A Model for Visual Memory Encoding

Memory encoding engages multiple concurrent and sequential processes. While the individual processes involved in successful encoding have been examined in many studies, a sequence of events and the importance of modules associated with memory encoding has not been established. For this reason, we sought to perform a comprehensive examination of the network for memory encoding using data driven methods and to determine the directionality of the information flow in order to build a viable model of visual memory encoding. Forty healthy controls ages 19–59 performed a visual scene encoding task. FMRI data were preprocessed using SPM8 and then processed using independent component analysis (ICA) with the reliability of the identified components confirmed using ICASSO as implemented in GIFT. The directionality of the information flow was examined using Granger causality analyses (GCA). All participants performed the fMRI task well above the chance level (>90% correct on both active and control conditions) and the post-fMRI testing recall revealed correct memory encoding at 86.33±5.83%. ICA identified involvement of components of five different networks in the process of memory encoding, and the GCA allowed for the directionality of the information flow to be assessed, from visual cortex via ventral stream to the attention network and then to the default mode network (DMN). Two additional networks involved in this process were the cerebellar and the auditory-insular network. This study provides evidence that successful visual memory encoding is dependent on multiple modules that are part of other networks that are only indirectly related to the main process. This model may help to identify the node(s) of the network that are affected by a specific disease processes and explain the presence of memory encoding difficulties in patients in whom focal or global network dysfunction exists.     

A Probabilistic Model of Perception

The development of electronic computers has made it possible to transfer to them a number of the "intellectual" functions of man. At present such machines are being employed to model broader and broader regions of man's psychological activity. One of the most important problems in this connection is the building of "perceptive machines," such as reading machines, "the printing phonograph" and so forth. Perceptive machines are a necessary element in the development of automatic equipment to operate in complicated environments.     

A Probabilistic Model of RNA Conformational Space

The increasing importance of non-coding RNA in biology and medicine has led to a growing interest in the problem of RNA 3-D structure prediction. As is the case for proteins, RNA 3-D structure prediction methods require two key ingredients: an accurate energy function and a conformational sampling procedure. Both are only partly solved problems. Here, we focus on the problem of conformational sampling. The current state of the art solution is based on fragment assembly methods, which construct plausible conformations by stringing together short fragments obtained from experimental structures. However, the discrete nature of the fragments necessitates the use of carefully tuned, unphysical energy functions, and their non-probabilistic nature impairs unbiased sampling. We offer a solution to the sampling problem that removes these important limitations: a probabilistic model of RNA structure that allows efficient sampling of RNA conformations in continuous space, and with associated probabilities. We show that the model captures several key features of RNA structure, such as its rotameric nature and the distribution of the helix lengths. Furthermore, the model readily generates native-like 3-D conformations for 9 out of 10 test structures, solely using coarse-grained base-pairing information. In conclusion, the method provides a theoretical and practical solution for a major bottleneck on the way to routine prediction and simulation of RNA structure and dynamics in atomic detail.     

Visual Search Elicits the Electrophysiological Marker of Visual Working Memory

Background

Although limited in capacity, visual working memory (VWM) plays an important role in many aspects of visually-guided behavior. Recent experiments have demonstrated an electrophysiological marker of VWM encoding and maintenance, the contralateral delay activity (CDA), which has been shown in multiple tasks that have both explicit and implicit memory demands. Here, we investigate whether the CDA is evident during visual search, a thoroughly-researched task that is a hallmark of visual attention but has no explicit memory requirements.

Methodology/Principal Findings

The results demonstrate that the CDA is present during a lateralized search task, and that it is similar in amplitude to the CDA observed in a change-detection task, but peaks slightly later. The changes in CDA amplitude during search were strongly correlated with VWM capacity, as well as with search efficiency. These results were paralleled by behavioral findings showing a strong correlation between VWM capacity and search efficiency.

Conclusions/Significance

We conclude that the activity observed during visual search was generated by the same neural resources that subserve VWM, and that this activity reflects the maintenance of previously searched distractors.     

Objective Identification of Functional Units of Short-term Memory

The idea of functional memory units may, without exaggeration, be called the central aspect of any theory concerning the training and organization of memory. It was introduced to distinguish between the nominal elements, i.e., the elements presented for memorization, of the content of a stimulus and their internal representations. The elements of a series presented by an experimenter are often apprehended and remembered as parts of larger units. One functional unit may combine two or more elements of a series to be memorized, and the key to its recall will then be only a part of the nominal stimulus (Miller, 1964; Klatsky, 1978; Underwood, 1963; Mandler, 1967; Bower, 1977; et al.).     

Attribute Pair-Based Visual Recognition and Memory

Background

In the human visual system, different attributes of an object, such as shape, color, and motion, are processed separately in different areas of the brain. This raises a fundamental question of how are these attributes integrated to produce a unified perception and a specific response. This “binding problem” is computationally difficult because all attributes are assumed to be bound together to form a single object representation. However, there is no firm evidence to confirm that such representations exist for general objects.

Methodology/Principal Findings

Here we propose a paired-attribute model in which cognitive processes are based on multiple representations of paired attributes. In line with the model''s prediction, we found that multiattribute stimuli can produce an illusory perception of a multiattribute object arising from erroneous integration of attribute pairs, implying that object recognition is based on parallel perception of paired attributes. Moreover, in a change-detection task, a feature change in a single attribute frequently caused an illusory perception of change in another attribute, suggesting that multiple pairs of attributes are stored in memory.

Conclusions/Significance

The paired-attribute model can account for some novel illusions and controversial findings on binocular rivalry and short-term memory. Our results suggest that many cognitive processes are performed at the level of paired attributes rather than integrated objects, which greatly facilitates the binding problem and provides simpler solutions for it.     

A Part-Based Probabilistic Model for Object Detection with Occlusion

The part-based method has been a fast rising framework for object detection. It is attracting more and more attention for its detection precision and partial robustness to the occlusion. However, little research has been focused on the problem of occlusion overlapping of the part regions, which can reduce the performance of the system. This paper proposes a part-based probabilistic model and the corresponding inference algorithm for the problem of the part occlusion. The model is based on the Bayesian theory integrally and aims to be robust to the large occlusion. In the stage of the model construction, all of the parts constitute the vertex set of a fully connected graph, and a binary variable is assigned to each part to indicate its occlusion status. In addition, we introduce a penalty term to regularize the argument space of the objective function. Thus, the part detection is formulated as an optimization problem, which is divided into two alternative procedures: the outer inference and the inner inference. A stochastic tentative method is employed in the outer inference to determine the occlusion status for each part. In the inner inference, the gradient descent algorithm is employed to find the optimal positions of the parts, in term of the current occlusion status. Experiments were carried out on the Caltech database. The results demonstrated that the proposed method achieves a strong robustness to the occlusion.     

Visual Memory and Models of Human Information-Processing

Interest in the structural organization of memory existed long before an analogy was drawn between the brain and computers, Nevertheless, we have the efforts of proponents of information theory [18, 22, 371 to thank for the foundations of numerous structural models of memory, whose common feature is the notion that there are three linked units of information storage and retention (Figure 1). Thus, the most sophisticated model of this sort, that of R. Atkinson & R. Shiffrin [27, 281 distinguishes: (1) modality-specific sensory registers, (2) short-term (primary) memory, and (3) long-term (secondary) memory.     

Reconstructions of Information in Visual Spatial Working Memory Degrade with Memory Load

1. Download : Download high-res image (296KB)
2. Download : Download full-size image

     

Contrast and Strength of Visual Memory and Imagery Differentially Affect Visual Perception

Visual short-term memory (VSTM) and visual imagery have been shown to modulate visual perception. However, how the subjective experience of VSTM/imagery and its contrast modulate this process has not been investigated. We addressed this issue by asking participants to detect brief masked targets while they were engaged either in VSTM or visual imagery. Subjective experience of memory/imagery (strength scale), and the visual contrast of the memory/mental image (contrast scale) were assessed on a trial-by-trial basis. For both VSTM and imagery, contrast of the memory/mental image was positively associated with reporting target presence. Consequently, at the sensory level, both VSTM and imagery facilitated visual perception. However, subjective strength of VSTM was positively associated with visual detection whereas the opposite pattern was found for imagery. Thus the relationship between subjective strength of memory/imagery and visual detection are qualitatively different for VSTM and visual imagery, although their impact at the sensory level appears similar. Our results furthermore demonstrate that imagery and VSTM are partly dissociable processes.     

Feature-Specific Encoding Flexibility in Visual Working Memory

The current study examined selective encoding in visual working memory by systematically investigating interference from task-irrelevant features. The stimuli were objects defined by three features (color, shape, and location), and during a delay period, any of the features could switch between two objects. Additionally, single- and whole-probe trials were randomized within experimental blocks to investigate effects of memory retrieval. A series of relevant-feature switch detection tasks, where one feature was task-irrelevant, showed that interference from the task-irrelevant feature was only observed in the color-shape task, suggesting that color and shape information could be successfully filtered out, but location information could not, even when location was a task-irrelevant feature. Therefore, although location information is added to object representations independent of task demands in a relatively automatic manner, other features (e.g., color, shape) can be flexibly added to object representations.     

A Zebrafish Model of Diabetes Mellitus and Metabolic Memory

Diabetes mellitus currently affects 346 million individuals and this is projected to increase to 400 million by 2030. Evidence from both the laboratory and large scale clinical trials has revealed that diabetic complications progress unimpeded via the phenomenon of metabolic memory even when glycemic control is pharmaceutically achieved. Gene expression can be stably altered through epigenetic changes which not only allow cells and organisms to quickly respond to changing environmental stimuli but also confer the ability of the cell to "memorize" these encounters once the stimulus is removed. As such, the roles that these mechanisms play in the metabolic memory phenomenon are currently being examined.We have recently reported the development of a zebrafish model of type I diabetes mellitus and characterized this model to show that diabetic zebrafish not only display the known secondary complications including the changes associated with diabetic retinopathy, diabetic nephropathy and impaired wound healing but also exhibit impaired caudal fin regeneration. This model is unique in that the zebrafish is capable to regenerate its damaged pancreas and restore a euglycemic state similar to what would be expected in post-transplant human patients. Moreover, multiple rounds of caudal fin amputation allow for the separation and study of pure epigenetic effects in an in vivo system without potential complicating factors from the previous diabetic state. Although euglycemia is achieved following pancreatic regeneration, the diabetic secondary complication of fin regeneration and skin wound healing persists indefinitely. In the case of impaired fin regeneration, this pathology is retained even after multiple rounds of fin regeneration in the daughter fin tissues. These observations point to an underlying epigenetic process existing in the metabolic memory state. Here we present the methods needed to successfully generate the diabetic and metabolic memory groups of fish and discuss the advantages of this model.     

A Probabilistic Model to Predict Clinical Phenotypic Traits from Genome Sequencing

Genetic screening is becoming possible on an unprecedented scale. However, its utility remains controversial. Although most variant genotypes cannot be easily interpreted, many individuals nevertheless attempt to interpret their genetic information. Initiatives such as the Personal Genome Project (PGP) and Illumina''s Understand Your Genome are sequencing thousands of adults, collecting phenotypic information and developing computational pipelines to identify the most important variant genotypes harbored by each individual. These pipelines consider database and allele frequency annotations and bioinformatics classifications. We propose that the next step will be to integrate these different sources of information to estimate the probability that a given individual has specific phenotypes of clinical interest. To this end, we have designed a Bayesian probabilistic model to predict the probability of dichotomous phenotypes. When applied to a cohort from PGP, predictions of Gilbert syndrome, Graves'' disease, non-Hodgkin lymphoma, and various blood groups were accurate, as individuals manifesting the phenotype in question exhibited the highest, or among the highest, predicted probabilities. Thirty-eight PGP phenotypes (26%) were predicted with area-under-the-ROC curve (AUC)>0.7, and 23 (15.8%) of these were statistically significant, based on permutation tests. Moreover, in a Critical Assessment of Genome Interpretation (CAGI) blinded prediction experiment, the models were used to match 77 PGP genomes to phenotypic profiles, generating the most accurate prediction of 16 submissions, according to an independent assessor. Although the models are currently insufficiently accurate for diagnostic utility, we expect their performance to improve with growth of publicly available genomics data and model refinement by domain experts.