Meet Chris
He loves playing video games in his spare time and spends hours in competitive lobbies, trying to become a better player than he was the day before. But these days, something is holding him back, as if he’s unable to hit the same marks he usually does.
Over the Spring 2026 semester at UC Berkeley as a part of a Human-AI Design course, our team of four partnered with HP to rethink what gaming technology can measure, moving past traditional hardware performance metrics toward an experience that interprets neural signals in real time with the ultimate goal of improving gameplay performance.
My hands were all over the pipeline: shaping interview questions, synthesizing research, ideating solutions, prototyping screens in Figma and front-end code, and integrating the Muse 2 headband into the system architecture.
THE DESIGN QUESTION
I observed and interviewed 19 players like Chris across the competitive spectrum, and three findings steered everything we built.
Research Synthesis
Putting an interviewee under a prototype EEG headset
01
Emotion drives performance
Players experience emotional volatility during high-intensity gameplay, where frustration negatively impacts focus and performance more than mechanical skill alone.
02
Immersion & Focus
Players create immersive gaming setups to improve focus and reaction time, with audio and haptic feedback helping maintain awareness and engagement.
03
Gaming as Stress Relief
For many players, gaming is a way to relax, decompress, and socialize, though competitive pressure can sometimes conflict with these goals.
Players wanted to improve and stay competitive, but they also wanted gaming to remain enjoyable and stress-relieving.
SYNTHESIS — USER RESEARCH
Ideation ranged wide. Using Chat-GPT’s experimental API, I stretched the concept space to 10 ideas, then narrowed hard to 2 against our timeline and technical constraints.
Using AI tools to evaluate our concepts based on usefulness, feasibility, and clarity, the team settled on a combination of these two initial ideas to further develop.
IDEA 1
Surround Visual Lighting
During the game, the lights mirror the colors and action on the screen for immersion. In moments of distress or potential negative snowballing, the lighting changes to a slow, calming breath-like glow.
IDEA 2
AI Coach Bot
Using all of the brain data from the headset EEG, an AI agent appears at the end of every game with a detailed summary of the performance based off of past experiences, with prompts that allow the user to explore more about a specific error or trend that they want to learn about.
01
EEG as input
Alpha and beta activity estimate mental states (calmness, alertness, stress, and loss of focus) while the player simply plays.
02
Feedback that whispers
When the system detects overwhelm, the interface shifts toward blue tones and offers short, supportive prompts — a reset, not an interruption.
03
Summaries, not scolding
AI-generated post-game summaries turn a session’s cognitive data into takeaways players can actually act on next match.
I connected Muse headband data through MuseLSL and used alpha/beta activity to estimate user’s state. The system creates a simple loop:
Feedback loop diagram
Demo of feedback loop
EEG signal → state detection → screen color feedback
Mid-fi prototypes for the post-game AI summary
The final prototype pairs the Muse 2 headband with an on-screen companion. As alpha and beta activity shift, NeuroLight estimates whether a player is calm, alert, stressed, or losing focus — and answers in the quietest way that works: the interface cools toward blue and a short supportive prompt helps the player reset mid-game.
After the session, an AI-generated summary translates the run’s cognitive data into plain-language takeaways. We built it as a working prototype — physical hardware included.
Working with physical hardware was new territory for me, but planning around the implications of every design decision made the path to a working prototype feel smooth and logical. Syncing with my teammates and our HP mentor showed me how design actually moves in a group setting, with 2-week, user-tested iteration cycles being essential in our overall direction. Next, I would run more play-testing to tune how early and how gently NeuroLight should step in. Overall, I am extremely proud with how much our team accomplished over the course of one semester!
The NeuroLight team at the showcase
PROJECT ROADMAP — JANUARY TO MAY 2026
