UCI Web Experiments

Tags: Behavioral Experiments · Human-Computer Interaction · JavaScript · Web Experiments · Decision Making · Wisdom of Crowds

This project is a collection of browser-based behavioral experiments developed at UC Irvine to study human perception, learning, and decision making. All experiments are implemented as lightweight web applications using JavaScript and HTML, and are designed for online deployment and data collection.

Together, these experiments explore how people form estimates, learn from feedback, and adapt their strategies under different informational and environmental constraints.

Experiments

1. Wisdom of Crowds — Temperature Estimation

Participants estimate the true temperature of a briefly presented visual sensor using information from multiple noisy thermometers. Across trials, feedback highlights the most accurate thermometer.

• 60 trials split into 3 randomized blocks
• Each block contains two expert information sources
• Tests whether participants learn to rely on experts or aggregate estimates
• Hosted using Firebase Hosting

2. Line Length Estimation (Baseline Data Collection)

A fast perceptual estimation task used to collect baseline human judgments. A black line is flashed briefly, followed by a ruler for reference, and participants estimate the line length using numeric keypresses.

• 100 ms stimulus presentation
• Numeric responses using keys 0–9
• 10 trials per participant
• Used as input data for downstream wisdom-of-crowds experiments

3. Agency and Instrumental Divergence Gambling Task

An online gambling task inspired by reinforcement learning and conditioning paradigms. Participants make repeated choices in environments with either high or low agency, defined by how much actions influence outcomes.

• High vs low instrumental divergence environments
• Learning and reward phases with changing outcome values
• Examines goal-directed versus habitual decision-making

Technical Overview

• Implemented using plain JavaScript and HTML
• Shared experiment structure with a public/ directory
• Centralized task logic in JavaScript files
• Designed for online deployment and scalable data collection

Links (Github repositories)

• Wisdom of Crowds Experiment
• Line Length Estimation Experiment
• Agency & Instrumental Divergence Experiment

Highlights

• Multiple experiments built on a shared web-based architecture
• Focus on estimation, learning, and adaptive decision-making
• Designed for real-world online data collection
• Clear progression from baseline perception to complex strategy learning