scholarly journals A Robust Model of Gated Working Memory

2019 ◽  
Author(s):  
Anthony Strock ◽  
Xavier Hinaut ◽  
Nicolas P. Rougier
Keyword(s):  
Working Memory ◽  
Computational Models ◽  
Extended Period ◽  
Inhibitory Neurons ◽  
Population Mixing ◽  
Robust Model ◽  
Global Weight ◽  
Electrophysiological Recordings ◽  
Real Value ◽  
Weight Scaling

AbstractGated working memory is defined as the capacity of holding arbitrary information at any time in order to be used at a later time. Based on electrophysiological recordings, several computational models have tackled the problem using dedicated and explicit mechanisms. We propose instead to consider an implicit mechanism based on a random recurrent neural network. We introduce a robust yet simple reservoir model of gated working memory with instantaneous updates. The model is able to store an arbitrary real value at random time over an extended period of time. The dynamics of the model is a line attractor that learns to exploit reentry and a non-linearity during the training phase using only a few representative values. A deeper study of the model shows that there is actually a large range of hyper parameters for which the results hold (number of neurons, sparsity, global weight scaling, etc.) such that any large enough population, mixing excitatory and inhibitory neurons can quickly learn to realize such gated working memory. In a nutshell, with a minimal set of hypotheses, we show that we can have a robust model of working memory. This suggests this property could be an implicit property of any random population, that can be acquired through learning. Furthermore, considering working memory to be a physically open but functionally closed system, we give account on some counter-intuitive electrophysiological recordings.

scholarly journals A Robust Model of Gated Working Memory

2020 ◽  
Vol 32 (1) ◽  
pp. 153-181
Author(s):  
Anthony Strock ◽  
Xavier Hinaut ◽  
Nicolas P. Rougier
Keyword(s):  
Working Memory ◽  
Computational Models ◽  
Extended Period ◽  
Inhibitory Neurons ◽  
Population Mixing ◽  
Robust Model ◽  
Global Weight ◽  
Electrophysiological Recordings ◽  
Real Value ◽  
Weight Scaling

Gated working memory is defined as the capacity of holding arbitrary information at any time in order to be used at a later time. Based on electrophysiological recordings, several computational models have tackled the problem using dedicated and explicit mechanisms. We propose instead to consider an implicit mechanism based on a random recurrent neural network. We introduce a robust yet simple reservoir model of gated working memory with instantaneous updates. The model is able to store an arbitrary real value at random time over an extended period of time. The dynamics of the model is a line attractor that learns to exploit reentry and a nonlinearity during the training phase using only a few representative values. A deeper study of the model shows that there is actually a large range of hyperparameters for which the results hold (e.g., number of neurons, sparsity, global weight scaling) such that any large enough population, mixing excitatory and inhibitory neurons, can quickly learn to realize such gated working memory. In a nutshell, with a minimal set of hypotheses, we show that we can have a robust model of working memory. This suggests this property could be an implicit property of any random population, that can be acquired through learning. Furthermore, considering working memory to be a physically open but functionally closed system, we give account on some counterintuitive electrophysiological recordings.

scholarly journals Fronto-medial theta coordinates posterior maintenance of working memory content

2021 ◽  
Author(s):  
Oliver Ratcliffe ◽  
Kimron Shapiro ◽  
Bernhard P. Staresina
Keyword(s):  
Working Memory ◽  
Computational Models ◽  
Medial Cortex ◽  
Duty Cycles ◽  
Machine Learning Approach ◽  
Theta Frequency ◽  
Oscillatory Power ◽  
Posterior Parietal ◽  
Memory Content ◽  
Memory Contents

AbstractHow does the human brain manage multiple bits of information to guide goal-directed behaviour? Successful working memory (WM) functioning has consistently been linked to oscillatory power in the theta frequency band (4-8 Hz) over fronto-medial cortex (fronto-medial theta, FMT). Specifically, FMT is thought to reflect the mechanism of an executive sub-system that coordinates maintenance of memory contents in posterior regions. However, direct evidence for the role of FMT in controlling specific WM content is lacking. Here we collected high-density Electroencephalography (EEG) data whilst participants engaged in load-varying WM tasks and then used multivariate decoding methods to examine WM content during the maintenance period. Higher WM load elicited a focal increase in FMT. Importantly, decoding of WM content was driven by posterior/parietal sites, which in turn showed load-induced functional theta coupling with fronto-medial cortex. Finally, we observed a significant slowing of FMT frequency with increasing WM load, consistent with the hypothesised broadening of a theta ‘duty cycle’ to accommodate additional WM items. Together these findings demonstrate that frontal theta orchestrates posterior maintenance of WM content. Moreover, the observed frequency slowing elucidates the function of FMT oscillations by specifically supporting phase-coding accounts of WM.Significance StatementHow does the brain juggle the maintenance of multiple items in working memory (WM)? Here we show that increased WM demands increase theta power (4-8 Hz) in fronto-medial cortex. Interestingly, using a machine learning approach, we found that the content held in WM could be read out not from frontal, but from posterior areas. These areas were in turn functionally coupled with fronto-medial cortex, consistent with the idea that frontal cortex orchestrates WM representations in posterior regions. Finally, we observed that holding an additional item in WM leads to significant slowing of the frontal theta rhythm, supporting computational models that postulate longer ‘duty cycles’ to accommodate additional WM demands.

scholarly journals Active information maintenance in working memory by a sensory cortex

2018 ◽  
Author(s):  
Xiaoxing Zhang ◽  
Wenjun Yan ◽  
Wenliang Wang ◽  
Hongmei Fan ◽  
Ruiqing Hou ◽  
...  
Keyword(s):  
Working Memory ◽  
Piriform Cortex ◽  
Delay Period ◽  
Sensory Cortex ◽  
Critical Function ◽  
Impaired Performance ◽  
Anterior Piriform Cortex ◽  
Electrophysiological Recordings ◽  
Dual Task Paradigm ◽  
The Brain

SummaryWorking memory is a critical function of the brain to maintain and manipulate information over delay periods of seconds. Sensory areas have been implicated in working memory; however, it is debated whether the delay-period activity of sensory regions is actively maintaining information or passively reflecting top-down inputs. We hereby examined the anterior piriform cortex, an olfactory cortex, in head-fixed mice performing a series of olfactory working memory tasks. Information maintenance is necessary in these tasks, especially in a dual-task paradigm in which mice are required to perform another distracting task while actively maintaining information during the delay period. Optogenetic suppression of the piriform cortex activity during the delay period impaired performance in all the tasks.Furthermore, electrophysiological recordings revealed that the delay-period activity of the anterior piriform cortex encoded odor information with or without the distracting task.Thus, this sensory cortex is critical for active information maintenance in working memory.

scholarly journals Neuronal and Astrocytic Regulations in Schizophrenia: A Computational Modelling Study

2021 ◽  
Vol 15 ◽  
Author(s):  
Lea Fritschi ◽  
Johanna Hedlund Lindmar ◽  
Florian Scheidl ◽  
Kerstin Lenk
Keyword(s):  
Computational Models ◽  
Glutamate Release ◽  
Signal Transmission ◽  
Computational Modelling ◽  
Atp Release ◽  
Network Activity ◽  
Inhibitory Neurons ◽  
Neuronal Network Activity ◽  
Cell Densities ◽  
Synapse Model

According to the tripartite synapse model, astrocytes have a modulatory effect on neuronal signal transmission. More recently, astrocyte malfunction has been associated with psychiatric diseases such as schizophrenia. Several hypotheses have been proposed on the pathological mechanisms of astrocytes in schizophrenia. For example, post-mortem examinations have revealed a reduced astrocytic density in patients with schizophrenia. Another hypothesis suggests that disease symptoms are linked to an abnormality of glutamate transmission, which is also regulated by astrocytes (glutamate hypothesis of schizophrenia). Electrophysiological findings indicate a dispute over whether the disorder causes an increase or a decrease in neuronal and astrocytic activity. Moreover, there is no consensus as to which molecular pathways and network mechanisms are altered in schizophrenia. Computational models can aid the process in finding the underlying pathological malfunctions. The effect of astrocytes on the activity of neuron-astrocyte networks has been analysed with computational models. These can reproduce experimentally observed phenomena, such as astrocytic modulation of spike and burst signalling in neuron-astrocyte networks. Using an established computational neuron-astrocyte network model, we simulate experimental data of healthy and pathological networks by using different neuronal and astrocytic parameter configurations. In our simulations, the reduction of neuronal or astrocytic cell densities yields decreased glutamate levels and a statistically significant reduction in the network activity. Amplifications of the astrocytic ATP release toward postsynaptic terminals also reduced the network activity and resulted in temporarily increased glutamate levels. In contrast, reducing either the glutamate release or re-uptake in astrocytes resulted in higher network activities. Similarly, an increase in synaptic weights of excitatory or inhibitory neurons raises the excitability of individual cells and elevates the activation level of the network. To conclude, our simulations suggest that the impairment of both neurons and astrocytes disturbs the neuronal network activity in schizophrenia.

scholarly journals Atypically larger variability of resource allocation accounts for visual working memory deficits in schizophrenia

2018 ◽  
Author(s):  
Yi-Jie Zhao ◽  
Tianye Ma ◽  
Xuemei Ran ◽  
Li Zhang ◽  
Ru-Yuan Zhang ◽  
...  
Keyword(s):  
Working Memory ◽  
Visual Working Memory ◽  
Computational Models ◽  
Variable Precision ◽  
Capacity Theory ◽  
Healthy Control ◽  
New Perspective ◽  
Almost All ◽  
Memory Precision ◽  
Memory Resources

AbstractSchizophrenia patients are known to have profound deficits in visual working memory (VWM), and almost all previous studies attribute the deficits to decreased memory capacity. This account, however, ignores the potential contributions of other VWM components (e.g., memory precision). Here, we measure the VWM performance of schizophrenia patients and healthy control subjects on two classical delay-estimation tasks. Moreover, we thoroughly evaluate several established computational models of VWM to compare the performance of the two groups. We find that the model assuming variable precision across items and trials is the best model to explain the performance of both groups. According to the variable-precision model, schizophrenia subjects exhibit abnormally larger variability of allocating memory resources rather than resources per se. These results invite a rethink of the widely accepted decreased-capacity theory and propose a new perspective on the diagnosis and rehabilitation of schizophrenia.

scholarly journals 0074 Inhibitory Neurons in the Dorsomedial Medulla Promote REM Sleep

SLEEP ◽  
2020 ◽  
Vol 43 (Supplement_1) ◽  
pp. A30-A30
Author(s):  
J Stucynski ◽  
A Schott ◽  
J Baik ◽  
J Hong ◽  
F Weber ◽  
...  
Keyword(s):  
Rem Sleep ◽  
Nrem Sleep ◽  
Inhibitory Neurons ◽  
Brain State ◽  
Sleep Regulation ◽  
Optogenetic Stimulation ◽  
Electrophysiological Recordings ◽  
Stimulation Of

Abstract Introduction The neural circuits controlling rapid eye movement (REM) sleep, and in particular the role of the medulla in regulating this brain state, remains an active area of study. Previous electrophysiological recordings in the dorsomedial medulla (DM) and electrical stimulation experiments suggested an important role of this area in the control of REM sleep. However the identity of the involved neurons and their precise role in REM sleep regulation are still unclear. Methods The properties of DM GAD2 neurons in mice were investigated through stereotaxic injection of CRE-dependent viruses in conjunction with implantation of electrodes for electroencephalogram (EEG) and electromyogram (EMG) recordings and optic fibers. Experiments included in vivo calcium imaging (fiber photometry) across sleep and wake states, optogenetic stimulation of cell bodies, chemogenetic excitation and suppression (DREADDs), and connectivity mapping using viral tracing and optogenetics. Results Imaging the calcium activity of DM GAD2 neurons in vivo indicates that these neurons are most active during REM sleep. Optogenetic stimulation of DM GAD2 neurons reliably triggered transitions into REM sleep from NREM sleep. Consistent with this, chemogenetic activation of DM GAD2 neurons increased the amount of REM sleep while inhibition suppressed its occurrence and enhanced NREM sleep. Anatomical tracing revealed that DM GAD2 neurons project to several areas involved in sleep / wake regulation including the wake-promoting locus coeruleus (LC) and the REM sleep-suppressing ventrolateral periaquaductal gray (vlPAG). Optogenetic activation of axonal projections from DM to LC, and DM to vlPAG was sufficient to induce REM sleep. Conclusion These experiments demonstrate that DM inhibitory neurons expressing GAD2 powerfully promote initiation of REM sleep in mice. These findings further characterize the dorsomedial medulla as a critical structure involved in REM sleep regulation and inform future investigations of the REM sleep circuitry. Support R01 HL149133

scholarly journals Making Working Memory Work: A Computational Model of Learning in the Prefrontal Cortex and Basal Ganglia

2006 ◽  
Vol 18 (2) ◽  
pp. 283-328 ◽  
Author(s):  
Randall C. O'Reilly ◽  
Michael J. Frank
Keyword(s):  
Working Memory ◽  
Prefrontal Cortex ◽  
Basal Ganglia ◽  
Computational Model ◽  
Computational Models ◽  
Memory Task ◽  
Learning Mechanisms ◽  
Subcortical Structures ◽  
Credit Assignment ◽  
Gating Mechanism

The prefrontal cortex has long been thought to subserve both working memory (the holding of information online for processing) and executive functions (deciding how to manipulate working memory and perform processing). Although many computational models of working memory have been developed, the mechanistic basis of executive function remains elusive, often amounting to a homunculus. This article presents an attempt to deconstruct this homunculus through powerful learning mechanisms that allow a computational model of the prefrontal cortex to control both itself and other brain areas in a strategic, task-appropriate manner. These learning mechanisms are based on subcortical structures in the midbrain, basal ganglia, and amygdala, which together form an actor-critic architecture. The critic system learns which prefrontal representations are task relevant and trains the actor, which in turn provides a dynamic gating mechanism for controlling working memory updating. Computationally, the learning mechanism is designed to simultaneously solve the temporal and structural credit assignment problems. The model's performance compares favorably with standard backpropagation-based temporal learning mechanisms on the challenging 1-2-AX working memory task and other benchmark working memory tasks.

scholarly journals Perceptual stimuli with novel bindings interfere with visual working memory

2021 ◽  
Author(s):  
Peter Shepherdson
Keyword(s):  
Working Memory ◽  
Change Detection ◽  
Visual Working Memory ◽  
Computational Models ◽  
Recall Performance ◽  
Visual Interference ◽  
Key Factor ◽  
Subsequent Recall ◽  
New Feature ◽  
Change Response

AbstractWhat influences the extent to which perceptual information interferes with the contents of visual working memory? In two experiments using a combination of change detection and continuous reproduction tasks, I show that binding novelty is a key factor in producing interference. In Experiment 2, participants viewed arrays of colored circles, then completed consecutive change detection and recall tests of their memory for stochastically independent items from the same array. When the probe used in the change detection test was novel (i.e., required a “change” response), subsequent recall performance was worse than in trials with matching (i.e., “no change”) probes, irrespective of whether or not the same item was tested in both phases. In Experiment 2, participants viewed arrays of oriented arrows, then completed a change detection (requiring memory) or direction judgement (not requiring memory) test, followed by recalling a stochastically independent item. Again, novel probes in the first phase led to worse recall, irrespective of whether the initial task required memory. This effect held whether the probe was wholly novel (i.e., a new feature presented at any location) or simply involved a novel binding (i.e., an old feature presented at a new location). These findings highlight the role of novelty in visual interference, consistent with the assumptions of computational models of WM, and suggest that new bindings of old information are sufficient to produce such interference.

scholarly journals Highly task-specific and distributed neural connectivity in working memory revealed by single-trial decoding in mice and humans

2021 ◽  
Author(s):  
Daniel Strahnen ◽  
Sampath K.T. Kapanaiah ◽  
Alexei M. Bygrave ◽  
Birgit Liss ◽  
David M. Bannerman ◽  
...  
Keyword(s):  
Working Memory ◽  
Activity Patterns ◽  
Brain Regions ◽  
Cognitive Domain ◽  
Neural Connectivity ◽  
Task Specificity ◽  
Electrophysiological Recordings ◽  
Connectivity Patterns ◽  
Unbiased Manner ◽  
Beta Frequency

AbstractWorking memory (WM), the capacity to briefly and intentionally maintain mental items, is key to successful goal-directed behaviour and impaired in a range of psychiatric disorders. To date, several brain regions, connections, and types of neural activity have been correlatively associated with WM performance. However, no unifying framework to integrate these findings exits, as the degree of their species- and task-specificity remains unclear. Here, we investigate WM correlates in three task paradigms each in mice and humans, with simultaneous multi-site electrophysiological recordings. We developed a machine learning-based approach to decode WM-mediated choices in individual trials across subjects from hundreds of electrophysiological measures of neural connectivity with up to 90% prediction accuracy. Relying on predictive power as indicator of correlates of psychological functions, we unveiled a large number of task phase-specific WM-related connectivity from analysis of predictor weights in an unbiased manner. Only a few common connectivity patterns emerged across tasks. In rodents, these were thalamus-prefrontal cortex delta- and beta-frequency connectivity during memory encoding and maintenance, respectively, and hippocampal-prefrontal delta- and theta-range coupling during retrieval, in rodents. In humans, task-independent WM correlates were exclusively in the gamma-band. Mostly, however, the predictive activity patterns were unexpectedly specific to each task and always widely distributed across brain regions. Our results suggest that individual tasks cannot be used to uncover generic physiological correlates of the psychological construct termed WM and call for a new conceptualization of this cognitive domain in translational psychiatry.

scholarly journals Uncertainty is Maintained and Used in Working Memory

2020 ◽  
Author(s):  
Aspen H. Yoo ◽  
Luigi Acerbi ◽  
Wei ji Ma
Keyword(s):  
Working Memory ◽  
Computational Models ◽  
Memory Task ◽  
Point Estimate ◽  
Memory Representation ◽  
Test Model ◽  
Specific Level ◽  
Uncertainty Model ◽  
Observer Model ◽  
Uncertainty Information

1AbstractWhat are the contents of working memory? In both behavioral and neural computational models, the working memory representation of a stimulus is typically described by a single number, namely a point estimate of that stimulus. Here, we asked if people also maintain the uncertainty associated with a memory, and if people use this uncertainty in subsequent decisions. We collected data in a two-condition orientation change detection task; while both conditions measured whether people used memory uncertainty, only one required maintaining it. For each condition, we compared an optimal Bayesian observer model, in which the observer uses an accurate representation of uncertainty in their decision, to one in which the observer does not. We find that this “Use Uncertainty” model fits better for all participants in both conditions. In the first condition, this result suggests that people use uncertainty optimally in a working memory task when that uncertainty information is available at the time of decision, confirming earlier results. Critically, the results of the second condition suggest that this uncertainty information was maintained in working memory. We test model variants and find that our conclusions do not depend on our assumptions about the observer’s encoding process, inference process, or decision rule. Our results provide evidence that people have uncertainty that reflects their memory precision at an item-specific level, maintain this information over a working memory delay, and use it implicitly in a way consistent with an optimal observer. These results challenge existing computational models of working memory to update their frameworks to represent uncertainty.

Sign in / Sign up

Export Citation Format

Share Document