Smartphone Pupillometry for Anxiety and Cognitive Load

ISEF Category: Cellular and Molecular Biology

Subcategory: Neurobiology  ·  Difficulty: Advanced  ·  Setup: School Lab  ·  Time: 1 to 2 Months

The Hook

Your pupils change before you even notice stress. They can get wider when your brain works harder, and smaller when light or focus changes. That makes pupil size a sneaky signal for cognitive load and anxiety. You can test that signal with a smartphone camera and basic machine learning.

What Is It?

Pupillometry means measuring pupil size and pupil changes over time. The pupil is the dark center of your eye. It acts like a camera aperture. When the brain gets busy, the pupil often changes in a way that tracks attention, effort, and stress.

This project asks whether a smartphone camera can capture those changes well enough to predict cognitive load or self-reported anxiety. You might compare tasks like arithmetic and the Stroop task, which asks you to name the ink color of a word instead of reading the word itself. That conflict forces your brain to work harder. If your measurements are clean, you can turn a video of the eye into numbers, then ask whether those numbers match what the student reports feeling.

Why This Is a Good Topic

This is a strong science fair topic because you can measure a real biological signal, change one task at a time, and test a clear prediction. You do not need invasive gear. A phone camera, careful lighting, and good analysis can give you usable data. The topic connects to stress, attention, and mental workload, which makes the results easy to explain and easy to care about. You can also build real skills in signal processing, experimental design, and machine learning.

Research Questions

• How does arithmetic difficulty change pupil diameter and pupil fluctuation patterns during task performance?
• What is the effect of Stroop task conflict on pupil dynamics compared with a matched control task?
• Does ambient lighting level change the accuracy of smartphone-based pupil tracking?
• To what extent do pupil fluctuation spectra predict self-reported anxiety scores after a cognitively demanding task?
• Which pupil features, mean diameter, peak dilation, or frequency-domain power, best separate high-load from low-load trials?
• What is the effect of using subject-specific calibration on the accuracy of a pupil-based prediction model?

Basic Materials

• Smartphone with video recording and manual focus control.
• Tripod or phone stand.
• Plain chair and table.
• Printed Stroop task sheets or a task app.
• Basic arithmetic problem set with graded difficulty.
• Consistent desk lamp or ring light.
• Dark, neutral background behind the participant.
• Ruler or printed calibration card for scale reference.
• Stopwatch or timer.
• Survey form for self-reported anxiety rating.

Advanced Materials

• Eye-tracking-capable smartphone camera or external macro lens attachment.
• Laptop for video processing and model training.
• ImageJ or Python for frame-by-frame pupil measurement.
• OpenCV for pupil detection and tracking.
• Standardized anxiety questionnaire.
• Light meter or lux sensor.
• Calibrated monitor for presenting visual tasks.
• MATLAB or Python for signal analysis.
• Statistical software for mixed-effects or classification analysis.
• Consent forms and participant information sheets.

Software & Tools

• ImageJ: Measures pupil diameter from video frames and helps you check tracking quality.
• Python: Processes pupil time series, extracts features, and trains prediction models.
• OpenCV: Detects the eye and tracks pupil edges across frames.
• R: Runs statistical tests and mixed-effects models on your task data.
• JASP: Gives you a free, easy way to compare groups and inspect effect sizes.

Experiment Steps

1. Define the exact signal you will measure, such as pupil diameter, dilation change, or fluctuation spectrum.
2. Choose one task contrast first, like easy versus hard arithmetic or Stroop versus control, so your design stays clean.
3. Plan a lighting and camera setup that keeps the eye visible and limits outside noise in the signal.
4. Build a calibration plan so you can convert image size into a comparable pupil measure across sessions.
5. Decide which features you will extract from the pupil trace before you collect data, such as mean response, peak response, and frequency content.
6. Preplan the model and statistics you will use to compare task conditions and anxiety scores.

Common Pitfalls

• Changing room light between participants, which makes pupil size shift for reasons unrelated to stress or effort.
• Letting the phone autofocus hunt during recording, which blurs the pupil edge and breaks tracking.
• Using tasks that are too easy or too hard, which compresses the pupil response and hides group differences.
• Comparing raw pupil size without calibration, which mixes body size, distance from the camera, and signal quality.
• Training a machine learning model on too few participants, which makes the predictions look better than they really are.

What Makes This Competitive

A class-level project often stops at showing that pupil size changes during hard tasks. A stronger project goes further. You can compare multiple pupil features, test whether frequency-domain measures add predictive power, and use separate training and validation sets. You can also control lighting, baseline anxiety, and task difficulty more carefully than a simple demo would. That turns a neat observation into a real biomarker study.

Project Variations

• Test whether pupil dynamics differ between math anxiety and general anxiety by comparing two self-report scales.
• Compare smartphone pupil tracking under white light, warm light, and screen-based tasks to see which setup gives the cleanest signal.
• Use the same pipeline on reading, Stroop, and arithmetic tasks to see which one best predicts cognitive load.

Learn More

• NIH PubMed: Search for review articles on pupillometry, cognitive load, and anxiety to find the core biology behind the project.
• NCBI Bookshelf: Look for open textbook chapters on autonomic nervous system function and pupil control.
• MIT OpenCourseWare: Search for free materials on signal processing, machine learning, and experimental design.
• ImageJ Documentation: Read the official guides for measuring objects in video frames and extracting image features.
• OpenCV Documentation: Find tutorials on detecting circles, edges, and motion in video streams.
• NIH National Institute of Mental Health: Use the anxiety overview pages to understand self-report measures and research context.