Debiasing Central Fixation Confounds Reveals a Peripheral "Sweet Spot" for Human-like Scanpaths in Hard-Attention Vision

• URL: http://arxiv.org/abs/2602.14834v1
• Date: Mon, 16 Feb 2026 15:25:26 GMT
• Title: Debiasing Central Fixation Confounds Reveals a Peripheral "Sweet Spot" for Human-like Scanpaths in Hard-Attention Vision
• Abstract summary: Task-driven hard-attention models for vision are often evaluated by how well their scanpaths match human gaze.<n>We show that a trivial center-fixation baseline achieves surprisingly strong scanpath scores, approaching many learned policies.<n>We propose GCS, a center-debiased composite metric augmented with movement similarity.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Human eye movements in visual recognition reflect a balance between foveal sampling and peripheral context. Task-driven hard-attention models for vision are often evaluated by how well their scanpaths match human gaze. However, common scanpath metrics can be strongly confounded by dataset-specific center bias, especially on object-centric datasets. Using Gaze-CIFAR-10, we show that a trivial center-fixation baseline achieves surprisingly strong scanpath scores, approaching many learned policies. This makes standard metrics optimistic and blurs the distinction between genuine behavioral alignment and mere central tendency. We then analyze a hard-attention classifier under constrained vision by sweeping foveal patch size and peripheral context, revealing a peripheral sweet spot: only a narrow range of sensory constraints yields scanpaths that are simultaneously (i) above the center baseline after debiasing and (ii) temporally human-like in movement statistics. To address center bias, we propose GCS (Gaze Consistency Score), a center-debiased composite metric augmented with movement similarity. GCS uncovers a robust sweet spot at medium patch size with both foveal and peripheral vision, that is not obvious from raw scanpath metrics or accuracy alone, and also highlights a "shortcut regime" when the field-of-view becomes too large. We discuss implications for evaluating active perception on object-centric datasets and for designing gaze benchmarks that better separate behavioral alignment from center bias.

Related papers

• What They Saw, Not Just Where They Looked: Semantic Scanpath Similarity via VLMs and NLP metric [50.699179933343686]
  We present a semantic scanpath similarity framework that integrates vision-language models into eye-tracking analysis.<n>Semantic similarity is then computed using embedding-based and lexical NLP metrics.<n>Experiments on free-viewing eye-tracking data demonstrate that semantic similarity captures partially independent variance from geometric alignment.
  arXiv  Detail & Related papers  (2026-04-09T17:36:22Z)
• Focus-Scan-Refine: From Human Visual Perception to Efficient Visual Token Pruning [78.75062483648243]
  Vision-language models (VLMs) often generate massive visual tokens that greatly increase inference latency and memory footprint.<n>We propose Focus-Scan-Refine (FSR), a human-inspired, plug-and-play pruning framework that mimics how humans answer visual questions.<n>FSR consistently improves the accuracy-efficiency trade-off over existing state-of-the-art pruning methods.
  arXiv  Detail & Related papers  (2026-02-05T16:02:48Z)
• A Trainable Centrality Framework for Modern Data [3.1788482285024107]
  Measuring how central or typical a data point is underpins robust estimation, ranking, and outlier detection.<n>We introduce Fused Unified centrality Score Estimation (FUSE), a neural centrality framework that operates on top of arbitrary representations.
  arXiv  Detail & Related papers  (2025-11-28T08:04:38Z)
• RoboView-Bias: Benchmarking Visual Bias in Embodied Agents for Robotic Manipulation [67.38036090822982]
  We propose RoboView-Bias, the first benchmark specifically designed to quantify visual bias in robotic manipulation.<n>We create 2,127 task instances that enable robust measurement of biases induced by individual visual factors and their interactions.<n>Our results highlight that systematic analysis of visual bias is a prerequisite for developing safe and reliable general-purpose embodied agents.
  arXiv  Detail & Related papers  (2025-09-26T13:53:25Z)
• Visual Attention Graph [21.860357478331107]
  We propose a new attention representation, called Attention Graph, to simultaneously code the visual saliency and scanpath.<n>In the attention graph, the semantic-based scanpath is defined by the path on the graph, while saliency of objects can be obtained by computing fixation density on each node.
  arXiv  Detail & Related papers  (2025-03-11T15:22:44Z)
• NCDD: Nearest Centroid Distance Deficit for Out-Of-Distribution Detection in Gastrointestinal Vision [13.349151639789902]
  The integration of deep learning tools in gastrointestinal vision holds the potential for significant advancements in diagnosis, treatment, and patient care.<n>A major challenge, however, is these tools' tendency to make overconfident predictions, even when encountering unseen or newly emerging disease patterns.<n>We propose a novel nearest-centroid distance deficit (NCCD) score in the feature space for gastrointestinal OOD detection.
  arXiv  Detail & Related papers  (2024-12-02T15:07:55Z)
• OAT: Object-Level Attention Transformer for Gaze Scanpath Prediction [0.2796197251957245]
  This paper introduces the Object-level Attention Transformer (OAT) OAT predicts human scanpaths as they search for a target object within a cluttered scene of distractors. We evaluate OAT on the Amazon book cover dataset and a new dataset for visual search that we collected.
  arXiv  Detail & Related papers  (2024-07-18T09:33:17Z)
• An Inter-observer consistent deep adversarial training for visual scanpath prediction [66.46953851227454]
  We propose an inter-observer consistent adversarial training approach for scanpath prediction through a lightweight deep neural network. We show the competitiveness of our approach in regard to state-of-the-art methods.
  arXiv  Detail & Related papers  (2022-11-14T13:22:29Z)
• Active Gaze Control for Foveal Scene Exploration [124.11737060344052]
  We propose a methodology to emulate how humans and robots with foveal cameras would explore a scene. The proposed method achieves an increase in detection F1-score of 2-3 percentage points for the same number of gaze shifts.
  arXiv  Detail & Related papers  (2022-08-24T14:59:28Z)
• The Center of Attention: Center-Keypoint Grouping via Attention for Multi-Person Pose Estimation [0.0]
  CenterGroup is an attention-based framework to estimate human poses from a set of identity-agnostic keypoints and person center predictions in an image. Our method obtains state-of-the-art performance with up to 2.5x faster inference time than competing bottom-up methods.
  arXiv  Detail & Related papers  (2021-10-11T10:22:04Z)
• On-Off Center-Surround Receptive Fields for Accurate and Robust Image Classification [54.73738206624343]
  We show that OOCS pathways complement each other in their response to light stimuli, ensuring this way a strong edge-detection capability. We provide empirical evidence showing that networks supplied with the OOCS edge representation gain accuracy and illumination-robustness compared to standard deep models.
  arXiv  Detail & Related papers  (2021-06-13T20:55:16Z)
• High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identification [84.43394420267794]
  We propose a novel framework by learning high-order relation and topology information for discriminative features and robust alignment. Our framework significantly outperforms state-of-the-art by6.5%mAP scores on Occluded-Duke dataset.
  arXiv  Detail & Related papers  (2020-03-18T12:18:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.