Deep Learning of Orthographic Representations in Baboons

What is the origin of our ability to learn orthographic knowledge? We use deep convolutional networks to emulate the primate''s ventral visual stream and explore the recent finding that baboons can be trained to discriminate English words from nonwords [1]. The networks were exposed to the exact same sequence of stimuli and reinforcement signals as the baboons in the experiment, and learned to map real visual inputs (pixels) of letter strings onto binary word/nonword responses. We show that the networks'' highest levels of representations were indeed sensitive to letter combinations as postulated in our previous research. The model also captured the key empirical findings, such as generalization to novel words, along with some intriguing inter-individual differences. The present work shows the merits of deep learning networks that can simulate the whole processing chain all the way from the visual input to the response while allowing researchers to analyze the complex representations that emerge during the learning process.     

Condensed Representations of Protein Secondary Structure Sequences and Their Application

Abstract

In this paper, we propose a nongraphical representation for protein secondary structures. By counting the frequency of occurrence of all possible four-tuples (i.e., four-letter words) of a protein secondary structure sequence, we construct a set of 3 × 3 matrices for the corresponding protein secondary structure sequence. Furthermore, the leading eigenvalues of these matrices are computed and considered as invariants for the protein secondary structure sequences. To illustrate the utility of our approach, we apply it to a set of real data to distinguish protein structural classes. The result indicates that it can be used to complement the classification of protein secondary structures.     

Manipulating Representations

The present paper proposes a definition for the complex polysemic concepts of consciousness and awareness (in humans as well as in other species), and puts forward the idea of a progressive ontological development of consciousness from a state of ‘childhood’ awareness, in order to explain that humans are not only able to manipulate objects, but also their mental representations. The paper builds on the idea of qualia intended as entities posing regular invariant requests to neural processes, trough the permanence of different properties. The concept of semantic differential introduces the properties of metaphorical qualia as an exclusively human ability. Furthermore this paper proposes a classification of qualia, according to the models–with different levels of abstraction–they are implied in, in a taxonomic perspective. This, in turn, becomes a source of categorization of divergent representations, sign systems, and forms of intentionality, relying always on biological criteria. New emerging image-of-the-world-devices are proposed, whose qualia are likely to be only accessible to humans: emotional qualia, where emotion accounts for the invariant and dominant property; and the qualic self where continuity, combined with the oneness of the self, accounts for the invariant and dominant property. The concept of congruence between different domains in a metaphor introduces the possibility of a general evaluation of truth and falsity of all kinds of metaphorical constructs, while the work of Matte Blanco enables us to classify conscious versus unconscious metaphors, both in individuals and in social organizations.     

The Influence of Direct and Indirect Speech on Mental Representations

Language can be viewed as a set of cues that modulate the comprehender’s thought processes. It is a very subtle instrument. For example, the literature suggests that people perceive direct speech (e.g., Joanne said: ‘I went out for dinner last night’) as more vivid and perceptually engaging than indirect speech (e.g., Joanne said that she went out for dinner last night). But how is this alleged vividness evident in comprehenders’ mental representations? We sought to address this question in a series of experiments. Our results do not support the idea that, compared to indirect speech, direct speech enhances the accessibility of information from the communicative or the referential situation during comprehension. Neither do our results support the idea that the hypothesized more vivid experience of direct speech is caused by a switch from the visual to the auditory modality. However, our results do show that direct speech leads to a stronger mental representation of the exact wording of a sentence than does indirect speech. These results show that language has a more subtle influence on memory representations than was previously suggested.     

The Structure of Microbial Community and Degradation of Diatoms in the Deep Near-Bottom Layer of Lake Baikal

Insight into the role of bacteria in degradation of diatoms is important for understanding the factors and components of silica turnover in aquatic ecosystems. Using microscopic methods, it has been shown that the degree of diatom preservation and the numbers of diatom-associated bacteria in the surface layer of bottom sediments decrease with depth; in the near-bottom water layer, the majority of bacteria are associated with diatom cells, being located either on the cell surface or within the cell. The structure of microbial community in the near-bottom water layer has been characterized by pyrosequencing of the 16S rRNA gene, which has revealed 149 208 unique sequences. According to the results of metagenomic analysis, the community is dominated by representatives of Proteobacteria (41.9%), Actinobacteria (16%); then follow Acidobacteria (6.9%), Cyanobacteria (5%), Bacteroidetes (4.7%), Firmicutes (2.8%), Nitrospira (1.6%), and Verrucomicrobia (1%); other phylotypes account for less than 1% each. For 18.7% of the sequences, taxonomic identification has been possible only to the Bacteria domain level. Many bacteria identified to the genus level have close relatives occurring in other aquatic ecosystems and soils. The metagenome of the bacterial community from the near-bottom water layer also contains 16S rRNA gene sequences found in previously isolated bacterial strains possessing hydrolytic enzyme activity. These data show that potential degraders of diatoms occur among the vast variety of microorganisms in the near-bottom water of Lake Baikal.     

The Development of the Mental Representations of the Magnitude of Fractions

We investigated the development of the mental representation of the magnitude of fractions during the initial stages of fraction learning in grade 5, 6 and 7 children as well as in adults. We examined the activation of global fraction magnitude in a numerical comparison task and a matching task. There were global distance effects in the comparison task, but not in the matching task. This suggests that the activation of the global magnitude representation of fractions is not automatic in all tasks involving magnitude judgments. The slope of the global distance effect increased during early fraction learning and declined by adulthood, demonstrating that the development of the fraction global distance effect differs from that of the integer distance effect.     

Improved 3-D Protein Structure Predictions using Deep ResNet Model

The Protein Journal - Protein Structure Prediction (PSP) is considered to be a complicated problem in computational biology. In spite of, the remarkable progress made by the co-evolution-based...     

Validation of X-ray Crystal Structure Ensemble Representations of SARS-CoV-2 Main Protease by Solution NMR Residual Dipolar Couplings

Considerable debate has focused on whether sampling of molecular dynamics trajectories restrained by crystallographic data can be used to develop realistic ensemble models for proteins in their natural, solution state. For the SARS-CoV-2 main protease, M^pro, we evaluated agreement between solution residual dipolar couplings (RDCs) and various recently reported multi-conformer and dynamic-ensemble crystallographic models. Although Phenix-derived ensemble models showed only small improvements in crystallographic R_free, substantially improved RDC agreement over fits to a conventionally refined 1.2-Å X-ray structure was observed, in particular for residues with above average disorder in the ensemble. For a set of six lower resolution (1.55–2.19 Å) M^pro X-ray ensembles, obtained at temperatures ranging from 100 to 310 K, no significant improvement over conventional two-conformer representations was found. At the residue level, large differences in motions were observed among these ensembles, suggesting high uncertainties in the X-ray derived dynamics. Indeed, combining the six ensembles from the temperature series with the two 1.2-Å X-ray ensembles into a single 381-member “super ensemble” averaged these uncertainties and substantially improved agreement with RDCs. However, all ensembles showed excursions that were too large for the most dynamic fraction of residues. Our results suggest that further improvements to X-ray ensemble refinement are feasible, and that RDCs provide a sensitive benchmark in such endeavors. Remarkably, a weighted ensemble of 350 PDB M^pro X-ray structures provided slightly better cross-validated agreement with RDCs than any individual ensemble refinement, implying that differences in lattice confinement also limit the fit of RDCs to X-ray coordinates.     

Native American Representations: First Encounters, Distorted Images, and Literary Appropriations.; Native American Literature: Boundaries & Sovereignties.

Native American Representations: First Encounters, Distorted Images, and Literary Appropriations. Gretchen Bataille. ed. Lincoln: University of Nebraska Press, 2001. 252 pp.
Native American Literature: Boundaries & Sovereignties. Kathryn W. Shanley. ed. Vashon Island, WA: Paradoxa, 2002. 312 pp.     

Deep Subsurface Microbial Biomass and Community Structure in Witwatersrand Basin Mines

The extreme environments of South Africa mines were investigated to determine microbial community structure and biomass in the deep subsurface. These community parameters were determined using phospholipid fatty acid (PLFA) technique. Air, water and rock samples were collected from several levels and shafts in eight different mines. Biomass estimates ranged over nine orders of magnitude. Biofilm samples exhibited the highest biomass with quantities ranging from 10 ³ to 10 ⁷ pmol PLFA g ^?1 . Rock samples had biomass ranging from 10 ³ to 10 ⁶ pmol PLFA g ^?1 . Mine service waters and rock fracture waters had biomass estimates ranging from 10 ⁰ to 10 ⁶ pmol PLFA L ^?1 . Air samples biomass values ranged from 10 ^?2 to 10 ⁰ pmol PLFA L ^?1 . The biomass estimates were similar to those estimates for other deep subsurface sites. Redundancy analysis of the PLFA profiles distinguished between the sample types, where signature lipid biomarkers for aerobic and anaerobic prokaryotes, sulfate-and metal-reducing bacteria were associated with biofilms. Rock samples were enriched in 18:1 ω 9 c , 18:2 ω 6, br17:1s and br18:1s, which are indicative of microeukaryotes and metal- reducing bacteria. Air samples were enriched with 22:0, 17:1, 18:1, and a polyunsaturated fatty acid. Service waters had monounsaturated fatty acids. Fracture waters contained i17:0 and 10Me18:0 which indicated gram-positive and other anaerobic bacteria. When the fracture and service water sample PLFA responses to changes in environmental parameters of temperature, pH, and anion concentrations were analyzed, service waters correlated with higher nitrate and sulfate concentrations and the PLFAs 18:1 ω 7 c and 16:1 ω 7 c . Dreifontein shaft 5 samples correlated with chloride concentrations and terminally branched saturated fatty acids and branched monounsaturated fatty acids. Kloof, Tau Tona, and Merriespruit fracture waters aligned with temperature and pH vectors and 18:0, 20:0 and 22:6 ω 3. The redundancy analysis provided a robust method to understand the PLFA responses to changes in environmental parameters.     

Resolving Anatomical and Functional Structure in Human Brain Organization: Identifying Mesoscale Organization in Weighted Network Representations

Human brain anatomy and function display a combination of modular and hierarchical organization, suggesting the importance of both cohesive structures and variable resolutions in the facilitation of healthy cognitive processes. However, tools to simultaneously probe these features of brain architecture require further development. We propose and apply a set of methods to extract cohesive structures in network representations of brain connectivity using multi-resolution techniques. We employ a combination of soft thresholding, windowed thresholding, and resolution in community detection, that enable us to identify and isolate structures associated with different weights. One such mesoscale structure is bipartivity, which quantifies the extent to which the brain is divided into two partitions with high connectivity between partitions and low connectivity within partitions. A second, complementary mesoscale structure is modularity, which quantifies the extent to which the brain is divided into multiple communities with strong connectivity within each community and weak connectivity between communities. Our methods lead to multi-resolution curves of these network diagnostics over a range of spatial, geometric, and structural scales. For statistical comparison, we contrast our results with those obtained for several benchmark null models. Our work demonstrates that multi-resolution diagnostic curves capture complex organizational profiles in weighted graphs. We apply these methods to the identification of resolution-specific characteristics of healthy weighted graph architecture and altered connectivity profiles in psychiatric disease.     

Representations of the human circulatory system

There is no agreement about the robustness of intuitive representations of the circulatory system and their susceptibility to change by instruction. In this paper, we analyse to what extent students with varying degrees of biology instruction and different ages (High School Health Science and Social Science students and first and final year University Psychology students) have been able to change their intuitive beliefs into scientific representations of the workings of the human circulatory system. We analysed two aspects of the representation of the circulatory system: models of the circulatory system (the relationship between the heart and lungs in the blood pathway) and the circulatory system pattern (the blood pathway between organs). In relation to the models of the circulatory system, we found differences among the different groups of students according to the amount of biological instruction they had. The majority of participants (High School Social Science students and both University Psychology student groups) had an incomplete model, which did not include the lungs while the High School Health Science students group had a scientific representation. As for the circulatory system pattern, we did not find differences among groups. However, we cannot conclude that the representations of the two aspects of the circulatory system (model and pattern) are independent. The data in this study show that compulsory human physiology instruction did not enhance the acquisition of a scientific representation of the human circulatory system.     

The Role of Task-Related Learned Representations in Explaining Asymmetries in Task Switching

Task switch costs often show an asymmetry, with switch costs being larger when switching from a difficult task to an easier task. This asymmetry has been explained by difficult tasks being represented more strongly and consequently requiring more inhibition prior to switching to the easier task. The present study shows that switch cost asymmetries observed in arithmetic tasks (addition vs. subtraction) do not depend on task difficulty: Switch costs of similar magnitudes were obtained when participants were presented with unsolvable pseudo-equations that did not differ in task difficulty. Further experiments showed that neither task switch costs nor switch cost asymmetries were due to perceptual factors (e.g., perceptual priming effects). These findings suggest that asymmetrical switch costs can be brought about by the association of some tasks with greater difficulty than others. Moreover, the finding that asymmetrical switch costs were observed (1) in the absence of a task switch proper and (2) without differences in task difficulty, suggests that present theories of task switch costs and switch cost asymmetries are in important ways incomplete and need to be modified.     

Representations of uncertainty in sensorimotor control

Uncertainty is ubiquitous in our sensorimotor interactions, arising from factors such as sensory and motor noise and ambiguity about the environment. Setting it apart from previous theories, a quintessential property of the Bayesian framework for making inference about the state of world so as to select actions, is the requirement to represent the uncertainty associated with inferences in the form of probability distributions. In the context of sensorimotor control and learning, the Bayesian framework suggests that to respond optimally to environmental stimuli the central nervous system needs to construct estimates of the sensorimotor transformations, in the form of internal models, as well as represent the structure of the uncertainty in the inputs, outputs and in the transformations themselves. Here we review Bayesian inference and learning models that have been successful in demonstrating the sensitivity of the sensorimotor system to different forms of uncertainty as well as recent studies aimed at characterizing the representation of the uncertainty at different computational levels.