The amygdala and hippocampus interact with thalamocortical systems to regulate cognitive-emotional learning, and lesions of amygdala, hippocampus, thalamus, and cortex have different effects depending on the phase of learning when they occur. In examining eyeblink conditioning data, several questions arise: Why is the hippocampus needed for trace conditioning where there is a temporal gap between the conditioned stimulus offset and the onset of the unconditioned stimulus, but not needed for delay conditioning where stimuli temporally overlap and co-terminate? Why do amygdala lesions made before or immediately after training decelerate conditioning while those made later have no impact on conditioned behavior? Why do thalamic lesions degrade trace conditioning more than delay conditioning? Why do hippocampal lesions degrade recent learning but not temporally remote learning? Why do cortical lesions degrade temporally remote learning, and cause amnesia, but not recent or post-lesion learning? How is temporally graded amnesia caused by ablation of medial prefrontal cortex? How are mechanisms of motivated attention and the emergent state of consciousness linked during conditioning? How do neurotrophins, notably Brain Derived Neurotrophic Factor (BDNF), influence memory formation and consolidation? A neural model, called neurotrophic START, or nSTART, proposes answers to these questions. The nSTART model synthesizes and extends key principles, mechanisms, and properties of three previously published brain models of normal behavior. These three models describe aspects of how the brain can learn to categorize objects and events in the world; how the brain can learn the emotional meanings of such events, notably rewarding and punishing events, through cognitive-emotional interactions; and how the brain can learn to adaptively time attention paid to motivationally important events, and when to respond to these events, in a context-appropriate manner. The model clarifies how hippocampal adaptive timing mechanisms and BDNF may bridge the gap between stimuli during trace conditioning and thereby allow thalamocortical and corticocortical learning to take place and be consolidated. The simulated data arise as emergent properties of several brain regions interacting together. The model overcomes problems of alternative memory models, notably models wherein memories that are initially stored in hippocampus move to the neocortex during consolidation.

Traumatic brain injury (TBI) is a serious public health concern. Although moderate and severe forms of TBI receive considerable attention, mild TBI accounts for the majority of all injuries. The first two aims of this work used a rodent model of mild blast to simulate primary injury (damage from the blast wave only). The first aim evaluated potential changes in interneurons containing the calcium-binding proteins calretinin or parvalbumin. In addition, morphological changes in astrocytes and microglia were assessed. Brains were analyzed 48 hours and one month following exposure to single or repeated blasts, with a focus on the hippocampus due to its integral role in learning and memory. Results showed significant region-specific alterations in microglia morphology 48 hours following blast. The absence of structural alterations in microglia one month following blast indicated that the regional hippocampal vulnerability may be transient. The second aim compared glial morphologies in the retina and brain (the lateral geniculate nucleus, superior colliculus, and visual cortex) 48 hours or one month following multiple blasts. Fiber degeneration has received considerable attention, however, less is known about the status of glia throughout the visual pathway following mild blasts. Although no structural alterations were detected, it is possible that alterations in glia occurred at a more acute time scale as changes in glia can be rapid and reversible. The final aim of this work focused on the immunocytochemical characterization of tau pathology in the visual cortices of human postmortem brains with advanced chronic traumatic encephalopathy (CTE). CTE is a devastating tauopathy associated with mild, repetitive TBIs. Although visual deficits are reported in CTE, the primary visual cortex is often spared. The main hypothesis under investigation was whether visual association areas would have tau pathology, despite sparing of primary visual cortex. In addition, a sub-class of interneurons containing parvalbumin was used to evaluate a potential cell-specific vulnerability. Results showed increased tau pathology in visual association areas in advanced CTE, that was largely absent from the primary visual cortex. There was no effect on parvalbumin positive interneurons. The results of this work provides valuable insight regarding potential cell-specific resistance to CTE pathology. / 2018-11-03T00:00:00Z

Neural networks are an important tool in the field of data science as well as in the study of the very structures they were inspired from i.e. the human nervous system. In this dissertation, we studied the application of neural networks in data modeling as well as their role in studying the properties of various structures in the nervous system. This dissertation has two foci: one relates to developing methods that help improve \gls{generalization} in data models and the other is to study the possible effects of the structure on the function. As the first focus of this dissertation, we proposed a set of heuristics that improve the \gls{generalization} capability of the neural network models in regression and classification problems. To do so, we explored applying apprioi information in the form of \gls{regularization} of the behavior of the models. We used smoothness and self-consistency as the two regularized attributes that were enforced on the behavior of the neural networks in our model. We used our proposed heuristics to improve the performance neural network ensembles in regression problems (more specifically in quantitative structure–activity relationship (QSAR) modeling problems). We demonstrated that these heuristics result in significant improvements in the performance of the models we used. In addition, we developed an anomaly detection method to identify and exclude the outliers among unknown cases presented to the model. This was to ensure that the data model only made a prediction about the outcome of the unknown cases that were within its domain of applicability. This filtering resulted in further improvement of the performance of the model in our experiments. Furthermore, and through some modifications, we extended the application of our proposed heuristics to classification problems. We evaluated the performance of the resulting classification models over several datasets and demonstrated that the \gls{regularization}s we employed in our heuristics, had a positive effect on the performance of the data model across various classification problems as well. In the second part of this dissertation, we focused on studying the relationship between the structure and the functionality in the nervous system. More specifically, whether or not the structure implies functionality. In studying these possible effects, we elected to study CA3b in Hippocampus. For this reason, we used current related literature to derive a physiologically plausible model of CA3b. To make our proposed model as close as possible to its counterpart in the nervous system, we used large scale neural simulations, in excess of 45,000 neurons, in our experiments. We used the collective firings of all the neurons in our proposed structure to produce a time series signal. We considered this time-series signal which is a way to demonstrate the overall output of the structure should it be monitored by an EEG probe as the output of the structure. In our simulations, the structure produced and maintained a low frequency rhythm. We believe that this rhythm is similar to the Theta rhythm which occurs naturally in CA3b. We used the fundamental frequency of this rhythm in our experiments to quantify the effects of modifications in the structure. That is, we modified various properties of our CA3b and measured the changes in the fundamental frequency of the signal. We conducted various experiments on the structural properties (the length of axons of the neurons, the density of connections around the neurons, etc.) of the simulated CA3b structure. Our results show that the structure was very resilient to such modifications. Finally, we studied the effects of lesions in such a resilient structure. For these experiments, we introduced two types of lesions: many lesions of small radius and a few lesions with large radii. We then increased the severity of these lesions by increasing the number of lesions in the case of former and increasing the radius of lesions in the case of the latter. Our results showed that many small lesions in the structure have a more pronounced effect on the fundamental frequency compared to the few lesions with large radii. / Graduate

Thesis (Ph.D.)--Boston University / Research in rodents and computational modeling work suggest a critical role for the hippocampus in representing overlapping memories. This thesis tested predictions that the hippocampus is important in humans for remembering overlapping spatial events, and that flexible navigation of spatial routes is supported by key prefrontal and striatal structures operating in conjunction with the hippocampus. The three experiments described in this dissertation used functional magnetic resonance imaging (fMRI) in healthy young people to examine brain activity during context-dependent navigation of virtual maze environments. Experiment 1 tested whether humans recruit the hippocampus and orbitofrontal cortex to successfully retrieve well-learned overlapping spatial routes. Participants navigated familiar virtual maze environments during fMRI scanning. Brain activity for flexible retrieval of overlapping spatial memories was contrasted with activity for retrieval of distinct non-overlapping memories. Results demonstrate the hippocampus is more strongly recruited for planning and retrieval of overlapping routes than non-overlapping routes, and the orbitofrontal cortex is recruited specifically for context-dependent navigational decisions. Experiment 2 examined whether the hippocampus, orbitofrontal cortex, and striatum interact cooperatively to support flexible navigation of overlapping routes. Using a functional connectivity analysis of fMRI data, we compared interactions between these structures during virtual navigation of overlapping and non-overlapping mazes. Results demonstrate the hippocampus interacts with the caudate more strongly for navigating overlapping than non-overlapping routes. Both structures cooperate with the orbitofrontal cortex specifically during context-dependent decision points, suggesting the orbitofrontal cortex mediates translation of contextual information into the flexible selection of behavior. Experiment 3 examined whether the hippocampus and caudate contribute to forming context-dependent memories. fMRI activity for learning new virtual mazes which overlap with familiar routes was compared with activity for learning completely distinct routes. Results demonstrate both the hippocampus and caudate are preferentially recruited for learning mazes which overlap with existing route memories. Furthermore, both areas update their responses to familiar route memories which become context-dependent, suggesting complementary roles in both learning and updating overlapping representations. Together, these studies demonstrate that navigational decisions based on overlapping representations rely on a network incorporating hippocampal function with the evaluation and selection of behavior in the prefrontal cortex and striatum.

The hippocampus and prefrontal cortex play distinct roles in the generation and retrieval of episodic memory. The hippocampus is crucial for binding inputs across behavioral timescales, whereas the prefrontal cortex is found to influence retrieval. Spiking of hippocampal principal neurons contains environmental information, including information about the presence of specific objects and their spatial or temporal position relative to environmental and behavioral cues. Neural activity in the prefrontal cortex is found to map behavioral sequences that share commonalities in sensory input, movement, and reward valence. Here I conducted a series of four experiments to test the hypothesis that external inputs from cortex update cell assemblies that are organized within the hippocampus. I propose that cortical inputs coordinate with CA3 to rapidly integrate information at fine timescales. Extracellular tetrode recordings of neurons in the orbitofrontal cortex were performed in rats during a task where object valences were dictated by the spatial context in which they were located. Orbitofrontal ensembles, during object sampling, were found to organize all measured task elements in inverse rank relative to the rank previously observed in the hippocampus, whereby orbitofrontal ensembles displayed greater differentiation for object valence and its contextual identity than spatial position. Using the same task, a follow-up experiment assessed coordination between prefrontal and hippocampal networks by simultaneously recording medial prefrontal and hippocampal activity. The circuit was found to coordinate at theta frequencies, whereby hippocampal theta engaged prefrontal signals during contextual sampling, and the order of engagement reversed during object sampling. Two additional experiments investigated hippocampal temporal representations. First, hippocampal patterns were found to represent conjunctions of time and odor during a head-fixed delayed match-to-sample task. Lastly, I assessed the dependence of hippocampal firing patterns on intrinsic connectivity during the delay period versus active navigation of spatial routes, as rats performed a delayed-alternation T-maze. Stimulation of the ventral hippocampal commissure induced remapping of hippocampal activity during the delay period selectively. Despite temporal reorganization, different hippocampal populations emerged to predict temporal position. These results show hippocampal representations are guided by stable cortical signals, but also, coordination between cortical and intrinsic circuitry stabilizes flexible CA1 temporal representations.

The hippocampus is a brain structure involved in memory as well as anxiety and depression-related behavior. One unique property of the hippocampus is that adult neurogenesis occurs in this region. Rodent studies in which adult hippocampal neurogenesis is ablated have shown a role for this process in the cognitive domain, specifically in pattern separation tasks, as well as in mediating the behavioral effects of antidepressants. These studies have furnished the intriguing hypothesis that increasing adult hippocampal neurogenesis may improve these functions and therefore serve as a target for novel treatments for cognitive impairments as well as depression and anxiety disorders. Here, we use both genetic and pharmacological models to increase adult neurogenesis in mice. Under baseline conditions, we find that increasing adult hippocampal neurogenesis is sufficient to improve performance in a fear-based pattern separation task, but has no effect on exploratory, anxiety or depression-related behavior. In mice exposed to voluntary exercise, increasing adult hippocampal neurogenesis increases exploration, without affecting anxiety or depression-related behavior. Finally, in mice treated with chronic corticosterone, a model of anxiety and depression, increasing adult hippocampal neurogenesis is sufficient to prevent the behavioral effect of CORT on anxiety and depression-related behavior. Here, we therefore describe dissociations between the effects of increasing adult hippocampal neurogenesis under baseline, voluntary exercise and chronic stress conditions. Together, our results suggest that increasing adult hippocampal neurogenesis has therapeutic potential for both cognitive, and anxiety and depression-related disorders.

The astounding capacity for learning ranks among the nervous system’s most impressive features. This thesis comprises studies employing varied approaches to improve understanding, at the level of neural circuits, of the brain’s capacity for learning. The first part of the thesis contains investigations of hippocampal circuitry – both theoretical work and experimental work in the mouse Mus musculus – as a model system for declarative memory. To begin, Chapter 2 presents a theory of hippocampal memory storage and retrieval that reflects nonlinear dendritic processing within hippocampal pyramidal neurons. As a prelude to the experimental work that comprises the remainder of this part, Chapter 3 describes an open source software platform that we have developed for analysis of data acquired with in vivo Ca2+ imaging, the main experimental technique used throughout the remainder of this part of the thesis. As a first application of this technique, Chapter 4 characterizes the content of signaling at synapses between GABAergic neurons of the medial septum and interneurons in stratum oriens of hippocampal area CA1. Chapter 5 then combines these techniques with optogenetic, pharmacogenetic, and pharmacological manipulations to uncover inhibitory circuit mechanisms underlying fear learning. The second part of this thesis focuses on the cerebellum-like electrosensory lobe in the weakly electric mormyrid fish Gnathonemus petersii, as a model system for non-declarative memory. In Chapter 6, we study how short-duration EOD motor commands are recoded into a complex temporal basis in the granule cell layer, which can be used to cancel Purkinje-like cell firing to the longer duration and temporally varying EOD-driven sensory responses. In Chapter 7, we consider not only the temporal aspects of the granule cell code, but also the encoding of body position provided from proprioceptive and efference copy sources. Together these studies clarify how the cerebellum-like circuitry of the electrosensory lobe combines information of different forms and then uses this combined information to predict the complex dependence of sensory responses on body position and timing relative to electric organ discharge.

The capacity for memory is one of the most profound features of the mammalian brain, and the proper encoding and retrieval of information are the processes that form the basis of learning. The goal of this thesis is to further our understanding of the network-level mechanisms supporting learning and memory in the mammalian brain. The hippocampus has been long recognized to play a central role in learning and memory. Although being one of the most extensively studied structures in the brain, the precise circuit mechanisms underlying its function remain elusive. Principal cells in the hippocampus form complex representations of an animal's environment, but in stark contrast to the interneuron population -- and despite the apparent need for functional segregation -- these cells are largely considered a homogeneous population of coding units. Much work, however, has indicated that principal cells throughout the hippocampus, from the input node of the dentate gyrus to the output node of area CA1, differ developmentally, genetically, anatomically, and functionally. By employing in vivo two-photon calcium imaging in awake, behaving mice, we attempted to characterize the role of dened subpopulations of neurons in memory-related behaviors. In the first part of this thesis, we focus on the dentate gyrus input node of the hippocampus. Chapter 2 compares the functional properties of adult-born and mature granule cells. Chapter 3 expands on this work by comparing granule cells with mossy cells, another glutamatergic but relatively understudied cell type. The second part of this thesis focuses on the hippocampal output node, area CA1. In chapter 4, we characterize an inhibitory microcircuit that differentially targets the sublayers of area CA1. And in chapter 5, we directly compare the contributions of these sublayers to episodic and semantic memory.

Recognizing and understanding where and when events occurred is essential for normal learning and memory of life experiences. Disruptions in the normal processing of spatial and episodic memories can have devastating consequences; in particular, this is one component of the debilitating cognitive deficits of schizophrenia. We are just now beginning to understand the molecular changes in schizophrenia, but still very little is known about how neural circuit are disrupted that lead to behavioral and cognitive dysfunction. In my thesis I will attempt to address two primary questions; how does hippocampal circuitry support spatial-episodic memories, and what goes wrong when these circuits and memories are impaired? First, how precisely do hippocampal circuits support spatial and episodic learning? In 1885 Hermann Ebbinghaus published the first results of a quantitative study of the psychology of memory, showing the predictable forgetting of items over time. Since then, psychologists and cognitive scientists have investigated, described, and defined the precise nature of memory and the behaviors it drives. We eventually realized that memory is not a unitary function of the brain, but that it is dissociable at it’s broadest level into explicit, recollectable memories and the implicit memory of learned skills and abilities. We have now identified networks of brain regions that are essential for these functions. The first functional imaging of the human brain further advanced out understanding of the particular brain regions active during memory tasks and technological advances have allowed us to generate higher resolution functional maps of the brain. Moving to rodent models, we are now getting closer to the memory engram, the set of changes that occur in the brain that store an object, event, or association for future recall. In some particular instances, such as spatial and episodic memories, we already have a very good understanding. But, which particular cells store this information and how does that memory come to be? In my primary thesis project, I will show that the stabilization of firing patterns in principal cells in hippocampal area CA1 supports learning of a spatial reward task. More specifically, as task demands shift pyramidal cells in CA1 specifically encode the reward zone by firing when the mouse is at the correct location. Finally, by modeling the shift of pyramidal cell activity throughout learning, I show the way in which the population of cells shift their firing activity to encode the reward zone. Second, what goes wrong in the normal processing of information that leads to disrupted memory storage and recall? Deficits in spatial and episodic memory are two of the primary cognitive dysfunctions in schizophrenia. While, hallucinations and delusions are perhaps the most widely recognized, they are in part treatable with antipsychotics, while the cognitive and memory deficits are not as well understood, untreatable, and the greatest barrier to rehabilitation. Cognitive deficits observed in schizophrenia patients are, at their core, neuronal circuit disruptions, spanning multiple brain regions and cognitive domains. What can we learn about the circuits underlying these behavioral symptoms? What goes wrong in the brain that is driving these disruptions? I focused on one particular well-characterized brain region (the hippocampus) by recording the activity of hippocampal area CA1 principal cells in an etiologically-validated mouse model of schizophrenia while the mice are actively engaged in a spatial learning task. I identified specific features of the place cell population that are disrupted and predict behavioral deficits - the day-to-day firing stability of the neuronal population and the lack of over-representation of the reward zone. Overall, my work used head-fixed two-photon functional imaging of awake mice to chronically record the activity of distinct components of the hippocampal memory system: long-range inhibitory projections from the entorhinal cortex to hippocampal area CA1, adult-born granule cells in the dentate gyrus, and large heterogeneous populations of CA1 principal cells. I recorded activity during hippocampal-dependent learning and memory tasks in both schizophrenia-mutant and wildtype mice in order to directly probe hippocampal circuits involved in spatial learning. These experiments provided new evidence of the underlying cellular substrates of both healthy and diseased spatial memory processing.

O lobo temporal mesial (LTM) é essencial para tarefas de memória e possui muitas conexões com diferentes áreas do cérebro. Pacientes com epilepsia do LTM, refratários ao tratamento medicamentoso, são candidatos à cirurgia para remoção do foco das crises. Portanto, antes da cirurgia, é essencial avaliar eventuais riscos de declínio das funções de memória, por meio de uma série de testes clínicos. Recentemente, abriu-se a possibilidade de estudar certos aspectos do funcionamento cerebral, de modo não invasivo, utilizando Imagens funcionais por Ressonância Magnética (fMRI). O objetivo deste trabalho foi desenvolver métodos que possibilitem a aplicação de protocolos de memória em estudos de fMRI, com vistas a pacientes com epilepsia. Para a manutenção da atenção durante os estudos de fMRI foi confeccionado um dispositivo infravermelho para registrar as respostas obtidas. Além disso, foi desenvolvido um programa (VOI Analyser) para a otimização das análises dos exames de fMRI. Tanto o dispositivo infravermelho como o programa foram amplamente utilizados em vários projetos de pesquisa permitindo o estudo de tarefas complexas. Neste estudo, a tarefa visava identificar as redes funcionais que participam do processo de codificação e recuperação de memória episódica utilizando tarefas visuais de identificação de cenas complexas. Foram estudados nesse estudo 12 voluntários assintomáticos e 7 pacientes com epilepsia do LTM. O estudo de grupo evidenciou o envolvimento de estruturas do LTM. A tarefa demonstrou ter um nível de dificuldade alta, em especial para pacientes, baseando-se na avaliação do tempo de resposta e nível de acertos. Além do estudo dos grupos, foi realizada uma análise por região de interesse (ROI), com ênfase no complexo amídala-hipocampo. Em seguida, o foco do estudo foi voltado para a assimetria hemisférica funcional, por meio do cálculo do índice de lateralização (IL). Além de rever os resultados obtidos pelo IL convencional, resultados preliminares levaram à proposta de um segundo índice corrigido, considerando a quantidade de voxels e a assimetria das ROI. A utilização do índice corrigido tornou a análise mais estável por diminuir a dependência do limiar estatístico considerado. A seguir, foi realizada uma subdivisão do hipocampo em porção anterior, central e posterior a qual indicou uma maior participação da região posterior na tarefa de codificação e da anterior na tarefa de recuperação, tanto entre os voluntários como em pacientes. / Medial temporal lobe (MTL) is essential for memory tasks and has many connections with different areas of the brain. Patients with MTL epilepsy refractory to medical treatment are candidates for surgery to remove the epileptiform tissue. Therefore, before surgery, it is essential to assess the risk of memory function decrease caused by the procedure, through a series of clinical trials. Recently, there is the possibility of studying certain aspects of brain functioning by using a non-invasive technique: functional Magnetic Resonance Imaging (fMRI). The aim of this work was to implement memory protocols in fMRI studies of epilepsy patients. For attention maintenance during the fMRI study an infrared device was built, in order to record the response times. In addition, a software was developed (VOI Analyser) to optimize the analysis of the fMRI examinations. Both have been widely used in several research projects enabling the study of complex tasks. In this study, the task was intended to identify the functional networks involved in the process of encoding and retrieving of episodic memory using a visual task involving complex scenes. 19 subjects were studied: 12 controls and 7 patients with refractory epilepsy. Group study showed the involvement of structures in MTL. The task has demonstrated a high level of difficulty, especially for patients, based on the analysis of response times and correct hits. In addition to the study of groups, an individual analysis was performed by region of interest (ROI), with emphasis on amigdala-hippocampus complex. Then, functional hemispheric asymmetry was studied, by means of the lateralization index (LI). In addition to the computation of conventional LI, an alternative LI was proposed, which considers voxels occupancy and ROI asymmetry. The use of such modified index tuned the analysis more stable by decreasing the dependence on considered statistical threshold. Moreover, LI was also computed on 3 portions of the hippocampus: anterior, middle and posterior. The results indicated a greater involvement of the posterior portion on the encoding task and anterior one in the recovery task, both for volunteers and patients.