Abstract: Working memory is the ability to maintain a limited amount of information after it has been removed from perception. It is a key cognitive ability, thought to play a role in other cognitive functions, including perception, attention and action. Given its importance, its accurate and efficient measurement is a major goal in working memory research. Here we introduce a novel working memory tracking paradigm, inspired by continuous psychophysics and multiple object tracking. Participants viewed a sequence of stimuli moving along variable paths and were asked to reproduce the path by tracing it on a touchscreen. This reproduction was then compared to the original stimulus to determine error and thus memory performance. Across three experiments, we found that this new method is efficient, reliable and powerful, with only ten trials per condition required for stable performance estimates. We have also shown that the method is only minimally affected by perceptual or attentional confounds. Most importantly, since performance was measured across the trial, this method also allows for the investigation of how working memory changes across time. By averaging equivalent time points across trials, we identified influences from both primacy and recency effects, and quantified performance around particularly important points along the motion path. The working memory tracking paradigm is therefore especially useful when experimental time is limited, experimental conditions are extensive or when the time-course is a key interest. The method also opens up the study of working memory with dynamic stimuli.

Abstract: Information retrieved from long-term memory (LTM) enters working memory (WM), and the amount of information that can be retrieved is constrained to the limits of WM (about three to four items; Fukuda & Woodman, Proceedings of the National Academy of Sciences, 114 (20), 5306-5311, 2017). Can LTM retrieval occur when WM is near capacity, without consequence to either the retrieved or the maintained information? Liu, Li, Theeuwes, and Wang (NeuroImage, 261: 119513, 2022) presented evidence that even when WM is near capacity, LTM items could still be reported. They argue that retrieved LTM items can bypass WM. We investigated this further by introducing continuous reporting of retrieved information and WM contents to their paradigm. If retrieval bypasses WM, then there should be no impairment of report accuracy to either WM contents or LTM-retrieved information. In the first experiment, WM reports suffered when an LTM item was retrieved. In the second, we found that when WM was near capacity (four items), the fidelity of LTM reports suffered compared to when WM was not (two items or no items). Additionally, WM contents were reported with lower fidelity when an LTM item was retrieved compared to a WM-only condition, under both two-item and four-item WM load. We conclude that LTM retrieval does not bypass WM.

Abstract: Cultural differences in cognition, including visual perception and long-term memory, may arise because typical visual environments differ across cultures, particularly in their spatial scale. Consequently, the influence of culture on cognitive processing depends on whether stimuli are presented at a large or small spatial scale. We tested North American and East Asian young adults to determine whether such cultural differences extend to short-term memory—testing, for the first time, whether spatial frequency information contributes to cross-cultural differences in memory. Test materials were images of natural and constructed scenes whose spatial structure was manipulated by low-pass filtering. Several seconds after briefly viewing a target scene, a subject saw three versions of that scene: the target itself and two variants whose low-pass filtering differed from the target. From these three, the subject selected the image identical to the target. The two groups did not differ in overall recognition accuracy but did in the way they mistook nonmatching images for certain targets. Specifically, North American subjects made reliably fewer errors in matching images whose high-frequency content was intact, providing evidence that cultural differences in prioritization of high spatial frequency information extend to short-term memory. Across both groups, subjects were highly accurate at recognizing images that retained all or most of their high-spatial frequency content and were highly sensitive to different levels of spatial filtering. These findings show that visual memory has sufficient fidelity to support fine discrimination of variation in spatial frequency.

Abstract: Background and aims Inferential comprehension difficulties and their impacts on reading comprehension are well documented in school-aged children with Developmental Language Disorder (DLD). In comparison, fewer studies have been conducted in young children with DLD prior to their formal schooling journey. In addition, the contribution of linguistic and cognitive skills to oral inferential comprehension abilities in preschoolers, notably receptive vocabulary, phonological short-term memory, and theory of mind (ToM), requires further investigation. The first aim of this study is to explore how young children presenting with DLD aged 5 to 6 years perform when compared to same-age and younger children presenting with typical language development (TLD) on measures of oral inferential comprehension, receptive vocabulary, ToM, and phonological short-term memory. The second aim is to analyze how these linguistic and cognitive skills contribute to oral inferential comprehension in both DLD and TLD preschool children. Methods A total of 112 preschool children participated in this study, including 21 ( n = 21) children with DLD aged 5 to 6 years and two TLD groups, 37 ( n = 37) younger children aged 4 to 5 years and 54 ( n = 54) same-age children. A narrative-based oral inferential comprehension task was administered to all children, in addition to measures of receptive vocabulary, phonological short-term memory, and ToM. Analysis of covariance (ANCOVAs) were used to compare performances between the three groups, followed by Pearson correlations and hierarchical regression analyses to examine the contribution of these variables to oral inferential comprehension abilities across the sample. Results After controlling for level of parental education (LPE) and biological sex, children with DLD performed significantly below the same-age TLD group on all four measures with large effect sizes ( p < .001; η 2 = .17–.44). Their performance was comparable to the younger TLD group on measure of oral inferential comprehension ( p = .234), and significantly below on measures of receptive vocabulary ( p = .008), phonological short-term memory ( p < .001), and ToM ( p = .028). Results from the regression analysis indicated that age, LPE, diagnosis condition, receptive vocabulary and ToM accounted for 53% of the total variance in oral inferential comprehension. Conclusions and implications This study reiterates the early listening comprehension difficulties experienced by preschool children with DLD when compared to children presenting with typical language development. The results also indicate that when controlling for age, LPE and diagnosis condition, children are likely to have better inferential comprehension abilities if they perform well on a measure of ToM. Considering that challenges related to language comprehension are acknowledged to be persistent and less responsive to intervention, these findings can help inform the development of evidence-based interventions aiming at supporting language comprehension of young children with DLD.

Abstract: According to recent studies, feed-forward Deep neural networks (DNNs) perform better than text-to-speech (TTS) systems that use decision-tree clustered context-dependent hidden Markov models (HMMs) [1, 4]. The feed-forward aspect of DNN-based models makes it difficult to incorporate the long-span contextual influence into spoken utterances. Another typical strategy in HMM-based TTS for establishing a continuous speech trajectory is using the dynamic characteristics to constrain the production of speech parameters [2]. Parametric time-to-speech synthesis is used in this study by capturing the co-occurrence or correlation data between any two points in a spoken phrase using time aware memory network cells. Based on our experiments, a combination of DNN and BLSTM-RNN is the best system to use. Upper hidden layers of this system use a bidirectional RNN structure of LSTM, the low layers use a simple, one way structure followed by additional layers. On objective and subjective metrics, it surpasses both the traditional decision-based tree HMM’s and the DNN-TTS system. Dynamic limitations are superfluous since the BLSTM-RNN TTS produces very smooth speech trajectories.

Abstract: Two experiments explored a finding that multiple, repeated presentations of the same, six-item colour-shape-location array results in no, or very slow improvements in change detection performance. This contrasts with studies showing clear learning from small numbers of repetitions of verbal and visual stimuli with memory tested by recall or reconstruction. Experiment 1 presented the same six-item visual array repeated across 120 trials for change detection between a study presentation and a test presentation that was identical or depicted a feature swap on 50% of trials. Longer (5000 ms) versus shorter (2000 ms and 500 ms) intervals between the study array and the test array resulted in more participants improving their performance, but only after at least 30 repetitions of the array. A study-test interval of 500 ms resulted in no improvement in performance across 180 repetitions of the same array. Experiment 2 showed that, regardless of the stimulus duration, presenting different arrays on each trial resulted in a lack of learning that was similar to that in Experiment 1 for a 500 ms study-test interval. Results appear consistent with reliance on a limited capacity temporary visual memory for change detection that retains the array only for the current trial and does not support learning across repetitions. With a sufficiently long study-test interval, it is proposed that on each trial there is, in addition, a weak long-term episodic trace of the array that gradually strengthens across repetitions until it is sufficiently strong to be useful for supplementing performance.

Abstract: With the introduction of goals for carbon neutrality and peak carbon emissions, Virtual Power Plants(VPPS) are evolving towards intelligence, automation, scaling, and diversification of resources. Load forecasting technology, when applied in VPPS, helps to improve the operational efficiency of the power system, optimize resource allocation, enhance the stability and reliability of the power grid, and also provides scientific data support and decision making basis for the operation of VPPS. This article focuses on the multi-objective optimization scheduling of VPPS, with economic and low-carbon goals at its core. An innovative short-term load forecasting method is proposed to address the key issue of poor accuracy in electric load forecasting. A load forecasting model based on the Sparrow Search Algorithm (SSA) optimized Long Short-Term Memory (LSTM) network is introduced. This model effectively enhances the accuracy and training efficiency of load forecasting by optimizing the key parameters of the LSTM network. Comparative experimental results show that the improved model significantly outperforms traditional methods in terms of error metrics, and is able to more accurately capture the trends of complex load changes. This provides high-quality input data support for the optimization of virtual power plant scheduling.

Abstract: ABSTRACT Memory is a crucial cognitive process involving several subsystems: sensory memory (SM), short‐term memory (STM), working memory (WM), and long‐term memory (LTM). While each has been extensively studied, the interaction between subsystems, particularly in relation to predicting temporal sequences, remains largely unexplored. This study investigates the association between WM and LTM, and how these relate to aging and musical training. Using three datasets with a total of 243 healthy volunteers across various age groups, we examined the impact of WM, age, and musical training on LTM recognition of novel and previously memorized musical sequences. Our results show that WM abilities are positively associated with the identification of novel sequences, but not with the recognition of memorized sequences. Additionally, musical training has a similar positive impact on the identification of novel sequences, while increasing age is associated with reduced memory performance. Different cognitive processes are involved in handling prediction errors compared to confirmatory predictions, and WM contributes to these processes differently. Future research should extend our investigation to populations with memory impairments and explore the underlying neural substrates.

Abstract: This study examined the role of domain-specific working memory and emotion regulation in the relation between mathematics anxiety and mathematics performance among 264 upper elementary students (Grades 3-5). Participants completed measures of mathematics testing and learning anxiety, verbal and numerical working memory, cognitive reappraisal, expressive suppression, general anxiety, mathematics self-efficacy, and calculation. Results showed that verbal working memory, but not numerical working memory, mediated the relation between mathematics testing anxiety and calculation. Higher verbal working memory exacerbated the negative effects of both mathematics testing and learning anxiety on calculation. Higher cognitive reappraisal exacerbated the negative effects of mathematics testing anxiety on calculation. These findings suggest that mathematics anxiety hinders calculation not by disrupting numerical processing but through verbal rumination and verbal information processing, especially in children with strong verbal working memory. For children who are still developing emotion regulation and foundational mathematics, cognitive reappraisal, a typically adaptive emotion regulation strategy, may paradoxically increase cognitive load, intensifying the adverse effects of mathematics anxiety during testing. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Abstract: Abstract The word-length effect refers to the finding that memory on many short-term/working memory tasks is better for words with fewer syllables than words with more syllables. The standard account attributes this result to a combination of decay offset by rehearsal: More short words can be rehearsed because they take less time to articulate. However, most studies have confounded length with lexical and other long-term memory factors that covary with length. In this paper, we reexamine word-length effects in serial recognition. Experiment 1 replicated previous findings of a word-length effect when short and long words also differed on numerous other dimensions. Experiment 2 found that when the short and long words were more fully equated, including being equated for orthographic and phonological neighborhood size, the word-length effect disappeared. Experiment 3 confirmed that memory was better for words with more orthographic and phonological neighbors than words with fewer neighbors, showing serial recognition is sensitive to at least some lexical/long-term memory factors. The results provide more evidence against the standard account of the word-length effect and instead are consistent with a growing body of work which shows that lexical and other long-term memory factors affect performance in short-term/working memory tasks.

Abstract: The aim of the present paper is to develop further a theory on the flow of linguistic variables making a sentence, namely, the transformation: (a) characters into words; (b) words into word intervals; (c) word intervals into sentences. The relationship between two linguistic variables is studied as a communication channel whose performance is determined by the slope of their regression line and by their correlation coefficient. The theory is applicable to any field/specialty in which a linear relationship holds between two variables. The signal–to–noise ratio Γ is a figure of merit of a channel being deterministic, i.e. a channel in which the scattering of the data around the regression line is negligible. The larger Γ is, the more the channel is deterministic. In conclusion, humans have invented codes whose sequences of symbols making words cannot vary very much for indicating single physical or mental objects of their experience (larger Γ). On the contrary, a large variability (smaller Γ) is achieved by introducing interpunctions to make word intervals, and word intervals to make sentences to communicate concepts

Abstract: Music emotion regression (MER) is a vital field that bridges psychology and music information retrieval. Music has the powerful ability to evoke a wide range of human emotions, from joy and sadness to anger and calmness. Understanding how music influences emotional states is essential for grasping its psychological effects on individuals. This research presents an innovative model that combines convolutional neural networks (CNNs) with bidirectional long short-term memory (BiLSTM) networks to analyze and predict the emotional impact of musical audio. The model uses CNNs to detect temporal patterns and BiLSTMs to interpret sequences in both forward and backward directions, enhancing its ability to capture the complex structure of musical data. Additionally, a multi-head attention mechanism is incorporated to improve the model’s expressiveness and generalizability, making it especially effective for handling intricate sequential tasks and large datasets. The model’s performance was evaluated through sentiment prediction using extensive, publicly available datasets comprising over 9,000 musical excerpts. Results show that the proposed model significantly outperforms existing methods in MER, achieving an R-squared value of 0.845, indicating an excellent fit with the empirical data.

Abstract: Background. Fluid intelligence is an integral cognitive ability that involves solving new non-standard problems. It strongly predicts academic and professional achievement, whereas a low level of fluid intelligence is an important predictor of learning problems. Clinical studies of fluid intelligence are of interest for the development of training programs in various groups of children with special needs. This article presents a study on fluid intelligence in children with learning disabilities. Objective. This study aimed to investigate characteristics of fluid intelligence and its relationships with other cognitive characteristics in children with learning disabilities. Design. This study involved 93 children, divided into two groups: 55 typically developing children (control group) and 38 children with learning disabilities (clinical group). To assess intelligence characteristics, this study employed the Kaufman Assessment Battery for Children (KABC-II) and the Wechsler Intelligence Scale for Children Fifth Edition (WISC-V). Results. A reduction was found in fluid intelligence, working memory, short-term memory, long-term memory, processing speed, visual-spatial abilities, and verbal abilities in the group of children with learning disabilities compared to the control group. In the clinical group, fluid intelligence was strongly associated with a greater number of cognitive parameters compared to the control group. Conclusions. It is possible to assume that a close connection of fluid intelligence with the assessed cognitive characteristics in the group of children with learning disabilities may be due to general challenges in cognitive development.

Abstract: Abstract Prior research has explored how working memory influences the formation of new long-term memories, but its role in modifying existing representations remains unclear. This study examines whether attentional prioritization and testing in working memory enhance long-term memory retrieval and investigates the underlying neural mechanisms. Eighty-six participants completed a three-phase memory task combining a long-term memory—with a working memory retro-cue paradigm. First, participants learned object-location associations. Next, during a working memory task, some objects have undergone attentional prioritization and testing, others have only been tested in working memory. Finally, participants retrieved the object locations from long-term memory. Three key findings emerged: (1) both attentional prioritization and testing in working memory improved long-term memory retrieval; (2) serving as a probe in working memory further contributed to long-term memory enhancement, with benefits observed at behavioral and neural levels; and (3) cross-phase decoding revealed a comparable representational format for location information across task phases, possibly explained by the neural reinstatement of location information across phases. These results suggest that working memory dynamically shapes long-term memory representations, playing a more active and integrated role in long-term memory formation than previously thought.

Abstract: Multimedia user interfaces incorporate various feedback methods using different modalities. Cognitive processing of audiovisual information requires the ability to recall visual and auditory information, either separately, or in combination. Short-term memory capabilities vary individually and depend on factors such as signal presentation and the number and type of visual and auditory items. In an experiment involving 40 subjects, we aimed to compare short-term auditory and visual capabilities in a serious game application. Subjects played the ‘Pairs’ game at different resolutions, using either visual icons or audio samples, while the total time cost and number of flips were recorded. The results indicate that visual memory is not superior, and female subjects performed better than males at higher levels in the visual task. Additionally, human sound samples, speech and familiar auditory icons were found to be easier to recall than artificial measurement signals.

Abstract: Every day, we make choices about how much effort we are willing and able to use to achieve the outcomes we desire against the backdrop of constantly shifting effort demands and available rewards. While factors like visual short-term memory and chronic stress levels can predict responses to stable cognitive effort demands, we do not yet know whether they constrain one's choices of higher effort trials for larger rewards when task demands and potential outcomes shift over time. Here, we examined whether these factors predicted the choice to deploy cognitive effort given increasing effort demands and the tendency to deploy effort given shifting reward availability. Undergraduate participants first performed an online visual short-term memory task to assess capacity for visuospatial short-term memory. They then completed a series of choice trials where they could choose between high-effort, high-reward or low-effort, low-reward trials. In two blocks, we varied either the effort required on high-effort trials or the reward offered on both trial types. We found that visual short-term memory predicted the likelihood of choosing high-effort trials given shifting rewards, while chronic stress and everyday preferences for cognitively effortful strategies predicted the tendency to deploy increasing amounts of effort for a stable reward. Furthermore, participants' subjective reports show a strong focus on attentional processes, and balancing rewards and losses, when making decisions about how much effort to deploy. These findings shed light on distinct trait-level factors associated with cognitive effort choices given shifting demands and outcomes.