1
|
Mishra B, Tarai S, Ratre V, Bit A. Processing of attentional and emotional stimuli depends on retrospective response of foot pressure: Conceptualizing neuron-cognitive distribution in human brain. Comput Biol Med 2023; 164:107186. [PMID: 37480678 DOI: 10.1016/j.compbiomed.2023.107186] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/24/2023] [Revised: 06/08/2023] [Accepted: 06/13/2023] [Indexed: 07/24/2023]
Abstract
Cognitive function of human brain requires temporal execution of emotional or attentional tasks, or their inter-dependence influences. Smooth execution of such tasks requires spontaneous distribution of cognitive load at specific regions of brain based on its classification. A strong connectivity between peripheral sensors and central nervous system is thought to assist the cognitive load distribution effectively. Novelty of current study evaluates the modulation of foot pressure and its mapping with distributed cognitive load while executing attentional biased emotional tasks. Emotional stimulus in form of happy and sad faces with attentional paradigm drawn on them were used in the study. Behavioral results were measured with respect to the analysis of response time (RT) and response accuracy (%). Neurological signals were acquired using 10-channel EEG data acquisition system, whereas, another 6 channels were used to measure foot pressure in the left and right feet at three different locations of foot. Acquired signals were further analyzed in time and frequency domains to interpret the cognitive load distribution, and the influence of foot pressure on distribution of cognitive loads. We found that the foot pressure accelerated the response accuracy rate in attending the local scope of attention, which was not in the case of global scope of attention. This means that the global attention does not require any pressure from peripheral sensory neurons. Our event related potential (ERP) results revealed that the early sensory negative N100 characterized the processing of global scope of attention coupled with high-foot pressure. However, the late positive peak of P300 and P600 associated with local scope of attention along with high-foot pressure. The global scope of attention with low-foot pressure modulates delta and theta oscillations. These results largely contribute to the literature on cognitive neuroscience of attention and it corelation with the peripheral sensory foot pressure.
Collapse
Affiliation(s)
- Bharti Mishra
- Department of Biomedical Engineering, National Institute of Technology, Raipur, India
| | - Shashikanta Tarai
- Department of Humanities and Social Sciences, National Institute of Technology, Raipur, India
| | - Vinod Ratre
- Department of Biomedical Engineering, National Institute of Technology, Raipur, India
| | - Arindam Bit
- Department of Biomedical Engineering, National Institute of Technology, Raipur, India.
| |
Collapse
|
2
|
Shi R, Bi K, Du K, Ma L, Fang F, Duan L, Jiang T, Huang T. PS-Net: human perception-guided segmentation network for EM cell membrane. Bioinformatics 2023; 39:btad464. [PMID: 37505461 PMCID: PMC10423022 DOI: 10.1093/bioinformatics/btad464] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/12/2023] [Revised: 05/19/2023] [Accepted: 07/27/2023] [Indexed: 07/29/2023] Open
Abstract
MOTIVATION Cell membrane segmentation in electron microscopy (EM) images is a crucial step in EM image processing. However, while popular approaches have achieved performance comparable to that of humans on low-resolution EM datasets, they have shown limited success when applied to high-resolution EM datasets. The human visual system, on the other hand, displays consistently excellent performance on both low and high resolutions. To better understand this limitation, we conducted eye movement and perceptual consistency experiments. Our data showed that human observers are more sensitive to the structure of the membrane while tolerating misalignment, contrary to commonly used evaluation criteria. Additionally, our results indicated that the human visual system processes images in both global-local and coarse-to-fine manners. RESULTS Based on these observations, we propose a computational framework for membrane segmentation that incorporates these characteristics of human perception. This framework includes a novel evaluation metric, the perceptual Hausdorff distance (PHD), and an end-to-end network called the PHD-guided segmentation network (PS-Net) that is trained using adaptively tuned PHD loss functions and a multiscale architecture. Our subjective experiments showed that the PHD metric is more consistent with human perception than other criteria, and our proposed PS-Net outperformed state-of-the-art methods on both low- and high-resolution EM image datasets as well as other natural image datasets. AVAILABILITY AND IMPLEMENTATION The code and dataset can be found at https://github.com/EmmaSRH/PS-Net.
Collapse
Affiliation(s)
- Ruohua Shi
- Advanced Institute of Information Technology, Peking University, Hangzhou, Zhejiang 310000, China
- National Engineering Research Center of Visual Technology, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University, Beijing 100871, China
- Beijing Academy of Artificial Intelligence, Beijing 100084, China
| | - Keyan Bi
- School of Psychological and Cognitive Sciences and Beijing Key Laboratory of Behavior and Mental Health, Peking University, Beijing 100871, China
- IDG/McGovern Institute for Brain Research, School of Psychological and Cognitive Sciences, Peking University, Beijing 100871, China
- Key Laboratory of Machine Perception (Ministry of Education), Peking University, Beijing 100871, China
- Peking-Tsinghua Center for Life Sciences, Peking University, Beijing 100084, China
| | - Kai Du
- Institute for Artificial Intelligence, Peking University, Beijing 100871, China
| | - Lei Ma
- National Engineering Research Center of Visual Technology, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University, Beijing 100871, China
- Beijing Academy of Artificial Intelligence, Beijing 100084, China
- National Biomedical Imaging Center, College of Future Technology, Peking University, Beijing 100871, China
| | - Fang Fang
- School of Psychological and Cognitive Sciences and Beijing Key Laboratory of Behavior and Mental Health, Peking University, Beijing 100871, China
- IDG/McGovern Institute for Brain Research, School of Psychological and Cognitive Sciences, Peking University, Beijing 100871, China
- Key Laboratory of Machine Perception (Ministry of Education), Peking University, Beijing 100871, China
- Peking-Tsinghua Center for Life Sciences, Peking University, Beijing 100084, China
| | - Lingyu Duan
- National Engineering Research Center of Visual Technology, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University, Beijing 100871, China
- Peng Cheng Laboratory, Shenzhen 518066, China
| | - Tingting Jiang
- Advanced Institute of Information Technology, Peking University, Hangzhou, Zhejiang 310000, China
- National Engineering Research Center of Visual Technology, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University, Beijing 100871, China
| | - Tiejun Huang
- National Engineering Research Center of Visual Technology, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University, Beijing 100871, China
- Beijing Academy of Artificial Intelligence, Beijing 100084, China
| |
Collapse
|
3
|
Lefebvre S, Beaucousin V. Seeing the forest or the tree depends on personality: Evidence from process communication model during global/local visual search task. PLoS One 2023; 18:e0284596. [PMID: 37083695 PMCID: PMC10121018 DOI: 10.1371/journal.pone.0284596] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/20/2022] [Accepted: 04/03/2023] [Indexed: 04/22/2023] Open
Abstract
In everyday life, we are continuously confronted with multiple levels of visual information processes (e.g., global information, the forest, and local information, the tree) and we must select information that has to be processed. In the present study, we investigated the relation between personality and the ability to process global and local visual information. Global precedence phenomenon was assessed by a standard global/local visual search task used in many visuo-spatial precedent studies, and the 77 participants were also presented with the standard Process Communication Model (PCM) questionnaire. Results suggest that the ability to process global and local properties of visual stimuli varied according to the Base type of participants. Even if four among six Base types (Thinker, Persister, Harmonizer and Promoter) presented a classical global visual precedence, the two other Base types (Rebel and Imaginer) presented only an effect of distractors and an effect of global advantage, respectively. Taken together, these results evidenced that each human being does not equally perceive the "forest" (global information) and the "tree" (local information). Even if objectively presented with similar visual stimuli, individual responses differ according to the Base, an inter-individual variability that could be taken into account during daily life situations.
Collapse
Affiliation(s)
- Sixtine Lefebvre
- Psychologist, PCM Trainer, PCM R&D Projects, Croisy-sur-Eure, France
| | | |
Collapse
|
4
|
Karim AKMR, Proulx MJ, de Sousa AA, Likova LT. Do we enjoy what we sense and perceive? A dissociation between aesthetic appreciation and basic perception of environmental objects or events. COGNITIVE, AFFECTIVE & BEHAVIORAL NEUROSCIENCE 2022; 22:904-951. [PMID: 35589909 PMCID: PMC10159614 DOI: 10.3758/s13415-022-01004-0] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Accepted: 03/27/2022] [Indexed: 05/06/2023]
Abstract
This integrative review rearticulates the notion of human aesthetics by critically appraising the conventional definitions, offerring a new, more comprehensive definition, and identifying the fundamental components associated with it. It intends to advance holistic understanding of the notion by differentiating aesthetic perception from basic perceptual recognition, and by characterizing these concepts from the perspective of information processing in both visual and nonvisual modalities. To this end, we analyze the dissociative nature of information processing in the brain, introducing a novel local-global integrative model that differentiates aesthetic processing from basic perceptual processing. This model builds on the current state of the art in visual aesthetics as well as newer propositions about nonvisual aesthetics. This model comprises two analytic channels: aesthetics-only channel and perception-to-aesthetics channel. The aesthetics-only channel primarily involves restricted local processing for quality or richness (e.g., attractiveness, beauty/prettiness, elegance, sublimeness, catchiness, hedonic value) analysis, whereas the perception-to-aesthetics channel involves global/extended local processing for basic feature analysis, followed by restricted local processing for quality or richness analysis. We contend that aesthetic processing operates independently of basic perceptual processing, but not independently of cognitive processing. We further conjecture that there might be a common faculty, labeled as aesthetic cognition faculty, in the human brain for all sensory aesthetics albeit other parts of the brain can also be activated because of basic sensory processing prior to aesthetic processing, particularly during the operation of the second channel. This generalized model can account not only for simple and pure aesthetic experiences but for partial and complex aesthetic experiences as well.
Collapse
Affiliation(s)
- A K M Rezaul Karim
- Department of Psychology, University of Dhaka, Dhaka, 1000, Bangladesh.
- Envision Research Institute, 610 N. Main St., Wichita, KS, USA.
- The Smith-Kettlewell Eye Research Institute, 2318 Fillmore St., San Francisco, CA, USA.
| | | | | | - Lora T Likova
- The Smith-Kettlewell Eye Research Institute, 2318 Fillmore St., San Francisco, CA, USA
| |
Collapse
|
5
|
Charles Leek E, Leonardis A, Heinke D. Deep neural networks and image classification in biological vision. Vision Res 2022; 197:108058. [PMID: 35487146 DOI: 10.1016/j.visres.2022.108058] [Citation(s) in RCA: 6] [Impact Index Per Article: 3.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/14/2021] [Revised: 04/12/2022] [Accepted: 04/13/2022] [Indexed: 10/18/2022]
Abstract
In this paper we consider recent advances in the use of deep convolutional neural networks to understanding biological vision. We focus on claims about the plausibility of feedforward deep convolutional neural networks (fDCNNs) as models of image classification in the biological system. Despite the putative similarity of these networks to some properties of the biological vision system, and the remarkable levels of performance accuracy of some fDCNNs, we argue that their plausibility as a framework for understanding image classification remains unclear. We highlight two key issues that we suggest are relevant to the evaluation of any form of DNN used to examine biological vision: (1) Network transparency under analysis - that is, the challenge of understanding what networks do, and how they do it. (2) Identifying appropriate benchmarks for comparing network performance and the biological system using both quantitative and qualitative performance measures. We show that there are important divergences between fDCNNs and biological vision that reflect fundamental differences in computational architectures, and representational structures, supporting image classification in these networks and the biological system.
Collapse
Affiliation(s)
| | | | - Dietmar Heinke
- School of Computer Science, University of Birmingham, UK
| |
Collapse
|
6
|
Chen T, Ding J, Yue GH, Liu H, Li J, Jiang C. Global-local consistency benefits memory-guided tracking of a moving target. Brain Behav 2022; 12:e2444. [PMID: 34859605 PMCID: PMC8785627 DOI: 10.1002/brb3.2444] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 11/22/2020] [Revised: 10/31/2021] [Accepted: 11/08/2021] [Indexed: 11/11/2022] Open
Abstract
INTRODUCTION Previous findings have demonstrated that several Gestalt principles do facilitate VSTM performance in change detection tasks. However, few studies have investigated the role of and time-course of global-local consistency in motion perception. METHODS Participants were required to track a moving target surrounded by three different backgrounds: blank, inconsistent, or consistent. Global-local objects were be bound to move together (covariation). During the PMT, participants had to follow the moving target with their eyes and react as fast as possible when the target had just vanished behind the obstruction or would arrive at a predetermined point of interception. Variable error (VE) and constant error (CE) of estimated time-to-contact (TTC) and gain of smooth pursuit eye movements were calculated in various conditions and analyzed qualitatively. RESULTS Experiment 1 established the basic finding that VSTM performance could benefit from global-local consistency. Experiment 2 extended this finding by eye-tracking device. Both in visible phase and in occluded phase, CEs were smaller for the target in a consistent background than for the target in an inconsistent background and for the target in a blank background, with both differences significant (ps < .05). However, the difference in VE among three conditions was not significant. At early stage (100-250 ms), later stage (2750-3000 ms), and termination stage (5750-6000 ms) of smooth pursuit, the velocity gains were higher in the trials with consistent backgrounds than in the trials with inconsistent backgrounds and blank backgrounds (ps < .001). With the exception of 100-250 ms phase, the means did not differ between the inconsistent background and the blank background trials (ps > .1). CONCLUSIONS Global-local consistency could be activated within the first few hundred milliseconds to prioritize the deployment of attention and eye movement to component target. Meanwhile, it also removes ambiguity from motion tracking and TTC estimation under some unpredictable conditions, leading to the consistency advantage during smooth-pursuit termination phase. Global-local consistency may act as an important information source to TTC estimation and oculomotor response in PMT.
Collapse
Affiliation(s)
- Tingting Chen
- School of EducationBeijing Dance AcademyBeijingP.R. China
| | - Jinhong Ding
- Beijing Key Laboratory of Learning and Cognition & School of PsychologyCapital Normal UniversityBeijingP.R. China
| | - Guang H. Yue
- Human Performance and Engineering Research, Kessler FoundationWest OrangeNew Jersey
| | - Haoqiang Liu
- School of EducationShangdong Woman UniversityJinanP.R. China
| | - Jie Li
- Institute of Psychological SciencesHangzhou Normal UniversityHangzhouP.R. China
| | - Changhao Jiang
- Beijing Key Lab of Physical Fitness Evaluation and Tech AnalysisCapital University of Physical Education and SportsBeijingP.R. China
| |
Collapse
|
7
|
Electrophysiological correlates of visual attention span in Chinese adults with poor reading fluency. Exp Brain Res 2021; 239:1987-1999. [PMID: 33893841 DOI: 10.1007/s00221-021-06115-7] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/14/2020] [Accepted: 04/15/2021] [Indexed: 10/21/2022]
Abstract
Individuals with reading fluency difficulty (RFD) show an impairment in the simultaneous processing of multiple elements, which could be reflected in their visual attention span (VAS) capacity. However, the relationship between VAS impairment and RFD is still controversial. A series of processes underlie VAS, such as the early stage of visual attentional processing and the late stage of allocating and maintaining attentional resources. Therefore, the present study explored the relationships between VAS skills and RFD through the event-related potential (ERP) technique to disentangle the contributing cognitive processes regarding VAS from a temporal perspective. Eighteen Chinese adults with poor reading fluency and 18 age-matched normal readers participated. Their VAS skills were measured by a visual one-back task with symbols as nonverbal stimuli and key pressing as nonverbal responses, while relevant electrophysiological signals were recorded. The results showed that lower d' values and abnormal electrophysiological activities (especially weak amplitudes in the N1 and P3 components) in the VAS task were observed for the nonfluent readers compared with the controls. These findings suggested that the low VAS capacity in adults with poor reading fluency could be reflected by problems both in directing selective attention to visually discriminate stimuli within a multielement string at the early processing stage and in allocating attention to further encode targets at the late processing stage. Alternative explanations were further discussed. The current results provide theoretical explanations of the VAS-RFD relationship from a temporal perspective and provide insights for future remediation of reading fluency difficulty.
Collapse
|
8
|
Pitchford B, Arnell KM. Individual Differences in Attentional Breadth Changes Over Time: An Event-Related Potential Investigation. Front Psychol 2021; 12:605250. [PMID: 33833706 PMCID: PMC8021726 DOI: 10.3389/fpsyg.2021.605250] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/14/2020] [Accepted: 03/02/2021] [Indexed: 11/13/2022] Open
Abstract
Event-related potentials (ERPs) to hierarchical stimuli have been compared for global/local target trials, but the pattern of results across studies is mixed with respect to understanding how ERPs differ with local and global bias. There are reliable interindividual differences in attentional breadth biases. This study addresses two questions. Can these interindividual differences in attentional breadth be predicted by interindividual ERP differences to hierarchical stimuli? Can attentional breadth changes over time within participants (i.e., intraindividual differences) be predicted by ERPs changes over time when viewing hierarchical stimuli? Here, we estimated attentional breadth and isolated ERPs in response to Navon letter stimuli presented at two time points. We found that interindividual differences in ERPs at Time 1 did not predict attentional breadth differences across individuals at Time 1. However, individual differences in changes to P1, N1, and P3 ERPs to hierarchical stimuli from Time 1 to Time 2 were associated with individual differences in changes in attentional breadth from Time 1 to Time 2. These results suggest that attentional breadth changes within individuals over time are reflected in changes in ERP responses to hierarchical stimuli such that smaller N1s and larger P3s accompany a shift to processing the newly prioritized level, suggesting that the preferred level required less perceptual processing and elicited more attention.
Collapse
Affiliation(s)
- Brent Pitchford
- Department of Psychology, Brock University, St. Catharines, ON, Canada
| | - Karen M Arnell
- Department of Psychology, Brock University, St. Catharines, ON, Canada
| |
Collapse
|
9
|
Scrivano RM, Kieffaber PD. Behavioral and electrophysiological correlates of Simon and flanker conflict interference in younger and older adults. AGING NEUROPSYCHOLOGY AND COGNITION 2021; 29:318-348. [PMID: 33472533 DOI: 10.1080/13825585.2021.1874278] [Citation(s) in RCA: 4] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 10/22/2022]
Abstract
Conflict processing and interference control have been popular topics of research in the study of pathological brain aging. However, there remains much to be learned about how these cognitive processes are altered in the course of healthy brain aging. Moreover, few studies have combined multiple measures of interference control using factorial designs. The aim of this study was to determine the nature of age-related changes in behavioral and electroencephalographic correlates of interference control using a factorial combination of the Simon and flanker interference conditions. Data were collected from a group of younger and high-functioning older adults. Behavioral results indicated the presence of conflict interference effects in both groups, that both Simon and flanker conflict effects are increased in high-functioning older adults, that the two types of conflict interference interact superadditively, and that older adults are more susceptible to the superadditive costs of multiple conflict types. ERP analyses revealed that early perceptual and response-selection processes are differentially modulated by flanker and Simon conflict respectively, however, there was no evidence that these early processes were impaired in older adults. Later components of the ERP in the P3 time range mirrored behavioral results, reflecting the increased susceptibility to flanker and Simon conflict in older adults.
Collapse
Affiliation(s)
- Rachel M Scrivano
- The College of Social Work, the Ohio State University, Columbus, United States
| | - Paul D Kieffaber
- Department of Psychological Sciences, College of William & Mary, Williamsburg, United States
| |
Collapse
|
10
|
Global precedence changes by environment: A systematic review and meta-analysis on effect of perceptual field variables on global-local visual processing. Atten Percept Psychophys 2020; 82:2348-2359. [PMID: 32189234 DOI: 10.3758/s13414-020-01997-1] [Citation(s) in RCA: 4] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/08/2022]
Abstract
Perceptual organization and, in particular, visual processing have been debated for many years. The global precedence effect in local-global visual processing, as introduced by David Navon, refers to the condition that global aspects of a scene are processed more rapidly than are local details. This perceptual dynamic is influenced by many factors that can be divided into two major categories: subjective or internal factors (e.g., age, disorder, culture) and the external factors called perceptual field variables (PFVs; e.g., stimulus size, eccentricity, sparsity). The aim of the current study was to identify the latter factors using a meta-analysis followed by a systematic literature review. In accordance of the standard framework suggested by PRISMA, 28 PFVs were observed through a literature search on articles published from 1982 to 2019, among which 10 factors have been qualified to be included in a meta-analysis. Subsequently, the random effects model proposed by Hedges and Olkin was used to estimate pooled effect sizes of PFVs. These effect sizes were used to compare and sort the PFVs on the basis of their intensity. According to Cohen's index, our analyses show that relevance, sparsity, and solidness type are categorized as small effects; visual field, level repetition, spatial frequency, and shape type are categorized as medium effects; and congruency, eccentricity, and size as large effect PFVs on global precedence.
Collapse
|
11
|
Elucidating the role of selective attention, divergent thinking, language abilities, and executive functions in metaphor generation. Neuropsychologia 2020; 142:107458. [PMID: 32275968 DOI: 10.1016/j.neuropsychologia.2020.107458] [Citation(s) in RCA: 8] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/19/2019] [Revised: 03/29/2020] [Accepted: 03/30/2020] [Indexed: 11/20/2022]
Abstract
Metaphoric language is one of the most common expressions of creative cognition in everyday life. However, the cognitive mechanisms underlying metaphor generation remain largely unexplained. In this study, we aimed to investigate the relationship between various cognitive functions and both novel and conventional metaphor generation. Ninety-five undergraduate students were administered a metaphor generation task that assesses novel and conventional metaphor generation, along with a battery of different cognitive measures: vocabulary; divergent thinking (Tel Aviv Creativity Test), working memory (WM) via digit span tests, executive functions (EFs) using the Behavioral Rating Inventory of Executive Function (BRIEF) questionnaire, and selective attention (lateralized global-local digit task). Results of a path analysis indicated that - whereas only selective attention contributed to conventional metaphor generation - selective attention, divergent thinking, and EFs contributed to novel metaphor generation beyond vocabulary and WM. Thus, the results indicate that although both novel and conventional metaphor generation are linked to attentional resources and inhibitory control, the greater creativity inherent in novel metaphor generation appears to reflect a more complex set of cognitive processes than conventional metaphor generation.
Collapse
|
12
|
Natekar P, Kori A, Krishnamurthi G. Demystifying Brain Tumor Segmentation Networks: Interpretability and Uncertainty Analysis. Front Comput Neurosci 2020; 14:6. [PMID: 32116620 PMCID: PMC7025464 DOI: 10.3389/fncom.2020.00006] [Citation(s) in RCA: 23] [Impact Index Per Article: 5.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/07/2019] [Accepted: 01/17/2020] [Indexed: 11/13/2022] Open
Abstract
The accurate automatic segmentation of gliomas and its intra-tumoral structures is important not only for treatment planning but also for follow-up evaluations. Several methods based on 2D and 3D Deep Neural Networks (DNN) have been developed to segment brain tumors and to classify different categories of tumors from different MRI modalities. However, these networks are often black-box models and do not provide any evidence regarding the process they take to perform this task. Increasing transparency and interpretability of such deep learning techniques is necessary for the complete integration of such methods into medical practice. In this paper, we explore various techniques to explain the functional organization of brain tumor segmentation models and to extract visualizations of internal concepts to understand how these networks achieve highly accurate tumor segmentations. We use the BraTS 2018 dataset to train three different networks with standard architectures and outline similarities and differences in the process that these networks take to segment brain tumors. We show that brain tumor segmentation networks learn certain human-understandable disentangled concepts on a filter level. We also show that they take a top-down or hierarchical approach to localizing the different parts of the tumor. We then extract visualizations of some internal feature maps and also provide a measure of uncertainty with regards to the outputs of the models to give additional qualitative evidence about the predictions of these networks. We believe that the emergence of such human-understandable organization and concepts might aid in the acceptance and integration of such methods in medical diagnosis.
Collapse
Affiliation(s)
- Parth Natekar
- Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India
| | - Avinash Kori
- Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India
| | | |
Collapse
|
13
|
Category learning can alter perception and its neural correlates. PLoS One 2019; 14:e0226000. [PMID: 31810079 PMCID: PMC6897555 DOI: 10.1371/journal.pone.0226000] [Citation(s) in RCA: 10] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/22/2019] [Accepted: 11/18/2019] [Indexed: 11/25/2022] Open
Abstract
Learned Categorical Perception (CP) occurs when the members of different categories come to look more dissimilar (“between-category separation”) and/or members of the same category come to look more similar (“within-category compression”) after a new category has been learned. To measure learned CP and its physiological correlates we compared dissimilarity judgments and Event Related Potentials (ERPs) before and after learning to sort multi-featured visual textures into two categories by trial and error with corrective feedback. With the same number of training trials and feedback, about half the subjects succeeded in learning the categories (“Learners”: criterion 80% accuracy) and the rest did not (“Non-Learners”). At both lower and higher levels of difficulty, successful Learners showed significant between-category separation—and, to a lesser extent, within-category compression—in pairwise dissimilarity judgments after learning, compared to before; their late parietal ERP positivity (LPC, usually interpreted as decisional) also increased and their occipital N1 amplitude (usually interpreted as perceptual) decreased. LPC amplitude increased with response accuracy and N1 amplitude decreased with between-category separation for the Learners. Non-Learners showed no significant changes in dissimilarity judgments, LPC or N1, within or between categories. This is behavioral and physiological evidence that category learning can alter perception. We sketch a neural net model predictive of this effect.
Collapse
|
14
|
Adelhöfer N, Gohil K, Passow S, Beste C, Li SC. Lateral prefrontal anodal transcranial direct current stimulation augments resolution of auditory perceptual-attentional conflicts. Neuroimage 2019; 199:217-227. [DOI: 10.1016/j.neuroimage.2019.05.009] [Citation(s) in RCA: 8] [Impact Index Per Article: 1.6] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/27/2019] [Revised: 05/01/2019] [Accepted: 05/04/2019] [Indexed: 01/24/2023] Open
|
15
|
Leek EC, Roberts MV, Dundon NM, Pegna AJ. Early sensitivity of evoked potentials to surface and volumetric structure during the visual perception of three-dimensional object shape. Eur J Neurosci 2018; 52:4453-4467. [PMID: 30447162 DOI: 10.1111/ejn.14270] [Citation(s) in RCA: 5] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/09/2017] [Revised: 10/11/2018] [Accepted: 10/15/2018] [Indexed: 11/26/2022]
Abstract
This study used event-related potentials (ERPs) to elucidate how the human visual system processes three-dimensional (3-D) object shape structure. In particular, we examined whether the perceptual mechanisms that support the analysis of 3-D shape are differentially sensitive to higher order surface and volumetric part structure. Observers performed a whole-part novel object matching task in which part stimuli comprised sub-regions of closed edge contour, surfaces or volumetric parts. Behavioural response latency data showed an advantage in matching surfaces and volumetric parts to whole objects over contours, but no difference between surfaces and volumes. ERPs were analysed using a convergence of approaches based on stimulus dependent amplitude modulations of evoked potentials, topographic segmentation, and spatial frequency oscillations. The results showed early differential perceptual processing of contours, surfaces, and volumetric part stimuli. This was first reliably observed over occipitoparietal electrodes during the N1 (140-200 ms) with a mean peak latency of 170 ms, and continued on subsequent P2 (220-260 ms) and N2 (260-320 ms) components. The differential sensitivity in perceptual processing during the N1 was accompanied by distinct microstate patterns that distinguished among contours, surfaces and volumes, and predominant theta band activity around 4-7 Hz over right occipitoparietal and orbitofrontal sites. These results provide the first evidence of early differential perceptual processing of higher order surface and volumetric shape structure within the first 200 ms of stimulus processing. The findings challenge theoretical models of object recognition that do not attribute functional significance to surface and volumetric object structure during visual perception.
Collapse
Affiliation(s)
- E Charles Leek
- School of Psychology, Institute of Life and Human Sciences, University of Liverpool, Liverpool, L69 7ZA, UK
| | | | - Neil M Dundon
- Brain Imaging Center, Department of Psychological and Brain Sciences, University of California, Santa Barbara, CA, USA.,Department of Child and Adolescent Psychiatry, Psychotherapy and Psychosomatics, University of Freiburg, Freiburg, Germany
| | - Alan J Pegna
- School of Psychology, University of Queensland, Saint Lucia, Qld, Australia
| |
Collapse
|
16
|
Children inhibit global information when the forest is dense and local information when the forest is sparse. J Exp Child Psychol 2018; 173:155-167. [PMID: 29723754 DOI: 10.1016/j.jecp.2018.03.020] [Citation(s) in RCA: 9] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/30/2017] [Revised: 01/19/2018] [Accepted: 03/31/2018] [Indexed: 11/24/2022]
Abstract
Visual environments are composed of global shapes and local details that compete for attentional resources. In adults, the global level is processed more rapidly than the local level, and global information must be inhibited in order to process local information when the local information and global information are in conflict. Compared with adults, children present less of a bias toward global visual information and appear to be more sensitive to the density of local elements that constitute the global level. The current study aimed, for the first time, to investigate the key role of inhibition during global/local processing in children. By including two different conditions of global saliency during a negative priming procedure, the results showed that when the global level was salient (dense hierarchical figures), 7-year-old children and adults needed to inhibit the global level to process the local information. However, when the global level was less salient (sparse hierarchical figures), only children needed to inhibit the local level to process the global information. These results confirm a weaker global bias and the greater impact of saliency in children than in adults. Moreover, the results indicate that, regardless of age, inhibition of the most salient hierarchical level is systematically required to select the less salient but more relevant level. These findings have important implications for future research in this area.
Collapse
|
17
|
Gerlach C, Poirel N. Navon's classical paradigm concerning local and global processing relates systematically to visual object classification performance. Sci Rep 2018; 8:324. [PMID: 29321634 PMCID: PMC5762637 DOI: 10.1038/s41598-017-18664-5] [Citation(s) in RCA: 20] [Impact Index Per Article: 3.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/10/2017] [Accepted: 12/15/2017] [Indexed: 11/23/2022] Open
Abstract
Forty years ago David Navon tried to tackle a central problem in psychology concerning the time course of perceptual processing: Do we first see the details (local level) followed by the overall outlay (global level) or is it rather the other way around? He did this by developing a now classical paradigm involving the presentation of compound stimuli; large letters composed of smaller letters. Despite the usefulness of this paradigm it remains uncertain whether effects found with compound stimuli relate directly to visual object recognition. It does so because compound stimuli are not actual objects but rather formations of elements and because the elements that form the global shape of compound stimuli are not features of the global shape but rather objects in their own right. To examine the relationship between performance on Navon’s paradigm and visual object processing we derived two indexes from Navon’s paradigm that reflect different aspects of the relationship between global and local processing. We find that individual differences on these indexes can explain a considerable amount of variance in two standard object classification paradigms; object decision and superordinate categorization, suggesting that Navon’s paradigm does relate to visual object processing.
Collapse
Affiliation(s)
- Christian Gerlach
- Department of Psychology, University of Southern Denmark, Odense, Denmark.
| | - Nicolas Poirel
- LaPsyDÉ, UMR 8240, CNRS, Université Paris Descartes, Université Caen Normandie, Paris, France.,Institut Universitaire de France (IUF), Paris, France
| |
Collapse
|
18
|
The influence of visual and phonological features on the hemispheric processing of hierarchical Navon letters. Neuropsychologia 2018; 109:75-85. [DOI: 10.1016/j.neuropsychologia.2017.12.013] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/19/2017] [Revised: 12/06/2017] [Accepted: 12/07/2017] [Indexed: 11/23/2022]
|
19
|
Black E, Stevenson JL, Bish JP. The Role of Musical Experience in Hemispheric Lateralization of Global and Local Auditory Processing. Perception 2017; 46:956-975. [DOI: 10.1177/0301006616685954] [Citation(s) in RCA: 6] [Impact Index Per Article: 0.9] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/17/2022]
Abstract
The global precedence effect is a phenomenon in which global aspects of visual and auditory stimuli are processed before local aspects. Individuals with musical experience perform better on all aspects of auditory tasks compared with individuals with less musical experience. The hemispheric lateralization of this auditory processing is less well-defined. The present study aimed to replicate the global precedence effect with auditory stimuli and to explore the lateralization of global and local auditory processing in individuals with differing levels of musical experience. A total of 38 college students completed an auditory-directed attention task while electroencephalography was recorded. Individuals with low musical experience responded significantly faster and more accurately in global trials than in local trials regardless of condition, and significantly faster and more accurately when pitches traveled in the same direction (compatible condition) than when pitches traveled in two different directions (incompatible condition) consistent with a global precedence effect. In contrast, individuals with high musical experience showed less of a global precedence effect with regards to accuracy, but not in terms of reaction time, suggesting an increased ability to overcome global bias. Further, a difference in P300 latency between hemispheres was observed. These findings provide a preliminary neurological framework for auditory processing of individuals with differing degrees of musical experience.
Collapse
Affiliation(s)
- Emily Black
- Department of Neurobiology and Anatomy, Drexel University College of Medicine, Philadelphia, PA, USA
| | | | | |
Collapse
|
20
|
Leek EC, Roberts M, Oliver ZJ, Cristino F, Pegna AJ. Early differential sensitivity of evoked-potentials to local and global shape during the perception of three-dimensional objects. Neuropsychologia 2016; 89:495-509. [DOI: 10.1016/j.neuropsychologia.2016.07.006] [Citation(s) in RCA: 15] [Impact Index Per Article: 1.9] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/08/2016] [Revised: 06/29/2016] [Accepted: 07/06/2016] [Indexed: 11/26/2022]
|
21
|
Xia C, Qi F, Shi G. Bottom-Up Visual Saliency Estimation With Deep Autoencoder-Based Sparse Reconstruction. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2016; 27:1227-1240. [PMID: 26800552 DOI: 10.1109/tnnls.2015.2512898] [Citation(s) in RCA: 11] [Impact Index Per Article: 1.4] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 06/05/2023]
Abstract
Research on visual perception indicates that the human visual system is sensitive to center-surround (C-S) contrast in the bottom-up saliency-driven attention process. Different from the traditional contrast computation of feature difference, models based on reconstruction have emerged to estimate saliency by starting from original images themselves instead of seeking for certain ad hoc features. However, in the existing reconstruction-based methods, the reconstruction parameters of each area are calculated independently without taking their global correlation into account. In this paper, inspired by the powerful feature learning and data reconstruction ability of deep autoencoders, we construct a deep C-S inference network and train it with the data sampled randomly from the entire image to obtain a unified reconstruction pattern for the current image. In this way, global competition in sampling and learning processes can be integrated into the nonlocal reconstruction and saliency estimation of each pixel, which can achieve better detection results than the models with separate consideration on local and global rarity. Moreover, by learning from the current scene, the proposed model can achieve the feature extraction and interaction simultaneously in an adaptive way, which can form a better generalization ability to handle more types of stimuli. Experimental results show that in accordance with different inputs, the network can learn distinct basic features for saliency modeling in its code layer. Furthermore, in a comprehensive evaluation on several benchmark data sets, the proposed method can outperform the existing state-of-the-art algorithms.
Collapse
|
22
|
Krakowski CS, Borst G, Pineau A, Houdé O, Poirel N. You can detect the trees as well as the forest when adding the leaves: evidence from visual search tasks containing three-level hierarchical stimuli. Acta Psychol (Amst) 2015; 157:131-43. [PMID: 25796055 DOI: 10.1016/j.actpsy.2015.03.001] [Citation(s) in RCA: 7] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/23/2014] [Revised: 02/16/2015] [Accepted: 03/04/2015] [Indexed: 10/23/2022] Open
Abstract
The present study investigated how multiple levels of hierarchical stimuli (i.e., global, intermediate and local) are processed during a visual search task. Healthy adults participated in a visual search task in which a target was either present or not at one of the three levels of hierarchical stimuli (global geometrical form made by intermediate forms themselves constituted by local forms). By varying the number of distractors, the results showed that targets presented at global and intermediate levels were detected efficiently (i.e., the detection times did not vary with the number of distractors) whereas local targets were processed less efficiently (i.e., the detection times increased with the number of distractors). Additional experiments confirmed that these results were not due to the size of the target elements or to the spatial proximity among the structural levels. Taken together, these results show that the most local level is always processed less efficiently, suggesting that it is disadvantaged during the competition for attentional resources compared to higher structural levels. The present study thus supports the view that the processing occurring in visual search acts dichotomously rather than continuously. Given that pure structuralist and pure space-based models of attention cannot account for the pattern of our findings, we discuss the implication for perception, attentional selection and executive control of target position on hierarchical stimuli.
Collapse
|