Recentemente observou-se experimentalmente, Johnson et al. (2010), que fatias organotípicas corticais de rato são capazes de completar padrões espaço-temporais, após serem treinadas. Embora se especule que mecanismos de plasticidade sináptica e homeostática estejam por trás do fenômeno, ainda não existe nenhuma explicação detalhada sobre o assunto. Com o intuito de propor uma explicação clara e consistente para os mecanismos que permeiam a resposta da rede aos estímulos como um todo, nos propomos a estudar este fenômeno por meio de uma rede de neurônios de integração-e-disparo dotada de mecanismos de homeostase intrínseca e de plasticidade sináptica dependente de disparos. O sistema construído foi explorado, de modo a determinar em que condições a rede poderia comportar-se como o sistema real, e treinado de forma similar `a realizada experimentalmente por Johnson et al. (2010). / Recently it has been observed experimentally, Johnson et al. (2010), that organotypic cortical slices of rat are capable of completing spatio-temporal patterns after training. Although it is speculated that synaptic and homeostatic plasticity may have an important role in this phenomenon, there is still no detailed explanation about this subject. In order to propose a clear and consistent explanation for the mechanisms that underlie the network response to stimuli as a whole, we propose to study this phenomenon through a network of integrate-and-fire neurons endowed with intrinsic homeostasis and spike-timing dependent plasticity mechanisms. The constructed system was explored, aiming to determine in which conditions the network could behave as the real system, and trained in a way similar as the experimental one done by Johnson et al. (2010).

O estudo de redes complexas tem despertado muita atenção nos últimos anos, principalmente pela sua capacidade de permitir a análise dos mais diversificados sistemas através de um mesmo conjunto de ferramentas matemáticas e computacionais. Até pouco tempo a ênfase nessa área era sobre o estudo das propriedades estruturas e sua influência em característica globais da dinâmica ocorrendo sobre o sistema. Recentemente foi-se percebendo a riqueza de comportamentos que podemos estudar ao olharmos para grupos de vértices e as interações que ocorrem entre esses grupos, de forma que cada vez mais se torna necessária a criação de uma forma sistemática de quantificar, a nível de nó, como uma dinâmica é influenciada (ou diferenciada) pela estrutura do sistema. Apresentamos nesse trabalho um primeiro passo nessa direção, no qual definimos uma medida que, baseada em características dinâmicas obtidas em diversas condições iniciais, busca comparar o comportamento dos nós presentes no sistema. Através dessa medida encontramos a alta capacidade de modelos de rede geográficas de produzir, por simples flutuações estatísticas, grupos de nós de dinâmica muito distinta em relação ao demais, um fato que não ocorre para uma rede definida de forma semelhante mas não geográfica. Nessas redes conseguimos verificar também se um nó de topologia local distinta dos demais consegue possuir uma dinâmica diferenciada, o que encontramos foi que mesmo em regiões da rede extremamente regulares existe a formação de grupos dinâmicos devido ao efeito da borda dessa região. Saindo de modelos de redes para apresentar aplicações práticas do método, encontramos na rede neuronal do verme Caenorhabditis elegans um conjunto de nós de alta diferenciação dinâmica, que representam os interneurônios do cordão ventral do verme. Adicionalmente encontramos na rede cortical do macaco da família Cercopithecidae uma divisão em relação a regularidade dos sinais dinâmicos, o que indica a presença de comunidades funcionais nessa rede. Além da metodologia de diferenciação desenvolvemos ainda uma ferramenta que busca encontrar de forma totalmente automatizada as características dinâmicas mais relevantes do sistema em estudo, que foi capaz de representar medidas dinâmicas tradicionalmente utilizadas na área. Todos esses resultados abrem caminho para diversas vertentes de estudo, em especial citamos a influência de uma borda irregular no interior de uma região regular, o estudo da rede Caenorhabditis elegans com dinâmicas neuronais mais precisas e a aplicação sistemática de dinâmicas para encontrar a divisão funcional de comunidades em redes direcionadas, um tema que apresenta resultados promissores. / The study of complex networks has drawn much attention over the last years, mainly by virtue of its potential to characterize the most diverse systems through the same mathematical and computational tools. Not long ago the emphasis on this field mostly focused on the effects of the structural properties on the global behavior of a dynamical process taking place in the system. Recently, some studies started to unveil the richness of interactions that occur between groups of nodes when we look at the small scale of interactions occurring in the network. Such findings call for a new systematic methodology to quantify, at node level, how a dynamics is being influenced (or differentiated) by the structure of the underlying system. Here we present a first step towards this direction, in which we define a new measurement that, based on dynamical characteristics obtained for a series of initial conditions, compares the dynamical behavior of the nodes present in the system. Through this measurement we find the high capacity of networks generated by geographical models to exhibit, by means of statistical fluctuations, groups of nodes with very distinct dynamics compared to the rest of the network, a behavior that does not occur for a similar non-geographical network. We also verify if a large topological differentiation of a node necessarily reflects on its dynamics. We find that even in very regular regions of the network the nodes tend to form dynamical groups influenced by the border effects. In addition to the network models used, we present practical applications of the methodology by using the neuronal network of the nematode Caenorhabditis elegans, where we show that the interneurons of the ventral cord presents a very large dynamical differentiation when compared to the rest of the network. We also analyze the cortical network of the Cercopithecidae monkey by means of signal communication complexity, finding that it contains two well-defined functional communities. Besides the differentiation measurement, we also presents an useful mechanism for automatically obtaining the relevant dynamical characteristics of the nodes, which showed promising results by obtaining traditional measurements of the area with little effort. All these results pave the way to a range of different studies, of which we highlight the influence of an irregular border on the dynamics taking place inside a regular network region, the study of the neurons of Caenorhabditis elegans using more robust neuronal dynamics and the systematic application of different dynamics in order to find the functional community division of directed networks.

A stochastic leaky integrate-and-fire neuron model was implemented in this study to simulate the spiking activity of the electrosensory "P-unit" receptor neurons of the weakly electric fish Apteronotus leptorhynchus. In the context of sensory coding, these cells have been previously shown to respond in experiment to natural random narrowband signals with either a linear or nonlinear coding scheme, depending on the intrinsic firing rate of the cell in the absence of external stimulation. It was hypothesised in this study that this duality is due to the relation of the stimulus to the neuron's excitation threshold. This hypothesis was validated with the model by lowering the threshold of the neuron or increasing its intrinsic noise, or randomness, either of which made the relation between firing rate and input strength more linear. Furthermore, synchronous P-unit firing to a common input also plays a role in decoding the stimulus at deeper levels of the neural pathways. Synchronisation and desynchronisation between multiple model responses for different types of natural communication signals were shown to agree with experimental observations. A novel result of resonance-induced synchrony enhancement of P-units to certain communication frequencies was also found.

A stochastic leaky integrate-and-fire neuron model was implemented in this study to simulate the spiking activity of the electrosensory "P-unit" receptor neurons of the weakly electric fish Apteronotus leptorhynchus. In the context of sensory coding, these cells have been previously shown to respond in experiment to natural random narrowband signals with either a linear or nonlinear coding scheme, depending on the intrinsic firing rate of the cell in the absence of external stimulation. It was hypothesised in this study that this duality is due to the relation of the stimulus to the neuron's excitation threshold. This hypothesis was validated with the model by lowering the threshold of the neuron or increasing its intrinsic noise, or randomness, either of which made the relation between firing rate and input strength more linear. Furthermore, synchronous P-unit firing to a common input also plays a role in decoding the stimulus at deeper levels of the neural pathways. Synchronisation and desynchronisation between multiple model responses for different types of natural communication signals were shown to agree with experimental observations. A novel result of resonance-induced synchrony enhancement of P-units to certain communication frequencies was also found.

A stochastic leaky integrate-and-fire neuron model was implemented in this study to simulate the spiking activity of the electrosensory "P-unit" receptor neurons of the weakly electric fish Apteronotus leptorhynchus. In the context of sensory coding, these cells have been previously shown to respond in experiment to natural random narrowband signals with either a linear or nonlinear coding scheme, depending on the intrinsic firing rate of the cell in the absence of external stimulation. It was hypothesised in this study that this duality is due to the relation of the stimulus to the neuron's excitation threshold. This hypothesis was validated with the model by lowering the threshold of the neuron or increasing its intrinsic noise, or randomness, either of which made the relation between firing rate and input strength more linear. Furthermore, synchronous P-unit firing to a common input also plays a role in decoding the stimulus at deeper levels of the neural pathways. Synchronisation and desynchronisation between multiple model responses for different types of natural communication signals were shown to agree with experimental observations. A novel result of resonance-induced synchrony enhancement of P-units to certain communication frequencies was also found.

Ce travail de thèse concerne la modélisation mathématique et l’étude du comportement d’une population de neurones. Dans tout ce travail on s’arrêtera principalement sur une population de neurones auto-excitateurs où chaque cellule du réseau est supposée suivre la loi de l’intègre et tire. Néanmoins nous aborderons au détour d’un chapitre la modélisation d’une population de neurones inhibiteurs, et dans une dernière partie, nous discuterons la modélisation d’une population de neurones obéissant au modèle Ermentrout-Kopell aussi appelé le théta-neurone. L’angle de vue adopté dans cette thèse est donné par l’approche densité de population. Cette approche, dont nous rappellerons en détail les hypothèses et la construction, a été introduite il ya maintenant plus d’une dizaine d’années afin de faciliter la simulation d’une grande population de neurones. Dit plus précisément, une telle approche donne une équation aux dérivées partielles sur la densité de population de neurones dans l’espace d’état formé des potentiels admissibles du neurone. Nous ferons de plus l’hypothèse que la réponse d’un neurone à l’arrivée d’une impulsion est une dépolarisation instantanée, autrement dit un saut de potentiel. Comme nous le verrons,cette équation aux dérivées partielles est non linéaire (à cause du couplage de la population) et non locale (à cause du saut de potentiel). Si cette idée est compliquée et abstraite, elle anéanmoins prouvé tout au long de ces dix dernières années son importance dans la simulation numérique des grands réseaux.Il s’agit avant tout dans ce travail de thèse de donner un cadre mathématique adéquat aux équations aux dérivées partielles qui surgissent d’une telle approche. Ainsi nous discuterons,selon les différents choix de modélisation, du caractère bien posé du modèle par densité de populationet de sa possible explosion en temps fini. Nous discuterons comment la prise en compte d’hypothèses réalistes supplémentaires dans la modélisation, comme le retard entre l’émission d’un potentiel d’action et sa réception ou encore la période réfractaire peut stopper l’explosionen temps fini et garantir l’existence d’une solution globale. Un autre aspect abordé dans ce travail concerne les explications et la prédiction de la synchronisation des neurones. Deux définitions de la synchronisation seront explicitées selon encoreune fois les choix de modélisation. Nous verrons qu’en interprétant l’explosion en temps fini dela solution comme l’arrivée d’une masse de Dirac dans le taux de décharge de la populationon peut relier l’explosion à la synchronisation. Toutefois, avec des hypothèses de modélisation plus réalistes, comme les retards et la période réfractaire, ce phénomène est exclu. Nous verrons néanmoins qu’avec ces paramètres physiques supplémentaires des solutions périodiques apparaissent offrant différents rythmes de décharge de la population. Encore une fois, l’apparition de ces oscillations sera perçue comme la synchronisation de la population. / This thesis concerns the mathematical modelling and the study of the behavior of a population of neurons. In this work we will mainly consider a population of excitatory neurons whe reall the cells of the network follow the integrate-and-fire model. Nonetheless, we will tackle in a chapter the modelling of an inhibitory population of neurons, and we will discuss in the lastchapter the modelling of a population of neurons that follows the Ermentrout-Koppell model.The point of view of this thesis is given by the population density approach that has beenintroduced more than a decade ago in order to facilitate the simulation of a large assembly ofneurons. More precisely, this approach gives a partial differential equation that describes thedensity of neurons in the state space that is the set of all admissible potential of a neuron. We will assume that when receiving an action potential, the potential of the neuron makes a small jump. As we will see this partial differential equation is non linear (due to the coupling betweenneurons) and non-local (due to the potential jump). If this idea is complicated and abstract, itallows to simulate easily a large neural network.First of all, the thesis gives a mathematical framework for the equations that arise from thisthe population density approach. Then we will discuss the existence and the possible blow upin finite time of the solution. We will discuss how the consideration of more realistic modellingassumptions, as the refractory period and the delay between the emission and the reception ofan action potential can stop the blow up of the solution and give a well posed model.We will also try to caracterise the occurence of synchronization of the neural network. Twodifferent ways of seeing the synchronization will be describe. One relates the blow up in finitetime of the solution to the occurence of a Dirac mass in the firing rate of the population.Nonetheless, taking into account the delays, this kind of blow up will not be observed anymore.Nonetheless, as we will see, with this additional features the model will generate some periodicalsolutions that can also be related to the synchronization of the population.

Recentemente observou-se experimentalmente, Johnson et al. (2010), que fatias organotípicas corticais de rato são capazes de completar padrões espaço-temporais, após serem treinadas. Embora se especule que mecanismos de plasticidade sináptica e homeostática estejam por trás do fenômeno, ainda não existe nenhuma explicação detalhada sobre o assunto. Com o intuito de propor uma explicação clara e consistente para os mecanismos que permeiam a resposta da rede aos estímulos como um todo, nos propomos a estudar este fenômeno por meio de uma rede de neurônios de integração-e-disparo dotada de mecanismos de homeostase intrínseca e de plasticidade sináptica dependente de disparos. O sistema construído foi explorado, de modo a determinar em que condições a rede poderia comportar-se como o sistema real, e treinado de forma similar `a realizada experimentalmente por Johnson et al. (2010). / Recently it has been observed experimentally, Johnson et al. (2010), that organotypic cortical slices of rat are capable of completing spatio-temporal patterns after training. Although it is speculated that synaptic and homeostatic plasticity may have an important role in this phenomenon, there is still no detailed explanation about this subject. In order to propose a clear and consistent explanation for the mechanisms that underlie the network response to stimuli as a whole, we propose to study this phenomenon through a network of integrate-and-fire neurons endowed with intrinsic homeostasis and spike-timing dependent plasticity mechanisms. The constructed system was explored, aiming to determine in which conditions the network could behave as the real system, and trained in a way similar as the experimental one done by Johnson et al. (2010).

Recentemente observou-se experimentalmente, Johnson et al. (2010), que fatias organotípicas corticais de rato são capazes de completar padrões espaço-temporais, após serem treinadas. Embora se especule que mecanismos de plasticidade sináptica e homeostática estejam por trás do fenômeno, ainda não existe nenhuma explicação detalhada sobre o assunto. Com o intuito de propor uma explicação clara e consistente para os mecanismos que permeiam a resposta da rede aos estímulos como um todo, nos propomos a estudar este fenômeno por meio de uma rede de neurônios de integração-e-disparo dotada de mecanismos de homeostase intrínseca e de plasticidade sináptica dependente de disparos. O sistema construído foi explorado, de modo a determinar em que condições a rede poderia comportar-se como o sistema real, e treinado de forma similar `a realizada experimentalmente por Johnson et al. (2010). / Recently it has been observed experimentally, Johnson et al. (2010), that organotypic cortical slices of rat are capable of completing spatio-temporal patterns after training. Although it is speculated that synaptic and homeostatic plasticity may have an important role in this phenomenon, there is still no detailed explanation about this subject. In order to propose a clear and consistent explanation for the mechanisms that underlie the network response to stimuli as a whole, we propose to study this phenomenon through a network of integrate-and-fire neurons endowed with intrinsic homeostasis and spike-timing dependent plasticity mechanisms. The constructed system was explored, aiming to determine in which conditions the network could behave as the real system, and trained in a way similar as the experimental one done by Johnson et al. (2010).

O estudo de redes complexas tem despertado muita atenção nos últimos anos, principalmente pela sua capacidade de permitir a análise dos mais diversificados sistemas através de um mesmo conjunto de ferramentas matemáticas e computacionais. Até pouco tempo a ênfase nessa área era sobre o estudo das propriedades estruturas e sua influência em característica globais da dinâmica ocorrendo sobre o sistema. Recentemente foi-se percebendo a riqueza de comportamentos que podemos estudar ao olharmos para grupos de vértices e as interações que ocorrem entre esses grupos, de forma que cada vez mais se torna necessária a criação de uma forma sistemática de quantificar, a nível de nó, como uma dinâmica é influenciada (ou diferenciada) pela estrutura do sistema. Apresentamos nesse trabalho um primeiro passo nessa direção, no qual definimos uma medida que, baseada em características dinâmicas obtidas em diversas condições iniciais, busca comparar o comportamento dos nós presentes no sistema. Através dessa medida encontramos a alta capacidade de modelos de rede geográficas de produzir, por simples flutuações estatísticas, grupos de nós de dinâmica muito distinta em relação ao demais, um fato que não ocorre para uma rede definida de forma semelhante mas não geográfica. Nessas redes conseguimos verificar também se um nó de topologia local distinta dos demais consegue possuir uma dinâmica diferenciada, o que encontramos foi que mesmo em regiões da rede extremamente regulares existe a formação de grupos dinâmicos devido ao efeito da borda dessa região. Saindo de modelos de redes para apresentar aplicações práticas do método, encontramos na rede neuronal do verme Caenorhabditis elegans um conjunto de nós de alta diferenciação dinâmica, que representam os interneurônios do cordão ventral do verme. Adicionalmente encontramos na rede cortical do macaco da família Cercopithecidae uma divisão em relação a regularidade dos sinais dinâmicos, o que indica a presença de comunidades funcionais nessa rede. Além da metodologia de diferenciação desenvolvemos ainda uma ferramenta que busca encontrar de forma totalmente automatizada as características dinâmicas mais relevantes do sistema em estudo, que foi capaz de representar medidas dinâmicas tradicionalmente utilizadas na área. Todos esses resultados abrem caminho para diversas vertentes de estudo, em especial citamos a influência de uma borda irregular no interior de uma região regular, o estudo da rede Caenorhabditis elegans com dinâmicas neuronais mais precisas e a aplicação sistemática de dinâmicas para encontrar a divisão funcional de comunidades em redes direcionadas, um tema que apresenta resultados promissores. / The study of complex networks has drawn much attention over the last years, mainly by virtue of its potential to characterize the most diverse systems through the same mathematical and computational tools. Not long ago the emphasis on this field mostly focused on the effects of the structural properties on the global behavior of a dynamical process taking place in the system. Recently, some studies started to unveil the richness of interactions that occur between groups of nodes when we look at the small scale of interactions occurring in the network. Such findings call for a new systematic methodology to quantify, at node level, how a dynamics is being influenced (or differentiated) by the structure of the underlying system. Here we present a first step towards this direction, in which we define a new measurement that, based on dynamical characteristics obtained for a series of initial conditions, compares the dynamical behavior of the nodes present in the system. Through this measurement we find the high capacity of networks generated by geographical models to exhibit, by means of statistical fluctuations, groups of nodes with very distinct dynamics compared to the rest of the network, a behavior that does not occur for a similar non-geographical network. We also verify if a large topological differentiation of a node necessarily reflects on its dynamics. We find that even in very regular regions of the network the nodes tend to form dynamical groups influenced by the border effects. In addition to the network models used, we present practical applications of the methodology by using the neuronal network of the nematode Caenorhabditis elegans, where we show that the interneurons of the ventral cord presents a very large dynamical differentiation when compared to the rest of the network. We also analyze the cortical network of the Cercopithecidae monkey by means of signal communication complexity, finding that it contains two well-defined functional communities. Besides the differentiation measurement, we also presents an useful mechanism for automatically obtaining the relevant dynamical characteristics of the nodes, which showed promising results by obtaining traditional measurements of the area with little effort. All these results pave the way to a range of different studies, of which we highlight the influence of an irregular border on the dynamics taking place inside a regular network region, the study of the neurons of Caenorhabditis elegans using more robust neuronal dynamics and the systematic application of different dynamics in order to find the functional community division of directed networks.

A stochastic leaky integrate-and-fire neuron model was implemented in this study to simulate the spiking activity of the electrosensory "P-unit" receptor neurons of the weakly electric fish Apteronotus leptorhynchus. In the context of sensory coding, these cells have been previously shown to respond in experiment to natural random narrowband signals with either a linear or nonlinear coding scheme, depending on the intrinsic firing rate of the cell in the absence of external stimulation. It was hypothesised in this study that this duality is due to the relation of the stimulus to the neuron's excitation threshold. This hypothesis was validated with the model by lowering the threshold of the neuron or increasing its intrinsic noise, or randomness, either of which made the relation between firing rate and input strength more linear. Furthermore, synchronous P-unit firing to a common input also plays a role in decoding the stimulus at deeper levels of the neural pathways. Synchronisation and desynchronisation between multiple model responses for different types of natural communication signals were shown to agree with experimental observations. A novel result of resonance-induced synchrony enhancement of P-units to certain communication frequencies was also found.