Linking working memory and long-term memory: a computational model of the learning of new words

The nonword repetition (NWR) test has been shown to be a good predictor of children's vocabulary size. NWR performance has been explained using phonological working memory, which is seen as a critical component in the learning of new words. However, no detailed specification of the link between phonological working memory and long-term memory (LTM) has been proposed. In this paper, we present a computational model of children's vocabulary acquisition (EPAM-VOC) that specifies how phonological working memory and LTM interact. The model learns phoneme sequences, which are stored in LTM and mediate how much information can be held in working memory. The model's behaviour is compared with that of children in a new study of NWR, conducted in order to ensure the same nonword stimuli and methodology across ages. EPAM-VOC shows a pattern of results similar to that of children: performance is better for shorter nonwords and for wordlike nonwords, and performance improves with age. EPAM-VOC also simulates the superior performance for single consonant nonwords over clustered consonant nonwords found in previous NWR studies. EPAM-VOC provides a simple and elegant computational account of some of the key processes involved in the learning of new words: it specifies how phonological working memory and LTM interact; makes testable predictions; and suggests that developmental changes in NWR performance may reflect differences in the amount of information that has been encoded in LTM rather than developmental changes in working memory capacity.     

The computational implementation of the landscape model: Modeling inferential processes and memory representations of text comprehension

The complexity of text comprehension demands a computational approach to describe the cognitive processes involved. In this article, we present the computational implementation of the landscape model of reading. This model captures both on-line comprehension processes during reading and the off-line memory representation after reading is completed, incorporating both memory-based and coherence-based mechanisms of comprehension. The overall architecture and specific parameters of the program are described, and a running example is provided. Several studies comparing computational and behavioral data indicate that the implemented model is able to account for cycle-by-cycle comprehension processes and memory for a variety of text types and reading situations.     

Developing the theory of working memory

The theory of working memory was devised to explain the effects of a concurrent memory load in various experimental situations in terms of the operation of a central executive processor and a phonemic response buffer. It also explains the effects of phonemic similarity, articulatory suppression, word length, and unattended speech. Experiment 1 demonstrated that a concurrent memory load markedly reduced the phonemic similarity effect in immediate serial recall, which was taken to support the concept of a limited-capacity phonemic response buffer. A more detailed analysis of the results suggested that a concurrent memory load may affect the storage capacity of the central executive processor and the translation of orthographic stimuli into phonological representations, as well as the storage capacity of the phonemic response buffer. Experiment 2 showed that a concurrent free-recall task reduced the phonemic similarity effect in immediate serial recall, but only in the case of visually presented sequences of items. Moreover, unattended speech was found to have no effect upon performance in immediate free recall. These results were taken to imply that the phonemic response buffer contributes only to performance in cognitive tasks that require the accurate retention of serial-order information.     

Interactions between frontal cortex and basal ganglia in working memory: a computational model

The frontal cortex and the basal ganglia interact via a relatively well understood and elaborate system of interconnections. In the context of motor function, these interconnections can be understood as disinhibiting, or “releasing the brakes,” on frontal motor action plans: The basal ganglia detect appropriate contexts for performing motor actions and enable the frontal cortex to execute such actions at the appropriate time. We build on this idea in the domain of working memory through the use of computational neural network models of this circuit. In our model, the frontal cortex exhibits robust active maintenance, whereas the basal ganglia contribute a selective, dynamic gating function that enables frontal memory representations to be rapidly updated in a task-relevant manner. We apply the model to a novel version of the continuous performance task that requires subroutine-like selective working memory updating and compare and contrast our model with other existing models and theories of frontal-cortex-basal-ganglia interactions.     

Modeling individual differences in working memory performance: a source activation account

Working memory resources are needed for processing and maintenance of information during cognitive tasks. Many models have been developed to capture the effects of limited working memory resources on performance. However, most of these models do not account for the finding that different individuals show different sensitivities to working memory demands, and none of the models predicts individual subjects' patterns of performance. We propose a computational model that accounts for differences in working memory capacity in terms of a quantity called source activation , which is used to maintain goal-relevant information in an available state. We apply this model to capture the working memory effects of individual subjects at a fine level of detail across two experiments. This, we argue, strengthens the interpretation of source activation as working memory capacity.     

Modeling working memory: An interference model of complex span

This article introduces a new computational model for the complex-span task, the most popular task for studying working memory. SOB-CS is a two-layer neural network that associates distributed item representations with distributed, overlapping position markers. Memory capacity limits are explained by interference from a superposition of associations. Concurrent processing interferes with memory through involuntary encoding of distractors. Free time in-between distractors is used to remove irrelevant representations, thereby reducing interference. The model accounts for benchmark findings in four areas: (1) effects of processing pace, processing difficulty, and number of processing steps; (2) effects of serial position and error patterns; (3) effects of different kinds of item-distractor similarity; and (4) correlations between span tasks. The model makes several new predictions in these areas, which were confirmed experimentally.     

Irrelevant sensory stimuli interfere with working memory storage: Evidence from a computational model of prefrontal neurons

The encoding of irrelevant stimuli into the memory store has previously been suggested as a mechanism of interference in working memory (e.g., Lange & Oberauer, Memory, 13, 333–339, 2005; Nairne, Memory & Cognition, 18, 251–269, 1990). Recently, Bancroft and Servos (Experimental Brain Research, 208, 529–532, 2011) used a tactile working memory task to provide experimental evidence that irrelevant stimuli were, in fact, encoded into working memory. In the present study, we replicated Bancroft and Servos’s experimental findings using a biologically based computational model of prefrontal neurons, providing a neurocomputational model of overwriting in working memory. Furthermore, our modeling results show that inhibition acts to protect the contents of working memory, and they suggest a need for further experimental research into the capacity of vibrotactile working memory.     

A computational model of frontal lobe dysfunction: working memory and the Tower of Hanoi task

A symbolic computer model, employing the perceptual strategy, is presented for solving Tower of Hanoi problems. The model is calibrated—in terms of the number of problems solved, time taken, and number of moves made—to the performance of 20 normal subjects. It is then "lesioned" by increasing the decay rate of elements in working memory to model the performance of 20 patients with lesions to the prefrontal cortex. The model captures both the main effects of subject groups (patients and normal controls) performance, and the subject groups (patients and normal controls) by problem difficulty interactions. This leads us to support the working memory hypothesis of frontal lobe functions, but for a narrow range of problems.     

Concurrent guidance of attention by multiple working memory items: Behavioral and computational evidence

During visual search, task-relevant representations in visual working memory (VWM), known as attentional templates, are assumed to guide attention. A current debate concerns whether only one (Single-Item-Template hypothesis; SIT) or multiple (Multiple-Item-Template hypothesis; MIT) items can serve as attentional templates simultaneously. The current study was designed to test these two hypotheses. Participants memorized two colors, prior to a visual-search task in which the target and the distractor could match or not match the colors held in VWM. Robust attentional guidance was observed when one of the memory colors was presented as the target (reduced response times (RTs) on target-match trials) or the distractor (increased RTs on distractor-match trials). We constructed two drift-diffusion models that implemented the MIT and SIT hypotheses, which are similar in their predictions about overall RTs, but differ in their predictions about RTs on individual trials. Critically, simulated RT distributions and error rates revealed a better match of the MIT hypothesis to the observed data than the SIT hypothesis. Taken together, our findings provide behavioral and computational evidence for the concurrent guidance of attention by multiple items in VWM.

     

Conceptual similarity effects on working memory in sentence contexts: Testing a theory of anaphora

The degree of semantic similarity between an anaphoric noun phrase (e.g., the bird) and its antecedent (e.g., a robin) is known to affect the anaphor resolution process, but the mechanisms that underlie this effect are not known. One proposal (Almor, 1999) is that semantic similarity triggers interference effects in working memory and makes two crucial assumptions: First, semantic similarity impairs working memory just as phonological similarity does (e.g., Baddeley, 1992), and, second, this impairment interferes with processes of sentence comprehension. We tested these assumptions in two experiments that compared recall accuracy between phonologically similar, semantically similar, and control words in sentence contexts. Our results do not provide support for Almor's claims: Phonological overlap decreased recall accuracy in sentence contexts, but semantic similarity did not. These results shed doubt on the idea that semantic interference in working memory is an underlying mechanism in anaphor resolution.     

What's working in working memory: a role for the Central Executive

Two experiments are reported which seek to explore the nature of the Central Executive of Baddeley's (Baddeley and Hitch, 1974) model of working memory. Experiment 1 uses a dual task methodology to compare performance on two storage tasks, making use of the two slave systems, the Phonological Loop (PL) and the Visuo-Spatial Scratch Pad (VSSP). Results suggest that storage can be thought of as based on the use of slave system independent resources. Experiment 2 uses a similar approach to compare performance on 2 processing tasks, again one PL based, the other VSSP based. These results suggest that processing is based on a central, shared set of resources. The two experiments support both a multiple resources model of working memory and more specifically identify a role for the Central Executive.     

Analogous mechanisms of selection and updating in declarative and procedural working memory: Experiments and a computational model

The article investigates the mechanisms of selecting and updating representations in declarative and procedural working memory (WM). Declarative WM holds the objects of thought available, whereas procedural WM holds representations of what to do with these objects. Both systems consist of three embedded components: activated long-term memory, a central capacity-limited component for building structures through temporary bindings, and a single-element focus of attention. Five experiments test the hypothesis of analogous mechanisms in declarative and procedural WM, investigating repetition effects across trials for individual representations (objects and responses) and for sets (memory sets and task sets), as well as set-congruency effects. Evidence for analogous processes was obtained from three phenomena: (1) Costs of task switching and of list switching are reduced with longer preparation interval. (2) The effects of task congruency and of list congruency are undiminished with longer preparation interval. (3) Response repetition interacts with task repetition in procedural WM; here we show an analogous interaction of list repetition with item repetition in declarative WM. All three patterns were reproduced by a connectionist model implementing the assumed selection and updating mechanisms. The model consists of two modules, an item-selection module selecting individual items from a memory set, or responses from a task set, and a set-selection module for selecting memory sets or task sets. The model codes the matrix of binding weights in the item-selection module as a pattern of activation in the set-selection module, thereby providing a mechanism for building chunks in LTM, and for unpacking them as structures into working memory.     

A capacity theory of comprehension: individual differences in working memory.

A theory of the way working memory capacity constrains comprehension is proposed. The theory proposes that both processing and storage are mediated by activation and that the total amount of activation available in working memory varies among individuals. Individual differences in working memory capacity for language can account for qualitative and quantitative differences among college-age adults in several aspects of language comprehension. One aspect is syntactic modularity: The larger capacity of some individuals permits interaction among syntactic and pragmatic information, so that their syntactic processes are not informationally encapsulated. Another aspect is syntactic ambiguity: The larger capacity of some individuals permits them to maintain multiple interpretations. The theory is instantiated as a production system model in which the amount of activation available to the model affects how it adapts to the transient computational and storage demands that occur in comprehension.     

Spatial and object working memory impairments in schizophrenia patients: a Bayesian item-response theory analysis

This study reports evidence that schizophrenia patients are significantly impaired in both spatial and object (shape) working memory. A 3-s delay between exposure and recall of targets was used and Bayesian item-response theory was applied to compensate for the tasks' differential difficulty while simultaneously taking account of missing data from participant attrition. Weaker evidence was found that in schizophrenia both domains are equally impaired on average, that spatial and object working memory appear to be more highly correlated with each other in the schizophrenia population than in the normal population, and that schizophrenia patients show greater variability in spatial than object working memory performance.     

The personalisation method applied to a working memory task: Evidence of long-term working memory effects

Ericsson and Kintsch (1995) proposed that, in situations of expertise, individuals can overcome working memory limitations by using long-term working memory. It allows a greater capacity than working memory thanks to long-term memory encoding and retrieving. To test this characteristic, an adaptation of Daneman and Carpenter's (1980) reading span was used. To operationalise expertise, the personalisation method (Guida & Tardieu, 2005) was employed. In Experiment 1, a personalised group, which read reading span sentences that mentioned familiar locations, was compared to a nonpersonalised group, which read sentences with unfamiliar locations. In Experiment 2, a personalised group, which read reading span sentences with neutral locations, was encouraged to mentally personalise these locations by thinking about known locations. This group was compared to a nonpersonalised group, which was encouraged to think about unknown locations. The personalised groups were expected to store and retrieve information in long-term memory via long-term working memory more easily than the nonpersonalised groups, which had to count massively on working memory. The results showed that personalisation enhanced reading span and confirmed one implication of the long-term working memory theory: high- and low-reading-span differences could also be due to long-term memory retrieval. Finally, these results are interpreted in terms of interaction between working memory size and long-term memory knowledge, showing that participants with a lower reading span benefited more from high domain knowledge than participants with a higher reading span.     

A computational model of fMRI activity in the intraparietal sulcus that supports visual working memory

A computational model was developed to explain a pattern of results of fMRI activation in the intraparietal sulcus (IPS) supporting visual working memory for multiobject scenes. The model is based on the hypothesis that dendrites of excitatory neurons are major computational elements in the cortical circuit. Dendrites enable formation of a competitive queue that exhibits a gradient of activity values for nodes encoding different objects, and this pattern is stored in working memory. In the model, brain imaging data are interpreted as a consequence of blood flow arising from dendritic processing. Computer simulations showed that the model successfully simulates data showing the involvement of inferior IPS in object individuation and spatial grouping through representation of objects’ locations in space, along with the involvement of superior IPS in object identification through representation of a set of objects’ features. The model exhibits a capacity limit due to the limited dynamic range for nodes and the operation of lateral inhibition among them. The capacity limit is fixed in the inferior IPS regardless of the objects’ complexity, due to the normalization of lateral inhibition, and variable in the superior IPS, due to the different encoding demands for simple and complex shapes. Systematic variation in the strength of self-excitation enables an understanding of the individual differences in working memory capacity. The model offers several testable predictions regarding the neural basis of visual working memory.     

Training the removal of negative information from working memory: A preliminary investigation of a working memory bias modification task

Rumination in depressed adults is associated with a bias toward retaining negative information in working memory. We developed a task designed to modify this cognitive bias by having subjects repeatedly practice removing negative words from working memory, thereby enabling them to retain positive and neutral words. To assess the efficacy of this task, we recruited 60 adults who reported elevated repetitive negative thought (RNT) and randomly assigned them to receive a single administration of either the working memory bias modification (WMBM) task or a control task. Subjects in the WMBM condition exhibited greater reduction in proactive interference for negative information than did those in the control condition. These results suggest that the WMBM task reduces biased retention of negative information in working memory and, thus, may be useful in investigating the possible causal role of this cognitive bias in RNT or depression.