Perceptual Boundaries Research Articles

Enhanced boundary perception and streamlined instance segmentation

Instance segmentation is a crucial task in computer vision that aims to simultaneously identify and segment individual objects within images. While existing approaches such as Mask R-CNN have shown promise, they often struggle with accurate boundary detection, especially for complex objects. In this paper, we introduce BorderMask, a novel framework that enhances boundary perception and streamlines instance segmentation. BorderMask comprises three key innovations: the Multiscale Boundary Perception Enhanced Attention (MBPEA) module, which iteratively optimizes multi-scale boundary features; the Cross-modal Link Structure (CMLS), which enables information exchange between detection and segmentation branches; and the Equilibrium Map loss function, which mitigates class imbalance issues. Extensive experiments on benchmark datasets including MS COCO, PASCAL VOC 2012, and Cityscapes demonstrate that BorderMask significantly outperforms state-of-the-art methods, achieving an AP of 44.7% on MS COCO, underscoring its robustness and effectiveness. The code will be available at https://gitee.com/shi-junyong/BorderMask.

Perceptual event boundaries cause mnemonic trade-offs between temporal order memory and source Memory: The role of semantic relatedness among items.

Perceptual event boundaries cause mnemonic trade-offs between temporal order memory and source Memory: The role of semantic relatedness among items.

Improved Semantic Segmentation of High‐Resolution Remote Sensing Data: Utilizing a Novel Enhanced Boundary‐Aware Network for Efficient Processing

ABSTRACTWith the recent advancements in deep learning and remote sensing technology, several semantic segmentation methods based on convolutional neural networks have been utilized for the semantic segmentation of remote sensing images. However, these models tend to lose valuable shallow detail information, particularly boundary information, due to feature compression during the encoding process. To address this issue, we propose an enhanced semantic segmentation model, EBANet, that focuses on boundary perception. EBANet consists of two parts—boundary path and spatial path, with boundary prediction serving as an independent subtask to enhance the segmentation model. Our model effectively aggregates boundary features and semantic features to achieve boundary enhancement segmentation. To achieve this, we employ a boundary filtering module (BFM) based on the gating mechanism that allows gating the low‐level boundary features through high‐level semantic understanding, thus effectively eliminating noisy data. The boundary‐guided contextual aggregation module (BCAM) enhances the interaction of boundary features with semantic features by establishing non‐local response relationships from boundary features to semantic features. We evaluated EBANet on high‐resolution remote sensing datasets Vaihingen and Potsdam, and the results showed that EBANet outperformed other models in terms of mIoU, F1‐Score, and OA, achieving the highest accuracy levels. On the Vaihingen dataset, EBANet's accuracy levels reached 68.31%, 81.05%, and 84.31%, respectively, while on the Potsdam dataset, it achieved 73.06%, 86.37%, and 89.64%, respectively.

“Express Yourself!” But Which Self? Investigating Role Negotiations and Boundaries on Journalists’ Personal Instagram Accounts

ABSTRACT This study explores journalists’ role negotiations and perceptions of boundaries on their personal (yet public) Instagram profiles. While journalism research often assumes that such profiles function as professional self-branding tools, such assumptions overlook the platform culture of Instagram and the fact that many journalists – especially younger ones – initially used the platform for private purposes and may continue to do so. Drawing on 15 qualitative interviews, the study investigates the roles journalists believe to perform on their profiles as well as the relevance they attribute to it (RQ1) and the contextual factors that influence the performance of their journalistic role (RQ2). The findings reveal a wide spectrum: Some journalists treat their profiles as journalistic spaces, some consider them private, and others navigate in between. Role negotiations often occur in moments of context collapse, as audiences from private and professional spheres converge. In such moments, journalists reassess their self-presentation, which can result either in a consolidation of the private focus or a reorientation towards their professional identity. Four contextual factors were found to influence these negotiations: Perception of a representative function, felt responsibility through visibility, pressure to protect oneself and one's journalistic image, and the need for personal branding.

IEAM: Integrating Edge Enhancement and Attention Mechanism with Multi-Path Complementary Features for Salient Object Detection in Remote Sensing Images

Prominent target detection in optical remote sensing images (RSI-SOD) focuses on segmenting key targets that capture human attention. However, most SOD methods prioritize detection accuracy at the cost of memory. Complex backgrounds, occlusions, and noise distort segmented target boundaries, while large memory demands increase computational cost, and reduced memory impairs segmentation accuracy. To address these challenges, we integrate edge enhancement and attention mechanisms with multi-path complementary features for salient object detection in remote sensing images (IEAM), aiming to improve salient target accuracy, boundary detection, and memory efficiency. The architecture utilizes a structured feature fusion strategy, combining spatial channel attention mechanisms with adaptive merging to enhance multi-scale feature representation and suppress background noise. The Spatially Adaptive Edge Embedded Module (SAEM) refines object boundary perception, the SCAAP module dynamically selects relevant spatial and channel features while balancing adaptive and maximal pooling, and the Spatial Adaptive Guidance (SAG) module enhances feature localization in cluttered environments to mitigate semantic dilution in U-shaped networks. Extensive experiments on the EORSSD and ORSSD benchmark datasets demonstrate that IEAM outperforms 21 state-of-the-art methods, achieving an inference speed of 48 FPS at 103.2 G FLOP, making it suitable for real-time applications. The proposed model is robust and excels in multiple aspects.

How final is final: The production and perception of utterance-medial and utterance-final boundaries

We examine the production and perception of two types of phrase-final prosodic boundaries, specifically, utterance-medial and utterance-final intonation phrase (IP) boundaries in German. These two types of boundaries are expected to differ in terms of general properties of the prosodic hierarchy, and properties of turn-taking and speech planning. In an articulatory magnetometer study and a perceptual rating study, we examine these boundaries in read speech, testing for temporal, spatial, and intonational properties of the boundaries. Only small and inconsistent differences were found in the temporal and spatial domains. The only robust difference is a lower f0 in the rhyme of the intonation contour for utterance-final IP boundaries compared to utterance-medial IP boundaries for five of the eight speakers. This is consistent with the results of the perception study, which indicate that listeners perceive subtle differences in the boundary-specific production, and that mean f0 during the rhyme and peak velocity were the information listeners used to determine utterance finality for the speakers producing the difference.

Symbolic boundary work among Syrian refugees: perceived stigmas, responses, and cultural repertoires

ABSTRACT Although the Turkish public initially welcomed Syrian refugees, signs of hardening boundaries have emerged over time. This study, based on in-depth interviews with Syrian refugees in Istanbul, examines both their perceptions of symbolic boundaries drawn against them by the majority Turkish society and their responses. The Syrian refugees in this study perceive three main stigmas: “uncultured,” “traitor,” and “freeloader.” They respond to these stigmas situationally by assuming individual responsibility, managing their self-presentation, and sometimes choosing not to respond. They also engage in discursive boundary drawing, emphasizing a strong work ethic as a defining characteristic of their group compared to Turkish workers. I argue that the historical cultural repertoire of anti-Arabism inherent in the conception of Turkish nationhood feeds these stigmas, especially in the context of the current economic crisis and political polarization. Meanwhile, the precarious position and fragile status of Syrian refugees constrain their situational responses. Their moral boundary work helps them reassert their dignity and legitimize their presence, but it also hinders potential solidarity with Turkish workers against a governing strategy that systematically exploits both groups.

Probing sensitivity to statistical structure in rapid sound sequences using deviant detection tasks.

Statistical structures and our ability to exploit them are a ubiquitous component of daily life. Yet, we still do not fully understand how we track these sophisticated statistics and the role they play in sensory processing. Predictive coding frameworks hypothesize that for stimuli that can be accurately anticipated based on prior experience, we rely more strongly on our internal model of the sensory world and are more "surprised" when that expectation is unmet. The present study used this phenomenon to probe listeners' sensitivity to probabilistic structures generated using rapid 50 ms tone-pip sequences that precluded conscious prediction of upcoming stimuli. Over three experiments, we measured listeners' sensitivity and response time to deviants of a frequency outside the expected range. Predictable sequences were generated using either a triplet-based or network-style structure, and deviant detection contrasted against the same set of tones but in a random, unpredictable order. All experiments found structured sequences enhanced deviant detection relative to random sequences. Additionally, Experiment 2 used three different instantiations of the community structure to demonstrate that the level of uncertainty in the structured sequences modulated deviant saliency. Finally, Experiment 3 placed the deviant within an established community or immediately after a transition between communities, where the perceptual boundary should generate momentary uncertainty. However, this manipulation did not impact performance. Together, these results demonstrate that probabilistic contexts generated from statistical structures modulate the processing of an ongoing auditory signal, leading to an improved ability to detect unexpected deviant stimuli, consistent with the predictive coding framework. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

On the mental representations of L2 English /z/ among L1 Chinese speakers.

On the mental representations of L2 English /z/ among L1 Chinese speakers.

Multi-perspective dynamic consistency learning for semi-supervised medical image segmentation

Semi-supervised learning (SSL) is an effective method for medical image segmentation as it alleviates the dependence on clinical pixel-level annotations. Among the SSL methods, pseudo-labels and consistency regularization play a key role as the dominant paradigm. However, current consistency regularization methods based on shared encoder structures are prone to trap the model in cognitive bias, which impairs the segmentation performance. Furthermore, traditional fixed-threshold-based pseudo-label selection methods lack the utilization of low-confidence pixels, making the model’s initial segmentation capability insufficient, especially for confusing regions. To this end, we propose a multi-perspective dynamic consistency (MPDC) framework to mitigate model cognitive bias and to fully utilize the low-confidence pixels. Specially, we propose a novel multi-perspective collaborative learning strategy that encourages the sub-branch networks to learn discriminative features from multiple perspectives, thus avoiding the problem of model cognitive bias and enhancing boundary perception. In addition, we further employ a dynamic decoupling consistency scheme to fully utilize low-confidence pixels. By dynamically adjusting the threshold, more pseudo-labels are involved in the early stages of training. Extensive experiments on several challenging medical image segmentation datasets show that our method achieves state-of-the-art performance, especially on boundaries, with significant improvements.

Personalized Neural State Segmentation: Validating the Greedy State Boundary Search Algorithm for Individual-level Functional Magnetic Resonance Imaging Data.

Humans segment experience into a nested series of discrete events, separated by neural state transitions that can be identified in fMRI data collected during passive movie viewing. Current neural state segmentation techniques manage the noisiness of fMRI data by modeling groups of participants at once. However, the perception of event boundaries is itself idiosyncratic. As such, we developed a denoising pipeline to separate meaningful signal from noise and validated the Greedy State Boundary Search algorithm for use in individual participants. We applied the Greedy State Boundary Search to publicly available (1) young adult and (2) developmental fMRI data sets. After extensive denoising, we confirmed that personalized young adult neural state transitions exhibited a canonical temporal cortical hierarchy and were related to normative behavioral boundaries across time in key regions such as posterior parietal cortex. Furthermore, we used machine learning to show that the strongest neural transitions from across cortex could be used to predict the timing of normative boundary judgments. Results from the developmental data set also demonstrated important boundary conditions for estimating personalized neural state transitions. Nonetheless, some brain-behavior relations were still apparent in individually modeled developmental data. Finally, we ran two individual differences analyses demonstrating the utility of our method. These validations pave the way for applying personalized fMRI modeling to the study of event segmentation; what meaningful insights could we be missing when we average away what makes each of us unique?

Prosodic boundary processing is language-dependent

This paper provides empirical data to support the claim that prosodic boundary processing is language-dependent. Prosody plays an important role in determining the meaning of the utterance. Despite their importance in speech recognition, prosodic boundaries in spontaneous speech are understudied, and the finding that syntactic and prosodic boundaries are not isomorphic complicates automatic speech recognition. This paper considers how prosodic boundaries are perceived in spontaneous speech via perception experiments. We will consider Japanese first. It is a mora-timed pitch language and typologically different from stress languages like English. Japanese prosody is complex; it is compositionally formed by the lexical pitch accents H*L, phrasal tones, and boundary tones (L%, H%, LH%, HL%). Though Japanese literature has a long history of prosodic studies on the word-level and the phrase-level, prosody above the φ-level is understudied and no standard view for phrasal patterns has ever been established. We conducted perception experiments on spontaneous Japanese via Rapid Prosody Transcription (RPT) and conducted multiple regression analyses between boundary marking and cue candidates. Our findings are that the primary cue for boundary perception in Japanese is post-boundary pause, followed by syntactic higher categories. On the other hand, prosody of stress languages is well-studied, and the studies claim that the prosodic boundary cues in such languages are either acoustic (e.g. French, etc.) or syntactic (e.g. American English, Estonian, etc.). If boundary cues vary between Japanese and American English, it is expected that Japanese learners of English process prosodic boundaries differently from native American English listeners. We did a comparative study and analyzed how native listeners and learners would perceive the same English natural speech. We reanalyzed the RPT data in Cole et al. 2010 and Mizuguchi et al. 2016 and found that native listeners and learners use different perception cues.

Effect of force-rate on continuous kinesthetic force discrimination.

The effect of force-rate, i.e., rate of change of force stimuli, on continuous kinesthetic force perception has not been investigated and remains an open area for research. Previous studies do not account for the force-rate in the Weber fraction of kinesthetic force discrimination. However, this force-rate agnostic Weber fraction fails to explain continuous kinesthetic force discrimination fully. For example, if the signal changes very slowly, a participant may get accustomed to the change, and hence, a larger, just noticeable difference (JND) is expected. Conversely, for a fast-varying haptic force, a smaller JND is expected. In this work, we aim to explore the relationship between the Weber fraction and the force-rate. For this purpose, we designed an extensive psychophysical experiment where a participant is exposed to a linearly increasing kinesthetic force stimulus and is asked to react to the change. We utilize a machine learning-based approach to study the effect of force-rate on recorded haptic responses (perceived and non-perceived) of 10 participants while varying the force-rate stimuli in the range [1,5] N/s. We determine the perceptual boundary between the perceived and non-perceived recorded responses using different classifiers based on linear and power functions of force-rate. The result indicates that the Weber fraction decreases significantly as the force-rate increases. The random forest classifier also confirms the significance of the utilized features in both perceptual boundaries. These findings may be useful in many virtual reality applications and telepresence and teleaction systems (TPTA).

Prima Donna in the Jungle: Opera, Class, and Race in RKO Pictures’ Hitting a New High (1937)

Abstract This article interrogates why critics panned RKO's Hitting a New High, an operatic musical starring French soprano Lily Pons in the role of an aspiring prima donna who assumes the guise of an “African bird girl” for a publicity stunt. Evidence shows that the film transgressed audience expectations due to (1) its depiction of a “lowbrow,” raced jungle queen, and (2) a shifting sociopolitical climate heightening perceptions of class boundaries and a Black/white binary.

Association of Risk-Taking Behaviors, Vestibular Provocation and Action Boundary Perception Following Sport-Related Concussion in Adolescents.

Background/Objectives: The purpose of this study was to evaluate the association between risk-taking behaviors, vestibular symptoms/impairment and perception-action coupling behavior in recently concussed adolescents. Methods: This study utilized a cross-sectional design to evaluate the early effects of concussion on 12-18-year-old adolescents (n = 47) recruited from a concussion specialty clinic at their presenting clinical appointment. The Perception-Action Coupling Task (PACT) was used to assess action boundary perception by evaluating the participant's ability to quickly and accurately determine whether a virtual "ball" fits in a virtual "hole". Accuracy, response time and inverse efficiency were evaluated at the 0.8 and 1.2 ratios of ball-hole pairings, where 0.8 indicates the ball was slightly smaller than the hole and 1.2 indicates the ball was slightly larger than the hole. The Balloon Analog Risk Task (BART) is a computerized test which measures risk-taking behavior by "pumping" up a balloon. Each pump provides a small amount of virtual money into their bank; the goal is to make as much money as possible without popping the virtual balloon. The Vestibular Ocular Motor Screening (VOMS) tool is a brief screening tool designed to identify ocular or vestibular dysfunction following sport-related concussion, where horizontal/vertical vestibular ocular reflex (VOR) and visual motion sensitivity (VMS) are the primary vestibular outcomes. Pearson correlation matrices were developed to evaluate the association between BART, VOMS and PACT outcomes within the study cohort of concussed adolescents. Results: PACT inverse efficiency at the 1.2 ball-hole ratio was significantly correlated with all three VOMS outcomes (r = 0.33-0.37). The standard deviation of pump reaction time during BART was significantly correlated with accuracy (r = -0.47) and inverse efficiency (r = 0.42) at the 1.2 ratio. The standard deviation of the total number of pumps during BART was significantly correlated with PACT response time at the 1.2 ratio (r = 0.34). Horizontal VOR correlated with balloons collected (r = -0.30) and balloons popped (r = -0.30). Conclusions: The results of this study suggest that risk-taking behaviors and vestibular symptoms/impairment are associated with worse action boundary perception in adolescents following concussion. This relationship is more pronounced in male adolescents than females.

Blurry dense SAR object detection algorithm based on collaborative boundary refinement and differential feature enhancement

ABSTRACT The coherent speckle noise of the SAR image blurs the boundaries between the objects. In dense scenes, the objects are easy to stick and overlap, making it difficult to be accurately positioned and distinguished. In order to detect blurry and dense SAR objects, this paper proposes a collaborative boundary refinement network based on differential feature enhancement strategy. Firstly, a collaborative structure for applying unsupervised denoising algorithms in object detection networks is developed. Mutual information between original image and low-noise image features is learnt through the collaboration of multiple boundary refinement units and coherent patch noise filters. The boundary refinement unit captures features that are insensitive to noise and recovers the boundary details of blurred dense objects. Additionally, to improve the detailed representation, a fine-grained differential feature enhancement module is built to mine the mutations and detailed features of the edge context by using the differential information. At the same time, the decoupled detection head is used to separate the classification and positioning tasks. And as a result, the enhanced boundary perception features are paid more attention and precise positioning of blurry and dense objects can be achieved. The proposed algorithm is verified on a typical blurry dense SAR object dataset. The experimental results show that the performance of the proposed method is better than that of the existing single-modal methods and representative cross-modal methods.

Does Working Out-of-the-Office Work for Employees? Alternative Work Arrangements, Ambiguity, and Job Satisfaction in the Federal Workforce

Public employees working away from the office has become more common in recent years, but the impact of these changes is not fully understood. This study considers whether out-of-office work culture impacts how employees come to make sense of their organizations, jobs, and work life, and, in turn, how this may affect job satisfaction. Using Stata 15’s random sampling tool, the authors employ Structural Equation Modeling of three random samples of 2,000 respondents from the 2022 Federal Employee Viewpoint Survey. Findings indicate that employees are likely to perceive more goal clarity and support for work–life balance where out-of-office work proliferates, leading to higher job satisfaction. By altering how employees interact with their colleagues, working away from the office often leaves employees with clearer perceptions of organizational goals and boundaries around their work lives. In turn, this supports higher job satisfaction as public servants have a stronger understanding of their work identities. This adds further dimension to understanding how the push toward telework or similar arrangements impacts organizations and employees in the public sector.

Children’s emerging sociolinguistic expectations around social roles: a triangulated approach

Abstract This paper adds to current work on developmental sociolinguistics and the sociopragmatic turn in anglicism research through a triangulated design meant to uncover children’s sociolinguistic expectations of social roles. A sample of 25 participants, aged between 7 and 13, completed three tasks probing assumptions on the use of English elements for three English-oriented roles (rapper, gamer, and soccer player) and two Dutch-oriented roles (prime minister and farmer). Specifically, participants took part in a closed rating task, a role-play performance task, and an open-ended interview. The results show that children share adult expectations of Dutch- versus English-oriented roles (RQ1), which they implement in their role performances, using more English words for the latter (RQ2). Participants, lastly, reflect on their performance to varying degrees, reporting changes to their language use in those performances, though only the oldest girls specifically mention English words (RQ3). Overall, the comparison of the participants and tasks invites a reflection on developmental pathways and on the relationship between metalinguistic awareness, sociolinguistic expectations, and the perception of boundaries between languages or varieties.

The Interaction Between Vowel Quality and Intensity in Loudness Perception of Short Vowels in Mongolian.

This study investigated the influence of vowel quality on loudness perception and stress judgment in Mongolian, an agglutinative language with free word stress. We aimed to explore the effects of intrinsic vowel features, presentation order, and intensity conditions on loudness perception and stress assignment. Eight Mongolian short vowel phonemes (/ɐ/, /ə/, /i/, /ɪ/, /ɔ/, /o/, /ʊ/, and /u/) were recorded by a native Mongolian speaker of the Urad subdialect (the Chahar dialect group) in Inner Mongolia. The short vowels were paired under different intensity conditions. Native Mongolian listeners from Inner Mongolia participated in two loudness perception experiments: Experiment 1 examined the effects of presentation order and different intensity conditions on loudness perception using pairs of vowels. Experiment 2 explored how different vowel pairs influence perceptual outcomes and identified specific thresholds and perceptual boundaries for loudness perception. The findings revealed that intensity significantly affected loudness perception, modulated by vowel quality. Presentation order of vowels affected loudness perception, and vowel centralization and lip rounding play crucial roles as well. Central vowels, particularly /ə/, were perceived as more prominent, whereas rounded vowels were more likely to be judged as stressed under equated intensity conditions. The study also identified a perceptual tendency toward final prominence, influenced by sonority and vowel positioning. This study highlights the intricate relationship among vowel quality, intensity, and stress perception in Mongolian. Different vowels exhibited distinct loudness perceptions at the same intensity level, emphasizing the importance of vowel quality in stress assignment. Vowels with higher sonority indices or those positioned peripherally in the vowel space are more likely to be perceived as prominent. These findings contribute to a broader understanding of the phonological processes and perceptual mechanisms in agglutinative languages and highlight the need for further research across diverse dialects.

A proof-of-concept methodology for identifying topical scientific issues in new publications whose citations have not yet been established

Identification of topical research issues using bibliometric data is complicated by the fact that the citation of publications from recent years has not yet been formed. In this paper, it is proposed to use the average citation of the journal over two years rather than the article citation to estimate to estimate the weight of the keyword occurring in the sample under consideration. In order to identify the terms that characterize relevant research topics, it is proposed to represent the term co-occurrence network in coordinates of the average occurrence of the term per year and the average normalized citation of the term to visualize the graph. Furthermore, this methodology proposes the use of preprocessing of keywords using a lemmatization dictionary. 3,696 bibliometric records for 2022–2024 from the ScienceDirect platform on the topic of industry digitalization were used for the analysis. The VOSviewer and Scimago Graphica programs were used sequentially. The former was used to display the overall landscape of the study, while the latter was used to analyze in more detail the individual slices of bibliometric data obtained with VOSviewer. A ‘convex hull’ was used to facilitate the perception of cluster boundaries. After analysing the data and highlighting the terms, it is proposed to provide context by quoting strings from publications and defining of lesser-known terms. The industry digitalization is not only a technical and technological issue but also an economic one, as evidenced by terms such as ‘digital economy’ and ‘Industry 5.0’.