Synaptic reverberation underlying mnemonic persistent activity

Stimulus-specific persistent neural activity is the neural process underlying active (working) memory. Since its discovery 30 years ago, mnemonic activity has been hypothesized to be sustained by synaptic reverberation in a recurrent circuit. Recently, experimental and modeling work has begun to test the reverberation hypothesis at the cellular level. Moreover, theory has been developed to describe memory storage of an analog stimulus (such as spatial location or eye position), in terms of continuous 'bump attractors' and 'line attractors'. This review summarizes new studies, and discusses insights and predictions from biophysically based models. The stability of a working memory network is recognized as a serious problem; stability can be achieved if reverberation is largely mediated by NMDA receptors at recurrent synapses.     

Molecular mechanisms of working memory

Working memory is a process for temporary active maintenance of information and the role of prefrontal cortex in this memory has been known since the pioneering experiments of Fulton in the early 20th century. Sustained firing of prefrontal neurons during the delay period is considered the neural correlate of working memory. Evidence in literature suggests the involvement of areas beyond the frontal lobe and illustrate that working memory involves parallel, distributed neuronal networks. Prefrontal cortex is part of a complex neural circuit that includes both cortical and subcortical components and many of these regions play vital roles in working memory function. In this article, we review the current understanding of the neural mechanisms of memory maintenance in the brain.     

NR2B subunit in the prefrontal cortex: A double-edged sword for working memory function and psychiatric disorders

The prefrontal cortex (PFC) is a brain region featured with working memory function. The exact mechanism of how working memory operates within the PFC circuitry is unknown, but persistent neuronal firing recorded from prefrontal neurons during a working memory task is proposed to be the neural correlate of this mnemonic encoding. The PFC appears to be specialized for sustaining persistent firing, with N-methyl-d-aspartate (NMDA) receptors, especially slow-decay NR2B subunits, playing an essential role in the maintenance of sustained activity and normal working memory function. However, the NR2B subunit serves as a double-edged sword for PFC function. Because of its slow kinetics, NR2B endows the PFC with not only “neural psychic” properties, but also susceptibilities for neuroexcitotoxicity and psychiatric disorders. This review aims to clarify the interplay among working memory, the PFC, and NMDA receptors; demonstrate the importance of NR2B in the maintenance of persistent activity; understand the risks and vulnerabilities of how NR2B is related to the development of neuropsychiatric disorders; identify gaps that currently exist in our understanding of these processes; and provide insights regarding future directions that may clarify these issues. We conclude that the PFC is a specialized brain region with distinct delayed maturation, unique neuronal circuitry, and characteristic NMDA receptor function. The unique properties and development of NMDA receptors, especially enrichment of NR2B subunits, endow the PFC with not only the capability to generate sustained activity for working memory, but also serves as a major vulnerability to environmental insults and risk factors for psychiatric disorders.     

From distributed resources to limited slots in multiple-item working memory: a spiking network model with normalization

Recent behavioral studies have given rise to two contrasting models for limited working memory capacity: a "discrete-slot" model in which memory items are stored in a limited number of slots, and a "shared-resource" model in which the neural representation of items is distributed across a limited pool of resources. To elucidate the underlying neural processes, we investigated a continuous network model for working memory of an analog feature. Our model network fundamentally operates with a shared resource mechanism, and stimuli in cue arrays are encoded by a distributed neural population. On the other hand, the network dynamics and performance are also consistent with the discrete-slot model, because multiple objects are maintained by distinct localized population persistent activity patterns (bump attractors). We identified two phenomena of recurrent circuit dynamics that give rise to limited working memory capacity. As the working memory load increases, a localized persistent activity bump may either fade out (so the memory of the corresponding item is lost) or merge with another nearby bump (hence the resolution of mnemonic representation for the merged items becomes blurred). We identified specific dependences of these two phenomena on the strength and tuning of recurrent synaptic excitation, as well as network normalization: the overall population activity is invariant to set size and delay duration; therefore, a constant neural resource is shared by and dynamically allocated to the memorized items. We demonstrate that the model reproduces salient observations predicted by both discrete-slot and shared-resource models, and propose testable predictions of the merging phenomenon.     

Context-dependent retrieval of information by neural-network dynamics with continuous attractors.

Memory retrieval in neural networks has traditionally been described by dynamic systems with discrete attractors. However, recent neurophysiological findings of graded persistent activity suggest that memory retrieval in the brain is more likely to be described by dynamic systems with continuous attractors. To explore what sort of information processing is achieved by continuous-attractor dynamics, keyword extraction from documents by a network of bistable neurons, which gives robust continuous attractors, is examined. Given an associative network of terms, a continuous attractor led by propagation of neuronal activation in this network appears to represent keywords that express underlying meaning of a document encoded in the initial state of the network-activation pattern. A dominant hypothesis in cognitive psychology is that long-term memory is archived in the network structure, which resembles associative networks of terms. Our results suggest that keyword extraction by the neural-network dynamics with continuous attractors might symbolically represent context-dependent retrieval of short-term memory from long-term memory in the brain.     

A computational model for spatial working memory deficits in schizophrenia

Cognitive deficits in schizophrenia have been hypothesized to be caused by altered synaptic transmission in circuits of the prefrontal cortex. 2 main hypotheses have been put forward: reduced inhibition and hypofunctional NMDA receptors. Recently, Lee et al. (2008) found that spatial working memory deficits in schizophrenic patients include a disproportionately high incidence of high-confidence error responses. Here, we have studied what synaptic dysfunction can generate this specific behavioral deficit using a computational network model of spatial working memory. We developed quantitative behavioral readout from our network simulations, which reflected the qualitative properties of underlying neural dynamics. We then analyzed the behavioral effect of the GABAergic and glutamatergic hypotheses on our network simulations. We found that reduction in inhibitory transmission in the network caused a reduction in performance through an increase of high-confidence errors, as in the experimental data. In contrast, a concerted reduction in NMDA-receptor-dependent transmission reduced performance via increased low-confidence errors. Only when NMDA receptors were specifically depleted in interneurons did the behavioral read-out of our network mimic the behavioral results for schizophrenic patients. Thus, dynamics in our model network support a role of both global inhibition reduction and hypofunctional NMDA receptors in interneurons in generating the behavioral deficits of simple spatial working memory tasks in schizophrenia.     

The Effects of Guanfacine and Phenylephrine on a Spiking Neuron Model of Working Memory

We use a spiking neural network model of working memory (WM) capable of performing the spatial delayed response task (DRT) to investigate two drugs that affect WM: guanfacine (GFC) and phenylephrine (PHE). In this model, the loss of information over time results from changes in the spiking neural activity through recurrent connections. We reproduce the standard forgetting curve and then show that this curve changes in the presence of GFC and PHE, whose application is simulated by manipulating functional, neural, and biophysical properties of the model. In particular, applying GFC causes increased activity in neurons that are sensitive to the information currently being remembered, while applying PHE leads to decreased activity in these same neurons. Interestingly, these differential effects emerge from network-level interactions because GFC and PHE affect all neurons equally. We compare our model to both electrophysiological data from neurons in monkey dorsolateral prefrontal cortex and to behavioral evidence from monkeys performing the DRT.     

Common neural mechanisms supporting spatial working memory, attention and motor intention

The prefrontal cortex (PFC) and posterior parietal cortex (PPC) are critical neural substrates for working memory. Neural activity persists in these regions during the maintenance of a working memory representation. Persistent activity, therefore, may be the neural mechanism by which information is temporarily maintained. However, the nature of the representation or what is actually being represented by this persistent activity is not well understood. In this review, we summarize the recent functional magnetic resonance imaging (fMRI) studies conducted in our laboratory that test hypotheses about the nature of persistent activity during a variety of spatial cognition tasks. We find that the same areas in the PFC and PPC that show persistent activity during the maintenance of a working memory representation also show persistent activity during the maintenance of spatial attention and the maintenance of motor intention. Therefore, we conclude that persistent activity is not specific to working memory, but instead, carries information that can be used generally to support a variety of cognitions. Specifically, activity in topographically organized maps of prioritized space in PFC and PPC could be read out to guide attention allocation, spatial memory, and motor planning.     

Visual and spatial working memory: from boxes to networks

It is shown that visuo-spatial working memory is better characterized as processes operating on sensory information (visual appearance) and on spatial location (environmental coordinates) in a distributed network than as unitary slave system. Results from passive (short-term) and active memory tasks (imagery) disclose the properties (capacity, content) and the components of this network. The prefrontal cortex is a control structure (dorsal prefers active, ventral passive tasks) and it contributes to spatial memory by a prospective spatial code (eye movements). Visual appearance (including dynamic aspects) is represented as features and object files (bound features) within content-specific areas in the ventral occipital cortex. Spatial coordinates are represented in the parietal cortex (modality-unspecific), when used in spatio-temporal tasks (Corsi) they are closely related to attention. Imagery of objects activates occipito-temporal structures, spatial transformations and mental rotation the parietal cortex (specifically the intraparietal sulcus). Perception, working memory, and imagery use the same neural network. Differences between the tasks are explained by different demands and states of the neural network, and differences in the configuration of the anterior–posterior neural circuits.     

Noninvasively decoding the contents of visual working memory in the human prefrontal cortex within high-gamma oscillatory patterns

The temporal maintenance and subsequent retrieval of information that no longer exists in the environment is called working memory. It is believed that this type of memory is controlled by the persistent activity of neuronal populations, including the prefrontal, temporal, and parietal cortex. For a long time, it has been controversially discussed whether, in working memory, the PFC stores past sensory events or, instead, its activation is an extramnemonic source of top-down control over posterior regions. Recent animal studies suggest that specific information about the contents of working memory can be decoded from population activity in prefrontal areas. However, it has not been shown whether the contents of working memory during the delay periods can be decoded from EEG recordings in the human brain. We show that by analyzing the nonlinear dynamics of EEG oscillatory patterns it is possible to noninvasively decode with high accuracy, during encoding and maintenance periods, the contents of visual working memory information within high-gamma oscillations in the human PFC. These results are thus in favor of an active storage function of the human PFC in working memory; this, without ruling out the role of PFC in top-down processes. The ability to noninvasively decode the contents of working memory is promising in applications such as brain computer interfaces, together with computation of value function during planning and decision making processes.     

Imaging and neural modelling in episodic and working memory processes.

Neuroimaging studies using positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) have revealed the involvement of distributed brain regions in memory processes mainly by the use of subtraction strategy based data analyses. Covariance analysis based data analysis strategies have been introduced more recently which allow functional interactions between brain regions of a neuronal network to be assessed. This contribution focuses on studies aiming to (1) establish the functional topography of episodic and working memory processes in young and old normal volunteers, (2) to assess functional interactions between modules of networks of brain regions by means of covariance based analyses and systems level modelling, (3) to characterise the temporal dynamics by the use of magnetoencephalography (MEG) and (4) to relate neuroimaging data to the underpinning neural networks. Male normal young and old volunteers without neurological or psychiatric illness participated in neuroimaging studies (PET, fMRI, MEG). Studies were approved by the ethical committee and federal authorities. Our results in young volunteers show distributed brain areas that are involved in memory processes (episodic and working memory) and show much of an overlap with respect to the network components. Systems level modelling analyses support the hypothesis of bihemispheric, asymmetric networks subserving memory processes and revealed both similarities in general and differences in the interactions between brain regions during episodic encoding and retrieval as well as working memory. Changes in memory function with ageing are evident from functional topographic studies in old volunteers activating more brain regions as compared to young volunteers. There are more and stronger influences of prefrontal regions in elderly volunteers comparing the functional models between old and young subjects. We discuss the way that the systems level models of the PET and fMRI results have implications for the underlying neural network functioning of the brain. This is done by developing simplifying assumptions, which lead from the equations describing the activities of the coupled neural modules to the systems level model equations. The resulting implications for the neural interactions are then discussed, in terms of a set of synaptically coupled neural modules. Finally, we consider how a similar analysis could be extended from the spatial to the temporal domain thus including the EEG and MEG results. The implication of preliminary MEG results presented here for the temporality arising in the interaction between the coupled neural modules in a working memory paradigm is discussed in terms of the previously developed neural network models arising from the PET and fMRI data.     

Right parietal dysfunction in children with attention deficit hyperactivity disorder, combined type: a functional MRI study

Attention deficit hyperactivity disorder, combined type (ADHD-CT) is associated with spatial working memory deficits. These deficits are known to be subserved by dysfunction of neural circuits involving right prefrontal, striatal and parietal brain regions. This study determines whether decreased right prefrontal, striatal and parietal activation with a mental rotation task shown in adolescents with ADHD-CT is also evident in children with ADHD-CT. A cross-sectional study of 12 pre-pubertal, right-handed, 8-12-year-old boys with ADHD-CT and 12 pre-pubertal, right-handed, performance IQ-matched, 8-12-year-old healthy boys, recruited from local primary schools, was completed. Participants underwent functional magnetic resonance imaging while performing a mental rotation task that requires spatial working memory. The two groups did not differ in their accuracy or response times for the mental rotation task. The ADHD-CT group showed significantly less activation in right parieto-occipital areas (cuneus and precuneus, BA 19), the right inferior parietal lobe (BA 40) and the right caudate nucleus. Our findings with a child cohort confirm previous reports of right striatal-parietal dysfunction in adolescents with ADHD-CT. This dysfunction suggests a widespread maturational deficit that may be developmental stage independent.     

Neurodynamics of the prefrontal cortex during conditional visuomotor associations

The prefrontal cortex is believed to be important for cognitive control, working memory, and learning. It is known to play an important role in the learning and execution of conditional visuomotor associations, a cognitive task in which stimuli have to be associated with actions by trial-and-error learning. In our modeling study, we sought to integrate several hypotheses on the function of the prefrontal cortex using a computational model, and compare the results to experimental data. We constructed a module of prefrontal cortex neurons exposed to two different inputs, which we envision to originate from the inferotemporal cortex and the basal ganglia. We found that working memory properties do not describe the dominant dynamics in the prefrontal cortex, but the activation seems to be transient, probably progressing along a pathway from sensory to motor areas. During the presentation of the cue, the dynamics of the prefrontal cortex is bistable, yielding a distinct activation for correct and error trails. We find that a linear change in network parameters relates to the changes in neural activity in consecutive correct trials during learning, which is important evidence for the underlying learning mechanisms.     

Impact of complex genetic variation in COMT on human brain function

Catechol-O-methyltransferase (COMT) has been shown to be critical for prefrontal dopamine flux, prefrontal cortex-dependent cognition and activation. Several potentially functional variants in the gene have been identified, but considerable controversy exists regarding the contribution of individual alleles and haplotypes to risk for schizophrenia, partly because clinical phenotypes are ill-defined and preclinical studies are limited by lack of adequate models. Here, we propose a neuroimaging approach to overcome these limitations by characterizing the functional impact of ambiguous haplotypes on a neural system-level intermediate phenotype in humans. Studying 126 healthy control subjects during a working-memory paradigm, we find that a previously described risk variant in a functional Val158Met (rs4680) polymorphism interacts with a P2 promoter region SNP (rs2097603) and an SNP in the 3' region (rs165599) in predicting inefficient prefrontal working memory response. We report evidence that the nonlinear response of prefrontal neurons to dopaminergic stimulation is a neural mechanism underlying these nonadditive genetic effects. This work provides an in vivo approach to functional validation in brain of the biological impact of complex genetic variations within a gene that may be critical for its clinical association.     

Load-related brain activation predicts spatial working memory performance in youth aged 9–12 and is associated with executive function at earlier ages

Spatial working memory is a central cognitive process that matures through adolescence in conjunction with major changes in brain function and anatomy. Here we focused on late childhood and early adolescence to more closely examine the neural correlates of performance variability during this important transition period. Using a modified spatial 1-back task with two memory load conditions in an fMRI study, we examined the relationship between load-dependent neural responses and task performance in a sample of 39 youth aged 9–12 years. Our data revealed that between-subject differences in task performance was predicted by load-dependent deactivation in default network regions, including the ventral anterior cingulate cortex (vACC) and posterior cingulate cortex (PCC). Although load-dependent increases in activation in prefrontal and posterior parietal regions were only weakly correlated with performance, increased prefrontal–parietal coupling was associated with better performance. Furthermore, behavioral measures of executive function from as early as age 3 predicted current load-dependent deactivation in vACC and PCC. These findings suggest that both task positive and task negative brain activation during spatial working memory contributed to successful task performance in late childhood/early adolescence. This may serve as a good model for studying executive control deficits in developmental disorders.     

Spatial Memory Sequence Encoding and Replay During Modeled Theta and Ripple Oscillations

Spatial learning involves the storage and replay of temporally ordered spatial information. The hippocampus is a key brain structure involved in spatial learning in rats. Temporally ordered spatial memories are encoded and replayed by the firing rate and phase of hippocampal pyramidal cells and inhibitory interneurons with respect to ongoing network theta and ripple oscillations paced by intra- and extrahippocampal areas. Theta oscillations (4–7 Hz) may contribute to memory formation, whereas fast ripple oscillations to temporally compressed forward and reverse replay of previously stored memories. Different classes of CA1 excitatory and inhibitory neurons and medial septal inhibitory neurons have been shown to differentially phase their activities with respect to theta and ripples. Understanding how the different hippocampal and extrahippocampal areas and their neuronal classes interact during these network oscillations and how they facilitate the storage and replay of spatiotemporal memories is of great importance. A computational model of the hippocampal CA1 microcircuit that uses biophysical representations of the major cell types, including pyramidal cells and four types of inhibitory interneurons, is extended. Inputs to the network come from the entorhinal cortex (EC), the CA3 Schaffer collaterals and the medial septum. A biophysical mechanism of spike timing-dependent plasticity (STDP) is used for learning spatial memory patterns in the correct order. The model addresses two important issues: (1) How are the storage and replay (forward and reverse) of temporally ordered memory patterns controlled in the CA1 microcircuit during theta and ripples? (2) What roles do the various types of inhibitory interneurons play in these processes?     

A neurocomputational theory of the dopaminergic modulation of working memory functions.

The dopaminergic modulation of neural activity in the prefrontal cortex (PFC) is essential for working memory. Delay-activity in the PFC in working memory tasks persists even if interfering stimuli intervene between the presentation of the sample and the target stimulus. Here, the hypothesis is put forward that the functional role of dopamine in working memory processing is to stabilize active neural representations in the PFC network and thereby to protect goal-related delay-activity against interfering stimuli. To test this hypothesis, we examined the reported dopamine-induced changes in several biophysical properties of PFC neurons to determine whether they could fulfill this function. An attractor network model consisting of model neurons was devised in which the empirically observed effects of dopamine on synaptic and voltage-gated membrane conductances could be represented in a biophysically realistic manner. In the model, the dopamine-induced enhancement of the persistent Na+ and reduction of the slowly inactivating K+ current increased firing of the delay-active neurons, thereby increasing inhibitory feedback and thus reducing activity of the "background" neurons. Furthermore, the dopamine-induced reduction of EPSP sizes and a dendritic Ca2+ current diminished the impact of intervening stimuli on current network activity. In this manner, dopaminergic effects indeed acted to stabilize current delay-activity. Working memory deficits observed after supranormal D1-receptor stimulation could also be explained within this framework. Thus, the model offers a mechanistic explanation for the behavioral deficits observed after blockade or after supranormal stimulation of dopamine receptors in the PFC and, in addition, makes some specific empirical predictions.     

The role of short-term depression in sustained neural activity in the prefrontal cortex: a simulation study.

Recent experimental researches have suggested that sustained neural activity in the prefrontal cortex is a process of memory retention in decision making. Previous theoretical studies indicate that a balance between recurrent excitation and feedback inhibition is important for sustaining the activity. To investigate a plausible balancing mechanism, we simulated a biophysically realistic network model. Our model shows that short-term depression (STD) enables the network to sustain its activity despite the presence of long-term inhibition by GABA(B) receptors and that the sustained firing rates have a bell-shaped dependence on the degree of STD. By analyzing the neural network dynamics, we show that the bell-shaped dependence on STD is formed by destabilizing the balance with either excessive or insufficient STD. We also show that the optimal degree of STD has a linear relationship with the neural network size. These results suggest that STD provides a balancing mechanism and controls levels of sustained activities of various size networks.     

Cooperation and biased competition model can explain attentional filtering in the prefrontal cortex

Recent neurophysiological experimental results suggest that the prefrontal cortex plays an important role in filtering out unattended visual inputs. Here we propose a neurodynamical computational model of a part of the prefrontal cortex to account for the neural mechanisms defining this attentional filtering effect. Similar models have been employed to explain experimental results obtained during the performance of attention and working memory tasks. In this previous work the principle of biased competition was shown to successfully account for the experimental data. To model the attentional filtering effect, the biased competition model was extended to enable cooperation between stimulus selective neurons. We show that, in a biological relevant minimal model, competition and cooperation between the neurons are sufficient conditions for reproducing the attentional effect. Furthermore, a characterization of the parameter regime where the cooperation effect is observed is presented. Finally, we also reveal parameter regimes where the network has different modes of operation: selective working memory, attentional filtering, pure competition and noncompetitive amplification.     

Specific relationship between prefrontal neuronal N-acetylaspartate and activation of the working memory cortical network in schizophrenia

OBJECTIVE: Abnormal activation of the dorsolateral prefrontal cortex and a related cortical network during working memory tasks has been demonstrated in patients with schizophrenia, but the responsible mechanism has not been identified. The present study was performed to determine whether neuronal pathology of the dorsolateral prefrontal cortex is linked to the activation of the working memory cortical network in patients with schizophrenia. METHOD: The brains of 13 patients with schizophrenia and 13 comparison subjects were studied with proton magnetic resonance spectroscopic ((1)H-MRS) imaging (to measure N-acetylaspartate as a marker of neuronal pathology) and with [(15)O]water positron emission tomography (PET) during performance of the Wisconsin Card Sorting Test (to measure activation of the working memory cortical network). An independent cohort of patients (N=7) was also studied in a post hoc experiment with (1)H-MRS imaging and with the same PET technique during performance of another working memory task (the "N-back" task). RESULTS: Measures of N-acetylaspartate in the dorsolateral prefrontal cortex strongly correlated with activation of the distributed working memory network, including the dorsolateral prefrontal, temporal, and inferior parietal cortices, during both working memory tasks in the two independent groups of patients with schizophrenia. In contrast, N-acetylaspartate in other cortical regions and in comparison subjects did not show these relationships. CONCLUSIONS: These findings directly implicate a population of dorsolateral prefrontal cortex neurons as selectively accounting for the activity of the distributed working memory cortical network in schizophrenia and complement other evidence that dorsolateral prefrontal cortex connectivity is fundamental to the pathophysiology of the disorder.