Neural mechanisms of memory: synaptic and genomic hypotheses

Memorizing of new facts and events means that entering signals produce definite changes within the brain. According to the commonly accepted hypothesis, traces of memory are stored through modifications in the strength of synaptic connections, resulting in formations of new patterns of neural activity. This synaptic hypothesis of memory determines the main direction of experimental studies in the field. It is shown in this review that the synaptic hypothesis can hardly explain the mechanism of long-term (often life-long) memory storage as well as memory resistance to both uncontrolled synaptic activity (epileptic seizures) and various adverse effects on the brain (anesthesia, injury, concussion, etc.). Arguments for an alternative hypothesis are given that long-term memory is mainly formed at the intraneural level through modifications of DNA molecules and associated proteins. This genomic hypothesis allows for a new approach to understanding the etiology ofAlzheimer's disease, whose initial symptom is solely memory impairment.     

Cellular and molecular mechanisms of memory

There has been nearly a century of interest in the idea that information is encoded in the brain as specific spatio-temporal patterns of activity in distributed networks and stored as changes in the efficacy of synaptic connections on neurons that are activated during learning. The discovery and detailed report of the phenomenon generally known as long-term potentiation opened a new chapter in the study of synaptic plasticity in the vertebrate brain, and this form of synaptic plasticity has now become the dominant model in the search for the cellular bases of learning and memory. To date, the key events in the cellular and molecular mechanisms underlying synaptic plasticity are starting to be identified. They require the activation of specific receptors and of several molecular cascades to convert extracellular signals into persistent functional changes in neuronal connectivity. Accumulating evidence suggests that the rapid activation of the genetic machinery is a key mechanism underlying the enduring modification of neural networks required for the laying down of memory. The recent developments in the search for the cellular and molecular mechanisms of memory storage are reviewed.     

Efficient Partitioning of Memory Systems and Its Importance for Memory Consolidation

Long-term memories are likely stored in the synaptic weights of neuronal networks in the brain. The storage capacity of such networks depends on the degree of plasticity of their synapses. Highly plastic synapses allow for strong memories, but these are quickly overwritten. On the other hand, less labile synapses result in long-lasting but weak memories. Here we show that the trade-off between memory strength and memory lifetime can be overcome by partitioning the memory system into multiple regions characterized by different levels of synaptic plasticity and transferring memory information from the more to less plastic region. The improvement in memory lifetime is proportional to the number of memory regions, and the initial memory strength can be orders of magnitude larger than in a non-partitioned memory system. This model provides a fundamental computational reason for memory consolidation processes at the systems level.     

Differential Involvement of the Dentate Gyrus in Adaptive Forgetting in the Rat

How does the brain discriminate essential information aimed to be stored permanently from information required only temporarily, and that needs to be cleared away for not saturating our precious memory space? Reference Memory (RM) refers to the long-term storage of invariable information whereas Working Memory (WM) depends on the short-term storage of trial-unique information. Previous work has revealed that WM tasks are very sensitive to proactive interference. In order to prevent such interference, irrelevant old memories must be forgotten to give new ones the opportunity to be stabilized. However, unlike memory, physiological processes underlying this adaptive form of forgetting are still poorly understood. Here, we precisely ask what specific brain structure(s) could be responsible for such process to occur. To answer this question, we trained rats in a radial maze using three paradigms, a RM task and two WM tasks involving or not the processing of interference but strictly identical in terms of locomotion or motivation. We showed that an inhibition of the expression of Zif268 and c-Fos, two indirect markers of neuronal activity and synaptic plasticity, was observed in the dentate gyrus of the dorsal hippocampus when processing such interfering previously stored information. Conversely, we showed that inactivating the dentate gyrus impairs both RM and WM, but improves the processing of interference. Altogether, these results strongly suggest for the first time that the dentate gyrus could be a key structure involved in adaptive forgetting.     

A Three-Threshold Learning Rule Approaches the Maximal Capacity of Recurrent Neural Networks

Understanding the theoretical foundations of how memories are encoded and retrieved in neural populations is a central challenge in neuroscience. A popular theoretical scenario for modeling memory function is the attractor neural network scenario, whose prototype is the Hopfield model. The model simplicity and the locality of the synaptic update rules come at the cost of a poor storage capacity, compared with the capacity achieved with perceptron learning algorithms. Here, by transforming the perceptron learning rule, we present an online learning rule for a recurrent neural network that achieves near-maximal storage capacity without an explicit supervisory error signal, relying only upon locally accessible information. The fully-connected network consists of excitatory binary neurons with plastic recurrent connections and non-plastic inhibitory feedback stabilizing the network dynamics; the memory patterns to be memorized are presented online as strong afferent currents, producing a bimodal distribution for the neuron synaptic inputs. Synapses corresponding to active inputs are modified as a function of the value of the local fields with respect to three thresholds. Above the highest threshold, and below the lowest threshold, no plasticity occurs. In between these two thresholds, potentiation/depression occurs when the local field is above/below an intermediate threshold. We simulated and analyzed a network of binary neurons implementing this rule and measured its storage capacity for different sizes of the basins of attraction. The storage capacity obtained through numerical simulations is shown to be close to the value predicted by analytical calculations. We also measured the dependence of capacity on the strength of external inputs. Finally, we quantified the statistics of the resulting synaptic connectivity matrix, and found that both the fraction of zero weight synapses and the degree of symmetry of the weight matrix increase with the number of stored patterns.     

The synaptic plasticity and memory hypothesis: encoding,storage and persistence

The synaptic plasticity and memory hypothesis asserts that activity-dependent synaptic plasticity is induced at appropriate synapses during memory formation and is both necessary and sufficient for the encoding and trace storage of the type of memory mediated by the brain area in which it is observed. Criteria for establishing the necessity and sufficiency of such plasticity in mediating trace storage have been identified and are here reviewed in relation to new work using some of the diverse techniques of contemporary neuroscience. Evidence derived using optical imaging, molecular-genetic and optogenetic techniques in conjunction with appropriate behavioural analyses continues to offer support for the idea that changing the strength of connections between neurons is one of the major mechanisms by which engrams are stored in the brain.     

Inducible and reversible NR1 knockout reveals crucial role of the NMDA receptor in preserving remote memories in the brain

Long-term storage of information is a hallmark feature of the brain, yet routine turnover of synaptic receptors appears to be intrinsically paradoxical to this capability. To investigate how the brain preserves its delicate synaptic efficacies, we generated inducible and reversible knockout mice in which the NMDA receptor can be temporarily switched off in the forebrain specifically during the storage stage. Retention of 9-month contextual and cued fear memories is severely disrupted by prolonged, but not transient, loss of the NMDA receptor that occurs 6 months after initial training and at least 2 months prior to memory retrieval. Normal learning and memory function in subsequent tasks following the 9-month retention tests suggest that the observed retention deficits did not result from recall or performance impairment. Thus, our study reveals a hitherto unrecognized role of the NMDA receptor in dynamically maintaining the long-term synaptic stability of memory storage circuits in the brain.     

Structural Synaptic Plasticity Has High Memory Capacity and Can Explain Graded Amnesia,Catastrophic Forgetting,and the Spacing Effect

Although already William James and, more explicitly, Donald Hebb''s theory of cell assemblies have suggested that activity-dependent rewiring of neuronal networks is the substrate of learning and memory, over the last six decades most theoretical work on memory has focused on plasticity of existing synapses in prewired networks. Research in the last decade has emphasized that structural modification of synaptic connectivity is common in the adult brain and tightly correlated with learning and memory. Here we present a parsimonious computational model for learning by structural plasticity. The basic modeling units are “potential synapses” defined as locations in the network where synapses can potentially grow to connect two neurons. This model generalizes well-known previous models for associative learning based on weight plasticity. Therefore, existing theory can be applied to analyze how many memories and how much information structural plasticity can store in a synapse. Surprisingly, we find that structural plasticity largely outperforms weight plasticity and can achieve a much higher storage capacity per synapse. The effect of structural plasticity on the structure of sparsely connected networks is quite intuitive: Structural plasticity increases the “effectual network connectivity”, that is, the network wiring that specifically supports storage and recall of the memories. Further, this model of structural plasticity produces gradients of effectual connectivity in the course of learning, thereby explaining various cognitive phenomena including graded amnesia, catastrophic forgetting, and the spacing effect.     

Memory Capacity of Networks with Stochastic Binary Synapses

In standard attractor neural network models, specific patterns of activity are stored in the synaptic matrix, so that they become fixed point attractors of the network dynamics. The storage capacity of such networks has been quantified in two ways: the maximal number of patterns that can be stored, and the stored information measured in bits per synapse. In this paper, we compute both quantities in fully connected networks of N binary neurons with binary synapses, storing patterns with coding level , in the large and sparse coding limits (). We also derive finite-size corrections that accurately reproduce the results of simulations in networks of tens of thousands of neurons. These methods are applied to three different scenarios: (1) the classic Willshaw model, (2) networks with stochastic learning in which patterns are shown only once (one shot learning), (3) networks with stochastic learning in which patterns are shown multiple times. The storage capacities are optimized over network parameters, which allows us to compare the performance of the different models. We show that finite-size effects strongly reduce the capacity, even for networks of realistic sizes. We discuss the implications of these results for memory storage in the hippocampus and cerebral cortex.     

A Taste for Learning

Understanding mechanisms of learning and memory storage in thehuman brain will contribute to designing strategies for optimizingthis functionality in both biological and electronic neuralnetworks. Studies of memory-deficient patients have localizedsome of the critical brain areas for memory storage and establishedthat different types of information can be stored in anatomicallyseparate locations. Studies of cellular and biochemical eventsduring associative learning in several molluscan neural networkshave produced detailed hypotheses about the causative eventsleading to changes in synaptic function during learning. Useof recently developed preparations of mammalian CNS should allowdirect tests of the generality of the molluscan mechanisms forsynaptic plasticity during learning in the mammalian brain.     

Molecular mechanisms of memory storage in Aplysia

Cellular studies of implicit and explicit memory suggest that experience-dependent modulation of synaptic strength and structure is a fundamental mechanism by which these memories are encoded, processed, and stored within the brain. In this review, we focus on recent advances in our understanding of the molecular mechanisms that underlie short-term, intermediate-term, and long-term forms of implicit memory in the marine invertebrate Aplysia californica, and consider how the conservation of common elements in each form may contribute to the different temporal phases of memory storage.     

The Autapse: A Simple Illustration of Short-Term Analog Memory Storage by Tuned Synaptic Feedback

According to a popular hypothesis, short-term memories are stored as persistent neural activity maintained by synaptic feedback loops. This hypothesis has been formulated mathematically in a number of recurrent network models. Here we study an abstraction of these models, a single neuron with a synapse onto itself, or autapse. This abstraction cannot simulate the way in which persistent activity patterns are distributed over neural populations in the brain. However, with proper tuning of parameters, it does reproduce the continuously graded, or analog, nature of many examples of persistent activity. The conditions for tuning are derived for the dynamics of a conductance-based model neuron with a slow excitatory autapse. The derivation uses the method of averaging to approximate the spiking model with a nonspiking, reduced model. Short-term analog memory storage is possible if the reduced model is approximately linear and if its feedforward bias and autapse strength are precisely tuned.     

Efficacité de la mémoire associative inhérente à la potentiation post-tétanique

An associative memory is modeled in networks of cells that are assumed to have the short-term plasticity of the neuromuscular junction of the frog. The data relating synaptic transmission efficiency and stimulation frequency for post-tetanic potentiation of the neuromuscular junction are represented by polynomial expansions. Simulation of storage and retrieval demonstrates that functional associative memory is feasible based on this particular synaptic plasticity. Retrieval reaches a maximum efficiency at a delay of three minutes after storage and is lost after about 9 min. The signal to noise ratio of the retrieved pattern drops steadily as additional associations are stored in memory but retrieval appears to be possible with up to four stored associations. Although the data are derived from synapses not normally proposed as a basis for memory functions, the results here will generalize to other synaptic junctions located more centrally that have similar characteristics. This simulation technique allows the efficiency of associative memory based on various types of synaptic plasticity to be evaluated.     

Common molecular mechanisms in explicit and implicit memory

Cellular and molecular studies of both implicit and explicit memory suggest that experience-dependent modulation of synaptic strength and structure is a fundamental mechanism by which these memories are encoded and stored within the brain. In this review, we focus on recent advances in our understanding of two types of memory storage: (i) sensitization in Aplysia, a simple form of implicit memory, and (ii) formation of explicit spatial memories in the mouse hippocampus. These two processes share common molecular mechanisms that have been highly conserved through evolution.     

Physical models of neural networks

The state of art in computer modelling of neural networks with associative memory is reviewed. The available experimental data are considered on learning and memory of small neural systems, on isolated synapses and on molecular level. Computer simulations demonstrate that realistic models of neural ensembles exhibit properties which can be interpreted as image recognition, categorization, learning, prototype forming, etc. A bilayer model of associative neural network is proposed. One layer corresponds to the short-term memory, the other one to the long-term memory. Patterns are stored in terms of the synaptic strength matrix. We have studied the relaxational dynamics of neurons firing and suppression within the short-term memory layer under the influence of the long-term memory layer. The interaction among the layers has found to create a number of novel stable states which are not the learning patterns. These synthetic patterns may consist of elements belonging to different non-intersecting learning patterns. Within the framework of a hypothesis of selective and definite coding of images in brain one can interpret the observed effect as the "idea? generating" process.     

The persistence of long-term memory: a molecular approach to self-sustaining changes in learning-induced synaptic growth

Recent cellular and molecular studies of both implicit and explicit memory storage suggest that experience-dependent modulation of synaptic strength and structure is a fundamental mechanism by which these diverse forms of memory are encoded and stored. For both forms of memory storage, some type of synaptic growth is thought to represent the stable cellular change that maintains the long-term process. In this review, we discuss recent findings on the molecular events that underlie learning-related synaptic growth in Aplysia and discuss the possibility that an active, prion-based mechanism is important for the maintenance of the structural change and for the persistence of long-term memory.     

Long term memory storage capacity of multiconnected neural networks

Quantitative expressions of long-term memory storage capacities of complex neural network are derived. The networks are made of neurons connected by synapses of any order, of the axono-axonal type considered by Kandel et al. for example. The effect of link deletion possibly related to aging, is also considered. The central result of this study is that, within the framework of Hebb's laws, the number of stored bits is proportional to the number of synapses. The proportionality factor however, decreases when the order of involved synaptic contact increases. This tends to favor neural architectures with low-order synaptic connectivities. It is finally shown that the memory storage capacities can be optimized by a partition of the network into neuron clusters with size comparable with that observed for cortical microcolumns.     

Soft-bound Synaptic Plasticity Increases Storage Capacity

Accurate models of synaptic plasticity are essential to understand the adaptive properties of the nervous system and for realistic models of learning and memory. Experiments have shown that synaptic plasticity depends not only on pre- and post-synaptic activity patterns, but also on the strength of the connection itself. Namely, weaker synapses are more easily strengthened than already strong ones. This so called soft-bound plasticity automatically constrains the synaptic strengths. It is known that this has important consequences for the dynamics of plasticity and the synaptic weight distribution, but its impact on information storage is unknown. In this modeling study we introduce an information theoretic framework to analyse memory storage in an online learning setting. We show that soft-bound plasticity increases a variety of performance criteria by about 18% over hard-bound plasticity, and likely maximizes the storage capacity of synapses.     

Samuel Butler and human long term memory: Is the cupboard bare?

Memory studies in biological systems distinguish three informational processes that are generally sequential—production/acquisition, storage, and retrieval/use. Identification of DNA as a storage form for hereditary information accelerated progress in that field. Assuming the path of successful elucidation in one memory field (heredity) to be heuristic for elucidation in another (brain), then progress in neuroscience should accelerate when a storage form is identified. In the 19th century Ewald Hering and Samuel Butler held that heredity and brain memory both involved the storage of information and that the two forms of storage were the same. Hering specified storage as ‘molecular vibrations’ but, while making a fuller case, Butler was less committal. In the 20th century, the ablation studies of Karl Lashley failed to identify unique sites for storage of brain information, and Donald Hebb's ‘synaptic plasticity’ hypothesis of distributed storage over a neuronal network won favor. In the 21st century this has come under attack, and the idea that brain and hereditary information are stored as DNA is advocated. Thus, albeit without attribution, Butler's idea is reinstated. Yet, while the case is still open, the synaptic plasticity and DNA hypotheses have problems. Two broad alternatives remain on the table. Long term memory is located: (1) in the brain, either in some other macromolecular form (e.g. protein, lipid) or in some sub-molecular form (e.g. quantum computing and ‘brain as holograph’ hypotheses) or (2) outside the brain. The suggestion of the medieval physician Avicenna that the brain ‘cupboard’ is bare—i.e. the brain is a perceptual, not storage, organ—is consistent with a mysterious ‘universe as holograph’ model. Understanding how Butler came to contribute could be heuristic for future progress in a field fraught with ‘fractionation and disunity’.     

Dendritic mRNA: transport, translation and function

Many cellular functions require the synthesis of a specific protein or functional cohort of proteins at a specific time and place in the cell. Local protein synthesis in neuronal dendrites is essential for understanding how neural activity patterns are transduced into persistent changes in synaptic connectivity during cortical development, memory storage and other long-term adaptive brain responses. Regional and temporal changes in protein levels are commonly coordinated by an asymmetric distribution of mRNAs. This Review attempts to integrate current knowledge of dendritic mRNA transport, storage and translation, placing particular emphasis on the coordination of regulation and function during activity-dependent synaptic plasticity in the adult mammalian brain.