About

Eric Ragan is an Associate Professor at the University of Florida in the Department of Computer & Information Science & Engineering (CISE). He leads the Indie (Interactive Data and Immersive Environments) research lab, which conducts research in human-computer interaction (HCI), human-centered computing (HCC), information visualization, virtual reality, 3D interaction, visual analytics, and trust in intelligent systems. The Indie Lab focuses on the design and evaluation of applications and techniques that support effective interaction and understanding of data, information, and virtual environments. Their research also includes explainable AI, and the group involves undergraduate and graduate students from multiple departments, collaborating actively with faculty across the university. Prior to his current position, Eric Ragan was an assistant professor at Texas A&M University in the Department of Visualization and the Department of Computer Science & Engineering. Before entering academia, he worked as a visual analytics research scientist at Oak Ridge National Laboratory as part of the Situation Awareness and Visual Analytics research team. He earned his Ph.D. in Computer Science from Virginia Tech. His work centers on human-centered research of interactive visualizations and immersive environments, contributing to advancing understanding and interaction with complex data and virtual spaces.

Research topics

• Computer Science
• Cognitive science
• World Wide Web
• Cognitive psychology
• Human–computer interaction
• Psychology
• Mathematics

Selected publications

• A Systematic Survey of Empirical User Studies of Unintentional Information Disclosure in Everyday Digital Interaction

  International Journal of Human-Computer Interaction · 2026-03-27

  articleSenior author
  PublisherDOI

• A Systematic Review on Human Roles, Solutions, and Methodological Approaches to Address Bias in AI

  ACM Computing Surveys · 2026-02-21

  articleOpen accessSenior author

  People play a significant role in designing, developing, and employing artificial intelligence (AI) systems. They can consider contextual information beyond the scope of AI models, thereby influencing system outcomes. At the same time, people’s choices or biases can introduce problems into the systems. This paradoxical scenario, in which people can both introduce and contribute to relieving the inherited machine bias, demands comprehensive and multidisciplinary approaches involving informed human interventions to improve systems’ performances and reduce their biases. Researchers across various communities have investigated multifaceted methods to reduce and mitigate bias in AI systems. Regardless of the method, humans are always involved in the debiasing method in one way or another, emphasizing the importance of human intervention during AI systems development. In this systematic review, we analyzed 100 peer-reviewed publications from various human-computer interaction (HCI) and machine learning (ML) venues. We discuss their research efforts to minimize data bias and algorithmic bias from three angles. First, we present a comprehensive taxonomy of bias mitigation solutions, analyzing the research methodologies and standard benchmarks for evaluating these solutions, highlighting the human researcher’s role in developing and evaluating solutions to address bias. Next, we identify humans’ roles in alleviating biases and specify how, when, and where their involvement occurs within the AI lifecycle. Finally, we summarize the research focus and methodologies across research disciplines. Our review revealed that, while technical solutions are essential, addressing bias requires a broad perspective that integrates human oversight, ethical frameworks, and interdisciplinary collaboration.

  PublisherDOI

• Assessment and Rehabilitation of Post-Stroke Visual Field Impairments: Current Approaches and Emerging Technologies

  Neurorehabilitation and neural repair · 2026-04-03

  article

  BackgroundVisual field defect following stroke can severely impair activities of daily living, such as visual exploration, reading, and mobility - significantly reducing quality of life. While neuroplasticity enables partial visual recovery in the first few months after stroke, the potential for recovery diminishes over time. Therefore, timely identification and treatment of visual field defects is critical.ObjectiveThis narrative review explores traditional methods and emerging technologies in visual assessment and rehabilitation. We examine traditional visual field screening methods, including static and kinetic perimetry. We also review current approaches to visual field rehabilitation, which include (1) visual substitution therapy, (2) eye movement-based therapy, and (3) visual restitution therapies such as border-field training, blindsight training, and brain stimulation. To address the limitations of current modalities, we explored technologies like wearable virtual reality headsets.ResultsVisual substitution therapies such as prisms expand the perceived visual field but are limited by adherence and side effects. Eye movement-based training improves scanning efficiency and functional performance. Visual restitution approaches, such as border-field and blindsight training, show inconsistent evidence for visual field restoration. Technologies such as virtual reality may offer more accessible and precise approaches to visual field screening and rehabilitation.ConclusionCurrent rehabilitation strategies show variable effectiveness in restoring visual fields. Further studies are needed to evaluate the effectiveness of virtual reality technologies.

  PublisherDOI

• CReLeRI: Explainable, Concept-centric, Representation, Learning, Reasoning, and Interaction Video Analysis System

  2025-10-25

  article

  Existing video analysis models often lack explainability, perform poorly on long videos, and frequently hallucinate. Commercial solutions are closed-source and costly. We introduce CReLeRI, an open-source system for action detection in untrimmed videos. CReLeRI segments videos using scene and action transitions, detects actions and their arguments and grounds them in 3D space to improve interpretability and reduce hallucinations. The system promotes transparency and trust in AI-driven analysis of complex, real-world videos. A demonstration video is also available.

  PublisherDOI

• User judgment of an AI model is biased by its description: A study in a job interview training context

  International Journal of Human-Computer Studies · 2025-11-20

  articleSenior authorCorresponding
  PublisherDOI

• Detection of Translation Gain is Decreased When Virtual Reality Users Are Unaware of Its Presence

  2025-11-12 · 1 citations

  articleSenior author

  The prevalent evaluation methods used to estimate detection of redirected walking are based on methods from psychophysics that require users to know their virtual movements are being manipulated. However, this higher-than-normal level of attention toward their movements yields conservative detection thresholds. We find that participants who were unaware that redirected walking (translation gain) was applied detected the technique at a significantly higher gain than users who were aware (at gains of 1.73 and 1.38, respectively). We provide evidence that redirected walking-based navigation solutions may be able to leverage gain values that are larger than the current threshold guidelines would suggest.

  PublisherDOI

• Empirical Study of Virtual Reality and Desktop Systems for Qualitative Editing of 3D Meshes: Impacts of Expertise and Context

  2025-11-12 · 1 citations

  articleSenior author

  For editing 3D spatial data, 3D interaction through virtual reality (VR) can be a viable alternative to 2D interfaces: research indicates that model editing in VR provides a more enjoyable experience and is fast to learn. These advantages make VR an appealing option for training new users’ spatial understanding before transitioning to standard 2D tools, like Blender and Maya. But how much does model editing in VR benefit the trained user? Our experiment compares the modeling accuracy of non-modelers, casual users, and formally trained artists for objects of varying complexity in desktop and VR. For users with no prior modeling experience, the study found significant improvements in qualitative accuracy and efficiency of aesthetic edits using VR. Importantly, improvements decreased with higher user experience and varied with types of editing for different surface features. The findings suggest that adaptation of traditional desktop modeling tools to VR should be situational decisions based on specific modeling scenarios.

  PublisherDOI

• That Ain’t Right: Assessing LLM Performance on QA in African American and West African English Dialects

  2025-01-01

  articleOpen access

  As Large Language Models (LLMs) gain mainstream public usage, understanding how users interact with them becomes increasingly important.Limited variety in training data raises concerns about LLM reliability across different language inputs.To explore this, we test several LLMs using functionally equivalent prompts expressed in different English dialects.We frame this analysis using Question-Answer (QA) pairs, which allow us to detect and evaluate appropriate and anomalous model behavior.We contribute a cross-LLM testing method and a new QA dataset translated into AAVE and WAPE variants.Results show a notable drop in accuracy for one dialect relative to the baseline.

  PublisherOA PDFDOI

• Privacy-by-design: Case studies in interactive record linkage using a hybrid human-computer system

  International Journal of Medical Informatics · 2025-07-31 · 1 citations

  articleOpen access

  OBJECTIVE: High-quality patient matching from several sources without a common identifier (ID) requires interactive record linkage (RL) using a hybrid human-computer system. MiNDFIRL (MInimum Necessary Disclosure For Interactive Record Linkage) is a hybrid prototype software system that facilitates maximizing linkage accuracy while minimizing information disclosure. We present and evaluate MiNDFIRL using two real-world case studies. MATERIALS AND METHODS: Two user studies were conducted linking 10,000 data pairs from EHR data and 18,240 unique patient IDs from patient generated data. After automated RL, manual review was conducted by three teams of four reviewers (12 total) using MiNDFIRL to resolve potential matches that required human judgment. Reviews for matches were conducted independently and disagreements were resolved through consensus. The teams then participated in a group discussion about MiNDFIRL using a semi-structured interview format. RESULTS AND DISCUSSION: The best algorithm, Random Forest, found 388 and 539 matches each for EHR and patient generated data algorithmically, but also output an additional 303 and 187 potential pairs that required manual review. 232 and 84 more matches were confirmed manually from these uncertain pairs respectively. Among the full uncertain pairs, only 30% of available identifying information was needed in MiNDFIRL to separate out 77% (232/303) and 45% (84/187) true linkages respectively. When available, first names and emails were the most frequently used fields in making RL decisions. CONCLUSION: On-demand access and masking techniques along with risk quantification through a hybrid human-computer system can significantly reduce disclosure while still minimizing false positives and false negatives in real-world RL.

  PublisherDOI

• Challenges of Precueing Instructions for Compound Task Procedures in Mixed Reality

  2025-05-26

  articleOpen access

  Augmented reality (AR) and virtual reality (VR) can enhance task guidance by overlaying visual information to improve efficiency and reduce errors. However, challenges remain in designing the appropriate presentation format and amount of information for real-time assistance. Prior research has shown benefits of visual cues in procedural tasks, but these findings are limited to simplified scenarios, highlighting a gap in understanding their effectiveness for complex, real-world applications. Therefore, we study visual design and cue effectiveness in the context of compound procedures encompassing subtasks and heterogeneous instructions. We present an experiment assessing different visual cues in VR to test a user’s ability to harness distinct information streams for different tasks, separating cues for object search and object placement for multi-step procedures. The results show that even for compound tasks requiring processing of multiple types of information, the addition of simple interaction cues for individual subtasks did significantly improved task performance for both time and errors. However, in contrast to prior studies showing successful precueing of future steps in more simplistic tasks, the study did not find evidence of precueing with the more complex tasks.

  PublisherDOI

Recent grants

• III: Medium: Collaborative Research: Towards Effective Interpretation of Deep Learning: Prediction, Representation, Modeling and Utilization

  NSF · $216k · 2019–2024

• CRII: III: Evaluating Provenance Visualizations for the Presentation and Communication of Investigative Data Analysis Processes

  NSF · $81k · 2018–2020

• CRII: III: Evaluating Provenance Visualizations for the Presentation and Communication of Investigative Data Analysis Processes

  NSF · $181k · 2016–2019

Frequent coauthors

• Doug A. Bowman

  26 shared

• Sina Mohseni

  Nvidia (United States)

  21 shared

• Mahsan Nourani

  Universidad del Noreste

  20 shared

• E. Świerczyński

  Nicolaus Copernicus University

  18 shared

• M. Mikołajewski

  Fraunhofer Institute for Industrial Engineering

  16 shared

• C. Gałan

  16 shared

• T. Tomov

  Alexandrovska Hospital

  16 shared

• P. Wychudzki

  Nicolaus Copernicus University

  16 shared