王榮 成昕鴻 周峰
摘 要:基于FitzHughNagumo神經(jīng)元模型,構(gòu)建了電磁輻射環(huán)境下突觸連接動(dòng)態(tài)變化的自組織神經(jīng)元網(wǎng)絡(luò)模型;采用神經(jīng)動(dòng)力學(xué)理論和復(fù)雜網(wǎng)絡(luò)方法,研究了不同電磁輻射強(qiáng)度下,自組織神經(jīng)元網(wǎng)絡(luò)的連接密度、因果關(guān)系、模塊化程度、網(wǎng)絡(luò)效率、同步行為以及興奮性。發(fā)現(xiàn)電磁輻射增強(qiáng)了磁場(chǎng)對(duì)神經(jīng)元的負(fù)反饋,減弱了神經(jīng)元之間的競(jìng)爭(zhēng),降低了神經(jīng)元之間的因果關(guān)系;但對(duì)神經(jīng)元網(wǎng)絡(luò)的連接密度、模塊化程度和網(wǎng)絡(luò)效率的影響較復(fù)雜;存在最佳輻射強(qiáng)度,使自組織神經(jīng)元網(wǎng)絡(luò)的局部效率和全局效率顯著升高。同時(shí),電磁輻射增強(qiáng)了自組織神經(jīng)元網(wǎng)絡(luò)固有的同步能力,產(chǎn)生更高的神經(jīng)元同步放電活動(dòng)。最后,合適的電磁輻射強(qiáng)度可以增強(qiáng)自組織經(jīng)元網(wǎng)絡(luò)的興奮性,而更高的輻射強(qiáng)度降低了神經(jīng)元網(wǎng)絡(luò)的興奮性。結(jié)果表明:電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)的形成具有復(fù)雜的調(diào)控作用,但存在合適的輻射強(qiáng)度,使自組織神經(jīng)元網(wǎng)絡(luò)在結(jié)構(gòu)上具有較強(qiáng)的信息傳遞能力,在動(dòng)力學(xué)上具有較高的同步放電行為。關(guān)鍵詞:電磁輻射;同步;自組織網(wǎng)絡(luò);復(fù)雜網(wǎng)絡(luò)中圖分類號(hào):N 93
文獻(xiàn)標(biāo)志碼:A
文章編號(hào):1672-9315(2021)02-0375-09
DOI:10.13800/j.cnki.xakjdxxb.2021.0223開放科學(xué)(資源服務(wù))標(biāo)識(shí)碼(OSID):
Topological properties and dynamics of selforganized
neuronal networks under electromagnetic radiation
WANG Rong1,CHENG Xinhong1,ZHOU Feng2
(1.College ofSciences,Xian University of Science and Technology,Xian 710054,China;
2.College of Geomatics,Xian University of Science and Technology,Xian 710054,China)
Abstract:Based on the FitzHughNagumo neuronal model,the neuronal network with dynamically variable synapses under electromagnetic radiation environment was constructed.The neurodynamics theory and complex network method were used to study the connection density,causal relationship,modularity,network efficiencies,synchronization and excitability of selforganized neuronal networks under different electromagnetic radiation strengths.It is pointed that the electromagnetic radiation enhances the negative feedback of magnetic field on neurons,weakens the competition among neurons and reduces the causal relationship between them,but it has complex effects on the connection density,modularity and network efficiencies.Specifically,an optimal strength of electromagnetic radiation produces a higher local efficiency and global efficiency in selforganized neuronal networks.In addition,the electromagnetic radiation improves the intrinsic synchronizability of selforganized neuronal networks which supports higher synchronized firing activities of neurons.Finally,the electromagnetic radiation with suitable strengths can increase the excitability of selforganized neuronal networks,but it has the contrary effect for higher strengths.The results indicate that the electromagnetic radiation has complex modulations on the formation of selforganized neuronal networks,but for the suitable strengths,it can induce selforganized neuronal networks to achieve higher information transferring efficiency in structure and stronger synchronized firing activities in dynamics.Key words:electromagnetic radiation;synchronization;selforganized network;complex network
0 引 言
神經(jīng)元是神經(jīng)系統(tǒng)基本的結(jié)構(gòu)與功能單位,其通過突觸連接傳遞神經(jīng)信息,構(gòu)成復(fù)雜的網(wǎng)絡(luò)結(jié)構(gòu) [1-2]?;谕挥|可塑性,外界神經(jīng)刺激會(huì)重塑突觸連接,使神經(jīng)元網(wǎng)絡(luò)結(jié)構(gòu)動(dòng)態(tài)改變以滿足大腦認(rèn)知功能的需求[3]。這種局部神經(jīng)回路與全局動(dòng)力學(xué)之間的反饋耦合構(gòu)成了大腦演化、學(xué)習(xí)和記憶的基礎(chǔ)[4]。研究表明,動(dòng)態(tài)變化的神經(jīng)元網(wǎng)絡(luò)具有自組織結(jié)構(gòu),在局部回路和全局回路都有較高的信息傳遞效率[5]。李秀敏等發(fā)現(xiàn)與全局連接和隨機(jī)連接網(wǎng)絡(luò)相比,自組織神經(jīng)元網(wǎng)絡(luò)可以產(chǎn)生更高的相干共振,隨機(jī)共振以及信息傳遞效率[6-7]。因此,研究自組織神經(jīng)元網(wǎng)絡(luò)的形成,對(duì)于理解大腦演化以及學(xué)習(xí)記憶的形成機(jī)理具有重要意義。神經(jīng)系統(tǒng)受到電磁輻射的影響,并且電磁場(chǎng)刺激在治療以及研究諸多神經(jīng)系統(tǒng)紊亂中得到廣泛應(yīng)用,如經(jīng)顱磁刺激(transcranial magnetic stimulation,TMS)可以在特定腦區(qū)產(chǎn)生磁場(chǎng),從而調(diào)節(jié)大腦神經(jīng)活動(dòng)[8]。AKIYAMA等通過實(shí)驗(yàn)發(fā)現(xiàn)電磁場(chǎng)可以誘發(fā)CA1椎體神經(jīng)元膜的超極化[9]以及大腦神經(jīng)組織中產(chǎn)生胞外電場(chǎng)[10]。李佳佳等通過數(shù)值模擬發(fā)現(xiàn)手機(jī)電磁輻射可以誘發(fā)神經(jīng)元由周期性放電轉(zhuǎn)遷到簇放電模式[2]。但是,電磁輻射對(duì)神經(jīng)元放電活動(dòng)的調(diào)節(jié)作用尚不完全清楚,尤其是通過將電磁輻射等效成施加在神經(jīng)元上的刺激電流進(jìn)行的,沒有體現(xiàn)電場(chǎng)-磁場(chǎng)耦合這一物理過程[2]。馬軍等基于麥克斯維爾電磁感應(yīng)理論,提出神經(jīng)元電磁感應(yīng)理論模型[11-12]。當(dāng)帶電離子跨膜運(yùn)輸,神經(jīng)元產(chǎn)生動(dòng)作電位時(shí),離子濃度的波動(dòng)必然會(huì)影響神經(jīng)元的放電活動(dòng),變化的電場(chǎng)又會(huì)產(chǎn)生變化的磁場(chǎng),形成電磁感應(yīng)。并且憶阻器可以描述神經(jīng)元膜電位與磁場(chǎng)之間的反饋耦合關(guān)系[11]?;谶@一理論,大量研究發(fā)現(xiàn),神經(jīng)元上自發(fā)性的電磁感應(yīng)能顯著影響神經(jīng)元的放電活動(dòng)[11,13-15],如誘發(fā)多種放電模式[16],促進(jìn)相干共振和抑制隨機(jī)共振等[14]。同時(shí),神經(jīng)元之間的磁場(chǎng)耦合為神經(jīng)信息傳遞提供了一種空間通道[13,17],進(jìn)而調(diào)控神經(jīng)元的放電活動(dòng)[13,18],如誘發(fā)神經(jīng)元同步[19-20],觸發(fā)復(fù)雜放電模式之間的轉(zhuǎn)遷[21],甚至可以抵消離子通道堵塞對(duì)神經(jīng)元集群動(dòng)力學(xué)行為的影響[22]。HU等發(fā)現(xiàn)自發(fā)性電磁感應(yīng)可以誘發(fā)神經(jīng)元產(chǎn)生多種混沌動(dòng)力學(xué)行為,并且在Pspice電路實(shí)驗(yàn)中驗(yàn)證了這一結(jié)果[23]。更重要的是,神經(jīng)元中自發(fā)性電磁感應(yīng)現(xiàn)象為研究外界電磁輻射對(duì)神經(jīng)系統(tǒng)的影響提供了合理可行的理論模型[24-27]。張曉涵等構(gòu)建了電磁輻射環(huán)境下的小世界神經(jīng)元網(wǎng)絡(luò)模型,發(fā)現(xiàn)最優(yōu)化的電磁輻射密度可以誘發(fā)隨機(jī)共振,最大化系統(tǒng)對(duì)外界弱信號(hào)的響應(yīng)[25]。馬軍等發(fā)現(xiàn)合適的電磁輻射密度可以誘發(fā)神經(jīng)系統(tǒng)中等程度的同步,而更強(qiáng)的密度可以誘發(fā)系統(tǒng)紊亂[24]。GE等發(fā)現(xiàn)高-低頻電磁輻射可以刺激神經(jīng)系統(tǒng)產(chǎn)生復(fù)雜的放電活動(dòng)[26],而WU等表明電磁輻射會(huì)誘發(fā)神經(jīng)元在靜息態(tài)、峰放電和簇放電之間轉(zhuǎn)遷,并且會(huì)調(diào)控神經(jīng)元混沌放電參數(shù)區(qū)域[27]。目前關(guān)于電磁輻射對(duì)神經(jīng)系統(tǒng)影響的研究都是假設(shè)神經(jīng)元之間的突觸連接是恒定的。然而,由于突觸可塑性,神經(jīng)元之間的突觸連接受到全局動(dòng)力學(xué)的影響。因此,構(gòu)建電磁輻射環(huán)境下突觸連接動(dòng)態(tài)變化的神經(jīng)元網(wǎng)絡(luò)模型,研究電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)的結(jié)構(gòu)與動(dòng)力學(xué)特性的影響,對(duì)于理解神經(jīng)系統(tǒng)的演化以及揭示電磁場(chǎng)刺激治療神經(jīng)系統(tǒng)紊亂的機(jī)理具有重要意義。
1 動(dòng)力學(xué)模型
1.1 神經(jīng)元網(wǎng)絡(luò)模型采用(fitz hugh nagumo,F(xiàn)HN)神經(jīng)元構(gòu)建神經(jīng)元網(wǎng)絡(luò)模型[28-30]
ε
dVidt=Vi-V3i3-Wi+Iext+Isyni+IMegi
dWidt=Vi+a-biWi
(1)
式中 Vi為第i個(gè)神經(jīng)元的膜電位,mV;Wi為對(duì)應(yīng)的膜恢復(fù)變量;ε1,a和bi為無量綱參數(shù),決定了單個(gè)神經(jīng)元的動(dòng)力學(xué)特性。Iext為背景電流;
Isyni為神經(jīng)元i接收到的總化學(xué)突觸電流,滿足如下方程
Isyni=-∑Nj(i≠j)gijsj(Vi-Vsyn)
dSjdt=α(Vj)(1-sj)-βsj
α(Vj)=α0/(1+e-Vj/Vshp)
(2)
式中 gij為從神經(jīng)元j到神經(jīng)元i的突觸權(quán)重。突觸變量sj由突觸激活和衰減過程組成,α(Vj)為突觸恢復(fù)函數(shù);β為突觸衰減速率。當(dāng)前突觸神經(jīng)元的膜電位低于突觸激活閾值,即Vj<0 mV,突觸無法激活,只能以速率β衰減;此時(shí),sj可以簡(jiǎn)化為
dsjdt=-βsj。相反,突觸可以被快速激活作用于后突觸神經(jīng)元,sj迅速上升到1,即sj=1。突觸電流還依賴于突觸類型,即興奮性突觸或者抑制性突觸,由突觸反轉(zhuǎn)電位Vsyn確定?;谝延形墨I(xiàn)[7],興奮性突觸反轉(zhuǎn)電位設(shè)為Vsyn=0 mV,抑制性突觸反轉(zhuǎn)電位設(shè)為
Vsyn=2 mV。
1.2 電磁輻射模型神經(jīng)元中帶電離子的跨膜運(yùn)輸不可避免的產(chǎn)生變化的電磁場(chǎng),其對(duì)神經(jīng)元的放電活動(dòng)具有顯著影響[11,13]。憶阻器描述了磁場(chǎng)與電場(chǎng)之間的物理對(duì)應(yīng)關(guān)系[31-32],可以用來描述神經(jīng)元上自發(fā)性磁場(chǎng)與膜電位之間的耦合關(guān)系[11-12,33]。自發(fā)性磁場(chǎng)對(duì)膜電位的反饋電流為[11-12]
IMegi=-k1ρ(i)Vi
(3)
式中 負(fù)號(hào)為自發(fā)性磁場(chǎng)對(duì)膜電位的抑制性影響;k1刻畫了反饋強(qiáng)度。ρ(i)為憶阻器的憶導(dǎo),滿足公式ρ(i)=c+3d2i,其中c和d為常數(shù)[34],c=0.1,d=0.02。i為磁場(chǎng)的狀態(tài)變量,滿足如下方程[13,18,21]
didt
=k3Vi-k2i+A
(4)
式中 A為無量綱的外界磁場(chǎng)輻射強(qiáng)度。當(dāng)A=0時(shí),神經(jīng)元的放電活動(dòng)驅(qū)動(dòng)自發(fā)性磁場(chǎng)的產(chǎn)生,驅(qū)動(dòng)強(qiáng)度為k3,而磁場(chǎng)在均勻介質(zhì)中以速率k2衰減。外界磁場(chǎng)又會(huì)驅(qū)動(dòng)神經(jīng)元自身的磁場(chǎng)發(fā)生改變,從而調(diào)控神經(jīng)元的放電行為。
1.3 突觸可塑性規(guī)則當(dāng)前突觸神經(jīng)元i在ti時(shí)刻放電,后突觸神經(jīng)元j在tj時(shí)刻放電,依賴峰間隙可塑性規(guī)則(spike timing dependent plasticity,STDP)根據(jù)突觸前后神經(jīng)元的放電時(shí)間差更新突觸權(quán)重[7,28],更新過程如下
Δgij=gijF(Δt)
F(Δt)
=
A+exp(-Δt/τ+) if Δt<0
-A-exp(Δt/τ-) if Δt>0
0if Δt=0
(5)
式中 Δt=ti-tj為放電時(shí)間差;
τ+和τ-為突觸更新的時(shí)間窗口;A+和A-決定了突觸更新的最大值。放電時(shí)間差Δt在更新窗口內(nèi)進(jìn)行測(cè)量,一旦通過時(shí)間窗口,突觸權(quán)重進(jìn)行更新。另外,為了保證STDP算法的收斂性,突觸權(quán)重限制在[0,gmax]。模型的參數(shù)選擇如下:
ε=0.08,Iext=0.1,
a=0.7,α0=2,β=1,Vshp=0.05,A+=0.05,A-=0052 5,τ+=τ-=2,gmax=0.1,k1=0.1,k2=1和k3=1。
式(1)中的b決定了神經(jīng)元的興奮性程度,更小的b值對(duì)應(yīng)更高的神經(jīng)元興奮性,這些神經(jīng)元在突觸更新過程中占據(jù)主導(dǎo)地位[7]。為了促進(jìn)神經(jīng)元之間的競(jìng)爭(zhēng),b均勻分布在[0.35,065]。
2 數(shù)值模擬結(jié)果與討論數(shù)值模擬采用EulerMaruyama算法,時(shí)間步長(zhǎng)
0.05 ms,總模擬時(shí)間200 ms。初始時(shí)刻,80個(gè)興奮性神經(jīng)元和20個(gè)抑制性神經(jīng)元通過化學(xué)突觸全局耦合,興奮性突觸連接強(qiáng)度為gmax/2,抑制性突觸連接強(qiáng)度為3gmax/2[6-7,28]。在自組織演化過程中,興奮性突觸權(quán)重由STDP規(guī)則更新,抑制性突觸權(quán)重保持不變[28,35]。同時(shí),為了便于描述,采用P0代替突觸權(quán)重在[0,0.1gmax]范圍內(nèi)的突觸比例,P1代替權(quán)重在[0.9gmax,gmax]范圍內(nèi)的突觸比例,P2代替其它突觸的比例。
2.1 自組織神經(jīng)元網(wǎng)絡(luò)的突觸連接特性圖1給出了自組織過程中神經(jīng)元網(wǎng)絡(luò)結(jié)構(gòu)以及突觸權(quán)重概率分布。由于神經(jīng)元的異質(zhì)性,神經(jīng)元之間的突觸連接受到STDP規(guī)則的調(diào)節(jié)。從高興奮性神經(jīng)元到低興奮性神經(jīng)元的突觸連接得到加強(qiáng),而從低興奮性神經(jīng)元到高興奮性神經(jīng)元之間的突觸連接減弱[6-7,28]。因此,隨著自組織過程的進(jìn)行,P2逐漸減小,而P0和P1逐漸增加。當(dāng)模擬時(shí)間超過150 ms以后,神經(jīng)元網(wǎng)絡(luò)結(jié)構(gòu)達(dá)到動(dòng)態(tài)穩(wěn)定狀態(tài)[15]。文中所有結(jié)果都是基于從150 ms到200 ms的動(dòng)態(tài)神經(jīng)元網(wǎng)絡(luò)進(jìn)行的,采樣步長(zhǎng)0.05 ms。
圖2給出了不同電磁輻射強(qiáng)度下,自組織神經(jīng)元網(wǎng)絡(luò)中突觸權(quán)重比例P0,P1和P2的演化規(guī)律。隨著輻射強(qiáng)度的增加,弱突觸比例P0逐漸降低,在輻射強(qiáng)度超過5后,突然下降;強(qiáng)突觸比例P1整體呈下降趨勢(shì),但下降幅度較小,在A=4處出現(xiàn)極小值;而P2呈現(xiàn)整體上升趨勢(shì)。該結(jié)果表明電磁輻射誘發(fā)突觸調(diào)整比例下降。這是因?yàn)榇艌?chǎng)對(duì)神經(jīng)元的負(fù)反饋降低了神經(jīng)元的興奮性,從而減弱了神經(jīng)元之間的競(jìng)爭(zhēng)以及突觸權(quán)重的調(diào)整[15],而外界電磁輻射進(jìn)一步降低了神經(jīng)元的興奮性,使得突觸更新比例下降。
圖3為不同電磁輻射強(qiáng)度下自組織神經(jīng)元網(wǎng)絡(luò)的平均突觸權(quán)重。隨著輻射強(qiáng)度的增加,平均突觸權(quán)重整體上升,表明自組織神經(jīng)元網(wǎng)絡(luò)連接密度升高;但在A=4處,平均突觸權(quán)重有極小值,且遠(yuǎn)小于無電磁輻射時(shí)的值。對(duì)比圖2和圖3可知,弱突觸比例下降幅度明顯高于強(qiáng)突觸比例下降幅度,造成平均突觸權(quán)重的增加。但在A=4處,弱突觸比例基本沒有降低,而強(qiáng)突觸比例卻顯著降低,導(dǎo)致平均突觸權(quán)重出現(xiàn)極小值。以上結(jié)果表明電磁輻射通過抑制神經(jīng)元的興奮性,降低神經(jīng)元之間的競(jìng)爭(zhēng),對(duì)突觸連接具有復(fù)雜的影響。
2.2 自組織神經(jīng)元網(wǎng)絡(luò)的拓?fù)涮匦宰越M織神經(jīng)元網(wǎng)絡(luò)為有向加權(quán)網(wǎng)絡(luò),神經(jīng)信息從前突觸神經(jīng)元傳遞到后突觸神經(jīng)元,反映了神經(jīng)元之間的因果關(guān)系。電磁輻射調(diào)控突觸連接,必然會(huì)影響神經(jīng)元之間的因果關(guān)系。采用因果流(causal flow,CF)刻畫神經(jīng)元之間的因果關(guān)系[36],其公式如下
CF(i)=kouti-kini
(6)
式中 kouti=∑j∈Ngij神經(jīng)元i的出度;
kini=∑j∈Ngij
為神經(jīng)元i的入度;N為神經(jīng)元的個(gè)數(shù)。高興奮性神經(jīng)元具有更高的出度和更大的正因果流,稱為“因果源”;而低興奮性神經(jīng)元具有更小的出度和更小的負(fù)因果流,稱為“因果匯”[15]。圖4為神經(jīng)元的平均因果流隨電磁輻射強(qiáng)度的變化,正因果流在因果源神經(jīng)元中進(jìn)行平均,負(fù)因果流在因果匯神經(jīng)元中進(jìn)行平均。隨著電磁輻射強(qiáng)度的增加,正負(fù)因果流都下降,且在A=4處有一個(gè)極小值,表明從低興奮性神經(jīng)元到高興奮性神經(jīng)元之間的突觸連接增強(qiáng)。這是因?yàn)殡姶泡椛浣档土松窠?jīng)元的興奮性,導(dǎo)致自組織神經(jīng)元網(wǎng)絡(luò)的異質(zhì)性降低,因果關(guān)系減弱。而負(fù)因果流增加是因?yàn)榈团d奮性神經(jīng)元接收到了更多的輸入。與此同時(shí),在A=4處,強(qiáng)突觸連接比例出現(xiàn)極小值(圖2),強(qiáng)興奮性神經(jīng)元對(duì)其它神經(jīng)元的因果關(guān)系減弱,導(dǎo)致負(fù)因果流出現(xiàn)極小值。
在自組織演化過程中,具有相似放電特性的神經(jīng)元會(huì)聚集在一起形成模塊,模塊內(nèi)的神經(jīng)元具有更強(qiáng)的突觸連接,而模塊之間的連接較弱。模塊化結(jié)構(gòu)對(duì)神經(jīng)元網(wǎng)絡(luò)的動(dòng)力學(xué)行為具有重要意義[37]。采用模塊化(modularity,Q)刻畫網(wǎng)絡(luò)中模塊的密度,其公式如下[38-39]
Q=1l∑i,j∈N
gij-koutikinil
δmi,mj
(7)
式中 l=∑i,j∈Ngij為所有突觸連接權(quán)重的總和;
δmi,mj
為δ函數(shù);mi為包含神經(jīng)元i的模塊。當(dāng)神經(jīng)元i和j在同一模塊,
δmi,mj=1,否則,δmi,mj=0。Q越大,表明網(wǎng)絡(luò)中模塊密度越高。圖5為不同電磁輻射強(qiáng)度下,自組織神經(jīng)元網(wǎng)絡(luò)的模塊化程度。隨著輻射強(qiáng)度的增加,模塊化程度先增加,后減小。在A=4處,自組織神經(jīng)元網(wǎng)絡(luò)具有最高的模塊化程度。文獻(xiàn)[15]發(fā)現(xiàn)在自發(fā)性電磁場(chǎng)環(huán)境下,隨著因果關(guān)系的減弱,神經(jīng)元模塊化程度降低,與文中強(qiáng)電磁輻射條件下的因果流和模塊化變化的關(guān)系一致。
神經(jīng)元網(wǎng)絡(luò)具有典型的小世界特性,從而保證信息在局部回路和全局網(wǎng)絡(luò)中的高效傳輸[5]。采用局部效率(local efficiency,Eloc)刻畫信息在局部神經(jīng)回路的傳遞效率[38-39],定義為
Eloc=12N∑i∈N
∑j,h∈N,j≠i(gij+gji)
(kini+kouti)
(gih+ghi)([djh(i)]-1+[dhj(i)]-1)
(kini+kini-1)-2∑j∈Ngijgji
(8)
式中 djh(i)為從神經(jīng)元j到h,且經(jīng)過神經(jīng)元i的最短路徑。全局效率(global efficiency,Eglob)定義為[38-39]
Eglob=1N∑i∈N
∑j∈N,j≠i(dij)-1
N-1
(9)
式中 dij為從神經(jīng)元i到j(luò)的最短路徑。圖6給出了自組織神經(jīng)元網(wǎng)絡(luò)效率隨電磁輻射強(qiáng)度的變化趨勢(shì)。隨著輻射強(qiáng)度的增加,局部效率呈整體上升趨勢(shì),但在A=4時(shí)出現(xiàn)極小值,且顯著小于無電磁輻射時(shí)的局部效率。而整體效率先下降后上升,同樣在A=4處出現(xiàn)極小值,并且只有在A=7時(shí),全局效率才大于無電磁輻射時(shí)的值。圖6結(jié)果表明電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)效率的影響依賴于輻射強(qiáng)度,在A=4時(shí),降低網(wǎng)絡(luò)的局部效率與全局效率,而在A=7時(shí),增強(qiáng)網(wǎng)絡(luò)的局部效率與全局效率。對(duì)比圖3和圖6可知,局部效率與平均突觸權(quán)重變化趨勢(shì)類似,與文獻(xiàn)[15]中的結(jié)果一致。這是因?yàn)榫植可窠?jīng)信息傳遞效率依賴于局部神經(jīng)回路中突觸連接強(qiáng)度,在A=4處,神經(jīng)元網(wǎng)絡(luò)中強(qiáng)突觸比例突然下降,而弱突觸比例變化很小,從而導(dǎo)致局部效率的下降。之后,弱突觸比例突然下降,而強(qiáng)突觸比例變化較小,導(dǎo)致局部神經(jīng)回路的增強(qiáng),局部效率上升。而對(duì)比圖5和圖6可知,全局效率與模塊化的變化趨勢(shì)相反。這是因?yàn)橥挥|連接在模塊內(nèi)部較強(qiáng),從而維持信息在模塊內(nèi)部的高效傳遞,而突觸連接在模塊之間較弱,導(dǎo)致信息在整個(gè)網(wǎng)絡(luò)的傳遞效率下降;自組織神經(jīng)元網(wǎng)絡(luò)模塊化程度升高,造成模塊之間的信息傳遞下降,全局效率降低。
2.3 自組織神經(jīng)元網(wǎng)絡(luò)的同步特性神經(jīng)元網(wǎng)絡(luò)結(jié)構(gòu)的拓?fù)涮匦詫?duì)神經(jīng)元的同步行為有至關(guān)重要的作用,SHESHBOLOUKI等基于網(wǎng)絡(luò)的特征值,提出2種結(jié)構(gòu)性的測(cè)量指標(biāo)來刻畫有向加權(quán)網(wǎng)絡(luò)的同步能力[40]。自組織神經(jīng)元網(wǎng)絡(luò)對(duì)應(yīng)的拉普拉斯矩陣為
L=D-G,其中G為突觸連接矩陣;D為矩陣G的對(duì)角矩陣。拉普拉斯矩陣的特征值為λk=λrk+jλik,k=1,…,N,j=-1并按照實(shí)部進(jìn)行排序0=λr1≤λr2…≤λrN。第1個(gè)測(cè)量指標(biāo)為特征值比,定義為實(shí)部非零最小特征值與實(shí)部最大特征值的比值[40]
Rλ=λrminλrmax
(10)
更大的Rλ表明網(wǎng)絡(luò)具有更高的同步能力。第2個(gè)測(cè)量指標(biāo)為特征值的歸一化標(biāo)準(zhǔn)差[62]
σλ=1d2(N-N0)
∑i=Ni=N0|λi-|2
(11)
式中 =
1N-N0∑i=Ni=N0λi,d=1N
∑i∑j≠igij,N0為零特征值的個(gè)數(shù)。更小的σλ表明網(wǎng)絡(luò)具有更高的同步能力。圖7為自組織神經(jīng)元網(wǎng)絡(luò)的同步能力隨電磁輻射強(qiáng)度的變化。隨著輻射強(qiáng)度的增加,Rλ逐漸上升,表明網(wǎng)絡(luò)的同步能力逐漸增強(qiáng)。同時(shí),σλ隨輻射強(qiáng)度逐漸下降,進(jìn)一步證明了電磁輻射提高了自組織神經(jīng)元網(wǎng)絡(luò)的同步能力。而在A=4處,自組織神經(jīng)元網(wǎng)絡(luò)的同步能力出現(xiàn)極大值。對(duì)比圖2和圖7可知,自組織神經(jīng)元網(wǎng)絡(luò)中弱突觸連接和強(qiáng)突觸連接比例的減小,使網(wǎng)絡(luò)具有較高的同步能力。尤其在A=4處,弱突觸連接比例基本不變,而強(qiáng)突觸連接比例出現(xiàn)極小值,此時(shí)網(wǎng)絡(luò)的同步能力出現(xiàn)極大值,表明強(qiáng)突觸連接顯著影響自組織神經(jīng)元網(wǎng)絡(luò)的同步能力,與文獻(xiàn)[15]中結(jié)果類似。對(duì)于異質(zhì)性神經(jīng)元,且突觸為單向連接,強(qiáng)突觸或弱突觸耦合都不利于神經(jīng)元的同步,合適的耦合強(qiáng)度才能促進(jìn)同步,且神經(jīng)元不能達(dá)到完全同步。與此同時(shí),對(duì)比圖4和圖7可知,神經(jīng)元之間的因果關(guān)系與自組織網(wǎng)絡(luò)的同步能力密切相關(guān),因果關(guān)系減弱,伴隨著網(wǎng)絡(luò)同步能力的增強(qiáng),與文獻(xiàn)[15]中結(jié)果相符。表明對(duì)于異質(zhì)性神經(jīng)元,有向突觸連接差距越小,越有利于同步。
神經(jīng)元的同步活動(dòng)不僅與網(wǎng)絡(luò)結(jié)構(gòu)有關(guān),還與突觸連接以及神經(jīng)元的動(dòng)力學(xué)行為有關(guān)[37]。圖8給出了不同電磁輻射強(qiáng)度下,神經(jīng)元的放電模式,圖中黑點(diǎn)代表神經(jīng)元在某一時(shí)刻處于放電狀態(tài),即膜電位超過閾值0 mV。在A=4時(shí),神經(jīng)元的放電規(guī)則性明顯高于A=0,表明增強(qiáng)的同步放電活動(dòng)。但在A=5處,神經(jīng)元的放電規(guī)則性下降,伴隨著網(wǎng)絡(luò)同步程度降低。當(dāng)A=8時(shí),神經(jīng)元的放電規(guī)則性明顯高于上面3種情況。該結(jié)果表明電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)的同步程度有復(fù)雜的影響。為了進(jìn)一步定量刻畫神經(jīng)元放電同步程度,計(jì)算神經(jīng)元放電時(shí)間序列之間的絕對(duì)相關(guān)系數(shù)
Rij=
∑Lt=1(Vi(t)-i)(Vj(t)-j)
∑Lt=1(Vi(t)-i)2∑Lt=1(Vj(t)-j)2
(12)
式中 Vi為第i個(gè)神經(jīng)元的放電時(shí)間序列,采樣范圍從150 ms到200 ms,采樣間隔0.05 ms;L為時(shí)間序列長(zhǎng)度;t為時(shí)間點(diǎn)。網(wǎng)絡(luò)的平均相關(guān)系數(shù)
R=1N(N-1)∑Ni=1∑Nj=1,j≠iRij
(13)
刻畫了網(wǎng)絡(luò)的同步程度。R越大,表明神經(jīng)元網(wǎng)絡(luò)的同步程度越高。從圖9可知,R隨電磁輻射強(qiáng)度增加而上升,在A=4時(shí)有極大值,表明電磁輻射增強(qiáng)了自組織神經(jīng)元網(wǎng)絡(luò)的同步行為,與圖8中結(jié)果完全相符。對(duì)比圖7和圖9可知,電磁輻射環(huán)境下,自組織神經(jīng)元網(wǎng)絡(luò)擁有更高的同步能力,并且由于全局同步動(dòng)力學(xué)與局部突觸連接的反饋耦合,該網(wǎng)絡(luò)也能夠產(chǎn)生更高的神經(jīng)元同步放電活動(dòng)。
從圖8可以看出,抑制性神經(jīng)元的放電程度明顯低于興奮性神經(jīng)元,這主要是因?yàn)橐种菩陨窠?jīng)元之間的突觸連接不受STDP規(guī)則的調(diào)節(jié),從而導(dǎo)致抑制性神經(jīng)元之間相互抑制放電。同時(shí),隨著電磁輻射強(qiáng)度的增加,興奮性神經(jīng)元的放電程度發(fā)生明顯變化,如A=8時(shí)的放電程度明顯低于其它3種情況,表明電磁輻射影響神經(jīng)元網(wǎng)絡(luò)興奮性程度。進(jìn)一步計(jì)算神經(jīng)元的放電概率刻畫網(wǎng)絡(luò)的興奮性程度[1]
F=1L∑Lt=1f(t)=
1L∑Lt=1
1N|{i|Vi(t)≥0}|
(14)
式中 f(t)刻畫了在時(shí)刻t,膜電位超過閾值0 mV的神經(jīng)元比例。更大的F表明放電神經(jīng)元個(gè)數(shù)越多,網(wǎng)絡(luò)興奮性程度越高。圖10分別給出了所有神經(jīng)元,興奮性神經(jīng)元和抑制性神經(jīng)元的放電概率。從圖10可以看出,興奮性神經(jīng)元的放電概率顯著高于抑制性神經(jīng)元,與圖8結(jié)果一致,而且興奮性神經(jīng)元的放電概率隨電磁輻射強(qiáng)度變化趨勢(shì)與整個(gè)網(wǎng)絡(luò)的放電概率變化趨勢(shì)基本一致。這是因?yàn)榫W(wǎng)絡(luò)的興奮性程度主要與興奮性神經(jīng)元有關(guān),且只有興奮性突觸連接受神經(jīng)元放電活動(dòng)的調(diào)節(jié),導(dǎo)致電磁輻射主要通過調(diào)節(jié)興奮性神經(jīng)元與興奮性突觸之間的反饋,控制自組織神經(jīng)元網(wǎng)絡(luò)的興奮性程度。同時(shí),在圖10中,隨著電磁輻射強(qiáng)度增加,神經(jīng)元的放電概率先增加后減小,在A=4處出現(xiàn)極大值,表明電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)興奮性程度的復(fù)雜調(diào)控作用,過大的電磁輻射強(qiáng)度降低了神經(jīng)元網(wǎng)絡(luò)的興奮性,與圖4中結(jié)果相符,而合適的電磁輻射強(qiáng)度提高了神經(jīng)元網(wǎng)絡(luò)的興奮性。對(duì)比圖9和圖10可知,在A=4處,網(wǎng)絡(luò)同步和放電概率都出現(xiàn)極大值,而在A=8處,同步因子最大,放電概率最小,表明了2種不同的同步機(jī)制。
3 結(jié) 論
1)電磁輻射通過增強(qiáng)磁場(chǎng)對(duì)神經(jīng)元的負(fù)反饋,抑制神經(jīng)元的興奮性,降低神經(jīng)元之間的競(jìng)爭(zhēng)以及突觸連接更新比例,從而減弱神經(jīng)元之間的因果關(guān)系。
2)電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)的模塊化程度和網(wǎng)絡(luò)效率的影響依賴于輻射強(qiáng)度;在A=7處,神經(jīng)元網(wǎng)絡(luò)具有增強(qiáng)的局部效率與全局效率。3)電磁輻射增強(qiáng)了自組織神經(jīng)元網(wǎng)絡(luò)的同步能力,能夠誘發(fā)更高的神經(jīng)元同步放電活動(dòng)。4)由于神經(jīng)元之間復(fù)雜的交互作用,電磁輻射對(duì)自組織神經(jīng)元網(wǎng)絡(luò)興奮性影響依賴于輻射強(qiáng)度。合適的輻射強(qiáng)度能夠增強(qiáng)神經(jīng)元網(wǎng)絡(luò)的興奮性,而過高的輻射強(qiáng)度降低神經(jīng)元網(wǎng)絡(luò)的興奮性。
參考文獻(xiàn)(References):
[1]
王榮,吳瑩,劉少寶.隨機(jī)中毒對(duì)神經(jīng)元網(wǎng)絡(luò)時(shí)空動(dòng)力學(xué)行為的影響[J].物理學(xué)報(bào),2013,62(22):56-63.WANG Rong,WU Ying,LIU Shaobao.Effects of random poisoning on spatiotemporal dynamics of neuronal networks[J].Acta Physica Sinica,2013,62(22):56-63.
[2]李佳佳.電磁輻射誘發(fā)神經(jīng)元放電節(jié)律轉(zhuǎn)遷的動(dòng)力學(xué)行為研究[J].物理學(xué)報(bào),2015,64(3):224-230.LI Jiajia.Dynamic behavior of neuronal discharge rhythm transition induced by electromagnetic radiation[J].Acta Physica Sinica,2015,64(3):224-230.
[3]WANG Y H,WANG R B,ZHU Y T.Optimal pathfinding through mental exploration based on neural energy field gradients[J].Cognitive Neurodynamics,2017,11(1):99-111.
[4]DAMOUR J A,F(xiàn)ROEMKE R C.Inhibitory and excitatory spiketimingdependent plasticity in the auditory cortex[J].Neuron,2015,86(2):514-528.
[5]BASSETT D S,BULLMORE E T.Smallworld brain networks revisited[J].Neuroscientist,2017,23(5):499-516.
[6]LI X M,SMALL M.Neuronal avalanches of a selforganized neural network with activeneurondominant structure[J].Chaos,2012,22(2):023104.
[7]LI X M,ZHANG J,SMALL M.Selforganization of a neural network with heterogeneous neurons enhances coherence and stochastic resonance[J].Chaos,2009,19:013126.
[8]GRAHAM L J.Mechanisms of magnetic stimulation of central nervous system neurons[J].PLoS Computational Biology,2011,7(3):e1002022.
[9]AKIYAMA H.Extracellular DC electric fields induce nonuniform membrane polarization in rat hippocampal CA1 pyramidal neurons[J].Brain Research,2011,1383:22-35.
[10]YI G S.Spikefrequency adaptation of a twocompartment neuron modulated by extracellular electric fields[J].Biologaical Cybernetics,2015,109(3):287-306.
[11]LV M.Model of electrical activity in a neuron under magnetic flow effect[J].Nonlinear Dynamics,2016,85(3):1479-1490.
[12]MA J,TANG J.A review for dynamics in neuron and neuronal network[J].Nonlinear Dynamics,2017,89(3):1569-1578.
[13]XU Y.Collective responses in electrical activities of neurons under field coupling[J].Scientific Reports,2018,8(1):1349.
[14]WU F Q.Dynamical responses in a new neuron model subjected to electromagnetic induction and phase noise[J].Physica A,2017,469:81-88.
[15]WANG R,F(xiàn)AN Y C,WU Y.Spontaneous electromagnetic induction promotes the formation of economical neuronal network structure via selforganization process[J].Scientific Reports,2019,9(1):9698.
[16]WANG J.Bursting types and bifurcation analysis in the prebtzinger complex respiratory rhythm neuron[J].International Journal of Bifurcation and Chaos,2017,27(1):1750010.
[17]ZHAO Y.Phase synchronization dynamics of coupled neurons with coupling phase in the electromagnetic field[J].Nonlinear Dynamics,2018,93(3):1315-1324.
[18]GUO S L.Collective response,synapse coupling and field coupling in neuronal network[J].Chaos Soliton Fract,2017,105:120-127.
[19]MA J.Phase synchronization between two neurons induced by coupling of electromagnetic field[J].Applied Mathematics and Computation,2017,307:321-328.
[20]XU Y.Synchronization between neurons coupled by memristor[J].Chaos Soliton Fract,2017,104:435-442.
[21]REN G D,XU Y,WANG C N.Synchronization behavior of coupled neuron circuits composed of memristors[J].Nonlinear Dynamics,2017,88(2):893-901.
[22]XU Y.Effects of ion channel blocks on electrical activity of stochastic HodgkinHuxley neural network under electromagnetic induction[J].Neurocomputing,2017,283:196-204.
[23]HU X Y.Chaotic dynamics in a neural network under electromagnetic radiation[J].Nonlinear Dynamics,2017,91(3):1541-1554.
[24]MA J,WU F Q,WANG C N.Synchronization behaviors of coupled neurons under electromagnetic radiation[J].International Journal of Modern Physics B,2017,31(2):1650251.
[25]ZHANG X H,LIU S Q.Stochastic resonance and synchronization behaviors of excitatoryinhibitory smallworld network subjected to electromagnetic induction[J].Chinese Physics B,2018,27(4):40501.
[26]GE M Y.Mode transition in electrical activities of neuron driven by high and low frequency stimulus in the presence of electromagnetic induction and radiation[J].Nonlinear Dynamics,2017,91(1):515-523.
[27]WU J,XU Y,MA J.Levy noise improves the electrical activity in a neuron under electromagnetic radiation[J].PLoS ONE,2017,12(3):e0174330.
[28]WANG R.Structure and dynamics of selforganized neuronal network with an improved STDP rule[J].Nonlinear Dynamics,2017,88(3):1855-1868.
[29]WANG Q Y.Multiple firing coherence resonances in excitatory and inhibitory coupled neurons[J].Communications in Nonlinear Science and Numerical Simulation,2012,17(10):3979-3988.
[30]ZHAO Z G,JIA B,GU H G.Bifurcations and enhancement of neuronal firing induced by negative feedback[J].Nonlinear Dynamics,2016,86(3):1-12.
[31]LI Q D.On hyperchaos in a small memristive neural network[J].Nonlinear Dynamics,2014,78(2):1087-1099.
[32]ZHANG J H,LIAO X F.Synchronization and chaos in coupled memristorbased FitzHughNagumo circuits with memristor synapse[J].AEUEInternational Journal of Electronics and Communications,2017,75:82-90.
[33]FENG P H,WU Y,ZHANG J Z.A route to chaotic behavior of single neuron exposed to external electromagnetic radiation[J].Frontiers in Computational Neuroscience,2017,11:94.
[34]LI Q D,ZENG H Z,LI J.Hyperchaos in a 4D memristive circuit with infinitely many stable equilibria[J].Nonlinear Dynamics,2015,79(4):2295-2308.
[35]PREZIOSO M.Selfadaptive spiketimedependent plasticity of MetalOxide memristors[J].Scientific Reports,2016,6:21331.
[36]KONG W Z.Investigating driver fatigue versus alertness using the granger causality network[J].Sensors,2015,15(8):19181-19198.
[37]JALILI M.Enhancing synchronizability of diffusively coupled dynamical networks:a survey[J].IEEE Transactions on Neural Networks and Learning Systems,2013,24(7):1009-1022.
[38]RUBINOV M,SPORNS O.Complex network measures of brain connectivity:Uses and interpretations[J].NeuroImage,2010,52(3):1059-1069.
[39]LEICHT E A,NEWMAN M E.Community structure in directed networks[J].Physical Review Letters,2008,100(11):118703.
[40]SHESHBOLOUKI A,ZAREI M,SARBAZIAZAD H.Are feedback loops destructive to synchronization?[J].Europhysics Letters,2015,111(4):40010.