A neural network model of hippocampal contributions to category learning

In addition to its critical role in encoding individual episodes, the hippocampus is capable of extracting regularities across experiences. This ability is central to category learning, and a growing literature indicates that the hippocampus indeed makes important contributions to this kind of learning. Using a neural network model that mirrors the anatomy of the hippocampus, we investigated the mechanisms by which the hippocampus may support novel category learning. We simulated three category learning paradigms and evaluated the network's ability to categorize and to recognize specific exemplars in each. We found that the trisynaptic pathway within the hippocampus-connecting entorhinal cortex to dentate gyrus, CA3, and CA1-was critical for remembering individual exemplars, reflecting the rapid binding and pattern separation functions of this circuit. The monosynaptic pathway from entorhinal cortex to CA1, in contrast, was responsible for detecting the regularities that define category structure, made possible by the use of distributed representations and a slower learning rate. Together, the simulations provide an account of how the hippocampus and its constituent pathways support novel category learning.

Prenatal stress programs behavioral pattern separation in adult mice

Stress is an unavoidable condition in human life. Stressful events experienced during development, including in utero, have been suggested as one major pathophysiological mechanism for developing vulnerability towards neuropsychiatric and neurodevelopmental disorders in adulthood. One cardinal feature of such disorders is impaired cognitive ability, which may in part rely on abnormal structure and function of the hippocampus. In the hippocampus, the dentate gyrus is a site of continuous neurogenesis, a process that has been recently implicated in spatial pattern separation, a cognitive phenomenon that serves to reduce the degree of overlap in the incoming information to facilitate its storage with minimal interference. We previously reported that adult neurogenesis is altered by prenatal stress allowing us to hypothesize that prenatal stress may possibly lead to impairment in pattern separation. To test this hypothesis, both control (C) and prenatally stressed (PS) adult mice were tested for metric and contextual discrimination abilities. We report for the first time that prenatal stress impairs pattern separation process, a deficit that may underlie their cognitive alterations and that may result in defective behaviors reminiscent of psychiatric illness such as post-traumatic stress disorder.

Working memory and pattern separation in founder strains of the BXD recombinant inbred mouse panel

Working memory and pattern separation are fundamental cognitive abilities which, when impaired, significantly diminish quality of life. Discovering genetic mechanisms underlying innate and disease-induced variation in these cognitive abilities is a critical step towards treatments for common and devastating neurodegenerative conditions such as Alzheimer's disease. In this regard, the trial-unique nonmatching-to-location assay (TUNL) is a touchscreen operant conditioning procedure allowing simultaneous quantification of working memory and pattern separation in mice and rats. In the present study, we used the TUNL assay to quantify these cognitive abilities in C57BL/6J and DBA/2J mice. These strains are the founders of the BXD recombinant inbred mouse panel which enables discovery of genetic mechanisms underlying phenotypic variation. TUNL testing revealed that pattern separation was significantly influenced by mouse strain, whereas working memory was not. Moreover, horizontal distance and vertical distance between choice-phase stimuli had dissociable effects on TUNL performance. These findings provide novel data on mouse strain differences in pattern separation and support previous findings of equivalent working memory performance in C57BL/6J and DBA/2J mice. Although working memory of the BXD founder strains was equivalent in this study, working memory of BXD strains may be divergent because of transgressive segregation. Collectively, data presented here indicate that pattern separation is heritable in the mouse and that the BXD panel can be used to identify mechanisms underlying variation in pattern separation.

An emergent temporal basis set robustly supports cerebellar time-series learning

Learning plays a key role in the function of many neural circuits. The cerebellum is considered a learning machine essential for time interval estimation underlying motor coordination and other behaviors. Theoretical work has proposed that the cerebellar input recipient structure, the granule cell layer (GCL), performs pattern separation of inputs that facilitates learning in Purkinje cells (P-cells). However, the relationship between input reformatting and learning outcomes has remained debated, with roles emphasized for pattern separation features from sparsification to decorrelation. We took a novel approach by training a minimalist model of the cerebellar cortex to learn complex time-series data from naturalistic inputs, in contrast to traditional classification tasks. The model robustly produced temporal basis sets from naturalistic inputs, and the resultant GCL output supported learning of temporally complex target functions. Learning favored surprisingly dense granule cell activity, yet the key statistical features in GCL population activity that drove learning differed from those seen previously for classification tasks. Moreover, different cerebellar tasks were supported by diverse pattern separation features that matched the demands of the tasks. These findings advance testable hypotheses for mechanisms of temporal basis set formation and predict that population statistics of granule cell activity may differ across cerebellar regions to support distinct behaviors.

A role for Lin28a in aging-associated decline of adult hippocampal neurogenesis

Hippocampal neurogenesis declines with aging. Wnt ligands and antagonists within the hippocampal neurogenic niche regulate the proliferation of neural progenitor cells and the development of new neurons, and the changes of their levels in the niche mediate aging-associated decline of neurogenesis. We found that RNA-binding protein Lin28a remained existent in neural progenitor cells and granule neurons in the adult hippocampus, and decreased with aging. Loss of Lin28a inhibited the responsiveness of neural progenitor cells to niche Wnt agonist and reduced neurogenesis, thus impairing pattern separation. Overexpression of Lin28a increased the proliferation of neural progenitor cells, promoted the functional integration of newborn neurons, restored neurogenesis in Wnt-deficient dentate gyrus, and rescued the impaired pattern separation in aging mice. Our data suggest that Lin28a regulates adult hippocampal neurogenesis as an intracellular mechanism by responding to niche Wnt signals, and its decrease is involved in aging-associated decline of hippocampal neurogenesis as well as related cognitive functions.

Acute and chronic physical activity improve spatial pattern separation in humans

Physical activity benefits both fitness and cognition. However, its effect on long-term memory is unclear. Successful memory involves not only remembering information over time but also keeping memories distinct and less confusing. The ability to separate similar experiences into distinct memories is one of the main features of episodic memory. In this work, we evaluated the effect of acute and chronic physical activity on a new task to assess spatial pattern separation in a 3D virtual reality environment. We manipulated the load of memory similarity and found that 25 minutes of cycling after encoding - but not before retrieval - was sufficient to improve similar, but not dissimilar memories, 24 hours after encoding. Furthermore, we found that participants who engaged in regular physical activity, but not sedentary subjects, showed memory for the similar condition the next day. Thus, physical activity could be a simple way to improve discrimination of spatial memories in humans.

Why do hippocampal mossy cells matter? It depends on the frequency and context

The discrimination of similar episodes and places, and their representation as distinct memories, depend on a process called pattern separation that relies on the circuitry of the hippocampal dentate gyrus (DG). Mossy cells (MCs) are key neurons in the circuitry, but how they influence DG network dynamics, function, and seizure risk has not been fully elucidated. We found the net impact of MCs was inhibitory at physiological frequencies connected with learning and behaviour, and their absence associated with deficits in pattern separation and spatial memory; at higher frequencies, their net impact was excitatory, and their absence protected against seizures. Thus, MCs influence DG outputs in a highly dynamic manner that varies with frequency and context.One-Sentence SummaryHippocampal mossy cells are required for learning and memory; but their absence protects against seizures.

Structured connectivity in the cerebellum enables noise-resilient pattern separation

The cerebellum is thought to detect and correct errors between intended and executed commands1-3 and is critical for social behaviors, cognition and emotion4-6. Computations for motor control must be performed quickly to correct errors in real time and should be sensitive to small differences between patterns for fine error correction while being resilient to noise7. Influential theories of cerebellar information processing have largely assumed random network connectivity, which increases the encoding capacity of the network's first layer8-13. However, maximizing encoding capacity reduces resiliency to noise7. To understand how neuronal circuits address this fundamental tradeoff, we mapped the feedforward connectivity in the mouse cerebellar cortex using automated large-scale transmission electron microscopy (EM) and convolutional neural network-based image segmentation. We found that both the input and output layers of the circuit exhibit redundant and selective connectivity motifs, which contrast with prevailing models. Numerical simulations suggest these redundant, non-random connectivity motifs increase discriminability of similar input patterns at a minimal cost to the network's overall encoding capacity. This work reveals how neuronal network structure can balance encoding capacity and redundancy, unveiling principles of biological network architecture with implications for artificial neural network design.