Working Memory Buffer Research Articles

Reward-based option competition in human dorsal stream and transition from stochastic exploration to exploitation in continuous space.

Primates exploring and exploiting a continuous sensorimotor space rely on dynamic maps in the dorsal stream. Two complementary perspectives exist on how these maps encode rewards. Reinforcement learning models integrate rewards incrementally over time, efficiently resolving the exploration/exploitation dilemma. Working memory buffer models explain rapid plasticity of parietal maps but lack a plausible exploration/exploitation policy. The reinforcement learning model presented here unifies both accounts, enabling rapid, information-compressing map updates and efficient transition from exploration to exploitation. As predicted by our model, activity in human frontoparietal dorsal stream regions, but not in MT+, tracks the number of competing options, as preferred options are selectively maintained on the map, while spatiotemporally distant alternatives are compressed out. When valuable new options are uncovered, posterior β1/α oscillations desynchronize within 0.4 to 0.7 s, consistent with option encoding by competing β1-stabilized subpopulations. Together, outcomes matching locally cached reward representations rapidly update parietal maps, biasing choices toward often-sampled, rewarded options.

The dynamics of auditory working memory impairment in logopenic aphasia and Alzheimer’s disease

AbstractBackgroundImpaired auditory verbal working memory is a diagnostic hallmark and integral driver of the clinical phenotype in logopenic aphasia. However, the physiology of the working memory buffer in this syndrome is poorly characterised. Here we addressed the temporal dynamics of auditory verbal working memory in patients with logopenic aphasia and typical Alzheimer’s disease.MethodIn a cohort of patients with logopenic aphasia and typical Alzheimer’s disease and in healthy age‐matched controls, we assessed how temporal manipulations of standard auditory verbal working memory tasks (forward and reverse digit span and phrasal repetition) affected performance. We varied tempo of delivery and inter‐trial gap and performed a detailed analysis of error types. Participants also had pure tone audiometry to assess peripheral hearing function and a comprehensive general neuropsychological assessment.ResultCompared with healthy controls, patients with logopenic aphasia and Alzheimer’s disease showed increased sensitivity to temporal manipulation of auditory verbal working memory tasks and error profiles suggesting a dynamic physiological lesion of the working memory buffer. These effects were particularly marked in the logopenic group.ConclusionOur findings open a novel physiological window on working memory dynamics in Alzheimer’s disease syndromes. Further work is warranted to assess how this dynamic deficit impacts communication in patients’ daily lives, how it can best inform management interventions and its potential as a novel, rapid read‐out of neural function in the era of disease‐modifying therapies.

Parental working memory buffers associations between COVID-19 hardships and child mental health

The coronavirus (COVID-19) pandemic has impaired young children's mental health, underscoring the need for research on protective factors. Using a cross-sectional design, we examined whether parental working memory (WM) buffered relations between COVID-19 hardships (home-life, economic, and quarantine) and child internalizing and externalizing symptoms. Female parents (N = 339; 83.19% White/Caucasian, 8.85% Black/African American, 3.54% Asian, 1.47% Native American, and 2.36% mixed race; 7.67% Hispanic/Latinx ethnicity) of children 2–5-years-old reported COVID-19 hardships and child internalizing and externalizing symptoms and completed a backward digit span task to measure WM. All types of COVID-19 hardships were positively related to child internalizing and externalizing symptoms. Higher parental WM weakened positive relations between all types of hardships and child internalizing symptoms, and between home-life and economic hardships and externalizing symptoms. Results suggest that parental WM, a malleable target for intervention, may buffer associations between the detrimental effects of COVID-19 and young children's mental health.

White Matter Correlates of Domain-Specific Working Memory.

Prior evidence suggests domain-specific working memory (WM) buffers for maintaining phonological (i.e., speech sound) and semantic (i.e., meaning) information. The phonological WM buffer's proposed location is in the left supramarginal gyrus (SMG), whereas semantic WM has been related to the left inferior frontal gyrus (IFG), the middle frontal gyrus (MFG), and the angular gyrus (AG). However, less is known about the white matter correlates of phonological and semantic WM. We tested 45 individuals with left hemisphere brain damage on single word processing, phonological WM, and semantic WM tasks and obtained T1 and diffusion weighted neuroimaging. Virtual dissections were performed for each participants' arcuate fasciculus (AF), inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), middle longitudinal fasciculus (MLF), and uncinate fasciculus (UF), which connect the proposed domain-specific WM buffers with perceptual or processing regions. The results showed that the left ILF, MLF, IFOF, and the direct segment of the AF were related to semantic WM performance. Phonological WM was related to both the left ILF and the MLF. This work informs our understanding of the white matter correlates of WM, especially semantic WM, which has not previously been investigated. In addition, this work helps to adjudicate between theories of verbal WM, providing some evidence for separate pathways supporting phonological and semantic WM.

Active Working Memory and Simple Cognitive Operations.

Working memory is thought to serve as a buffer for ongoing cognitive operations, even in tasks that have no obvious memory requirements. This conceptualization has been supported by dual-task experiments, in which interference is observed between a primary task involving short-term memory storage and a secondary task that presumably requires the same buffer as the primary task. Little or no interference is typically observed when the secondary task is very simple. Here, we test the hypothesis that even very simple tasks require the working memory buffer, but interference can be minimized by using activity-silent representations to store the information from the primary task. We tested this hypothesis using dual-task paradigm in which a simple discrimination task was interposed in the retention interval of a change detection task. We used contralateral delay activity (CDA) to track the active maintenance of information for the change detection task. We found that the CDA was massively disrupted after the interposed task. Despite this disruption of active maintenance, we found that performance in the change detection task was only slightly impaired, suggesting that activity-silent representations were used to retain the information for the change detection task. A second experiment replicated this result and also showed that automated discriminations could be performed without producing a large CDA disruption. Together, these results suggest that simple but non-automated discrimination tasks require the same processes that underlie active maintenance of information in working memory.

Can posthypnotic suggestions boost updating in working memory? Behavioral and ERP evidence

Updating is an essential executive function (EF), responsible for storing, retrieving, and substituting information in working memory (WM). Here we investigated whether posthypnotic suggestions (PHS) given to high-hypnotizable participants can enhance updating in WM and measured neural correlates of the observed effects by recording event-related brain potentials (ERP). In a tone-monitoring task different syllables were presented in random order, requiring a response to every fourth presentation of a given syllable. Experiment 1 (n = 19) established the relationship between performance and several ERP components across updating load (different numbers of syllables). In Experiment 2 (n = 18), a no-hypnosis (NH) and a hypnosis-plus-PHS session were administrated in counterbalanced order. Task instructions, presented at the beginning of the sessions, emphasized a cognitive strategy, demanding imagination of visual counters, a strategy that was also emphasized during PHS. PHS additionally contained suggestions stimulating cognitive simulation of the task, where participants were advised to apply the suggested strategy. Relative to the NH session, PHS enhanced WM performance with medium to large effect sizes. In ERPs, PHS increased the P2 and P3 components, indicating the proactive recruitment of control-related attention and updating-related cognitive control processes, respectively. PHS also reduced updating load effects in the posterior recognition component, suggesting diminished demands on WM buffers. These ERP findings suggest that PHS enhanced updating in WM by strengthening proactive control, which may have diminished the necessity for reactive control. Hence, the present results suggest that our PHS had worked like mental practice helping participants to develop an efficient and context-dependent trigger-action contingency. Consequentially, the present study provides a new framework for employing PHSs, which may be used as a basis for developing new training regimes for modifying WM or other EFs.

Relation Between Working Memory Capacity of Biological Movements and Fluid Intelligence.

Studies have revealed that there is an independent buffer for holding biological movements (BM) in working memory (WM), and this BM-WM has a unique link to our social ability. However, it remains unknown as to whether the BM-WM also correlates to our cognitive abilities, such as fluid intelligence (Gf). Since BM processing has been considered as a hallmark of social cognition, which distinguishes from canonical cognitive abilities in many ways, it has been hypothesized that only canonical object-WM (e.g., memorizing color patches), but not BM-WM, emerges to have an intimate relation with Gf. We tested this prediction by measuring the relationship between WM capacity of BM and Gf. With two Gf measurements, we consistently found moderate correlations between BM-WM capacity, the score of both Raven’s advanced progressive matrix (RAPM), and the Cattell culture fair intelligence test (CCFIT). This result revealed, for the first time, a close relation between WM and Gf with a social stimulus, and challenged the double-dissociation hypothesis for distinct functions of different WM buffers.

Parietal low beta rhythm provides a dynamical substrate for a working memory buffer

Working memory (WM) is a component of the brain's memory systems vital for interpretation of sequential sensory inputs and consequent decision making. Anatomically, WM is highly distributed over the prefrontal cortex (PFC) and the parietal cortex (PC). Here we present a biophysically detailed dynamical systems model for a WM buffer situated in the PC, making use of dynamical properties believed to be unique to this area. We show that the natural beta1 rhythm (12 to 20 Hz) of the PC provides a substrate for an episodic buffer that can synergistically combine executive commands (e.g., from PFC) and multimodal information into a flexible and updatable representation of recent sensory inputs. This representation is sensitive to distractors, it allows for a readout mechanism, and it can be readily terminated by executive input. The model provides a demonstration of how information can be usefully stored in the temporal patterns of activity in a neuronal network rather than just synaptic weights between the neurons in that network.

Children’s suffix effects for verbal working memory reflect phonological coding and perceptual grouping

When listeners recall order of presentation for sequences of unrelated words, recall is most accurate for first and final items. When a speech suffix is appended to the list, however, the advantage for final items is diminished. The usual interpretation is that listeners recover phonological structure from speech signals and use that structure to store items in a working memory buffer; the process of recovering phonological structure for a suffix interrupts that processing for the final list item. Although not mutually exclusive, another hypothesis suggests that perceptual grouping of list items and suffix based on common acoustic structure is necessary for the effect to occur. To evaluate these accounts as well as potential age-related differences, adults and 8-year-old children were asked to recall order of presentation for a closed set of nouns in five suffix conditions: none, auditory go, lipread go, a tone, and a colored circle. Overt articulation was prohibited, but attention to the suffix was mandated. Children’s serial recall was generally poorer than that of adults, but patterns across list positions were similar for both age groups. Participants showed stronger effects for speech suffixes than for nonspeech suffixes regardless of whether suffixes were seen or heard, but effects were not restricted to final list items. And although effects of heard and lipread suffixes were similar for early list items, heard speech exerted greater effects on late list items. Outcomes suggest that some effect of heard and lipread speech suffixes arises from their shared phonological structure, but this effect is strongest when perceptual grouping occurs.

How the Brain Represents Language and Answers Questions? Using an AI System to Understand the Underlying Neurobiological Mechanisms.

To understand the computations that underlie high-level cognitive processes we propose a framework of mechanisms that could in principle implement START, an AI program that answers questions using natural language. START organizes a sentence into a series of triplets, each containing three elements (subject, verb, object). We propose that the brain similarly defines triplets and then chunks the three elements into a spatial pattern. A complete sentence can be represented using up to 7 triplets in a working memory buffer organized by theta and gamma oscillations. This buffer can transfer information into long-term memory networks where a second chunking operation converts the serial triplets into a single spatial pattern in a network, with each triplet (with corresponding elements) represented in specialized subregions. The triplets that define a sentence become synaptically linked, thereby encoding the sentence in synaptic weights. When a question is posed, there is a search for the closest stored memory (having the greatest number of shared triplets). We have devised a search process that does not require that the question and the stored memory have the same number of triplets or have triplets in the same order. Once the most similar memory is recalled and undergoes 2-level dechunking, the sought for information can be obtained by element-by-element comparison of the key triplet in the question to the corresponding triplet in the retrieved memory. This search may require a reordering to align corresponding triplets, the use of pointers that link different triplets, or the use of semantic memory. Our framework uses 12 network processes; existing models can implement many of these, but in other cases we can only suggest neural implementations. Overall, our scheme provides the first view of how language-based question answering could be implemented by the brain.

Competing inhibition-stabilized networks in sensory and memory processing.

In simplified models of neocortical circuits, inhibition is either modeled in a feedforward manner or through mutual inhibitory interactions that provide for competition between neuronal populations. By contrast, recent work has suggested a critical role for recurrent inhibition as a negative feedback element that stabilizes otherwise unstable recurrent excitation. Here, we show how models based upon a motif of recurrently connected "E-I" pairs of excitatory and inhibitory units can be used to describe experimental observations in sensory and memory networks. In a sensory network model of binocular rivalry, a model based on competing E-I motifs captures psychophysical observations about how incongruous images presented to the two eyes compete. In a model of cortical working memory, an architecturally similar model with modified synaptic time constants can mathematically accumulate signals into a working memory buffer in a manner that is robust to the abrupt removal of cells. These results suggest the inhibition-stabilized E-I motif as a fundamental building block for models of a wide array of neocortical dynamics.

鄂伦春族的视空间工作记忆能力优势：生态环境和生产方式的影响

<p id="C2">Working memory refers to a system of temporary holding and manipulation of information during the performance of a range of cognitive tasks, such as comprehension, learning, and reasoning. According to <xref xml:base="bibr" rid="b4">Baddeley (1974)</xref>, the architecture of working memory comprises four separated components, namely, phonological loop, visual-spatial working memory, central executive, and episodic buffer. The visual-spatial working memory is involved in the temporary retention of visual-spatial information. <p id="C101">The Oroqen is one of the oldest ethnic groups in northeast China. Most Oroqens live in the Oroqen Autonomous Banner in the Greater Hinggan Mountains. The area is bestowed with rich resources with a variety of wild animals. For generations, the Oroqens lived a life of hunting and fishing in the forests. They went on hunting expeditions in groups. Researchers explored the ecological environment and mode of production that affected the performance of cognition. The present study aims to explore whether the natural environment, which has formed the Oroqen’s hunting culture and hunting-gathering stage of economic production, affects their visual-spatial working memory. <p id="C102">In Experiment 1, four experiments were conducted to investigate the effect of the ecological environment and mode of production on the visuospatial working memory Oroqens. Oroqen (<italic>n</italic> = 32) and Han (<italic>n</italic> = 30) high school students performed simple storage and complex visual-spatial span tasks. Simple storage tasks were distinguished into three presentation formats: (i) visual, which involved maintaining irregular figures; (ii) spatial-sequential, which involved maintaining sequentially-presented locations; and (iii) spatial-simultaneous, which involved maintaining patterns of locations. The results proved that Oroqens performed significantly better than Hans in all the four visual-spatial working memory tasks. <p id="C103">In Experiment 2, the Corsi block test presented in a computer was employed to investigate whether the structural, quantitative, and path complexity influenced the visual-spatial working memory of Oroqen (<italic>n</italic> = 30) and Han (<italic>n</italic> = 30) students. Results showed that superiority of the Oroqen on visual-spatial working memory still persisted when the memory materials differ in terms of difficulty. <p id="C104">In summary, the ecological environment and mode of production have a significant impact on visual-spatial working memory. The Oroqen people’s ecological environment and traditional mode of production have become a genetic factor that may affect the visual-spatial working memory of their offspring.

Refining understanding of working memory buffers through the construct of binding: Evidence from a single case informs theory and clinical practise

Binding operations carried out in working memory enable the integration of information from different sources during online performance. While available evidence suggests that working memory may involve distinct binding functions, whether or not they all involve the episodic buffer as a cognitive substrate remains unclear. Similarly, knowledge about the neural underpinnings of working memory buffers is limited, more specifically regarding the involvement of medial temporal lobe structures. In the present study, we report on the case of patient KA, with developmental amnesia and selective damage to the whole hippocampal system. We found that KA was unable to hold shape-colours associations (relational binding) in working memory. In contrast, he could hold integrated coloured shapes (conjunctive binding) in two different tasks. Otherwise, and as expected, KA was impaired on three relational memory tasks thought to depend on the hippocampus that are widely used in the early detection of Alzheimer's disease. Our results emphasize a dissociation between two binding processes within working memory, suggesting that the visuo-spatial sketchpad could support conjunctive binding, and may rely upon a large cortical network including sub-hippocampal structures. By contrast, we found evidence for a selective impairment of relational binding in working memory when the hippocampal system is compromised, suggesting that the long-term memory deficit observed in amnesic patients may be related to impaired short-term relational binding at encoding. Finally, these findings may inform research on the early detection of Alzheimer's disease as the preservation of conjunctive binding in KA is in sharp contrast with the impaired performance demonstrated very early in this disease.

Novelty modulates human striatal activation and prefrontal\u2013striatal effective connectivity during working memory encoding

The functional role of the basal ganglia (BG) in the gating of suitable motor responses to the cortex is well established. Growing evidence supports an analogous role of the BG during working memory encoding, a task phase in which the “input-gating” of relevant materials (or filtering of irrelevant information) is an important mechanism supporting cognitive capacity and the updating of working memory buffers. One important aspect of stimulus relevance is the novelty of working memory items, a quality that is understudied with respect to its effects on corticostriatal function and connectivity. To this end, we used functional magnetic resonance imaging (fMRI) in 74 healthy volunteers performing an established Sternberg working memory task with different task phases (encoding vs. retrieval) and degrees of stimulus familiarity (novel vs. previously trained). Activation analyses demonstrated a highly significant engagement of the anterior striatum, in particular during the encoding of novel working memory items. Dynamic causal modeling (DCM) of corticostriatal circuit connectivity identified a selective positive modulatory influence of novelty encoding on the connection from the dorsolateral prefrontal cortex (DLPFC) to the anterior striatum. These data extend prior research by further underscoring the relevance of the BG for human cognitive function and provide a mechanistic account of the DLPFC as a plausible top-down regulatory element of striatal function that may facilitate the “input-gating” of novel working memory materials.

The anterior-ventrolateral temporal lobe contributes to boosting visual working memory capacity for items carrying semantic information

Working memory (WM) is a buffer that temporarily maintains information, be it visual or auditory, in an active state, caching its contents for online rehearsal or manipulation. How the brain enables long-term semantic knowledge to affect the WM buffer is a theoretically significant issue awaiting further investigation. In the present study, we capitalise on the knowledge about famous individuals as a ‘test-case’ to study how it impinges upon WM capacity for human faces and its neural substrate. Using continuous theta-burst transcranial stimulation combined with a psychophysical task probing WM storage for varying contents, we provide compelling evidence that (1) faces (regardless of familiarity) continued to accrue in the WM buffer with longer encoding time, whereas for meaningless stimuli (colour shades) there was little increment; (2) the rate of WM accrual was significantly more efficient for famous faces, compared to unknown faces; (3) the right anterior-ventrolateral temporal lobe (ATL) causally mediated this superior WM storage for famous faces. Specifically, disrupting the ATL (a region tuned to semantic knowledge including person identity) selectively hinders WM accrual for celebrity faces while leaving the accrual for unfamiliar faces intact. Further, this ‘semantically-accelerated’ storage is impervious to disruption of the right middle frontal gyrus and vertex, supporting the specific and causative contribution of the right ATL. Our finding advances the understanding of the neural architecture of WM, demonstrating that it depends on interaction with long-term semantic knowledge underpinned by the ATL, which causally expands the WM buffer when visual content carries semantic information.

Visual working memory buffers information retrieved from visual long-term memory

Human memory is thought to consist of long-term storage and short-term storage mechanisms, the latter known as working memory. Although it has long been assumed that information retrieved from long-term memory is represented in working memory, we lack neural evidence for this and need neural measures that allow us to watch this retrieval into working memory unfold with high temporal resolution. Here, we show that human electrophysiology can be used to track information as it is brought back into working memory during retrieval from long-term memory. Specifically, we found that the retrieval of information from long-term memory was limited to just a few simple objects' worth of information at once, and elicited a pattern of neurophysiological activity similar to that observed when people encode new information into working memory. Our findings suggest that working memory is where information is buffered when being retrieved from long-term memory and reconcile current theories of memory retrieval with classic notions about the memory mechanisms involved.

P187 Individual differences in parietal theta frequency drive spatial working memory capacity

The speed of fronto-parietal theta oscillatory activity is thought to play a key role in determining working memory (WM) capacity. Individual differences in the length of a theta cycle (ranging between 4 Hz and 7 Hz) might act as an envelope for nested gamma cycles (>30 Hz). This would provide the fundamental neurophysiological mechanism for working memory storage accounting for individual differences in the maximum number of items (number of gamma cycles) retained in the working memory buffer (one theta cycle). Accordingly, it would be expected that slower theta frequencies bind a higher number of gamma cycles per memory buffer resulting in increased memory capacity. Conversely, faster theta frequencies would bind a comparatively smaller number of nested gamma cycles, corresponding to decreased memory capacity. We directly tested this hypothesis by means of right parietal transcranial Alternating Current Stimulation (tACS). Specifically, we tested for the impact of down-regulation (4 Hz tACS) and up-regulation (7 Hz tACS) of theta oscillations in a spatial WM paradigm. To this aim eleven participants underwent 4 Hz, 7 Hz and sham stimulation while performing a variation of the Sternberg task. In line with our hypotheses, results showed that down-regulation of right parietal theta by 4 Hz tACS enhanced WM capacity while up-regulation of right parietal theta frequency by 7 Hz tACS reduced WM capacity compared to sham stimulation. These effects were found only for items presented to the left hemifield, i.e. contralateral to the stimulation site, thus providing causal evidence for frequency and spatially tuned WM capacity organization. In conclusion, we provide the first causal evidence that inducing slower parietal theta frequency activity results in higher working memory capacity, while inducing faster parietal theta frequency activity results in reduced working memory capacity, in line with the theta-gamma phase coupling theory of WM capacity.

Editorial: Turning the Mind's Eye Inward: The Interplay Between Selective Attention and Working Memory.

Editorial: Turning the Mind's Eye Inward: The Interplay Between Selective Attention and Working Memory.

The eye-voice span during reading aloud.

Although eye movements during reading are modulated by cognitive processing demands, they also reflect visual sampling of the input, and possibly preparation of output for speech or the inner voice. By simultaneously recording eye movements and the voice during reading aloud, we obtained an output measure that constrains the length of time spent on cognitive processing. Here we investigate the dynamics of the eye-voice span (EVS), the distance between eye and voice. We show that the EVS is regulated immediately during fixation of a word by either increasing fixation duration or programming a regressive eye movement against the reading direction. EVS size at the beginning of a fixation was positively correlated with the likelihood of regressions and refixations. Regression probability was further increased if the EVS was still large at the end of a fixation: if adjustment of fixation duration did not sufficiently reduce the EVS during a fixation, then a regression rather than a refixation followed with high probability. We further show that the EVS can help understand cognitive influences on fixation duration during reading: in mixed model analyses, the EVS was a stronger predictor of fixation durations than either word frequency or word length. The EVS modulated the influence of several other predictors on single fixation durations (SFDs). For example, word-N frequency effects were larger with a large EVS, especially when word N-1 frequency was low. Finally, a comparison of SFDs during oral and silent reading showed that reading is governed by similar principles in both reading modes, although EVS maintenance and articulatory processing also cause some differences. In summary, the EVS is regulated by adjusting fixation duration and/or by programming a regressive eye movement when the EVS gets too large. Overall, the EVS appears to be directly related to updating of the working memory buffer during reading.

Working memory capacity of biological movements predicts empathy traits.

Working memory (WM) and empathy are core issues in cognitive and social science, respectively. However, no study so far has explored the relationship between these two constructs. Considering that empathy takes place based on the others' observed experiences, which requires extracting the observed dynamic scene into WM and forming a coherent representation, we hypothesized that a sub-type of WM capacity, i.e., WM for biological movements (BM), should predict one's empathy level. Therefore, WM capacity was measured for three distinct types of stimuli in a change detection task: BM of human beings (BM; Experiment 1), movements of rectangles (Experiment 2), and static colors (Experiment 3). The first two stimuli were dynamic and shared one WM buffer which differed from the WM buffer for colors; yet only the BM conveyed social information. We found that BM-WM capacity was positively correlated with both cognitive and emotional empathy, with no such correlations for WM capacity of movements of rectangles or of colors. Thus, the current study is the first to provide evidence linking a specific buffer of WM and empathy, and highlights the necessity for considering different WM capacities in future social and clinical research.