Enactive AI Experimental Research Platforms

My research program is supported by the development of experimental computational platforms that function as instruments for studying adaptive cognition and human–AI interaction. To investigate enactive AI architectures empirically, I develop experimental computational platforms that function as research instruments. Each system explores a different dimension of adaptive intelligence, ranging from drift-aware decision architectures to participatory human–AI creativity and therapeutic interaction dynamics. 

Adaptive Decision Systems

• Emergence Machine

Co-Creative Interaction Systems

• Aether

• AI Drawing Partner

Human Regulation and Therapeutic Systems

• Enactive Art Therapy Interfaces

Emergence Machine

Adaptive Decision Architecture for Dynamic Environments

The Emergence Machine is a real-time adaptive decision architecture designed to operate in environments where statistical structure evolves over time. Instead of relying solely on predictive optimization, the system continuously monitors the structural coherence of incoming data streams and detects drift in regime dynamics.

The architecture maintains multi-scale representations of behavioral regimes and identifies deviations that signal structural change. When drift is detected, the system reorganizes its behavioral posture—adjusting decision strategies, engagement levels, or response patterns to remain aligned with the current environment.

This approach reframes intelligence as the ability to maintain effective action under structural drift, rather than simply improving prediction accuracy. The system provides a testbed for studying adaptive decision-making in domains such as financial markets, autonomous systems, and other real-time dynamic environments.

The Emergence Machine operates entirely online, monitoring structural signals (EEG, EKG, infrastructure and energy consumption, stocks and bitcoin) and identifies regime shifts in real time, and adapts behavioral posture without reliance on static training assumptions. It learns through interaction on a real-time data stream. It operates primarily online, adapting behavior through interaction with a live data stream rather than relying on a fixed offline training dataset. Its objective is not simply to forecast outcomes, but to remain coherent under change.

The results demonstrate that the model successfully adapts to diverse dynamic environments across physiological and infrastructure datasets while maintaining low prediction error. Importantly, there was no offline training process. Across all domains, the system identifies multiple regimes and attractor structures, indicating sensitivity to evolving temporal patterns rather than reliance on a single stationary model. Physiological signals such as EEG exhibit a large number of attractors and regime transitions, reflecting the rapid and complex dynamics of neural activity, whereas energy infrastructure datasets exhibit fewer regimes and attractors consistent with slower structural change. The variation in model horizons further illustrates the system’s capacity to operate across multiple temporal scales—from sub-second neural dynamics to multi-hour infrastructure signals—supporting the claim that the architecture can maintain coherent behavior under structural drift across heterogeneous real-world environments.

Aether: Co-Creative Drawing AI

Interactive Platform for Participatory Sense-Making

Aether is an interactive co-creative drawing agent designed to explore how humans and AI can collaboratively construct meaning through creative interaction. The system participates in drawing sessions alongside a human partner, responding to strokes, patterns, and evolving visual structure in real time.

Rather than generating images independently, Aether engages in participatory sense-making, adapting its contributions based on the developing interaction dynamics between human and machine. The system analyzes spatial motifs, gesture patterns, and compositional structure in order to produce context-sensitive responses during the creative process.

This platform provides a controlled research environment for studying human–AI creative collaboration, enabling the investigation of how coordination, influence, and shared meaning emerge through interactive artistic practice.

AI Drawing Partner

Quantified Co-Creative Research Platform

The AI Drawing Partner extends the Aether system into a broader research platform for measuring and modeling human–AI co-creative interaction. The system captures detailed behavioral traces during drawing sessions, including stroke timing, spatial patterns, motif development, and interaction sequences between human and AI contributions.

These data streams allow researchers to analyze the cognitive, interactional, and collaborative dynamics of co-creation. The platform supports experiments investigating how humans interpret AI contributions, how collaborative structures evolve, and how creative agency is distributed across participants.

By quantifying these processes, the system enables empirical study of human–AI collaboration in creative domains, contributing new methods for evaluating co-creative AI systems.

Art Therapy Interfaces

Quantifying Regulation and Creative Interaction

The Art Therapy Interfaces project explores how interactive AI systems can support and analyze therapeutic creative processes. These interfaces combine drawing-based interaction with computational models that track behavioral dynamics during artistic activity.

The systems monitor features such as gesture patterns, interaction pacing, collaborative engagement, and evolving visual structure. These signals provide insight into regulation, engagement, and cognitive-emotional dynamics during creative practice.

By quantifying interaction patterns during artistic activity, the platform opens new possibilities for studying how co-creation, expression, and regulation interact within therapeutic contexts. The long-term goal is to develop adaptive AI systems that support mental health interventions while also providing measurable insight into the dynamics of creative therapy.

These platforms allow theoretical concepts from enactive cognitive science to be operationalized and empirically examined in real-time interactive environments.