Parallel Information Processing Research Articles

A Cognitive-Affective Approach to Foregrounding Categorical-Thematic Patterns in Popular Cinema

The article sets out to discuss disruptions of the embodied flow of movie perception triggered by foregrounded categorical-thematic patterns. First, categorical-thematic patterns are framed in a cognitive perspective and tied to categorical (or parallel) information processing as opposed schematic (sequential). I argue that the former are not prototypical of embodied movie perception and tend to be disruptive if foregrounded, as they are more prevalent in art cinema. Next, I indicate how categorical-thematic patterns may encourage a type of non-habitual pattern recognition producing a number of emotional and aesthetic effects: delight at pattern isolation, wonder emotions, emotional focus of a story theme, and intensification or modulation of global and empathetic emotions. Finally, I turn to illustrate these points usingPan’s Labyrinth, a film that systematically foregrounds categorical-thematic patterns yet naturalizes them, alleviating disruption of movie perception. This, I believe, marks an effective strategy of importing avant-garde film techniques into popular cinema.

Analog programing of conducting-polymer dendritic interconnections and control of their morphology

Although materials and processes are different from biological cells’, brain mimicries led to tremendous achievements in parallel information processing via neuromorphic engineering. Inexistent in electronics, we emulate dendritic morphogenesis by electropolymerization in water, aiming in operando material modification for hardware learning. Systematic study of applied voltage-pulse parameters details on tuning independently morphological aspects of micrometric dendrites’: fractal number, branching degree, asymmetry, density or length. Growths time-lapse image processing shows spatial features to be dynamically dependent, and expand distinctively before and after conductive bridging with two electro-generated dendrites. Circuit-element analysis and impedance spectroscopy confirms their morphological control in temporal windows where growth kinetics is finely perturbed by the input frequency and duty cycle. By the emulation of one’s most preponderant mechanisms for brain’s long-term memory, its implementation in vicinity of sensing arrays, neural probes or biochips shall greatly optimize computational costs and recognition required to classify high-dimensional patterns from complex environments.

Cell and circuit origins of fast network oscillations in the mammalian main olfactory bulb.

Neural synchrony generates fast network oscillations throughout the brain, including the main olfactory bulb (MOB), the first processing station of the olfactory system. Identifying the mechanisms synchronizing neurons in the MOB will be key to understanding how network oscillations support the coding of a high-dimensional sensory space. Here, using paired recordings and optogenetic activation of glomerular sensory inputs in MOB slices, we uncovered profound differences in principal mitral cell (MC) vs. tufted cell (TC) spike-time synchrony: TCs robustly synchronized across fast- and slow-gamma frequencies, while MC synchrony was weaker and concentrated in slow-gamma frequencies. Synchrony among both cell types was enhanced by shared glomerular input but was independent of intraglomerular lateral excitation. Cell-type differences in synchrony could also not be traced to any difference in the synchronization of synaptic inhibition. Instead, greater TC than MC synchrony paralleled the more periodic firing among resonant TCs than MCs and emerged in patterns consistent with densely synchronous network oscillations. Collectively, our results thus reveal a mechanism for parallel processing of sensory information in the MOB via differential TC vs. MC synchrony, and further contrast mechanisms driving fast network oscillations in the MOB from those driving the sparse synchronization of irregularly firing principal cells throughout cortex.

Mimicking Neuroplasticity via Ion Migration in van der Waals Layered Copper Indium Thiophosphate.

Artificial synaptic devices are the essential components of neuromorphic computing systems, which are capable of parallel information storage and processing with high area and energy efficiencies, showing high promise in future storage systems and in-memory computing. Analogous to the diffusion of neurotransmitter between neurons, ion-migration-based synaptic devices are becoming promising for mimicking synaptic plasticity, though the precise control of ion migration is still challenging. Due to the unique 2D nature and highly anisotropic ionic transport properties, van der Waals layered materials are attractive for synaptic device applications. Here, utilizing the high conductivity from Cu+ -ion migration, a two-terminal artificial synaptic device based on layered copper indium thiophosphate is studied. By controlling the migration of Cu+ ions with an electric field, the device mimics various neuroplasticity functions, such as short-term plasticity, long-term plasticity, and spike-time-dependent plasticity. The Pavlovian conditioning and activity-dependent synaptic plasticity involved neural functions are also successfully emulated. These results show a promising opportunity to modulate ion migration in 2D materials through field-driven ionic processes, making the demonstrated synaptic device an intriguing candidate for future low-power neuromorphic applications.

Evidence for parallel processing of relational information in visual search

Evidence for parallel processing of relational information in visual search

Parallel and distributed encoding of speech across human auditory cortex

Speech perception is thought to rely on a cortical feedforward serial transformation of acoustic into linguistic representations. Using intracranial recordings across the entire human auditory cortex, electrocortical stimulation, and surgical ablation, we show that cortical processing across areas is not consistent with a serial hierarchical organization. Instead, response latency and receptive field analyses demonstrate parallel and distinct information processing in the primary and nonprimary auditory cortices. This functional dissociation was also observed where stimulation of the primary auditory cortex evokes auditory hallucination but does not distort or interfere with speech perception. Opposite effects were observed during stimulation of nonprimary cortex in superior temporal gyrus. Ablation of the primary auditory cortex does not affect speech perception. These results establish a distributed functional organization of parallel information processing throughout the human auditory cortex and demonstrate an essential independent role for nonprimary auditory cortex in speech processing.

Predicting Tunnel Squeezing Using the SVM-BP Combination Model

Rock squeezing has a large influence on tunnel construction safety; thus, when designing and constructing tunnels it is highly important to use a reliable method for predicting tunnel squeezing from incomplete data. In this study, a combination SVM-BP (support vector machine-back-propagation) model is proposed to classify the deformation caused by surrounding rock squeezing. We design different characteristic parameters and three types of classifiers (a SVM model, a BP model, and the proposed SVM-BP model) for the tunnel-squeezing prediction experiments and analyse the accuracy of predictions by different models and the influences of characteristic parameters on the prediction results. In contrast to other prediction methods, the proposed SVM-BP model is verified to be reliable. The results show that four characteristics: tunnel diameter (D), tunnel buried depth (H), rock quality index (Q) and support stiffness (K) reflect the effect of rock squeezing sufficiently for classification. The SVM-BP model combines the advantages of both an SVM and a BP neural network. It possesses flexible nonlinear modelling ability and the ability to perform parallel processing of large-scale information. Therefore, the SVM-BP model achieves better classification performance than do the SVM or BP models separately. Moreover, coupling D, H, and K has a significant impact on the predicted results of tunnel squeezing.

Neurocognitive functioning of children with mild to borderline intellectual disabilities and psychiatric disorders: profile characteristics and predictors of behavioural problems.

BackgroundThe aim of the current study was twofold: first, to uncover a neurocognitive profile of normative and relative strengths and weaknesses that characterises an extremely vulnerable group of children with mild to borderline intellectual disabilities (MBID) and co‐morbid psychiatric disorders, and second, to investigate the relevance of these neurocognitive functions explaining internalising and externalising symptoms.MethodWe recruited 45 children (M age = 9.5, SD age = 1.7; range 6–13 years) with MBID (Full‐Scale IQ 50–85) and at least one psychiatric disorder. Neurocognitive functioning was examined utilising the Wechsler Intelligence Scale for Children – Fifth Edition (WISC‐V) indices and the Cognitive Task Application (COTAPP), a comprehensive computerised self‐paced task designed in such a manner that ‘g’ (an overall tendency of children with MBID to execute tasks with a slower reaction time and a higher error rate) has been corrected for in the administration of the task (i.e. completely self‐paced) and in the operationalisation of outcome measures. Behavioural problems were measured using the CBCL and TRF. One‐sample t‐tests and binomial tests were carried out to compare performance with normative data. Regression analyses were used to examine the relationship between neurocognitive parameters and mental health.ResultsCompared with normative data, very small to very large effect sizes were found, indicating clear heterogeneity amongst neurocognitive domains relevant for children with MBID. Two prominent neurocognitive weaknesses emerged: processing speed – characterised by slowness and unstableness combined with a high drift rate and delayed processing of the previous trial, particularly under higher cognitive demands – and working memory – in terms of a weaker central executive and ‘slave’ systems to temporarily store information. Both domains were not clearly predictive of internalising or externalising problems.ConclusionChildren with MBID and psychiatric disorders are hampered by a strongly diminished processing speed and working memory capacity, together resulting in an overall limited processing capacity that may underlie the general developmental delays on domains that depend on fast and parallel processing of information (i.e. language, reading, mathematics and more complex forms of social cognition). Neurocognitive vulnerabilities are neither necessary nor sufficient to explain internalising and externalising problems; rather, a mismatch between the support needs and adaptations these children need, arising from their diminished processing capacity, and the inadequacy of the environment to compensate for this vulnerability may be of relevance.

Causal neural mechanisms of context-based object recognition.

Objects can be recognized based on their intrinsic features, including shape, color, and texture. In daily life, however, such features are often not clearly visible, for example when objects appear in the periphery, in clutter, or at a distance. Interestingly, object recognition can still be highly accurate under these conditions when objects are seen within their typical scene context. What are the neural mechanisms of context-based object recognition? According to parallel processing accounts, context-based object recognition is supported by the parallel processing of object and scene information in separate pathways. Output of these pathways is then combined in downstream regions, leading to contextual benefits in object recognition. Alternatively, according to feedback accounts, context-based object recognition is supported by (direct or indirect) feedback from scene-selective to object-selective regions. Here, in three pre-registered transcranial magnetic stimulation (TMS) experiments, we tested a key prediction of the feedback hypothesis: that scene-selective cortex causally and selectively supports context-based object recognition before object-selective cortex does. Early visual cortex (EVC), object-selective lateral occipital cortex (LOC), and scene-selective occipital place area (OPA) were stimulated at three time points relative to stimulus onset while participants categorized degraded objects in scenes and intact objects in isolation, in different trials. Results confirmed our predictions: relative to isolated object recognition, context-based object recognition was selectively and causally supported by OPA at 160-200 ms after onset, followed by LOC at 260-300 ms after onset. These results indicate that context-based expectations facilitate object recognition by disambiguating object representations in the visual cortex.

Оценка вероятности спонтанного синтеза вычислительных структур применительно к реализации параллельной обработки информации

Мы рассматриваем модель спонтанного формирования вычислительной структуры в мозге человека для решения заданного класса задач в процессе выполнения серии однотипных заданий. Модель основана на специальном определении числовой меры сложности алгоритма решения. Эта мера обладает информационным свойством: сложность вычислительной структуры, состоящей из двух независимых структур, равна сумме сложностей этих структур. Тогда вероятность спонтанного возникновения структуры экспоненциально зависит от сложности структуры. Коэффициент при экспоненте требует экспериментального определения для каждого типа задач. Он может зависеть от формы предъявления исходных данных и от процедуры выдачи результата. Этот метод оценки применен к результатам серии экспериментов, в которых определялась стратегия решения человеком серии однотипных задач с растущим числом исходных данных. Эти эксперименты были описаны в ранее изданных работах. Рассматривались две основные стратегии: последовательное выполнение вычислительного алгоритма или использование параллельных вычислений в тех задачах, где это эффективно. Эти стратегии различаются схемами проведения вычислений. Используя оценку сложности схем, можно по эмпирической вероятности одной из стратегий рассчитать вероятность другой. Проведенные вычисления показали хорошее совпадение расчетной и эмпирической вероятности. Это подтверждает гипотезу о спонтанном формировании структур, решающих задачу, в процессе начальной тренировки человека. Работа содержит краткое описание экспериментов, подробные вычислительные схемы и строгое определение меры сложности вычислительных структур и вывод зависимости вероятности формирования структуры от ее сложности.

Photonic Matrix Computing: From Fundamentals to Applications

In emerging artificial intelligence applications, massive matrix operations require high computing speed and energy efficiency. Optical computing can realize high-speed parallel information processing with ultra-low energy consumption on photonic integrated platforms or in free space, which can well meet these domain-specific demands. In this review, we firstly introduce the principles of photonic matrix computing implemented by three mainstream schemes, and then review the research progress of optical neural networks (ONNs) based on photonic matrix computing. In addition, we discuss the advantages of optical computing architectures over electronic processors as well as current challenges of optical computing and highlight some promising prospects for the future development.

Generation and Characterization of a Cell Type-Specific, Inducible Cre-Driver Line to Study Olfactory Processing.

In sensory systems of the brain, mechanisms exist to extract distinct features from stimuli to generate a variety of behavioral repertoires. These often correspond to different cell types at various stages in sensory processing. In the mammalian olfactory system, complex information processing starts in the olfactory bulb, whose output is conveyed by mitral cells (MCs) and tufted cells (TCs). Despite many differences between them, and despite the crucial position they occupy in the information hierarchy, Cre-driver lines that distinguish them do not yet exist. Here, we sought to identify genes that are differentially expressed between MCs and TCs of the mouse, with an ultimate goal to generate a cell type-specific Cre-driver line, starting from a transcriptome analysis using a large and publicly available single-cell RNA-seq dataset (Zeisel et al., 2018). Many genes were differentially expressed, but only a few showed consistent expressions in MCs and at the specificity required. After further validating these putative markers using ISH, two genes (i.e., Pkib and Lbdh2) remained as promising candidates. Using CRISPR/Cas9-mediated gene editing, we generated Cre-driver lines and analyzed the resulting recombination patterns. This indicated that our new inducible Cre-driver line, Lbhd2-CreERT2, can be used to genetically label MCs in a tamoxifen dose-dependent manner, both in male and female mice, as assessed by soma locations, projection patterns, and sensory-evoked responses in vivo. Hence, this is a promising tool for investigating cell type-specific contributions to olfactory processing and demonstrates the power of publicly accessible data in accelerating science.SIGNIFICANCE STATEMENT In the brain, distinct cell types play unique roles. It is therefore important to have tools for studying unique cell types specifically. For the sense of smell in mammals, information is processed first by circuits of the olfactory bulb, where two types of cells, mitral cells and tufted cells, output different information. We generated a transgenic mouse line that enables mitral cells to be specifically labeled or manipulated. This was achieved by looking for genes that are specific to mitral cells using a large and public gene expression dataset, and creating a transgenic mouse using the gene editing technique, CRISPR/Cas9. This will allow scientists to better investigate parallel information processing underlying the sense of smell.

Neuromorphic extreme learning machines with bimodal memristive synapses

The biology-inspired intelligent computing system for the neuromorphic hardware implementation is useful in high-speed parallel information processing. However, the traditional Von Neumann computer architecture and the unsatisfactory signal transmission approach have jointly limited the overall performance of the specific hardware implementation. In this paper, a compact extreme learning machine (ELM) architecture synthesized with the spintronic memristor-based synaptic circuit, the biasing circuit, and the activation function circuit is presented. Notably, due to the threshold characteristic of the memristive device, the synaptic circuit has a bimodal behavior. Namely, it is capable to provide the constant and adjustable network weights between the adjacent layers in the ELM. Furthermore, two major limitations (process variations and sneak path issue) are taken into account for the detailed robustness analysis of the whole network. Finally, the entire scheme is verified with case studies in single image super-resolution (SR) reconstruction.

Fast Post-Hoc Normalization for Brain Inspired Sparse Coding on a Neuromorphic Device

Exploration of novel computational platforms is critical for the advancement of artificial intelligence as we approach the physical limitations of traditional hardware. Biologically accurate, energy efficient neuromorphic systems are particularly promising for enabling future breakthroughs because of their ability to process information in parallel and to scale using extremely low power. Sparse coding is a signal processing technique which has been known to model the information encoding in the primary visual cortex. When sparse solutions are solved using local neuron competition along with the unsupervised dictionary learning that mimics cortical development, we can build an end to end, hardware to software, brain inspired solution to a machine learning problem. In this article, we perform a detailed comparison of sparse coding solutions generated classically by orthogonal matching pursuit (OMP) implemented on a conventional digital processor with spike-based solutions obtained using the Intel Loihi neuromorphic processor. A novel “post-hoc” normalization technique to shorten simulation time for Loihi is presented along with analysis of optimal parameter selection, reconstruction errors, and unsupervised dictionary learning for Loihi approaches and their classical counterparts. Preliminary results show that both the Loihi full simulation approach and the post-hoc normalization approach are well suited to neuromorphic processors and operate in a size, weight and power regime that is not accessible by classical approaches. Ultimately, the use of this normalization technique allows for faster and, often, better solutions than demonstrated previously.

Drosophila Fezf functions as a transcriptional repressor to direct layer-specific synaptic connectivity in the fly visual system

The layered compartmentalization of synaptic connections, a common feature of nervous systems, underlies proper connectivity between neurons and enables parallel processing of neural information. However, the stepwise development of layered neuronal connections is not well understood. The medulla neuropil of the Drosophila visual system, which comprises 10 discrete layers (M1 to M10), where neural computations underlying distinct visual features are processed, serves as a model system for understanding layered synaptic connectivity. The first step in establishing layer-specific connectivity in the outer medulla (M1 to M6) is the innervation by lamina (L) neurons of one of two broad, primordial domains that will subsequently expand and transform into discrete layers. We previously found that the transcription factor dFezf cell-autonomously directs L3 lamina neurons to their proper primordial broad domain before they form synapses within the developing M3 layer. Here, we show that dFezf controls L3 broad domain selection through temporally precise transcriptional repression of the transcription factor slp1 (sloppy paired 1). In wild-type L3 neurons, slp1 is transiently expressed at a low level during broad domain selection. When dFezf is deleted, slp1 expression is up-regulated, and ablation of slp1 fully rescues the defect of broad domain selection in dFezf-null L3 neurons. Although the early, transient expression of slp1 is expendable for broad domain selection, it is surprisingly necessary for the subsequent L3 innervation of the M3 layer. DFezf thus functions as a transcriptional repressor to coordinate the temporal dynamics of a transcriptional cascade that orchestrates sequential steps of layer-specific synapse formation.

Symmetry-Induced Error Filtering in a Photonic Lieb Lattice.

Quantum computation promises intrinsically parallel information processing capacity by harnessing the superposition and entanglement of quantum states. However, it is still challenging to realize universal quantum computation due that the reliability and scalability are limited by unavoidable noises on qubits. Nontrivial topological properties like quantum Hall phases are found capable of offering protection, but require stringent conditions of topological band gaps and broken time-reversal symmetry. Here, we propose and experimentally demonstrate a symmetry-induced error filtering scheme, showing a more general role of geometry in protection mechanism and applications. We encode qubits in a superposition of two spatial modes on a photonic Lieb lattice. The geometric symmetry endows the system with topological properties featuring a flat band touching, leading to distinctive transmission behaviors of π-phase qubits and 0-phase qubits. The geometry exhibits a significant effect on filtering phase errors, which also enables it to monitor phase deviations in real time. The symmetry-induced error filtering can be a key element for encoding and protecting quantum states, suggesting an emerging field of symmetry-protected universal quantum computation and noisy intermediate-scale quantum technologies.

Memristor-based BAM circuit implementation for image associative memory and filling-in

Associative memory and filling-in are two essential functions of the human brain. To implement these two brain-inspired functions in hardware, we proposed a memristor-based bidirectional associative memory (BAM) circuit in this paper. This circuit combines an online algorithm with a memristor array adjustment process, thus makes the circuit more universal for various tasks. The proposed circuit is constructed out of memristive synaptic circuits, IN modules and ACT modules. The memristive synaptic circuits utilize memristor arrays to represent weight matrix and operate corresponding operations hence make computing-in-memory and process information in parallel, which simplifies the complexity of circuit and improves the processing speed. The IN modules employ transistors as switches to choose the input layer hence can get initial information flow bidirectionally. The ACT modules perform activation function and can output continuous arbitrary real numbers. Thereby, both binary and gray-scale images can be tested in the proposed circuit. In addition to the hetero-association and filling-in results given in detail, the retrieval rates of the proposed circuit with the impact of different degrees of noise and the number of stored patterns are also evaluated and compared with software-based BAM. The simulation has experimented via MATLAB and PSpice, and the corresponding results show a remarkable performance of the proposed circuit. The influence of memristor’s stuck-at-fault is also considered. In comparison with software-based BAM and similar memristor-based neural network circuit, the proposed circuit performs better in processing speed.

Artificial synapses that exploit ionic modulation for perception and integration

We propose the first two-terminal sodium-ion artificial synapse (SIAS) with nanoparticle/electrolyte architecture to emulate behaviors of a biological synapse. The device emulates essential synaptic functions such as postsynaptic current and paired-pulse facilitation by employing sodium-ion doping. The synaptic event uses only 0.55 pW at an ultra-low operating voltage of −6 mV, sensitive responsiveness. Pain perception nociceptors, with pain sensitization/desensitization properties, are realized using SIAS. Artificial neural components with multiple inputs also implement dendritic integration functions such as spatiotemporal learning rules and classical conditional learning. Furthermore, benefiting from the nanostructured active layer and solid-state polymer electrolyte, the flexible device on stainless steel substrate induced different deformation curvatures. The proposed SIAS device with capabilities of parallel information processing and pain perception could apply to future bioinspired flexible sensing-processing nervous systems.

Parallel Social Information Processing Circuits Are Differentially Impacted in Autism

Parallel processing circuits are thought to dramatically expand the network capabilities of the nervous system. Magnocellular and parvocellular oxytocin neurons have been proposed to subserve two parallel streams of social information processing, which allow a single molecule to encode a diverse array of ethologically distinct behaviors. Here we provide the first comprehensive characterization of magnocellular and parvocellular oxytocin neurons in male mice, validated across anatomical, projection target, electrophysiological, and transcriptional criteria. We next use novel multiple feature selection tools in Fmr1-KO mice to provide direct evidence that normal functioning of the parvocellular but not magnocellular oxytocin pathway is required for autism-relevant social reward behavior. Finally, we demonstrate that autism risk genes are enriched in parvocellular compared with magnocellular oxytocin neurons. Taken together, these results provide the first evidence that oxytocin-pathway-specific pathogenic mechanisms account for social impairments across a broad range of autism etiologies.

Intrinsic non-hub connectivity predicts human inter-temporal decision-making.

Inter-temporal decision-making is ubiquitous in daily life and has been considered as a critical characteristic associated with an individual's success. Such decisions require us to tradeoff between short-term and long-term benefits. Prior studies have indicated that inter-temporal decision involves various brain regions that tend to occupy the central hubs. However, it is unclear whether the functional connectivities among hub as well as non-hub regions can predict discounting behaviors. Here, we combined with graph-theoretical algorithm and multivariate pattern analysis to explore whether voxel-wise functional connectivity strength in the whole brain could predict discounting rates (indexed as logk, based on the adaptive delay-discounting task) in a relatively large sample (n= 429) of young adults. Results revealed that short- and long-distance as well as all-range non-hub functional connectivity strength in the limbic system (i.e., medial orbitofrontal cortex and parahippocampus) were inversely associated with discounting rates. Furthermore, these results were robust and did not appear to be due to potential confounding factors. Above weight-based degree metric is commonly indicative of the communication pattern of local and global parallel information processing, and it therefore provides novel insights into the neural mechanisms underlying inter-temporal decision-making from the perspective of human brain topological organizations.