Modello matematico della memoria spaziale nell'ippocampo: Simulazioni e analisi di sensitivita

E' stata presentata un'analisi di sensitività di un modello matematico della memoria spaziale nell'ippocampo. Il modello usa le masse neuronali per simulare le oscillazioni theta delle place cells dell'ippocampo. I risultati mostrano che il modello è robusto a variazioni di diversi parametri.

The relationship between the field-shifting phenomenon and representational coherence of place cells in CA1 and CA3 in a cue-altered environment.

Subfields of the hippocampus display differential dynamics in processing a spatial environment, especially when changes are introduced to the environment. Specifically, when familiar cues in the environment are spatially rearranged, place cells in the CA3 subfield tend to rotate with a particular set of cues (e.g., proximal cues), maintaining a coherent spatial representation. Place cells in CA1, in contrast, display discordant behaviors (e.g., rotating with different sets of cues or remapping) in the same condition. In addition, on average, CA3 place cells shift their firing locations (measured by the center of mass, or COM) backward over time when the animal encounters the changed environment for the first time, but not after that first experience. However, CA1 displays an opposite pattern, in which place cells exhibit the backward COM-shift only from the second day of experience, but not on the first day. Here, we examined the relationship between the environment-representing behavior (i.e., rotation vs. remapping) and the COM-shift of place fields in CA1 and CA3. Both in CA1 and CA3, the backward (as well as forward) COM-shift phenomena occurred regardless of the rotating versus remapping of the place cell. The differential, daily time course of the onset/offset of backward COM-shift in the cue-altered environment in CA1 and CA3 (on day 1 in CA1 and from day 2 onward in CA3) stems from different population dynamics between the subfields. The results suggest that heterogeneous, complex plasticity mechanisms underlie the environment-representating behavior (i.e., rotate/remap) and the COM-shifting behavior of the place cell.

Altered oscillatory dynamics of CA1 parvalbumin-basket cells during theta-gamma rhythmopathies of temporal lobe epilepsy

Recent reports in human demonstrate a role of theta– gamma coupling in memory for spatial episodes and a lack of coupling in people experiencing temporal lobe epilepsy, but the mechanisms are unknown. Using multisite silicon probe recordings of epileptic rats engaged in episodic-like object recognition tasks, we sought to evaluate the role of theta– gamma coupling in the absence of epileptiform activities. Our data reveal a specific association between theta– gamma (30 – 60 Hz) coupling at the proximal stratum radiatum of CA1 and spatial memory deficits. We targeted the microcircuit mechanisms with a novel approach to identify putative interneuronal types in tetrode recordings (parvalbumin basket cells in particular) and validated classification criteria in the epileptic context with neurochemical identification of intracellularly recorded cells. In epileptic rats, putative parvalbumin basket cells fired poorly modulated at the falling theta phase, consistent with weaker inputs from Schaffer collaterals and attenuated gamma oscillations, as evaluated by theta-phase decomposition of current–source density signals. We propose that theta– gamma interneuronal rhythmopathies of the temporal lobe are intimately related to episodic memory dysfunction in this condition.

Slowness and sparseness lead to place, head-direction, and spatial-view cells.

We present a model for the self-organized formation of place cells, head-direction cells, and spatial-view cells in the hippocampal formation based on unsupervised learning on quasi-natural visual stimuli. The model comprises a hierarchy of Slow Feature Analysis (SFA) nodes, which were recently shown to reproduce many properties of complex cells in the early visual system []. The system extracts a distributed grid-like representation of position and orientation, which is transcoded into a localized place-field, head-direction, or view representation, by sparse coding. The type of cells that develops depends solely on the relevant input statistics, i.e., the movement pattern of the simulated animal. The numerical simulations are complemented by a mathematical analysis that allows us to accurately predict the output of the top SFA layer.

Modello matematico dell'integrazione audio-visiva e del suo sviluppo in un ambiente multisensoriale

L'integrazione multisensoriale è la capacità da parte del sistema nervoso centrale di integrare informazioni provenienti da diverse sorgenti sensoriali. In particolare, l'integrazione audio-visiva avviene anche nelle cortecce visive e acustiche, in principio ritenute puramente unisensoriali. L'integrazione audio-visiva non è un'abilità innata, ma si sviluppa progressivamente nel tempo sulla base dell'esperienza. In questa Tesi viene presentato un modello matematico di rete neurale in grado di simulare lo sviluppo in un ambiente multisensoriale dei neuroni delle cortecce primarie visive e uditive nei primi mesi di vita di un neonato, e gli effetti dell'integrazione audio-visiva successivi a tale addestramento. In particolare il modello vuole mostrare, a partire da una condizione basale in cui i neuroni visivi e acustici non sono in grado di discriminare spazialmente la posizione degli input esterni e in cui non sussiste alcuna correlazione tra le due aree corticali primarie visive e uditive, come l'addestramento migliori la precisione della percezione spaziale degli stimoli esterni e come si stabiliscano tra le due aree in esame dei collegamenti stabili a lungo termine. Terminato l'addestramento, si verifica se la rete sia in grado di riprodurre gli effetti di integrazione audio-visiva nella corteccia primaria, quali la "cattura" dello stimolo acustico da parte di quello visivo (ventriloquismo) e il rafforzamento della risposta neurale allo stimolo acustico se contemporaneamente accompagnato da uno stimolo visivo nella stessa posizione spaziale (enhancement). Il modello potrebbe essere utilizzato in futuro anche per simulare altri fenomeni illusori come il fenomeno offline del ventriloquismo e il ventriloquismo a livello temporale.

Meccanismi neurali per la rappresentazione semantica e lessicale: modello di una rete neurale per apprendere il significato di oggetti e parole

Lo studio della memoria semantica attraverso modelli teorici (qualitativi o matematici) gioca un ruolo importante nelle neuroscienze cognitive. In questa tesi viene presentato un modello della memoria semantica e lessicale. Il principio alla base del modello è che la semantica di un oggetto è descritta attraverso una collezione di proprietà, che sintetizzano la percezione dell’oggetto nelle diverse regioni corticali. Gli aspetti semantici e lessicali sono memorizzati in regioni separate del cervello, ma reciprocamente connesse sulla base dell’esperienza passata, secondo un meccanismo di apprendimento Hebbiano. L’obiettivo del lavoro è indagare i meccanismi che portano alla formazione di categorie, analizzare il ruolo delle diverse proprietà nella costruzione di un concetto e le connessioni tra queste proprietà e la parola corrispondente al nome dell’oggetto. Durante l’addestramento per ogni oggetto sono presentate alcune proprietà in input, con una data frequenza: alcune sono condivise, cioè appartengono a più di un concetto e permettono la formazione di categorie, altre sono distintive, cioè sono tipiche di un concetto e consentono la rappresentazione dei membri di una categoria. Un ulteriore aspetto riguardante le proprietà è la distinzione fra proprietà salienti, che sono spontaneamente evocate, e proprietà marginali. E’ stata utilizzata una tassonomia composta da 11 parole che identificano 11 animali. La salienza è stabilita dalla frequenza con cui si verifica ciascuna proprietà. La distinzione tra proprietà condivise e distintive, e tra proprietà salienti e non salienti, è stata ottenuta mediante l’uso della regola di Hebb con una diversa soglia presinaptica e postsinaptica, producendo patterns di sinapsi asimmetriche all’interno della rete semantica (auto-associazione) e tra la rete semantica e lessicale (etero-associazione). La rete addestrata è in grado di risolvere compiti di riconoscimento e denominazione di oggetti, mantenendo una distinzione tra le categorie e i suoi membri, e fornendo un diverso ruolo per le proprietà salienti rispetto alle proprietà marginali.

Hebbian analysis of the transformation of medial entorhinal grid-cell inputs to hippocampal place fields.

The discovery of grid cells in the medial entorhinal cortex (MEC) permits the characterization of hippocampal computation in much greater detail than previously possible. The present study addresses how an integrate-and-fire unit driven by grid-cell spike trains may transform the multipeaked, spatial firing pattern of grid cells into the single-peaked activity that is typical of hippocampal place cells. Previous studies have shown that in the absence of network interactions, this transformation can succeed only if the place cell receives inputs from grids with overlapping vertices at the location of the place cell's firing field. In our simulations, the selection of these inputs was accomplished by fast Hebbian plasticity alone. The resulting nonlinear process was acutely sensitive to small input variations. Simulations differing only in the exact spike timing of grid cells produced different field locations for the same place cells. Place fields became concentrated in areas that correlated with the initial trajectory of the animal; the introduction of feedback inhibitory cells reduced this bias. These results suggest distinct roles for plasticity of the perforant path synapses and for competition via feedback inhibition in the formation of place fields in a novel environment. Furthermore, they imply that variability in MEC spiking patterns or in the rat's trajectory is sufficient for generating a distinct population code in a novel environment and suggest that recalling this code in a familiar environment involves additional inputs and/or a different mode of operation of the network.

A biophysical model of synaptic plasticity and metaplasticity can account for the dynamics of the backward shift of hippocampal place fields.

Hippocampal place cells in the rat undergo experience-dependent changes when the rat runs stereotyped routes. One such change, the backward shift of the place field center of mass, has been linked by previous modeling efforts to spike-timing-dependent plasticity (STDP). However, these models did not account for the termination of the place field shift and they were based on an abstract implementation of STDP that ignores many of the features found in cortical plasticity. Here, instead of the abstract STDP model, we use a calcium-dependent plasticity (CaDP) learning rule that can account for many of the observed properties of cortical plasticity. We use the CaDP learning rule in combination with a model of metaplasticity to simulate place field dynamics. Without any major changes to the parameters of the original model, the present simulations account both for the initial rapid place field shift and for the subsequent slowing down of this shift. These results suggest that the CaDP model captures the essence of a general cortical mechanism of synaptic plasticity, which may underlie numerous forms of synaptic plasticity observed both in vivo and in vitro.

Dominance of the proximal coordinate frame in determining the locations of hippocampal place cell activity during navigation.

The place-specific activity of hippocampal cells provides downstream structures with information regarding an animal's position within an environment and, perhaps, the location of goals within that environment. In rodents, recent research has suggested that distal cues primarily set the orientation of the spatial representation, whereas the boundaries of the behavioral apparatus determine the locations of place activity. The current study was designed to address possible biases in some previous research that may have minimized the likelihood of observing place activity bound to distal cues. Hippocampal single-unit activity was recorded from six freely moving rats as they were trained to perform a tone-initiated place-preference task on an open-field platform. To investigate whether place activity was bound to the room- or platform-based coordinate frame (or both), the platform was translated within the room at an "early" and at a "late" phase of task acquisition (Shift 1 and Shift 2). At both time points, CA1 and CA3 place cells demonstrated room-associated and/or platform-associated activity, or remapped in response to the platform shift. Shift 1 revealed place activity that reflected an interaction between a dominant platform-based (proximal) coordinate frame and a weaker room-based (distal) frame because many CA1 and CA3 place fields shifted to a location intermediate to the two reference frames. Shift 2 resulted in place activity that became more strongly bound to either the platform- or room-based coordinate frame, suggesting the emergence of two independent spatial frames of reference (with many more cells participating in platform-based than in room-based representations).

Experience-dependent, asymmetric expansion of hippocampal place fields

Theories of sequence learning based on temporally asymmetric, Hebbian long-term potentiation predict that during route learning the spatial firing distributions of hippocampal neurons should enlarge in a direction opposite to the animal’s movement. On a route AB, increased synaptic drive from cells representing A would cause cells representing B to fire earlier and more robustly. These effects appeared within a few laps in rats running on closed tracks. This provides indirect evidence for Hebbian synaptic plasticity and a functional explanation for why place cells become directionally selective during route following, namely, to preserve the synaptic asymmetry necessary to encode the sequence direction.

Caracterização de células de lugar no hipocampo e de suas relações com oscilações do potencial de campo local

The main inputs to the hippocampus arise from the entorhinal cortex (EC) and form a loop involving the dentate gyrus, CA3 and CA1 hippocampal subfields and then back to EC. Since the discovery that the hippocampus is involved in memory formation in the 50's, this region and its circuitry have been extensively studied. Beyond memory, the hippocampus has also been found to play an important role in spatial navigation. In rats and mice, place cells show a close relation between firing rate and the animal position in a restricted area of the environment, the so-called place field. The firing of place cells peaks at the center of the place field and decreases when the animal moves away from it, suggesting the existence of a rate code for space. Nevertheless, many have described the emergence of hippocampal network oscillations of multiple frequencies depending on behavioral state, which are believed to be important for temporal coding. In particular, theta oscillations (5-12 Hz) exhibit a spatio-temporal relation with place cells known as phase precession, in which place cells consistently change the theta phase of spiking as the animal traverses the place field. Moreover, current theories state that CA1, the main output stream of the hippocampus, would interplay inputs from EC and CA3 through network oscillations of different frequencies, namely high gamma (60-100 Hz; HG) and low gamma (30-50 Hz; LG), respectively, which tend to be nested in different phases of the theta cycle. In the present dissertation we use a freely available online dataset to make extensive computational analyses aimed at reproducing classical and recent results about the activity of place cells in the hippocampus of freely moving rats. In particular, we revisit the debate of whether phase precession is due to changes in firing frequency or space alone, and conclude that the phenomenon cannot be explained by either factor independently but by their joint influence. We also perform novel analyses investigating further characteristics of place cells in relation to network oscillations. We show that the strength of theta modulation of spikes only marginally affects the spatial information content of place cells, while the mean spiking theta phase has no influence on spatial information. Further analyses reveal that place cells are also modulated by theta when they fire outside the place field. Moreover, we find that the firing of place cells within the theta cycle is modulated by HG and LG amplitude in both CA1 and EC, matching cross-frequency coupling results found at the local field potential level. Additionally, the phase-amplitude coupling in CA1 associated with spikes inside the place field is characterized by amplitude modulation in the 40-80 Hz range. We conclude that place cell firing is embedded in large network states reflected in local field potential oscillations and suggest that their activity might be seen as a dynamic state rather than a fixed property of the cell.

Large Scale Simulations of Hippocampal-Neocortical Interactions in a Parallel Version of Genesis

A hippocampal-CA3 memory model was constructed with PGENESIS, a recently developed version of GENESIS that allows for distributed processing of a neural network simulation. A number of neural models of the human memory system have identified the CA3 region of the hippocampus as storing the declarative memory trace. However, computational models designed to assess the viability of the putative mechanisms of storage and retrieval have generally been too abstract to allow comparison with empirical data. Recent experimental evidence has shown that selective knock-out of NMDA receptors in the CA1 of mice leads to reduced stability of firing specificity in place cells. Here a similar reduction of stability of input specificity is demonstrated in a biologically plausible neural network model of the CA3 region, under conditions of Hebbian synaptic plasticity versus an absence of plasticity. The CA3 region is also commonly associated with seizure activity. Further simulations of the same model tested the response to continuously repeating versus randomized nonrepeating input patterns. Each paradigm delivered input of equal intensity and duration. Non-repeating input patterns elicited a greater pyramidal cell spike count. This suggests that repetitive versus non-repeating neocortical inpus has a quantitatively different effect on the hippocampus. This may be relevant to the production of independent epileptogenic zones and the process of encoding new memories.

Grid Cell Hexagonal Patterns Formed by Fast Self-Organized Learning within Entorhinal Cortex

Grid cells in the dorsal segment of the medial entorhinal cortex (dMEC) show remarkable hexagonal activity patterns, at multiple spatial scales, during spatial navigation. How these hexagonal patterns arise has excited intense interest. It has previously been shown how a selforganizing map can convert firing patterns across entorhinal grid cells into hippocampal place cells that are capable of representing much larger spatial scales. Can grid cell firing fields also arise during navigation through learning within a self-organizing map? A neural model is proposed that converts path integration signals into hexagonal grid cell patterns of multiple scales. This GRID model creates only grid cell patterns with the observed hexagonal structure, predicts how these hexagonal patterns can be learned from experience, and can process biologically plausible neural input and output signals during navigation. These results support a unified computational framework for explaining how entorhinal-hippocampal interactions support spatial navigation.