Short-term Memory Representations Research Articles

Verbal Working Memory Error Patterns and Speech-Language Outcomes in Youth With Cochlear Implants.

Verbal working memory (VWM) delays are commonly found in prelingually deaf youth with cochlear implants (CIs), albeit with considerable interindividual variability. However, little is known about the neurocognitive information-processing mechanisms underlying these delays and how these mechanisms relate to spoken language outcomes. The goal of this study was to use error analysis of the letter-number sequencing (LNS) task to test the hypothesis that VWM delays in CI users are due, in part, to fragile, underspecified phonological representations in short-term memory. Fifty-one CI users aged 7-22 years and 53 normal hearing (NH) peers completed a battery of speech, language, and neurocognitive tests. LNS raw scores and error profiles were compared between samples, and a hierarchical regression model was used to test for associations with measures of speech, language, and hearing. Youth with CIs scored lower on the LNS test than NH peers and committed a significantly higher number of errors involving phonological confusions (recalling an incorrect letter/digit in place of a phonologically similar one). More phonological errors were associated with poorer performance on measures of nonword repetition and following spoken directions but not with hearing quality. Study findings support the hypothesis that poorer VWM in deaf children with CIs is due, in part, to fragile, underspecified phonological representations in short-term/working memory, which underlie spoken language delays. Programs aimed at strengthening phonological representations may improve VWM and spoken language outcomes in CI users.

Exploring the Impact of COVID-19 on Social Life by Deep Learning

Due to the globalisation of the COVID-19 pandemic, and the expansion of social media as the main source of information for many people, there have been a great variety of different reactions surrounding the topic. The World Health Organization (WHO) announced in December 2020 that they were currently fighting an “infodemic” in the same way as they were fighting the pandemic. An “infodemic” relates to the spread of information that is not controlled or filtered, and can have a negative impact on society. If not managed properly, an aggressive or negative tweet can be very harmful and misleading among its recipients. Therefore, authorities at WHO have called for action and asked the academic and scientific community to develop tools for managing the infodemic by the use of digital technologies and data science. The goal of this study is to develop and apply natural language processing models using deep learning to classify a collection of tweets that refer to the COVID-19 pandemic. Several simpler and widely used models are applied first and serve as a benchmark for deep learning methods, such as Long Short-Term Memory (LSTM) and Bidirectional Encoder Representations from Transformers (BERT). The results of the experiments show that the deep learning models outperform the traditional machine learning algorithms. The best approach is the BERT-based model.

The role of phonological and semantic representations in verbal short-term memory and delayed retention.

It has been suggested that phonological representations play a central role in verbal short-term memory, but when semantic knowledge has been investigated, it has also been shown to influence verbal short-term memory. Explaining this interaction between verbal short-term memory and the linguistic system has produced different theoretical positions: whether semantic knowledge is used to redintegrate phonological traces or if there is direct activation of both phonological and semantic knowledge upon encountering a word. The present study employed a new paradigm to systematically examine phonological and semantic representations in verbal short-term memory as well as long-term impacts. Across two experiments, a list of words was presented sequentially, followed by a probe word. Participants were to judge whether the probe word rhymed or was synonymous with any items on the list. Delayed memory was also tested. In Experiment 1, we found that immediate performance was better for synonym than rhyme judgements, and this continued to be the case after a brief delay. In Experiment 2, under a fast-encoding, running-span paradigm, we found similar activation of phonological and semantic knowledge. Nevertheless, accuracy was again higher for items probed with the semantic than rhyme cue in the long term. Results showed that indeed there are short-term semantic effects, in addition to phonological effects. Further, semantic processing can occur in a highly automatic and rapid manner, with strong influence on long-term memory. These findings provide a new perspective on viewing verbal short-term memory as operating more dynamically within the context of a complex linguistic system.

Sex-Specific Association of Endogenous PCSK9 With Memory Function in Elderly Subjects at High Cardiovascular Risk.

Background: Growing evidence indicates that cognitive decline and cardiovascular diseases (CVDs) share common vascular risk factors. Protease proprotein convertase subtilisin/kexin type 9 (PCSK9) is associated with CV disease risk and has been also involved in neuronal differentiation.Aim: Evaluate whether in patients at high CV risk cognitive function is related to PCSK9 levels.Methods. One hundred sixty-six patients (67 female) were enrolled. A detailed neuropsychological (NP) assessment was performed. PCSK9 levels were measured with ELISA.Results: Men had significantly higher short-term memory, executive function, and praxic and mental representation skills, as reflected by Forward Digit Span (FDS) (p = 0.005), Trail Making Test-A (TMT-A) (p = 0.047), Clock Drawing Test (CDT) (0.016). Endogenous PCSK9 levels were higher in female (p = 0.005). On linear regression analysis PCSK9 predicts short term memory only in females (Beta = 0.408, p = 0.001), with an interaction between PCSK9 and gender (p = 0.004 for interaction PCSK9 by sex). The association of PCSK9 with FDS in female was partially mediated by waist circumference (mediation effect 8.5%).Conclusions: In patients at high CV risk short term memory was directly related to PCSK9 levels only in women, revealing the relevance of sex in this relationship. The association of PCSK9 with memory function may be mediated, at least in part, by waist circumference.

Object and Spatial Context Representations in Visual Short-Term Memory.

Highlighted Research Paper: The Role of Location-Context Binding in Nonspatial Visual Working Memory, by Ying Cai, Jacqueline M. Fulvio, Qing Yu, Andrew D. Sheldon, and Bradley R. Postle (2020).

Maintaining verbal short-term memory representations in non-perceptual parietal regions

Buffer accounts of verbal short-term memory (STM) assume dedicated buffers for maintaining different types of information (e.g., phonological, visual) whereas embedded processes accounts argue against the existence of buffers and claim that STM consists of the activated portion of long-term memory (LTM). We addressed this debate by determining whether STM recruits the same neural substrate as LTM, or whether additional regions are involved in short-term storage. Using fMRI with representational similarity analysis (RSA), we examined the representational correspondence of multi-voxel neural activation patterns with the theoretical predictions for the maintenance of both phonological and semantic codes in STM. We found that during the delay period of a phonological STM task, phonological representations could be decoded in the left supramarginal gyrus (SMG) but not the superior temporal gyrus (STG), a speech processing region, for word stimuli. Whereas the pattern in the SMG was specific to phonology, a different region in the left angular gyrus showed RSA decoding evidence for the retention of either phonological or semantic codes, depending on the task context. Taken together, the results provide clear support for a dedicated buffer account of phonological STM, although evidence for a semantic buffer is equivocal.

Masks are memorable: ERP evidence on visual short-term memory and individuation in object substitution masking

ABSTRACT Object substitution masking (OSM) involves reduced perceptibility by a briefly lingering mask despite no spatial target-mask overlap. Two primary accounts for this phenomenon posit differences in object representation in visual short-term memory (VSTM). In the object substitution account a single representation resides in VSTM: a lone mask when attention is not focused on the target, and a lone target+mask when attention is focused on the target. The object updating account assumes individualization, with mask-alone and target+mask representations concurrently present in VSTM in correct delayed-offset OSM trials. We explored object individuation in OSM, focusing on the quantity of representations in VSTM in two OSM tasks with low (Easy Search) and high (Difficult Search) attentional demands, using event-related potentials (ERPs). Individuation was indicated in correct trials, with ERPs showing greater negativity in the sustained posterior contralateral negativity (SPCN) component for delayed compared to co-terminating offset conditions, consistent with greater VSTM load in delayed-offset. This effect, indicating target+mask and mask-alone as distinct memory representations, was found in Easy Search only. Additionally, no SPCN difference was found between correct and incorrect delayed trials in Difficult Search. We discuss our results in the context of individuation and perceptual load in OSM.

Orienting Attention to Short-Term Memory Representations via Sensory Modality and Semantic Category Retro-Cues.

There is growing interest in characterizing the neural mechanisms underlying the interactions between attention and memory. Current theories posit that reflective attention to memory representations generally involves a fronto-parietal attentional control network. The present study aimed to test this idea by manipulating how a particular short-term memory (STM) representation is accessed, that is, based on its input sensory modality or semantic category, during functional magnetic resonance imaging (fMRI). Human participants performed a novel variant of the retro-cue paradigm, in which they were presented with both auditory and visual non-verbal stimuli followed by Modality, Semantic, or Uninformative retro-cues. Modality and, to a lesser extent, Semantic retro-cues facilitated response time relative to Uninformative retro-cues. The univariate and multivariate pattern analyses (MVPAs) of fMRI time-series revealed three key findings. First, the posterior parietal cortex (PPC), including portions of the intraparietal sulcus (IPS) and ventral angular gyrus (AG), had activation patterns that spatially overlapped for both modality-based and semantic-based reflective attention. Second, considering both the univariate and multivariate analyses, Semantic retro-cues were associated with a left-lateralized fronto-parietal network. Finally, the experimental design enabled us to examine how dividing attention cross-modally within STM modulates the brain regions involved in reflective attention. This analysis revealed that univariate activation within bilateral portions of the PPC increased when participants simultaneously attended both auditory and visual memory representations. Therefore, prefrontal and parietal regions are flexibly recruited during reflective attention, depending on the representational feature used to selectively access STM representations.

A single, simple, statistical mechanism explains resource distribution and temporal updating in visual short-term memory

Investigations into the way that information is held and integrated within the visual system provides some basis for understanding how visual information is represented and processed. Just over sixty years ago, Swets, Shipley, McKey, and Green (1959) demonstrated that performance within an auditory detection task increases as a function of the square root of the number of stimulus observation intervals, following the predictions of basic sampling theory, indicating the efficient perceptual integration of stimulus information. This principle of observer performance contingent on a constant rate of stimulus sampling also forms the basis of the sample-size model (Palmer, 1990; Sewell, Lilburn, & Smith, 2014) which seeks to provide an account of how memory resources might be divided among item representations in visual short-term memory (VSTM). In this article, we combine the multiple observations paradigm of Swets and colleagues with the VSTM paradigm of Sewell and colleagues and show that the sample-size relationship accounts for both the increase in performance with the number of presentation intervals and the way that performance changes as a function of the number of items in memory. The model provides an account of both the overall information limit of VSTM and an account of the dynamics of that limit, demonstrating not only that observers can selectively update specific representations in memory but that performance in this task is accounted for by a simple statistical constraint. We discuss the implications for models of VSTM capacity and architecture generally, focusing on the implications for objecthood and the characteristics of encoding to and retrieval from memory.

A Path Toward Explainable AI and Autonomous Adaptive Intelligence: Deep Learning, Adaptive Resonance, and Models of Perception, Emotion, and Action.

Biological neural network models whereby brains make minds help to understand autonomous adaptive intelligence. This article summarizes why the dynamics and emergent properties of such models for perception, cognition, emotion, and action are explainable, and thus amenable to being confidently implemented in large-scale applications. Key to their explainability is how these models combine fast activations, or short-term memory (STM) traces, and learned weights, or long-term memory (LTM) traces. Visual and auditory perceptual models have explainable conscious STM representations of visual surfaces and auditory streams in surface-shroud resonances and stream-shroud resonances, respectively. Deep Learning is often used to classify data. However, Deep Learning can experience catastrophic forgetting: At any stage of learning, an unpredictable part of its memory can collapse. Even if it makes some accurate classifications, they are not explainable and thus cannot be used with confidence. Deep Learning shares these problems with the back propagation algorithm, whose computational problems due to non-local weight transport during mismatch learning were described in the 1980s. Deep Learning became popular after very fast computers and huge online databases became available that enabled new applications despite these problems. Adaptive Resonance Theory, or ART, algorithms overcome the computational problems of back propagation and Deep Learning. ART is a self-organizing production system that incrementally learns, using arbitrary combinations of unsupervised and supervised learning and only locally computable quantities, to rapidly classify large non-stationary databases without experiencing catastrophic forgetting. ART classifications and predictions are explainable using the attended critical feature patterns in STM on which they build. The LTM adaptive weights of the fuzzy ARTMAP algorithm induce fuzzy IF-THEN rules that explain what feature combinations predict successful outcomes. ART has been successfully used in multiple large-scale real world applications, including remote sensing, medical database prediction, and social media data clustering. Also explainable are the MOTIVATOR model of reinforcement learning and cognitive-emotional interactions, and the VITE, DIRECT, DIVA, and SOVEREIGN models for reaching, speech production, spatial navigation, and autonomous adaptive intelligence. These biological models exemplify complementary computing, and use local laws for match learning and mismatch learning that avoid the problems of Deep Learning.

Forgetting Rates on the Recency Portion of a Word List Predict Conversion from Mild Cognitive Impairment to Alzheimer's Disease.

Amnestic mild cognitive impairment has a greater risk of progressing to Alzheimer's disease (AD). Consistent with AD patients' distinctive deficit in consolidating new memory traces, in a recent study we demonstrated that the forgetting rate on the recency portion of a word list differentiates AD from other forms of dementia. In line with this finding, the aim of this study was to investigate whether increased recency forgetting could be a reliable index for predicting amnestic mild cognitive impairment (MCI) patients' conversion to AD. For this purpose, we compared accuracy in immediate and delayed recall from different portions of a word list in a group of patients with amnestic MCI who converted (C-MCI) or did not convert (S-MCI) to AD during a three-year follow-up period and in a group of normal controls. The results of the present study show that the forgetting from the recency portion of the list (operationalized as a ratio between immediate and delayed recall) was significantly larger in C-MCI than in S-MCI patients. Consistently, the hierarchical logistic regression analyses demonstrated that the recency ratio is a strong predictor of group membership. Similar to what occurs in full-blown AD patients, the results of our study suggest that the increased forgetting rate from the recency portion of the list in C-MCI patients is due to severely reduced efficiency in converting transitory short-term memory representations into stable long-term memory traces. This is consistent with prominent involvement of neuropathological changes in the cortical areas of the medial-temporal lobes and suggests that the recency ratio is a cognitive marker able to identify MCI patients who have a greater likelihood of progressing to AD.

Fast and effective biomedical named entity recognition using temporal convolutional network with conditional random field.

Biomedical named entity recognition (Bio-NER) is the prerequisite for mining knowledge from biomedical texts. The state-of-the-art models for Bio-NER are mostly based on bidirectional long short-term memory (BiLSTM) and bidirectional encoder representations from transformers (BERT) models. However, both BiLSTM and BERT models are extremely computationally intensive. To this end, this paper proposes a temporal convolutional network (TCN) with a conditional random field (TCN-CRF) layer for Bio-NER. The model uses TCN to extract features, which are then decoded by the CRF to obtain the final result. We improve the original TCN model by fusing the features extracted by convolution kernel with different sizes to enhance the performance of Bio-NER. We compared our model with five deep learning models on the GENIA and CoNLL-2003 datasets. The experimental results show that our model can achieve comparative performance with much less training time. The implemented code has been made available to the research community.

Orienting Attention to Auditory and Visual Short-term Memory: The Roles of Age, Hearing Loss, and Cognitive Status

ABSTRACTBackground/Study Context: Attention can be reflectively oriented to a visual or auditory representation in short-term memory, but it is not clear how aging and hearing acuity affects reflective attention. The purpose of the present study was to examine whether performance in auditory and visual reflective attention tasks varies as a function of participants’ age and hearing status.Methods: Young (19 to 33 years) and older adults with normal or mild to moderate hearing loss (62–90 years) completed a delayed match-to-sample task in which participants were first presented with a memory array of four different digits to hold in memory. Two digits were presented visually (left and right hemifield), and two were presented aurally (left and right ears simultaneously). During the retention interval, participants were presented with a cue (dubbed retro-cue), which could be either uninformative or indicated to the participants to retrospectively orient their attention to either auditory short-term memory (ASTM) or visual short-term memory (VSTM). The cue was followed by another delay, after which a single item was presented (i.e., test probe) for comparison (match or no match) with the items held in ASTM and/or VSTM.Results: Overall, informative retro-cue yielded faster response time than uninformative retro-cue. The retro-cue benefit in response time was comparable for auditory and visual-orienting retro-cue and similar in young and older adults. Regression analyses showed that only the auditory-orienting retro-cue benefit was predicted by hearing status rather than age per se.Conclusion: Both younger and older adults can benefit from visual and auditory-orienting retro-cues, but the auditory-orienting retro-cue benefit decreases with poorer hearing acuity. This finding highlights changes in cognitive processes that come with age even in those with just mild-to-moderate hearing loss, and suggest that older adults’ performance in working memory tasks is sensitive to low level auditory scene analysis (i.e., concurrent sound segregation).

Short-term memory based on activated long-term memory: A review in response to Norris (2017).

Short-term memory (STM), the limited information temporarily in a state of heightened accessibility, includes just-presented events and recently retrieved information. Norris (2017) argued for a prominent class of theories in which STM depends on the brain keeping a separate copy of new information, and against alternatives in which the information is held only in a portion of long-term memory (LTM) that is currently activated (aLTM). Here I question premises of Norris' case for separate-copy theories in the following ways. (a) He did not allow for implications of the common assumption (e.g., Cowan, 1999; Cowan & Chen, 2009) that aLTM can include new, rapidly formed LTM records of a trial within an STM task. (b) His conclusions from pathological cases of impaired STM along with intact LTM are tenuous; these rare cases can be explained by impairments in encoding, processing, or retrieval related to LTM rather than passive maintenance. (c) Although Norris reasonably allowed structured pointers to aLTM instead of separate copies of the actual item representations in STM, the same structured pointers may well be involved in long-term learning. (d) Last, models of STM storage can serve as the front end of an LTM learning system rather than being separate. I summarize evidence for these premises and an updated version of an alternative theory in which storage depends on aLTM (newly clarified), and, embedded within it, information enhanced by the current focus of attention (Cowan, 1988, 1999), with no need for a separate STM copy. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Short-term memory problems for phonemes’ serial order in adults with dyslexia: Evidence from a different analysis of the Nonword repetition task

AbstractMany studies show that poor readers make more errors in nonword repetition than better readers. Although this finding is generally linked to the lower quality of poor readers’ phonological representations in verbal short-term memory, the nature of this poor performance remains unclear. We addressed this issue by focusing on two types of phoneme-related performance in a nonword repetition task: (a) recall of phonemes irrespective of their serial order (phoneme identity) and (b) recall of correctly reproduced phonemes’ serial order (serial order). We tested 91 young adults with and without dyslexia. Generalized linear mixed-effects models demonstrated that controls outperformed individuals with dyslexia in the recall of phonemes’ serial order but failed to detect a difference in the recall of phonemes’ identity. These findings are discussed not only in terms of the nature of or access to phonological representations but also in terms of another concept that has recently been advanced in the literature: a specialized serial order mechanism in verbal short-term memory. We also consider the possibility that individuals with dyslexia may be less sensitive to phonotactic constraints.

Ensemble coding of memory strength in recognition tests.

Recent work by Benjamin and colleagues (Psychological Review, 116 (1), 84-115, 2009; Journal of Experimental Psychology: Learning, Memory, and Cognition, 39(5), 1601-1608, 2013) suggests that recognition memory decisions are corrupted by random variability in decision criteria. This conclusion, which explains several anomalies in the recognition literature, was based on fits of the Noisy Decision Theory of Signal Detection (NDT) to a novel task: ensemble recognition. In the ensemble task, participants make Old/New decisions to ensembles of items rather than single items. The NDT assumption that criteria are fixed across ensembles was criticized by Kellen, Klauer, and Singmann (Psychological Review, 119 (3), 457-479, 2012), and defended by Benjamin (Psychological Review, 120, 720-726, 2013). Little attention, however, has been paid to the assumption of the best-fitting NDT model that participants solve the ensemble task by calculating the average memory strength of items in the probe display. We review evidence of summary statistical representation in visual perception and short-term memory that suggests the aggregation hypothesis is plausible, and hold it up to test in three experiments using the direct ratings procedure. Although we conclude that participants can produce estimates of average probe memory strength at test, in line with the assumptions of NDT, the mechanisms and strategies used to produce such estimates remain unclear.

Theory of visual attention thalamic model for visual short-term memory capacity and top-down control: Evidence from a thalamo-cortical structural connectivity analysis

In the theory of visual attention (TVA), it is suggested that objects in a visual scene compete for representation in a visual short-term memory (vSTM) store. The race towards the store is assumed to be biased by top-down controlled weighting of the objects according to their task relevance. Only objects that reach the store before its capacity limitation is reached are represented consciously in a given instant. TVA-based computational modeling of participants’ performance in whole- and partial-report tasks permits independent parameters of individual efficiency of top-down control α and vSTM storage capacity K to be extracted. The neural interpretation of the TVA proposes recurrent loops between the posterior thalamus and posterior visual cortices to be relevant for generating attentional weights for competing objects and for maintaining selected objects in vSTM. Accordingly, we tested whether structural connectivity between posterior thalamus and occipital cortices (PT-OC) is associated with estimates of top-down control and vSTM capacity. We applied whole- and partial-report tasks and probabilistic tractography in a sample of 37 healthy adults. We found vSTM capacity K to be associated with left PT-OC structural connectivity and a trend-wise relation between top-down control α and right PT-OC structural connectivity. These findings support the assumption of the relevance of thalamic structures and their connections to visual cortex for top-down control and vSTM capacity.

Central tendency representation and exemplar matching in visual short-term memory.

I address a recent extension of the generalized context model (GCM), a model which excludes prototypes, to the visual short-term memory (VSTM) literature, which is currently deluged with prototype effects. The paper includes a brief review whose aim is to discuss the background and key findings suggesting that prototypes have an obligatory influence on visual short-term memory responses in the same VSTM task that the GCM's random walk extension, EBRW, was extended to account for: Sternberg scanning. I present a new model that incorporates such "central tendency representations" in memory, as well as several other regularities of the literature, and compare its prediction and postdictions to those of the GCM on some unpublished Sternberg scanning data. The GCM cannot account for the pattern in those data without post hoc modifications but the pattern is predicted nicely by the central tendency representation model. Although the new model is certainly wrong, the review and modeling exercise suggest a reconsideration of prototype models may be warranted, at least in the VSTM literature.

Decay of internal reference information in duration discrimination: Intertrial interval modulates the Type B effect

Psychophysical evidence suggests that human perception of a stimulus is assimilated towards previous stimuli. The internal reference model (IRM) explains such assimilation through an internal reference (IR), which integrates past and present stimulus representations and thus might be conceived as a form of perceptual memory. In this study, we investigated whether the IR decays with time, as previously shown for perceptual memory representations in general. One specific prediction of IRM is higher discrimination sensitivity when a constant standard precedes rather than follows a variable comparison in a two-alternative forced-choice (2AFC) discrimination task. Furthermore, the magnitude of this so-called negative Type B effect should decrease with decreasing weighting of past stimulus information in the integration process. Therefore, decay of the IR should result in a reduced Type B effect. To examine this prediction, we carried out a 2AFC duration discrimination experiment with a short (1,600 ms) and a long (3,200 ms) intertrial interval (ITI). As expected, a reduced negative Type B effect was observed at the long compared with the short ITI, consistent with the idea that humans rely on the immediate past when evaluating current sensory input, however, less so when the IR incorporating the perceptual short-term memory representation of these past stimuli has already decayed.

Temporal Expectation Modulates the Cortical Dynamics of Short-Term Memory.

Increased memory load is often signified by enhanced neural oscillatory power in the alpha range (8-13 Hz), which is taken to reflect inhibition of task-irrelevant brain regions. The corresponding neural correlates of memory decay, however, are not yet well understood. In the current study, we investigated auditory short-term memory decay in humans using a delayed matching-to-sample task with pure-tone sequences. First, in a behavioral experiment, we modeled memory performance over six different delay-phase durations. Second, in a MEG experiment, we assessed alpha-power modulations over three different delay-phase durations. In both experiments, the temporal expectation for the to-be-remembered sound was manipulated so that it was either temporally expected or not. In both studies, memory performance declined over time, but this decline was weaker when the onset time of the to-be-remembered sound was expected. Similarly, patterns of alpha power in and alpha-tuned connectivity between sensory cortices changed parametrically with delay duration (i.e., decrease in occipitoparietal regions, increase in temporal regions). Temporal expectation not only counteracted alpha-power decline in heteromodal brain areas (i.e., supramarginal gyrus), but also had a beneficial effect on memory decay, counteracting memory performance decline. Correspondingly, temporal expectation also boosted alpha connectivity within attention networks known to play an active role during memory maintenance. The present data show how patterns of alpha power orchestrate short-term memory decay and encourage a more nuanced perspective on alpha power across brain space and time beyond its inhibitory role.SIGNIFICANCE STATEMENT Our sensory memories of the physical world fade quickly. We show here that this decay of short-term memory can be counteracted by so-called temporal expectation; that is, knowledge of when to expect a sensory event that an individual must remember. We also show that neural oscillations in the "alpha" (8-13 Hz) range index both the degree of memory decay (for brief sound patterns) and the respective memory benefit from temporal expectation. Spatially distributed cortical patterns of alpha power show opposing effects in auditory versus visual sensory cortices. Moreover, alpha-tuned connectivity changes within supramodal attention networks reflect the allocation of neural resources as short-term memory representations fade.