Article
Multidisciplinary Sciences
Yonatan Vanunu, Jared M. Hotaling, Mike E. Le Pelley, Ben R. Newell
Summary: The study shows that initial attention in risky choice is driven by perceptual properties of the stimulus, while subsequent choices are more influenced by goal-driven factors. Options with the highest values and largest font sizes have the greatest impact on choice, while distractors may attract attention but do not affect actual decision-making.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2021)
Review
Multidisciplinary Sciences
Deryn O. LeDuke, Matilde Borio, Raymundo Miranda, Kay M. Tye
Summary: The functional interaction between bottom-up and top-down processing allows individuals to respond appropriately to the changing environment. These processing modalities can be represented using a dynamical systems model of the brain. The transition between attractor states depends on stability, depth, neuromodulatory tone, and plasticity changes. In disease states like anxiety or depression, understanding the relationship between these states is crucial. This study examines bottom-up and top-down processing from Marr's computational perspective to understand depressive and anxious disease states, using examples of amygdala and prefrontal cortex signaling.
ANNALS OF THE NEW YORK ACADEMY OF SCIENCES
(2023)
Article
Computer Science, Information Systems
Yingwei Pan, Yehao Li, Ting Yao, Tao Mei
Summary: This article discusses a technique for image captioning that utilizes a bottom-up and top-down attention mechanism to select and describe salient objects based on human subjective experience. The proposed approach uses the object sequence of interest as top-down signals to guide caption generation and incorporates an attention mechanism to integrate bottom-up and top-down signals. It also introduces a contrastive learning-based objective to ensure reliable and explainable cross-modal reasoning.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Ashkan Taghipour, Hassan Ghassemian
Summary: The paper introduces a visual attention approach to leverage hyperspectral anomaly detection, integrating spatial and spectral anomaly features, and outperforms six state-of-the-art anomaly detection methods in the experiment.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2021)
Article
Neurosciences
Leena Ali Ibrahim, Shuhan Huang, Marian Fernandez-Otero, Mia Sherer, Yanjie Qiu, Spurti Vemuri, Qing Xu, Robert Machold, Gabrielle Pouchelon, Bernardo Rudy, Gord Fishell
Summary: The study reveals that the establishment of top-down connectivity in the primary visual cortex critically depends on early thalamic inputs during postnatal development. Thalamic inputs predominantly target neurogliaform cells in layer 1, facilitating the subsequent strengthening of top-down inputs from the anterior cingulate cortex onto these cells. Sensory deprivation or removal of thalamic afferents blocks this phenomenon, highlighting the importance of bottom-up inputs for the development of top-down connections in the visual cortex.
Article
Computer Science, Artificial Intelligence
Zhixiong Nan, Jingjing Jiang, Xiaofeng Gao, Sanping Zhou, Weiliang Zuo, Ping Wei, Nanning Zheng
Summary: This paper presents a cognitively-explanatory deep neural network model to predict task-driven attention, which is evaluated on two datasets and further analyzed through ablation studies. The experiment results demonstrate the effectiveness of the proposed model.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Biochemistry & Molecular Biology
Prachi Patel, Kiki van der Heijden, Stephan Bickel, Jose L. Herrero, Ashesh D. Mehta, Nima Mesgarani
Summary: The representation of spatially separated simultaneous talkers and the modulation of neural representations of attended and unattended speech by talkers' locations and voices in the human auditory cortex are still unclear. This study measured neural responses from patients with implanted electrodes and found that spatial separation between talkers caused preferential encoding of contralateral speech. The location and spectrotemporal features of talkers were encoded in different aspects of neural response.
Article
Engineering, Electrical & Electronic
Cagdas Bas, Nazli Ikizler-Cinbis
Summary: This paper proposes an end-to-end, top-down and bottom-up attentional deep multiple instance learning approach for still image action recognition. The approach does not rely on attribute or object labels, and can successfully select action-related image regions and create fine-grained pixel-level action masks.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2022)
Article
Multidisciplinary Sciences
Juan Diaz-Colunga, Nanxi Lu, Alicia Sanchez-Gorostiaga, Chang-Yu Chang, Helen S. Cai, Joshua E. Goldford, Mikhail Tikhonov, Alvaro Sanchez
Summary: Microbial communities frequently invade one another as a whole, a phenomenon known as community coalescence. This study provides direct evidence that collective invasions can lead to ecological coselection in microbial communities due to cross-feeding interactions at the community level.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)
Article
Neurosciences
Qiang Li, Dinghong Gong, Yuan Zhang, Hongyi Zhang, Guangyuan Liu
Summary: This study investigates the cortical processing process of tonal working memory and finds it to be a bottom-up information transfer process. It also reveals that changes in tonal working memory load do not alter local brain activity patterns or response strength, but are linked to top-down attention control mechanisms involving the auditory cortex and an attention control network.
FRONTIERS IN NEUROSCIENCE
(2022)
Article
Multidisciplinary Sciences
Victor Pando-Naude, Agata Patyczek, Leonardo Bonetti, Peter Vuust
Summary: The human brain has the ability to integrate information from the environment with internally generated content, but the integration of top-down and bottom-up processes during complex multi-modal human activities is yet to be fully understood. Music listening leads to the urge to move, and music making entails both playing and listening, requiring different brain structures to process similar information.
SCIENTIFIC REPORTS
(2021)
Article
Automation & Control Systems
T. Burr, S. Croft, A. Favalli, T. Krieger, B. Weaver
Summary: This paper reviews progress and presents new results for uncertainty quantification in nuclear safeguards, focusing on data-driven choices in SD estimation, the use of approximate Bayesian computation, computational calibration, revisions to the GUM, and critique of a Unified Theory of Measurement Errors and Uncertainties.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2021)
Article
Biology
Hajo Eicken, Finn Danielsen, Josephine-Mary Sam, Maryann Fidel, Noor Johnson, Michael K. Poulsen, Olivia A. Lee, Katie Spellman, Lisbeth Iversen, Peter Pulsifer, Martin Enghoff
Summary: Effective responses to rapid environmental change rely on observations to inform planning and decision-making. Comparing top-down, large-scale program driven approaches with bottom-up approaches initiated and steered at the community level can yield benefits, especially when connecting to Indigenous and local knowledge.
Article
Chemistry, Multidisciplinary
Gyurim Park, Dong Yeon Jeong, Seung Yeon Yu, Jong Jin Park, Jong H. Kim, Hoichang Yang, Youngmin You
Summary: This research demonstrates an advancement in the engineering of emitters with circularly polarized luminescence (CPL) properties. By utilizing a combined top-down and bottom-up strategy, the photoluminescence quantum yield and the luminescence dissymmetry factor were simultaneously amplified. The integration of chiral anions and the formation of helical assemblies of square-planar Pt(II) complexes played crucial roles in activating the CPL-active metal-metal-to-ligand charge transfer transition.
ANGEWANDTE CHEMIE-INTERNATIONAL EDITION
(2023)
Article
Environmental Sciences
Andreas Froemelt, Arne Geschke, Thomas Wiedmann
Summary: This article discusses a highly detailed, spatially-resolved modelling framework that quantifies local activities and analyzes system-wide environmental and economic effects of planned interventions. Two case studies were conducted to examine production-based greenhouse gas emissions, consumption-based carbon footprints, and regional differences in Switzerland. The importance of providing regionalized information along economic value chains and the potential benefits and drawbacks of detailed scenarios were highlighted.
ENVIRONMENTAL RESEARCH LETTERS
(2021)