Synchronization properties, malleability of synchronization, and nonstationary states ein neural networks
Resumo
Resumo: A pesar dos grandes avancos, os mecanismos de funcionam ento por trás dos interessantes fenôm enos observados no cerebro ain d a carecem de u m a te o ria fu n d am en tal que de u m a descricao m a te m a tic a e explique os fenom enos neurais e a N eurociencia com o um todo. Investigacoes recentes forneceram inform açoes sobre as prop riedad es individuais dos neuronios, seus padroes de ativ id ad e e ate m esm o como eles estao conectados. No en tan to , os m ecanism os p o r trá s do com p o rtam en to coletivo dos sistem as neurais ainda nao sao entendidos. P o r o u tro lado, e im p o rta n te en fatizar que os avanços recentes nas analises ex perim entais, com binados com esforços interdisciplinares, p ro d u zira m ideias interessantes que abrem grandes cam inhos p ara os práxim os anos de pesquisa. E sta tese e desenvolvida considerando este am biente frutífero, onde o uso de m odelagem m atem atica, ideias físicas e análise com putacional abrem oportunidades p ara avancos im portantes para um a m elhor com preensão das questoes da neurociencia. Nesse sentido, o com portam ento coletivo dos sistem as n eurais assum e g ran d e im p o rtan cia, onde se observam fenom enos interessantes, como oscilacoes, sincronizacao e m etaestabilid ad e, que estao relacionados ta n to a características saudíveis quanto nao saudíveis do cerebro. P ortanto, a anílise desse tip o de fenom eno te m grande im p o rtân cia do p o n to de v ista d a física e da neurociencia. D esta forma, os objetivos desta tese sao descritos em três linhas principais de pesquisa: (i) a relaçao entre os padroes de atividade neural individuais e a sincronizacao da rede; (ii) a resposta dos sistem as neurais devido a pequenas m udanças nos estímulos de externos; (iii) o papel da topologia da rede nas propriedades de sincronização e interm itència observada em sistem as neurais. P a ra investigar essas questães, os sistem as neurais sao m odelados p o r redes com plexas, onde cad a no rep resen ta um neuronio e as arestas rep resen tam as sinapses. A lem disso, sao considerados diferentes m odelos neurais e a rq u ite tu ra s de conexao, onde o sistem a e resolvido p or meio de simulacoes num ericas e, a seguir, o nível de sincronizacao e suas prop riedad es sao estu dado s com base nos dados sim ulados. P o r fim, os resu ltado s aqui obtidos indicam características im p o rtan tes do co m po rtam ento coletivo de sistem as neurais, onde se en co n tram estados com diferentes prop riedad es de sincronizacão e padrães espaco-tem porais. A lem disso, um a relacão en tre a variabilidade neural individual e a sincronizaçcaão íe encontrada: quanto m enor a variabilidade da atividade neural individual, m aior o nível de sincronização considerando sinapses fracas. Alem disso, um equilíbrio entre conexães nao-locais e locais leva a rede neural a representar um a alta m aleabilidade de sincronizacao devido a pequenas m udancas nos estím ulos externos. Por outro lado, o processo de transicao p ara a sincronizacao pode representar a interm itència d a sincronizacao, onde estad os nao estacionarios p o d em ser observados e a topologia desem p en ha u m p ap el im p o rtan te. P o rta n to , os resu ltad o s aqui descritos con tribu em p a ra um a m elhor com preensao do com portam ento coletivo em sistem as neurais e abrem questoes e possibilidades im po rtan tes p a ra fu tu ras aplicaçoes. Abstract: T D espite great advances, the functioning m echanisms behind the interesting phenom ena observed in th e b rain still lacking a fu n d am en tal theory, w hich gives a m ath em a tic al description, su p p o rts and explains th e n eu ral p hen om ena and neuroscience as a whole. Recent investigations have provided inform ation about th e individual properties of neurons, their p attern s of activity, and even how they are connected. Yet, the m echanisms behind the collective behavior of neural system s still lacking. Otherwise, it is im p ortan t to em phasize th a t recent advances in experim ental analyses, combined w ith interdisciplinary efforts have p rod uced in terestin g ideas th a t open large avenues for th e n ex t years of research. T his thesis is developed considering th is fruitful environm ent, w here th e use of m ath em atical m odeling, physics ideas, and co m p u tatio n al analysis open o p p o rtu n itie s for im p o rta n t advances tow ard a b e tte r understanding of neuroscience issues. In this sense, th e collective behavior of neural system s is im p o rtan t, where interesting phenom ena are observed, such as oscillations, synchronization, an d m etastability, which are related to b o th healthy and unhealthy characteristics in th e brain. Therefore, the analysis of this kind of phenom enon has huge im portance from the point-of-view of bo th, physics and neuroscience. Then, the goals of this thesis are described in three m ain research lines: (i) th e relationship between individual neural activity p a tte rn s an d th e netw ork synchronization; (ii) th e response of neural system s due to small changes in the input stim uli; (iii) the role of network topology in th e synchronization properties and interm itten cy observed in neural system s. In order to investigate th ese questions, n eu ral system s are m odeled by com plex netw orks, w here each node represents a neuron, and th e edges represent th e synapses. Moreover, different n eural m odels and connection architectu res are used, w here th e problem is solved using num erical sim ulations, and, then, the level of synchronization and its properties are studied based on th e sim ulated data. At last, th e results obtained here indicate im po rtant features of th e collective behavior of neu ral system s, w here states w ith different synchronization p ro p erties and sp atio tem p o ral p a tte rn s are found. F urtherm ore, a relatio nship betw een in dividual n eu ral v ariability an d synchronization is found: th e lower th e v ariability of individual neural activity th e higher the level of synchronization for weak synapses. Also, a balance between nonlocal and local connections leads the neural network to depict a high m alleability of synchronization due to tiny changes in the external stimuli. Besides this, in th e process of transitio n to synchronization, synchronization interm ittency can be observed, w here n o n statio n ary sta te s can be observed and th e topology plays an im p o rta n t role. Therefore, th e results here contribute to a b e tte r understanding of th e collective behavior in neural system s and open im p o rtan t issues and possibilities for fu rth er application.
Collections
- Teses [122]