Limited Working Memory Capacity Research Articles

Working memory contributions to reinforcement learning impairments in schizophrenia.

Previous research has shown that patients with schizophrenia are impaired in reinforcement learning tasks. However, behavioral learning curves in such tasks originate from the interaction of multiple neural processes, including the basal ganglia- and dopamine-dependent reinforcement learning (RL) system, but also prefrontal cortex-dependent cognitive strategies involving working memory (WM). Thus, it is unclear which specific system induces impairments in schizophrenia. We recently developed a task and computational model allowing us to separately assess the roles of RL (slow, cumulative learning) mechanisms versus WM (fast but capacity-limited) mechanisms in healthy adult human subjects. Here, we used this task to assess patients' specific sources of impairments in learning. In 15 separate blocks, subjects learned to pick one of three actions for stimuli. The number of stimuli to learn in each block varied from two to six, allowing us to separate influences of capacity-limited WM from the incremental RL system. As expected, both patients (n = 49) and healthy controls (n = 36) showed effects of set size and delay between stimulus repetitions, confirming the presence of working memory effects. Patients performed significantly worse than controls overall, but computational model fits and behavioral analyses indicate that these deficits could be entirely accounted for by changes in WM parameters (capacity and reliability), whereas RL processes were spared. These results suggest that the working memory system contributes strongly to learning impairments in schizophrenia.

The steady-state visual evoked potential reveals neural correlates of the items encoded into visual working memory

Visual working memory (VWM) capacity limitations are estimated to be ~4 items. Yet, it remains unclear why certain items from a given memory array may be successfully retrieved from VWM and others are lost. Existing measures of the neural correlates of VWM cannot address this question because they measure the aggregate processing of the entire stimulus array rather than neural signatures of individual items. Moreover, this cumulative processing is usually measured during the delay period, thereby reflecting the allocation of neural resources during VWM maintenance. Here, we use the steady-state visual evoked potential (SSVEP) to identify the neural correlates of individual stimuli at VWM encoding and test two distinct hypotheses: the focused-resource hypothesis and the diffuse-resource hypothesis, for how the allocation of neural resources during VWM encoding may contribute to VWM capacity limitations. First, we found that SSVEP amplitudes were larger for stimuli that were later remembered than for items that were subsequently forgotten. Second, this pattern generalized so that the SSVEP amplitudes were also larger for the unprobed stimuli in correct compared to incorrect trials. These data are consistent with the diffuse-resource view in which attentional resources are broadly allocated across the whole stimulus array. These results illustrate the important role encoding mechanisms play in limiting the capacity of VWM.

Focused attention improves working memory: implications for flexible-resource and discrete-capacity models

Performance in working memory (WM) tasks depends on the capacity for storing objects and on the allocation of attention to these objects. Here, we explored how capacity models need to be augmented to account for the benefit of focusing attention on the target of recall. Participants encoded six colored disks (Experiment 1) or a set of one to eight colored disks (Experiment 2) and were cued to recall the color of a target on a color wheel. In the no-delay condition, the recall-cue was presented after a 1,000-ms retention interval, and participants could report the retrieved color immediately. In the delay condition, the recall-cue was presented at the same time as in the no-delay condition, but the opportunity to report the color was delayed. During this delay, participants could focus attention exclusively on the target. Responses deviated less from the target's color in the delay than in the no-delay condition. Mixture modeling assigned this benefit to a reduction in guessing (Experiments 1 and 2) and transposition errors (Experiment 2). We tested several computational models implementing flexible or discrete capacity allocation, aiming to explain both the effect of set size, reflecting the limited capacity of WM, and the effect of delay, reflecting the role of attention to WM representations. Both models fit the data better when a spatially graded source of transposition error is added to its assumptions. The benefits of focusing attention could be explained by allocating to this object a higher proportion of the capacity to represent color.

Cognitive Processing of Information with Visitor Value in Cultural Heritage Environments. The Case of the SEE TCP SAGITTARIUS 2011–2014

The South East Europe Transnational Cooperation Project SAGITTARIUS is a research and developmental project aiming to effectively communicate values of cultural heritage to non-captives audiences in an entertaining and participatory way. A significance assessment facilitates the selection of assets with interpretive potential. Information architecture presupposes a limited working memory capacity to deal with visual and auditory material and a long-term memory able to hold mental representations that vary in their degree of automation. A communication pattern has been developed and adapted to the conditions regulating the recreational learning environment. It facilitates attention and information retention, reinforces the association chain enabling new cognitive content to relate to prior knowledge. A Roving Museum offers a scarce time-budget audience access to interpretive narratives via a portable infrastructure, a mobile telephony application and a social media driven platform.

Movement planning and attentional control of visuospatial working memory: evidence from a grasp-to-place task

In this study, we have investigated the influence of available attentional resources on the dual-task costs of implementing a new action plan and the influence of movement planning on the transfer of information into visuospatial working memory. To approach these two questions, we have used a motor-memory dual-task design in which participants grasped a sphere and planned a placing movement toward a left or right target according to a directional arrow. Subsequently, they encoded a centrally presented memory stimulus (4×4 symbol matrix). While maintaining the information in working memory, a visual stay/change cue (presented on the left, center or right) either confirmed or reversed the planned movement direction. That is, participants had to execute either the prepared or the re-planned movement and finally reported the symbols at leisure. The results show that both, shifts of spatial attention required to process the incongruent stay/change cues and movement re-planning, constitute processing bottlenecks as they both reduced visuospatial working memory performance. Importantly, the spatial attention shifts and movement re-planning appeared to be independent of each other. Further, we found that the initial preparation of the placing movement influenced the report pattern of the central working memory stimulus. Preparing a leftward movement resulted in better memory performance for the left stimulus side, while the preparation of a rightward movement resulted in better memory performance for the right stimulus side. Hence, movement planning influenced the transfer of information into the capacity-limited working memory store. Therefore, our results suggest complex interactions in that the processes involved in movement planning, spatial attention and visuospatial working memory are functionally correlated but not linked in a mandatory fashion.

The magic number 70 (plus or minus 20): Variables determining performance in the Rodent Odor Span Task

The olfactory span task (OST) uses an incrementing non-matching to sample procedure such that the number of stimuli to remember increases during the session. The number of consecutive correct responses (span length) and percent correct as a function of the memory load have been viewed as defining rodent working memory capacity limitations in several studies using the OST. However, the procedural parameters of the OST vary across experiments and their effects are not well understood. For example, in several studies, the number of stimuli to remember is confounded with the number of comparison stimuli displayed in the test arena. Experiment 1 addressed whether performance is influenced by the number of comparison choices available on any given trial (2, 5, 10) as well as the number of odor stimuli to remember during a session (12, 24, 36). Performance was most accurate when the number of stimuli to remember was low, as would be expected from a working memory interpretation of OST. However, accuracy was also affected by the number of comparison stimulus choices. High levels of accuracy were seen even with 36 odors, suggesting that the capacity for odor memory in rats was greater than suggested by previous research. Experiment 2 attempted to define this capacity by programming sessions with 36, 48 or 72 stimuli to remember in a group of rats that had previously received extensive OST training. Highly accurate performance (80% correct or better) was sustained throughout the session at even the greatest memory loads, arguing strongly against the notion that the OST models the limited capacity of human working memory. Experiment 3 explored the possibility that stimulus control in the OST is based on relative stimulus familiarity, rather than recognition of stimuli not yet presented during the current session. Number of odor cups visited increased with the number of comparisons in the arena, but rats rarely sampled all of the comparison odors before responding. However, on probe trials which included only stimuli that had been presented during the session, latency to respond and number of comparisons sampled was sharply increased. These data suggest that responding in the OST is determined not just by relative familiarity, but rather by a more specific “what-when” or perhaps “how long ago” form of stimulus control.

Temporal buffering and visual capacity: the time course of object formation underlies capacity limits in visual cognition.

Capacity limits are a hallmark of visual cognition. The upper boundary of our ability to individuate and remember objects is well known but—despite its central role in visual information processing—not well understood. Here, we investigated the role of temporal limits in the perceptual processes of forming “object files.” Specifically, we examined the two fundamental mechanisms of object file formation—individuation and identification—by selectively interfering with visual processing by using forward and backward masking with variable stimulus onset asynchronies. While target detection was almost unaffected by these two types of masking, they showed distinct effects on the two different stages of object formation. Forward “integration” masking selectively impaired object individuation, whereas backward “interruption” masking only affected identification and the consolidation of information into visual working memory. We therefore conclude that the inherent temporal dynamics of visual information processing are an essential component in creating the capacity limits in object individuation and visual working memory.

Keeping timbre in mind: Working memory for complex sounds that can't be verbalized.

Properties of auditory working memory for sounds that lack strong semantic associations and are not readily verbalized or sung are poorly understood. We investigated auditory working memory capacity for lists containing 2-6 easily discriminable abstract sounds synthesized within a constrained timbral space, at delays of 1-6 s (Experiment 1), and the effect of greater perceptual variability among list items on capacity estimates at delays of 1-6 s (Experiment 2). Working memory capacity estimates of 1-2 items were found in all conditions and increased significantly as the perceptual variability among the list items increased. Nonetheless, the capacity estimates were smaller than the commonly observed average working memory capacity limit of 3-5 items. Decay profiles in both experiments were comparable with those previously reported in the verbal and auditory working memory literature. The results help define boundary conditions on capacity estimates for nonverbalizable timbres that lack strong long-term memory associations.

Effects of children's working memory capacity and processing speed on their sentence imitation performance

More limited working memory capacity and slower processing for language and cognitive tasks are characteristics of many children with language difficulties. Individual differences in processing speed have not consistently been found to predict language ability or severity of language impairment. There are conflicting views on whether working memory and processing speed are integrated or separable abilities. To evaluate four models for the relations of individual differences in children's processing speed and working memory capacity in sentence imitation. The models considered whether working memory and processing speed are integrated or separable, as well as the effect of the number of operations required per sentence. The role of working memory as a mediator of the effect of processing speed on sentence imitation was also evaluated. Forty-six children with varied language and reading abilities imitated sentences. Working memory was measured with the Competing Language Processing Task (CLPT), and processing speed was measured with a composite of truth-value judgment and rapid automatized naming tasks. Mixed-effects ordinal regression models evaluated the CLPT and processing speed as predictors of sentence imitation item scores. A single mediator model evaluated working memory as a mediator of the effect of processing speed on sentence imitation total scores. Working memory was a reliable predictor of sentence imitation accuracy, but processing speed predicted sentence imitation only as a component of a processing speed by number of operations interaction. Processing speed predicted working memory capacity, and there was evidence that working memory acted as a mediator of the effect of processing speed on sentence imitation accuracy. The findings support a refined view of working memory and processing speed as separable factors in children's sentence imitation performance. Processing speed does not independently explain sentence imitation accuracy for all sentence types, but contributes when the task requires more mental operations. Processing speed also has an indirect effect on sentence imitation by contributing to working memory capacity.

Serial recall of visuospatial and verbal information with and without material-specific interference: Implications for contemporary models of working memory

A longstanding question in working memory (WM) research concerns the fractionation of verbal and nonverbal processing. Although some contemporary models include both domain-specific and general-purpose mechanisms, the necessity to postulate differential processing of verbal and nonverbal material remains unclear. In the present two-experiment series we revisit the order reconstruction paradigm that Jones, Farrand, Stuart, and Morris (1995) used to support a unitary model of WM. Goals were to assess (1) whether serial position curves for dot positions differ from curves for letter names; and (2) whether selective interference can be demonstrated. Although we replicated Jones et al.'s finding of similar serial position curves for the two tasks, this similarity could reflect the demands of the order reconstruction paradigm rather than undifferentiated processing of verbal and nonverbal stimuli. Both generalised and material-specific interference was found, which can be attributed to competition between primary and secondary tasks for attentional resources. As performance levels for the combined primary and secondary tasks exceed active WM capacity limits, primary task items apparently are removed from active memory during processing of the secondary list and held temporarily in maintenance storage. We conclude that active WM is multimodal but maintenance stores may be domain specific.

Working memory capacity: the need for process task-analysis

Working memory capacity: the need for process task-analysis

Working memory inefficiency: Minimal information is utilized in visual recognition tasks.

Can people make perfect use of task-relevant information in working memory (WM)? Specifically, when questioned about an item in an array that does not happen to be in WM, can participants take into account other items that are in WM, eliminating them as response candidates? To address this question, an ideal-responder model that assumes perfect use of items in a capacity-limited WM was tested against a minimal-responder model that assumes use of only information about the queried item. Three different WM tasks were adopted: change detection, identity recognition, and location recognition. The change-detection task produced benchmark WM results. The 2 novel tasks showed that only the minimal responder model provided convergence with this benchmark. This finding was replicable even when the change-detection task was replaced by a feature-switch detection task. Thus, it appears that people do not make full use of information in WM.

Evidence against decay in verbal working memory.

The article tests the assumption that forgetting in working memory for verbal materials is caused by time-based decay, using the complex-span paradigm. Participants encoded 6 letters for serial recall; each letter was preceded and followed by a processing period comprising 4 trials of difficult visual search. Processing duration, during which memory could decay, was manipulated via search set size. This manipulation increased retention interval by up to 100% without having any effect on recall accuracy. This result held with and without articulatory suppression. Two experiments using a dual-task paradigm showed that the visual search process required central attention. Thus, even when memory maintenance by central attention and by articulatory rehearsal was prevented, a large delay had no effect on memory performance, contrary to the decay notion. Most previous experiments that manipulated the retention interval and the opportunity for maintenance processes in complex span have confounded these variables with time pressure during processing periods. Three further experiments identified time pressure as the variable that affected recall. We conclude that time-based decay does not contribute to the capacity limit of verbal working memory.

Effective Visual Working Memory Capacity: An Emergent Effect from the Neural Dynamics in an Attractor Network

The study of working memory capacity is of outmost importance in cognitive psychology as working memory is at the basis of general cognitive function. Although the working memory capacity limit has been thoroughly studied, its origin still remains a matter of strong debate. Only recently has the role of visual saliency in modulating working memory storage capacity been assessed experimentally and proved to provide valuable insights into working memory function. In the computational arena, attractor networks have successfully accounted for psychophysical and neurophysiological data in numerous working memory tasks given their ability to produce a sustained elevated firing rate during a delay period. Here we investigate the mechanisms underlying working memory capacity by means of a biophysically-realistic attractor network with spiking neurons while accounting for two recent experimental observations: 1) the presence of a visually salient item reduces the number of items that can be held in working memory, and 2) visually salient items are commonly kept in memory at the cost of not keeping as many non-salient items.Our model suggests that working memory capacity is determined by two fundamental processes: encoding of visual items into working memory and maintenance of the encoded items upon their removal from the visual display. While maintenance critically depends on the constraints that lateral inhibition imposes to the mnemonic activity, encoding is limited by the ability of the stimulated neural assemblies to reach a sufficiently high level of excitation, a process governed by the dynamics of competition and cooperation among neuronal pools. Encoding is therefore contingent upon the visual working memory task and has led us to introduce the concept of effective working memory capacity (eWMC) in contrast to the maximal upper capacity limit only reached under ideal conditions.

From distributed resources to limited slots in multiple-item working memory: a spiking network model with normalization.

Recent behavioral studies have given rise to two contrasting models for limited working memory capacity: a "discrete-slot" model in which memory items are stored in a limited number of slots, and a "shared-resource" model in which the neural representation of items is distributed across a limited pool of resources. To elucidate the underlying neural processes, we investigated a continuous network model for working memory of an analog feature. Our model network fundamentally operates with a shared resource mechanism, and stimuli in cue arrays are encoded by a distributed neural population. On the other hand, the network dynamics and performance are also consistent with the discrete-slot model, because multiple objects are maintained by distinct localized population persistent activity patterns (bump attractors). We identified two phenomena of recurrent circuit dynamics that give rise to limited working memory capacity. As the working memory load increases, a localized persistent activity bump may either fade out (so the memory of the corresponding item is lost) or merge with another nearby bump (hence the resolution of mnemonic representation for the merged items becomes blurred). We identified specific dependences of these two phenomena on the strength and tuning of recurrent synaptic excitation, as well as network normalization: the overall population activity is invariant to set size and delay duration; therefore, a constant neural resource is shared by and dynamically allocated to the memorized items. We demonstrate that the model reproduces salient observations predicted by both discrete-slot and shared-resource models, and propose testable predictions of the merging phenomenon.

Understanding Mathematics: Some Key Factors

<p style="text-align:justify">Mathematics is well known as a subject area where there can be problems in terms of understanding as well as retaining positive attitudes. In a large study involving 813 school students (ages approximately 10-12) drawn from two different school systems in Pakistan, the effect of limited working memory capacity on performance in mathematics was explored along with a survey of areas of difficulty and student attitudes. This involved looking at student perceptions of their experiences, the nature of the difficulties they have with mathematics and possible reasons for these difficulties. The overall aim is to explore the extent of the effect of working memory and to gain insights so that practical ways forward to enhance mathematics education can be identified. It was found that limited working memory capacity has a very strong influence on performance, confirming other studies. Indeed, if the cognitive load exceeds the capacity of working memory, understanding becomes a casualty, with consequent attitude deterioration. Students need to be able to see that mathematics has a purpose in being able to be applied to real-life situations. However, attempts to develop applications may often generate further working memory overload. Curricula devised by those outside the classroom can sometimes be inappropriate while topics causing the greatest problems at these ages and include areas of geometry, statistics and the applications of mathematics.</p>

How much of reinforcement learning is working memory, not reinforcement learning? A behavioral, computational, and neurogenetic analysis

Instrumental learning involves corticostriatal circuitry and the dopaminergic system. This system is typically modeled in the reinforcement learning (RL) framework by incrementally accumulating reward values of states and actions. However, human learning also implicates prefrontal cortical mechanisms involved in higher level cognitive functions. The interaction of these systems remains poorly understood, and models of human behavior often ignore working memory (WM) and therefore incorrectly assign behavioral variance to the RL system. Here we designed a task that highlights the profound entanglement of these two processes, even in simple learning problems. By systematically varying the size of the learning problem and delay between stimulus repetitions, we separately extracted WM-specific effects of load and delay on learning. We propose a new computational model that accounts for the dynamic integration of RL and WM processes observed in subjects' behavior. Incorporating capacity-limited WM into the model allowed us to capture behavioral variance that could not be captured in a pure RL framework even if we (implausibly) allowed separate RL systems for each set size. The WM component also allowed for a more reasonable estimation of a single RL process. Finally, we report effects of two genetic polymorphisms having relative specificity for prefrontal and basal ganglia functions. Whereas the COMT gene coding for catechol-O-methyl transferase selectively influenced model estimates of WM capacity, the GPR6 gene coding for G-protein-coupled receptor 6 influenced the RL learning rate. Thus, this study allowed us to specify distinct influences of the high-level and low-level cognitive functions on instrumental learning, beyond the possibilities offered by simple RL models.

The Processing of Formulaic Language

It is generally accepted that we store representations of individual words in our mental lexicon. There is growing agreement that the lexicon also contains formulaic language (How are you? kick the bucket). In fact, there are compelling reasons to think that the brain represents formulaic sequences in long-term memory, bypassing the need to compose them online through word selection and grammatical sequencing in capacity-limited working memory. The research surveyed in this chapter strongly supports the position that there is an advantage in the way that native speakers process formulaic language compared to nonformulaic language. This advantage extends to the access and use of different types of formulaic language, including idioms, binomials, collocations, and lexical bundles. However, the evidence is mixed for nonnative speakers. While very proficient nonnatives sometimes exhibit processing advantages similar to natives, less proficient learners often have been shown to process formulaic language in a word-by-word manner similar to nonformulaic language. Furthermore, if the formulaic language is idiomatic (where the meaning cannot be understood from the component words), the figurative meanings can be much more difficult to process for nonnatives than nonidiomatic, nonformulaic language.

What is simulation for?

What is simulation for?

Does high memory load kick task-irrelevant information out of visual working memory?

The limited capacity of visual working memory (VWM) requires the existence of an efficient information selection mechanism. While it has been shown that under low VWM load, an irrelevant simple feature can be processed, its fate under high load (e.g., six objects) remains unclear. We explored this issue by probing the "irrelevant-change distracting effect," in which the change of a stored irrelevant feature affects performance. Simple colored shapes were used as stimuli, with color as the target. Using a whole-probe method (presenting six objects in both the memory and test arrays), in Experiment 1 we found that a change to one of the six shapes led to a significant distracting effect. Using a partial-probe method (presenting the probe either at the screen center or at a location selected from the memory array), in Experiment 2 we showed the distracting effect again. These results suggest that irrelevant simple features can be stored into VWM, regardless of memory load.