No description available.

No description available.

Why are some memories easy to retrieve, while others are more difficult to access? Here, we tested whether we could bias memory replay, a process whereby newly learned information is reinforced by reinstating the neuronal patterns of activation that were present during learning, towards particular memory traces. The goal of this biasing is to strengthen some memory traces, making them more easily retrieved. To test this, participants were scanned during interleaved periods of encoding and rest. Throughout the encoding runs, participants learned triplets of images that were paired with semantically related sound cues. During two of the three rest periods, novel, irrelevant sounds were played. During one critical rest period, however, the sound cues learned in the preceding encoding period were played in an effort to preferentially increase reactivation of the associated visual images, a manipulation known as targeted memory reactivation. Representational similarity analyses were used to compare multi-voxel patterns of hippocampal activation across encoding and rest periods. Our index of reactivation was selectively enhanced for memory traces that were targeted for preferential reactivation during offline rest, both compared to information that was not targeted for preferential reactivation and compared to a baseline rest period. Importantly, this neural effect of targeted reactivation was related to the difference in delayed order memory for information that was cued versus uncued, suggesting that preferential replay may be a mechanism by which specific memory traces can be selectively strengthened for enhanced subsequent memory retrieval. We also found partial evidence of discrimination of unique temporal sequences within the hippocampus. Over time, multi-voxel patterns associated with a given triplet sequence became more dissimilar to the patterns associated with the other sequences. Furthermore, this neural marker of sequence preservation was correlated with the difference in delayed order memory for cued versus uncued triplets, signifying that the ability to reactivate particular temporal sequences within the hippocampus may be related to enhanced temporal order memory for the cued information. Taken together, these findings support the claim that awake replay can be biased towards preferential reactivation of particular memory traces and also suggest that this preferential reactivation, as well as representations of reactivated temporal sequences, can be detected within patterns of hippocampal activation. / Psychology

Episodic memories are neurally coded records of personally experienced events across a lifetime. These records are encoded via medial temporal lobe structures in the brain, including the hippocampus, and are commonly called "representations" or "memory traces". Existing studies indicate that information about the neural signal corresponding to a memory representation can be found in functional magnetic resonance imaging (fMRI) data when the pattern across its smallest units (voxels, often 3mm3 sections of the brain) is measured. Many prior studies have measured these voxel patterns in response to stimuli as if they are a spontaneous brain function, regardless of cognitive factors. These studies sometimes find that similarity in the voxel patterns across repetition of a to-be-remembered event predicts later memory retrieval, but the results are inconsistent. The current fMRI study investigated the possibility that cognitive goals during encoding affect the type of neural representation (voxel pattern) that will later support memory retrieval. This seems likely because prior behavioral studies indicate that cognitive variability across repetitions of an event benefits later memory retrieval, which is difficult to reconcile with the common finding that voxel pattern variability across repetitions of an event harms later memory. We tested this hypothesis by comparing voxel patterns that support later memory retrieval to those associated with forgotten items in the medial temporal lobe, including the hippocampus, and lateral occipital cortex. Overall, as previously demonstrated, the behavioral results showed that exposure to variable cognitive goals across repetition of events during encoding benefited subsequent memory retrieval. Voxel patterns in the hippocampus indicated a significant interaction between cognitive goals (variable vs. consistent) and memory (remembered vs. forgotten) such that less voxel pattern similarity for the repeated events with variable cognitive goals, but not consistent cognitive goals, supported later memory success. In other words, variable hippocampal neural activations for the same events under different cognitive goals predicted better later memory performance. However, there was no significant interaction in neural pattern similarity between cognitive goals and memory success in medial temporal cortices or lateral occipital lobe. Instead, higher similarity in voxel patterns in right medial temporal cortices was associated with later memory retrieval, regardless of cognitive goals. In the lateral occipital lobe, the main effects of cognitive goals, hemisphere, and memory success were found but no interactions. In conclusion, we found that the relationship between pattern similarity and memory success in the hippocampus (but not the medial temporal lobe cortex) changes when the cognitive goal during encoding does or does not vary across repetitions of the event. / Master of Science / Episodic memory is a long-term memory of personal experiences which are encoded via the medial temporal lobe in the brain, primarily in the hippocampus. The records of personal experiences in these areas are commonly called "patterns", "representations", or "memory traces". Prior investigations indicate that the way of measuring the neural signals corresponding to personal events is functional magnetic resonance imaging (fMRI). The brain images taken by an fMRI scanner represent the patterns of the smallest unit (voxels, often 3mm3 sections of the brain). Many prior investigations of episodic memory used the voxel patterns but showed mixed results in whether similarity in the voxel patterns across repetition of a repeated event leads to subsequent memory retrieval. One of the possible explanations for mixed results is that the cognitive factors during encoding were neglected. Therefore, the current fMRI study examined how cognitive goals during encoding influence the voxel patterns that later support memory retrieval. During encoding, participants were shown an image repeated with the same or different questions and answered the question on the screen in an fMRI scanner. After 10 days, they were invited to the item memory test on the images that they were given during the encoding phase. The voxel patterns in the medial temporal lobe, including the hippocampus, and the lateral occipital lobe were compared across the repetitions of each image. The behavioral results showed that variable cognitive goals across repeated events during encoding benefited later memory retrieval. Furthermore, less similar voxel patterns in the hippocampus for the images repeated with different questions, but not the same questions, during encoding predicted better later memory success. In the right medial temporal cortices, higher similarity in voxel patterns was significantly associated with later memory retrieval, regardless of cognitive goals. In the lateral occipital lobe, higher voxel pattern similarity was found in the right hemisphere, for images repeated with the same question, and for images successfully retrieved later. In conclusion, we found that the relationship between voxel pattern similarity and memory success in the hippocampus (but not the medial temporal lobe cortex) changes when the cognitive goal during encoding does or does not vary across repetitions of the event.

No description available.

No description available.

Dans les systèmes cognitifs, le rôle de la mémoire de travail est crucial pour le raisonnement visuel et la prise de décision. D’énormes progrès ont été réalisés dans la compréhension des mécanismes de la mémoire de travail humain/animal, ainsi que dans la formulation de différents cadres de réseaux de neurones artificiels à mémoire augmentée. L’objectif global de notre projet est de former des modèles de réseaux de neurones artificiels capables de consolider la mémoire sur une courte période de temps pour résoudre une tâche de mémoire et les relier à l’activité cérébrale des humains qui ont résolu la même tâche. Le projet est de nature interdisciplinaire en essayant de relier les aspects de l’intelligence artificielle (apprentissage profond) et des neurosciences. La tâche cognitive utilisée est la tâche N-back, très populaire en neurosciences cognitives dans laquelle les sujets sont présentés avec une séquence d’images, dont chacune doit être identifiée pour savoir si elle a déjà été vue ou non. L’ensemble de données d’imagerie fonctionnelle (IRMf) utilisé a été collecté dans le cadre du projet Courtois Neurmod. Nous étudions plusieurs variantes de modèles de réseaux neuronaux récurrents qui apprennent à résoudre la tâche de mémoire de travail N-back en les entraînant avec des séquences d’images. Ces réseaux de neurones entraînés optimisés pour la tâche de mémoire sont finalement utilisés pour générer des représentations de caractéristiques pour les images de stimuli vues par les sujets humains pendant leurs enregistrements tout en résolvant la tâche. Les représentations dérivées de ces réseaux de neurones servent ensuite à créer un modèle de codage pour prédire l’activité IRMf BOLD des sujets. On comprend alors la relation entre le modèle de réseau neuronal et l’activité cérébrale en analysant cette capacité prédictive du modèle dans différentes zones du cerveau impliquées dans la mémoire de travail. Ce travail présente une manière d’utiliser des réseaux de neurones artificiels pour modéliser le comportement et le traitement de l’information de la mémoire de travail du cerveau et d’utiliser les données d’imagerie cérébrale capturées sur des sujets humains lors de la tâche N-back pour potentiellement comprendre certains mécanismes de mémoire du cerveau en relation avec ces modèles de réseaux de neurones artificiels. / In cognitive systems, the role of working memory is crucial for visual reasoning and decision making. Tremendous progress has been made in understanding the mechanisms of the human/animal working memory, as well as in formulating different frameworks of memory augmented artificial neural networks. The overall objective of our project is to train artificial neural network models that are capable of consolidating memory over a short period of time to solve a memory task and relate them to the brain activity of humans who solved the same task. The project is of interdisciplinary nature in trying to bridge aspects of Artificial Intelligence (deep learning) and Neuroscience. The cognitive task used is the N-back task, a very popular one in Cognitive Neuroscience in which the subjects are presented with a sequence of images, each of which needs to be identified as to whether it was already seen or not. The functional imaging (fMRI) dataset used has been collected as a part of the Courtois Neurmod Project. We study multiple variants of recurrent neural network models that learn to remember input images across timesteps. These trained neural networks optimized for the memory task are ultimately used to generate feature representations for the stimuli images seen by the human subjects during their recordings while solving the task. The representations derived from these neural networks are then to create an encoding model to predict the fMRI BOLD activity of the subjects. We then understand the relationship between the neural network model and brain activity by analyzing this predictive ability of the model in different areas of the brain that are involved in working memory. This work presents a way of using artificial neural networks to model the behavior and information processing of the working memory of the brain and to use brain imaging data captured from human subjects during the N-back task to potentially understand some memory mechanisms of the brain in relation to these artificial neural network models.