The relational bottleneck as an inductive bias for efficient abstraction.

A central challenge for cognitive science is to explain how abstract concepts are acquired from limited experience. This has often been framed in terms of a dichotomy between connectionist and symbolic cognitive models. Here, we highlight a recently emerging line of work that suggests a novel reconciliation of these approaches, by exploiting an inductive bias that we term the relational bottleneck. In that approach, neural networks are constrained via their architecture to focus on relations between perceptual inputs, rather than the attributes of individual inputs. We review a family of models that employ this approach to induce abstractions in a data-efficient manner, emphasizing their potential as candidate models for the acquisition of abstract concepts in the human mind and brain.

Images with harder-to-reconstruct visual representations leave stronger memory traces.

Much of what we remember is not because of intentional selection, but simply a by-product of perceiving. This raises a foundational question about the architecture of the mind: how does perception interface with and influence memory? Here, inspired by a classic proposal relating perceptual processing to memory durability, the level-of-processing theory, we present a sparse coding model for compressing feature embeddings of images, and show that the reconstruction residuals from this model predict how well images are encoded into memory. In an open memorability dataset of scene images, we show that reconstruction error not only explains memory accuracy, but also response latencies during retrieval, subsuming, in the latter case, all of the variance explained by powerful vision-only models. We also confirm a prediction of this account with 'model-driven psychophysics'. This work establishes reconstruction error as an important signal interfacing perception and memory, possibly through adaptive modulation of perceptual processing.

Shallow neural networks trained to detect collisions recover features of visual loom-selective neurons.

Animals have evolved sophisticated visual circuits to solve a vital inference problem: detecting whether or not a visual signal corresponds to an object on a collision course. Such events are detected by specific circuits sensitive to visual looming, or objects increasing in size. Various computational models have been developed for these circuits, but how the collision-detection inference problem itself shapes the computational structures of these circuits remains unknown. Here, inspired by the distinctive structures of LPLC2 neurons in the visual system of Drosophila, we build anatomically-constrained shallow neural network models and train them to identify visual signals that correspond to impending collisions. Surprisingly, the optimization arrives at two distinct, opposing solutions, only one of which matches the actual dendritic weighting of LPLC2 neurons. Both solutions can solve the inference problem with high accuracy when the population size is large enough. The LPLC2-like solutions reproduces experimentally observed LPLC2 neuron responses for many stimuli, and reproduces canonical tuning of loom sensitive neurons, even though the models are never trained on neural data. Thus, LPLC2 neuron properties and tuning are predicted by optimizing an anatomically-constrained neural network to detect impending collisions. More generally, these results illustrate how optimizing inference tasks that are important for an animal's perceptual goals can reveal and explain computational properties of specific sensory neurons.

Bone marrow aspirate collection and preparation--a comparison of three methods.

PURPOSE: Preparing bone marrow smears using non-anticoagulated bone marrow aspirate is a traditional practice but many laboratories now use anticoagulated aspirate samples in K-EDTA. There are no published studies comparing the effectiveness of these two methods. This report compares the readability of slides, prepared using non-anticoagulated and anticoagulated methods, from three laboratories in Hamilton Ontario. METHODS: A blinded set of 129 aspirate slides prepared using anticoagulated and non-anticoagulated methodologies (using K-EDTA) was reviewed by three reviewers. Slides were classified as unreadable if two of the three observers rejected them based on a standardized survey. RESULTS: The proportion of slides classed as unreadable varied widely (5.0% to 46.9%) depending on collection and slide preparation methods. Degree of coagulation did not affect readability. CONCLUSION: A measurable advantage to using non-anticoagulated bone marrow was not demonstrated. Immediate anticoagulation of bone marrow samples, with laboratory personnel at the bedside to assess sample quality, followed by slide preparation in the laboratory provided the best results.

The huge Package for High-dimensional Undirected Graph Estimation in R.

We describe an R package named huge which provides easy-to-use functions for estimating high dimensional undirected graphs from data. This package implements recent results in the literature, including Friedman et al. (2007), Liu et al. (2009, 2012) and Liu et al. (2010). Compared with the existing graph estimation package glasso, the huge package provides extra features: (1) instead of using Fortan, it is written in C, which makes the code more portable and easier to modify; (2) besides fitting Gaussian graphical models, it also provides functions for fitting high dimensional semiparametric Gaussian copula models; (3) more functions like data-dependent model selection, data generation and graph visualization; (4) a minor convergence problem of the graphical lasso algorithm is corrected; (5) the package allows the user to apply both lossless and lossy screening rules to scale up large-scale problems, making a tradeoff between computational and statistical efficiency.

A multicenter trial of the effectiveness of zeta-globin enzyme-linked immunosorbent assay and hemoglobin H inclusion body screening for the detection of alpha0-thalassemia trait.

Routine laboratories use a hemoglobin H (HbH) screen to detect alpha-thalassemia carriers of fatal hemoglobin Bart's hydrops fetalis. This test is laborious and has sensitivity concerns. A commercial zeta-globin enzyme-linked immunosorbent assay (ELISA) is effective in detecting Southeast Asian (SEA) alpha-thalassemia. We present results of a study of the effectiveness of carrier detection of ELISA and a shortened HbH screen compared with gap polymerase chain reaction. ELISA was superior to the HbH screen for the SEA alpha0-thalassemia trait. The ELISA and H screen were equal for detection of all carriers encountered and combined were more effective than either test alone. A positive zeta-globin ELISA result is diagnostic of SEA alpha-thalassemia, and routine use of the zeta-globin ELISA in combination with a shortened HbH screen will improve the efficacy of prenatal screening for carriers of hemoglobin Bart's hydrops fetalis through improved detection and referral for follow-up DNA testing.

Prevalence of thalassemia in patients with microcytosis referred for hemoglobinopathy investigation in Ontario: a prospective cohort study.

In Ontario, Canada, beta-thalassemia is easily detected through measurement of hemoglobin A2, but most laboratories do not do exhaustive DNA investigations for alpha-thalassemia. Therefore, the prevalence of thalassemia in microcytic samples for hemoglobinopathy investigation in Ontario is unknown. To address this, we performed a prospective cohort study in which samples referred for hemoglobinopathy investigation were also evaluated for alpha-thalassemia by DNA testing. Of 800 samples submitted, 664 were evaluable. Of the 664 patients represented, 163 (24.5%) were beta-thalassemia major carriers, 68 (10.2%) were hemoglobin Bart's hydrops fetalis carriers and, in total, 361 (54.4%) had some form of thalassemia. We conclude that microcytosis due to thalassemia is common in Ontario and that major forms of thalassemia, including forms predisposing to hemoglobin Bart's hydrops fetalis and beta-thalassemia major, are frequent. This illustrates the importance of adequate prenatal and laboratory investigation for these abnormalities in Ontario and other similar multiethnic jurisdictions worldwide.

Proficiency testing of reticulocyte counting in Ontario.

The reticulocyte count reflects the erythropoietic activity of the bone marrow and is thus useful in both the approach to the diagnosis of anemia and in monitoring bone marrow response to therapy. Traditionally, reticulocyte quantitation relied upon microscopic techniques; recently automated reticulocyte quantitation has become widely available. External quality assessment (EQA) of reticulocyte quantitation has not been widely reported; this paper presents data from reticulocyte EQA surveys conducted in Ontario between November 2000 and October 2004. Samples with normal and increased reticulocyte levels were obtained from adult donors and analyzed within 48 hours. We found that despite improved accuracy and precision, automated reticulocyte counting techniques do not translate into markedly improved inter-laboratory accuracy and precision. Further, we found that both microscopic and automated techniques can effectively differentiate between patients with normal and those with increased reticulocyte counts. Finally, we ascertained that reference intervals submitted with the initial surveys revealed that laboratories could have made erroneous clinical interpretations of their normal or elevated reticulocyte counts. The reestablishment of reference intervals by participants resulted in a marked improvement of the reference intervals submitted on subsequent surveys.

Mixed-membership models of scientific publications.

PNAS is one of world's most cited multidisciplinary scientific journals. The PNAS official classification structure of subjects is reflected in topic labels submitted by the authors of articles, largely related to traditionally established disciplines. These include broad field classifications into physical sciences, biological sciences, social sciences, and further subtopic classifications within the fields. Focusing on biological sciences, we explore an internal soft-classification structure of articles based only on semantic decompositions of abstracts and bibliographies and compare it with the formal discipline classifications. Our model assumes that there is a fixed number of internal categories, each characterized by multinomial distributions over words (in abstracts) and references (in bibliographies). Soft classification for each article is based on proportions of the article's content coming from each category. We discuss the appropriateness of the model for the PNAS database as well as other features of the data relevant to soft classification.

Good practice guidelines for laboratory investigation of hemoglobinopathies.

A review of the Quality Management Program-Laboratory Services (QMP-LS) hemoglobin fraction quantitation/hemoglobinopathy investigation surveys revealed recurrent errors in diagnosis of carrier genotypes for serious hemoglobinopathies. To address these concerns, QMP-LS developed guidelines for laboratory investigation of hemoglobinopathy syndromes. The guidelines are based on current practice in the province of Ontario, expert opinion in the field, and a review of recent literature. They include an overview of the clinical significance of the hemoglobinopathies, the indications for a hemoglobinopathy investigation, key components of a thorough laboratory hemoglobinopathy investigation, interpretation of investigation results, diagnosis of rare hemoglobin variants, and indications for DNA investigation of hemoglobinopathies. The guidelines present various tables and figures outlining mean cell volume reference intervals, recommended hemoglobinopathy investigative techniques, recommendations for further investigation of abnormal findings, and algorithms for interpretation of the various laboratory hemoglobinopathy investigative techniques. The guidelines are intended for use in the processing of specimens when a clinical or laboratory physician has ordered a hemoglobinopathy investigation (hemoglobin electrophoresis).

Beta-thalassemia in association with a new delta-chain hemoglobin variant [delta116(g18)Arg-->Leu]: implications for carrier screening and prenatal diagnosis.

We describe a complicated genetic counseling and prenatal diagnostic case involving an East Indian couple that had lost two consecutive pregnancies. Hemoglobinopathy screening was conducted to investigate the possibility of Hb Bart's hydrops fetalis or Hb H hydrops fetalis. The initial work-up indicated that alpha-thalassemia was not a contributing factor, with both parents being carriers of single gene deletions (-alpha(3.7)/alphaalpha). However, the Hb electrophoresis results indicated that the couple might be at risk for having children with Hb E/Hb Lepore disease. Subsequent DNA testing demonstrated that the father carried the Hb E mutation, but failed to confirm that the mother carries the Hb Lepore deletion. Sequence analysis revealed that the mother was heterozygous for a common East Indian beta(0)-thalassemia mutation, yet had a normal level of Hb A(2). The mother also carried a previously unreported missense mutation of the delta-globin gene, in cis with the beta(0)-thalassemia mutation, which gave rise to the minor Hb variant originally misidentified as Hb Lepore. This case illustrates the importance of comprehensive molecular analyses for accurate assessment of genetic risks for hemoglobinopathy syndromes.

Fetal-maternal hemorrhage detection in Ontario.

The results from fetal-maternal hemorrhage (FMH) detection and quantitation external quality assessment surveys conducted in Ontario indicate that the rosette test had a sensitivity and specificity for an FMH of more than 10 mL of 1.0 and 0.75, respectively, compared with 0.96 and 0.92, respectively, for acid elution. With FMH quantitation, the percentage error of the mean from the target FMH was 20% or more in 7 of 8 surveys, and coefficients of variation ranged from 39.5% to 71.8%. Inadequate Rho(D) immune globulin prophylaxis could have occurred in 19.4% of the challenges with an FMH of more than 10 mL. The rosette and acid elution techniques are both effective for the detection or exclusion of FMH, but acid elution lacks adequate accuracy and precision for reliable FMH quantitation. Furthermore, a strategy of prescribing an extra 1,500-IU Rho(D) immune globulin dose, in addition to the dose required to treat the volume of fetal blood detected, is an effective strategy to overcome the limitations of FMH quantitation by acid elution.

Evaluation of a dual hemoglobin A(2)/A(1c) quantitation kit on the bio-rad variant II automated hemoglobin analyzer.

CONTEXT: Quantitation of hemoglobin (Hb) A(1c) and investigation of hemoglobinopathy on the Bio-Rad Variant analyzers require a switch between 2 separate kits that is time consuming and causes errors. OBJECTIVE: Evaluation of a new Variant II HbA(2)/HbA(1c) Dual kit capable of both Hb A(1c) quantitation and hemoglobinopathy investigation on a single kit. DESIGN: We evaluated Hb A(1c), Hb A(2), and Hb F quantitation for precision, linearity, and correlation with current methodology. We also evaluated detection of Hb variants and correlation of Hb Barts quantitation. SETTING: Hamilton Regional Laboratory Medicine Program, Provincial Hemoglobinopathy Laboratory, St Joseph's Healthcare Site, Hamilton, Ontario. PATIENTS: Patient blood samples submitted for Hb A(1c) quantitation or hemoglobinopathy investigation. MAIN OUTCOME MEASURES: Precision, linearity, linear regression, and reference interval validation. RESULTS: We provide tables and figures illustrating precision, linearity, linear regression, and quantitation of Hb variants. We validated reference intervals for Hb A(1c), Hb A(2), and Hb F. CONCLUSIONS: The dual kit provides precise Hb A(1c), Hb A(2), and Hb F quantitation. The results show good linearity and correlate well with the results of current methods. We detected all clinically important Hb variants and a wide variety of rare variants. The dual kit has several advantages: it eliminates the need for extensive kit switch over; improves utility for newborn screening because of its quantification of Hb Barts; permits quantification of Hb A(1c) using the beta-Thal method; and eliminates the need for separate Hb A(2) reference intervals for patients with Hb S because of its accurate quantitation of Hb A(2) in the presence of Hb S.