Physiol. 507, 237–247 (2010).
 25.
Haas, J. S., Nowotny, T. & Abarbanel, H. D. Spiketimingdependent plasticity of inhibitory synapses in the entorhinal cortex. J. Neurophysiol. 96, 3305–3313 (2006).
 26.
Froemke, R. C. & Dan, Y. Spiketimingdependent synaptic modification induced by natural spike trains. Nat. 416, 433 (2002).
 27.
Nishimura, Y., Perlmutter, S. I., Eaton, R. W. & Fetz, E. E. Spiketimingdependent plasticity in primate corticospinal connections induced during free behavior. Neuron 80, 1301–1309 (2013).
 28.
Huang, S., Huganir, R. L. & Kirkwood, A. Adrenergic gating of hebbian spiketimingdependent plasticity in cortical interneurons. J. Neurosci. 33, 13171 (2013).
 29.
Gerstner, W., Kempter, R., van Hemmen, J. L. & Wagner, H. A neuronal learning rule for submillisecond temporal coding. Nat. 383, 76–78 (1996).
 30.
Bi, G.q. & Poo, M.m. Synaptic modifications in cultured hippocampal neurons: Dependence on spike timing, synaptic strength, and postsynaptic cell type. J.Neurosci. 18, 10464–10472 (1998).
 31.
Subhani, A. R. et al. Mitigation of stress: new treatment alternatives. Cogn. Neurodyn. 12, 1–20 (2018).
 32.
Markram, H., Gerstner, W. & Sjöström, P. J. A history of spiketimingdependent plasticity. Front. Synaptic Neurosci. 3, 4 (2011).
 33.
Kim, S. J. & Linden, D. J. Ubiquitous plasticity and memory storage. Neuron 56, 582–592 (2007).
 34.
Lv, M., Wang, C., Ren, G., Ma, J. & Song, X. Model of electrical activity in a neuron under magnetic flow effect. Nonlinear Dyn. 85, 1479–1490 (2016).
 35.
Ma, J. & Tang, J. A review for dynamics of collective behaviors of network of neurons. Sci. China Technol. Sc. 58, 2038–2045 (2015).
 36.
Li, J., Liu, S., Liu, W., Yu, Y. & Wu, Y. Suppression of firing activities in neuron and neurons of network induced by electromagnetic radiation. Nonlinear Dyn. 83, 801–810 (2016).
 37.
Deli, E., Tozzi, A. & Peters, J. F. Relationships between short and fast brain timescales. Cogn. Neurodyn. 11, 539–552 (2017).
 38.
Wang, Y., Ma, J., Xu, Y., Wu, F. & Zhou, P. The electrical activity of neurons subject to electromagnetic induction and gaussian white noise. Int. J. Bifurcat. Chaos 27, 1750030 (2017).
 39.
Zhan, F. & Liu, S. Response of electrical activity in an improved neuron model under electromagnetic radiation and noise. Front. Comput. Neurosci. 11, 107 (2017).
 40.
Wu, F., Wang, C., Jin, W. & Ma, J. Dynamical responses in a new neuron model subjected to electromagnetic induction and phase noise. Phys. A 469, 81–88 (2017).
 41.
Zhang, X. & Liu, S. Nonlinear delayed feedback control of synchronization in an excitatory–inhibitory coupled neuronal network. Nonlinear Dyn. (2019).
 42.
Wang, J., Lu, B., Liu, S. & Jiang, X. Bursting types and bifurcation analysis in the prebötzinger complex respiratory rhythm neuron. Int. J. Bifurcat. Chaos 27, 1750010 (2017).
 43.
Feng, P., Wu, Y. & Zhang, J. A route to chaotic behavior of single neuron exposed to external electromagnetic radiation. Front. Comput. Neurosci. 11, 94 (2017).
 44.
Rostami, Z. & Jafari, S. Defects formation and spiral waves in a network of neurons in presence of electromagnetic induction. Cogn. Neurodyn. 12, 235–254 (2018).
 45.
Xu, Y., Jia, Y., Ma, J., Hayat, T. & Alsaedi, A. Collective responses in electrical activities of neurons under field coupling. Sci. Rep. 8, 1349 (2018).
 46.
Zhao, Y., Sun, X., Liu, Y. & Kurths, J. Phase synchronization dynamics of coupled neurons with coupling phase in the electromagnetic field. Nonlinear Dyn. 93, 1315–1324 (2018).
 47.
Guo, S. et al. Collective response, synapse coupling and field coupling in neuronal network. Chaos Soliton. Fract. 105, 120–127 (2017).
 48.
Ma, J., Wu, F. & Wang, C. Synchronization behaviors of coupled neurons under electromagnetic radiation. Int. J. Mod. Phys. B 31, 1650251 (2017).
 49.
Xu, Y., Jia, Y., Ma, J., Alsaedi, A. & Ahmad, B. Synchronization between neurons coupled by memristor. Chaos Soliton. Fract. 104, 435–442 (2017).
 50.
Lu, L., Jia, Y., Liu, W. & Yang, L. Mixed stimulusinduced mode selection in neural activity driven by high and low frequency current under electromagnetic radiation. Complex. 2017, 1–11 (2017).
 51.
Mvogo, A., Takembo, C. N., Ekobena Fouda, H. P. & Kofané, T. C. Pattern formation in diffusive excitable systems under magnetic flow effects. Phys. Lett. A 381, 2264–2271 (2017).
 52.
Ren, G., Xu, Y. & Wang, C. Synchronization behavior of coupled neuron circuits composed of memristors. Nonlinear Dyn. 88, 893–901 (2017).
 53.
Hu, X., Liu, C., Liu, L., Ni, J. & Yao, Y. Chaotic dynamics in a neural network under electromagnetic radiation. Nonlinear Dyn. 91, 1541–1554 (2017).
 54.
Xu, Y. et al. Effects of ion channel blocks on electrical activity of stochastic hodgkin–huxley neural network under electromagnetic induction. Neurocomputing 283, 196–204 (2017).
 55.
XiaoHan, Z. & ShenQuan, L. Stochastic resonance and synchronization behaviors of excitatoryinhibitory smallworld network subjected to electromagnetic induction. Chin. Phys. B 27, 040501 (2018).
 56.
Ge, M., Jia, Y., Xu, Y. & Yang, L. Mode transition in electrical activities of neuron driven by high and low frequency stimulus in the presence of electromagnetic induction and radiation. Nonlinear Dyn. 91, 515–523 (2017).
 57.
Wu, J., Xu, Y. & Ma, J. Levy noise improves the electrical activity in a neuron under electromagnetic radiation. PLoS ONE 12, e0174330 (2017).
 58.
Rostami, Z., Jafari, S., Perc, M. & Slavinec, M. Elimination of spiral waves in excitable media by magnetic induction. Nonlinear Dyn. 94, 679–692 (2018).
 59.
Ma, J., Wu, F., Hayat, T., Zhou, P. & Tang, J. Electromagnetic induction and radiationinduced abnormality of wave propagation in excitable media. Phys. A 486, 508–516 (2017).
 60.
Fu, Y. X., Kang, Y. M. & Xie, Y. Subcritical hopf bifurcation and stochastic resonance of electrical activities in neuron under electromagnetic induction. Front. Comput. Neurosci. 12, 6 (2018).
 61.
Kong, W., Lin, W., Babiloni, F., Hu, S. & Borghini, G. Investigating driver fatigue versus alertness using the granger causality network. Sensors 15, 19181–19198 (2015).
 62.
Bayati, M., Valizadeh, A., Abbassian, A. & Cheng, S. Selforganization of synchronous activity propagation in neuronal networks driven by local excitation. Front. Comput. Neurosc. 9, 69 (2015).
 63.
Wang, R. et al. Hierarchical connectome modes and critical state jointly maximize human brain functional diversity. Phys. Rev. Lett. (2019).
 64.
Jalili, M. Enhancing synchronizability of diffusively coupled dynamical networks: A survey. IEEE T. Neural Net. Lear. 24, 1009–1022 (2013).
 65.
ZamoraLopez, G., Chen, Y., Deco, G., Kringelbach, M. L. & Zhou, C. Functional complexity emerging from anatomical constraints in the brain: the significance of network modularity and richclubs. Sci. Rep. 6, 38424 (2016).
 66.
Ma, J., Mi, L., Zhou, P., Xu, Y. & Hayat, T. Phase synchronization between two neurons induced by coupling of electromagnetic field. Appl. Math. Comput. 307, 321–328 (2017).
 67.
Kim, S. Y. & Lim, W. Dynamical responses to external stimuli for both cases of excitatory and inhibitory synchronization in a complex neuronal network. Cogn. Neurodyn. 11, 395–413 (2017).
 68.
Wang, R., Feng, P., Fan, Y. & Wu, Y. Spontaneous electromagnetic induction modulating neuronal dynamical response. Int. J. Bifurcat. Chaos 29, 1950005 (2019).
 69.
Ma, J. & Tang, J. A review for dynamics in neuron and neuronal network. Nonlinear Dyn. 89, 1569–1578 (2017).
 70.
Qin, H., Wang, C., Cai, N., An, X. & Alzahrani, F. Field couplinginduced pattern formation in twolayer neuronal network. Phys. A 501, 141–152 (2018).
 71.
Sun, X. J., Lei, J. Z., Perc, M., Lu, Q. S. & Lv, S. J. Effects of channel noise on firing coherence of smallworld hodgkinhuxley neuronal networks. Eur. Phys. J. B 79, 61–66 (2011).
 72.
Ma, J., Wu, Y., Ying, H. P. & Jia, Y. Channel noiseinduced phase transition of spiral wave in networks of hodgkinhuxley neurons. Chin. Sci. Bull. 56, 151–157 (2011).
 73.
Sheshbolouki, A., Zarei, M. & SarbaziAzad, H. Are feedback loops destructive to synchronization? EPLEurophys. Lett. 111, 40010 (2015).
 74.
Karim, E. L. & Martin, B. Synaptic energy drives the information processing mechanisms in spiking neural networks. Math. Biosci. Eng. 11, 233–256 (2017).
 75.
Wang, Z. & Wang, R. Energy distribution property and energy coding of a structural neural network. Front. Comput. Neurosci. 8, 14 (2014).
 76.
Wang, R., Wang, Z. & Zhu, Z. The essence of neuronal activity from the consistency of two different neuron models. Nonlinear Dyn. 92, 973–982 (2018).
 77.
Wang, R. & Zhu, Y. Can the activities of the large scale cortical network be expressed by neural energy? a brief review. Cogn. Neurodyn. 10, 1–5 (2016).
 78.
Zheng, H., Wang, R., Qiao, L. & Du, Y. The molecular dynamics of neural metabolism during the action potential. Sci. China Technol. Sc. 57, 857–863 (2014).
 79.
Wang, Y., Wang, R. & Xu, X. Neural energy supplyconsumption properties based on hodgkinhuxley model. Neural Plas. 2017, 1–11 (2017).
 80.
Wang, G., Wang, R., Kong, W. & Zhang, J. Simulation of retinal ganglion cell response using fast independent component analysis. Cogn. Neurodyn. 12, 615–624 (2018).
 81.
Wang, Q., Zhang, H., Perc, M. & Chen, G. Multiple firing coherence resonances in excitatory and inhibitory coupled neurons. Commun. Nonlinear. Sci. Numer. Simulat. 17, 3979–3988 (2012).
 82.
Zhao, Z., Jia, B. & Gu, H. Bifurcations and enhancement of neuronal firing induced by negative feedback. Nonlinear Dyn. 86, 1549–1560 (2016).
 83.
Wang, R., Zhang, Z. & Chen, G. Energy coding and energy functions for local activities of brain. Neurocomputing 73, 139–150 (2009).
 84.
Rubinov, M. & Sporns, O. Complex network measures of brain connectivity: Uses and interpretations. NeuroImage 52, 1059–1069 (2010).
Acknowledgements
This project was partially supported by the National Natural Science Foundation of China (Nos 11802229 and 11772242) and Outstanding Youth Science Fund of Xi’an University of Science and Technology (No. 2019YQ311).
Author information
Affiliations
Contributions
R.W. and Y.W. conceived the study, R.W. and Y.F. programmed the numerical simulations, R.W. wrote the manuscript, and all authors reviewed the manuscript.
Corresponding author
Ethics declarations
Competing Interests
The authors declare no competing interests.
Additional information
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wang, R., Fan, Y. & Wu, Y. Spontaneous electromagnetic induction promotes the formation of economical neuronal network structure via selforganization process. Sci Rep 9, 9698 (2019). https://doi.org/10.1038/s4159801946104z
Received:
Accepted:
Published:
Further reading

A Geometric Milieu Inside the Brain
Foundations of Science (2021)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.