How does the brain convert sensory information into abstract representations that can support complex behaviours? The rodent head-direction (HD) system, whose cell ensembles represent head direction in the horizontal plane, is a striking example of a “cognitive” representation without a direct sensory correlate. It can be updated by sensory inputs fromdifferentmodalities, yet persists in the absence of external input. Together with cells tuned for place, the HD system is thought to be fundamental for navigation and spatial information processing. However, relatively few studies have sought to characterise the connection between the HD system and spatial behaviour directly, and their overall outcome has been inconclusive. In the experiments that make up the first part of this thesis, we approach this issue by isolating the self-motion component of the HDsystem. We developed an (angular) path integration task in which we show that rats rely on their internal sense of direction to return to a trial-unique starting location, allowing us to investigate the contribution of the HD system to this behaviour without influences from uncontrolled external cues. Using this path integration task, we show that rats with bilateral lesions of the lateral mammillary nuclei (LMN) are significantly impaired compared to sham-operated controls. Lesions of the LMN, which contains HDcells, are known to abolish directional firing in downstream HD areas, suggesting that impairment on the task is due to loss of HD activity. We also recorded HD cell activity as rats are performing the path integration task, and found the HD representation to correlate with the rats’ choice of return journey. Thus, we provide both causal and correlational experimental evidence for a critical role of the HD system in path integration. For the second part of this thesis, we implemented a computational model of how the HD system is updated by head movements during path integration, providing a novel explanation for HD cells’ ability to anticipate the animal’s head direction. The model predicts that such anticipatory time intervals (ATIs) should depend on the frequency spectrum of the rats’ head movements. In direct comparison with experimental recording data, we show that the model can explain up to 80% of the experimentally observed variance, where none was explained by previous models. We also consider the effects of propagating the HD signal through multiple layers, identifying several potential sources of anticipation and lag. In summary, this thesis provides behavioural, lesion, and unit-recording evidence that during path integration, rats use a directional signal provided by the head direction system. The neural mechanisms responsible for the generation and maintenance of this signal are explored computationally. The finding that ATIs depend on the statistics of head movements has methodological implications and constrains models of the HD system.

Rats learn to navigate to a specific location faster in a familiar environment (Keith and Mcvety 1988). It has been proposed that place learning does not require specific reward signals, but rather, that it occurs automatically. One of the strongest pieces of evidence for the automatic nature of place learning comes from the observation that place and head direction cells reference their receptive fields to prominent landmarks in an environment without needing a reward signal (O’Keefe and Conway 1978; Taube et al. 1990b). It has also been proposed that an allocentric representation of an environment would be bound to the landmarks with the greatest relative stability to guide its orientation (O’Keefe and Nadel 1978). The first two parts of this thesis explore whether place and head direction cells automatically use the most coherent landmarks for orientation. Head direction cells have been shown to orient their preferred firing directs coherently when being exposed to conflicting landmarks in an environment (Yoganarasimha et al. 2006). A model of head direction cells was thus used to explore the necessary mechanisms required to implement an allocentric system that selects landmarks based on their relative stability. We found that the simple addition of Hebbian projections combined with units representing the orientation of landmarks to the head direction cell system is sufficient for the system to exhibit such a capacity. We then recorded both entorhinal head direction cells and CA1 place cells and at the same time subjected the rats to repeated experiences of landmark conflicts. During the conflicts a subset of landmarks always maintained a fixed relative relationship with each other. We found that the visual landmarks retained their ability to control the place and head direction cells even after repeated experience of conflict and that the simultaneously recorded place cells exhibited coherent representations between conflicts. However, the ’stable landmarks’ did not show significantly greater control over the place and head direction cells when comparing to the unstable landmarks. This argues against the hypothesis that the relative stability between landmarks is encoded automatically. We did observe a trend that, with more conflict experience, the ’stable landmarks’ appeared to exert greater control over the cells. The last part of the thesis explores whether goal sensitive cells (Ainge et al. 2007a) discovered from CA1 of hippocampus are developed due to familiarity with the environment or from the demands for rats to perform a win-stay behaviour. We used the same win-stay task as in Ainge et al. and found that there were few or no goal sensitive cells on the first day of training. Subsequent development of goal sensitive activity correlated significantly with the rat’s performance during the learning phase of the task. The correlation provides support to the hypothesis that the development of goal sensitive cells is associated to the learning of the win-stay task though it does not rule out the possibility that these goal sensitive cells are developed due to the accumulated experience on the maze. In summary, this thesis explores what kind of spatial information is encoded by place and head direction cells and finds that relative stability between landmarks without a reward signal is not automatically encoded. On the other hand, when additional information is required to solve a task, CA1 place cells adapt their spatial code to provide the necessary information to guide successful navigation.

Head direction cells signal the orientation of the head in the horizontal plane. This thesis shows how some of the known head direction cell response properties might develop through learning. The research methodology employed is the computer simulation of neural network models of head direction cells that self-organize through learning. The preferred firing directions of head direction cells will change in response to the manipulation of distal visual cues, but not in response to the manipulation of proximal visual cues. Simulation results are presented of neural network models that learn to form separate representations of distal and proximal visual cues that are presented simultaneously as visual input to the network. These results demonstrate the computation required for a subpopulation of head direction cells to learn to preferentially respond to distal visual cues. Within a population of head direction cells, the angular distance between the preferred firing directions of any two cells is maintained across different environments. It is shown how a neural network model can learn to maintain the angular distance between the learned preferred firing directions of head direction cells across two different visual training environments. A population of head direction cells can update the population representation of the current head direction, in the absence of visual input, using internal idiothetic (self-generated) motion signals alone. This is called the path integration of head direction. It is important that the head direction cell system updates its internal representation of head direction at the same speed as the animal is rotating its head. Neural network models are simulated that learn to perform the path integration of head direction, using solely idiothetic signals, at the same speed as the head is rotating.

L’orientation spatiale et la fonction de navigation sont des processus contrôlés par des circuits et éléments neuronaux bien précis. Le présubiculum, aire cortical de transition de la région parahippocampique, est situé entre l’hippocampe et le cortex entorhinal. Le présubiculum est impliqué dans la navigation spatiale à la fois chez l’animal et l’Homme. Plus de la moitié des neurones du présubiculum sont des cellules de direction de la tête qui déchargent en fonction de la direction prise par la tête de l’animal. Le présubiculum est un carrefour majeur pour le transfert d’information de direction de la tête et de l’information visuelle aux régions de la formation hippocampique et parahippocampique et sous-corticale. Malgré son importance fonctionnelle, le traitement de l’information au sein du circuit présubiculaire à 6 couches reste encore peu connu. Au cours de ma thèse, j’ai étudié les éléments inhibiteurs qui composent le microcircuit présubiculaire à partir de tranches aigües de cerveau de souris en utilisant la technique du patch-clamp. J’ai caractérisé les propriétés anatomique et électriques des interneurones ainsi que leur connectivité locale et à distances avec d’autres régions corticales.Dans un premier temps, j’ai étudié la diversité des interneurones exprimant la parvalbumine et la somatostatine à partir de lignées de souris transgéniques exprimant une protéine fluorescente dans les interneurones. J’ai montré l’existence des cellules en panier à décharge rapide exprimant la parvalbumine et des cellules de Martinotti à bas seuil d’activation exprimant la somatostatine. J’ai également décrit un troisième groupe atypique avec des propriétés électriques intermédiaires et des morphologies hétérogènes. L’existence de ce groupe transitionnel pourrait s’expliquer par la présence d’interneurones exprimant à la fois la parvalbumine et la somatostatine. Ainsi, le microcircuit inhibiteur du présubiculum semble partager toute la complexité des autres aires corticales. Dans un second temps, je me suis intéressée à l’intégration des entrées thalamiques par les neurones excitateurs et inhibiteurs dans les couches superficielles du présubiculum à l’aide de la technique du double patch-clamp. J’ai montré que les axones thalamiques innervent sélectivement les couches superficielles et plus particulièrement, contactent directement les cellules de projection vers le cortex entorhinal ainsi que les interneurons exprimant la parvalbumine dans la couche 3 du présubiculum. En revanche, les interneurons exprimant la somatostatine sont indirectement recrutés par les cellules pyramidales du microcircuit. Ces interneurones joueraient un double rôle à la fois dans l’inhibition latérale et le maintien d’une décharge soutenue des cellules principales. Du fait de la forte probabilité de connexion entre les cellules principales et les interneurones exprimant la parvalbumine, ces derniers seraient impliqués dans l’inhibition de type feed-forward. Mon travail de thèse a permis d’apporter des connaissances fondamentales concernant l’inhibition au sein du présubiculum. Il a permis de dévoiler une diversité d’interneurones GABAergiques et de montrer l’existence de circuits neuronaux canoniques de type « feedforward » et « feedback » qui seraient recrutés à différents moments de la signalisation de la direction de la tête. / Spatial orientation and navigation are controlled by specific neuronal circuits and elements. The presubiculum, a transitional cortical area of the parahippocampal formation, is located between the hippocampus and the entorhinal cortex, and it participates in spatial navigation in animals and humans. More than half of presubicular neurons are head direction cells that fire as a function of the directional heading. The presubiculum is thought to be a crucial node for transferring directional heading information to the entorhinal-hippocampal network, and feeding back visual landmark information to upstream regions of the head directional circuit. Despite its functional importance, information processing within the 6-layered presubicular microcircuit remains not completely understood. During my PhD, I studied inhibitory neurons of the presubicular microcircuit in the slice preparation using patch-clamp recordings. I characterized their anatomo-physiological properties as well as their functional connectivity with local principal neurons. In the first part, I examined the diversity of two major populations of GABAergic neurons, the parvalbumin (PV) and somatostatin (SOM) expressing interneurons in mouse presubiculum. Using transgenic mouse strains Pvalb-Cre, Sst-Cre and X98, where interneurons were fluorescently labeled, I showed the existence of typical PV fast-spiking basket-like interneurons mainly in the Pvalb-Cre line and SOM low-threshold spiking Martinotti cell-like interneurons in the X98 and Sst-Cre line. Unsupervised cluster analysis based on electrophysiological parameters further revealed a transitional group containing interneurons from either Pvalb-Cre or Sst-cre lines with quasi-fast-spiking properties and heterogeneous morphologies. A small subpopulation of ~6% of interneurons co-expressed PV and SOM in mouse presubiculum. The presubiculum appears to share the whole complexity of other cortical areas in term of inhibition. In the second part, I investigated the integration of thalamic inputs by principal neurons as well as PV and SST interneurons in the presubiculum using double patch-clamp recordings. I found that thalamic axons selectively innervated superficial layers and made direct synaptic contacts with pyramidal neurons that project to medial entorhinal cortex and also with PV interneurons in superficial layer 3. In contrast, SST interneurons were indirectly recruited by presubicular pyramidal cells in a facilitating and frequency dependent manner. They may mediate lateral inhibition onto nearby principal cells, and at the same time, preserve sustained firing of principal neurons. In paired recording experiments, I found that PV cells inhibit neighboring pyramidal neurons with a high connection probability. PV interneurons are rapidly recruited by thalamic excitation and mediate feed-forward inhibition in presubicular pyramidal neurons. My PhD work brought fundamental knowledge about the presubicular inhibitory microcircuit. It has unraveled different populations of GABAergic interneurons and revealed canonical feedforward and feedback inhibitory motifs that are likely to be recruited at different times during head direction signaling.

The ability to self-localize and navigate in both stable and changing environments is crucial for the survival of many species. Research conducted on the non-human mammalian hippocampus and surrounding brain structures has uncovered several classes of spatial related cells. These cells provide the rest of the brain with knowledge of the animal’s location and direction—knowledge that is subsequently used in spatial navigation. This thesis provides an overview of three types of cells underlying this ability in rodents. First, place cells located in the hippocampus encode the animal’s specific location in the environment. Second, head direction cells found throughout the Papez circuit convey the angular direction of the animal’s head. Last, grid cells in the medial entorhinal cortex generate a regular triangular grid spanning the entire explored setting. The focus of this review lies on the most salient features of these types of cells. It is also considered how the cells respond to manipulations of external and internal information, as well as how different lesions affect their activity.

The hippocampal formation has been implicated in spatial formation for many decades. The hippocampus proper has received the most attention but other regions of the hippocampal formation contribute largely to spatial cognition. This thesis concentrated on one such region, the postsubiculum. The postsubiculum is considered important because it contains head direction cells and because it thought to be a major input to the hippocampus, via the entorhinal cortex. This thesis aims to test the functional role of the rat postsubiculum under two types of situation: one where the rat must rely on idiothetic cues for navigation, and another where the rat has visual cues present and can rely on these for orientation. The thesis also investigates hippocampal place cells and their stability over time after short exposures to novel environments. Chapter 3 of this thesis aimed to test whether the postsubiculum is necessary for path integration during a homing task. Rats were trained on a homing task on a circular platform maze. Once the task was acquired, rats were given lesions of the postsubiculum or sham lesions and then re-tested on the path integration task. The homing performance of rats with lesions of the postsubiculum was as good as that of the sham rats. A series of manipulations suggests that the rats were homing by path integration, confirmed by probe tests. The rats were then tested on a forced-choice delayed alternation T-maze task that revealed a significant impairment in alternation with delays of 5, 30, and 60 seconds. This suggests that the postsubiculum is not necessary for path integration in a homing task but is necessary for avoiding previously visited locations as is necessary in an alternation task. The experiments in Chapters 4 and 5 of this thesis aimed to investigate the effects of postsubiculum pharmacological inactivation on hippocampal CA1 place cells when rats were introduced to a novel environment with visual cues. A necessary first step was to assess place cells without any manipulation of the postsubiculum (Chapter 4) and then use information gained from this in the design of experiments in Chapter 5. Rats chronically implanted with recording electrodes in the CA1 region of the hippocampus were exposed to novel cue-rich environments whilst place fields were recorded. Following delays of 3, 6, or 24 hours, the same cells were recorded again in the same environment but with the cues rotated by 90°. Pixel-by-pixel correlations of the place fields show that stability of the place fields was significantly lower at 24 hours than at 3 hours. Stability after 6 hours was not significantly different from 3 hours. In the third set of experiments, rats were implanted with drug infusion cannulae in the postsubiculum and recording electrodes in CA1. Following infusions of either the AMPA receptor antagonist CXQX, the NMDA receptor antagonist D-AP5 or a control infusion of ACSF, place field stability was assessed as rats were exposed to a cylindrical environment with a single polarising cue card for 3 x 10 minute sessions and then again 6 hours later. There were no differences in place field correlations between the 3 drug conditions, although there was evidence of larger changes in spatial information content between cells in the CNQX and AP5 drug condition, but not the ACSF condition. The results suggest that, under the present testing conditions, place fields stability did not depend upon AMPA receptor-mediated transmission nor did it depend on NMDA receptor-mediated synaptic plasticity.

It has been proposed that as animals explore their environment they build and maintain a cognitive map, an internal representation of their surroundings (Tolman, 1948). We tested this hypothesis using a task designed to assess the ability of rats to make a spatial inference (take a novel shortcut)(Roberts et al., 2007). Our findings suggest that rats are unable to make a spontaneous spatial inference. Furthermore, they bear similarities to experiments which have been similarly unable to replicate or support Tolman’s (1948) findings. An inability to take novel shortcuts suggests that rats do not possess a cognitive map (Bennett, 1996). However, we found evidence of alternative learning strategies, such as latent learning (Tolman & Honzik, 1930b) , which suggest that rats may still be building such a representation, although it does not appear they are able to utilise this information to make complex spatial computations. Neurons found in the hippocampus show remarkable spatial modulation of their firing rate and have been suggested as a possible neural substrate for a cognitive map (O'Keefe & Nadel, 1978). However, the firing of these place cells often appears to be modulated by features of an animal’s behaviour (Ainge, Tamosiunaite, et al., 2007; Wood, Dudchenko, Robitsek, & Eichenbaum, 2000). For instance, previous experiments have demonstrated that the firing rate of place fields in the start box of some mazes are predictive of the animal’s final destination (Ainge, Tamosiunaite, et al., 2007; Ferbinteanu & Shapiro, 2003). We sought to understand whether this prospective firing is in fact related to the goal the rat is planning to navigate to or the route the rat is planning to take. Our results provide strong evidence for the latter, suggesting that rats may not be aware of the location of specific goals and may not be aware of their environment in the form of a contiguous map. However, we also found behavioural evidence that rats are aware of specific goal locations, suggesting that place cells in the hippocampus may not be responsible for this representation and that it may reside elsewhere (Hok, Chah, Save, & Poucet, 2013). Unlike their typical activity in an open field, place cells often have multiple place fields in geometrically similar areas of a multicompartment environment (Derdikman et al., 2009; Spiers et al., 2013). For example, Spiers et al. (2013) found that in an environment composed of four parallel compartments, place cells often fired similarly in multiple compartments, despite the active movement of the rat between them. We were able to replicate this phenomenon, furthermore, we were also able to show that if the compartments are arranged in a radial configuration this repetitive firing does not occur as frequently. We suggest that this place field repetition is driven by inputs from Boundary Vector Cells (BVCs) in neighbouring brain regions which are in turn greatly modulated by inputs from the head direction system. This is supported by a novel BVC model of place cell firing which predicts our observed results accurately. If place cells form the neural basis of a cognitive map one would predict spatial learning to be difficult in an environment where repetitive firing is observed frequently (Spiers et al., 2013). We tested this hypothesis by training animals on an odour discrimination task in the maze environments described above. We found that rats trained in the parallel version of the task were significantly impaired when compared to the radial version. These results support the hypothesis that place cells form the neural basis of a cognitive map; in environments where it is difficult to discriminate compartments based on the firing of place cells, rats find it similarly difficult to discriminate these compartments as shown by their behaviour. The experiments reported here are discussed in terms of a cognitive map, the likelihood that such a construct exists and the possibility that place cells form the neural basis of such a representation. Although the results of our experiments could be interpreted as evidence that animals do not possess a cognitive map, ultimately they suggest that animals do have a cognitive map and that place cells form a more than adequate substrate for this representation.

No description available.

In dieser Doktorarbeit wird Langsamkeit als unüberwachtes Lernprinzip in sensorischen Systemen untersucht. Dabei wird zwei Aspekten besondere Aufmerksamkeit gewidmet: der mathematischen Analyse von Slow Feature Analysis - einer Implementierung des Langsamkeitsprinzips - und der Frage, wie das Langsamkeitsprinzip biologisch umgesetzt werden kann. Im ersten Teil wird zunächst eine mathematische Theorie für Slow Feature Analysis entwickelt, die zeigt, dass die optimalen Funktionen für Slow Feature Analysis die Lösungen einer partiellen Differentialgleichung sind. Die Theorie erlaubt, das Verhalten komplizierter Anwendungen analytisch vorherzusagen und intuitiv zu verstehen. Als konkrete Anwendungen wird das Erlernen von Orts- und Kopfrichtungszellen, sowie von komplexen Zellen im primären visuellen Kortex vorgestellt. Im Rahmen einer technischen Anwendung werden die theoretischen Ergebnisse verwendet, um einen neuen Algorithmus für nichtlineare blinde Quellentrennung zu entwickeln und zu testen. Als Abschluss des ersten Teils wird die Beziehung zwischen dem Langsamkeitsprinzip und dem Lernprinzip der verhersagenden Kodierung mit Hilfe eines informationstheoretischen Ansatzes untersucht. Der zweite Teil der Arbeit befasst sich mit der Frage der biologischen Implementierung des Langsamkeitsprinzips. Dazu wird zunächst gezeigt, dass Spikezeit-abhängige Plastizität unter bestimmten Bedingungen als Implementierung des Langsamkeitsprinzips verstanden werden kann. Abschließend wird gezeigt, dass sich die Lerndynamik sowohl von gradientenbasiertem Langsamkeitslernen als auch von Spikezeit-abhängiger Plastizität mathematisch durch Reaktions-Diffusions-Gleichungen beschreiben lässt. / In this thesis, we investigate slowness as an unsupervised learning principle of sensory processing. Two aspects are given particular emphasis: (a) the mathematical analysis of Slow Feature Analysis (SFA) as one particular implementation of slowness learning and (b) the question, how slowness learning can be implemented in a biologically plausible fashion. In the first part of the thesis, we develop a mathematical framework for SFA and show that the optimal functions for SFA are the solutions of a partial differential eigenvalue problem. The theory allows (a) to make analytical predictions for the behavior of complicated applications and (b) an intuitive understanding of how the statistics of the input data are reflected in the optimal functions of SFA. The theory is applied to the learning of place and head-direction representations and to the learning of complex cell receptive fields as found in primary visual cortex. As a technical application, we use the theoretical results to develop and test a new algorithm for nonlinear blind source separation. The first part of the thesis is concluded by an information-theoretic analysis of the relation between slowness learning and predictive coding. In the second part of the thesis, we study the question, how slowness learning could be implemented in a biologically plausible manner. To this end, we first show that spike timing-dependent plasticity can under certain conditions be interpreted as an implementation of slowness learning. Finally, we show that both gradient-based slowness learning and spike timing-dependent plasticity lead to receptive field dynamics that can be described in terms of reaction-diffusion equations.