Visual Attention and Saliency Detection

This study investigates how spatial attention affects performance in a deep neural network model of the visual system, finding that suppression in unattended regions is more effective than enhancement in attended regions under certain conditions.

The paper investigates the electrophysiological markers of statistically learned attentional enhancement and finds it differs from top-down attention by lacking alpha lateralization but showing increased N2pc amplitudes at high-probability locations.

The paper examines how target expectations influence electrophysiological indices of distractor processing in visual search, showing that distractor positivity (PD) is attenuated when target features vary randomly.

This study models human eye movements during immersive visual search using a reinforcement learning framework, finding that model predictions align with human gaze behavior.

The paper investigates how distractor-location frequency affects visual search performance and finds that local distractor frequency, not global frequency, primarily influences distractor interference.

The paper introduces an attentional search model inspired by visual pathways to locate targets in 3D environments using classifier and camera motion networks.

The paper shows that semantically inconsistent objects attract more fixations than consistent ones, but contextualized meaning maps fail to capture this relationship.

The paper uses a computational model to explain an ERP pattern associated with proactive suppression in visual search, offering a mechanistic account distinct from prior interpretations.

The paper develops a neurally constrained Bayesian model to explain differences in change detection success during a change blindness task by analyzing gaze metrics.

This study examines gaze behavior during a virtual navigation task, finding that participants fixate more on target objects than distractors and adjust their search patterns under divided attention.

The paper challenges the traditional interpretation of spatial-cueing effects by showing attention is deployed when search-relevant context appears, not immediately upon a salient cue.

The paper shows that spatial suppression of salient distractors improves target selection during visual search, as evidenced by reaction time and neural signal changes.

This study investigates neural communication dynamics in the Ventral Attention Network using EEG and fMRI, revealing context-specific alpha-band interactions during saliency processing.

The paper tests adaptive gain theory by linking pupil size to exploration and exploitation behaviors in a visual search task using eye movements.

The paper shows that EEG activity during visual search increases when targets appear in contextually appropriate locations in real scenes, and this neural response correlates with behavioral decisions.

The paper develops a mathematical model of exploration and exploitation in natural scene viewing and demonstrates its predictive power using Bayesian inference.

This study shows deep neural networks exhibit human-like limitations in visual search tasks due to object recognition, as evidenced by inverse correlation with search difficulty scores from natural images.

This study compares meaning maps and saliency models in predicting human fixations, finding that deep neural networks outperform meaning maps in accuracy and responsiveness to meaning changes.

The paper develops a dual visual processing model that improves saccade selection efficiency by combining foveal accuracy with peripheral visual field analysis.

This paper introduces a model of reflexive attention that simulates neural attractor states and their role in visual processing, linking them to behavioral and neural correlates like accuracy and EEG components.

The paper develops a neurodynamic model of V1 to predict visual attention by simulating retinal, thalamic, and cortical processes, and demonstrates its superiority over existing models in predicting saccade sequences.

The paper challenges the standard interpretation of spatial-cueing effects and supports the Priority Accumulation Framework by showing search performance deteriorates with increased search difficulty on valid-cue trials.

This paper introduces a recurrent generative model that mimics human visual attention by minimizing predictive uncertainty through saccadic eye movements.

The paper designs a brain-inspired CNN based on the ventral visual pathway to extract category-consistent features and shows it predicts attention control better than other models.

The paper shows that participants proactively suppress frequent distractor locations in visual search before stimuli are presented, as indicated by alpha oscillations and event-related potentials.

The paper shows individual differences in visual salience along semantic dimensions, with stable patterns across images and countries.

This study predicts eye movements from spatial priority maps reconstructed from fMRI data using a deep CNN model pre-trained for scene categorization.

This study investigates whether bottom-up attentional capture is influenced by prior expectations, finding no evidence of modulation by predictability.

The paper uses MEG decoding to show that spatial attention emerges later than category-based attention during visual search.

The paper shows that meaning, not just salience, guides attention in real-world scenes by comparing fixation patterns with meaning maps and salience maps.