Magnetic tunnel junction random number generators applied to dynamically tuned probability trees driven by spin orbit torque.

Perpendicular magnetic tunnel junction (pMTJ)-based true-random number generators (RNGs) can consume orders of magnitude less energy per bit than CMOS pseudo-RNGs. Here, we numerically investigate with a macrospin Landau-Lifshitz-Gilbert equation solver the use of pMTJs driven by spin-orbit torque to directly sample numbers from arbitrary probability distributions with the help of a tunable probability tree. The tree operates by dynamically biasing sequences of pMTJ relaxation events, called 'coinflips', via an additional applied spin-transfer-torque current. Specifically, using a single, ideal pMTJ device we successfully draw integer samples on the interval [0, 255] from an exponential distribution based onp-value distribution analysis. In order to investigate device-to-device variations, the thermal stability of the pMTJs are varied based on manufactured device data. It is found that while repeatedly using a varied device inhibits ability to recover the probability distribution, the device variations average out when considering the entire set of devices as a 'bucket' to agnostically draw random numbers from. Further, it is noted that the device variations most significantly impact the highest level of the probability tree, with diminishing errors at lower levels. The devices are then used to draw both uniformly and exponentially distributed numbers for the Monte Carlo computation of a problem from particle transport, showing excellent data fit with the analytical solution. Finally, the devices are benchmarked against CMOS and memristor RNGs, showing faster bit generation and significantly lower energy use.

Intermittent fasting enhances long-term memory consolidation, adult hippocampal neurogenesis, and expression of longevity gene Klotho.

Daily calorie restriction (CR) and intermittent fasting (IF) enhance longevity and cognition but the effects and mechanisms that differentiate these two paradigms are unknown. We examined whether IF in the form of every-other-day feeding enhances cognition and adult hippocampal neurogenesis (AHN) when compared to a matched 10% daily CR intake and ad libitum conditions. After 3 months under IF, female C57BL6 mice exhibited improved long-term memory retention. IF increased the number of BrdU-labeled cells and neuroblasts in the hippocampus, and microarray analysis revealed that the longevity gene Klotho (Kl) was upregulated in the hippocampus by IF only. Furthermore, we found that downregulating Kl in human hippocampal progenitor cells led to decreased neurogenesis, whereas Kl overexpression increased neurogenesis. Finally, histological analysis of Kl knockout mice brains revealed that Kl is required for AHN, particularly in the dorsal hippocampus. These data suggest that IF is superior to 10% CR in enhancing memory and identifies Kl as a novel candidate molecule that regulates the effects of IF on cognition likely via AHN enhancement.

Regulation and function of adult neurogenesis: from genes to cognition.

Adult neurogenesis in the hippocampus is a notable process due not only to its uniqueness and potential impact on cognition but also to its localized vertical integration of different scales of neuroscience, ranging from molecular and cellular biology to behavior. This review summarizes the recent research regarding the process of adult neurogenesis from these different perspectives, with particular emphasis on the differentiation and development of new neurons, the regulation of the process by extrinsic and intrinsic factors, and their ultimate function in the hippocampus circuit. Arising from a local neural stem cell population, new neurons progress through several stages of maturation, ultimately integrating into the adult dentate gyrus network. The increased appreciation of the full neurogenesis process, from genes and cells to behavior and cognition, makes neurogenesis both a unique case study for how scales in neuroscience can link together and suggests neurogenesis as a potential target for therapeutic intervention for a number of disorders.

Dopaminergic inputs in the dentate gyrus direct the choice of memory encoding.

Rewarding experiences are often well remembered, and such memory formation is known to be dependent on dopamine modulation of the neural substrates engaged in learning and memory; however, it is unknown how and where in the brain dopamine signals bias episodic memory toward preceding rather than subsequent events. Here we found that photostimulation of channelrhodopsin-2-expressing dopaminergic fibers in the dentate gyrus induced a long-term depression of cortical inputs, diminished theta oscillations, and impaired subsequent contextual learning. Computational modeling based on this dopamine modulation indicated an asymmetric association of events occurring before and after reward in memory tasks. In subsequent behavioral experiments, preexposure to a natural reward suppressed hippocampus-dependent memory formation, with an effective time window consistent with the duration of dopamine-induced changes of dentate activity. Overall, our results suggest a mechanism by which dopamine enables the hippocampus to encode memory with reduced interference from subsequent experience.

Computing with Spikes: The Advantage of Fine-Grained Timing.

Neural-inspired spike-based computing machines often claim to achieve considerable advantages in terms of energy and time efficiency by using spikes for computation and communication. However, fundamental questions about spike-based computation remain unanswered. For instance, how much advantage do spike-based approaches have over conventional methods, and under what circumstances does spike-based computing provide a comparative advantage? Simply implementing existing algorithms using spikes as the medium of computation and communication is not guaranteed to yield an advantage. Here, we demonstrate that spike-based communication and computation within algorithms can increase throughput, and they can decrease energy cost in some cases. We present several spiking algorithms, including sorting a set of numbers in ascending/descending order, as well as finding the maximum or minimum or median of a set of numbers. We also provide an example application: a spiking median-filtering approach for image processing providing a low-energy, parallel implementation. The algorithms and analyses presented here demonstrate that spiking algorithms can provide performance advantages and offer efficient computation of fundamental operations useful in more complex algorithms.

A Combinatorial Model for Dentate Gyrus Sparse Coding.

The dentate gyrus forms a critical link between the entorhinal cortex and CA3 by providing a sparse version of the signal. Concurrent with this increase in sparsity, a widely accepted theory suggests the dentate gyrus performs pattern separation-similar inputs yield decorrelated outputs. Although an active region of study and theory, few logically rigorous arguments detail the dentate gyrus's (DG) coding. We suggest a theoretically tractable, combinatorial model for this action. The model provides formal methods for a highly redundant, arbitrarily sparse, and decorrelated output signal.To explore the value of this model framework, we assess how suitable it is for two notable aspects of DG coding: how it can handle the highly structured grid cell representation in the input entorhinal cortex region and the presence of adult neurogenesis, which has been proposed to produce a heterogeneous code in the DG. We find tailoring the model to grid cell input yields expansion parameters consistent with the literature. In addition, the heterogeneous coding reflects activity gradation observed experimentally. Finally, we connect this approach with more conventional binary threshold neural circuit models via a formal embedding.

Molecular layer perforant path-associated cells contribute to feed-forward inhibition in the adult dentate gyrus.

New neurons, which have been implicated in pattern separation, are continually generated in the dentate gyrus in the adult hippocampus. Using a genetically modified rabies virus, we demonstrated that molecular layer perforant pathway (MOPP) cells innervated newborn granule neurons in adult mouse brain. Stimulating the perforant pathway resulted in the activation of MOPP cells before the activation of dentate granule neurons. Moreover, activation of MOPP cells by focal uncaging of glutamate induced strong inhibition of granule cells. Together, these results indicate that MOPP cells located in the molecular layer of the dentate gyrus contribute to feed-forward inhibition of granule cells via perforant pathway activation.

New neurons and new memories: how does adult hippocampal neurogenesis affect learning and memory?

The integration of adult-born neurons into the circuitry of the adult hippocampus suggests an important role for adult hippocampal neurogenesis in learning and memory, but its specific function in these processes has remained elusive. In this article, we summarize recent progress in this area, including advances based on behavioural studies and insights provided by computational modelling. Increasingly, evidence suggests that newborn neurons might be involved in hippocampal functions that are particularly dependent on the dentate gyrus, such as pattern separation. Furthermore, newborn neurons at different maturation stages may make distinct contributions to learning and memory. In particular, computational studies suggest that, before newborn neurons are fully mature, they might function as a pattern integrator by introducing a degree of similarity to the encoding of events that occur closely in time.

Development of GABAergic inputs controls the contribution of maturing neurons to the adult hippocampal network.

New neurons are continuously generated in the dentate gyrus (DG) in the adult hippocampus, and new granule cells (GCs) have been shown to be necessary for several aspects of learning and memory. Nonetheless, the limited information available regarding the anatomical and physiological development of synaptic inputs onto maturing neurons has restricted our understanding of how new GCs affect cognition. Here, we use photostimulation to demonstrate the time course by which anatomically isolated inhibitory inputs develop onto maturing GCs. We then show that the gradual development of inhibition is sufficient in a computational model to drive learning of novel information in young neurons. Finally, we validate this model observation by using slice physiology to show how inhibition regulates firing probability and plasticity in young GCs. Combined, these data demonstrate that the unique connectivity of immature GCs affords them a functional role that is different from mature neurons in the DG circuit, a distinction that potentially underlies many of the proposed functions of new neurons in the hippocampal network.

Overcoming the noise in neural computing.

Circuit strategies can enable noisy analog hardware to achieve high precision.

Probabilistic Neural Computing with Stochastic Devices.

The brain has effectively proven a powerful inspiration for the development of computing architectures in which processing is tightly integrated with memory, communication is event-driven, and analog computation can be performed at scale. These neuromorphic systems increasingly show an ability to improve the efficiency and speed of scientific computing and artificial intelligence applications. Herein, it is proposed that the brain's ubiquitous stochasticity represents an additional source of inspiration for expanding the reach of neuromorphic computing to probabilistic applications. To date, many efforts exploring probabilistic computing have focused primarily on one scale of the microelectronics stack, such as implementing probabilistic algorithms on deterministic hardware or developing probabilistic devices and circuits with the expectation that they will be leveraged by eventual probabilistic architectures. A co-design vision is described by which large numbers of devices, such as magnetic tunnel junctions and tunnel diodes, can be operated in a stochastic regime and incorporated into a scalable neuromorphic architecture that can impact a number of probabilistic computing applications, such as Monte Carlo simulations and Bayesian neural networks. Finally, a framework is presented to categorize increasingly advanced hardware-based probabilistic computing technologies.

Modeling new neuron function: a history of using computational neuroscience to study adult neurogenesis.

Adult neurogenesis is a sophisticated biological process whose function has remained a mystery to neuroscience researchers. To address this question, a number of unique modeling studies have explored the computational implications of adding new neurons to the adult dentate gyrus. Models of neurogenesis fall into two broad categories: abstract models that explore the function of new neurons in simple networks, and biologically based models that investigate the role of new neurons in networks based on the anatomy of the hippocampus. In this review, we summarize the strategies and results of these different modeling approaches, and we discuss their conclusions and limitations in the face of new biological findings.

Synapse formation on neurons born in the adult hippocampus.

Although new and functional neurons are produced in the adult brain, little is known about how they integrate into mature networks. Here we explored the mechanisms of synaptogenesis on neurons born in the adult mouse hippocampus using confocal microscopy, electron microscopy and live imaging. We report that new neurons, similar to mature granule neurons, were contacted by axosomatic, axodendritic and axospinous synapses. Consistent with their putative role in synaptogenesis, dendritic filopodia were more abundant during the early stages of maturation and, when analyzed in three dimensions, the tips of all filopodia were found within 200 nm of preexisting boutons that already synapsed on other neurons. Furthermore, dendritic spines primarily synapsed on multiple-synapse boutons, suggesting that initial contacts were preferentially made with preexisting boutons already involved in a synapse. The connectivity of new neurons continued to change until at least 2 months, long after the formation of the first dendritic protrusions.

Put them out to pasture? What are old granule cells good for, anyway...?

In this issue, Alme and colleagues propose a novel hypothesis that young adult-born neurons represent the functional population of the dentate gyrus, and that old granule cells "retire" from relevance. Here, we present several implications of this interesting theory for the function of both neurogenesis and the dentate gyrus.

Potential role for adult neurogenesis in the encoding of time in new memories.

The dentate gyrus in the hippocampus is one of two brain regions with lifelong neurogenesis in mammals. Despite an increasing amount of information about the characteristics of the newborn granule cells, the specific contribution of their robust generation to memory formation by the hippocampus remains unclear. We describe here a possible role that this population of young granule cells may have in the formation of temporal associations in memory. Neurogenesis is a continuous process; the newborn population is only composed of the same cells for a short period of time. As time passes, the young neurons mature or die and others are born, gradually changing the identity of this young population. We discuss the possibility that one cognitive impact of this gradually changing population on hippocampal memory formation is the formation of the temporal clusters of long-term episodic memories seen in some human psychological studies.

Crossing the Cleft: Communication Challenges Between Neuroscience and Artificial Intelligence.

Historically, neuroscience principles have heavily influenced artificial intelligence (AI), for example the influence of the perceptron model, essentially a simple model of a biological neuron, on artificial neural networks. More recently, notable recent AI advances, for example the growing popularity of reinforcement learning, often appear more aligned with cognitive neuroscience or psychology, focusing on function at a relatively abstract level. At the same time, neuroscience stands poised to enter a new era of large-scale high-resolution data and appears more focused on underlying neural mechanisms or architectures that can, at times, seem rather removed from functional descriptions. While this might seem to foretell a new generation of AI approaches arising from a deeper exploration of neuroscience specifically for AI, the most direct path for achieving this is unclear. Here we discuss cultural differences between the two fields, including divergent priorities that should be considered when leveraging modern-day neuroscience for AI. For example, the two fields feed two very different applications that at times require potentially conflicting perspectives. We highlight small but significant cultural shifts that we feel would greatly facilitate increased synergy between the two fields.

Cholinergic input is required during embryonic development to mediate proper assembly of spinal locomotor circuits.

Rhythmic limb movements are controlled by pattern-generating neurons within the ventral spinal cord, but little is known about how these locomotor circuits are assembled during development. At early stages of embryogenesis, motor neurons are spontaneously active, releasing acetylcholine that triggers the depolarization of adjacent cells in the spinal cord. To investigate whether acetylcholine-driven activity is required for assembly of the central pattern-generating (CPG) circuit, we studied mice lacking the choline acetyltransferase (ChAT) enzyme. Our studies show that a rhythmically active spinal circuit forms in ChAT mutants, but the duration of each cycle period is elongated, and right-left and flexor-extensor coordination are abnormal. In contrast, blocking acetylcholine receptors after the locomotor network is wired does not affect right-left or flexor-extensor coordination. These findings suggest that the cholinergic neurotransmitter pathway is involved in configuring the CPG during a transient period of development.

IGF-I instructs multipotent adult neural progenitor cells to become oligodendrocytes.

Adult multipotent neural progenitor cells can differentiate into neurons, astrocytes, and oligodendrocytes in the mammalian central nervous system, but the molecular mechanisms that control their differentiation are not yet well understood. Insulin-like growth factor I (IGF-I) can promote the differentiation of cells already committed to an oligodendroglial lineage during development. However, it is unclear whether IGF-I affects multipotent neural progenitor cells. Here, we show that IGF-I stimulates the differentiation of multipotent adult rat hippocampus-derived neural progenitor cells into oligodendrocytes. Modeling analysis indicates that the actions of IGF-I are instructive. Oligodendrocyte differentiation by IGF-I appears to be mediated through an inhibition of bone morphogenetic protein signaling. Furthermore, overexpression of IGF-I in the hippocampus leads to an increase in oligodendrocyte markers. These data demonstrate the existence of a single molecule, IGF-I, that can influence the fate choice of multipotent adult neural progenitor cells to an oligodendroglial lineage.