Browsing by Subject "Visual Attention"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Bridging Visual Perception and Reasoning: A Visual Attention Perspective
    (2023-06) Chen, Shi
    One of the fundamental goals of Artificial Intelligence (AI) is to develop visual systems that can reason with the complexity of the world. Advances in machine learning have revolutionized many fields in computer vision, achieving human-level performance among several benchmark tasks and industrial applications. While the performance gap between machines and humans seems to be closing, the recent debates on the discrepancies between machine and human intelligence have also received a considerable amount of attention. Studies argue that existing vision models tend to use tactics different from human perception, and are vulnerable to even a tiny shift in visual domains. Evidence also suggests they commonly exploit statistical priors, instead of genuinely reasoning on the visual observations, and have yet to develop the capability to overcome issues resulting from spurious data biases. These contradictory observations strike the very heart of AI research, and bring attention to the question: How can AI systems understand the comprehensive range of visual concepts and reason with them to accomplish various real-life tasks, as we do on a daily basis? Humans learn much from little. With just a few relevant experiences, we are able to adapt to different situations. We also take advantage of inductive biases that can easily generalize, and avoid distraction from all kinds of statistical biases. The innate generalizability is a result of not only our profound understanding of the world but also the ways we perceive and reason with visual information. For instance, unlike machines that develop holistic understanding by scanning through the whole visual scene, humans prioritize their attention with a sequence of eye fixations. Guided by visual stimuli and the structured reasoning process, we progressively locate the regions of interest, and understand their semantic relationships as well as connections to the overall task. Despite the lack of comprehensive understanding of human vision, research on humans' visual behavior can provide abundant insights into the developments of vision models, and have the potential of contributing to AI systems that are practical for real-world scenarios. With an overarching goal of building visual systems with human-like reasoning capability, we focus on understanding and enhancing the integration between visual perception and reasoning. We leverage visual attention as an interface for studying how humans and machines prioritize their focus when reasoning with diverse visual scenes. We tackle the challenges by making progress from three distinct perspectives: From the visual perception perspective, we study the relationship between the accuracy of attention and the performance related to visual understanding; From the reasoning perspective, we pay attention to the connections between reasoning and visual perception, and study the roles of attention throughout the continuous decision-making process; Humans not only capture and reason on important information with high accuracy, but can also justify their rationales with supporting evidence. From the perspective of explainability, we explore the use of multi-modal explanations for justifying the rationales behind models' decisions. Our efforts provide an extensive collection of observations for demystifying the integration between perception and reasoning, and more importantly, they offer insights into the development of trustworthy AI systems with the help of human vision.
  • Thumbnail Image
    Item
    An Eye-Tracking Study of Experience-Driven Attention and Transfer to Related Tasks
    (2016-09) Salovich, Nikita A
    Spatial attention is frequently influenced by previous experiences, often without explicit awareness. This influence of previous experiences on spatial attention can lead to statistical learning and the formation of habitual attention––the tendency to prioritize locations that were frequently attended to in the past. The present study evaluated whether habitual attention transfers from a relatively impoverished task to a more realistic task as a first step in exploring the real-world applications of trained statistical learning. We induced habitual attention by training participants with a simple visual search task, which involved searching for the letter T amongst many letter Ls. This task was interleaved with a more realistic visual search task, where participants searched for an arrow against a road scene. Consistent with previous research, participants acquired habitual attention within T-among-L search task. Analyses of first saccadic eye movement, but not reaction time, showed a short-term transfer of habitual attention between the T-among-L search task and the map search task. Keywords: habitual attention, statistical learning, probability cuing, visual search
  • Thumbnail Image
    Item
    Visual-Motor Strategies During a Bilateral Visually-Guided Reaching Task in Typically-Developing Children
    (2023-05) Richardson, Alexandria
    Background. Bilateral coordination is developed in childhood and is critical for performing many everyday motor skills. Reaching is a fundamental component of many bilateral skills and may be performed either symmetrically to targets of the same distance and angle, or asymmetrically to targets requiring unique reaching trajectories. During a unilateral reach, vision is primarily directed to the end target. However, during a bilateral reach there is competition between end targets for visual resources. It is not well-characterized how children utilize visual-motor strategies during a bilateral reach. Objective. The goal of this study was to compare kinematic performance and gaze behavior between symmetrical and asymmetrical reaches in typically developing children ages 8-17 years old. Methods. Participants (n = 20) performed a bilateral visually-guided reaching task using the KINARM Exoskeleton robot with an integrated gaze tracker. Outcome metrics were designed to characterize the spatiotemporal performance of hand kinematics and gaze behavior. Results. Spatial error was high in asymmetrical reaches compared to symmetrical reaches, and for the non-dominant arm compared to the dominant arm. Regardless of target symmetry, arms were tightly coupled at movement onset but became desynchronized with a bias towards the dominant arm reaching movement offset first. The number of eye movements did not differ between symmetrical and asymmetrical trials. A directional bias for gaze was found towards the dominant end target. Conclusions. Despite greater error in asymmetrical reaches, a common gaze strategy may be used for both symmetrical and asymmetrical trials in which the number of eye movements does not change, and vision is primarily directed towards the dominant end target first and for a larger overall percentage of the trial.