Serial Bottleneck Research Articles

Is there a serial bottleneck in visual object recognition?

Divided attention has little effect for simple tasks, such as luminance detection, but it has large effects for complex tasks, such as semantic categorization of masked words. Here, we asked whether the semantic categorization of visual objects shows divided attention effects as large as those observed for words, or as small as those observed for simple feature judgments. Using a dual-task paradigm with nameable object stimuli, performance was compared with the predictions of serial and parallel models. At the extreme, parallel processes with unlimited capacity predict no effect of divided attention; alternatively, an all-or-none serial process makes two predictions: a large divided attention effect (lower accuracy for dual-task trials, compared to single-task trials) and a negative response correlation in dual-task trials (a given response is more likely to be incorrect when the response about the other stimulus is correct). These predictions were tested in two experiments examining object judgments. In both experiments, there was a large divided attention effect and a small negative correlation in responses. The magnitude of these effects was larger than for simple features, but smaller than for words. These effects were consistent with serial models, and rule out some but not all parallel models. More broadly, the results help establish one of the first examples of likely serial processing in perception.

Parallel and serial processes in number-to-quantity conversion

Converting a multi-digit number to quantity requires processing not only the digits but also the number's decimal structure, thus raising several issues. First, are all the digits processed in parallel, or serially from left to right? Second, given that the same digit at different places can represent different quantities (e.g., “2” can mean 2, 20, etc.), how is each digit assigned to its correct decimal role? We presented participants with two-digit numbers and asked them to point at the corresponding locations on a number line, while we recorded their pointing trajectory. Crucially, on some trials, the decade and unit digits did not appear simultaneously. When the decade digit was delayed, the decade effect on finger movement was delayed by the same amount. However, a lag in presenting the unit digit delayed the unit effect by 35 ms less than the lag duration, a pattern reminiscent of the psychological refractory period, indicating an idle time window of 35 ms in the units processing pathway. When a lag transiently caused a display of just one digit on screen, the unit effect increased and the decade effect decreased, suggesting errors in binding digits to decimal roles. We propose that a serial bottleneck is imposed by the creation of a syntactic frame for the multidigit number, a process launched by the leftmost digit. All other stages, including the binding of digits to decimal roles, quantification, and merging them into a whole-number quantity, appear to operate in parallel across digits, suggesting a remarkable degree of parallelism in expert readers.

Visual word recognition: Evidence for a serial bottleneck in lexical access

Reading is a demanding task, constrained by inherent processing capacity limits. Do those capacity limits allow for multiple words to be recognized in parallel? In a recent study, we measured semantic categorization accuracy for nouns presented in pairs. The words were replaced by post-masks after an interval that was set to each subject’s threshold, such that with focused attention they could categorize one word with ~80% accuracy. When subjects tried to divide attention between both words, their accuracy was so impaired that it supported a serial processing model: on each trial, subjects could categorize one word but had to guess about the other. In the experiments reported here, we investigated how our previous result generalizes across two tasks that require lexical access but vary in the depth of semantic processing (semantic categorization and lexical decision), and across different masking stimuli, word lengths, lexical frequencies and visual field positions. In all cases, the serial processing model was supported by two effects: (1) a sufficiently large accuracy deficit with divided compared to focused attention; and (2) a trial-by-trial stimulus processing tradeoff, meaning that the response to one word was more likely to be correct if the response to the other was incorrect. However, when the task was to detect colored letters, neither of those effects occurred, even though the post-masks limited accuracy in the same way. Altogether, the results are consistent with the hypothesis that visual processing of words is parallel but lexical access is serial.

Humans can efficiently look for but not select multiple visual objects.

The human brain recurrently prioritizes task-relevant over task-irrelevant visual information. A central question is whether multiple objects can be prioritized simultaneously. To answer this, we let observers search for two colored targets among distractors. Crucially, we independently varied the number of target colors that observers anticipated, and the number of target colors actually used to distinguish the targets in the display. This enabled us to dissociate the preparation of selection mechanisms from the actual engagement of such mechanisms. Multivariate classification of electroencephalographic activity allowed us to track selection of each target separately across time. The results revealed only small neural and behavioral costs associated with preparing for selecting two objects, but substantial costs when engaging in selection. Further analyses suggest this cost is the consequence of neural competition resulting in limited parallel processing, rather than a serial bottleneck. The findings bridge diverging theoretical perspectives on capacity limitations of feature-based attention.

Parallel spatial channels converge at a bottleneck in anterior word-selective cortex

In most environments, the visual system is confronted with many relevant objects simultaneously. That is especially true during reading. However, behavioral data demonstrate that a serial bottleneck prevents recognition of more than one word at a time. We used fMRI to investigate how parallel spatial channels of visual processing converge into a serial bottleneck for word recognition. Participants viewed pairs of words presented simultaneously. We found that retinotopic cortex processed the two words in parallel spatial channels, one in each contralateral hemisphere. Responses were higher for attended than for ignored words but were not reduced when attention was divided. We then analyzed two word-selective regions along the occipitotemporal sulcus (OTS) of both hemispheres (subregions of the visual word form area, VWFA). Unlike retinotopic regions, each word-selective region responded to words on both sides of fixation. Nonetheless, a single region in the left hemisphere (posterior OTS) contained spatial channels for both hemifields that were independently modulated by selective attention. Thus, the left posterior VWFA supports parallel processing of multiple words. In contrast, activity in a more anterior word-selective region in the left hemisphere (mid OTS) was consistent with a single channel, showing (i) limited spatial selectivity, (ii) no effect of spatial attention on mean response amplitudes, and (iii) sensitivity to lexical properties of only one attended word. Therefore, the visual system can process two words in parallel up to a late stage in the ventral stream. The transition to a single channel is consistent with the observed bottleneck in behavior.

A Robust Distributed Big Data Clustering-based on Adaptive Density Partitioning using Apache Spark

Unsupervised machine learning and knowledge discovery from large-scale datasets have recently attracted a lot of research interest. The present paper proposes a distributed big data clustering approach-based on adaptive density estimation. The proposed method is developed-based on Apache Spark framework and tested on some of the prevalent datasets. In the first step of this algorithm, the input data is divided into partitions using a Bayesian type of Locality Sensitive Hashing (LSH). Partitioning makes the processing fully parallel and much simpler by avoiding unneeded calculations. Each of the proposed algorithm steps is completely independent of the others and no serial bottleneck exists all over the clustering procedure. Locality preservation also filters out the outliers and enhances the robustness of the proposed approach. Density is defined on the basis of Ordered Weighted Averaging (OWA) distance which makes clusters more homogenous. According to the density of each node, the local density peaks will be detected adaptively. By merging the local peaks, final cluster centers will be obtained and other data points will be a member of the cluster with the nearest center. The proposed method has been implemented and compared with similar recently published researches. Cluster validity indexes achieved from the proposed method shows its superiorities in precision and noise robustness in comparison with recent researches. Comparison with similar approaches also shows superiorities of the proposed method in scalability, high performance, and low computation cost. The proposed method is a general clustering approach and it has been used in gene expression clustering as a sample of its application.

PDDS: Scalable Sensor Data Distribution for Cyber-Physical Systems Using GPGPUs

In many cyber-physical systems (CPS), one of the primary challenges is how to disseminate many sensor streams to collaborating entities in a timely, reliable, and scalable manner. However, in CPS, communication and related computation is often unstable and constitutes a serial bottleneck when multiple sensor streams need to be distributed. In this paper, we propose a novel publish-subscribe middleware architecture, called parallel data distribution service (PDDS), to address this problem. To bypass this serial bottleneck effectively, PDDS exploits state-space models for sensor streams both at a publisher and its subscribers. These models are used to estimate true states of monitored physical processes without actual communication. This approach not only reduces the communication overhead significantly, but also provides the tolerance against potential delays and instability in communication. Further, with the model-based methods, many sensor streams can be accessed in parallel since sensor streams can bypass a serial bottleneck, such as network stacks. At the middleware layer, PDDS processes the model-related computation, such as the state estimation, in parallel using modern embedded general-purpose computing on graphics processing units, enabling highly scalable and energy-efficient processing of massive sensor streams. The proposed approach is implemented in a data distribution service middleware using CUDA. Our evaluation results show that the proposed approach can achieve significant energy savings and much higher scalability compared to baseline approaches.

An analysis on the traffic processing efficiency of a combination of serial and parallel bottlenecks

By means of the Nagel–Schreckenberg model, we have investigated into the maximum vehicular flow rate of traffic processing bottlenecks. The evaluated analytical form of this flow rate is found to give quantitative insights into the underlying physics of collective vehicular motions constrained by these bottlenecks. Our analysis shows that for large-scale expansion, a new class of processing bottleneck known as the serial bottleneck is more efficient than the conventional parallel bottleneck in the absence of human driving behavior. When characteristics such as slow-to-start is considered in the model, the consequential delay due to human reaction time not only degrade the overall efficiency, it also diminishes the efficacy of serial processing such that a serial bottleneck is no longer tenable for traffic processing. These results point to the fundamental importance of optimizing traffic efficiency, which we illustrate by elucidating the detailed mechanisms with which vehicles interact collectively in the bottlenecks. In particular, we demonstrate that by constructing combinations of serial and parallel bottlenecks, optimal efficiencies are achieved via configurations with few (many) lanes of a large (small) number of serial units when the processing time is short (long). A direct implication of these results is that autonomous self-driving vehicles could serve to improve the transportation capacity for the densely populated urban cities of the future, due to the intrinsically more efficient collective vehicular motions through these bottlenecks.

Syntactic Recursion Facilitates and Working Memory Predicts Recursive Theory of Mind.

In this study, we focus on the possible roles of second-order syntactic recursion and working memory in terms of simple and complex span tasks in the development of second-order false belief reasoning. We tested 89 Turkish children in two age groups, one younger (4;6–6;5 years) and one older (6;7–8;10 years). Although second-order syntactic recursion is significantly correlated with the second-order false belief task, results of ordinal logistic regressions revealed that the main predictor of second-order false belief reasoning is complex working memory span. Unlike simple working memory and second-order syntactic recursion tasks, the complex working memory task required processing information serially with additional reasoning demands that require complex working memory strategies. Based on our results, we propose that children’s second-order theory of mind develops when they have efficient reasoning rules to process embedded beliefs serially, thus overcoming a possible serial processing bottleneck.

Temporal discrimination of one's own reaction times in dual-task performance: Context effects and methodological constraints.

In this study we used the method of constant stimuli to investigate introspective reaction times in the psychological refractory period (PRP) paradigm under different temporal contexts. Previous introspective PRP studies have mostly used visual analogue scales to assess introspective reaction times and found that participants were largely unaware of the typical dual-task costs that arise in this paradigm (PRP effect). This apparent limitation of introspection has been taken as evidence for a serial processing bottleneck that encompasses response selection as well as conscious perception. In our study, in each trial participants first performed the PRP task and were then presented with a comparison interval that they had to compare with their reaction time to the second task (RT2). Across three experiments, we observed that the subjective estimates of RT2 (i.e., the points of subjective equality) did not reflect the objective pattern but were almost completely biased toward the center of the comparison intervals (asymmetry effect). In a control experiment in which participants discriminated RT2s of other participants without performing the PRP task, this bias was largely reduced. We interpret these results as indicating that in dual-task performance participants acquire only poor temporal representations of their own reaction times, and the apparent unawareness of the PRP effect may reflect disturbed timing abilities rather than a conscious perception bottleneck.

The source of dual-task limitations: Serial or parallel processing of multiple response selections?

Although it is generally recognized that the concurrent performance of two tasks incurs costs, the sources of these dual-task costs remain controversial. The serial bottleneck model suggests that serial postponement of task performance in dual-task conditions results from a central stage of response selection that can only process one task at a time. Cognitive-control models, by contrast, propose that multiple response selections can proceed in parallel, but that serial processing of task performance is predominantly adopted because its processing efficiency is higher than that of parallel processing. In the present study, we empirically tested this proposition by examining whether parallel processing would occur when it was more efficient and financially rewarded. The results indicated that even when parallel processing was more efficient and was incentivized by financial reward, participants still failed to process tasks in parallel. We conclude that central information processing is limited by a serial bottleneck.

General Cognitive Ability and the Psychological Refractory Period

Identifying the precise locus of general cognitive ability (g) in the flow of information between perception and action is an important goal of differential psychology. To localize the negative correlation between g and reaction time to a specific processing stage, we administered a speeded number-comparison task to two groups differing in average g. The participants had to respond to two stimuli in each trial, which produced the well-known slowing of the second reaction time known as the psychological refractory period. The difference in the second reaction time favoring the high-g group doubled as the stimulus onsets became very close together. This finding affirms that the faster reaction times of higher-g individuals reflect an advantage exclusively in the serial bottleneck of central processing and not in the parallel peripheral stages.

Pruned Bit-Reversal Permutations: Mathematical Characterization, Fast Algorithms and Architectures

A mathematical characterization of serially pruned permutations (SPPs) employed in variable-length permuters and their associated fast pruning algorithms and architectures are proposed. Permuters are used in many signal processing systems for shuffling data and in communication systems as an adjunct to coding for error correction. Typically, only a small set of discrete permuter lengths are supported. Serial pruning is a simple technique to alter the length of a permutation to support a wider range of lengths, but results in a serial processing bottleneck. In this paper, parallelizing SPPs is formulated in terms of recursively computing sums involving integer floor functions using integer operations, in a fashion analogous to evaluating Dedekind sums. A mathematical treatment for bit-reversal permutations (BRPs) is presented, and closed-form expressions for BRP statistics including descents/ascents, major index, excedances/descedances, inversions, and serial correlations are derived. It is shown that BRP sequences have weak correlation properties. Moreover, a new statistic called permutation inliers that characterizes the pruning gap of pruned interleavers is proposed. Using this statistic, a recursive algorithm that computes the minimum inliers count of a pruned BR interleaver (PBRI) in logarithmic time is presented. This algorithm enables parallelizing a serial PBRI algorithm by any desired parallelism factor by computing the pruning gap in lookahead rather than a serial fashion, resulting in significant reduction in interleaving latency and memory overhead. Extensions to 2-D block and stream interleavers, as well as applications to pruned fast Fourier transforms and LTE turbo interleavers, are also presented. Moreover, hardware-efficient architectures for the proposed algorithms are developed. Simulation results of interleavers employed in modern communication standards demonstrate three to four orders of magnitude improvement in interleaving time compared to existing approaches.

Rhythmic Fluctuations in Evidence Accumulation during Decision Making in the Human Brain

Categorical choices are preceded by the accumulation of sensory evidence in favor of one action or another. Current models describe evidence accumulation as a continuous process occurring at a constant rate, but this view is inconsistent with accounts of a psychological refractory period during sequential information processing. During multisample perceptual categorization, we found that the neural encoding of momentary evidence in human electrical brain signals and its subsequent impact on choice fluctuated rhythmically according to the phase of ongoing parietal delta oscillations (1-3 Hz). By contrast, lateralized beta-band power (10-30 Hz) overlying human motor cortex encoded the integrated evidence as a response preparation signal. These findings draw a clear distinction between central and motor stages of perceptual decision making, with successive samples of sensory evidence competing to pass through a serial processing bottleneck before being mapped onto action.

Rhythmic fluctuations in evidence accumulation during decision making in the human brain

Categorical choices are preceded by the accumulation of sensory evidence in favor of one action or another. Current models describe evidence accumulation as a continuous process occurring at a constant rate, but this view is inconsistent with accounts of a psychological refractory period during sequential information processing. During multisample perceptual categorization, we found that the neural encoding of momentary evidence in human electrical brain signals and its subsequent impact on choice fluctuated rhythmically according to the phase of ongoing parietal delta oscillations (1-3 Hz). By contrast, lateralized beta-band power (10-30 Hz) overlying human motor cortex encoded the integrated evidence as a response preparation signal. These findings draw a clear distinction between central and motor stages of perceptual decision making, with successive samples of sensory evidence competing to pass through a serial processing bottleneck before being mapped onto action.

NoC-Based Hardware Accelerator for Breakpoint Phylogeny

Maximum Parsimony phylogenetic tree reconstruction is based on finding the breakpoint median, given a set of species, and is represented by a bounded edge-weight graph model. This reduces the breakpoint median problem to one of solving multiple instances of the Traveling Salesman Problem (TSP), which is a classical NP-complete problem in graph theory. Exponential time algorithms that apply efficient runtime heuristics, such as branch-and-bound, to dynamically prune the search space are used to solve TSP. In this paper, we present the design and performance evaluation of a network-on-chip (NoC)-based implementation for solving TSP under the bounded edge-weight model, as used in the computation of breakpoint phylogeny. Our approach takes advantage of fine-grain parallelism from the multiple processing elements (PEs) and uses efficient NoC architecture for inter-PE communication. To accelerate the application on hardware, our PE design optimizes a particular lower bound calculation operation which typically tends to be the serial bottleneck in computation of a TSP solution. We also explore two representative NoC architectures-mesh and quad-tree-and show that the latter is more energy-efficient for this application domain. Experimental results show that this new implementation is able to achieve speedups of up to three orders of magnitude over state-of-the-art multithreaded software implementations.

Parallel lookahead algorithms for pruned interleavers

In this letter, the design of efficient parallel pruned channel and turbo interleavers for Ultra Mobile Broadband (UMB) physical layer standard [1] is addressed. Channel interleaving is based on a bit-reversal algorithm in which addresses are mapped from linear order into bit-reversed order. Turbo interleaving is based on filling a 2D array row by row, interleaving each row independently using a linear congruential sequence (LCS), bit-reversing the order of the rows, and then reading the interleaved addresses column by column. To accommodate for flexible codeword lengths L, interleaving is done using a mother interleaver of length M = 2 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</sup> , where n is the smallest integer such that L ⩽ M, such that outlier interleaved addresses greater than L - 1 get pruned away. This pruning operation creates a serial bottleneck since the interleaved address of a linear address χ is now a function of the interleaving operation as well as the number of pruned addresses up to χ. A generic parallel lookahead pruned interleaving scheme that breaks this dependency is proposed. The efficiency of the proposed scheme is demonstrated in the context of both UMB interleavers. An iterative pruned bit-reversal algorithm that interleaves any address in O(log L) steps is presented. Moreover, an iterative pruned turbo interleaving algorithm based on LCSs that interleaves any address in O(log <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> L) steps is presented.

Brain Mechanisms of Serial and Parallel Processing during Dual-Task Performance

The psychological refractory period (PRP) refers to the fact that humans typically cannot perform two tasks at once. Behavioral experiments have led to the proposal that, in fact, peripheral perceptual and motor stages continue to operate in parallel, and that only a central decision stage imposes a serial bottleneck. We tested this model using neuroimaging methods combined with innovative time-sensitive analysis tools. Subjects performed a dual-task visual-auditory paradigm in which a delay of 300 ms was injected into the auditory task either within or outside of the dual-task interference period. Event-related potentials indicated that the first approximately 250 ms of processing were insensitive to dual-task interference, and that the PRP was mainly reflected in a delayed global component. By a clustering analysis based on time-resolved functional magnetic resonance imaging, we identified networks with qualitatively different timing properties: sensory areas tracked the objective time of stimulus presentation, a bilateral parietoprefrontal network correlated with the PRP delay, and an extended bilateral network that included bilateral posterior parietal cortex, premotor cortex, supplementary motor area, anterior part of the insula, and cerebellum was shared by both tasks during the extent of dual-task performance. The results provide physiological evidence for the coexistence of serial and parallel processes within a cognitive task.

Repeated Bottleneck Transfers Can Lead to Non-cytocidal Forms of a Cytopathic Virus: Implications for Viral Extinction

Several biological subclones of a biological clone of foot-and-mouth disease virus (FMDV) have been subjected to many plaque-to-plaque (serial bottleneck) transfers in cell culture. At transfer 190 to 409, clones underwent a transition towards a non-cytolytic (NC) phenotype in which the virus was unable to produce plaques, representing at least a 140-fold reduction in specific infectivity relative to the parental biological clone. NC clones, however, were competent in RNA replication and established a persistent infection in cell culture without an intervening cytolytic phase. In one clone, the transition to the NC phenotype was associated with the elongation of an internal oligodenylate tract that precedes the second functional AUG translation initiation codon. The pattern of mutations and their distribution along the FMDV genome of the clones subjected to serial bottleneck transfers were compared with the pattern of mutations in FMDV clones subjected to large population passages. Both the corrected ratios of non-synonymous to synonymous mutations and some specific mutations in coding and non-coding regions suggest participation of positive selection during large population passages and not during bottleneck transfers. Some mutations in the clones that attained the NC phenotype were located in genomic regions affecting the capacity of FMDV to kill BHK-21 cells. The resistance to extinction of clones subjected to plaque-to-plaque transfers marks a striking contrast with regard to the ease of extinction mediated by lethal mutagenesis. The results document a major phenotypic transition of a virus as a result of serial bottleneck events.

Contrasted patterns of genetic variation in the dogwhelk Nucella lapillus along two putative post-glacial expansion routes

During the last glacial maximum, the dogwhelk Nucella lapillus, like many other rocky shore species, was eradicated from the west and largely confined to southern refugia in the east Atlantic. Using microsatellite markers, we studied the geographical distribution of population genetic diversity of N. lapillus along 2 putative post-glacial expansion routes: northward along the east shores of Europe, and westward across the via Iceland to North America. The patterns of genetic diversity and genetic structure along these 2 routes indicate that the trans- Atlantic expansion involved a stepping-stone migration from Europe to Iceland to North America, with rare long-distance dispersal causing serial bottleneck effects. Along the east route, however, long-distance dispersal seems to have occurred more often. These observations, under- pinned by hydrography, suggest that rafting could be a major agent of dispersal for marine species lacking planktonic life stages.