Memory Reorganization Research Articles

Reorganization of associative memory in humans with long-standing hippocampal damage

Conflicting theories have been advanced to explain why hippocampal lesions affect distinct memory domains and spare others. Recent findings in monkeys suggest that lesion-induced plasticity may contribute to the seeming preservation of some of these domains. We tested this hypothesis by investigating visuo-spatial associative memory in two patient groups with similar surgical lesions to the right medial temporal lobe, but different preoperative disease courses (benign brain tumours, mean: 1.8 +/- 0.6 years, n = 5, age: 28.2 +/- 4.0 years; hippocampal sclerosis, mean: 16.8 +/- 1.9 years, n = 9, age: 38.9 +/- 4.1 years). Compared to controls (n = 14), tumour patients showed a significant delay-dependent deficit in memory of colour-location associations. No such deficit was observed in hippocampal sclerosis patients, which appeared to benefit from a compensatory mechanism that was inefficient in tumour patients. These results indicate that long-standing hippocampal damage can yield significant functional reorganization of the neural substrate underlying memory in the human brain. We suppose that this process accounts for some of the discrepancies between results from previous lesion studies of the human medial temporal lobe.

Memory Breakdown

The phenomenon of memory reconsolidation has made people question the traditional view that long-term memories become more stable and resistant to perturbation with time. Reconsolidation indicates that memory change is a continuous process and that change is initiated by retrieval experiences themselves. However, the cellular events and mechanisms underlying this phenomenon have not been clear. Lee et al . provide evidence for degradation of postsynaptic proteins in hippocampal synapses thought to participate in the formation of contextual fear memories. Blockade of this degradation is accompanied by blockade of the retrieval-induced reorganization of the original memories. Thus, reconsolidation is like a breakdown of original memories while new elements are incorporated by new protein synthesis. S.-H. Lee, J.-H. Choi, N. Lee, H.-R. Lee, J.-I. Kim, N.-K. Yu, S.-L. Choi, S.-H. Lee, H. Kim, B.-K. Kaang, Synaptic protein degradation underlies destabilization of retrieved fear memory. Science 319 , 1253-1256 (2008). [Abstract] [Full Text]

Reorganization of Verbal and Nonverbal Memory in Temporal Lobe Epilepsy Due to Unilateral Hippocampal Sclerosis

Purpose: Patients with temporal lobe epilepsy (TLE) due to hippocampal sclerosis (HS) often suffer from material-specific memory impairments. The purpose of this study was to use functional magnetic resonance imaging (fMRI) to study the organization of specific memory functions in these patients.Methods: We report 14 patients with unilateral TLE and HS, and 10 controls, performing an fMRI memory paradigm of word, picture, and face encoding.Results: Compared with controls, patients with left TLE demonstrated less left MTL and greater right MTL activation and patients with right TLE demonstrated less right MTL and greater left MTL activation. Correlations between fMRI activation and memory performance revealed greater activation in the damaged left hippocampus to be correlated with better verbal memory performance in left TLE patients and greater right hippocampal activation to be correlated with better nonverbal memory in right TLE patients. Conversely, greater fMRI activation in the contralateral hippocampus correlated with worse memory performance.Conclusions: Our findings suggest that memory function in unilateral TLE is better when it is sustained by activation within the damaged hippocampus and that reorganization to the undamaged MTL is an inefficient process, incapable of preserving memory function.

Grading the gradient: Evidence for time-dependent memory reorganization in experimental animals

In humans, damage limited to the hippocampus produces temporally graded retrograde amnesia, with relative sparing of remote compared to recent memory. This observation forms the cornerstone of the idea that as memories mature they are reorganized in a time-dependent manner. In this paper, we evaluate evidence for similar time-dependent reorganization in experimental animals. In the majority of behavioral paradigms examined, these studies provide evidence that memories are gradually reorganized over time, with the hippocampus and cortex playing preferential (although not necessarily mutually exclusive) roles in the expression of recent and remote memory, respectively. This pattern is not observed in all tasks, however. For example, in the water maze hippocampal activity is always necessary for memory expression. Identifying situations when the hippocampus is, and is not, required for remote memory expression will help us to understand hippocampal contributions to memory, and, more generally, whether changes in memory organization lead to qualitative changes in the nature of memory.

Sleep reorganizes declarative memories from hippocampal towards cortical storing sites

Declarative memory consolidation is supposed to rely on a time-dependent process that gradually incorporates recently acquired hippocampal memory traces into cortical long-term knowledge structures. Besides the mere passage of time sleep appears to provide central nervous conditions that facilitate declarative memory consolidation. However, whether sleep also contributes to the reorganization of declarative memories within hippocampal-neocortical networks is unknown. Here we show that sleep makes memories less dependent on the hippocampus while the anterior cingulate and superior temporal cortex become significantly more engaged. Moreover, we found a negative correlation between retrieval-related brain activity in the hippocampus and both of these cortical regions selectively after sleep. Whereas these differential patterns of neuronal activity were related to comparable levels of memory performance additional experiments revealed that only sleep protects the initially vulnerable memory traces against behavioral interference. Our results indicate that sleep consolidates declarative memories by reorganizing them from temporary hippocampal towards cortical long-term storing sites.

Simple conditions for forming triangular grids

We have used simple learning rules to study how firing maps containing triangular grids—as found in in vivo experiments—can be developed by Hebbian means in realistic robotic simulations. We started from typical non-local postrhinal neuronal responses. We found that anti-Hebbian weight pruning can develop triangular grids under certain conditions. Experimental evidences and the present study suggest that within this model, whitening is a bottom-up process, whereas weight pruning and possibly the non-linear extension of whitening to bottom-up information maximization are guided by top-down influences that reorganize entorhinal responses. We connect our model to the computational model of the entorhinal–hippocampal region of Lőrincz and Buzsáki. In the joined model, the hippocampus is the origin of response reorganization. The joined model may provide insights for memory reorganization guided by hippocampal supervision.

Frontoparietal activation during delayed visuospatial recall in patients with epilepsy due to hippocampal sclerosis

We hypothesized that brain activation during encoding and retrieval of visual material differed between epilepsy patients with hippocampal sclerosis (HS) and healthy controls. Eleven patients with epilepsy and HS and nine age- and education-matched control subjects were tested during functional MRI recording. A three-block design for visuospatial memory encoding and retrieval and an interference interval longer than 1 minute without memory tasks were used. All subjects revealed parietal, occipital, and prefrontal activation patterns during encoding. Interference revealed parietal more than occipital activation, whereas retrieval revealed asymmetrical frontal and parietal activation. Patients demonstrated a relative increase in occipitoparietal versus frontal cortical activation as compared with controls. Memory performance did not differ between patients and controls. The increased activation in occipitoparietal versus frontal areas in the patients suggests cortical reorganization of visuospatial recognition memory in epilepsy patients with HS. The study is limited by other factors that may contribute to the results, for example, antiepileptic drugs, effects of greater cognitive effort allocated in patients than controls, and possibly subclinical epileptic activity. However, normal visuospatial memory performance in our patients with HS suggests successful network plasticity.

Fast track to the medial prefrontal cortex

Studies of brain-damaged patients have provided remarkable insights into how our memories are organized (1). In particular, these studies have established that memory consolidation is a dynamic process involving gradual (but quite dramatic) reorganization of the circuits supporting memory over time (2, 3). A pioneer in this field was the French psychologist Theodule Ribot. In the late part of the 19th century, Ribot described how memory loss after brain insult tended to be related to the age of the memory: the effect on more recent memories was typically greater than that on older (or more remote) memories (4). Later studies established that damage to the medial temporal lobe (5), and more specifically the hippocampus (1), is responsible for this typical graded amnesia. This Ribot gradient has suggested that the hippocampus is only temporarily involved in the storage and recall of certain types of memory, and that these functions must be subsumed by extrahippocampal structures as memories mature (2, 6). Hippocampal damage in experimental animals also preferentially disrupts recent memory. Such examples, from mice to monkeys, suggest that time-dependent memory reorganization is an evolutionarily conserved process for memory consolidation (3). However, differences do exist between human and experimental animal studies, and these differences have made it difficult to come up with a unified mechanistic account of memory consolidation. For example, the lengths of the gradients vary dramatically from days or weeks (for example, in rodents) to years (in humans), suggesting that the rate of reorganization may vary greatly across species. Second, although studies in experimental animals have begun to successfully identify extrahippocampal brain regions that support remote memory recall, the evidence from human studies of remote memory has been much less consistent, and it has even been suggested that autobiographical memories may never become truly independent of the …

Memory reorganization in adult brain: observations in three patients with temporal lobe epilepsy

We present three patients with left-sided temporal lobe epilepsy who exhibited preoperatively a neuropsychological pattern characteristic for interhemispheric language transfer (marked nonverbal memory deficits, relatively preserved verbal memory and language performance). The Wada test indicated atypical language dominance in two patients, but one patient was clearly left hemispheric language dominant. All patients showed a marked recovery of nonverbal memory after left-sided surgery. Results are discussed with respect to memory transfer and plasticity for memory functions in the adult brain.

Remote Reference Counting: Distributed Garbage Collection with Low Communication and Computation Overhead

Automatic storage management is important in distributed programming environments because programmer controlled reclamation is highly prone to errors. The main disadvantage of existing memory reclamation schemes is their high communication cost, a cost that is proportional to the number of pointer operations. We present a novel, efficient distributed garbage collection (GC) algorithm called Remote Reference Counting (RRC). This algorithm has small network traffic and computation overheads, essentially eliminating all communication when there is no active collection. The efficiency is provided by locality of RRC computations. RRC sends an average of only two to three messages from every node in order to reach global consensus on reclamation of any object. Moreover, no GC messages are sent as a result of a pointer operation. Messages for different objects can be combined and piggybacked on non-GC messages, trading promptness of garbage reclamation for additional communication efficiency. The low cost of local garbage identification is provided by a reference counting strategy which is independent of the heap size. RRC works correctly for asynchronous environments where messages may experience arbitrary delays on the way to their destinations. In addition, when applied in granularity of pages, the communication and memory overhead is not inflated when the average allocation size is small, and the memory reorganization required due to the GC operations is simplified. RRC was implemented as a WIN32 library on Windows-NT on top of a distributed shared memory system called millipage. We confirm RRC efficiency by providing performance results measured on a set of widely accepted benchmarks.

Mechanisms of memory reorganization during retrieval of acquired behavioral experience in chicks: the effects of protein synthesis inhibition in the brain.

According to current concepts, memory can be disrupted by administration of protein synthesis inhibitors over a relatively short time period before and after learning. However, data have been obtained indicating that protein synthesis inhibitors can induce amnesia when given long after learning if administration is performed in reminder conditions, i.e., when the animal is presented with one of the environmental components which previously formed the learning situation. The aim of the present work was to confirm the possibility of inducing memory disruption in chicks at late post-learning stages by administering the protein synthesis inhibitor cycloheximide in association with a reminder procedure. Day-old chicks were trained to perform a standard passive avoidance task. Chicks were given cycloheximide (20 microg, intracerebrally) 5 min before the reminder procedure, which was performed 2, 24, or 48 h after training. Testing was conducted 0.5, 1, 3, 24, and 48 h after the reminder. Administration of cycloheximide in association with the reminder procedure induced the development of temporary amnesia, whose duration gradually decreased as the interval between training and reminding increased. These data led to the hypothesis that a memory reactivated by a reminder undergoes a process of reorganization and reconsolidation, which depends on the synthesis of new proteins. The quenching of the ability of protein synthesis inhibition during the reminder to disrupt memory demonstrates the existence of a gradual process resulting in consolidation of memory between 2 and 48 h of learning.

Reorganization of verbal memory and language: a case of dissociation.

Left-to-right reorganization of verbal memory following early left hemisphere damage has been reported in patients whose expressive language is governed by the right hemisphere. We present a case in which verbal memory performance was intact, despite severe left mesial temporal damage, and despite aphasia on left internal carotid sodium amytal ablation. The distribution and degree of left mesial temporal damage was assessed visually and quantitatively on MRI. These findings raise the possibility that verbal memory may shift to the language-nondominant hemisphere as a result of early left mesial temporal damage.

Further evidence for hierarchical chunking in rat spatial memory.

In Experiment 1, rats were given a test to determine the order of preference among 3 types of food. Two groups of rats then were trained on a 12-arm radial maze in Experiment 2, with the 3 foods placed in fixed-arm locations for 1 group and in locations that varied randomly across sessions for the other group. The results replicated those of Dallal and Meck (1990) by showing faster learning and more clustering of arm choices by food type in the fixed-locations group than in the random-locations group. Two further experiments were performed to test the chunking hypothesis. Observations of working memory in Experiment 3 and the reorganization of reference memory in Experiment 4 both supported the chunking hypothesis by showing superior spatial memory and arm chunking by food type when chunk integrity was maintained than when it was compromised.

Reactivation of an old memory during sleep and wakefulness

In a series of four experiments, the effects of the reactivation of an old memory in different states of vigilance were investigated. Rats were trained to run in a six-unit spatial discrimination maze for food reward with low-level electrical stimulation delivered to the ear when they entered blind alleys. Twenty-five days after training, electrotactile stimulation was given to animals as a reactivating treatment just before a retention test. When the cuing treatment was given during wakefulness, retention performance was improved. When it was given in slow-wave sleep, it had no effect. When it was given in paradoxical sleep, performance was poorer. The use of control groups permitted an assessment of the specificity of the cuing effect in each case: Pretest presentation of the same stimulus did not modify performance when it had no cue value. Together, these results show that the effect of the reactivation of an old memory depends upon the vigilance state in which reactivation takes place. They suggest that reorganization of an old memory could be accomplished during paradoxical sleep.

Differential effects of several retrieval cues over time: Evidence for time-dependent reorganization of memory

The retention performance following partial training (15 trials) of a brightness-discrimination avoidance task has been shown to fluctuate over time, with a drop in performance 1 h after training (Kamin effect), a long-term spontaneous improvement (LTSI) after 3 days, and long-term spontaneous forgetting after 21 days. The purpose of this paper was to determine if these time-dependent modulations of retention performance reflect time-dependent modifications in the organization of the attributes that constitute the memory trace. We studied the relative effectiveness of several pretest cuings on retention performance when they were delivered just before a retention test occurring 1 h, 3 days, or 21 days following initial training. Some cuing treatments were related to a particular training event (conditioned or unconditioned stimulus, experimental context). Compound cuings, composed of two or more training events, were also studied to test a possible additive effect between retrieval cues. To demonstrate time-dependent modifications in the memory trace, a differential effectiveness over time was expected for at least some retrieval cues. The results show that cuing may compensate for performance deficits (1 h or 21 days), but does not further enhance performance when spontaneously improved (3 days). No detectable additive effects of retrieval cues were obtained. However, these results provide information that suggests a time-dependent effectiveness for some cuing treatments. In the last experiment, we investigated this possibility by studying the effects of each of these treatments when delivered after either a 1-h or a 21-day retention interval. The results confirm a time-dependent decrease in effectiveness of a pretest exposure to the CS and an increase in effectiveness over time of a pretest exposure to the experimental context or to a well-ordered sequence of events. Such differential retrievability of a partially learned episode according to both the nature of the cues and the length of the retention interval suggests a time-dependent reorganization of memory attributes.

Memory Development and Neurophysiology: Accomplishments and Limitations

The review of Teyler and Fountain offers developmental psychologists an update on the status of neurophysiological theorizing and findings related to memory and learning. Recent findings of LTP and of localization of different types of memory in different brain areas have potential for enriching our understanding of memory development. However, we note several limitations in Teyler and Fountain's presentation in that they do not: distinguish between learning and memory, nor between storage and retrieval; address the role of knowledge-based or top-down influences in memory and learning; employ concepts that can accommodate such developmental phenomena as stages in the hierarchical reorganization of memory. We conclude that even when the age-old search for the "engram" is accomplished, these issues will remain, and that different levels of neural modeling will be required to accommodate them. It is important for developmental psychologists and neuroscientists to maintain communication for the purpose of mutual refinement of models as their knowledge bases continue to grow.

Memory Development and Neurophysiology: Accomplishments and Limitations

The review of Teyler and Fountain offers developmental psychologists an update on the status of neurophysiological theorizing and findings related to memory and learning. Recent findings of LTP and of localization of different types of memory in different brain areas have potential for enriching our understanding of memory development. However, we note several limitations in Teyler and Fountain's presentation in that they do not: distinguish between learning and memory, nor between storage and retrieval; address the role of knowledge-based or top-down influences in memory and learning; employ concepts that can accommodate such developmental phenomena as stages in the hierarchical reorganization of memory. We conclude that even when the age-old search for the "engram" is accomplished, these issues will remain, and that different levels of neural modeling will be required to accommodate them. It is important for developmental psychologists and neuroscientists to maintain communication for the purpose of mutual refinement of models as their knowledge bases continue to grow.

Maintaining organization in a dynamic long-term memory

As new unanticipated items are added to a memory, it must be able to reorganize itself, integrating the new items into its structure. The reorganization process must maintain the memory's structure and also build up the knowledge retrieval strategies need to search that structure. This study will present an algorithm for knowledge-based memory reorganization. Included in that algorithm are processes for directed generalization and generalization refinement. A fact retrieval system called CYRUS which uses the algorithm is also presented. Conclusions are drawn about maintaining accessibility in a conceptual memory, organizing generalized knowledge with respect to specific knowledge, and expected retrieval failures due to change over time in the memory's organization.

Maintaining Organization in a Dynamic Long‐Term Memory*

As new unanticipated items are added to a memory, it must be able to reorganize itself, integrating the new items into its structure. The reorganization process must maintain the memory's structure and also build up the knowledge retrieval strategies need to search that structure. This study will present an algorithm for knowledge‐based memory reorganization. Included in that algorithm are processes for directed generalization and generalization refinement. A fact retrieval system called CYRUS which uses the algorithm is also presented. Conclusions are drawn about maintaining accessibility in a conceptual memory, organizing generalized knowledge with respect to specific knowledge, and expected retrieval failures due to change over time in the memory's organization.

Object salience and code separation in picture naming

Picture naming requires access to three kinds of codes in permanent memory: visual-perceptual, semantic, and lexical. When a picture is named a second time, there is a substantial latency drop. This is assumed to reflect a limited but specific reorganization of memory called “update.” Research was designed to explore three questions: (1) Are the code systems independent? (2) Does update occur in the visual system? If so, what proportion of update is visual? (3) How does the salience of objects (i.e., frequency, ecological importance) determine the magnitude of the latency drop? The naming latencies for pictures previously processed only visually were compared with pictures renamed and those not previously presented. Pictures were calibrated on one index of salience (uncertainty). Results supported the conclusion that the visual system is at least partially independent of the others. Update occurs in the visual system; visual-node priming may be responsible for approximately one-third of the total latency drop. Finally, visual update appears insensitive to object salience, whereas the nonvisual components primed, lexical and/or semantic, interact with salience.