EyeLearn: Revolutionizing Learning Assessment Through Visual Cognition

Version 1

published

• Created

  April 30, 2025 18:57 by kxm5924

• Updated

  April 30, 2025 18:57 by [unknown user]

• Updated Description, Publication Date Show Changes

  April 30, 2025 18:59 by kxm5924

  Description

  • Learning disabilities represent a significant educational challenge affecting millions of students worldwide, yet current gold standard assessments like standardized testing and paper-based evaluations are fraught with critical limitations including cultural bias, outcome-centered evaluation, and inability to capture cognitive processes. Our novel approach leveraging eye tracking in virtual reality (VR) learning environments provides unprecedented access to previously invisible cognitive mechanisms, enabling objective, process-based assessment of learning capabilities where traditional methods fail. In our study, we developed a VR car reconstruction environment using Unity with integrated Tobii eye-tracking technology to capture participants' visual attention patterns during assembly tasks. The VR system accurately tracked gaze coordinates, fixation durations, and object interactions at 250Hz sampling rate while participants reconstructed vehicle components following a predefined sequence. We applied machine learning (ML) algorithms to analyze visual attention patterns and identify characteristic gaze behaviors associated with different learning abilities. Our analysis revealed a statistically significant increase in gaze pattern organization over time (p<0.01), indicating more structured visual strategies as learning progresses. The computational models successfully differentiated between learning patterns, with fast learners demonstrating targeted, non-linear scanning strategies focusing on critical components, while slower learners exhibited sequential, linear exploration patterns requiring greater visual coverage. These findings demonstrate the potential for eye-tracking data combined with advanced ML algorithms to automatically assess learning capabilities and identify potential learning difficulties earlier and more accurately than traditional methods.

  Publication Date

  • 2025-04
• Updated Acknowledgments Show Changes

  April 30, 2025 19:00 by kxm5924

  Acknowledgments

  • The authors of this work would like to acknowledge the NSF grants IIS-2302834 and MCB-1856132 for funding this research. Any opinions, findings, or conclusions found in this paper originate from the authors and do not necessarily reflect the views of the sponsor.
• Added Creator Kevin Mekulu

  April 30, 2025 19:00 by kxm5924

• Added GRE 2025.pdf

  April 30, 2025 19:02 by kxm5924

• Updated License Show Changes

  April 30, 2025 19:10 by kxm5924

  License

  • https://rightsstatements.org/page/InC/1.0/
• Published

  April 30, 2025 19:10 by kxm5924

• Updated

  April 30, 2025 22:05 by [unknown user]

• Updated Description Show Changes

  May 07, 2025 19:47 by avs5190

  Description

  • Learning disabilities represent a significant educational challenge affecting millions of students worldwide, yet current gold standard assessments like standardized testing and paper-based evaluations are fraught with critical limitations including cultural bias, outcome-centered evaluation, and inability to capture cognitive processes. Our novel approach leveraging eye tracking in virtual reality (VR) learning environments provides unprecedented access to previously invisible cognitive mechanisms, enabling objective, process-based assessment of learning capabilities where traditional methods fail. In our study, we developed a VR car reconstruction environment using Unity with integrated Tobii eye-tracking technology to capture participants' visual attention patterns during assembly tasks. The VR system accurately tracked gaze coordinates, fixation durations, and object interactions at 250Hz sampling rate while participants reconstructed vehicle components following a predefined sequence. We applied machine learning (ML) algorithms to analyze visual attention patterns and identify characteristic gaze behaviors associated with different learning abilities. Our analysis revealed a statistically significant increase in gaze pattern organization over time (p<0.01), indicating more structured visual strategies as learning progresses. The computational models successfully differentiated between learning patterns, with fast learners demonstrating targeted, non-linear scanning strategies focusing on critical components, while slower learners exhibited sequential, linear exploration patterns requiring greater visual coverage. These findings demonstrate the potential for eye-tracking data combined with advanced ML algorithms to automatically assess learning capabilities and identify potential learning difficulties earlier and more accurately than traditional methods.
  • Learning disabilities represent a significant educational challenge affecting millions of students worldwide, yet current gold standard assessments like standardized testing and paper-based evaluations are fraught with critical limitations including cultural bias, outcome-centered evaluation, and inability to capture cognitive processes. Our novel approach leveraging eye tracking in virtual reality (VR) learning environments provides unprecedented access to previously invisible cognitive mechanisms, enabling objective, process-based assessment of learning capabilities where traditional methods fail. In our study, we developed a VR car reconstruction environment using Unity with integrated Tobii eye-tracking technology to capture participants' visual attention patterns during assembly tasks. The VR system accurately tracked gaze coordinates, fixation durations, and object interactions at 250Hz sampling rate while participants reconstructed vehicle components following a predefined sequence. We applied machine learning (ML) algorithms to analyze visual attention patterns and identify characteristic gaze behaviors associated with different learning abilities. Our analysis revealed a statistically significant increase in gaze pattern organization over time (p<0.01), indicating more structured visual strategies as learning progresses. The computational models successfully differentiated between learning patterns, with fast learners demonstrating targeted, non-linear scanning strategies focusing on critical components, while slower learners exhibited sequential, linear exploration patterns requiring greater visual coverage. These findings demonstrate the potential for eye-tracking data combined with advanced ML algorithms to automatically assess learning capabilities and identify potential learning difficulties earlier and more accurately than traditional methods.
  • This poster was presented and the 2025 Graduate Exhibition.
• Added Creator Faisal Aqlan

  May 07, 2025 19:47 by avs5190

• Added Creator Hui Yang

  May 07, 2025 19:47 by avs5190

• Updated Description Show Changes

  May 07, 2025 19:49 by avs5190

  Description

  • Learning disabilities represent a significant educational challenge affecting millions of students worldwide, yet current gold standard assessments like standardized testing and paper-based evaluations are fraught with critical limitations including cultural bias, outcome-centered evaluation, and inability to capture cognitive processes. Our novel approach leveraging eye tracking in virtual reality (VR) learning environments provides unprecedented access to previously invisible cognitive mechanisms, enabling objective, process-based assessment of learning capabilities where traditional methods fail. In our study, we developed a VR car reconstruction environment using Unity with integrated Tobii eye-tracking technology to capture participants' visual attention patterns during assembly tasks. The VR system accurately tracked gaze coordinates, fixation durations, and object interactions at 250Hz sampling rate while participants reconstructed vehicle components following a predefined sequence. We applied machine learning (ML) algorithms to analyze visual attention patterns and identify characteristic gaze behaviors associated with different learning abilities. Our analysis revealed a statistically significant increase in gaze pattern organization over time (p<0.01), indicating more structured visual strategies as learning progresses. The computational models successfully differentiated between learning patterns, with fast learners demonstrating targeted, non-linear scanning strategies focusing on critical components, while slower learners exhibited sequential, linear exploration patterns requiring greater visual coverage. These findings demonstrate the potential for eye-tracking data combined with advanced ML algorithms to automatically assess learning capabilities and identify potential learning difficulties earlier and more accurately than traditional methods.
  • This poster was presented and the 2025 Graduate Exhibition.
  • This poster was presented at the 2025 Graduate Exhibition.