Holo Flower Brain

Give invisible signals a living form.

Holo Flower Brain turns live EEG signals into a responsive flower. As research lead, I shaped the interaction and visual system, connecting a brain-sensing headband to a Unity-powered experience and a holographic display.

The design question was how to make an unfamiliar stream of data feel observable and personal. I explored a living form that people could watch change, alongside trained commands that let them rotate and lift it.

Developed at Harvard in 2019.

Role
Research Lead
Responsibilities
Interaction design, EEG-to-form mapping, visual development, Unity implementation, and iterative testing.
Platform
Collaboration with Emotiv EEG Headband 2019
Holo Flower displayed in a transparent holographic pyramid
Holo Flower Brain · Holographic display

A living form for live signals.

I mapped 14 EEG streams to the central lines of individual petals. This gave the changing form a repeatable structure: each channel had a place in the flower, rather than driving an arbitrary visual effect.

The flower became both a designed object and a responsive surface. The challenge was to preserve its identity while allowing the incoming data to deform it.

Signal → structure

FRONT AF3 F7 F3 FC5 T7 P7 O1 O2 P8 T8 FC6 F4 F8 AF4 P3 · CMSP4 · DRL14 sensing channels + 2 references · top view AF3F7F3FC5T7P7O1O2P8T8FC6F4F8AF4 Shared center
Hover over an electrode to reveal its petal.

Form → projection

One responsive form.
Four orientations.

Source
Deformed flower mesh
Layout
Radial arrangement
Display
Holographic pyramid
Original flower rendered in four orientations for the holographic pyramid01020304

Separate the signal from the command.

Continuous EEG input deforms the petals. Trained mental commands rotate and lift the flower. Keeping these behaviors distinct created two interaction paths: observing an incoming signal and intentionally changing the object.

I connected incoming values to the mesh through pressure-based deformation, then made the force adjustable to tune the response. The recordings show rotation and deformation alongside the adjustable-force iteration, keeping the flower’s form and behavior connected throughout testing.

HOLO / Brain interface

EEG input

Illustrative signals

AF3
F7
F3
FC5
T7
P7
O1
O2
P8
T8
FC6
F4
F8
AF4

Spatial response

Holo Flower rotation and deformation recording
Rotation and deformation
Holo Flower adjustable deformation force recording
Adjustable deformation force
Original motion recordings alongside illustrative EEG signals.

From response to understanding.

A facial-expression test produced a strong visual response. That observation exposed the next design challenge: a changing form can make a signal perceptible, but people still need a way to understand what caused the change.

The system established the sensing-to-form loop. The next step would be to test how people distinguish passive signal changes from intentional commands, and which feedback helps them learn that relationship.

Demonstrated
Live sensing, mapped deformation, trained commands, and a tablet interface.
Next question
What can people consistently understand from the changing form, and what remains ambiguous?

Emotion dimensions

ArousalHigh energy · +4
Low energy · −4
Unpleasant · −4ValencePleasant · +4
Select a positionSelect a point to inspect.
Valence
—
Arousal
—

Pair the signal with lived experience.

The research connected two scales of experience: live EEG feedback and a person’s longer-term account of sleep and dreaming. Holo Flower made the signal visible; Dream BCI explored the personal context around it.

The research included 102 survey responses and 10 follow-up interviews. Participants wanted more than a daily data report: they wanted patterns over time, flexible ways to record dreams, and reliable saving and export.

This led to a personal profile combining measured signals with self-reported context. The survey and sample-journal analysis below formed part of the same research.

Sleepiness by situation

Share of responses (%)0 5 10 15 20 25 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3Restingin afternoonPassengerin a carSittingand readingWatchingtelevisionSitting in apublic placeSitting andtalkingStoppedin trafficResponse likelihood: 0 Never / 1 Slight / 2 Moderate / 3 HighFall-asleep grouping 0.1 0.2 0.25 0.5 1 2 2+
Survey response distribution.

Emotion over time

Emotion intensity-10 -5 0 5 10 2017201820192020Date of dream entry Arousal Valence
Arousal and valence across journal entries.