My midterm, Pitchy Bird, was a voice-controlled version of the classic Flappy Bird where the user’s vocal pitch determined the bird’s flight height. People leaned into microphones, experimenting with their vocal range to navigate obstacles. It was a fun experience. But for my final project, I want to take what learned to push it into physical computing and psychological testing.
I’m currently actively prototyping Flappy IAT, an endless runner controlled by an Arduino-based accelerometer that gamifies the Implicit Association Test (IAT).
The Concept
Instead of voice, players use a custom handheld controller (powered by an MMA8452Q accelerometer or a joystick). The mechanics require high-speed multitasking to bypass conscious cognitive filtering:
- Navigation (“Bump”): Players must physically “bump” the controller upward to flap the bird’s wings and maintain altitude against gravity.
- Categorization (“Tilt”): Every pipe obstacle displays a text stimulus (e.g., words like “Happy,” “Sad,” or category pairs like “Female Doctor”). To pass through the pipe safely, the player must tilt the controller Left (Green/Good) or Right (Red/Bad) to correctly categorize the word.
If it works, this game will be an active research tool more than only about survival. By forcing players to make categorization decisions under the pressure of keeping the bird afloat, the system exposes unconscious biases. The game tracks specific metrics via serial communication to p5.js:
- Reaction Time (RT): The milliseconds between seeing the word and tilting the device.
- Response Intensity: Measuring the angular velocity and degree of the tilt to detect hesitation or certainty.
- Cognitive Load: Increasing difficulty across three levels to force players into a “flow state” where implicit biases are harder to suppress.
Technical Implementation
The project utilizes a bidirectional data flow. The Arduino handles raw gesture detection (calibrating “rest” position to detect relative Bumps and Tilts), while p5.js manages the game state, visual feedback (green checks for correct tilts, red flashes for errors), and data logging. At the end of a session, the game exports a CSV file detailing the player’s performance and implicit bias metrics.
