Search

The predictive processing theory of cognition and neural encoding dictates that hierarchical regions in the neocortex learn and encode predictive hypotheses of current and future stimuli. To better handle uncertainty these regions must also learn, infer, and encode the precision of stimuli. In this treatment we investigate the potential of handling uncertainty within a single learned predictive model. We exploit the rich predictive models formed by the learning of temporal sequences within a Hierarchical Temporal Memory (HTM) framework, a cortically inspired connectionist system of self-organizing predictive cells. We weight a cell’s feedforward response by the degree of its own temporal expectations of a response. We test this model on data that has been saturated with temporal or spatial noise, and show significant improvements over traditional HTM systems. In addition we perform an experiment based on the Posner cuing task and show that the system displays phenomena consistent with attention and biased competition. We conclude that the observed effects are similar to those of precision encoding.

It has become increasingly popular to view the brain as a prediction machine. This view has informed a number of theories of brain function, the most prominent being predictive processing, where generative hypotheses are iteratively updated by error signals. In this treatment we take a lower level approach by examining the hierarchical temporal memory framework, which views individual pyramidal cells as the primary predictive unit of a self-organizing networked sequence learning system. Within this computational framework, the cell behaviour is constrained by a number of parameters which are static and shared across all cells. To further increase the adaptability of the cells, we shift away from this paradigm by introducing the concept of dynamic thresholds. This allows for the activation threshold (the amount of activity on a distal dendrite needed to form a prediction) to be adjusted continuously and individually for each cell. As a metric we use the prior, or unconditional, probability of activity on t...

Attention is both ubiquitous throughout and key to our cognitive experience. It has been shown to filter out mundane stimuli, while simultaneously communicating specific stimuli from the lowest levels of perception through to the highest levels of cognition. In this paper we present a connectionist system with mechanisms that produce both exogenous (bottom-up) and endogenous (top-down) attention. The foundational algorithm of our system is the Temporal Pooler (TP), a neocortically inspired algorithm that learns and predicts temporal sequences. We make a number of modifications to the Temporal Pooler and place it in a framework which is inspired by predictive coding. We use a novel technique in which feedback connections elicit endogenous attention by disrupting the learned representations of attended sequences. Our experiments show that this approach successfully filters attended stimuli and suppresses unattended stimuli.

Fragile X syndrome (FXS) is the most common cause of inherited intellectual disability and a variant of autism spectrum disorder (ASD). The FXS population is quite heterogeneous with respect to comorbidities, which implies the need for a personalized medicine approach, relying on biomarkers or endophenotypes to guide treatment. There is evidence that quantitative electroencephalography (EEG) endophenotype-guided treatments can support increased clinical benefit by considering the patient's neurophysiological profile. We describe a case series of 11 children diagnosed with FXS, aged one to 14 years, mean 4.6 years. Case data are based on longitudinal clinically-observed reports by attending physicians for comorbid symptoms including awake and asleep EEG profiles. We tabulate the comorbid EEG symptoms in this case series, and relate them to the literature on EEG endophenotypes and associated treatment options. The two most common endophenotypes in the data were diffuse slow oscillations and epileptiform EEG, which have been associated with attention and epilepsy respectively. This observation agrees with reported prevalence of comorbid behavioral symptoms for FXS. In this sample of FXS children, attention problems were found in 37% (4 of 11), and epileptic seizures in 45% (5 of 11). Attention problems were found to associate with the epilepsy endophenotype. From the synthesis of this case series and literature review, we argue that the evidence-based personalized treatment approach, exemplified by neurofeedback, could benefit FXS children by focusing on observable, specific characteristics of comorbid disease symptoms.

Aim HLA match is important to solid organ transplantation, especially to kidney and thoracic transplant. However, HLA antibody and matches to liver transplant are still critical topics figure out whether liver graft can develop antibody mediated rejection or protect other solid organ transplant. This study is focusing on monitoring pre- and post-transplant DSA of liver transplant to figure out the importance of HLA CI and CII antibody for liver and other organ transplantation. Methods Eight patients with liver transplant only and one patient received liver and kidney. All donors were no DCD donor but either EBV or CMV positive. Solid phase assay: Luminex HLA single antigen beads. Results 1. Two patients had pre HLA CI DSA. After 2–4 days of transplant, CI DSAs were disappeared and other CI antibodies within the same CREG of DSA were disappeared as well. One patient had DSA as B60 as in Fig. and DSA B60 disappeared after 4 days of transplant. 2. Three patients had pre HLA CII DSA. After liver transplanted, CII DSAs were not disappeared and CII DSA’s MFIs went up higher than pre transplant; there was one patient without DSA, when liver function became abnormal, this patient developed HLA CI and CII DSA; for the patient with liver and kidney transplant, liver graft can’t protect the kidney’s AMR caused from CII DSA (see Table). Conclusions The results indicate that if the patients with liver transplant have HLA CI DSA and functional liver, CI DSA will be cleaned soon after grafted and other CI antibody within the same CREG of DSA will be cleaned as well. If the liver graft failed or infected, the patient will develop both CI and CII DSA. There is a high rate of CII DSA detected after liver transplant. More cases will be collected in order to confirm the indication above. Download high-res image (474KB) Download full-size image Download high-res image (112KB) Download full-size image

The Hierarchical Temporal Memory (HTM) framework is a deep learning system inspired by the functioning of the human neocortex. In this paper we investigate the feasibility of this framework by evaluating the performance of one component, the spatial pooler. Using a recently developed implementation, the augmented spatial pooler (ASP), as a single layer feature detector, we test its performance using a standard image classification pipeline. The main contributions of the paper are the implementation and evaluation of modifications to ASP that enable it to form overcomplete representations of the input and to form connections with multiple data channels. Our results show that these modifications significantly improve the utility of ASP, making its performance competitive with more traditional feature detectors such as sparse restricted Boltzmann machines and sparse auto-encoders.

Sparse coding of visual information has been of interest to the neuroscientific community for many decades and it is widely recognised that sparse codes should exhibit a high degree of statistical independence, typically measured by the kurtosis of the response distributions. In this paper we extend work on the hierarchical temporal memory model by studying the suitability of the augmented spatial pooling (ASP) sparse coding algorithm in comparison with independent component analysis (ICA) when applied to the recognition of handwritten digits. We present an extension to the ASP algorithm that forms synaptic receptive fields located closer to their respective columns and show that this produces lower Naive Bayes classification errors than both ICA and the original ASP algorithm. In evaluating kurtosis as a predictor of classification performance, we also show that additional measures of dispersion and mutual information are needed to reliably distinguish between competing approaches.