Computer simulations of global memory models

Computer simulation models of memory—including the so-called global memory models—have had and continue to have a profound impact on current memory research and theory. Unfortunately, no memory textbooks published before 1998 present these models to students. The arguments for and against these types of models are summarized along with reasons why the models should be taught at both the undergraduate and graduate levels. A set of freely available computer programs that can facilitate classroom presentation is then briefly described.     

Testing global memory models using ROC curves.

Global memory models are evaluated by using data from recognition memory experiments. For recognition, each of the models gives a value of familiarity as the output from matching a test item against memory. The experiments provide ROC (receiver operating characteristic) curves that give information about the standard deviations of familiarity values for old and new test items in the models. The experimental results are consistent with normal distributions of familiarity (a prediction of the models). However, the results also show that the new-item familiarity standard deviation is about 0.8 that of the old-item familiarity standard deviation and independent of the strength of the old items (under the assumption of normality). The models are inconsistent with these results because they predict either nearly equal old and new standard deviations or increasing values of old standard deviation with strength. Thus, the data provide the basis for revision of current models or development of new models.     

Implementation intention encoding does not automatize prospective memory responding

An implementation intention encoding, one that specifies the concrete situation that is appropriate for initiating an intended action and links that situational cue to the intended action, has been shown to improve prospective memory. One proposed mechanism is that implementation intentions create automatized prospective remembering. This view anticipates that implementation intentions should prevent prospective memory decline in highly cognitively demanding situations. Contrary to this expectation, although implementation intention encoding enhanced prospective memory (Experiments 1 and 2), implementation-intention encoding did not buffer against significant prospective memory decline in high-cognitive-demand conditions (Experiments 1–3), and in Experiment 3, implementation intention encoding produced lower levels of prospective memory performance than did behavioral practice in the high-cognitive-demand situation. We suggest that although implementation intentions may stimulate a strong associative encoding (between an anticipated environmental cue and an intended action), that encoding does not support a completely automatized prospective memory response.     

Time course of item and associative information: implications for global memory models

The time course of availability of associative and item information was examined by using a response signal procedure. Associative information discriminates between a studied pair of words and a pair with words from two different studied pairs. Item information is sufficient to discriminate between a studied pair and a pair not studied. In two experiments, discriminations that require associative information are delayed relative to those based on item information. Two additional experiments discount alternative explanations in terms of the time to encode the test items or task strategies. Examination of the global memory models of Gillund and Shiffrin (1984), Hintzman (1988), and Murdock (1982) shows that the models treat item and associative information inseparably. Modifications to these models which can produce separate contributions for item and associative information do not predict any difference in their availability. Two possible mechanisms for the delayed availability of associative information are considered: the involvement of recall in recognition and the time required to form a compound cue.     

Understanding memory impairment with memory models and hierarchical Bayesian analysis

The study of human episodic memory is a topic that interests cognitive and mathematical psychologists as well as clinicians interested in the diagnosis and assessment of Alzheimer’s disease and related disorders (ADRD). In this paper, we use simple cognitive models for the recognition and recall tasks typically applied in clinical assessments of ADRD to study memory performance in ADRD patients. Our models make use of hierarchical Bayesian methods as a way to model individual differences in patient performance and to facilitate the modeling of performance changes that occur during multiple recall tasks. We show how the models are able to account for different aspects of patient performance, and also discuss some of the predictive capabilities of the model. We conclude with a discussion on the scope to improve on our results by discussing the link between memory theory in psychology and clinical practice.     

Distinctiveness models of memory and absolute identification: evidence for local, not global, effects

Many models of memory assume that the probability of remembering an item is related to how distinctive that item is relative to all the other items in the set, with no distinction made between the contributions of near or far items. These “global” distinctiveness models do well in accounting for the ubiquitous serial position effects observed in numerous memory paradigms, including absolute identification. Here, we provide experimental confirmation of Bower's (1971) suggestion that, contrary to a fundamental prediction of global distinctiveness models, midseries items can be more discriminable than their immediate neighbours. We show that such data are consistent with a revised distinctiveness account in which the factor affecting discrimination performance is primarily the distinctiveness of an item relative to its close neighbours.     

Distinctiveness models of memory and absolute identification: Evidence for local,not global,effects

Many models of memory assume that the probability of remembering an item is related to how distinctive that item is relative to all the other items in the set, with no distinction made between the contributions of near or far items. These “global” distinctiveness models do well in accounting for the ubiquitous serial position effects observed in numerous memory paradigms, including absolute identification. Here, we provide experimental confirmation of Bower's (1971) suggestion that, contrary to a fundamental prediction of global distinctiveness models, midseries items can be more discriminable than their immediate neighbours. We show that such data are consistent with a revised distinctiveness account in which the factor affecting discrimination performance is primarily the distinctiveness of an item relative to its close neighbours.     

Complexity of symbolic representation in working memory of Transformer correlates with the complexity of a task

Even though Transformers are extensively used for Natural Language Processing tasks, especially for machine translation, they lack an explicit memory to store key concepts of processed texts. This paper explores the properties of the content of symbolic working memory added to the Transformer model decoder. Such working memory enhances the quality of model predictions in machine translation task and works as a neural-symbolic representation of information that is important for the model to make correct translations. The study of memory content revealed that translated text keywords are stored in the working memory, pointing to the relevance of memory content to the processed text. Also, the diversity of tokens and parts of speech stored in memory correlates with the complexity of the corpora for machine translation task.     

The influence of symbolic literacy on memory: testing Plato's hypothesis.

The present study examined the influence of the production of external symbols on memory strategies. Plato hypothesized that dependency on writing as an external memory store would be deterimental to memory. Three experiments were conducted to explore this hypothesis. Participants played Concentration, a memory game where players must find matching pairs of cards placed face down in an array. Participants were allowed to make notes to aid their performance under some experimental conditions, while under other conditions they could not. In Experiments 1 and 2, the unexpected removal of participants' notes revealed that the performance benefit was due to notes acting as a form of external memory storage, rather than as an aid to encoding information in memory. Experiment 3 qualified these findings by demonstrating that the identity of each card was retained in memory, while the location of each card tended to be stored in the participants' external notations. These data suggest a modified interpretation of Plato's hypothesis in that symbolic literacy may change how we remember information. Rather than storing all information in memory, we only have to retain the information necessary to use the much larger storage capacity of the external system. Thus, the introduction of external symbols allows for a change in how memory is adaptively distributed.     

Animal models of source memory

Source memory is the aspect of episodic memory that encodes the origin (i.e., source) of information acquired in the past. Episodic memory (i.e., our memories for unique personal past events) typically involves source memory because those memories focus on the origin of previous events. Source memory is at work when, for example, someone tells a favorite joke to a person while avoiding retelling the joke to the friend who originally shared the joke. Importantly, source memory permits differentiation of one episodic memory from another because source memory includes features that were present when the different memories were formed. This article reviews recent efforts to develop an animal model of source memory using rats. Experiments are reviewed which suggest that source memory is dissociated from other forms of memory. The review highlights strengths and weaknesses of a number of animal models of episodic memory. Animal models of source memory may be used to probe the biological bases of memory. Moreover, these models can be combined with genetic models of Alzheimer's disease to evaluate pharmacotherapies that ultimately have the potential to improve memory.     

Relative influence of sincere,insincere, and neutral symbolic models

The study investigated how the sincerity of models affects their influence, using a delay-of-reward paradigm. Subjects were 40 fifth and sixth graders (ages 9 and 10 years, respectively), selected for delayed gratification patterns. In the experimental manipulation, sincerity or insincerity was explicitly attributed to the model or, in a neutral-model condition, no information about the model's sincerity was given. Subjects then observed the symbolically presented model's immediatereward choices, opposed to their own preferences. Immediately after treatment, children in all model conditions and in a fourth, no-model control group responded to a second delay-preference test. It was found that the influence of models varied substantially as a function of experimentally attributed sincerity. Insincere models had considerably less effect than sincere and neutral models, who provoked substantial and equivalent changes. Insincere models did not evoke significantly different change than occurred in the control group. A postexperimental inquiry showed the experimental manipulation had affected perceptions of the model's sincerity and liking for the model, with the insincere model rated considerably lower than the sincere and neutral models. The relevance of cognitive appraisal in observational learning was discussed.     

Processing of global and local information in memory

The experiments reported in this paper were designed to test how global and local information are processed by the memory system. When subjects are required to match a given letter with either a previously presented large capital letter or the small capital letters comprising it, (1) responses to the global level (i.e. the big letter) are faster than responses to the local level (i.e. the small letters), and (2) responses to the latter level only are affected by the consistency between the large and the small letters (Experiment 2), a pattern similar to that obtained in perception (Experiment 1). Such results obtain when subjects are required to attend to only one level with a short ISI between the first and second stimulus, but not when a longer ISI is used (Experiment 5) or when subjects are required to attend to both levels at the same time (Experiments 3 and 4). The results are discussed in the light of a model that postulates a temporal precedence of the global information over the local one at the perceptual level.     

Harm that does not hurt: Humour in coping with self-threat

We examined the affective and cognitive impact of humour on coping with self-threat. Research was based on an incongruity concept of humour that specifies humour as a state resulting from appraising an aversive incident as both harmful and as acceptable. An appraisal related procedural priming paradigm was used to induce humour. In Study 1 (N = 41 female students) the impact of humour on positive and negative affect following self-threat was examined. In Study 2 (N = 52 students; 94% women) we investigated the consequences of humour for a self-serving interpretation of failure, the awareness of harm, and subsequent performance. Relative to the control condition, humour increased positive affect, while not exclusively affecting negative affect, and increased the tendency for an external attribution of failure, while harm was clearly recognized. However, humour led to poorer subsequent performance, suggesting that humour may also have its costs.     

Computational models of working memory: putting long-term memory into context

Detailed computational modeling of human memory has typically been aimed at either short-term (working) memory or long-term memory in isolation. However, recent research highlights the importance of interactions between these systems for both item and order information. At the same time, computational models of both systems are beginning to converge onto a common framework in which items are associated with an evolving "context" signal and subsequently compete with one another at recall. We review some of these models, and discuss a common mechanism capable of modelling working memory and its interaction with long-term memory, focussing on memory for verbal sequences.     

Primary and secondary memory: multistore versus process models of memory

Healy, et al. recently distinguished between recency and "penultimate" effects when contrasting recall performance. They argued that the penultimate effect was significantly different from the typical recency effect, which point was taken as support for distinguishing between primary and secondary memory systems. The current study adapted the Healy, et al. methodology to control for the occurrence of intentional learning strategies. 40 participants were asked to recall the names of 42 U.S. Presidents by their term in office (secondary memory) or in a pseudorandom order (primary memory); the recency and penultimate effects were observed for each condition. With intentional learning controlled for, contrast analyses yielded a pattern of performance similar to that observed by Healy, et al.     

Priming effects that span an intervening unrelated word: implications for models of memory representation and retrieval.

Several studies have reported priming effects that span an intervening unrelated word (Davelaar & Coltheart, 1975; Meyer, Schvaneveldt, & Ruddy, 1972). More recently, other investigators have suggested that such relatedness effects are the result of postaccess processes (Gough, Alford, & Holley-Wilcox, 1981; Masson, 1991; Ratcliff & McKoon, 1988). In fact, these investigators claim that when procedures are used that discourage the use of postaccess processes, relatedness effects do not span intervening unrelated words. The present experiments demonstrate reliable relatedness effects with procedures that eliminate postaccess processes. These results are consistent with the notion of spreading activation among local representations in memory. Implications for the issue of local versus distributed representations are discussed.     

Retrieval does not induce reconsolidation of inhibitory avoidance memory

It has been suggested that retrieval during a nonreinforced test induces reconsolidation instead of extinction of the mnemonic trace. Reconsolidation would preserve the original memory from the labilization induced by its nonreinforced recall through a hitherto uncharacterized mechanism requiring protein synthesis. Given the importance that such a process would have in terms of maintaining, as part of the animal behavioral repertoire, a learned response that has been devalued by experience, we analyzed its existence for the memory associated with a one-trial, step-down inhibitory avoidance task (IA), a memory whose consolidation and extinction require protein synthesis in the CA1 region of the dorsal hippocampus (CA1) and involve the participation of the basolateral amygdala (BLA) and entorhinal cortex (ENT). Rats were trained in IA, and 24 h later they were submitted either to a pure reactivation session (retrieval without stepping down), which was unable by itself to initiate extinction of the avoidance response, or to a second training session. Fifteen minutes before or 3 h after either the reactivation or the retraining sessions, animals were infused with the protein synthesis inhibitor anisomycin (ANI) into CA1, BLA, or ENT. Contrary to the prediction of the reconsolidation hypothesis, none of these treatments affected subsequent memory retention. Because reconsolidation is regarded to be a direct consequence of retrieval, one would expect that, when given before a retention test or a pure reactivation session, enhancers of memory expression should permanently improve retention and, therefore, facilitate retrieval both in that and in subsequent sessions. Using two well-known retrieval enhancers, noradrenaline and adrenocorticotropin(1-24), we could not find any evidence suggestive of reconsolidation. Hence, our results indicate that there is no retrieval-induced, protein synthesis-dependent process that would cause reconsolidation of IA memory.     

A cognitive architecture that combines internal simulation with a global workspace

This paper proposes a brain-inspired cognitive architecture that incorporates approximations to the concepts of consciousness, imagination, and emotion. To emulate the empirically established cognitive efficacy of conscious as opposed to non-conscious information processing in the mammalian brain, the architecture adopts a model of information flow from global workspace theory. Cognitive functions such as anticipation and planning are realised through internal simulation of interaction with the environment. Action selection, in both actual and internally simulated interaction with the environment, is mediated by affect. An implementation of the architecture is described which is based on weightless neurons and is used to control a simulated robot.