Jun 19, 2026 Luc Vanier Faculty Fellow, Fall 2025

Not Posture, but Development

Reflections on AI, movement learning, and student agency

This project began with a question about digital pedagogy: could AI help make aspects of movement learning more visible to students? Along the way, it changed how I understood movement itself.

F4I, or the Framework for Integration, is a movement framework built around the idea that human movement is not organized around maintaining a single ideal posture, but around the ability to cycle between different states of support and coordination. Rather than asking whether someone is in the “correct” position, the framework asks whether they can access multiple organizational states and move fluidly between them as conditions change.

When I first began exploring computer vision and pose estimation, I thought the challenge would be teaching a machine to recognize movement. What I discovered instead was that I needed to become much more precise about what I believed movement actually was.

Like many people working with posture and movement, I have inherited a cultural image of what “good posture” looks like. Most of us do. We tend to think mechanically: upright, aligned, balanced, stacked. Even when we claim to be teaching movement, we often find ourselves chasing positions.

I love that the machine exposed that assumption very quickly.

As I worked with systems like MediaPipe and later DeepLabCut, I realized that the problem wasn’t simply identifying body positions. The real question was whether a person could move between organizational states. A posture only tells us where someone is. It tells us very little about whether they can leave that state, enter another, and return anew again.

What I wanted the machine to learn was not posture. I wanted it to learn how movement develops.

In the F4I framework, the four phases—Beach Chair, Sandals, Ocean and Seagull—are not ideal positions to achieve. They are organizational states that the body cycles through as it coordinates support, movement, and attention. The value is not in remaining in one phase. The value is in being able to access all of them and transition between them when needed.

This realization gradually shifted my teaching. Instead of asking whether someone had good posture, I became more interested in whether they were adaptable. Could they access different organizational strategies? Could they transition easily between states? Could they respond to changing demands without becoming fixed in a single solution?

Importantly, I do not see this perspective as being in conflict with the expectations of the professional dance world. If anything, it helps explain them. Professional dancers are rarely valued simply because they can hold a particular shape. They are valued because they can adapt, respond, coordinate, and reorganize themselves under changing artistic and technical demands. The ability to move between organizational states may be one of the foundations that makes those qualities possible.

The technology reinforced this perspective. Every time the programming tried to reduce movement to a single correct position, the analysis became less meaningful. Every time I approached movement as a process of cycling, adaptation, and reorganization, new patterns emerged.

What excites me most about the project is that it may eventually provide a way to make this perspective visible. The original project, Digitally Reconstructing Ballet Pedagogy: AI Tools for Embodied Learning and Student Agency, was never simply about building a movement-analysis system. It was about exploring whether digital tools could help learners better understand their own embodied experience.

Although the project has not yet reached the stage of being shared with ballet dancers, the framework that has emerged points in that direction. The long-term goal is an application that can help make patterns of organization, adaptation, and movement variability more visible to learners. Rather than telling students how they should move, the aim is to give them richer information about how they are already organizing themselves.

In that sense, the project is fundamentally about agency. If students can perceive how they move between different organizational states—and where they may be becoming fixed in one while trying to reach another—they gain more opportunities to experiment, adapt, and make choices about their own movement. Rather than reinforcing a single image of “good posture,” the goal is to support a more developmental understanding of movement as an ongoing process of coordination and change.

For me, that has become the most important lesson of the project.

The goal is not to find the right posture.

The goal is not even to remain in the “best” state.

The goal is to remain capable of moving from state to state.