Spatial Experiential Embodiment Laboratory
SEE Lab is a research laboratory for Embodied Intelligence, directed by Sinclair Emoghene, Assistant Professor of Dance at The University of Texas at Austin. We study how movement carries knowledge, how that knowledge can be represented without removing it from the people and cultures that made it, and how computation can learn to read the body with greater precision and care.
The work begins with a simple problem. Dance and other forms of embodied knowledge are usually carried through practice, memory, repetition, and transmission from one body to another. Much of that knowledge disappears when the body stops moving because existing scholarly and computational systems were built to preserve text, images, and sound more easily than movement itself.
Schmidt Sciences
SEE Lab launches with support from Schmidt Sciences
Schmidt Sciences today named the Movement Language Model one of 19 projects funded through its Humanities and AI Virtual Institute (HAVI). The award supports building the Movement Language Model, the AI system at the center of SEE Lab, led by Sinclair Ogaga Emoghene and Dr. Lillian Chin at the University of Texas at Austin. The Movement Language Model opens as a beta in 2027. To request access, write to hello@seelab.dance.
Read the announcementPress and coverageWe are excited about where this work is going, and we would like to hear from you. If the Movement Language Model interests you, or you can see a collaboration in it, write to us at hello@seelab.dance. The research pages are where we post what the lab is doing as the work happens.
The lab brings together four connected parts: a movement corpus, a method for representing embodied knowledge, a computational model, and a publication platform where movement can become a citable scholarly object.
The corpus is built from movement recorded with the people and traditions that hold it. Every contribution enters on consented terms, and contributors retain authority over how their movement knowledge is stored, studied, published, and circulated.
The method is called Embodied Knowledge Quantization. It gives continuous embodied conditions a form that computation can work with while keeping interpretation with the people and cultures from which the movement comes. The structure can be represented. Its meaning cannot be reduced to a code.
The method works through a growing family of units, called emes, including:
- Kineme
- reads visible movement.
- Someme
- reads the body’s felt interior.
- Tereme
- reads the land and environment through which a body moves.
- Soneme
- reads sound.
- Arteme
- reads the movement recorded in made and inherited things.
- Opseme
- reads the expressive face.
- Ichneme
- reads the trace a body leaves on the surfaces it touches.
A separate unit, the scapeme, reads the moment when several emes converge in one moving body. This matters because movement never arrives as one condition at a time. A body moves through gravity, sensation, sound, culture, material, expression, and place together.
The Movement Language Model (MLM) is the computational system being developed from this research. It learns from human movement by compressing continuous movement into a vocabulary of 128 learned units called kinemes.
These units function as the smallest recurring structures through which larger movement phrases can be read, compared, and generated. The model also includes eight gravitational states that describe how a body negotiates weight, ground, and space:
- G-1 Grounding
- G-2 Rebounding
- G-3 Suspending
- G-4 Collapsing
- G-5 Resisting
- G-6 Yielding
- G-7 Propelling
- G-8 Hovering
The Movement Journal is the first peer-reviewed publication platform for scholarship in which movement is the primary research object.
The journal addresses a long-standing problem in dance scholarship. A dancer may hold and transmit significant knowledge, yet that knowledge often leaves no permanent object another scholar can cite. The Movement Journal gives published movement a Crossref DOI, preserves the record, and names the dancer or movement contributor as the author of the work.
The journal launched on 15 March 2026.
Publish your movementSEE Lab develops this research across four connected studios where movement can be captured, tokenized, compared, interpreted, and published.
The corpus currently includes movement recordings, with new material entering through ongoing fieldwork. Some recordings have already been published with permanent identifiers, while others remain under development, restricted access, or community review.
- 163
- Recordings
- 33
- Countries
- 75
- Traditions
- 38
- Published with a DOI
About
01 / 07