O objetivo deste trabalho é estudar a estrutura dos módulos sobre uma álgebra trivial de dimensão dois na variedade M das álgebras comutativas de potências associativas. Em particular classificamos os módulos irredutíveis. Estes resultados nos permitem compreender melhor a estrutura das nilálgebras comutativas de dimensão finita e nilíndice 4. Finalmente classificamos, sob isomorfismos, as nilálgebras comutativas de potências associativas de dimensão n e nilíndice n. / The aim of this work is to study the structure of the modules over a trivial algebra of dimension two in the variety M of commutative and power-associative algebras. In particular we classify the irreducible modules. These results enables us to understand better the structure of finite-dimensional power-associative nilalgebras of nilindex 4. Finally, we classify, up to isomorphism, commutative power associative nilalgebras of nilindex n and dimension n.

Comment les différentes formes de mémoire sont-elles encodées et comment interagissent-elles ? L'identification des réseaux neuronaux sous-jacents aux différentes formes de mémoire est une approche performante qui permet de mieux comprendre la dynamique des phases de mémoire et leurs interactions. La Drosophile représente un modèle de choix pour l'étude des mécanismes d'apprentissage et de mémorisation. Son cerveau est composé de 100.000 neurones et des outils génétiques permettent d'étudier fonctionnellement les circuits neuronaux avec une résolution proche du neurone unique. On utilise deux sortes d’apprentissage olfactif associatif : l’un, appétitif, repose sur l’association entre une odeur etdu sucre, et l’autre, aversif, associe une odeur et des chocs électriques. On distingue deux types de mémoires appétitives, une à court-terme et une à long-terme. Notre équipe a montré récemment que ces mémoires appétitives se forment indépendamment. La mémoire aversive a été décomposée en deux mémoires labiles à court-terme et à moyen terme, et deux mémoires consolidées : la mémoire à long terme (LTM) qui repose sur une synthèse protéique de novo et la mémoire résistante à l’anesthésie (ARM). Les corps pédonculés (CP), le centre de l'apprentissage et de la mémoire olfactive de la Drosophile, forment une structure bilatérale d’environ 2.000 neurones par hémisphère cérébral, appelés cellules de Kenyon. Celles-ci ont été classées en trois sous-types–γ, αβ et α’β’–en fonction des ramifications de leurs projections axonales qui définissent trois lobes médians (β, β’, γ) et deux verticaux (α, α’). Certains réseaux neuronaux responsables du rappel de la mémoire ont été caractérisés mais la correspondance entre ces réseaux et les phases de mémoire n’est encore que parcellaire. Grâce à des outils thermogénétiques nous pouvons inhiber une population neuronale restreinte pendant la phase de rappel spécifiquement. Au cours de ma thèse j'ai affiné la dissection des différentes phases de mémoire olfactive tant au niveau comportemental qu'au niveau des réseaux neuronaux. Nous avons pu attribuer à chaque phase de mémoire un réseau neuronal précis allant des neurones intrinsèques des CP aux neurones efférents impliqués dans le rappel de la mémoire. Nous avons ainsi mis en évidence une distribution spatio-temporelle de six phases de mémoires aversives mettant en jeu des réseaux neuronaux distincts et complémentaires. Nous avons montré que la ARM, jusqu’ici considérée comme une phase de mémoire unique est en réalité une mémoire composite. Nous avons identifié et localisé trois formes de ARM : la ARM immédiate, la ARM à moyen terme et la ARM à long-terme localisées respectivement dans les neurones αβ, γ et α’β’. Les deux formes séquentielles de mémoire labile immédiate et à moyen-terme ont été localisées dans les neurones γ et αβ respectivement. Nous avons aussi identifié les neurones extrinsèques de rappel de ces différentes formes de mémoire. Nous avons montré que les neurones MB-V2, précédemment décrits par notre équipe, assurent le rappel des mémoires localisées dans les neurones αβ et α’β’. Nous avons également identifié un nouveau type de neurones impliqués dans le rappel mnésique, les neurones MB-M6, efférents aux lobes γ et β’. Il existe un neurone MB-M6 par hémisphère. Ceux-ci sont impliqués dans le rappel des phases de mémoire aversive impliquant les neurones γ et α’β’. Nous avons aussi montré que les neurones MB-M6 sous tendent le rappel de la mémoire appétitive à court-terme. Ces découvertes permettent de dresser un tableau complet des circuits de sortie des corps pédonculés sous-tendant chaque phase de mémoire. Ce schéma global laisse apparaître qu’à un temps donné, deux formes de mémoire distinctes ne partagent jamais le même circuit. Nos travaux ouvrent la voie à l’étude comparée des modifications physiologiques encodant les différentes formes de mémoire associative chez la Drosophile. / A fly can form robust aversive associative olfactory memory after pairing an odor with electric shocks. Appetitive memory forms in a starved fly after pairing an odor with sugar delivery. Both types of olfactory memory rely on the mushroom bodies (MBs), a paired lobed structure of ∼2,000 neurons—the Kenyon cells (KCs)—per brain hemisphere. Based on their axonal morphology, KCs are classed into three different subtypes: axons from α/β and α′/β′ KCs branch into vertical (α and α′) and medial (β and β′) lobes, whereas axons from γ neurons form only a medial γ lobe. Drosophila can form two types of consolidated memories: LTM, which relies on de novo protein synthesis, and ARM, which does not. Droophila can also sequetially forms an immediate labile memory and a middle term labile memory.We have demonstrated a spatio-temporal distribution of six phases of aversive memories involving distinct and complementary neural networks. We have identified and located three forms of ARM: immediate ARM, middle term ARM and long-term ARM located respectively in αβ neurons, γ and α'β’. Both sequential forms of immediate labile memory and middle term memory were localized in γ and αβ neurons respectively. We also identified the extrinsic neurons required for the retrieval of these different forms of memory. We showed that MB-V2 neurons, previously described by our team, ensure the recall of memories localized in αβ and α'β’ neurons. We identified a specific pair of MB-efferent neurons, named M6 neurons. These are involved in the retrieval of aversive memories involving γ and α'β ‘ neurons. We have also shown that MB-M6 neurons are required for the retrieval of appetitive short-term memory.

Les méthodes non-invasives de neuro-imagerie et de neurostimulation peuvent être combinées pour mieux comprendre les connexions dans le cerveau. Pour la première fois, une étude combine de façon séquentielle l’IRM fonctionnelle (fMRI) et un protocole de TMS associative pairée cortico-corticale (TMS-PAScc) sur le cortex moteur primaire (M1) et sur le lobule pariétal inférieur (LPI) dans l’hémisphère gauche. La TMS module-t-elle le couplage neurovasculaire et permet-elle de renforcer une connexion fonctionnelle qui soit détectable à la fMRI à l’état de repos (RS-fMRI)? 10 sujets droitiers et en santé font une session de TMS-PAScc LPI-M1 de courte durée (180 paires d’impulsions, fréquence de stimulation à 0.02 Hz). Les mêmes sujets font 2 sessions de la RS-fMRI, avant et après le protocole PAScc. Les résultats montrent que la corrélation du signal BOLD entre les régions LPI-M1 avant et après la PAScc ne change pas de façon significative (avant-PAS=0.10±0.07 et après-PAS=0.09±0.07, p=0.64), tout comme l’amplitude des potentiels évoqués moteurs (PEM) des impulsions pairées LPI-M1 ne change pas de façon significative du début de la PAScc à 25 minutes après la PAScc (PASdébut=0.71±0.46mV, PASpost25min=0.72±0.89mV, p=0.338). Toutefois, les PEM des impulsions pairées LPI-M1 sont réduites par rapport aux PEM des impulsions simples M1, avant la PAScc et après la PAScc (PEM simples_pré et PASdébut, réduction de 0.32mV, p=0.05; PEM simples_post et PASpost25min, réduction de 0.39mV p=0.008), illustrant la présence d’un lien fonctionnel de nature inhibitrice entre LPI et M1. Toutefois, l’amplitude de cette inhibition n’est pas modulée de façon significative par la TMS-PAScc (ratio mesures pairées/mesures simples préPAS=0.9 et ratio postPAS=0.6, p=0.257). Dans l’ensemble, la TMS-PAScc ne montre pas d’effet soutenu sur la connectivité cérébrale telle que mesurée par la RS-fMRI et la TMS et ce, bien que le LPI montre un lien inhibiteur sur M1 de façon aigue. Plusieurs hypothèses peuvent expliquer cette absence d’effet soutenu, notamment, il est possible que l’altération de la connectivité ne soit visible que lorsque le réseau LPI-M1 est activement sollicité, comme durant l’exécution d’une tâche motrice. Il est aussi possible que le nombre de pairages soit insuffisant pour induire des changements mesurables, mais que la connectivité fonctionnelle suite à des sessions répétées de protocole PAScc pourrait modifier le couplage neurovasculaire et la plasticité cérébrale. / Abstract : Noninvasive neuroimagery and neurostimulation methods can be combined to further the understanding of the human brain connections. For the first time, resting state functional MRI (RS-fMRI) and paired associative cortico-cortical TMS (TMS-PAScc) of the motor cortex (M1) and the cortex of the inferior parietal lobule (LPI) of the left hemisphere are combined in a serial manner. Is TMS able to modify the neurovascular coupling as to facilitate LPI-M1 functional connectivity and change the fMRI BOLD signal? 10 right-handed and healthy subjects did a LPI-M1 TMS-PAScc session of short duration (180 paired pulses at 0.02 Hz, 15 min total). The same subjects underwent 2 fMRI sessions, before and after TMS-PAScc LPI-M1. Results show that the BOLD signal correlation between LPI-M1 does not change significantly before and after PAS (prePAS=0.10±0.07 et postPAS=0.09±0.07, p=0.64). TMS measures of motor evoked potentials (PEM) were taken before and after PAS LPI-M1. The paired pulse PEM measures did not change significantly from the start of PAScc to 25 minutes postPAS (PASstart=0.71 ± 0.46 mV, PASpost25min=0.72±0.89 mV, p=0.338). Paired PEM measures are statistically reduced from PAS PEM single measures, before and afterPAS (sPEM_pre et PASstart, significant 0.32mV reduction, p=0.05; PEMs_post et PASpost25min, 0.39mV reduction, p=0.008). PAScc did not show any significant neuroplasticity effect after 20 minutes because paired pulses did not change before and after PAScc. The PEM reduction of paired pulses is most likely related to the inhibiting effect of the conditioning stimulus of LPI on the test stimulus of M1 at 8ms. This inhibition is an effect limited to the measure itself and does not increase significantly with time (pairedpulse/singlepulsemeasures prePASratio=0.9 and postPASratio=0.6, p=0.257). TMSPAScc did not show a sustained effect on cerebral connectivity as measured by RS-fMRI although stimulation of LPI showed an acute inhibiting effect on M1 during paired measures. LPI-M1 TMS-PAScc did not show sustained connectivity and it could be because no task was involved in our study to actively solicit both cerebral regions during PAS. It is also possible that the number of paired stimulation was not enough to bring a change of connectivity and that PAS needs to be repeated on different days to eventually have a sustainable effect.

Notre travail s'inscrit dans l'intersection de deux disciplines qui sont la Théorie des Fonctions de Croyance (TFC) et la fouille de données. L'interaction pouvant exister entre la TFC et la fouille de données est étudiée sous deux volets.La première interaction souligne l'apport des règles associatives génériques au sein de la TFC. Nous nous sommes intéressés au problème de fusion de sources non fiables dont la principale conséquence est l'apparition de conflit lors de la combinaison. Une approche de gestion de conflit reposant sur les règles d'association génériques appelé ACM a été proposée.La deuxième interaction s'intéresse aux bases de données imparfaites en particulier les bases de données évidentielles. Les informations, représentées par des fonctions de masse, sont étudiées afin d'extraire des connaissances cachées par le biais des outils de fouille de données. L'extraction des informations pertinentes et cachées de la base se fait grâce à la redéfinition de la mesure du support et de la confiance. Ces mesures introduites ont été les fondements d'un nouveau classifieur associatif que nous avons appelé EDMA. / This thesis explores the relation between two domains which are the Belief Function Theory (BFT) and data mining. Two main interactions between those domain have been pointed out.The first interaction studies the contribution of the generic associative rules in the BFT. We were interested in managing conflict in case of fusing conflictual information sources. A new approach for conflict management based on generic association rules has been proposed called ACM.The second interation studies imperfect databases such as evidential databases. Those kind of databases, where information is represented by belief functions, are studied in order to extract hidden knowledges using data mining tools. The extraction of those knowledges was possible thanks to a new definition to the support and the confidence measures. Those measures were integrated into a new evidential associative classifier called EDMA.

La progression de l’espérance de vie observée au cours du XXième siècle a été accompagnée par une augmentation massive de l’incidence des maladies liées à l’âge et en particulier des maladies neurodégénératives. Malheureusement, les thérapeutiques actuelles ciblant principalement les anomalies d’agrégation protéique caractérisant ces maladies, tel que la maladie d’Alzheimer, ont échoué au niveau des essais cliniques. De récentes études épidémiologiques ont suggéré un lien entre la dysfonction métabolique et les maladies neurodégénératives. Par conséquent, une approche alternative pour développer des nouveaux médicaments serait se cibler les voies de signalisation métaboliques perturbées dans les modèles de maladie d’Alzheimer. L’AMPK (AMP activated protein kinase) est une enzyme activée par les bas niveaux d’énergie cellulaire via la détection du taux AMP : ATP. Une fois activée, l’AMPK allonge la durée de vie d’organismes modèles et protège contre le développement de pathologies liées à l’âge telle que les maladies neurodégénératives. De plus, l’AMPK régule l’homéostasie mitochondriale et les réseaux mitochondriaux chez les mammifères. Cependant, il reste à savoir si l’AMPK protège contre le développement de pathologies neurodégénératives via la régulation de la structure mitochondriale. Lors de ces travaux, nous avons utilisé un protocole d’apprentissage et de mémoire chez C. elegans pour mesurer la fonction neuronale. Nous avons montré que les nématodes exprimant le peptide amyloïde Aβ1-42 dans les neurones avait une capacité d’apprentissage détériorée. Ce déficit a pu être restauré par l’activation constitutionnelle de l’AMPK. Nous montrons également que l’activation de l’AMPK améliore les capacités d’apprentissage des nématodes sauvages en induisant la fusion des mitochondries. En effet, les vers mutés pour le gène responsable de la fusion mitochondriale ont une capacité d’apprentissage diminuée, laquelle peut être restaurée par le rétablissement de la fusion mitochondriale, spécifiquement dans les neurones. Des résultats supplémentaires suggèrent que l’AMPK induirait ses effets bénéfiques sur la fonction neuronale en inhibant le facteur de transcription CRTC-1 (CREB-regulated transcriptional co-activator 1). Nos résultats tendent à montrer que cibler le métabolisme cellulaire neuronal représenterait une option thérapeutique viable afin de maintenir les fonctions neuronales dans le cadre de pathologies neurodégénératives. / The dramatic increase in life expectancy during the 20th century was accompanied by a resultant epidemic of age-related pathologies including neurodegenerative diseases. Unfortunately, current therapeutics primarily focusing on protein misfolding aspects of diseases such as Alzheimer’s Disease (AD) have been unsuccessful in the clinical trials. Recent epidemiological studies have suggested a strong association between metabolic dysfunction and neurodegeneration. Therefore, an alternative approach is to target metabolic pathways disrupted in AD models for therapeutics. AMP activated protein kinase (AMPK) is activated in a low energy state via sensing the AMP: ATP ratio. Once active, AMPK promotes longevity in model organism and protects against a wide range of age related diseases including neurodegenerative diseases. In addition, AMPK regulates mitochondrial homeostasis and mitochondrial networks in mammals. However, whether mitochondrial regulation causally links AMPK to protection against neurodegenerative disease is unknown. Here we use a learning and memory protocol in C. elegans as readout of neuronal function. We show that nematodes expressing the toxic amyloid peptide Aβ1-42 in the neurons display impaired learning ability, which can be rescued by constitutive activation of AMPK (CA-AMPK). We further show that CA-AMPK enhances learning ability in wild type nematodes by promoting mitochondrial fusion. Indeed, fusion deficient worms show impaired learning, which can be rescued by restoring mitochondrial fusion specifically in the neurons. Additional results suggest that AMPK might promote its beneficial effects on neuronal function via inhibition of CREBregulated transcriptional co-activator 1 (CRTC-1). Our results show that targeting neuronal metabolism may be a viable therapeutic option to restore neuronal function in the context of neurodegenerative diseases.

Building an accurate and reliable model for prediction for different application domains, is one of the most significant challenges in knowledge discovery and data mining. This thesis focuses on building and enhancing a generic predictive model for estimating a future value by extracting association rules (knowledge) from a quantitative database. This model is applied to several data sets obtained from different benchmark problems, and the results are evaluated through extensive experimental tests. The thesis presents an incremental development process for the prediction model with three stages. Firstly, a Knowledge Discovery (KD) model is proposed by integrating Fuzzy C-Means (FCM) with Apriori approach to extract Fuzzy Association Rules (FARs) from a database for building a Knowledge Base (KB) to predict a future value. The KD model has been tested with two road-traffic data sets. Secondly, the initial model has been further developed by including a diversification method in order to improve a reliable FARs to find out the best and representative rules. The resulting Diverse Fuzzy Rule Base (DFRB) maintains high quality and diverse FARs offering a more reliable and generic model. The model uses FCM to transform quantitative data into fuzzy ones, while a Multiple Support Apriori (MSapriori) algorithm is adapted to extract the FARs from fuzzy data. The correlation values for these FARs are calculated, and an efficient orientation for filtering FARs is performed as a post-processing method. The FARs diversity is maintained through the clustering of FARs, based on the concept of the sharing function technique used in multi-objectives optimization. The best and the most diverse FARs are obtained as the DFRB to utilise within the Fuzzy Inference System (FIS) for prediction. The third stage of development proposes a hybrid prediction model called Fuzzy Associative Classification Rule Mining (FACRM) model. This model integrates the ii improved Gustafson-Kessel (G-K) algorithm, the proposed Fuzzy Associative Classification Rules (FACR) algorithm and the proposed diversification method. The improved G-K algorithm transforms quantitative data into fuzzy data, while the FACR generate significant rules (Fuzzy Classification Association Rules (FCARs)) by employing the improved multiple support threshold, associative classification and vertical scanning format approaches. These FCARs are then filtered by calculating the correlation value and the distance between them. The advantage of the proposed FACRM model is to build a generalized prediction model, able to deal with different application domains. The validation of the FACRM model is conducted using different benchmark data sets from the University of California, Irvine (UCI) of machine learning and KEEL (Knowledge Extraction based on Evolutionary Learning) repositories, and the results of the proposed FACRM are also compared with other existing prediction models. The experimental results show that the error rate and generalization performance of the proposed model is better in the majority of data sets with respect to the commonly used models. A new method for feature selection entitled Weighting Feature Selection (WFS) is also proposed. The WFS method aims to improve the performance of FACRM model. The prediction performance is improved by minimizing the prediction error and reducing the number of generated rules. The prediction results of FACRM by employing WFS have been compared with that of FACRM and Stepwise Regression (SR) models for different data sets. The performance analysis and comparative study show that the proposed prediction model provides an effective approach that can be used within a decision support system.

La formation et le stockage à long terme des souvenirs mettent en jeu le processus de consolidation mnésique. S’il est maintenant bien admis que ce processus requiert une interaction entre la formation hippocampique et différentes régions corticales dépositaires des souvenirs, les mécanismes qui sous-tendent ce dialogue restent encore mal connus. En combinant chez le rat des approches comportementale, d’imagerie cellulaire et d’inactivations pharmacologiques des voies de signalisation intracérébrales impliquées dans l’épigenèse, nous avons cherché dans ce travail de thèse à élucider certains des mécanismes responsables de la formation des souvenirs au niveau cortical. Dans la perspective de pouvoir appréhender de façon temporellement précise le dialogue hippocampo-cortical au cours du processus de consolidation à l’échelle systémique, notre premier objectif a été de valider une épreuve comportementale adaptée à l’étude de ce processus, la transmission sociale de préférence alimentaire. Nos résultats ont montré que cette tâche, qui ne nécessite qu’une phase d’acquisition ponctuelle, induit une mémoire robuste et durable. Cette mémoire s’appuie sur des stimuli olfactifs de nature non spatiale. Son caractère associatif nécessite l’implication fonctionnelle de l’hippocampe et de régions corticales spécifiques comme le cortex orbitofrontal qui joue un rôle crucial dans le traitement d’informations de nature olfactive. Dans une deuxième série d’expériences, une approche d’imagerie cellulaire utilisant le facteur de transcription c-fos couplée à une approche pharmacologique d’inactivation transitoire région-spécifique a révélé le rôle crucial du cortex orbitofrontal dans le rappel d’informations anciennes (délai de 30 jours) mais pas récentes (délai de 1 jour). Nous avons par ailleurs mis en évidence que la consolidation des informations dans cette structure s’accompagnait de changements progressifs de l’architecture des réseaux neuronaux comme la formation de nouvelles synapses (synaptogénèse) ou l’augmentation du nombre d’épines dendritiques. En accord avec le modèle standard de la consolidation mnésique, ce recrutement cortical était associé à un désengagement de l’hippocampe, confirmant le rôle transitoire de cette structure dans le rappel à long terme d’informations olfactives associatives. Dans une troisième série d’expériences, nous nous sommes intéressés aux mécanismes pouvant sous-tendre l’établissement de la mémoire à long terme au niveau cortical. Nos résultats apportent un éclairage nouveau sur la dynamique des interactions hippocampo-corticales pendant la consolidation systémique en démontrant la nécessité d’un étiquetage des assemblées neuronales du cortex orbitofrontal dès l’encodage des informations. Un blocage de cet étiquetage par une inactivation de ce cortex au moment de l’interaction sociale (phase d’acquisition) a perturbé le rappel à long terme et empêché les modifications de l’architecture des réseaux neuronaux corticaux normalement associés au stockage à long terme des informations olfactives. Sur le plan cellulaire, cet étiquetage requiert l’activation des récepteurs NMDA et de la voie des MAPK, ainsi que l’acétylation des protéines histones impliquées dans la régulation de l’état transcriptionnel de la chromatine. En modifiant leur état d’acétylation, nous avons pu moduler positivement ou négativement le rappel à long terme des informations olfactives relatives à la préférence alimentaire. Ainsi, nos données soulignent l’importance du dialogue hippocampo-cortical dans l’établissement de la mémoire à long terme. / Abstract :

This thesis investigates the potential of wafer-scale integration (WSI) for the implementation of low-cost fine-grain parallel processing computer systems. As WSI is a relatively new subject, there was little work on which to base investigations. Indeed, most WSI architectures existed only as untried and sometimes vague proposals. Accordingly, the research strategy approached this problem by identifying a representative WSI structure and architecture on which to base investigations. An analysis of architectural proposals identified associative memory to be general purpose parallel processing component used in a wide range of WSI architectures. Furthermore, this analysis provided a set of WSI-level design requirements to evaluate the sustainability of different architectures as research vehicles. The WSI-ASP (WASP) device, which has a large associative memory as its main component is shown to meet these requirements and hence was chosen as the research vehicle. Consequently, this thesis addresses WSI potential through an in-depth investigation into the feasibility of implementing a large associative memory for the WASP device that meets the demanding technological constraints of WSI. Overall, the thesis concludes that WSI offers significant potential for the implementation of low-cost fine-grain parallel processing computer systems. However, due to the dual constraints of thermal management and the area required for the power distribution network, power density is a major design constraint in WSI. Indeed, it is shown that WSI power densities need to be an order of magnitude lower than VLSI power densities. The thesis demonstrates that for associative memories at least, VLSI designs are unsuited to implementation in WSI. Rather, it is shown that WSI circuits must be closely matched to the operational environment to assure suitable power densities. These circuits are significantly larger than their VLSI equivalents. Nonetheless, the thesis demonstrates that by concentrating on the most power intensive circuits, it is possible to achieve acceptable power densities with only a modest increase in area overheads.

In this thesis we use computational neural network models to examine the dynamics and functionality of the CA3 region of the mammalian hippocampus. The emphasis of the project is to investigate how the dynamic control structures provided by inhibitory circuitry and cellular modification may effect the CA3 region during the recall of previously stored information. The CA3 region is commonly thought to work as a recurrent auto-associative neural network due to the neurophysiological characteristics found, such as, recurrent collaterals, strong and sparse synapses from external inputs and plasticity between coactive cells. Associative memory models have been developed using various configurations of mathematical artificial neural networks which were first developed over 40 years ago. Within these models we can store information via changes in the strength of connections between simplified model neurons (two-state). These memories can be recalled when a cue (noisy or partial) is instantiated upon the net. The type of information they can store is quite limited due to restrictions caused by the simplicity of the hard-limiting nodes which are commonly associated with a binary activation threshold. We build a much more biologically plausible model with complex spiking cell models and with realistic synaptic properties between cells. This model is based upon some of the many details we now know of the neuronal circuitry of the CA3 region. We implemented the model in computer software using Neuron and Matlab and tested it by running simulations of storage and recall in the network. By building this model we gain new insights into how different types of neurons, and the complex circuits they form, actually work. The mammalian brain consists of complex resistive-capacative electrical circuitry which is formed by the interconnection of large numbers of neurons. A principal cell type is the pyramidal cell within the cortex, which is the main information processor in our neural networks. Pyramidal cells are surrounded by diverse populations of interneurons which have proportionally smaller numbers compared to the pyramidal cells and these form connections with pyramidal cells and other inhibitory cells. By building detailed computational models of recurrent neural circuitry we explore how these microcircuits of interneurons control the flow of information through pyramidal cells and regulate the efficacy of the network. We also explore the effect of cellular modification due to neuronal activity and the effect of incorporating spatially dependent connectivity on the network during recall of previously stored information. In particular we implement a spiking neural network proposed by Sommer and Wennekers (2001). We consider methods for improving associative memory recall using methods inspired by the work by Graham and Willshaw (1995) where they apply mathematical transforms to an artificial neural network to improve the recall quality within the network. The networks tested contain either 100 or 1000 pyramidal cells with 10% connectivity applied and a partial cue instantiated, and with a global pseudo-inhibition.We investigate three methods. Firstly, applying localised disynaptic inhibition which will proportionalise the excitatory post synaptic potentials and provide a fast acting reversal potential which should help to reduce the variability in signal propagation between cells and provide further inhibition to help synchronise the network activity. Secondly, implementing a persistent sodium channel to the cell body which will act to non-linearise the activation threshold where after a given membrane potential the amplitude of the excitatory postsynaptic potential (EPSP) is boosted to push cells which receive slightly more excitation (most likely high units) over the firing threshold. Finally, implementing spatial characteristics of the dendritic tree will allow a greater probability of a modified synapse existing after 10% random connectivity has been applied throughout the network. We apply spatial characteristics by scaling the conductance weights of excitatory synapses which simulate the loss in potential in synapses found in the outer dendritic regions due to increased resistance. To further increase the biological plausibility of the network we remove the pseudo-inhibition and apply realistic basket cell models with differing configurations for a global inhibitory circuit. The networks are configured with; 1 single basket cell providing feedback inhibition, 10% basket cells providing feedback inhibition where 10 pyramidal cells connect to each basket cell and finally, 100% basket cells providing feedback inhibition. These networks are compared and contrasted for efficacy on recall quality and the effect on the network behaviour. We have found promising results from applying biologically plausible recall strategies and network configurations which suggests the role of inhibition and cellular dynamics are pivotal in learning and memory.

Conventional functional electrical stimulation (FES) therapy, if provided shortly after an incomplete spinal cord injury, is able to help an individual to restore voluntary hand function. This is thought to occur through the induction of neuroplasticity. However, conventional FES therapy employs a push-button-based control scheme, which does not fully require the recipient to generate volitional movements. The first study in this thesis therefore sought to determine, in an early proof-of-concept test with able-bodied participants, whether control strategies which are triggered by volitional activity (including an electroencephalography-based brain-machine interface (BMI-FES) and an electromyogram-based control scheme (EMG-FES)) might provide greater benefits to hand function. The results offer relatively weak evidence to suggest that BMI-FES, and especially EMG-FES, were able to induce greater neuroplasticity than conventional treatments in the corticospinal tract leading to the hands, but that this did not immediately translate to more functional improvements such as maximum grip force. ii The second study in this thesis focussed on spinal associative stimulation (SAS), which involves paired stimulation pulses at both the head (via transcranial magnetic stimulation), and the wrist (via peripheral nerve stimulation). The purpose of this, as with the first study, was to induce neuroplasticity and upregulate the corticospinal tract leading to the hands. While limited research has suggested that it is possible to produce neuroplasticity through SAS, all such studies have provided stimulation at a fixed frequency of 0.1 or 0.2 Hz. The present study therefore sought to compare the effectiveness of a typical 0.1 Hz paradigm with a 1 Hz paradigm, and a paradigm which provided stimulation in 5 Hz “bursts”. None of the paradigms were able to successfully induce neuroplasticity in a consistent manner. The increased variability in this study as compared to the previous one, despite the nearly identical assessment methodology, suggests that responses to the SAS treatment may have been highly individual. This serves to highlight a potential limitation of the treatment, which is that its effectiveness may not be universal, but rather dependent on each specific recipient. This may be a challenge faced by SAS should it continue to be tested as a novel therapy.