Visual Working Memory Research Articles

Comprehensive exploration of visual working memory mechanisms using large-scale behavioral experiment

Abstract Two decades of research on visual working memory have produced substantial yet fragmented knowledge. This study aims to integrate these findings into a cohesive framework. Drawing on a large-scale behavioral experiment involving 40 million responses to 10,000 color patterns, a quasi-comprehensive exploration model of visual working memory, termed QCE-VWM, is developed. Despite its significantly reduced complexity (57 parameters versus 30,796), QCE-VWM outperforms neural networks in data fitting. The model provides an integrative framework for understanding human visual working memory, incorporating a dozen mechanisms—some directly adopted from previous studies, some modified, and others newly identified. This work underscores the value of large-scale behavioral experiments in advancing comprehensive models of cognitive mechanisms.

Temporal attention modulates distraction resistance of visual working memory representations.

The regulation of Visual Working Memory (VWM) distraction resistance by internal attention remains debated with four hypotheses: the null hypothesis (attentional priorities don't affect distraction resistance), protection hypothesis (higher priority, greater distraction resistance), vulnerability hypothesis (higher priority, lower distraction resistance), and available resource threshold hypothesis (distraction resistance depends on attentional resource allocation exceed required thresholds). A recent study found that temporal attention can influence VWM priorities, yet this hasn't been explored from the perspective of temporal attention. This study used a continuous reporting task to examine these issues. Experiment 1 established stable attentional priority using an auditory cue, while Experiment 2 removed this cue to introduce dynamic priority changes. To explore neural mechanisms, Experiment 3 employed EEG to measure contingent negative variation (CNV) and decode priority representations during the delay period. Behavioral results confirmed that visual distractors increased memory deviation during maintenance, but deviations were smaller for anticipated high-priority items, suggesting better memory accuracy. High-priority items showed greater resistance to distraction at long intervals. Without the auditory cue, high-priority items resisted distraction better at short intervals. EEG results revealed enhanced CNV before long interval targets and better decoding of priority with distractors during long intervals. In summary, distraction affects different priority items equally when resources exceed requirements. High-priority items resist interference when adequately resourced but suffer if resources are insufficient, supporting the available resource threshold hypothesis. This study highlights the temporal dynamics of distraction resistance in VWM representations.

A Pilot Study on Video Game Training Effects on Visual Working Memory: Behavioral and Neural Insights

Background/Objectives: Recent research suggests that video games may serve as cognitive training tools to enhance visual working memory (VWM) capacity. However, the effectiveness of game-based cognitive training remains debated, and the underlying neural mechanisms, as well as the relationship between training efficacy and game design factors, are unclear. This study aimed to evaluate the impact of video game training on VWM capacity and explore its neural correlates. Methods: Two groups underwent 56 daily 20 min training sessions with two distinct video games targeting different cognitive skills: a reaction-time training game and a VWM-specific training game. Behavioral assessments included accuracy, hit response times, correct rejection response times, and Cowan’s K values. Neural correlates were measured through Negative Slow Wave (NSW) activity using EEG. Decision tree classification analyses were applied to NSW data across sessions and set sizes to identify patterns linked to VWM capacity. Results: Preliminary results are that both groups showed improvements in behavioral measures (accuracy, response times, and Cowan’s K values). NSW analyses revealed a main effect of set size in both groups, and classification results indicated that NSW patterns differed between groups, across sessions, and set sizes, supporting the relationship between NSW and VWM capacity. Conclusions: These findings contribute to understanding NSW as a neurophysiological correlate of VWM capacity, demonstrating its plasticity through video game training. Simple video games could effectively enhance behavioral and neural aspects of VWM, encouraging their potential as accessible cognitive training tools.

Dissociating the roles of alpha oscillation sub-bands in visual working memory.

Alpha oscillations play a critical role in visual working memory (VWM), but the specific contributions of lower and upper alpha sub-bands remain unclear. To address this, we employed a whole-field change detection paradigm to investigate how alpha power modulation and decoding accuracy differ between these sub-bands in response to varying set sizes and spatial extents of memory arrays. Our results revealed that lower alpha (8-9 Hz) exhibits widespread event-related desynchronization (ERD) during the early maintenance phase, which increases with set size and reflects attentional allocation to individual memory items. In contrast, upper alpha (10-12 Hz) demonstrates posteriorly localized ERD that is strongly associated with the spatial extent of memory arrays. During the late maintenance phase, upper alpha transitions to event-related synchronization (ERS), suggesting a role in suppression of irrelevant sensory inputs and enhancement of alertness. Multivariate decoding analyses showed that all alpha sub-bands accurately decoded both set size and spatial extent across time windows, with lower alpha achieving better decoding performance during the early maintenance phase and upper alpha excelling in later stages. These findings suggest that alpha oscillations encode both the number and spatial distribution of memory items, with lower and upper alpha sub-bands serving complementary roles in encoding and maintaining VWM representations.

Network Segregation and Integration Changes in Healthy Aging: Evidence From EEG Subbands During the Visual Short-Term Memory Binding Task.

Working memory, which tends to be the most vulnerable cognitive domain to aging, is thought to depend on a functional brain network for efficient communication. The dynamic communication within this network is represented by segregation and integration. This study aimed to investigate healthy aging by examining age effect on outcomes of graph theory analysis during the visual short-term memory binding (VSTMB) task. VSTMB tasks rely on the integration of visual features and are less sensitive to semantic and verbal strategies. Effects of age on neuropsychological test scores, along with the EEG graph-theoretical integration, segregation and global organization metrics in frequencies from delta to gamma band were investigated. Neuropsychological assessment showed low sensitivity as a measure of age-related changes. EEG results indicated that network architecture changed more effectively during middle age, while this effectiveness appears to vanish or show compensatory mechanisms in the elderly. These differences were further found to be related to cognitive domain scores. This study is the first to demonstrate differences in working memory network architecture across a broad age range.

Exploring the role of sex, sex steroids, menstrual cycle, and hormonal contraception use in visual working memory: Insights from behavioral and EEG analyses.

Sex hormones have been shown to influence cognitive and emotional processes, yet their effects on visual working memory (VWM) are poorly understood. This study aimed to investigate the relationship between VWM, sex, and female hormonal status in participants aged 18-35 years. We recruited 32 males (M) and 133 females, categorized into four groups: naturally cycling females in the early follicular (NCF, n = 33) and mid-luteal (NCL, n = 35) phases of the menstrual cycle, oral contraceptive (OC, n = 37), and intrauterine device users (IUD, n = 28). Participants completed a bilateral change detection task while behavioral and EEG data were recorded. We evaluated VWM performance and associated brain electrophysiological responses, specifically Contralateral Delay Activity (CDA). Salivary levels of testosterone, progesterone, and estradiol were assessed. We found no systematic differences in VWM task performance or CDA between groups, nor any correlations with hormone levels. However, an exception to this was that NCF females performed worse than OC users when recalling four items. Age emerged as a significant covariate, with greater age being linked to poorer performance. An interaction between age and group in memory capacity highlighted differential patterns of age-related cognitive decline across sexes and female hormonal status groups. These findings provide valuable insights into the broader relationship between sex, sex hormones, and cognition. They suggest that in studies employing a between-subject design, hormone-dependent differences in more complex processes, such as visuospatial performance, are unlikely to stem from the role of sex hormones in VWM and may instead arise from other factors.

Working Memory as the Focus of the Bilingual Effect in Executive Functions

The bilingual effect on executive functions (EFs) has garnered considerable attention, with most studies focusing on the visual domain and largely overlooking the auditory domain. Furthermore, research has predominantly concentrated on specific subcomponents of executive functions, with few studies systematically examining all three key subcomponents. This raises two important questions: (a) Is the bilingual effect specific to certain modalities (modality-specific), or a more general phenomenon (modality-general)? (b) Is the bilingual effect concentrated in a specific component of executive functions (process-specific), or does it extend to all three components (process-general)? To explore these questions, this study recruited monolingual Chinese and bilingual Chinese–English participants, using matched visual and auditory Stroop, N-back, and task-switching tasks to assess inhibitory control, working memory, and cognitive flexibility in both groups. The results showed that, after controlling for variables like intelligence, socioeconomic status, and age, bilingualism significantly predicted performance in both auditory and visual working memory tasks, explaining 34% and 19% of the variance, respectively. However, no evidence was found to support a bilingual effect in inhibitory control or cognitive flexibility. In conclusion, these results suggest that bilingual effects are not only process-specific (affecting only working memory) but also modality-general (providing advantages in both visual and auditory modalities).

Representational geometry explains puzzling error distributions in behavioral tasks

Measuring and interpreting errors in behavioral tasks is critical for understanding cognition. Conventional wisdom assumes that encoding/decoding errors for continuous variables in behavioral tasks should naturally have Gaussian distributions, so that deviations from normality in the empirical data indicate the presence of more complex sources of noise. This line of reasoning has been central for prior research on working memory. Here, we reassess this assumption and find that even in ideal observer models with Gaussian encoding noise, the error distribution is generally non-Gaussian, contrary to the commonly held belief. Critically, we find that the shape of the error distribution is determined by the geometrical structure of the encoding manifold via a simple rule. In the case of a high-dimensional geometry, the error distributions naturally exhibit flat tails. Using this insight, we apply our theory to visual short-term memory tasks, and find that it can account for a large array of experimental data with only two free parameters. Our results challenge the dominant view in the mechanisms and capacity constraints of working memory systems. They instead suggest that the Bayesian framework, which explains various aspects of perceptual behavior, also provides an excellent account of working memory. Overall, our results establish a direct connection between neural manifold geometry and behavior, and call attention to the geometry of the representation as a critically important, yet underappreciated factor in determining the character of errors in human behavior.

Development of compositionality through interactive learning of language and action of robots.

Humans excel at applying learned behavior to unlearned situations. A crucial component of this generalization behavior is our ability to compose/decompose a whole into reusable parts, an attribute known as compositionality. One of the fundamental questions in robotics concerns this characteristic: How can linguistic compositionality be developed concomitantly with sensorimotor skills through associative learning, particularly when individuals only learn partial linguistic compositions and their corresponding sensorimotor patterns? To address this question, we propose a brain-inspired neural network model that integrates vision, proprioception, and language into a framework of predictive coding and active inference on the basis of the free-energy principle. The effectiveness and capabilities of this model were assessed through various simulation experiments conducted with a robot arm. Our results show that generalization in learning to unlearned verb-noun compositions is significantly enhanced when training variations of task composition are increased. We attribute this to self-organized compositional structures in linguistic latent state space being influenced substantially by sensorimotor learning. Ablation studies show that visual attention and working memory are essential to accurately generate visuomotor sequences to achieve linguistically represented goals. These insights advance our understanding of mechanisms underlying development of compositionality through interactions of linguistic and sensorimotor experience.

Neural deterioration and compensation in visual short-term memory among individuals with amnestic mild cognitive impairment.

Visual short-term memory (VSTM) is a critical indicator of Alzheimer's disease (AD), but whether its neural substrates could adapt to early disease progression and contribute to cognitive resilience in amnestic mild cognitive impairment (aMCI) has been unclear. Fifty-five aMCI patients and 68 normal controls (NC) performed a change-detection task and underwent multimodal neuroimaging scanning. Among the atrophic brain regions in aMCI, VSTM performance correlated with the volume of the right prefrontal cortex (PFC) but not the medial temporal lobe (MTL), and this correlation was mainly present in patients with greater MTL atrophy. Furthermore, VSTM was primarily correlated with frontal structural connectivity in aMCI but was correlated with more distributed frontal and MTL connectivity in NC. This study provided evidence on neural adaptation in the precursor stages of AD, highlighting the compensatory role of PFC as MTL deteriorated and suggesting potential targets in early intervention for cognitive preservation. Atrophic left medial temporal lobe (MTL) no longer correlated with visual short-term memory (VSTM) in amnestic mild cognitive impairment (aMCI). Atrophic right middle frontal area continued to correlate with VSTM in aMCI. Frontal brain-behavior correlation was mainly present in the aMCI subgroup with greater medial temporal lobe (MTL) atrophy. Reliance of VSTM on frontal connectivity increased in compensation for MTL dysfunction.

Running after two hares in visual working memory: Exploring retrospective attention to multiple items using simulation, behavioral outcomes, and eye tracking.

Multi-item retro-cueing effects refer to better working memory performance for multiple items when they are cued after their offset compared to a neutral condition in which all items are cued. However, several studies have reported boundary conditions, and findings have also sometimes failed to replicate. We hypothesized that a strategy to focus on only one of the cued items could possibly yield these inconsistent patterns. In Study 1, a Monte Carlo simulation showed that randomly selecting one of the cued items as the focus in each trial increased the chance of obtaining significant "multi-item retro-cueing effects" on the mean accuracy over the trials, providing an incorrect conclusion if interpreted as evidence for attending all the cued items. These high rates to obtain such data fit with inconsistent patterns in the literature. To try and circumvent this situation, we conducted two new experiments (Studies 2A and 2B) where participants were explicitly instructed to fixate their gaze on all the cued positions, verified through eye tracking (Study 2B). These produced robust multi-item retro-cueing effects regardless of previously identified boundary conditions. Notably, gazes were clearly fixated to multiple cued positions within each trial. Nevertheless, simulation revealed that our accuracy patterns could also in principle be produced by single-item enhancement on each trial. The present study forms the first step to disentangle overt gaze-based allocation of attention from single-item focusing strategies while also highlighting the need for improved methodologies to probe genuine multiplicity in working memory. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Screen time exposure and executive functions in preschool children

Increased screen time (ST) among preschool children is becoming a matter of concern globally. Although gadgets such as phones, tablets and computers might be of educational use in this population, excessive ST might impair cognitive function among preschoolers. As data on this topic in preschool children are scarce, this study sought to investigate the relationship between ST and executive functions (EFs) in this population. A total of 1016 preschool children aged 5–6 years (M = 70.8 months, SD = 4.5) were tested using the Developmental Neuropsychological Assessment (NEPSY-II) and the Dimensional Change Card Sort tests for cognitive flexibility (CF), visual working memory (WM), verbal WM, inhibition and motor persistence with parental consent, while ST was reported via a questionnaire completed by their mothers. Participants spent approximately 2 h in both active and passive ST per day, with boys spending much more time in active ST than girls. There was a very weak negative correlation between CF and both active and passive ST, and a weak negative correlation between verbal WM and both active and passive ST. Additionally, there was a very weak negative correlation between inhibition and passive ST on weekday. To ensure proper development of EFs in preschool children, ST should be limited to ≤ 1 h per day of high-quality educational/interactive content as previously determined by eminent pediatric institutions worldwide. Proven remedies that enhance EFs in children, such as physical activity and cognitive training, should be practiced regularly.

Predictive value of an unsupervised web-based assessment of the neuropsychological function

Neuropsychological assessment has to consider the subjective and objective functional deficits of help-seeking individuals in several cognitive domains. Due to time constraints in clinical practice, several web-based approaches have been developed. The current study examined whether functional deficits in the mnestic and attentive domain can be predicted based on an unsupervised self-administered online assessment neuropsychological online screening (NOS): This screening includes self-reports and psychometric memory tests (face-name association, visual short-term memory). Data of help-seeking individuals (n = 213, mean age: 48.2 years) running an extensive in-person assessment were analyzed. A functional deficit in at least one cognitive domain was detected in 48 individuals. This classification was supported by the preceding online screening (sensitivity = 0.75, specificity = 0.80), and a linear regression model identified two significant predictors (free recall performance, form discrimination in visual short-term memory). The predictive quality was enhanced for individuals with selective deficits in the mnestic domain (n = 23: sensitivity = 0.78 and specificity = 0.78) as compared to the attentive domain (n = 25: sensitivity = 0.68 and specificity = 0.75). Our results show that a neuropsychological in-person assessment cannot be replaced by an unsupervised self-administered online test. However, a specific pattern of results in the online test might indicate the necessity of an extensive assessment in help-seeking individuals.

Association between auditory mismatch negativity and visual working memory in school-age children with attention deficit/hyperactivity disorder.

Attention-deficit/hyperactivity disorder (ADHD) patients exhibit characteristics of impaired working memory (WM) and diminished sensory processing function. This study aimed to identify the neurophysiologic basis underlying the association between visual WM and auditory processing function in children with ADHD. The participants included 86 children with ADHD (aged 6-15 years, mean age 9.66 years, 70 boys, and 16 girls) and 90 typically developing (TD) children (aged 7-16 years, mean age 10.30 years, 66 boys, and 24 girls). Electroencephalograms were recorded from all participants while they performed an auditory discrimination task (oddball task). The visual WM capacity and ADHD symptom severity were measured for all participants. Compared with TD children, children with ADHD presented a poorer visual WM capacity and a smaller mismatch negativity (MMN) amplitude. Notably, the smaller MMN amplitude in children with ADHD predicted a less impaired WM capacity and milder inattention symptom severity. In contrast, the larger MMN amplitude in TD children predicted a better visual WM capacity. Our results suggest an intimate relationship and potential shared mechanism between visual WM and auditory processing function. We liken this shared mechanism to a total cognitive resource limit that varies between groups of children, which could drive correlated individual differences in auditory processing function and visual WM. Our findings provide a neurophysiological correlate for reports of WM deficits in ADHD patients and indicate potential effective markers for clinical intervention.

Verbal Encoding Strategies in Visuo-Spatial Working Memory.

Visual working memory and verbal storage are often investigated independently of one another. However, a growing body of evidence suggests that naming visual stimuli can provide an advantage in performance during visual working memory tasks. On the other hand, there is also evidence that labeling could lead to biases in recall. Here, we present an exploratory investigation of verbal labels associated with the memorization of simple visuo-spatial stimuli, and how the use of these labels informs recall behavior of the same stimuli in a separate working memory task. English-speaking participants performed a working memory task with orientation and location stimuli, followed by a separate naming task featuring the same stimuli. We found a diverse set of labels employed frequently and with a consistent distribution across stimulus types, the stimulus space, and among participants. The use of individual spatial words, predicted class 1 cardinal biases in memory (i.e. the observation that cardinal stimuli are more accurately recalled than non-cardinal ones). Conversely, words expressing uncertainty (e.g. 'slightly', 'near') predicted class 2 cardinal bias (i.e. recall biases away from the cardinal planes). This relationship between word use and recall biases is consistent with shared representational resources that are used for both visuo-spatial and verbal working memory.

Action consequences guide the use of visual working memory.

Visual working memory (VWM) is a store for temporary maintenance of visual information. It is often disregarded, though, that information is typically stored to enable actions. Therefore, the context of these actions is of great importance for how VWM is used. Here, we questioned whether the severity of the consequence of an action might affect how well information is memorized, and how cautiously it is utilized. We employed an (online) copying task, in which participants reproduced an example display comprised of six items in a grid, using a pool of items. Crucially, we manipulated the severity of penalties: participants had to wait 5 (high) or 0.5 (low error cost) s after an error. Additionally, we manipulated the accessibility of task-relevant information (a well-studied manipulation in this paradigm): participants had to wait 5 (high) or 0.5 (low sampling cost) s to inspect the example. Our results show that with higher error cost the number of inspections remained comparable, but the number of errors decreased. Furthermore, they show that with higher sampling cost the number of inspections halved, and the number of errors increased. Thus, more severe action consequences increase the reluctance to act on uncertain information in VWM, but do not lead to more attempts to store information in VWM. We conclude that, in contrast to the effect of the accessibility of information, action consequences do not affect how well information is memorized, but affect how cautiously this stored information is utilized. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

The effect of occlusion on the visual working memory pointer-system.

To access its online representations, visual working memory (VWM) relies on a pointer-system that creates correspondence between objects in the environment with their memory representations. This pointer-system allows VWM to modify its representations using a process called updating. When the pointer is invalidated, however, VWM triggers a process called resetting in which the no longer relevant representation and pointer are replaced. Past studies used the contralateral delay activity (CDA) to differentiate between updating and resetting and found that resetting is followed by a drop in the CDA amplitude. The current study aimed to investigate the effects of occlusion on VWM representations and the resetting process across four experiments. Experiment 1 examined whether resetting occurs with occluded changes and compared the CDA of occluded versus visible objects. The results indicated a decline in CDA amplitude during occlusion, but it was unclear if resetting occurred when the change was occluded due to the lack of time-locked changes. To better isolate the resetting process, Experiment 2 used a brief occluder appearances (100ms) and observed a CDA drop likely due to an ERP response to the sudden stimulus appearance. This drop occurred earlier than the resetting CDA drop and appeared even in conditions that did not trigger resetting, which indicates that it might be an ERP response to the short and sudden appearance of a stimulus. Experiment 3 further isolated this ERP response, confirming the early CDA drop as a reaction to the occluder's onset and offset. Experiment 4, which included occluders that did not flash to avoid ERP responses, found a CDA drop indicating that resetting can occur with inferred changes. These findings suggest that VWM maintains representations of occluded objects, and can update or reset these representations based on inferred changes, with brief stimuli eliciting ERP responses that affect CDA amplitude.

Neural correlates of olfactory working memory in the human brain.

Previous research has revealed that the insula, pallidum, thalamus, hippocampus, middle frontal gyrus, and supplementary motor area are activated during odor memory and that the performance of olfactory working memory is affected by the verbalization of odors. However, the neural mechanisms underlying olfactory working memory and the role of verbalization in olfactory working memory are not fully understood. Twenty-nine participants were enrolled in a study to complete olfactory and visual n-back tasks using high- and low-verbalizability stimuli while undergoing fMRI imaging. The behavioral results showed that the participants achieved greater accuracy in the visual rather than olfactory n-back task. We observed increased activation in the precentral gyrus, superior frontal gyrus, middle frontal gyrus, supplementary motor area, and inferior parietal gyrus during olfactory working memory. Interestingly, decreased activation was observed in the olfactory 2-back task versus the 0-back task. Moreover, the left angular gyrus and inferior parietal gyrus were more strongly activated during processing of olfactory working memory using high-verbalizability odors. In conclusion, olfactory working memory engages cross-modal regions to facilitate responses, is involved in the monitoring and manipulation of information during working memory, and boasts a unique activation pattern that is different from that of visual working memory. Semantic information supports the representation of odor information in the working memory system.

Electrophysiological correlates of visual short-term memory Binding deficits in community-dwelling seniors at risk of dementia

Electrophysiological correlates of visual short-term memory Binding deficits in community-dwelling seniors at risk of dementia

Orienting attention to auditory and visual working memory in older adults with cochlear implants.

Cochlear implantation is a well-established method for restoring hearing sensation in individuals with severe to profound hearing loss. It significantly improves verbal communication for many users, despite substantial variability in patients' reports and performance on speech perception tests and quality-of-life outcome measures. Such variability in outcome measures remains several years after implantation and could reflect difficulties in attentional regulation. The current study assessed the ability to use a cue to guide attention internally toward visual or auditory working memory (i.e., reflective attention) in cochlear implant (CI) users. Participants completed a cognitive task called the delayed match-to-sample task in which a visual gradient was presented on a computer screen and a piano tone was presented through speakers simultaneously. A visual cue (i.e., letter A or V) instructed participants to focus attention on the item held in auditory or visual working memory. After a delay following the cue presentation, participants were presented with a probe item and indicated by pressing a button whether it matched the cued item in working memory. CI users and age-matched normal hearing adults showed comparable benefit from having an informative cue relative to an uninformative cue (i.e., letter X). Although CI users have had a history of severe deafness and experience coarse sound information, they were able to retrospectively orient their attention to an item in auditory or visual working memory. These findings suggest that CI users with at least one year of CI experience can successfully regulate attention to a level that is comparable to that of normal hearing individuals.